diff --git a/.dockerignore b/.dockerignore new file mode 100644 index 000000000..f36f86fbb --- /dev/null +++ b/.dockerignore @@ -0,0 +1,5 @@ +Dockerfile +.travis.yml +.gitignore +demo/etc +tox.ini diff --git a/.github/ISSUE_TEMPLATE.md b/.github/ISSUE_TEMPLATE.md new file mode 100644 index 000000000..d2050a3e4 --- /dev/null +++ b/.github/ISSUE_TEMPLATE.md @@ -0,0 +1,47 @@ + + +### Description + +Describe here the problem that you are experiencing, or the feature you are requesting. + +### Steps to reproduce + +- For bugs, list the steps +- that reproduce the bug +- using hyphens as bullet points + +Describe how what happens differs from what you expected. + +If you can identify any relevant log snippets from _homeserver.log_, please include +those here (please be careful to remove any personal or private data): + +### Version information + + + +- **Homeserver**: Was this issue identified on matrix.org or another homeserver? + +If not matrix.org: +- **Version**: What version of Synapse is running? +- **Install method**: package manager/git clone/pip +- **Platform**: Tell us about the environment in which your homeserver is operating + - distro, hardware, if it's running in a vm/container, etc. diff --git a/.gitignore b/.gitignore index 491047c35..9f42a7568 100644 --- a/.gitignore +++ b/.gitignore @@ -1,5 +1,6 @@ *.pyc .*.swp +*~ .DS_Store _trial_temp/ @@ -13,6 +14,7 @@ docs/build/ cmdclient_config.json homeserver*.db homeserver*.log +homeserver*.log.* homeserver*.pid homeserver*.yaml @@ -32,6 +34,7 @@ demo/media_store.* demo/etc uploads +cache .idea/ media_store/ @@ -39,6 +42,8 @@ media_store/ *.tac build/ +venv/ +venv*/ localhost-800*/ static/client/register/register_config.js @@ -46,3 +51,6 @@ static/client/register/register_config.js env/ *.config + +.vscode/ +.ropeproject/ diff --git a/.travis.yml b/.travis.yml index 3ce93cb43..b34b17af7 100644 --- a/.travis.yml +++ b/.travis.yml @@ -1,14 +1,33 @@ sudo: false language: python -python: 2.7 # tell travis to cache ~/.cache/pip cache: pip -env: - - TOX_ENV=packaging - - TOX_ENV=pep8 - - TOX_ENV=py27 +before_script: + - git remote set-branches --add origin develop + - git fetch origin develop + +matrix: + fast_finish: true + include: + - python: 2.7 + env: TOX_ENV=packaging + + - python: 2.7 + env: TOX_ENV=pep8 + + - python: 2.7 + env: TOX_ENV=py27 + + - python: 3.6 + env: TOX_ENV=py36 + + - python: 3.6 + env: TOX_ENV=check_isort + + - python: 3.6 + env: TOX_ENV=check-newsfragment install: - pip install tox diff --git a/AUTHORS.rst b/AUTHORS.rst index 3dcb1c2a8..e13ac5ad3 100644 --- a/AUTHORS.rst +++ b/AUTHORS.rst @@ -60,3 +60,6 @@ Niklas Riekenbrauck Christoph Witzany * Add LDAP support for authentication + +Pierre Jaury +* Docker packaging \ No newline at end of file diff --git a/CHANGES.md b/CHANGES.md new file mode 100644 index 000000000..8edd15855 --- /dev/null +++ b/CHANGES.md @@ -0,0 +1,2470 @@ +Synapse 0.33.0 (2018-07-19) +=========================== + +Bugfixes +-------- + +- Disable a noisy warning about logcontexts. ([\#3561](https://github.com/matrix-org/synapse/issues/3561)) + +Synapse 0.33.0rc1 (2018-07-18) +============================== + +Features +-------- + +- Enforce the specified API for report\_event. ([\#3316](https://github.com/matrix-org/synapse/issues/3316)) +- Include CPU time from database threads in request/block metrics. ([\#3496](https://github.com/matrix-org/synapse/issues/3496), [\#3501](https://github.com/matrix-org/synapse/issues/3501)) +- Add CPU metrics for \_fetch\_event\_list. ([\#3497](https://github.com/matrix-org/synapse/issues/3497)) +- Optimisation to make handling incoming federation requests more efficient. ([\#3541](https://github.com/matrix-org/synapse/issues/3541)) + +Bugfixes +-------- + +- Fix a significant performance regression in /sync. ([\#3505](https://github.com/matrix-org/synapse/issues/3505), [\#3521](https://github.com/matrix-org/synapse/issues/3521), [\#3530](https://github.com/matrix-org/synapse/issues/3530), [\#3544](https://github.com/matrix-org/synapse/issues/3544)) +- Use more portable syntax in our use of the attrs package, widening the supported versions. ([\#3498](https://github.com/matrix-org/synapse/issues/3498)) +- Fix queued federation requests being processed in the wrong order. ([\#3533](https://github.com/matrix-org/synapse/issues/3533)) +- Ensure that erasure requests are correctly honoured for publicly accessible rooms when accessed over federation. ([\#3546](https://github.com/matrix-org/synapse/issues/3546)) + +Misc +---- + +- Refactoring to improve testability. ([\#3351](https://github.com/matrix-org/synapse/issues/3351), [\#3499](https://github.com/matrix-org/synapse/issues/3499)) +- Use `isort` to sort imports. ([\#3463](https://github.com/matrix-org/synapse/issues/3463), [\#3464](https://github.com/matrix-org/synapse/issues/3464), [\#3540](https://github.com/matrix-org/synapse/issues/3540)) +- Use parse and asserts from http.servlet. ([\#3534](https://github.com/matrix-org/synapse/issues/3534), [\#3535](https://github.com/matrix-org/synapse/issues/3535)). + +Synapse 0.32.2 (2018-07-07) +=========================== + +Bugfixes +-------- + +- Amend the Python dependencies to depend on attrs from PyPI, not attr ([\#3492](https://github.com/matrix-org/synapse/issues/3492)) + +Synapse 0.32.1 (2018-07-06) +=========================== + +Bugfixes +-------- + +- Add explicit dependency on netaddr ([\#3488](https://github.com/matrix-org/synapse/issues/3488)) + +Changes in synapse v0.32.0 (2018-07-06) +======================================= + +No changes since 0.32.0rc1 + +Synapse 0.32.0rc1 (2018-07-05) +============================== + +Features +-------- + +- Add blacklist & whitelist of servers allowed to send events to a room via `m.room.server_acl` event. +- Cache factor override system for specific caches ([\#3334](https://github.com/matrix-org/synapse/issues/3334)) +- Add metrics to track appservice transactions ([\#3344](https://github.com/matrix-org/synapse/issues/3344)) +- Try to log more helpful info when a sig verification fails ([\#3372](https://github.com/matrix-org/synapse/issues/3372)) +- Synapse now uses the best performing JSON encoder/decoder according to your runtime (simplejson on CPython, stdlib json on PyPy). ([\#3462](https://github.com/matrix-org/synapse/issues/3462)) +- Add optional ip\_range\_whitelist param to AS registration files to lock AS IP access ([\#3465](https://github.com/matrix-org/synapse/issues/3465)) +- Reject invalid server names in federation requests ([\#3480](https://github.com/matrix-org/synapse/issues/3480)) +- Reject invalid server names in homeserver.yaml ([\#3483](https://github.com/matrix-org/synapse/issues/3483)) + +Bugfixes +-------- + +- Strip access\_token from outgoing requests ([\#3327](https://github.com/matrix-org/synapse/issues/3327)) +- Redact AS tokens in logs ([\#3349](https://github.com/matrix-org/synapse/issues/3349)) +- Fix federation backfill from SQLite servers ([\#3355](https://github.com/matrix-org/synapse/issues/3355)) +- Fix event-purge-by-ts admin API ([\#3363](https://github.com/matrix-org/synapse/issues/3363)) +- Fix event filtering in get\_missing\_events handler ([\#3371](https://github.com/matrix-org/synapse/issues/3371)) +- Synapse is now stricter regarding accepting events which it cannot retrieve the prev\_events for. ([\#3456](https://github.com/matrix-org/synapse/issues/3456)) +- Fix bug where synapse would explode when receiving unicode in HTTP User-Agent header ([\#3470](https://github.com/matrix-org/synapse/issues/3470)) +- Invalidate cache on correct thread to avoid race ([\#3473](https://github.com/matrix-org/synapse/issues/3473)) + +Improved Documentation +---------------------- + +- `doc/postgres.rst`: fix display of the last command block. Thanks to @ArchangeGabriel! ([\#3340](https://github.com/matrix-org/synapse/issues/3340)) + +Deprecations and Removals +------------------------- + +- Remove was\_forgotten\_at ([\#3324](https://github.com/matrix-org/synapse/issues/3324)) + +Misc +---- + +- [\#3332](https://github.com/matrix-org/synapse/issues/3332), [\#3341](https://github.com/matrix-org/synapse/issues/3341), [\#3347](https://github.com/matrix-org/synapse/issues/3347), [\#3348](https://github.com/matrix-org/synapse/issues/3348), [\#3356](https://github.com/matrix-org/synapse/issues/3356), [\#3385](https://github.com/matrix-org/synapse/issues/3385), [\#3446](https://github.com/matrix-org/synapse/issues/3446), [\#3447](https://github.com/matrix-org/synapse/issues/3447), [\#3467](https://github.com/matrix-org/synapse/issues/3467), [\#3474](https://github.com/matrix-org/synapse/issues/3474) + +Changes in synapse v0.31.2 (2018-06-14) +======================================= + +SECURITY UPDATE: Prevent unauthorised users from setting state events in a room when there is no `m.room.power_levels` event in force in the room. (PR #3397) + +Discussion around the Matrix Spec change proposal for this change can be followed at . + +Changes in synapse v0.31.1 (2018-06-08) +======================================= + +v0.31.1 fixes a security bug in the `get_missing_events` federation API where event visibility rules were not applied correctly. + +We are not aware of it being actively exploited but please upgrade asap. + +Bug Fixes: + +- Fix event filtering in get\_missing\_events handler (PR #3371) + +Changes in synapse v0.31.0 (2018-06-06) +======================================= + +Most notable change from v0.30.0 is to switch to the python prometheus library to improve system stats reporting. WARNING: this changes a number of prometheus metrics in a backwards-incompatible manner. For more details, see [docs/metrics-howto.rst](docs/metrics-howto.rst#removal-of-deprecated-metrics--time-based-counters-becoming-histograms-in-0310). + +Bug Fixes: + +- Fix metric documentation tables (PR #3341) +- Fix LaterGauge error handling (694968f) +- Fix replication metrics (b7e7fd2) + +Changes in synapse v0.31.0-rc1 (2018-06-04) +=========================================== + +Features: + +- Switch to the Python Prometheus library (PR #3256, #3274) +- Let users leave the server notice room after joining (PR #3287) + +Changes: + +- daily user type phone home stats (PR #3264) +- Use iter\* methods for \_filter\_events\_for\_server (PR #3267) +- Docs on consent bits (PR #3268) +- Remove users from user directory on deactivate (PR #3277) +- Avoid sending consent notice to guest users (PR #3288) +- disable CPUMetrics if no /proc/self/stat (PR #3299) +- Consistently use six\'s iteritems and wrap lazy keys/values in list() if they\'re not meant to be lazy (PR #3307) +- Add private IPv6 addresses to example config for url preview blacklist (PR #3317) Thanks to @thegcat! +- Reduce stuck read-receipts: ignore depth when updating (PR #3318) +- Put python\'s logs into Trial when running unit tests (PR #3319) + +Changes, python 3 migration: + +- Replace some more comparisons with six (PR #3243) Thanks to @NotAFile! +- replace some iteritems with six (PR #3244) Thanks to @NotAFile! +- Add batch\_iter to utils (PR #3245) Thanks to @NotAFile! +- use repr, not str (PR #3246) Thanks to @NotAFile! +- Misc Python3 fixes (PR #3247) Thanks to @NotAFile! +- Py3 storage/\_base.py (PR #3278) Thanks to @NotAFile! +- more six iteritems (PR #3279) Thanks to @NotAFile! +- More Misc. py3 fixes (PR #3280) Thanks to @NotAFile! +- remaining isintance fixes (PR #3281) Thanks to @NotAFile! +- py3-ize state.py (PR #3283) Thanks to @NotAFile! +- extend tox testing for py3 to avoid regressions (PR #3302) Thanks to @krombel! +- use memoryview in py3 (PR #3303) Thanks to @NotAFile! + +Bugs: + +- Fix federation backfill bugs (PR #3261) +- federation: fix LaterGauge usage (PR #3328) Thanks to @intelfx! + +Changes in synapse v0.30.0 (2018-05-24) +======================================= + +\'Server Notices\' are a new feature introduced in Synapse 0.30. They provide a channel whereby server administrators can send messages to users on the server. + +They are used as part of communication of the server policies (see `docs/consent_tracking.md`), however the intention is that they may also find a use for features such as \"Message of the day\". + +This feature is specific to Synapse, but uses standard Matrix communication mechanisms, so should work with any Matrix client. For more details see `docs/server_notices.md` + +Further Server Notices/Consent Tracking Support: + +- Allow overriding the server\_notices user\'s avatar (PR #3273) +- Use the localpart in the consent uri (PR #3272) +- Support for putting %(consent\_uri)s in messages (PR #3271) +- Block attempts to send server notices to remote users (PR #3270) +- Docs on consent bits (PR #3268) + +Changes in synapse v0.30.0-rc1 (2018-05-23) +=========================================== + +Server Notices/Consent Tracking Support: + +- ConsentResource to gather policy consent from users (PR #3213) +- Move RoomCreationHandler out of synapse.handlers.Handlers (PR #3225) +- Infrastructure for a server notices room (PR #3232) +- Send users a server notice about consent (PR #3236) +- Reject attempts to send event before privacy consent is given (PR #3257) +- Add a \'has\_consented\' template var to consent forms (PR #3262) +- Fix dependency on jinja2 (PR #3263) + +Features: + +- Cohort analytics (PR #3163, #3241, #3251) +- Add lxml to docker image for web previews (PR #3239) Thanks to @ptman! +- Add in flight request metrics (PR #3252) + +Changes: + +- Remove unused update\_external\_syncs (PR #3233) +- Use stream rather depth ordering for push actions (PR #3212) +- Make purge\_history operate on tokens (PR #3221) +- Don\'t support limitless pagination (PR #3265) + +Bug Fixes: + +- Fix logcontext resource usage tracking (PR #3258) +- Fix error in handling receipts (PR #3235) +- Stop the transaction cache caching failures (PR #3255) + +Changes in synapse v0.29.1 (2018-05-17) +======================================= + +Changes: + +- Update docker documentation (PR #3222) + +Changes in synapse v0.29.0 (2018-05-16) +======================================= + +Not changes since v0.29.0-rc1 + +Changes in synapse v0.29.0-rc1 (2018-05-14) +=========================================== + +Notable changes, a docker file for running Synapse (Thanks to @kaiyou!) and a closed spec bug in the Client Server API. Additionally further prep for Python 3 migration. + +Potentially breaking change: + +- Make Client-Server API return 401 for invalid token (PR #3161). + + This changes the Client-server spec to return a 401 error code instead of 403 when the access token is unrecognised. This is the behaviour required by the specification, but some clients may be relying on the old, incorrect behaviour. + + Thanks to @NotAFile for fixing this. + +Features: + +- Add a Dockerfile for synapse (PR #2846) Thanks to @kaiyou! + +Changes - General: + +- nuke-room-from-db.sh: added postgresql option and help (PR #2337) Thanks to @rubo77! +- Part user from rooms on account deactivate (PR #3201) +- Make \'unexpected logging context\' into warnings (PR #3007) +- Set Server header in SynapseRequest (PR #3208) +- remove duplicates from groups tables (PR #3129) +- Improve exception handling for background processes (PR #3138) +- Add missing consumeErrors to improve exception handling (PR #3139) +- reraise exceptions more carefully (PR #3142) +- Remove redundant call to preserve\_fn (PR #3143) +- Trap exceptions thrown within run\_in\_background (PR #3144) + +Changes - Refactors: + +- Refactor /context to reuse pagination storage functions (PR #3193) +- Refactor recent events func to use pagination func (PR #3195) +- Refactor pagination DB API to return concrete type (PR #3196) +- Refactor get\_recent\_events\_for\_room return type (PR #3198) +- Refactor sync APIs to reuse pagination API (PR #3199) +- Remove unused code path from member change DB func (PR #3200) +- Refactor request handling wrappers (PR #3203) +- transaction\_id, destination defined twice (PR #3209) Thanks to @damir-manapov! +- Refactor event storage to prepare for changes in state calculations (PR #3141) +- Set Server header in SynapseRequest (PR #3208) +- Use deferred.addTimeout instead of time\_bound\_deferred (PR #3127, #3178) +- Use run\_in\_background in preference to preserve\_fn (PR #3140) + +Changes - Python 3 migration: + +- Construct HMAC as bytes on py3 (PR #3156) Thanks to @NotAFile! +- run config tests on py3 (PR #3159) Thanks to @NotAFile! +- Open certificate files as bytes (PR #3084) Thanks to @NotAFile! +- Open config file in non-bytes mode (PR #3085) Thanks to @NotAFile! +- Make event properties raise AttributeError instead (PR #3102) Thanks to @NotAFile! +- Use six.moves.urlparse (PR #3108) Thanks to @NotAFile! +- Add py3 tests to tox with folders that work (PR #3145) Thanks to @NotAFile! +- Don\'t yield in list comprehensions (PR #3150) Thanks to @NotAFile! +- Move more xrange to six (PR #3151) Thanks to @NotAFile! +- make imports local (PR #3152) Thanks to @NotAFile! +- move httplib import to six (PR #3153) Thanks to @NotAFile! +- Replace stringIO imports with six (PR #3154, #3168) Thanks to @NotAFile! +- more bytes strings (PR #3155) Thanks to @NotAFile! + +Bug Fixes: + +- synapse fails to start under Twisted \>= 18.4 (PR #3157) +- Fix a class of logcontext leaks (PR #3170) +- Fix a couple of logcontext leaks in unit tests (PR #3172) +- Fix logcontext leak in media repo (PR #3174) +- Escape label values in prometheus metrics (PR #3175, #3186) +- Fix \'Unhandled Error\' logs with Twisted 18.4 (PR #3182) Thanks to @Half-Shot! +- Fix logcontext leaks in rate limiter (PR #3183) +- notifications: Convert next\_token to string according to the spec (PR #3190) Thanks to @mujx! +- nuke-room-from-db.sh: fix deletion from search table (PR #3194) Thanks to @rubo77! +- add guard for None on purge\_history api (PR #3160) Thanks to @krombel! + +Changes in synapse v0.28.1 (2018-05-01) +======================================= + +SECURITY UPDATE + +- Clamp the allowed values of event depth received over federation to be \[0, 2\^63 - 1\]. This mitigates an attack where malicious events injected with depth = 2\^63 - 1 render rooms unusable. Depth is used to determine the cosmetic ordering of events within a room, and so the ordering of events in such a room will default to using stream\_ordering rather than depth (topological\_ordering). + + This is a temporary solution to mitigate abuse in the wild, whilst a long term solution is being implemented to improve how the depth parameter is used. + + Full details at + +- Pin Twisted to \<18.4 until we stop using the private \_OpenSSLECCurve API. + +Changes in synapse v0.28.0 (2018-04-26) +======================================= + +Bug Fixes: + +- Fix quarantine media admin API and search reindex (PR #3130) +- Fix media admin APIs (PR #3134) + +Changes in synapse v0.28.0-rc1 (2018-04-24) +=========================================== + +Minor performance improvement to federation sending and bug fixes. + +(Note: This release does not include the delta state resolution implementation discussed in matrix live) + +Features: + +- Add metrics for event processing lag (PR #3090) +- Add metrics for ResponseCache (PR #3092) + +Changes: + +- Synapse on PyPy (PR #2760) Thanks to @Valodim! +- move handling of auto\_join\_rooms to RegisterHandler (PR #2996) Thanks to @krombel! +- Improve handling of SRV records for federation connections (PR #3016) Thanks to @silkeh! +- Document the behaviour of ResponseCache (PR #3059) +- Preparation for py3 (PR #3061, #3073, #3074, #3075, #3103, #3104, #3106, #3107, #3109, #3110) Thanks to @NotAFile! +- update prometheus dashboard to use new metric names (PR #3069) Thanks to @krombel! +- use python3-compatible prints (PR #3074) Thanks to @NotAFile! +- Send federation events concurrently (PR #3078) +- Limit concurrent event sends for a room (PR #3079) +- Improve R30 stat definition (PR #3086) +- Send events to ASes concurrently (PR #3088) +- Refactor ResponseCache usage (PR #3093) +- Clarify that SRV may not point to a CNAME (PR #3100) Thanks to @silkeh! +- Use str(e) instead of e.message (PR #3103) Thanks to @NotAFile! +- Use six.itervalues in some places (PR #3106) Thanks to @NotAFile! +- Refactor store.have\_events (PR #3117) + +Bug Fixes: + +- Return 401 for invalid access\_token on logout (PR #2938) Thanks to @dklug! +- Return a 404 rather than a 500 on rejoining empty rooms (PR #3080) +- fix federation\_domain\_whitelist (PR #3099) +- Avoid creating events with huge numbers of prev\_events (PR #3113) +- Reject events which have lots of prev\_events (PR #3118) + +Changes in synapse v0.27.4 (2018-04-13) +======================================= + +Changes: + +- Update canonicaljson dependency (\#3095) + +Changes in synapse v0.27.3 (2018-04-11) +====================================== + +Bug fixes: + +- URL quote path segments over federation (\#3082) + +Changes in synapse v0.27.3-rc2 (2018-04-09) +=========================================== + +v0.27.3-rc1 used a stale version of the develop branch so the changelog overstates the functionality. v0.27.3-rc2 is up to date, rc1 should be ignored. + +Changes in synapse v0.27.3-rc1 (2018-04-09) +=========================================== + +Notable changes include API support for joinability of groups. Also new metrics and phone home stats. Phone home stats include better visibility of system usage so we can tweak synpase to work better for all users rather than our own experience with matrix.org. Also, recording \'r30\' stat which is the measure we use to track overal growth of the Matrix ecosystem. It is defined as:- + +Counts the number of native 30 day retained users, defined as:- \* Users who have created their accounts more than 30 days + +: - Where last seen at most 30 days ago + - Where account creation and last\_seen are \> 30 days\" + +Features: + +- Add joinability for groups (PR #3045) +- Implement group join API (PR #3046) +- Add counter metrics for calculating state delta (PR #3033) +- R30 stats (PR #3041) +- Measure time it takes to calculate state group ID (PR #3043) +- Add basic performance statistics to phone home (PR #3044) +- Add response size metrics (PR #3071) +- phone home cache size configurations (PR #3063) + +Changes: + +- Add a blurb explaining the main synapse worker (PR #2886) Thanks to @turt2live! +- Replace old style error catching with \'as\' keyword (PR #3000) Thanks to @NotAFile! +- Use .iter\* to avoid copies in StateHandler (PR #3006) +- Linearize calls to \_generate\_user\_id (PR #3029) +- Remove last usage of ujson (PR #3030) +- Use simplejson throughout (PR #3048) +- Use static JSONEncoders (PR #3049) +- Remove uses of events.content (PR #3060) +- Improve database cache performance (PR #3068) + +Bug fixes: + +- Add room\_id to the response of rooms/{roomId}/join (PR #2986) Thanks to @jplatte! +- Fix replication after switch to simplejson (PR #3015) +- 404 correctly on missing paths via NoResource (PR #3022) +- Fix error when claiming e2e keys from offline servers (PR #3034) +- fix tests/storage/test\_user\_directory.py (PR #3042) +- use PUT instead of POST for federating groups/m.join\_policy (PR #3070) Thanks to @krombel! +- postgres port script: fix state\_groups\_pkey error (PR #3072) + +Changes in synapse v0.27.2 (2018-03-26) +======================================= + +Bug fixes: + +- Fix bug which broke TCP replication between workers (PR #3015) + +Changes in synapse v0.27.1 (2018-03-26) +======================================= + +Meta release as v0.27.0 temporarily pointed to the wrong commit + +Changes in synapse v0.27.0 (2018-03-26) +======================================= + +No changes since v0.27.0-rc2 + +Changes in synapse v0.27.0-rc2 (2018-03-19) +=========================================== + +Pulls in v0.26.1 + +Bug fixes: + +- Fix bug introduced in v0.27.0-rc1 that causes much increased memory usage in state cache (PR #3005) + +Changes in synapse v0.26.1 (2018-03-15) +======================================= + +Bug fixes: + +- Fix bug where an invalid event caused server to stop functioning correctly, due to parsing and serializing bugs in ujson library (PR #3008) + +Changes in synapse v0.27.0-rc1 (2018-03-14) +=========================================== + +The common case for running Synapse is not to run separate workers, but for those that do, be aware that synctl no longer starts the main synapse when using `-a` option with workers. A new worker file should be added with `worker_app: synapse.app.homeserver`. + +This release also begins the process of renaming a number of the metrics reported to prometheus. See [docs/metrics-howto.rst](docs/metrics-howto.rst#block-and-response-metrics-renamed-for-0-27-0). Note that the v0.28.0 release will remove the deprecated metric names. + +Features: + +- Add ability for ASes to override message send time (PR #2754) +- Add support for custom storage providers for media repository (PR #2867, #2777, #2783, #2789, #2791, #2804, #2812, #2814, #2857, #2868, #2767) +- Add purge API features, see [docs/admin\_api/purge\_history\_api.rst](docs/admin_api/purge_history_api.rst) for full details (PR #2858, #2867, #2882, #2946, #2962, #2943) +- Add support for whitelisting 3PIDs that users can register. (PR #2813) +- Add `/room/{id}/event/{id}` API (PR #2766) +- Add an admin API to get all the media in a room (PR #2818) Thanks to @turt2live! +- Add `federation_domain_whitelist` option (PR #2820, #2821) + +Changes: + +- Continue to factor out processing from main process and into worker processes. See updated [docs/workers.rst](docs/workers.rst) (PR #2892 - \#2904, #2913, #2920 - \#2926, #2947, #2847, #2854, #2872, #2873, #2874, #2928, #2929, #2934, #2856, #2976 - \#2984, #2987 - \#2989, #2991 - \#2993, #2995, #2784) +- Ensure state cache is used when persisting events (PR #2864, #2871, #2802, #2835, #2836, #2841, #2842, #2849) +- Change the default config to bind on both IPv4 and IPv6 on all platforms (PR #2435) Thanks to @silkeh! +- No longer require a specific version of saml2 (PR #2695) Thanks to @okurz! +- Remove `verbosity`/`log_file` from generated config (PR #2755) +- Add and improve metrics and logging (PR #2770, #2778, #2785, #2786, #2787, #2793, #2794, #2795, #2809, #2810, #2833, #2834, #2844, #2965, #2927, #2975, #2790, #2796, #2838) +- When using synctl with workers, don\'t start the main synapse automatically (PR #2774) +- Minor performance improvements (PR #2773, #2792) +- Use a connection pool for non-federation outbound connections (PR #2817) +- Make it possible to run unit tests against postgres (PR #2829) +- Update pynacl dependency to 1.2.1 or higher (PR #2888) Thanks to @bachp! +- Remove ability for AS users to call /events and /sync (PR #2948) +- Use bcrypt.checkpw (PR #2949) Thanks to @krombel! + +Bug fixes: + +- Fix broken `ldap_config` config option (PR #2683) Thanks to @seckrv! +- Fix error message when user is not allowed to unban (PR #2761) Thanks to @turt2live! +- Fix publicised groups GET API (singular) over federation (PR #2772) +- Fix user directory when using `user_directory_search_all_users` config option (PR #2803, #2831) +- Fix error on `/publicRooms` when no rooms exist (PR #2827) +- Fix bug in quarantine\_media (PR #2837) +- Fix url\_previews when no Content-Type is returned from URL (PR #2845) +- Fix rare race in sync API when joining room (PR #2944) +- Fix slow event search, switch back from GIST to GIN indexes (PR #2769, #2848) + +Changes in synapse v0.26.0 (2018-01-05) +======================================= + +No changes since v0.26.0-rc1 + +Changes in synapse v0.26.0-rc1 (2017-12-13) +=========================================== + +Features: + +- Add ability for ASes to publicise groups for their users (PR #2686) +- Add all local users to the user\_directory and optionally search them (PR #2723) +- Add support for custom login types for validating users (PR #2729) + +Changes: + +- Update example Prometheus config to new format (PR #2648) Thanks to @krombel! +- Rename redact\_content option to include\_content in Push API (PR #2650) +- Declare support for r0.3.0 (PR #2677) +- Improve upserts (PR #2684, #2688, #2689, #2713) +- Improve documentation of workers (PR #2700) +- Improve tracebacks on exceptions (PR #2705) +- Allow guest access to group APIs for reading (PR #2715) +- Support for posting content in federation\_client script (PR #2716) +- Delete devices and pushers on logouts etc (PR #2722) + +Bug fixes: + +- Fix database port script (PR #2673) +- Fix internal server error on login with ldap\_auth\_provider (PR #2678) Thanks to @jkolo! +- Fix error on sqlite 3.7 (PR #2697) +- Fix OPTIONS on preview\_url (PR #2707) +- Fix error handling on dns lookup (PR #2711) +- Fix wrong avatars when inviting multiple users when creating room (PR #2717) +- Fix 500 when joining matrix-dev (PR #2719) + +Changes in synapse v0.25.1 (2017-11-17) +======================================= + +Bug fixes: + +- Fix login with LDAP and other password provider modules (PR #2678). Thanks to @jkolo! + +Changes in synapse v0.25.0 (2017-11-15) +======================================= + +Bug fixes: + +- Fix port script (PR #2673) + +Changes in synapse v0.25.0-rc1 (2017-11-14) +=========================================== + +Features: + +- Add is\_public to groups table to allow for private groups (PR #2582) +- Add a route for determining who you are (PR #2668) Thanks to @turt2live! +- Add more features to the password providers (PR #2608, #2610, #2620, #2622, #2623, #2624, #2626, #2628, #2629) +- Add a hook for custom rest endpoints (PR #2627) +- Add API to update group room visibility (PR #2651) + +Changes: + +- Ignore \ tags when generating URL preview descriptions (PR #2576) Thanks to @maximevaillancourt! +- Register some /unstable endpoints in /r0 as well (PR #2579) Thanks to @krombel! +- Support /keys/upload on /r0 as well as /unstable (PR #2585) +- Front-end proxy: pass through auth header (PR #2586) +- Allow ASes to deactivate their own users (PR #2589) +- Remove refresh tokens (PR #2613) +- Automatically set default displayname on register (PR #2617) +- Log login requests (PR #2618) +- Always return is\_public in the /groups/:group\_id/rooms API (PR #2630) +- Avoid no-op media deletes (PR #2637) Thanks to @spantaleev! +- Fix various embarrassing typos around user\_directory and add some doc. (PR #2643) +- Return whether a user is an admin within a group (PR #2647) +- Namespace visibility options for groups (PR #2657) +- Downcase UserIDs on registration (PR #2662) +- Cache failures when fetching URL previews (PR #2669) + +Bug fixes: + +- Fix port script (PR #2577) +- Fix error when running synapse with no logfile (PR #2581) +- Fix UI auth when deleting devices (PR #2591) +- Fix typo when checking if user is invited to group (PR #2599) +- Fix the port script to drop NUL values in all tables (PR #2611) +- Fix appservices being backlogged and not receiving new events due to a bug in notify\_interested\_services (PR #2631) Thanks to @xyzz! +- Fix updating rooms avatar/display name when modified by admin (PR #2636) Thanks to @farialima! +- Fix bug in state group storage (PR #2649) +- Fix 500 on invalid utf-8 in request (PR #2663) + +Changes in synapse v0.24.1 (2017-10-24) +======================================= + +Bug fixes: + +- Fix updating group profiles over federation (PR #2567) + +Changes in synapse v0.24.0 (2017-10-23) +======================================= + +No changes since v0.24.0-rc1 + +Changes in synapse v0.24.0-rc1 (2017-10-19) +=========================================== + +Features: + +- Add Group Server (PR #2352, #2363, #2374, #2377, #2378, #2382, #2410, #2426, #2430, #2454, #2471, #2472, #2544) +- Add support for channel notifications (PR #2501) +- Add basic implementation of backup media store (PR #2538) +- Add config option to auto-join new users to rooms (PR #2545) + +Changes: + +- Make the spam checker a module (PR #2474) +- Delete expired url cache data (PR #2478) +- Ignore incoming events for rooms that we have left (PR #2490) +- Allow spam checker to reject invites too (PR #2492) +- Add room creation checks to spam checker (PR #2495) +- Spam checking: add the invitee to user\_may\_invite (PR #2502) +- Process events from federation for different rooms in parallel (PR #2520) +- Allow error strings from spam checker (PR #2531) +- Improve error handling for missing files in config (PR #2551) + +Bug fixes: + +- Fix handling SERVFAILs when doing AAAA lookups for federation (PR #2477) +- Fix incompatibility with newer versions of ujson (PR #2483) Thanks to @jeremycline! +- Fix notification keywords that start/end with non-word chars (PR #2500) +- Fix stack overflow and logcontexts from linearizer (PR #2532) +- Fix 500 error when fields missing from power\_levels event (PR #2552) +- Fix 500 error when we get an error handling a PDU (PR #2553) + +Changes in synapse v0.23.1 (2017-10-02) +======================================= + +Changes: + +- Make \'affinity\' package optional, as it is not supported on some platforms + +Changes in synapse v0.23.0 (2017-10-02) +======================================= + +No changes since v0.23.0-rc2 + +Changes in synapse v0.23.0-rc2 (2017-09-26) +=========================================== + +Bug fixes: + +- Fix regression in performance of syncs (PR #2470) + +Changes in synapse v0.23.0-rc1 (2017-09-25) +=========================================== + +Features: + +- Add a frontend proxy worker (PR #2344) +- Add support for event\_id\_only push format (PR #2450) +- Add a PoC for filtering spammy events (PR #2456) +- Add a config option to block all room invites (PR #2457) + +Changes: + +- Use bcrypt module instead of py-bcrypt (PR #2288) Thanks to @kyrias! +- Improve performance of generating push notifications (PR #2343, #2357, #2365, #2366, #2371) +- Improve DB performance for device list handling in sync (PR #2362) +- Include a sample prometheus config (PR #2416) +- Document known to work postgres version (PR #2433) Thanks to @ptman! + +Bug fixes: + +- Fix caching error in the push evaluator (PR #2332) +- Fix bug where pusherpool didn\'t start and broke some rooms (PR #2342) +- Fix port script for user directory tables (PR #2375) +- Fix device lists notifications when user rejoins a room (PR #2443, #2449) +- Fix sync to always send down current state events in timeline (PR #2451) +- Fix bug where guest users were incorrectly kicked (PR #2453) +- Fix bug talking to IPv6 only servers using SRV records (PR #2462) + +Changes in synapse v0.22.1 (2017-07-06) +======================================= + +Bug fixes: + +- Fix bug where pusher pool didn\'t start and caused issues when interacting with some rooms (PR #2342) + +Changes in synapse v0.22.0 (2017-07-06) +======================================= + +No changes since v0.22.0-rc2 + +Changes in synapse v0.22.0-rc2 (2017-07-04) +=========================================== + +Changes: + +- Improve performance of storing user IPs (PR #2307, #2308) +- Slightly improve performance of verifying access tokens (PR #2320) +- Slightly improve performance of event persistence (PR #2321) +- Increase default cache factor size from 0.1 to 0.5 (PR #2330) + +Bug fixes: + +- Fix bug with storing registration sessions that caused frequent CPU churn (PR #2319) + +Changes in synapse v0.22.0-rc1 (2017-06-26) +=========================================== + +Features: + +- Add a user directory API (PR #2252, and many more) +- Add shutdown room API to remove room from local server (PR #2291) +- Add API to quarantine media (PR #2292) +- Add new config option to not send event contents to push servers (PR #2301) Thanks to @cjdelisle! + +Changes: + +- Various performance fixes (PR #2177, #2233, #2230, #2238, #2248, #2256, #2274) +- Deduplicate sync filters (PR #2219) Thanks to @krombel! +- Correct a typo in UPGRADE.rst (PR #2231) Thanks to @aaronraimist! +- Add count of one time keys to sync stream (PR #2237) +- Only store event\_auth for state events (PR #2247) +- Store URL cache preview downloads separately (PR #2299) + +Bug fixes: + +- Fix users not getting notifications when AS listened to that user\_id (PR #2216) Thanks to @slipeer! +- Fix users without push set up not getting notifications after joining rooms (PR #2236) +- Fix preview url API to trim long descriptions (PR #2243) +- Fix bug where we used cached but unpersisted state group as prev group, resulting in broken state of restart (PR #2263) +- Fix removing of pushers when using workers (PR #2267) +- Fix CORS headers to allow Authorization header (PR #2285) Thanks to @krombel! + +Changes in synapse v0.21.1 (2017-06-15) +======================================= + +Bug fixes: + +- Fix bug in anonymous usage statistic reporting (PR #2281) + +Changes in synapse v0.21.0 (2017-05-18) +======================================= + +No changes since v0.21.0-rc3 + +Changes in synapse v0.21.0-rc3 (2017-05-17) +=========================================== + +Features: + +- Add per user rate-limiting overrides (PR #2208) +- Add config option to limit maximum number of events requested by `/sync` and `/messages` (PR #2221) Thanks to @psaavedra! + +Changes: + +- Various small performance fixes (PR #2201, #2202, #2224, #2226, #2227, #2228, #2229) +- Update username availability checker API (PR #2209, #2213) +- When purging, don\'t de-delta state groups we\'re about to delete (PR #2214) +- Documentation to check synapse version (PR #2215) Thanks to @hamber-dick! +- Add an index to event\_search to speed up purge history API (PR #2218) + +Bug fixes: + +- Fix API to allow clients to upload one-time-keys with new sigs (PR #2206) + +Changes in synapse v0.21.0-rc2 (2017-05-08) +=========================================== + +Changes: + +- Always mark remotes as up if we receive a signed request from them (PR #2190) + +Bug fixes: + +- Fix bug where users got pushed for rooms they had muted (PR #2200) + +Changes in synapse v0.21.0-rc1 (2017-05-08) +=========================================== + +Features: + +- Add username availability checker API (PR #2183) +- Add read marker API (PR #2120) + +Changes: + +- Enable guest access for the 3pl/3pid APIs (PR #1986) +- Add setting to support TURN for guests (PR #2011) +- Various performance improvements (PR #2075, #2076, #2080, #2083, #2108, #2158, #2176, #2185) +- Make synctl a bit more user friendly (PR #2078, #2127) Thanks @APwhitehat! +- Replace HTTP replication with TCP replication (PR #2082, #2097, #2098, #2099, #2103, #2014, #2016, #2115, #2116, #2117) +- Support authenticated SMTP (PR #2102) Thanks @DanielDent! +- Add a counter metric for successfully-sent transactions (PR #2121) +- Propagate errors sensibly from proxied IS requests (PR #2147) +- Add more granular event send metrics (PR #2178) + +Bug fixes: + +- Fix nuke-room script to work with current schema (PR #1927) Thanks @zuckschwerdt! +- Fix db port script to not assume postgres tables are in the public schema (PR #2024) Thanks @jerrykan! +- Fix getting latest device IP for user with no devices (PR #2118) +- Fix rejection of invites to unreachable servers (PR #2145) +- Fix code for reporting old verify keys in synapse (PR #2156) +- Fix invite state to always include all events (PR #2163) +- Fix bug where synapse would always fetch state for any missing event (PR #2170) +- Fix a leak with timed out HTTP connections (PR #2180) +- Fix bug where we didn\'t time out HTTP requests to ASes (PR #2192) + +Docs: + +- Clarify doc for SQLite to PostgreSQL port (PR #1961) Thanks @benhylau! +- Fix typo in synctl help (PR #2107) Thanks @HarHarLinks! +- `web_client_location` documentation fix (PR #2131) Thanks @matthewjwolff! +- Update README.rst with FreeBSD changes (PR #2132) Thanks @feld! +- Clarify setting up metrics (PR #2149) Thanks @encks! + +Changes in synapse v0.20.0 (2017-04-11) +======================================= + +Bug fixes: + +- Fix joining rooms over federation where not all servers in the room saw the new server had joined (PR #2094) + +Changes in synapse v0.20.0-rc1 (2017-03-30) +=========================================== + +Features: + +- Add delete\_devices API (PR #1993) +- Add phone number registration/login support (PR #1994, #2055) + +Changes: + +- Use JSONSchema for validation of filters. Thanks @pik! (PR #1783) +- Reread log config on SIGHUP (PR #1982) +- Speed up public room list (PR #1989) +- Add helpful texts to logger config options (PR #1990) +- Minor `/sync` performance improvements. (PR #2002, #2013, #2022) +- Add some debug to help diagnose weird federation issue (PR #2035) +- Correctly limit retries for all federation requests (PR #2050, #2061) +- Don\'t lock table when persisting new one time keys (PR #2053) +- Reduce some CPU work on DB threads (PR #2054) +- Cache hosts in room (PR #2060) +- Batch sending of device list pokes (PR #2063) +- Speed up persist event path in certain edge cases (PR #2070) + +Bug fixes: + +- Fix bug where current\_state\_events renamed to current\_state\_ids (PR #1849) +- Fix routing loop when fetching remote media (PR #1992) +- Fix current\_state\_events table to not lie (PR #1996) +- Fix CAS login to handle PartialDownloadError (PR #1997) +- Fix assertion to stop transaction queue getting wedged (PR #2010) +- Fix presence to fallback to last\_active\_ts if it beats the last sync time. Thanks @Half-Shot! (PR #2014) +- Fix bug when federation received a PDU while a room join is in progress (PR #2016) +- Fix resetting state on rejected events (PR #2025) +- Fix installation issues in readme. Thanks @ricco386 (PR #2037) +- Fix caching of remote servers\' signature keys (PR #2042) +- Fix some leaking log context (PR #2048, #2049, #2057, #2058) +- Fix rejection of invites not reaching sync (PR #2056) + +Changes in synapse v0.19.3 (2017-03-20) +======================================= + +No changes since v0.19.3-rc2 + +Changes in synapse v0.19.3-rc2 (2017-03-13) +=========================================== + +Bug fixes: + +- Fix bug in handling of incoming device list updates over federation. + +Changes in synapse v0.19.3-rc1 (2017-03-08) +=========================================== + +Features: + +- Add some administration functionalities. Thanks to morteza-araby! (PR #1784) + +Changes: + +- Reduce database table sizes (PR #1873, #1916, #1923, #1963) +- Update contrib/ to not use syutil. Thanks to andrewshadura! (PR #1907) +- Don\'t fetch current state when sending an event in common case (PR #1955) + +Bug fixes: + +- Fix synapse\_port\_db failure. Thanks to Pneumaticat! (PR #1904) +- Fix caching to not cache error responses (PR #1913) +- Fix APIs to make kick & ban reasons work (PR #1917) +- Fix bugs in the /keys/changes api (PR #1921) +- Fix bug where users couldn\'t forget rooms they were banned from (PR #1922) +- Fix issue with long language values in pushers API (PR #1925) +- Fix a race in transaction queue (PR #1930) +- Fix dynamic thumbnailing to preserve aspect ratio. Thanks to jkolo! (PR #1945) +- Fix device list update to not constantly resync (PR #1964) +- Fix potential for huge memory usage when getting device that have changed (PR #1969) + +Changes in synapse v0.19.2 (2017-02-20) +======================================= + +- Fix bug with event visibility check in /context/ API. Thanks to Tokodomo for pointing it out! (PR #1929) + +Changes in synapse v0.19.1 (2017-02-09) +======================================= + +- Fix bug where state was incorrectly reset in a room when synapse received an event over federation that did not pass auth checks (PR #1892) + +Changes in synapse v0.19.0 (2017-02-04) +======================================= + +No changes since RC 4. + +Changes in synapse v0.19.0-rc4 (2017-02-02) +=========================================== + +- Bump cache sizes for common membership queries (PR #1879) + +Changes in synapse v0.19.0-rc3 (2017-02-02) +=========================================== + +- Fix email push in pusher worker (PR #1875) +- Make presence.get\_new\_events a bit faster (PR #1876) +- Make /keys/changes a bit more performant (PR #1877) + +Changes in synapse v0.19.0-rc2 (2017-02-02) +=========================================== + +- Include newly joined users in /keys/changes API (PR #1872) + +Changes in synapse v0.19.0-rc1 (2017-02-02) +=========================================== + +Features: + +- Add support for specifying multiple bind addresses (PR #1709, #1712, #1795, #1835). Thanks to @kyrias! +- Add /account/3pid/delete endpoint (PR #1714) +- Add config option to configure the Riot URL used in notification emails (PR #1811). Thanks to @aperezdc! +- Add username and password config options for turn server (PR #1832). Thanks to @xsteadfastx! +- Implement device lists updates over federation (PR #1857, #1861, #1864) +- Implement /keys/changes (PR #1869, #1872) + +Changes: + +- Improve IPv6 support (PR #1696). Thanks to @kyrias and @glyph! +- Log which files we saved attachments to in the media\_repository (PR #1791) +- Linearize updates to membership via PUT /state/ to better handle multiple joins (PR #1787) +- Limit number of entries to prefill from cache on startup (PR #1792) +- Remove full\_twisted\_stacktraces option (PR #1802) +- Measure size of some caches by sum of the size of cached values (PR #1815) +- Measure metrics of string\_cache (PR #1821) +- Reduce logging verbosity (PR #1822, #1823, #1824) +- Don\'t clobber a displayname or avatar\_url if provided by an m.room.member event (PR #1852) +- Better handle 401/404 response for federation /send/ (PR #1866, #1871) + +Fixes: + +- Fix ability to change password to a non-ascii one (PR #1711) +- Fix push getting stuck due to looking at the wrong view of state (PR #1820) +- Fix email address comparison to be case insensitive (PR #1827) +- Fix occasional inconsistencies of room membership (PR #1836, #1840) + +Performance: + +- Don\'t block messages sending on bumping presence (PR #1789) +- Change device\_inbox stream index to include user (PR #1793) +- Optimise state resolution (PR #1818) +- Use DB cache of joined users for presence (PR #1862) +- Add an index to make membership queries faster (PR #1867) + +Changes in synapse v0.18.7 (2017-01-09) +======================================= + +No changes from v0.18.7-rc2 + +Changes in synapse v0.18.7-rc2 (2017-01-07) +=========================================== + +Bug fixes: + +- Fix error in rc1\'s discarding invalid inbound traffic logic that was incorrectly discarding missing events + +Changes in synapse v0.18.7-rc1 (2017-01-06) +=========================================== + +Bug fixes: + +- Fix error in \#PR 1764 to actually fix the nightmare \#1753 bug. +- Improve deadlock logging further +- Discard inbound federation traffic from invalid domains, to immunise against \#1753 + +Changes in synapse v0.18.6 (2017-01-06) +======================================= + +Bug fixes: + +- Fix bug when checking if a guest user is allowed to join a room (PR #1772) Thanks to Patrik Oldsberg for diagnosing and the fix! + +Changes in synapse v0.18.6-rc3 (2017-01-05) +=========================================== + +Bug fixes: + +- Fix bug where we failed to send ban events to the banned server (PR #1758) +- Fix bug where we sent event that didn\'t originate on this server to other servers (PR #1764) +- Fix bug where processing an event from a remote server took a long time because we were making long HTTP requests (PR #1765, PR #1744) + +Changes: + +- Improve logging for debugging deadlocks (PR #1766, PR #1767) + +Changes in synapse v0.18.6-rc2 (2016-12-30) +=========================================== + +Bug fixes: + +- Fix memory leak in twisted by initialising logging correctly (PR #1731) +- Fix bug where fetching missing events took an unacceptable amount of time in large rooms (PR #1734) + +Changes in synapse v0.18.6-rc1 (2016-12-29) +=========================================== + +Bug fixes: + +- Make sure that outbound connections are closed (PR #1725) + +Changes in synapse v0.18.5 (2016-12-16) +======================================= + +Bug fixes: + +- Fix federation /backfill returning events it shouldn\'t (PR #1700) +- Fix crash in url preview (PR #1701) + +Changes in synapse v0.18.5-rc3 (2016-12-13) +=========================================== + +Features: + +- Add support for E2E for guests (PR #1653) +- Add new API appservice specific public room list (PR #1676) +- Add new room membership APIs (PR #1680) + +Changes: + +- Enable guest access for private rooms by default (PR #653) +- Limit the number of events that can be created on a given room concurrently (PR #1620) +- Log the args that we have on UI auth completion (PR #1649) +- Stop generating refresh\_tokens (PR #1654) +- Stop putting a time caveat on access tokens (PR #1656) +- Remove unspecced GET endpoints for e2e keys (PR #1694) + +Bug fixes: + +- Fix handling of 500 and 429\'s over federation (PR #1650) +- Fix Content-Type header parsing (PR #1660) +- Fix error when previewing sites that include unicode, thanks to kyrias (PR #1664) +- Fix some cases where we drop read receipts (PR #1678) +- Fix bug where calls to `/sync` didn\'t correctly timeout (PR #1683) +- Fix bug where E2E key query would fail if a single remote host failed (PR #1686) + +Changes in synapse v0.18.5-rc2 (2016-11-24) +=========================================== + +Bug fixes: + +- Don\'t send old events over federation, fixes bug in -rc1. + +Changes in synapse v0.18.5-rc1 (2016-11-24) +=========================================== + +Features: + +- Implement \"event\_fields\" in filters (PR #1638) + +Changes: + +- Use external ldap auth pacakge (PR #1628) +- Split out federation transaction sending to a worker (PR #1635) +- Fail with a coherent error message if /sync?filter= is invalid (PR #1636) +- More efficient notif count queries (PR #1644) + +Changes in synapse v0.18.4 (2016-11-22) +======================================= + +Bug fixes: + +- Add workaround for buggy clients that the fail to register (PR #1632) + +Changes in synapse v0.18.4-rc1 (2016-11-14) +=========================================== + +Changes: + +- Various database efficiency improvements (PR #1188, #1192) +- Update default config to blacklist more internal IPs, thanks to Euan Kemp (PR #1198) +- Allow specifying duration in minutes in config, thanks to Daniel Dent (PR #1625) + +Bug fixes: + +- Fix media repo to set CORs headers on responses (PR #1190) +- Fix registration to not error on non-ascii passwords (PR #1191) +- Fix create event code to limit the number of prev\_events (PR #1615) +- Fix bug in transaction ID deduplication (PR #1624) + +Changes in synapse v0.18.3 (2016-11-08) +======================================= + +SECURITY UPDATE + +Explicitly require authentication when using LDAP3. This is the default on versions of `ldap3` above 1.0, but some distributions will package an older version. + +If you are using LDAP3 login and have a version of `ldap3` older than 1.0 it is **CRITICAL to updgrade**. + +Changes in synapse v0.18.2 (2016-11-01) +======================================= + +No changes since v0.18.2-rc5 + +Changes in synapse v0.18.2-rc5 (2016-10-28) +=========================================== + +Bug fixes: + +- Fix prometheus process metrics in worker processes (PR #1184) + +Changes in synapse v0.18.2-rc4 (2016-10-27) +=========================================== + +Bug fixes: + +- Fix `user_threepids` schema delta, which in some instances prevented startup after upgrade (PR #1183) + +Changes in synapse v0.18.2-rc3 (2016-10-27) +=========================================== + +Changes: + +- Allow clients to supply access tokens as headers (PR #1098) +- Clarify error codes for GET /filter/, thanks to Alexander Maznev (PR #1164) +- Make password reset email field case insensitive (PR #1170) +- Reduce redundant database work in email pusher (PR #1174) +- Allow configurable rate limiting per AS (PR #1175) +- Check whether to ratelimit sooner to avoid work (PR #1176) +- Standardise prometheus metrics (PR #1177) + +Bug fixes: + +- Fix incredibly slow back pagination query (PR #1178) +- Fix infinite typing bug (PR #1179) + +Changes in synapse v0.18.2-rc2 (2016-10-25) +=========================================== + +(This release did not include the changes advertised and was identical to RC1) + +Changes in synapse v0.18.2-rc1 (2016-10-17) +=========================================== + +Changes: + +- Remove redundant event\_auth index (PR #1113) +- Reduce DB hits for replication (PR #1141) +- Implement pluggable password auth (PR #1155) +- Remove rate limiting from app service senders and fix get\_or\_create\_user requester, thanks to Patrik Oldsberg (PR #1157) +- window.postmessage for Interactive Auth fallback (PR #1159) +- Use sys.executable instead of hardcoded python, thanks to Pedro Larroy (PR #1162) +- Add config option for adding additional TLS fingerprints (PR #1167) +- User-interactive auth on delete device (PR #1168) + +Bug fixes: + +- Fix not being allowed to set your own state\_key, thanks to Patrik Oldsberg (PR #1150) +- Fix interactive auth to return 401 from for incorrect password (PR #1160, #1166) +- Fix email push notifs being dropped (PR #1169) + +Changes in synapse v0.18.1 (2016-10-05) +======================================= + +No changes since v0.18.1-rc1 + +Changes in synapse v0.18.1-rc1 (2016-09-30) +=========================================== + +Features: + +- Add total\_room\_count\_estimate to `/publicRooms` (PR #1133) + +Changes: + +- Time out typing over federation (PR #1140) +- Restructure LDAP authentication (PR #1153) + +Bug fixes: + +- Fix 3pid invites when server is already in the room (PR #1136) +- Fix upgrading with SQLite taking lots of CPU for a few days after upgrade (PR #1144) +- Fix upgrading from very old database versions (PR #1145) +- Fix port script to work with recently added tables (PR #1146) + +Changes in synapse v0.18.0 (2016-09-19) +======================================= + +The release includes major changes to the state storage database schemas, which significantly reduce database size. Synapse will attempt to upgrade the current data in the background. Servers with large SQLite database may experience degradation of performance while this upgrade is in progress, therefore you may want to consider migrating to using Postgres before upgrading very large SQLite databases + +Changes: + +- Make public room search case insensitive (PR #1127) + +Bug fixes: + +- Fix and clean up publicRooms pagination (PR #1129) + +Changes in synapse v0.18.0-rc1 (2016-09-16) +=========================================== + +Features: + +- Add `only=highlight` on `/notifications` (PR #1081) +- Add server param to /publicRooms (PR #1082) +- Allow clients to ask for the whole of a single state event (PR #1094) +- Add is\_direct param to /createRoom (PR #1108) +- Add pagination support to publicRooms (PR #1121) +- Add very basic filter API to /publicRooms (PR #1126) +- Add basic direct to device messaging support for E2E (PR #1074, #1084, #1104, #1111) + +Changes: + +- Move to storing state\_groups\_state as deltas, greatly reducing DB size (PR #1065) +- Reduce amount of state pulled out of the DB during common requests (PR #1069) +- Allow PDF to be rendered from media repo (PR #1071) +- Reindex state\_groups\_state after pruning (PR #1085) +- Clobber EDUs in send queue (PR #1095) +- Conform better to the CAS protocol specification (PR #1100) +- Limit how often we ask for keys from dead servers (PR #1114) + +Bug fixes: + +- Fix /notifications API when used with `from` param (PR #1080) +- Fix backfill when cannot find an event. (PR #1107) + +Changes in synapse v0.17.3 (2016-09-09) +======================================= + +This release fixes a major bug that stopped servers from handling rooms with over 1000 members. + +Changes in synapse v0.17.2 (2016-09-08) +======================================= + +This release contains security bug fixes. Please upgrade. + +No changes since v0.17.2-rc1 + +Changes in synapse v0.17.2-rc1 (2016-09-05) +=========================================== + +Features: + +- Start adding store-and-forward direct-to-device messaging (PR #1046, #1050, #1062, #1066) + +Changes: + +- Avoid pulling the full state of a room out so often (PR #1047, #1049, #1063, #1068) +- Don\'t notify for online to online presence transitions. (PR #1054) +- Occasionally persist unpersisted presence updates (PR #1055) +- Allow application services to have an optional \'url\' (PR #1056) +- Clean up old sent transactions from DB (PR #1059) + +Bug fixes: + +- Fix None check in backfill (PR #1043) +- Fix membership changes to be idempotent (PR #1067) +- Fix bug in get\_pdu where it would sometimes return events with incorrect signature + +Changes in synapse v0.17.1 (2016-08-24) +======================================= + +Changes: + +- Delete old received\_transactions rows (PR #1038) +- Pass through user-supplied content in /join/\$room\_id (PR #1039) + +Bug fixes: + +- Fix bug with backfill (PR #1040) + +Changes in synapse v0.17.1-rc1 (2016-08-22) +=========================================== + +Features: + +- Add notification API (PR #1028) + +Changes: + +- Don\'t print stack traces when failing to get remote keys (PR #996) +- Various federation /event/ perf improvements (PR #998) +- Only process one local membership event per room at a time (PR #1005) +- Move default display name push rule (PR #1011, #1023) +- Fix up preview URL API. Add tests. (PR #1015) +- Set `Content-Security-Policy` on media repo (PR #1021) +- Make notify\_interested\_services faster (PR #1022) +- Add usage stats to prometheus monitoring (PR #1037) + +Bug fixes: + +- Fix token login (PR #993) +- Fix CAS login (PR #994, #995) +- Fix /sync to not clobber status\_msg (PR #997) +- Fix redacted state events to include prev\_content (PR #1003) +- Fix some bugs in the auth/ldap handler (PR #1007) +- Fix backfill request to limit URI length, so that remotes don\'t reject the requests due to path length limits (PR #1012) +- Fix AS push code to not send duplicate events (PR #1025) + +Changes in synapse v0.17.0 (2016-08-08) +======================================= + +This release contains significant security bug fixes regarding authenticating events received over federation. PLEASE UPGRADE. + +This release changes the LDAP configuration format in a backwards incompatible way, see PR #843 for details. + +Changes: + +- Add federation /version API (PR #990) +- Make psutil dependency optional (PR #992) + +Bug fixes: + +- Fix URL preview API to exclude HTML comments in description (PR #988) +- Fix error handling of remote joins (PR #991) + +Changes in synapse v0.17.0-rc4 (2016-08-05) +=========================================== + +Changes: + +- Change the way we summarize URLs when previewing (PR #973) +- Add new `/state_ids/` federation API (PR #979) +- Speed up processing of `/state/` response (PR #986) + +Bug fixes: + +- Fix event persistence when event has already been partially persisted (PR #975, #983, #985) +- Fix port script to also copy across backfilled events (PR #982) + +Changes in synapse v0.17.0-rc3 (2016-08-02) +=========================================== + +Changes: + +- Forbid non-ASes from registering users whose names begin with \'\_\' (PR #958) +- Add some basic admin API docs (PR #963) + +Bug fixes: + +- Send the correct host header when fetching keys (PR #941) +- Fix joining a room that has missing auth events (PR #964) +- Fix various push bugs (PR #966, #970) +- Fix adding emails on registration (PR #968) + +Changes in synapse v0.17.0-rc2 (2016-08-02) +=========================================== + +(This release did not include the changes advertised and was identical to RC1) + +Changes in synapse v0.17.0-rc1 (2016-07-28) +=========================================== + +This release changes the LDAP configuration format in a backwards incompatible way, see PR #843 for details. + +Features: + +- Add purge\_media\_cache admin API (PR #902) +- Add deactivate account admin API (PR #903) +- Add optional pepper to password hashing (PR #907, #910 by KentShikama) +- Add an admin option to shared secret registration (breaks backwards compat) (PR #909) +- Add purge local room history API (PR #911, #923, #924) +- Add requestToken endpoints (PR #915) +- Add an /account/deactivate endpoint (PR #921) +- Add filter param to /messages. Add \'contains\_url\' to filter. (PR #922) +- Add device\_id support to /login (PR #929) +- Add device\_id support to /v2/register flow. (PR #937, #942) +- Add GET /devices endpoint (PR #939, #944) +- Add GET /device/{deviceId} (PR #943) +- Add update and delete APIs for devices (PR #949) + +Changes: + +- Rewrite LDAP Authentication against ldap3 (PR #843 by mweinelt) +- Linearize some federation endpoints based on (origin, room\_id) (PR #879) +- Remove the legacy v0 content upload API. (PR #888) +- Use similar naming we use in email notifs for push (PR #894) +- Optionally include password hash in createUser endpoint (PR #905 by KentShikama) +- Use a query that postgresql optimises better for get\_events\_around (PR #906) +- Fall back to \'username\' if \'user\' is not given for appservice registration. (PR #927 by Half-Shot) +- Add metrics for psutil derived memory usage (PR #936) +- Record device\_id in client\_ips (PR #938) +- Send the correct host header when fetching keys (PR #941) +- Log the hostname the reCAPTCHA was completed on (PR #946) +- Make the device id on e2e key upload optional (PR #956) +- Add r0.2.0 to the \"supported versions\" list (PR #960) +- Don\'t include name of room for invites in push (PR #961) + +Bug fixes: + +- Fix substitution failure in mail template (PR #887) +- Put most recent 20 messages in email notif (PR #892) +- Ensure that the guest user is in the database when upgrading accounts (PR #914) +- Fix various edge cases in auth handling (PR #919) +- Fix 500 ISE when sending alias event without a state\_key (PR #925) +- Fix bug where we stored rejections in the state\_group, persist all rejections (PR #948) +- Fix lack of check of if the user is banned when handling 3pid invites (PR #952) +- Fix a couple of bugs in the transaction and keyring code (PR #954, #955) + +Changes in synapse v0.16.1-r1 (2016-07-08) +========================================== + +THIS IS A CRITICAL SECURITY UPDATE. + +This fixes a bug which allowed users\' accounts to be accessed by unauthorised users. + +Changes in synapse v0.16.1 (2016-06-20) +======================================= + +Bug fixes: + +- Fix assorted bugs in `/preview_url` (PR #872) +- Fix TypeError when setting unicode passwords (PR #873) + +Performance improvements: + +- Turn `use_frozen_events` off by default (PR #877) +- Disable responding with canonical json for federation (PR #878) + +Changes in synapse v0.16.1-rc1 (2016-06-15) +=========================================== + +Features: None + +Changes: + +- Log requester for `/publicRoom` endpoints when possible (PR #856) +- 502 on `/thumbnail` when can\'t connect to remote server (PR #862) +- Linearize fetching of gaps on incoming events (PR #871) + +Bugs fixes: + +- Fix bug where rooms where marked as published by default (PR #857) +- Fix bug where joining room with an event with invalid sender (PR #868) +- Fix bug where backfilled events were sent down sync streams (PR #869) +- Fix bug where outgoing connections could wedge indefinitely, causing push notifications to be unreliable (PR #870) + +Performance improvements: + +- Improve `/publicRooms` performance(PR #859) + +Changes in synapse v0.16.0 (2016-06-09) +======================================= + +NB: As of v0.14 all AS config files must have an ID field. + +Bug fixes: + +- Don\'t make rooms published by default (PR #857) + +Changes in synapse v0.16.0-rc2 (2016-06-08) +=========================================== + +Features: + +- Add configuration option for tuning GC via `gc.set_threshold` (PR #849) + +Changes: + +- Record metrics about GC (PR #771, #847, #852) +- Add metric counter for number of persisted events (PR #841) + +Bug fixes: + +- Fix \'From\' header in email notifications (PR #843) +- Fix presence where timeouts were not being fired for the first 8h after restarts (PR #842) +- Fix bug where synapse sent malformed transactions to AS\'s when retrying transactions (Commits 310197b, 8437906) + +Performance improvements: + +- Remove event fetching from DB threads (PR #835) +- Change the way we cache events (PR #836) +- Add events to cache when we persist them (PR #840) + +Changes in synapse v0.16.0-rc1 (2016-06-03) +=========================================== + +Version 0.15 was not released. See v0.15.0-rc1 below for additional changes. + +Features: + +- Add email notifications for missed messages (PR #759, #786, #799, #810, #815, #821) +- Add a `url_preview_ip_range_whitelist` config param (PR #760) +- Add /report endpoint (PR #762) +- Add basic ignore user API (PR #763) +- Add an openidish mechanism for proving that you own a given user\_id (PR #765) +- Allow clients to specify a server\_name to avoid \'No known servers\' (PR #794) +- Add secondary\_directory\_servers option to fetch room list from other servers (PR #808, #813) + +Changes: + +- Report per request metrics for all of the things using request\_handler (PR #756) +- Correctly handle `NULL` password hashes from the database (PR #775) +- Allow receipts for events we haven\'t seen in the db (PR #784) +- Make synctl read a cache factor from config file (PR #785) +- Increment badge count per missed convo, not per msg (PR #793) +- Special case m.room.third\_party\_invite event auth to match invites (PR #814) + +Bug fixes: + +- Fix typo in event\_auth servlet path (PR #757) +- Fix password reset (PR #758) + +Performance improvements: + +- Reduce database inserts when sending transactions (PR #767) +- Queue events by room for persistence (PR #768) +- Add cache to `get_user_by_id` (PR #772) +- Add and use `get_domain_from_id` (PR #773) +- Use tree cache for `get_linearized_receipts_for_room` (PR #779) +- Remove unused indices (PR #782) +- Add caches to `bulk_get_push_rules*` (PR #804) +- Cache `get_event_reference_hashes` (PR #806) +- Add `get_users_with_read_receipts_in_room` cache (PR #809) +- Use state to calculate `get_users_in_room` (PR #811) +- Load push rules in storage layer so that they get cached (PR #825) +- Make `get_joined_hosts_for_room` use get\_users\_in\_room (PR #828) +- Poke notifier on next reactor tick (PR #829) +- Change CacheMetrics to be quicker (PR #830) + +Changes in synapse v0.15.0-rc1 (2016-04-26) +=========================================== + +Features: + +- Add login support for Javascript Web Tokens, thanks to Niklas Riekenbrauck (PR #671,\#687) +- Add URL previewing support (PR #688) +- Add login support for LDAP, thanks to Christoph Witzany (PR #701) +- Add GET endpoint for pushers (PR #716) + +Changes: + +- Never notify for member events (PR #667) +- Deduplicate identical `/sync` requests (PR #668) +- Require user to have left room to forget room (PR #673) +- Use DNS cache if within TTL (PR #677) +- Let users see their own leave events (PR #699) +- Deduplicate membership changes (PR #700) +- Increase performance of pusher code (PR #705) +- Respond with error status 504 if failed to talk to remote server (PR #731) +- Increase search performance on postgres (PR #745) + +Bug fixes: + +- Fix bug where disabling all notifications still resulted in push (PR #678) +- Fix bug where users couldn\'t reject remote invites if remote refused (PR #691) +- Fix bug where synapse attempted to backfill from itself (PR #693) +- Fix bug where profile information was not correctly added when joining remote rooms (PR #703) +- Fix bug where register API required incorrect key name for AS registration (PR #727) + +Changes in synapse v0.14.0 (2016-03-30) +======================================= + +No changes from v0.14.0-rc2 + +Changes in synapse v0.14.0-rc2 (2016-03-23) +=========================================== + +Features: + +- Add published room list API (PR #657) + +Changes: + +- Change various caches to consume less memory (PR #656, #658, #660, #662, #663, #665) +- Allow rooms to be published without requiring an alias (PR #664) +- Intern common strings in caches to reduce memory footprint (\#666) + +Bug fixes: + +- Fix reject invites over federation (PR #646) +- Fix bug where registration was not idempotent (PR #649) +- Update aliases event after deleting aliases (PR #652) +- Fix unread notification count, which was sometimes wrong (PR #661) + +Changes in synapse v0.14.0-rc1 (2016-03-14) +=========================================== + +Features: + +- Add event\_id to response to state event PUT (PR #581) +- Allow guest users access to messages in rooms they have joined (PR #587) +- Add config for what state is included in a room invite (PR #598) +- Send the inviter\'s member event in room invite state (PR #607) +- Add error codes for malformed/bad JSON in /login (PR #608) +- Add support for changing the actions for default rules (PR #609) +- Add environment variable SYNAPSE\_CACHE\_FACTOR, default it to 0.1 (PR #612) +- Add ability for alias creators to delete aliases (PR #614) +- Add profile information to invites (PR #624) + +Changes: + +- Enforce user\_id exclusivity for AS registrations (PR #572) +- Make adding push rules idempotent (PR #587) +- Improve presence performance (PR #582, #586) +- Change presence semantics for `last_active_ago` (PR #582, #586) +- Don\'t allow `m.room.create` to be changed (PR #596) +- Add 800x600 to default list of valid thumbnail sizes (PR #616) +- Always include kicks and bans in full /sync (PR #625) +- Send history visibility on boundary changes (PR #626) +- Register endpoint now returns a refresh\_token (PR #637) + +Bug fixes: + +- Fix bug where we returned incorrect state in /sync (PR #573) +- Always return a JSON object from push rule API (PR #606) +- Fix bug where registering without a user id sometimes failed (PR #610) +- Report size of ExpiringCache in cache size metrics (PR #611) +- Fix rejection of invites to empty rooms (PR #615) +- Fix usage of `bcrypt` to not use `checkpw` (PR #619) +- Pin `pysaml2` dependency (PR #634) +- Fix bug in `/sync` where timeline order was incorrect for backfilled events (PR #635) + +Changes in synapse v0.13.3 (2016-02-11) +======================================= + +- Fix bug where `/sync` would occasionally return events in the wrong room. + +Changes in synapse v0.13.2 (2016-02-11) +======================================= + +- Fix bug where `/events` would fail to skip some events if there had been more events than the limit specified since the last request (PR #570) + +Changes in synapse v0.13.1 (2016-02-10) +======================================= + +- Bump matrix-angular-sdk (matrix web console) dependency to 0.6.8 to pull in the fix for SYWEB-361 so that the default client can display HTML messages again(!) + +Changes in synapse v0.13.0 (2016-02-10) +======================================= + +This version includes an upgrade of the schema, specifically adding an index to the `events` table. This may cause synapse to pause for several minutes the first time it is started after the upgrade. + +Changes: + +- Improve general performance (PR #540, #543. \#544, #54, #549, #567) +- Change guest user ids to be incrementing integers (PR #550) +- Improve performance of public room list API (PR #552) +- Change profile API to omit keys rather than return null (PR #557) +- Add `/media/r0` endpoint prefix, which is equivalent to `/media/v1/` (PR #595) + +Bug fixes: + +- Fix bug with upgrading guest accounts where it would fail if you opened the registration email on a different device (PR #547) +- Fix bug where unread count could be wrong (PR #568) + +Changes in synapse v0.12.1-rc1 (2016-01-29) +=========================================== + +Features: + +- Add unread notification counts in `/sync` (PR #456) +- Add support for inviting 3pids in `/createRoom` (PR #460) +- Add ability for guest accounts to upgrade (PR #462) +- Add `/versions` API (PR #468) +- Add `event` to `/context` API (PR #492) +- Add specific error code for invalid user names in `/register` (PR #499) +- Add support for push badge counts (PR #507) +- Add support for non-guest users to peek in rooms using `/events` (PR #510) + +Changes: + +- Change `/sync` so that guest users only get rooms they\'ve joined (PR #469) +- Change to require unbanning before other membership changes (PR #501) +- Change default push rules to notify for all messages (PR #486) +- Change default push rules to not notify on membership changes (PR #514) +- Change default push rules in one to one rooms to only notify for events that are messages (PR #529) +- Change `/sync` to reject requests with a `from` query param (PR #512) +- Change server manhole to use SSH rather than telnet (PR #473) +- Change server to require AS users to be registered before use (PR #487) +- Change server not to start when ASes are invalidly configured (PR #494) +- Change server to require ID and `as_token` to be unique for AS\'s (PR #496) +- Change maximum pagination limit to 1000 (PR #497) + +Bug fixes: + +- Fix bug where `/sync` didn\'t return when something under the leave key changed (PR #461) +- Fix bug where we returned smaller rather than larger than requested thumbnails when `method=crop` (PR #464) +- Fix thumbnails API to only return cropped thumbnails when asking for a cropped thumbnail (PR #475) +- Fix bug where we occasionally still logged access tokens (PR #477) +- Fix bug where `/events` would always return immediately for guest users (PR #480) +- Fix bug where `/sync` unexpectedly returned old left rooms (PR #481) +- Fix enabling and disabling push rules (PR #498) +- Fix bug where `/register` returned 500 when given unicode username (PR #513) + +Changes in synapse v0.12.0 (2016-01-04) +======================================= + +- Expose `/login` under `r0` (PR #459) + +Changes in synapse v0.12.0-rc3 (2015-12-23) +=========================================== + +- Allow guest accounts access to `/sync` (PR #455) +- Allow filters to include/exclude rooms at the room level rather than just from the components of the sync for each room. (PR #454) +- Include urls for room avatars in the response to `/publicRooms` (PR #453) +- Don\'t set a identicon as the avatar for a user when they register (PR #450) +- Add a `display_name` to third-party invites (PR #449) +- Send more information to the identity server for third-party invites so that it can send richer messages to the invitee (PR #446) +- Cache the responses to `/initialSync` for 5 minutes. If a client retries a request to `/initialSync` before the a response was computed to the first request then the same response is used for both requests (PR #457) +- Fix a bug where synapse would always request the signing keys of remote servers even when the key was cached locally (PR #452) +- Fix 500 when pagination search results (PR #447) +- Fix a bug where synapse was leaking raw email address in third-party invites (PR #448) + +Changes in synapse v0.12.0-rc2 (2015-12-14) +=========================================== + +- Add caches for whether rooms have been forgotten by a user (PR #434) +- Remove instructions to use `--process-dependency-link` since all of the dependencies of synapse are on PyPI (PR #436) +- Parallelise the processing of `/sync` requests (PR #437) +- Fix race updating presence in `/events` (PR #444) +- Fix bug back-populating search results (PR #441) +- Fix bug calculating state in `/sync` requests (PR #442) + +Changes in synapse v0.12.0-rc1 (2015-12-10) +=========================================== + +- Host the client APIs released as r0 by on paths prefixed by `/_matrix/client/r0`. (PR #430, PR #415, PR #400) +- Updates the client APIs to match r0 of the matrix specification. + - All APIs return events in the new event format, old APIs also include the fields needed to parse the event using the old format for compatibility. (PR #402) + - Search results are now given as a JSON array rather than a JSON object (PR #405) + - Miscellaneous changes to search (PR #403, PR #406, PR #412) + - Filter JSON objects may now be passed as query parameters to `/sync` (PR #431) + - Fix implementation of `/admin/whois` (PR #418) + - Only include the rooms that user has left in `/sync` if the client requests them in the filter (PR #423) + - Don\'t push for `m.room.message` by default (PR #411) + - Add API for setting per account user data (PR #392) + - Allow users to forget rooms (PR #385) +- Performance improvements and monitoring: + - Add per-request counters for CPU time spent on the main python thread. (PR #421, PR #420) + - Add per-request counters for time spent in the database (PR #429) + - Make state updates in the C+S API idempotent (PR #416) + - Only fire `user_joined_room` if the user has actually joined. (PR #410) + - Reuse a single http client, rather than creating new ones (PR #413) +- Fixed a bug upgrading from older versions of synapse on postgresql (PR #417) + +Changes in synapse v0.11.1 (2015-11-20) +======================================= + +- Add extra options to search API (PR #394) +- Fix bug where we did not correctly cap federation retry timers. This meant it could take several hours for servers to start talking to ressurected servers, even when they were receiving traffic from them (PR #393) +- Don\'t advertise login token flow unless CAS is enabled. This caused issues where some clients would always use the fallback API if they did not recognize all login flows (PR #391) +- Change /v2 sync API to rename `private_user_data` to `account_data` (PR #386) +- Change /v2 sync API to remove the `event_map` and rename keys in `rooms` object (PR #389) + +Changes in synapse v0.11.0-r2 (2015-11-19) +========================================== + +- Fix bug in database port script (PR #387) + +Changes in synapse v0.11.0-r1 (2015-11-18) +========================================== + +- Retry and fail federation requests more aggressively for requests that block client side requests (PR #384) + +Changes in synapse v0.11.0 (2015-11-17) +======================================= + +- Change CAS login API (PR #349) + +Changes in synapse v0.11.0-rc2 (2015-11-13) +=========================================== + +- Various changes to /sync API response format (PR #373) +- Fix regression when setting display name in newly joined room over federation (PR #368) +- Fix problem where /search was slow when using SQLite (PR #366) + +Changes in synapse v0.11.0-rc1 (2015-11-11) +=========================================== + +- Add Search API (PR #307, #324, #327, #336, #350, #359) +- Add \'archived\' state to v2 /sync API (PR #316) +- Add ability to reject invites (PR #317) +- Add config option to disable password login (PR #322) +- Add the login fallback API (PR #330) +- Add room context API (PR #334) +- Add room tagging support (PR #335) +- Update v2 /sync API to match spec (PR #305, #316, #321, #332, #337, #341) +- Change retry schedule for application services (PR #320) +- Change retry schedule for remote servers (PR #340) +- Fix bug where we hosted static content in the incorrect place (PR #329) +- Fix bug where we didn\'t increment retry interval for remote servers (PR #343) + +Changes in synapse v0.10.1-rc1 (2015-10-15) +=========================================== + +- Add support for CAS, thanks to Steven Hammerton (PR #295, #296) +- Add support for using macaroons for `access_token` (PR #256, #229) +- Add support for `m.room.canonical_alias` (PR #287) +- Add support for viewing the history of rooms that they have left. (PR #276, #294) +- Add support for refresh tokens (PR #240) +- Add flag on creation which disables federation of the room (PR #279) +- Add some room state to invites. (PR #275) +- Atomically persist events when joining a room over federation (PR #283) +- Change default history visibility for private rooms (PR #271) +- Allow users to redact their own sent events (PR #262) +- Use tox for tests (PR #247) +- Split up syutil into separate libraries (PR #243) + +Changes in synapse v0.10.0-r2 (2015-09-16) +========================================== + +- Fix bug where we always fetched remote server signing keys instead of using ones in our cache. +- Fix adding threepids to an existing account. +- Fix bug with invinting over federation where remote server was already in the room. (PR #281, SYN-392) + +Changes in synapse v0.10.0-r1 (2015-09-08) +========================================== + +- Fix bug with python packaging + +Changes in synapse v0.10.0 (2015-09-03) +======================================= + +No change from release candidate. + +Changes in synapse v0.10.0-rc6 (2015-09-02) +=========================================== + +- Remove some of the old database upgrade scripts. +- Fix database port script to work with newly created sqlite databases. + +Changes in synapse v0.10.0-rc5 (2015-08-27) +=========================================== + +- Fix bug that broke downloading files with ascii filenames across federation. + +Changes in synapse v0.10.0-rc4 (2015-08-27) +=========================================== + +- Allow UTF-8 filenames for upload. (PR #259) + +Changes in synapse v0.10.0-rc3 (2015-08-25) +=========================================== + +- Add `--keys-directory` config option to specify where files such as certs and signing keys should be stored in, when using `--generate-config` or `--generate-keys`. (PR #250) +- Allow `--config-path` to specify a directory, causing synapse to use all \*.yaml files in the directory as config files. (PR #249) +- Add `web_client_location` config option to specify static files to be hosted by synapse under `/_matrix/client`. (PR #245) +- Add helper utility to synapse to read and parse the config files and extract the value of a given key. For example: + + $ python -m synapse.config read server_name -c homeserver.yaml + localhost + + (PR #246) + +Changes in synapse v0.10.0-rc2 (2015-08-24) +=========================================== + +- Fix bug where we incorrectly populated the `event_forward_extremities` table, resulting in problems joining large remote rooms (e.g. `#matrix:matrix.org`) +- Reduce the number of times we wake up pushers by not listening for presence or typing events, reducing the CPU cost of each pusher. + +Changes in synapse v0.10.0-rc1 (2015-08-21) +=========================================== + +Also see v0.9.4-rc1 changelog, which has been amalgamated into this release. + +General: + +- Upgrade to Twisted 15 (PR #173) +- Add support for serving and fetching encryption keys over federation. (PR #208) +- Add support for logging in with email address (PR #234) +- Add support for new `m.room.canonical_alias` event. (PR #233) +- Change synapse to treat user IDs case insensitively during registration and login. (If two users already exist with case insensitive matching user ids, synapse will continue to require them to specify their user ids exactly.) +- Error if a user tries to register with an email already in use. (PR #211) +- Add extra and improve existing caches (PR #212, #219, #226, #228) +- Batch various storage request (PR #226, #228) +- Fix bug where we didn\'t correctly log the entity that triggered the request if the request came in via an application service (PR #230) +- Fix bug where we needlessly regenerated the full list of rooms an AS is interested in. (PR #232) +- Add support for AS\'s to use v2\_alpha registration API (PR #210) + +Configuration: + +- Add `--generate-keys` that will generate any missing cert and key files in the configuration files. This is equivalent to running `--generate-config` on an existing configuration file. (PR #220) +- `--generate-config` now no longer requires a `--server-name` parameter when used on existing configuration files. (PR #220) +- Add `--print-pidfile` flag that controls the printing of the pid to stdout of the demonised process. (PR #213) + +Media Repository: + +- Fix bug where we picked a lower resolution image than requested. (PR #205) +- Add support for specifying if a the media repository should dynamically thumbnail images or not. (PR #206) + +Metrics: + +- Add statistics from the reactor to the metrics API. (PR #224, #225) + +Demo Homeservers: + +- Fix starting the demo homeservers without rate-limiting enabled. (PR #182) +- Fix enabling registration on demo homeservers (PR #223) + +Changes in synapse v0.9.4-rc1 (2015-07-21) +========================================== + +General: + +- Add basic implementation of receipts. (SPEC-99) +- Add support for configuration presets in room creation API. (PR #203) +- Add auth event that limits the visibility of history for new users. (SPEC-134) +- Add SAML2 login/registration support. (PR #201. Thanks Muthu Subramanian!) +- Add client side key management APIs for end to end encryption. (PR #198) +- Change power level semantics so that you cannot kick, ban or change power levels of users that have equal or greater power level than you. (SYN-192) +- Improve performance by bulk inserting events where possible. (PR #193) +- Improve performance by bulk verifying signatures where possible. (PR #194) + +Configuration: + +- Add support for including TLS certificate chains. + +Media Repository: + +- Add Content-Disposition headers to content repository responses. (SYN-150) + +Changes in synapse v0.9.3 (2015-07-01) +====================================== + +No changes from v0.9.3 Release Candidate 1. + +Changes in synapse v0.9.3-rc1 (2015-06-23) +========================================== + +General: + +- Fix a memory leak in the notifier. (SYN-412) +- Improve performance of room initial sync. (SYN-418) +- General improvements to logging. +- Remove `access_token` query params from `INFO` level logging. + +Configuration: + +- Add support for specifying and configuring multiple listeners. (SYN-389) + +Application services: + +- Fix bug where synapse failed to send user queries to application services. + +Changes in synapse v0.9.2-r2 (2015-06-15) +========================================= + +Fix packaging so that schema delta python files get included in the package. + +Changes in synapse v0.9.2 (2015-06-12) +====================================== + +General: + +- Use ultrajson for json (de)serialisation when a canonical encoding is not required. Ultrajson is significantly faster than simplejson in certain circumstances. +- Use connection pools for outgoing HTTP connections. +- Process thumbnails on separate threads. + +Configuration: + +- Add option, `gzip_responses`, to disable HTTP response compression. + +Federation: + +- Improve resilience of backfill by ensuring we fetch any missing auth events. +- Improve performance of backfill and joining remote rooms by removing unnecessary computations. This included handling events we\'d previously handled as well as attempting to compute the current state for outliers. + +Changes in synapse v0.9.1 (2015-05-26) +====================================== + +General: + +- Add support for backfilling when a client paginates. This allows servers to request history for a room from remote servers when a client tries to paginate history the server does not have - SYN-36 +- Fix bug where you couldn\'t disable non-default pushrules - SYN-378 +- Fix `register_new_user` script - SYN-359 +- Improve performance of fetching events from the database, this improves both initialSync and sending of events. +- Improve performance of event streams, allowing synapse to handle more simultaneous connected clients. + +Federation: + +- Fix bug with existing backfill implementation where it returned the wrong selection of events in some circumstances. +- Improve performance of joining remote rooms. + +Configuration: + +- Add support for changing the bind host of the metrics listener via the `metrics_bind_host` option. + +Changes in synapse v0.9.0-r5 (2015-05-21) +========================================= + +- Add more database caches to reduce amount of work done for each pusher. This radically reduces CPU usage when multiple pushers are set up in the same room. + +Changes in synapse v0.9.0 (2015-05-07) +====================================== + +General: + +- Add support for using a PostgreSQL database instead of SQLite. See [docs/postgres.rst](docs/postgres.rst) for details. +- Add password change and reset APIs. See [Registration](https://github.com/matrix-org/matrix-doc/blob/master/specification/10_client_server_api.rst#registration) in the spec. +- Fix memory leak due to not releasing stale notifiers - SYN-339. +- Fix race in caches that occasionally caused some presence updates to be dropped - SYN-369. +- Check server name has not changed on restart. +- Add a sample systemd unit file and a logger configuration in contrib/systemd. Contributed Ivan Shapovalov. + +Federation: + +- Add key distribution mechanisms for fetching public keys of unavailable remote home servers. See [Retrieving Server Keys](https://github.com/matrix-org/matrix-doc/blob/6f2698/specification/30_server_server_api.rst#retrieving-server-keys) in the spec. + +Configuration: + +- Add support for multiple config files. +- Add support for dictionaries in config files. +- Remove support for specifying config options on the command line, except for: + - `--daemonize` - Daemonize the home server. + - `--manhole` - Turn on the twisted telnet manhole service on the given port. + - `--database-path` - The path to a sqlite database to use. + - `--verbose` - The verbosity level. + - `--log-file` - File to log to. + - `--log-config` - Python logging config file. + - `--enable-registration` - Enable registration for new users. + +Application services: + +- Reliably retry sending of events from Synapse to application services, as per [Application Services](https://github.com/matrix-org/matrix-doc/blob/0c6bd9/specification/25_application_service_api.rst#home-server---application-service-api) spec. +- Application services can no longer register via the `/register` API, instead their configuration should be saved to a file and listed in the synapse `app_service_config_files` config option. The AS configuration file has the same format as the old `/register` request. See [docs/application\_services.rst](docs/application_services.rst) for more information. + +Changes in synapse v0.8.1 (2015-03-18) +====================================== + +- Disable registration by default. New users can be added using the command `register_new_matrix_user` or by enabling registration in the config. +- Add metrics to synapse. To enable metrics use config options `enable_metrics` and `metrics_port`. +- Fix bug where banning only kicked the user. + +Changes in synapse v0.8.0 (2015-03-06) +====================================== + +General: + +- Add support for registration fallback. This is a page hosted on the server which allows a user to register for an account, regardless of what client they are using (e.g. mobile devices). +- Added new default push rules and made them configurable by clients: + - Suppress all notice messages. + - Notify when invited to a new room. + - Notify for messages that don\'t match any rule. + - Notify on incoming call. + +Federation: + +- Added per host server side rate-limiting of incoming federation requests. +- Added a `/get_missing_events/` API to federation to reduce number of `/events/` requests. + +Configuration: + +- Added configuration option to disable registration: `disable_registration`. +- Added configuration option to change soft limit of number of open file descriptors: `soft_file_limit`. +- Make `tls_private_key_path` optional when running with `no_tls`. + +Application services: + +- Application services can now poll on the CS API `/events` for their events, by providing their application service `access_token`. +- Added exclusive namespace support to application services API. + +Changes in synapse v0.7.1 (2015-02-19) +====================================== + +- Initial alpha implementation of parts of the Application Services API. Including: + - AS Registration / Unregistration + - User Query API + - Room Alias Query API + - Push transport for receiving events. + - User/Alias namespace admin control +- Add cache when fetching events from remote servers to stop repeatedly fetching events with bad signatures. +- Respect the per remote server retry scheme when fetching both events and server keys to reduce the number of times we send requests to dead servers. +- Inform remote servers when the local server fails to handle a received event. +- Turn off python bytecode generation due to problems experienced when upgrading from previous versions. + +Changes in synapse v0.7.0 (2015-02-12) +====================================== + +- Add initial implementation of the query auth federation API, allowing servers to agree on whether an event should be allowed or rejected. +- Persist events we have rejected from federation, fixing the bug where servers would keep requesting the same events. +- Various federation performance improvements, including: + - Add in memory caches on queries such as: + + > - Computing the state of a room at a point in time, used for authorization on federation requests. + > - Fetching events from the database. + > - User\'s room membership, used for authorizing presence updates. + + - Upgraded JSON library to improve parsing and serialisation speeds. + +- Add default avatars to new user accounts using pydenticon library. +- Correctly time out federation requests. +- Retry federation requests against different servers. +- Add support for push and push rules. +- Add alpha versions of proposed new CSv2 APIs, including `/sync` API. + +Changes in synapse 0.6.1 (2015-01-07) +===================================== + +- Major optimizations to improve performance of initial sync and event sending in large rooms (by up to 10x) +- Media repository now includes a Content-Length header on media downloads. +- Improve quality of thumbnails by changing resizing algorithm. + +Changes in synapse 0.6.0 (2014-12-16) +===================================== + +- Add new API for media upload and download that supports thumbnailing. +- Replicate media uploads over multiple homeservers so media is always served to clients from their local homeserver. This obsoletes the \--content-addr parameter and confusion over accessing content directly from remote homeservers. +- Implement exponential backoff when retrying federation requests when sending to remote homeservers which are offline. +- Implement typing notifications. +- Fix bugs where we sent events with invalid signatures due to bugs where we incorrectly persisted events. +- Improve performance of database queries involving retrieving events. + +Changes in synapse 0.5.4a (2014-12-13) +====================================== + +- Fix bug while generating the error message when a file path specified in the config doesn\'t exist. + +Changes in synapse 0.5.4 (2014-12-03) +===================================== + +- Fix presence bug where some rooms did not display presence updates for remote users. +- Do not log SQL timing log lines when started with \"-v\" +- Fix potential memory leak. + +Changes in synapse 0.5.3c (2014-12-02) +====================================== + +- Change the default value for the content\_addr option to use the HTTP listener, as by default the HTTPS listener will be using a self-signed certificate. + +Changes in synapse 0.5.3 (2014-11-27) +===================================== + +- Fix bug that caused joining a remote room to fail if a single event was not signed correctly. +- Fix bug which caused servers to continuously try and fetch events from other servers. + +Changes in synapse 0.5.2 (2014-11-26) +===================================== + +Fix major bug that caused rooms to disappear from peoples initial sync. + +Changes in synapse 0.5.1 (2014-11-26) +===================================== + +See UPGRADES.rst for specific instructions on how to upgrade. + +> - Fix bug where we served up an Event that did not match its signatures. +> - Fix regression where we no longer correctly handled the case where a homeserver receives an event for a room it doesn\'t recognise (but is in.) + +Changes in synapse 0.5.0 (2014-11-19) +===================================== + +This release includes changes to the federation protocol and client-server API that is not backwards compatible. + +This release also changes the internal database schemas and so requires servers to drop their current history. See UPGRADES.rst for details. + +Homeserver: + +: - Add authentication and authorization to the federation protocol. Events are now signed by their originating homeservers. + - Implement the new authorization model for rooms. + - Split out web client into a seperate repository: matrix-angular-sdk. + - Change the structure of PDUs. + - Fix bug where user could not join rooms via an alias containing 4-byte UTF-8 characters. + - Merge concept of PDUs and Events internally. + - Improve logging by adding request ids to log lines. + - Implement a very basic room initial sync API. + - Implement the new invite/join federation APIs. + +Webclient: + +: - The webclient has been moved to a seperate repository. + +Changes in synapse 0.4.2 (2014-10-31) +===================================== + +Homeserver: + +: - Fix bugs where we did not notify users of correct presence updates. + - Fix bug where we did not handle sub second event stream timeouts. + +Webclient: + +: - Add ability to click on messages to see JSON. + - Add ability to redact messages. + - Add ability to view and edit all room state JSON. + - Handle incoming redactions. + - Improve feedback on errors. + - Fix bugs in mobile CSS. + - Fix bugs with desktop notifications. + +Changes in synapse 0.4.1 (2014-10-17) +===================================== + +Webclient: + +: - Fix bug with display of timestamps. + +Changes in synpase 0.4.0 (2014-10-17) +===================================== + +This release includes changes to the federation protocol and client-server API that is not backwards compatible. + +The Matrix specification has been moved to a separate git repository: + +You will also need an updated syutil and config. See UPGRADES.rst. + +Homeserver: + +: - Sign federation transactions to assert strong identity over federation. + - Rename timestamp keys in PDUs and events from \'ts\' and \'hsob\_ts\' to \'origin\_server\_ts\'. + +Changes in synapse 0.3.4 (2014-09-25) +===================================== + +This version adds support for using a TURN server. See docs/turn-howto.rst on how to set one up. + +Homeserver: + +: - Add support for redaction of messages. + - Fix bug where inviting a user on a remote home server could take up to 20-30s. + - Implement a get current room state API. + - Add support specifying and retrieving turn server configuration. + +Webclient: + +: - Add button to send messages to users from the home page. + - Add support for using TURN for VoIP calls. + - Show display name change messages. + - Fix bug where the client didn\'t get the state of a newly joined room until after it has been refreshed. + - Fix bugs with tab complete. + - Fix bug where holding down the down arrow caused chrome to chew 100% CPU. + - Fix bug where desktop notifications occasionally used \"Undefined\" as the display name. + - Fix more places where we sometimes saw room IDs incorrectly. + - Fix bug which caused lag when entering text in the text box. + +Changes in synapse 0.3.3 (2014-09-22) +===================================== + +Homeserver: + +: - Fix bug where you continued to get events for rooms you had left. + +Webclient: + +: - Add support for video calls with basic UI. + - Fix bug where one to one chats were named after your display name rather than the other person\'s. + - Fix bug which caused lag when typing in the textarea. + - Refuse to run on browsers we know won\'t work. + - Trigger pagination when joining new rooms. + - Fix bug where we sometimes didn\'t display invitations in recents. + - Automatically join room when accepting a VoIP call. + - Disable outgoing and reject incoming calls on browsers we don\'t support VoIP in. + - Don\'t display desktop notifications for messages in the room you are non-idle and speaking in. + +Changes in synapse 0.3.2 (2014-09-18) +===================================== + +Webclient: + +: - Fix bug where an empty \"bing words\" list in old accounts didn\'t send notifications when it should have done. + +Changes in synapse 0.3.1 (2014-09-18) +===================================== + +This is a release to hotfix v0.3.0 to fix two regressions. + +Webclient: + +: - Fix a regression where we sometimes displayed duplicate events. + - Fix a regression where we didn\'t immediately remove rooms you were banned in from the recents list. + +Changes in synapse 0.3.0 (2014-09-18) +===================================== + +See UPGRADE for information about changes to the client server API, including breaking backwards compatibility with VoIP calls and registration API. + +Homeserver: + +: - When a user changes their displayname or avatar the server will now update all their join states to reflect this. + - The server now adds \"age\" key to events to indicate how old they are. This is clock independent, so at no point does any server or webclient have to assume their clock is in sync with everyone else. + - Fix bug where we didn\'t correctly pull in missing PDUs. + - Fix bug where prev\_content key wasn\'t always returned. + - Add support for password resets. + +Webclient: + +: - Improve page content loading. + - Join/parts now trigger desktop notifications. + - Always show room aliases in the UI if one is present. + - No longer show user-count in the recents side panel. + - Add up & down arrow support to the text box for message sending to step through your sent history. + - Don\'t display notifications for our own messages. + - Emotes are now formatted correctly in desktop notifications. + - The recents list now differentiates between public & private rooms. + - Fix bug where when switching between rooms the pagination flickered before the view jumped to the bottom of the screen. + - Add bing word support. + +Registration API: + +: - The registration API has been overhauled to function like the login API. In practice, this means registration requests must now include the following: \'type\':\'m.login.password\'. See UPGRADE for more information on this. + - The \'user\_id\' key has been renamed to \'user\' to better match the login API. + - There is an additional login type: \'m.login.email.identity\'. + - The command client and web client have been updated to reflect these changes. + +Changes in synapse 0.2.3 (2014-09-12) +===================================== + +Homeserver: + +: - Fix bug where we stopped sending events to remote home servers if a user from that home server left, even if there were some still in the room. + - Fix bugs in the state conflict resolution where it was incorrectly rejecting events. + +Webclient: + +: - Display room names and topics. + - Allow setting/editing of room names and topics. + - Display information about rooms on the main page. + - Handle ban and kick events in real time. + - VoIP UI and reliability improvements. + - Add glare support for VoIP. + - Improvements to initial startup speed. + - Don\'t display duplicate join events. + - Local echo of messages. + - Differentiate sending and sent of local echo. + - Various minor bug fixes. + +Changes in synapse 0.2.2 (2014-09-06) +===================================== + +Homeserver: + +: - When the server returns state events it now also includes the previous content. + - Add support for inviting people when creating a new room. + - Make the homeserver inform the room via m.room.aliases when a new alias is added for a room. + - Validate m.room.power\_level events. + +Webclient: + +: - Add support for captchas on registration. + - Handle m.room.aliases events. + - Asynchronously send messages and show a local echo. + - Inform the UI when a message failed to send. + - Only autoscroll on receiving a new message if the user was already at the bottom of the screen. + - Add support for ban/kick reasons. + +Changes in synapse 0.2.1 (2014-09-03) +===================================== + +Homeserver: + +: - Added support for signing up with a third party id. + - Add synctl scripts. + - Added rate limiting. + - Add option to change the external address the content repo uses. + - Presence bug fixes. + +Webclient: + +: - Added support for signing up with a third party id. + - Added support for banning and kicking users. + - Added support for displaying and setting ops. + - Added support for room names. + - Fix bugs with room membership event display. + +Changes in synapse 0.2.0 (2014-09-02) +===================================== + +This update changes many configuration options, updates the database schema and mandates SSL for server-server connections. + +Homeserver: + +: - Require SSL for server-server connections. + - Add SSL listener for client-server connections. + - Add ability to use config files. + - Add support for kicking/banning and power levels. + - Allow setting of room names and topics on creation. + - Change presence to include last seen time of the user. + - Change url path prefix to /\_matrix/\... + - Bug fixes to presence. + +Webclient: + +: - Reskin the CSS for registration and login. + - Various improvements to rooms CSS. + - Support changes in client-server API. + - Bug fixes to VOIP UI. + - Various bug fixes to handling of changes to room member list. + +Changes in synapse 0.1.2 (2014-08-29) +===================================== + +Webclient: + +: - Add basic call state UI for VoIP calls. + +Changes in synapse 0.1.1 (2014-08-29) +===================================== + +Homeserver: + +: - Fix bug that caused the event stream to not notify some clients about changes. + +Changes in synapse 0.1.0 (2014-08-29) +===================================== + +Presence has been reenabled in this release. + +Homeserver: + +: - + + Update client to server API, including: + + : - Use a more consistent url scheme. + - Provide more useful information in the initial sync api. + + - Change the presence handling to be much more efficient. + - Change the presence server to server API to not require explicit polling of all users who share a room with a user. + - Fix races in the event streaming logic. + +Webclient: + +: - Update to use new client to server API. + - Add basic VOIP support. + - Add idle timers that change your status to away. + - Add recent rooms column when viewing a room. + - Various network efficiency improvements. + - Add basic mobile browser support. + - Add a settings page. + +Changes in synapse 0.0.1 (2014-08-22) +===================================== + +Presence has been disabled in this release due to a bug that caused the homeserver to spam other remote homeservers. + +Homeserver: + +: - Completely change the database schema to support generic event types. + - Improve presence reliability. + - Improve reliability of joining remote rooms. + - Fix bug where room join events were duplicated. + - Improve initial sync API to return more information to the client. + - Stop generating fake messages for room membership events. + +Webclient: + +: - Add tab completion of names. + - Add ability to upload and send images. + - Add profile pages. + - Improve CSS layout of room. + - Disambiguate identical display names. + - Don\'t get remote users display names and avatars individually. + - Use the new initial sync API to reduce number of round trips to the homeserver. + - Change url scheme to use room aliases instead of room ids where known. + - Increase longpoll timeout. + +Changes in synapse 0.0.0 (2014-08-13) +===================================== + +> - Initial alpha release diff --git a/CHANGES.rst b/CHANGES.rst deleted file mode 100644 index ce356a11b..000000000 --- a/CHANGES.rst +++ /dev/null @@ -1,2074 +0,0 @@ -Changes in synapse v0.22.0-rc1 (2017-06-26) -=========================================== - -Features: - -* Add a user directory API (PR #2252, and many more) -* Add shutdown room API to remove room from local server (PR #2291) -* Add API to quarantine media (PR #2292) -* Add new config option to not send event contents to push servers (PR #2301) - Thanks to @cjdelisle! - -Changes: - -* Various performance fixes (PR #2177, #2233, #2230, #2238, #2248, #2256, - #2274) -* Deduplicate sync filters (PR #2219) Thanks to @krombel! -* Correct a typo in UPGRADE.rst (PR #2231) Thanks to @aaronraimist! -* Add count of one time keys to sync stream (PR #2237) -* Only store event_auth for state events (PR #2247) -* Store URL cache preview downloads separately (PR #2299) - -Bug fixes: - -* Fix users not getting notifications when AS listened to that user_id (PR - #2216) Thanks to @slipeer! -* Fix users without push set up not getting notifications after joining rooms - (PR #2236) -* Fix preview url API to trim long descriptions (PR #2243) -* Fix bug where we used cached but unpersisted state group as prev group, - resulting in broken state of restart (PR #2263) -* Fix removing of pushers when using workers (PR #2267) -* Fix CORS headers to allow Authorization header (PR #2285) Thanks to @krombel! - - -Changes in synapse v0.21.1 (2017-06-15) -======================================= - -Bug fixes: - -* Fix bug in anonymous usage statistic reporting (PR #2281) - - -Changes in synapse v0.21.0 (2017-05-18) -======================================= - -No changes since v0.21.0-rc3 - - -Changes in synapse v0.21.0-rc3 (2017-05-17) -=========================================== - -Features: - -* Add per user rate-limiting overrides (PR #2208) -* Add config option to limit maximum number of events requested by ``/sync`` - and ``/messages`` (PR #2221) Thanks to @psaavedra! - - -Changes: - -* Various small performance fixes (PR #2201, #2202, #2224, #2226, #2227, #2228, - #2229) -* Update username availability checker API (PR #2209, #2213) -* When purging, don't de-delta state groups we're about to delete (PR #2214) -* Documentation to check synapse version (PR #2215) Thanks to @hamber-dick! -* Add an index to event_search to speed up purge history API (PR #2218) - - -Bug fixes: - -* Fix API to allow clients to upload one-time-keys with new sigs (PR #2206) - - -Changes in synapse v0.21.0-rc2 (2017-05-08) -=========================================== - -Changes: - -* Always mark remotes as up if we receive a signed request from them (PR #2190) - - -Bug fixes: - -* Fix bug where users got pushed for rooms they had muted (PR #2200) - - -Changes in synapse v0.21.0-rc1 (2017-05-08) -=========================================== - -Features: - -* Add username availability checker API (PR #2183) -* Add read marker API (PR #2120) - - -Changes: - -* Enable guest access for the 3pl/3pid APIs (PR #1986) -* Add setting to support TURN for guests (PR #2011) -* Various performance improvements (PR #2075, #2076, #2080, #2083, #2108, - #2158, #2176, #2185) -* Make synctl a bit more user friendly (PR #2078, #2127) Thanks @APwhitehat! -* Replace HTTP replication with TCP replication (PR #2082, #2097, #2098, - #2099, #2103, #2014, #2016, #2115, #2116, #2117) -* Support authenticated SMTP (PR #2102) Thanks @DanielDent! -* Add a counter metric for successfully-sent transactions (PR #2121) -* Propagate errors sensibly from proxied IS requests (PR #2147) -* Add more granular event send metrics (PR #2178) - - - -Bug fixes: - -* Fix nuke-room script to work with current schema (PR #1927) Thanks - @zuckschwerdt! -* Fix db port script to not assume postgres tables are in the public schema - (PR #2024) Thanks @jerrykan! -* Fix getting latest device IP for user with no devices (PR #2118) -* Fix rejection of invites to unreachable servers (PR #2145) -* Fix code for reporting old verify keys in synapse (PR #2156) -* Fix invite state to always include all events (PR #2163) -* Fix bug where synapse would always fetch state for any missing event (PR #2170) -* Fix a leak with timed out HTTP connections (PR #2180) -* Fix bug where we didn't time out HTTP requests to ASes (PR #2192) - - -Docs: - -* Clarify doc for SQLite to PostgreSQL port (PR #1961) Thanks @benhylau! -* Fix typo in synctl help (PR #2107) Thanks @HarHarLinks! -* ``web_client_location`` documentation fix (PR #2131) Thanks @matthewjwolff! -* Update README.rst with FreeBSD changes (PR #2132) Thanks @feld! -* Clarify setting up metrics (PR #2149) Thanks @encks! - - -Changes in synapse v0.20.0 (2017-04-11) -======================================= - -Bug fixes: - -* Fix joining rooms over federation where not all servers in the room saw the - new server had joined (PR #2094) - - -Changes in synapse v0.20.0-rc1 (2017-03-30) -=========================================== - -Features: - -* Add delete_devices API (PR #1993) -* Add phone number registration/login support (PR #1994, #2055) - - -Changes: - -* Use JSONSchema for validation of filters. Thanks @pik! (PR #1783) -* Reread log config on SIGHUP (PR #1982) -* Speed up public room list (PR #1989) -* Add helpful texts to logger config options (PR #1990) -* Minor ``/sync`` performance improvements. (PR #2002, #2013, #2022) -* Add some debug to help diagnose weird federation issue (PR #2035) -* Correctly limit retries for all federation requests (PR #2050, #2061) -* Don't lock table when persisting new one time keys (PR #2053) -* Reduce some CPU work on DB threads (PR #2054) -* Cache hosts in room (PR #2060) -* Batch sending of device list pokes (PR #2063) -* Speed up persist event path in certain edge cases (PR #2070) - - -Bug fixes: - -* Fix bug where current_state_events renamed to current_state_ids (PR #1849) -* Fix routing loop when fetching remote media (PR #1992) -* Fix current_state_events table to not lie (PR #1996) -* Fix CAS login to handle PartialDownloadError (PR #1997) -* Fix assertion to stop transaction queue getting wedged (PR #2010) -* Fix presence to fallback to last_active_ts if it beats the last sync time. - Thanks @Half-Shot! (PR #2014) -* Fix bug when federation received a PDU while a room join is in progress (PR - #2016) -* Fix resetting state on rejected events (PR #2025) -* Fix installation issues in readme. Thanks @ricco386 (PR #2037) -* Fix caching of remote servers' signature keys (PR #2042) -* Fix some leaking log context (PR #2048, #2049, #2057, #2058) -* Fix rejection of invites not reaching sync (PR #2056) - - - -Changes in synapse v0.19.3 (2017-03-20) -======================================= - -No changes since v0.19.3-rc2 - - -Changes in synapse v0.19.3-rc2 (2017-03-13) -=========================================== - -Bug fixes: - -* Fix bug in handling of incoming device list updates over federation. - - - -Changes in synapse v0.19.3-rc1 (2017-03-08) -=========================================== - -Features: - -* Add some administration functionalities. Thanks to morteza-araby! (PR #1784) - - -Changes: - -* Reduce database table sizes (PR #1873, #1916, #1923, #1963) -* Update contrib/ to not use syutil. Thanks to andrewshadura! (PR #1907) -* Don't fetch current state when sending an event in common case (PR #1955) - - -Bug fixes: - -* Fix synapse_port_db failure. Thanks to Pneumaticat! (PR #1904) -* Fix caching to not cache error responses (PR #1913) -* Fix APIs to make kick & ban reasons work (PR #1917) -* Fix bugs in the /keys/changes api (PR #1921) -* Fix bug where users couldn't forget rooms they were banned from (PR #1922) -* Fix issue with long language values in pushers API (PR #1925) -* Fix a race in transaction queue (PR #1930) -* Fix dynamic thumbnailing to preserve aspect ratio. Thanks to jkolo! (PR - #1945) -* Fix device list update to not constantly resync (PR #1964) -* Fix potential for huge memory usage when getting device that have - changed (PR #1969) - - - -Changes in synapse v0.19.2 (2017-02-20) -======================================= - -* Fix bug with event visibility check in /context/ API. Thanks to Tokodomo for - pointing it out! (PR #1929) - - -Changes in synapse v0.19.1 (2017-02-09) -======================================= - -* Fix bug where state was incorrectly reset in a room when synapse received an - event over federation that did not pass auth checks (PR #1892) - - -Changes in synapse v0.19.0 (2017-02-04) -======================================= - -No changes since RC 4. - - -Changes in synapse v0.19.0-rc4 (2017-02-02) -=========================================== - -* Bump cache sizes for common membership queries (PR #1879) - - -Changes in synapse v0.19.0-rc3 (2017-02-02) -=========================================== - -* Fix email push in pusher worker (PR #1875) -* Make presence.get_new_events a bit faster (PR #1876) -* Make /keys/changes a bit more performant (PR #1877) - - -Changes in synapse v0.19.0-rc2 (2017-02-02) -=========================================== - -* Include newly joined users in /keys/changes API (PR #1872) - - -Changes in synapse v0.19.0-rc1 (2017-02-02) -=========================================== - -Features: - -* Add support for specifying multiple bind addresses (PR #1709, #1712, #1795, - #1835). Thanks to @kyrias! -* Add /account/3pid/delete endpoint (PR #1714) -* Add config option to configure the Riot URL used in notification emails (PR - #1811). Thanks to @aperezdc! -* Add username and password config options for turn server (PR #1832). Thanks - to @xsteadfastx! -* Implement device lists updates over federation (PR #1857, #1861, #1864) -* Implement /keys/changes (PR #1869, #1872) - - -Changes: - -* Improve IPv6 support (PR #1696). Thanks to @kyrias and @glyph! -* Log which files we saved attachments to in the media_repository (PR #1791) -* Linearize updates to membership via PUT /state/ to better handle multiple - joins (PR #1787) -* Limit number of entries to prefill from cache on startup (PR #1792) -* Remove full_twisted_stacktraces option (PR #1802) -* Measure size of some caches by sum of the size of cached values (PR #1815) -* Measure metrics of string_cache (PR #1821) -* Reduce logging verbosity (PR #1822, #1823, #1824) -* Don't clobber a displayname or avatar_url if provided by an m.room.member - event (PR #1852) -* Better handle 401/404 response for federation /send/ (PR #1866, #1871) - - -Fixes: - -* Fix ability to change password to a non-ascii one (PR #1711) -* Fix push getting stuck due to looking at the wrong view of state (PR #1820) -* Fix email address comparison to be case insensitive (PR #1827) -* Fix occasional inconsistencies of room membership (PR #1836, #1840) - - -Performance: - -* Don't block messages sending on bumping presence (PR #1789) -* Change device_inbox stream index to include user (PR #1793) -* Optimise state resolution (PR #1818) -* Use DB cache of joined users for presence (PR #1862) -* Add an index to make membership queries faster (PR #1867) - - -Changes in synapse v0.18.7 (2017-01-09) -======================================= - -No changes from v0.18.7-rc2 - - -Changes in synapse v0.18.7-rc2 (2017-01-07) -=========================================== - -Bug fixes: - -* Fix error in rc1's discarding invalid inbound traffic logic that was - incorrectly discarding missing events - - -Changes in synapse v0.18.7-rc1 (2017-01-06) -=========================================== - -Bug fixes: - -* Fix error in #PR 1764 to actually fix the nightmare #1753 bug. -* Improve deadlock logging further -* Discard inbound federation traffic from invalid domains, to immunise - against #1753 - - -Changes in synapse v0.18.6 (2017-01-06) -======================================= - -Bug fixes: - -* Fix bug when checking if a guest user is allowed to join a room (PR #1772) - Thanks to Patrik Oldsberg for diagnosing and the fix! - - -Changes in synapse v0.18.6-rc3 (2017-01-05) -=========================================== - -Bug fixes: - -* Fix bug where we failed to send ban events to the banned server (PR #1758) -* Fix bug where we sent event that didn't originate on this server to - other servers (PR #1764) -* Fix bug where processing an event from a remote server took a long time - because we were making long HTTP requests (PR #1765, PR #1744) - -Changes: - -* Improve logging for debugging deadlocks (PR #1766, PR #1767) - - -Changes in synapse v0.18.6-rc2 (2016-12-30) -=========================================== - -Bug fixes: - -* Fix memory leak in twisted by initialising logging correctly (PR #1731) -* Fix bug where fetching missing events took an unacceptable amount of time in - large rooms (PR #1734) - - -Changes in synapse v0.18.6-rc1 (2016-12-29) -=========================================== - -Bug fixes: - -* Make sure that outbound connections are closed (PR #1725) - - -Changes in synapse v0.18.5 (2016-12-16) -======================================= - -Bug fixes: - -* Fix federation /backfill returning events it shouldn't (PR #1700) -* Fix crash in url preview (PR #1701) - - -Changes in synapse v0.18.5-rc3 (2016-12-13) -=========================================== - -Features: - -* Add support for E2E for guests (PR #1653) -* Add new API appservice specific public room list (PR #1676) -* Add new room membership APIs (PR #1680) - - -Changes: - -* Enable guest access for private rooms by default (PR #653) -* Limit the number of events that can be created on a given room concurrently - (PR #1620) -* Log the args that we have on UI auth completion (PR #1649) -* Stop generating refresh_tokens (PR #1654) -* Stop putting a time caveat on access tokens (PR #1656) -* Remove unspecced GET endpoints for e2e keys (PR #1694) - - -Bug fixes: - -* Fix handling of 500 and 429's over federation (PR #1650) -* Fix Content-Type header parsing (PR #1660) -* Fix error when previewing sites that include unicode, thanks to kyrias (PR - #1664) -* Fix some cases where we drop read receipts (PR #1678) -* Fix bug where calls to ``/sync`` didn't correctly timeout (PR #1683) -* Fix bug where E2E key query would fail if a single remote host failed (PR - #1686) - - - -Changes in synapse v0.18.5-rc2 (2016-11-24) -=========================================== - -Bug fixes: - -* Don't send old events over federation, fixes bug in -rc1. - -Changes in synapse v0.18.5-rc1 (2016-11-24) -=========================================== - -Features: - -* Implement "event_fields" in filters (PR #1638) - -Changes: - -* Use external ldap auth pacakge (PR #1628) -* Split out federation transaction sending to a worker (PR #1635) -* Fail with a coherent error message if `/sync?filter=` is invalid (PR #1636) -* More efficient notif count queries (PR #1644) - - -Changes in synapse v0.18.4 (2016-11-22) -======================================= - -Bug fixes: - -* Add workaround for buggy clients that the fail to register (PR #1632) - - -Changes in synapse v0.18.4-rc1 (2016-11-14) -=========================================== - -Changes: - -* Various database efficiency improvements (PR #1188, #1192) -* Update default config to blacklist more internal IPs, thanks to Euan Kemp (PR - #1198) -* Allow specifying duration in minutes in config, thanks to Daniel Dent (PR - #1625) - - -Bug fixes: - -* Fix media repo to set CORs headers on responses (PR #1190) -* Fix registration to not error on non-ascii passwords (PR #1191) -* Fix create event code to limit the number of prev_events (PR #1615) -* Fix bug in transaction ID deduplication (PR #1624) - - -Changes in synapse v0.18.3 (2016-11-08) -======================================= - -SECURITY UPDATE - -Explicitly require authentication when using LDAP3. This is the default on -versions of ``ldap3`` above 1.0, but some distributions will package an older -version. - -If you are using LDAP3 login and have a version of ``ldap3`` older than 1.0 it -is **CRITICAL to updgrade**. - - -Changes in synapse v0.18.2 (2016-11-01) -======================================= - -No changes since v0.18.2-rc5 - - -Changes in synapse v0.18.2-rc5 (2016-10-28) -=========================================== - -Bug fixes: - -* Fix prometheus process metrics in worker processes (PR #1184) - - -Changes in synapse v0.18.2-rc4 (2016-10-27) -=========================================== - -Bug fixes: - -* Fix ``user_threepids`` schema delta, which in some instances prevented - startup after upgrade (PR #1183) - - -Changes in synapse v0.18.2-rc3 (2016-10-27) -=========================================== - -Changes: - -* Allow clients to supply access tokens as headers (PR #1098) -* Clarify error codes for GET /filter/, thanks to Alexander Maznev (PR #1164) -* Make password reset email field case insensitive (PR #1170) -* Reduce redundant database work in email pusher (PR #1174) -* Allow configurable rate limiting per AS (PR #1175) -* Check whether to ratelimit sooner to avoid work (PR #1176) -* Standardise prometheus metrics (PR #1177) - - -Bug fixes: - -* Fix incredibly slow back pagination query (PR #1178) -* Fix infinite typing bug (PR #1179) - - -Changes in synapse v0.18.2-rc2 (2016-10-25) -=========================================== - -(This release did not include the changes advertised and was identical to RC1) - - -Changes in synapse v0.18.2-rc1 (2016-10-17) -=========================================== - -Changes: - -* Remove redundant event_auth index (PR #1113) -* Reduce DB hits for replication (PR #1141) -* Implement pluggable password auth (PR #1155) -* Remove rate limiting from app service senders and fix get_or_create_user - requester, thanks to Patrik Oldsberg (PR #1157) -* window.postmessage for Interactive Auth fallback (PR #1159) -* Use sys.executable instead of hardcoded python, thanks to Pedro Larroy - (PR #1162) -* Add config option for adding additional TLS fingerprints (PR #1167) -* User-interactive auth on delete device (PR #1168) - - -Bug fixes: - -* Fix not being allowed to set your own state_key, thanks to Patrik Oldsberg - (PR #1150) -* Fix interactive auth to return 401 from for incorrect password (PR #1160, - #1166) -* Fix email push notifs being dropped (PR #1169) - - - -Changes in synapse v0.18.1 (2016-10-05) -====================================== - -No changes since v0.18.1-rc1 - - -Changes in synapse v0.18.1-rc1 (2016-09-30) -=========================================== - -Features: - -* Add total_room_count_estimate to ``/publicRooms`` (PR #1133) - - -Changes: - -* Time out typing over federation (PR #1140) -* Restructure LDAP authentication (PR #1153) - - -Bug fixes: - -* Fix 3pid invites when server is already in the room (PR #1136) -* Fix upgrading with SQLite taking lots of CPU for a few days - after upgrade (PR #1144) -* Fix upgrading from very old database versions (PR #1145) -* Fix port script to work with recently added tables (PR #1146) - - -Changes in synapse v0.18.0 (2016-09-19) -======================================= - -The release includes major changes to the state storage database schemas, which -significantly reduce database size. Synapse will attempt to upgrade the current -data in the background. Servers with large SQLite database may experience -degradation of performance while this upgrade is in progress, therefore you may -want to consider migrating to using Postgres before upgrading very large SQLite -databases - - -Changes: - -* Make public room search case insensitive (PR #1127) - - -Bug fixes: - -* Fix and clean up publicRooms pagination (PR #1129) - - -Changes in synapse v0.18.0-rc1 (2016-09-16) -=========================================== - -Features: - -* Add ``only=highlight`` on ``/notifications`` (PR #1081) -* Add server param to /publicRooms (PR #1082) -* Allow clients to ask for the whole of a single state event (PR #1094) -* Add is_direct param to /createRoom (PR #1108) -* Add pagination support to publicRooms (PR #1121) -* Add very basic filter API to /publicRooms (PR #1126) -* Add basic direct to device messaging support for E2E (PR #1074, #1084, #1104, - #1111) - - -Changes: - -* Move to storing state_groups_state as deltas, greatly reducing DB size (PR - #1065) -* Reduce amount of state pulled out of the DB during common requests (PR #1069) -* Allow PDF to be rendered from media repo (PR #1071) -* Reindex state_groups_state after pruning (PR #1085) -* Clobber EDUs in send queue (PR #1095) -* Conform better to the CAS protocol specification (PR #1100) -* Limit how often we ask for keys from dead servers (PR #1114) - - -Bug fixes: - -* Fix /notifications API when used with ``from`` param (PR #1080) -* Fix backfill when cannot find an event. (PR #1107) - - -Changes in synapse v0.17.3 (2016-09-09) -======================================= - -This release fixes a major bug that stopped servers from handling rooms with -over 1000 members. - - -Changes in synapse v0.17.2 (2016-09-08) -======================================= - -This release contains security bug fixes. Please upgrade. - - -No changes since v0.17.2-rc1 - - -Changes in synapse v0.17.2-rc1 (2016-09-05) -=========================================== - -Features: - -* Start adding store-and-forward direct-to-device messaging (PR #1046, #1050, - #1062, #1066) - - -Changes: - -* Avoid pulling the full state of a room out so often (PR #1047, #1049, #1063, - #1068) -* Don't notify for online to online presence transitions. (PR #1054) -* Occasionally persist unpersisted presence updates (PR #1055) -* Allow application services to have an optional 'url' (PR #1056) -* Clean up old sent transactions from DB (PR #1059) - - -Bug fixes: - -* Fix None check in backfill (PR #1043) -* Fix membership changes to be idempotent (PR #1067) -* Fix bug in get_pdu where it would sometimes return events with incorrect - signature - - - -Changes in synapse v0.17.1 (2016-08-24) -======================================= - -Changes: - -* Delete old received_transactions rows (PR #1038) -* Pass through user-supplied content in /join/$room_id (PR #1039) - - -Bug fixes: - -* Fix bug with backfill (PR #1040) - - -Changes in synapse v0.17.1-rc1 (2016-08-22) -=========================================== - -Features: - -* Add notification API (PR #1028) - - -Changes: - -* Don't print stack traces when failing to get remote keys (PR #996) -* Various federation /event/ perf improvements (PR #998) -* Only process one local membership event per room at a time (PR #1005) -* Move default display name push rule (PR #1011, #1023) -* Fix up preview URL API. Add tests. (PR #1015) -* Set ``Content-Security-Policy`` on media repo (PR #1021) -* Make notify_interested_services faster (PR #1022) -* Add usage stats to prometheus monitoring (PR #1037) - - -Bug fixes: - -* Fix token login (PR #993) -* Fix CAS login (PR #994, #995) -* Fix /sync to not clobber status_msg (PR #997) -* Fix redacted state events to include prev_content (PR #1003) -* Fix some bugs in the auth/ldap handler (PR #1007) -* Fix backfill request to limit URI length, so that remotes don't reject the - requests due to path length limits (PR #1012) -* Fix AS push code to not send duplicate events (PR #1025) - - - -Changes in synapse v0.17.0 (2016-08-08) -======================================= - -This release contains significant security bug fixes regarding authenticating -events received over federation. PLEASE UPGRADE. - -This release changes the LDAP configuration format in a backwards incompatible -way, see PR #843 for details. - - -Changes: - -* Add federation /version API (PR #990) -* Make psutil dependency optional (PR #992) - - -Bug fixes: - -* Fix URL preview API to exclude HTML comments in description (PR #988) -* Fix error handling of remote joins (PR #991) - - -Changes in synapse v0.17.0-rc4 (2016-08-05) -=========================================== - -Changes: - -* Change the way we summarize URLs when previewing (PR #973) -* Add new ``/state_ids/`` federation API (PR #979) -* Speed up processing of ``/state/`` response (PR #986) - -Bug fixes: - -* Fix event persistence when event has already been partially persisted - (PR #975, #983, #985) -* Fix port script to also copy across backfilled events (PR #982) - - -Changes in synapse v0.17.0-rc3 (2016-08-02) -=========================================== - -Changes: - -* Forbid non-ASes from registering users whose names begin with '_' (PR #958) -* Add some basic admin API docs (PR #963) - - -Bug fixes: - -* Send the correct host header when fetching keys (PR #941) -* Fix joining a room that has missing auth events (PR #964) -* Fix various push bugs (PR #966, #970) -* Fix adding emails on registration (PR #968) - - -Changes in synapse v0.17.0-rc2 (2016-08-02) -=========================================== - -(This release did not include the changes advertised and was identical to RC1) - - -Changes in synapse v0.17.0-rc1 (2016-07-28) -=========================================== - -This release changes the LDAP configuration format in a backwards incompatible -way, see PR #843 for details. - - -Features: - -* Add purge_media_cache admin API (PR #902) -* Add deactivate account admin API (PR #903) -* Add optional pepper to password hashing (PR #907, #910 by KentShikama) -* Add an admin option to shared secret registration (breaks backwards compat) - (PR #909) -* Add purge local room history API (PR #911, #923, #924) -* Add requestToken endpoints (PR #915) -* Add an /account/deactivate endpoint (PR #921) -* Add filter param to /messages. Add 'contains_url' to filter. (PR #922) -* Add device_id support to /login (PR #929) -* Add device_id support to /v2/register flow. (PR #937, #942) -* Add GET /devices endpoint (PR #939, #944) -* Add GET /device/{deviceId} (PR #943) -* Add update and delete APIs for devices (PR #949) - - -Changes: - -* Rewrite LDAP Authentication against ldap3 (PR #843 by mweinelt) -* Linearize some federation endpoints based on (origin, room_id) (PR #879) -* Remove the legacy v0 content upload API. (PR #888) -* Use similar naming we use in email notifs for push (PR #894) -* Optionally include password hash in createUser endpoint (PR #905 by - KentShikama) -* Use a query that postgresql optimises better for get_events_around (PR #906) -* Fall back to 'username' if 'user' is not given for appservice registration. - (PR #927 by Half-Shot) -* Add metrics for psutil derived memory usage (PR #936) -* Record device_id in client_ips (PR #938) -* Send the correct host header when fetching keys (PR #941) -* Log the hostname the reCAPTCHA was completed on (PR #946) -* Make the device id on e2e key upload optional (PR #956) -* Add r0.2.0 to the "supported versions" list (PR #960) -* Don't include name of room for invites in push (PR #961) - - -Bug fixes: - -* Fix substitution failure in mail template (PR #887) -* Put most recent 20 messages in email notif (PR #892) -* Ensure that the guest user is in the database when upgrading accounts - (PR #914) -* Fix various edge cases in auth handling (PR #919) -* Fix 500 ISE when sending alias event without a state_key (PR #925) -* Fix bug where we stored rejections in the state_group, persist all - rejections (PR #948) -* Fix lack of check of if the user is banned when handling 3pid invites - (PR #952) -* Fix a couple of bugs in the transaction and keyring code (PR #954, #955) - - - -Changes in synapse v0.16.1-r1 (2016-07-08) -========================================== - -THIS IS A CRITICAL SECURITY UPDATE. - -This fixes a bug which allowed users' accounts to be accessed by unauthorised -users. - -Changes in synapse v0.16.1 (2016-06-20) -======================================= - -Bug fixes: - -* Fix assorted bugs in ``/preview_url`` (PR #872) -* Fix TypeError when setting unicode passwords (PR #873) - - -Performance improvements: - -* Turn ``use_frozen_events`` off by default (PR #877) -* Disable responding with canonical json for federation (PR #878) - - -Changes in synapse v0.16.1-rc1 (2016-06-15) -=========================================== - -Features: None - -Changes: - -* Log requester for ``/publicRoom`` endpoints when possible (PR #856) -* 502 on ``/thumbnail`` when can't connect to remote server (PR #862) -* Linearize fetching of gaps on incoming events (PR #871) - - -Bugs fixes: - -* Fix bug where rooms where marked as published by default (PR #857) -* Fix bug where joining room with an event with invalid sender (PR #868) -* Fix bug where backfilled events were sent down sync streams (PR #869) -* Fix bug where outgoing connections could wedge indefinitely, causing push - notifications to be unreliable (PR #870) - - -Performance improvements: - -* Improve ``/publicRooms`` performance(PR #859) - - -Changes in synapse v0.16.0 (2016-06-09) -======================================= - -NB: As of v0.14 all AS config files must have an ID field. - - -Bug fixes: - -* Don't make rooms published by default (PR #857) - -Changes in synapse v0.16.0-rc2 (2016-06-08) -=========================================== - -Features: - -* Add configuration option for tuning GC via ``gc.set_threshold`` (PR #849) - -Changes: - -* Record metrics about GC (PR #771, #847, #852) -* Add metric counter for number of persisted events (PR #841) - -Bug fixes: - -* Fix 'From' header in email notifications (PR #843) -* Fix presence where timeouts were not being fired for the first 8h after - restarts (PR #842) -* Fix bug where synapse sent malformed transactions to AS's when retrying - transactions (Commits 310197b, 8437906) - -Performance improvements: - -* Remove event fetching from DB threads (PR #835) -* Change the way we cache events (PR #836) -* Add events to cache when we persist them (PR #840) - - -Changes in synapse v0.16.0-rc1 (2016-06-03) -=========================================== - -Version 0.15 was not released. See v0.15.0-rc1 below for additional changes. - -Features: - -* Add email notifications for missed messages (PR #759, #786, #799, #810, #815, - #821) -* Add a ``url_preview_ip_range_whitelist`` config param (PR #760) -* Add /report endpoint (PR #762) -* Add basic ignore user API (PR #763) -* Add an openidish mechanism for proving that you own a given user_id (PR #765) -* Allow clients to specify a server_name to avoid 'No known servers' (PR #794) -* Add secondary_directory_servers option to fetch room list from other servers - (PR #808, #813) - -Changes: - -* Report per request metrics for all of the things using request_handler (PR - #756) -* Correctly handle ``NULL`` password hashes from the database (PR #775) -* Allow receipts for events we haven't seen in the db (PR #784) -* Make synctl read a cache factor from config file (PR #785) -* Increment badge count per missed convo, not per msg (PR #793) -* Special case m.room.third_party_invite event auth to match invites (PR #814) - - -Bug fixes: - -* Fix typo in event_auth servlet path (PR #757) -* Fix password reset (PR #758) - - -Performance improvements: - -* Reduce database inserts when sending transactions (PR #767) -* Queue events by room for persistence (PR #768) -* Add cache to ``get_user_by_id`` (PR #772) -* Add and use ``get_domain_from_id`` (PR #773) -* Use tree cache for ``get_linearized_receipts_for_room`` (PR #779) -* Remove unused indices (PR #782) -* Add caches to ``bulk_get_push_rules*`` (PR #804) -* Cache ``get_event_reference_hashes`` (PR #806) -* Add ``get_users_with_read_receipts_in_room`` cache (PR #809) -* Use state to calculate ``get_users_in_room`` (PR #811) -* Load push rules in storage layer so that they get cached (PR #825) -* Make ``get_joined_hosts_for_room`` use get_users_in_room (PR #828) -* Poke notifier on next reactor tick (PR #829) -* Change CacheMetrics to be quicker (PR #830) - - -Changes in synapse v0.15.0-rc1 (2016-04-26) -=========================================== - -Features: - -* Add login support for Javascript Web Tokens, thanks to Niklas Riekenbrauck - (PR #671,#687) -* Add URL previewing support (PR #688) -* Add login support for LDAP, thanks to Christoph Witzany (PR #701) -* Add GET endpoint for pushers (PR #716) - -Changes: - -* Never notify for member events (PR #667) -* Deduplicate identical ``/sync`` requests (PR #668) -* Require user to have left room to forget room (PR #673) -* Use DNS cache if within TTL (PR #677) -* Let users see their own leave events (PR #699) -* Deduplicate membership changes (PR #700) -* Increase performance of pusher code (PR #705) -* Respond with error status 504 if failed to talk to remote server (PR #731) -* Increase search performance on postgres (PR #745) - -Bug fixes: - -* Fix bug where disabling all notifications still resulted in push (PR #678) -* Fix bug where users couldn't reject remote invites if remote refused (PR #691) -* Fix bug where synapse attempted to backfill from itself (PR #693) -* Fix bug where profile information was not correctly added when joining remote - rooms (PR #703) -* Fix bug where register API required incorrect key name for AS registration - (PR #727) - - -Changes in synapse v0.14.0 (2016-03-30) -======================================= - -No changes from v0.14.0-rc2 - -Changes in synapse v0.14.0-rc2 (2016-03-23) -=========================================== - -Features: - -* Add published room list API (PR #657) - -Changes: - -* Change various caches to consume less memory (PR #656, #658, #660, #662, - #663, #665) -* Allow rooms to be published without requiring an alias (PR #664) -* Intern common strings in caches to reduce memory footprint (#666) - -Bug fixes: - -* Fix reject invites over federation (PR #646) -* Fix bug where registration was not idempotent (PR #649) -* Update aliases event after deleting aliases (PR #652) -* Fix unread notification count, which was sometimes wrong (PR #661) - -Changes in synapse v0.14.0-rc1 (2016-03-14) -=========================================== - -Features: - -* Add event_id to response to state event PUT (PR #581) -* Allow guest users access to messages in rooms they have joined (PR #587) -* Add config for what state is included in a room invite (PR #598) -* Send the inviter's member event in room invite state (PR #607) -* Add error codes for malformed/bad JSON in /login (PR #608) -* Add support for changing the actions for default rules (PR #609) -* Add environment variable SYNAPSE_CACHE_FACTOR, default it to 0.1 (PR #612) -* Add ability for alias creators to delete aliases (PR #614) -* Add profile information to invites (PR #624) - -Changes: - -* Enforce user_id exclusivity for AS registrations (PR #572) -* Make adding push rules idempotent (PR #587) -* Improve presence performance (PR #582, #586) -* Change presence semantics for ``last_active_ago`` (PR #582, #586) -* Don't allow ``m.room.create`` to be changed (PR #596) -* Add 800x600 to default list of valid thumbnail sizes (PR #616) -* Always include kicks and bans in full /sync (PR #625) -* Send history visibility on boundary changes (PR #626) -* Register endpoint now returns a refresh_token (PR #637) - -Bug fixes: - -* Fix bug where we returned incorrect state in /sync (PR #573) -* Always return a JSON object from push rule API (PR #606) -* Fix bug where registering without a user id sometimes failed (PR #610) -* Report size of ExpiringCache in cache size metrics (PR #611) -* Fix rejection of invites to empty rooms (PR #615) -* Fix usage of ``bcrypt`` to not use ``checkpw`` (PR #619) -* Pin ``pysaml2`` dependency (PR #634) -* Fix bug in ``/sync`` where timeline order was incorrect for backfilled events - (PR #635) - -Changes in synapse v0.13.3 (2016-02-11) -======================================= - -* Fix bug where ``/sync`` would occasionally return events in the wrong room. - -Changes in synapse v0.13.2 (2016-02-11) -======================================= - -* Fix bug where ``/events`` would fail to skip some events if there had been - more events than the limit specified since the last request (PR #570) - -Changes in synapse v0.13.1 (2016-02-10) -======================================= - -* Bump matrix-angular-sdk (matrix web console) dependency to 0.6.8 to - pull in the fix for SYWEB-361 so that the default client can display - HTML messages again(!) - -Changes in synapse v0.13.0 (2016-02-10) -======================================= - -This version includes an upgrade of the schema, specifically adding an index to -the ``events`` table. This may cause synapse to pause for several minutes the -first time it is started after the upgrade. - -Changes: - -* Improve general performance (PR #540, #543. #544, #54, #549, #567) -* Change guest user ids to be incrementing integers (PR #550) -* Improve performance of public room list API (PR #552) -* Change profile API to omit keys rather than return null (PR #557) -* Add ``/media/r0`` endpoint prefix, which is equivalent to ``/media/v1/`` - (PR #595) - -Bug fixes: - -* Fix bug with upgrading guest accounts where it would fail if you opened the - registration email on a different device (PR #547) -* Fix bug where unread count could be wrong (PR #568) - - - -Changes in synapse v0.12.1-rc1 (2016-01-29) -=========================================== - -Features: - -* Add unread notification counts in ``/sync`` (PR #456) -* Add support for inviting 3pids in ``/createRoom`` (PR #460) -* Add ability for guest accounts to upgrade (PR #462) -* Add ``/versions`` API (PR #468) -* Add ``event`` to ``/context`` API (PR #492) -* Add specific error code for invalid user names in ``/register`` (PR #499) -* Add support for push badge counts (PR #507) -* Add support for non-guest users to peek in rooms using ``/events`` (PR #510) - -Changes: - -* Change ``/sync`` so that guest users only get rooms they've joined (PR #469) -* Change to require unbanning before other membership changes (PR #501) -* Change default push rules to notify for all messages (PR #486) -* Change default push rules to not notify on membership changes (PR #514) -* Change default push rules in one to one rooms to only notify for events that - are messages (PR #529) -* Change ``/sync`` to reject requests with a ``from`` query param (PR #512) -* Change server manhole to use SSH rather than telnet (PR #473) -* Change server to require AS users to be registered before use (PR #487) -* Change server not to start when ASes are invalidly configured (PR #494) -* Change server to require ID and ``as_token`` to be unique for AS's (PR #496) -* Change maximum pagination limit to 1000 (PR #497) - -Bug fixes: - -* Fix bug where ``/sync`` didn't return when something under the leave key - changed (PR #461) -* Fix bug where we returned smaller rather than larger than requested - thumbnails when ``method=crop`` (PR #464) -* Fix thumbnails API to only return cropped thumbnails when asking for a - cropped thumbnail (PR #475) -* Fix bug where we occasionally still logged access tokens (PR #477) -* Fix bug where ``/events`` would always return immediately for guest users - (PR #480) -* Fix bug where ``/sync`` unexpectedly returned old left rooms (PR #481) -* Fix enabling and disabling push rules (PR #498) -* Fix bug where ``/register`` returned 500 when given unicode username - (PR #513) - -Changes in synapse v0.12.0 (2016-01-04) -======================================= - -* Expose ``/login`` under ``r0`` (PR #459) - -Changes in synapse v0.12.0-rc3 (2015-12-23) -=========================================== - -* Allow guest accounts access to ``/sync`` (PR #455) -* Allow filters to include/exclude rooms at the room level - rather than just from the components of the sync for each - room. (PR #454) -* Include urls for room avatars in the response to ``/publicRooms`` (PR #453) -* Don't set a identicon as the avatar for a user when they register (PR #450) -* Add a ``display_name`` to third-party invites (PR #449) -* Send more information to the identity server for third-party invites so that - it can send richer messages to the invitee (PR #446) -* Cache the responses to ``/initialSync`` for 5 minutes. If a client - retries a request to ``/initialSync`` before the a response was computed - to the first request then the same response is used for both requests - (PR #457) -* Fix a bug where synapse would always request the signing keys of - remote servers even when the key was cached locally (PR #452) -* Fix 500 when pagination search results (PR #447) -* Fix a bug where synapse was leaking raw email address in third-party invites - (PR #448) - -Changes in synapse v0.12.0-rc2 (2015-12-14) -=========================================== - -* Add caches for whether rooms have been forgotten by a user (PR #434) -* Remove instructions to use ``--process-dependency-link`` since all of the - dependencies of synapse are on PyPI (PR #436) -* Parallelise the processing of ``/sync`` requests (PR #437) -* Fix race updating presence in ``/events`` (PR #444) -* Fix bug back-populating search results (PR #441) -* Fix bug calculating state in ``/sync`` requests (PR #442) - -Changes in synapse v0.12.0-rc1 (2015-12-10) -=========================================== - -* Host the client APIs released as r0 by - https://matrix.org/docs/spec/r0.0.0/client_server.html - on paths prefixed by ``/_matrix/client/r0``. (PR #430, PR #415, PR #400) -* Updates the client APIs to match r0 of the matrix specification. - - * All APIs return events in the new event format, old APIs also include - the fields needed to parse the event using the old format for - compatibility. (PR #402) - * Search results are now given as a JSON array rather than - a JSON object (PR #405) - * Miscellaneous changes to search (PR #403, PR #406, PR #412) - * Filter JSON objects may now be passed as query parameters to ``/sync`` - (PR #431) - * Fix implementation of ``/admin/whois`` (PR #418) - * Only include the rooms that user has left in ``/sync`` if the client - requests them in the filter (PR #423) - * Don't push for ``m.room.message`` by default (PR #411) - * Add API for setting per account user data (PR #392) - * Allow users to forget rooms (PR #385) - -* Performance improvements and monitoring: - - * Add per-request counters for CPU time spent on the main python thread. - (PR #421, PR #420) - * Add per-request counters for time spent in the database (PR #429) - * Make state updates in the C+S API idempotent (PR #416) - * Only fire ``user_joined_room`` if the user has actually joined. (PR #410) - * Reuse a single http client, rather than creating new ones (PR #413) - -* Fixed a bug upgrading from older versions of synapse on postgresql (PR #417) - -Changes in synapse v0.11.1 (2015-11-20) -======================================= - -* Add extra options to search API (PR #394) -* Fix bug where we did not correctly cap federation retry timers. This meant it - could take several hours for servers to start talking to ressurected servers, - even when they were receiving traffic from them (PR #393) -* Don't advertise login token flow unless CAS is enabled. This caused issues - where some clients would always use the fallback API if they did not - recognize all login flows (PR #391) -* Change /v2 sync API to rename ``private_user_data`` to ``account_data`` - (PR #386) -* Change /v2 sync API to remove the ``event_map`` and rename keys in ``rooms`` - object (PR #389) - -Changes in synapse v0.11.0-r2 (2015-11-19) -========================================== - -* Fix bug in database port script (PR #387) - -Changes in synapse v0.11.0-r1 (2015-11-18) -========================================== - -* Retry and fail federation requests more aggressively for requests that block - client side requests (PR #384) - -Changes in synapse v0.11.0 (2015-11-17) -======================================= - -* Change CAS login API (PR #349) - -Changes in synapse v0.11.0-rc2 (2015-11-13) -=========================================== - -* Various changes to /sync API response format (PR #373) -* Fix regression when setting display name in newly joined room over - federation (PR #368) -* Fix problem where /search was slow when using SQLite (PR #366) - -Changes in synapse v0.11.0-rc1 (2015-11-11) -=========================================== - -* Add Search API (PR #307, #324, #327, #336, #350, #359) -* Add 'archived' state to v2 /sync API (PR #316) -* Add ability to reject invites (PR #317) -* Add config option to disable password login (PR #322) -* Add the login fallback API (PR #330) -* Add room context API (PR #334) -* Add room tagging support (PR #335) -* Update v2 /sync API to match spec (PR #305, #316, #321, #332, #337, #341) -* Change retry schedule for application services (PR #320) -* Change retry schedule for remote servers (PR #340) -* Fix bug where we hosted static content in the incorrect place (PR #329) -* Fix bug where we didn't increment retry interval for remote servers (PR #343) - -Changes in synapse v0.10.1-rc1 (2015-10-15) -=========================================== - -* Add support for CAS, thanks to Steven Hammerton (PR #295, #296) -* Add support for using macaroons for ``access_token`` (PR #256, #229) -* Add support for ``m.room.canonical_alias`` (PR #287) -* Add support for viewing the history of rooms that they have left. (PR #276, - #294) -* Add support for refresh tokens (PR #240) -* Add flag on creation which disables federation of the room (PR #279) -* Add some room state to invites. (PR #275) -* Atomically persist events when joining a room over federation (PR #283) -* Change default history visibility for private rooms (PR #271) -* Allow users to redact their own sent events (PR #262) -* Use tox for tests (PR #247) -* Split up syutil into separate libraries (PR #243) - -Changes in synapse v0.10.0-r2 (2015-09-16) -========================================== - -* Fix bug where we always fetched remote server signing keys instead of using - ones in our cache. -* Fix adding threepids to an existing account. -* Fix bug with invinting over federation where remote server was already in - the room. (PR #281, SYN-392) - -Changes in synapse v0.10.0-r1 (2015-09-08) -========================================== - -* Fix bug with python packaging - -Changes in synapse v0.10.0 (2015-09-03) -======================================= - -No change from release candidate. - -Changes in synapse v0.10.0-rc6 (2015-09-02) -=========================================== - -* Remove some of the old database upgrade scripts. -* Fix database port script to work with newly created sqlite databases. - -Changes in synapse v0.10.0-rc5 (2015-08-27) -=========================================== - -* Fix bug that broke downloading files with ascii filenames across federation. - -Changes in synapse v0.10.0-rc4 (2015-08-27) -=========================================== - -* Allow UTF-8 filenames for upload. (PR #259) - -Changes in synapse v0.10.0-rc3 (2015-08-25) -=========================================== - -* Add ``--keys-directory`` config option to specify where files such as - certs and signing keys should be stored in, when using ``--generate-config`` - or ``--generate-keys``. (PR #250) -* Allow ``--config-path`` to specify a directory, causing synapse to use all - \*.yaml files in the directory as config files. (PR #249) -* Add ``web_client_location`` config option to specify static files to be - hosted by synapse under ``/_matrix/client``. (PR #245) -* Add helper utility to synapse to read and parse the config files and extract - the value of a given key. For example:: - - $ python -m synapse.config read server_name -c homeserver.yaml - localhost - - (PR #246) - - -Changes in synapse v0.10.0-rc2 (2015-08-24) -=========================================== - -* Fix bug where we incorrectly populated the ``event_forward_extremities`` - table, resulting in problems joining large remote rooms (e.g. - ``#matrix:matrix.org``) -* Reduce the number of times we wake up pushers by not listening for presence - or typing events, reducing the CPU cost of each pusher. - - -Changes in synapse v0.10.0-rc1 (2015-08-21) -=========================================== - -Also see v0.9.4-rc1 changelog, which has been amalgamated into this release. - -General: - -* Upgrade to Twisted 15 (PR #173) -* Add support for serving and fetching encryption keys over federation. - (PR #208) -* Add support for logging in with email address (PR #234) -* Add support for new ``m.room.canonical_alias`` event. (PR #233) -* Change synapse to treat user IDs case insensitively during registration and - login. (If two users already exist with case insensitive matching user ids, - synapse will continue to require them to specify their user ids exactly.) -* Error if a user tries to register with an email already in use. (PR #211) -* Add extra and improve existing caches (PR #212, #219, #226, #228) -* Batch various storage request (PR #226, #228) -* Fix bug where we didn't correctly log the entity that triggered the request - if the request came in via an application service (PR #230) -* Fix bug where we needlessly regenerated the full list of rooms an AS is - interested in. (PR #232) -* Add support for AS's to use v2_alpha registration API (PR #210) - - -Configuration: - -* Add ``--generate-keys`` that will generate any missing cert and key files in - the configuration files. This is equivalent to running ``--generate-config`` - on an existing configuration file. (PR #220) -* ``--generate-config`` now no longer requires a ``--server-name`` parameter - when used on existing configuration files. (PR #220) -* Add ``--print-pidfile`` flag that controls the printing of the pid to stdout - of the demonised process. (PR #213) - -Media Repository: - -* Fix bug where we picked a lower resolution image than requested. (PR #205) -* Add support for specifying if a the media repository should dynamically - thumbnail images or not. (PR #206) - -Metrics: - -* Add statistics from the reactor to the metrics API. (PR #224, #225) - -Demo Homeservers: - -* Fix starting the demo homeservers without rate-limiting enabled. (PR #182) -* Fix enabling registration on demo homeservers (PR #223) - - -Changes in synapse v0.9.4-rc1 (2015-07-21) -========================================== - -General: - -* Add basic implementation of receipts. (SPEC-99) -* Add support for configuration presets in room creation API. (PR #203) -* Add auth event that limits the visibility of history for new users. - (SPEC-134) -* Add SAML2 login/registration support. (PR #201. Thanks Muthu Subramanian!) -* Add client side key management APIs for end to end encryption. (PR #198) -* Change power level semantics so that you cannot kick, ban or change power - levels of users that have equal or greater power level than you. (SYN-192) -* Improve performance by bulk inserting events where possible. (PR #193) -* Improve performance by bulk verifying signatures where possible. (PR #194) - - -Configuration: - -* Add support for including TLS certificate chains. - -Media Repository: - -* Add Content-Disposition headers to content repository responses. (SYN-150) - - -Changes in synapse v0.9.3 (2015-07-01) -====================================== - -No changes from v0.9.3 Release Candidate 1. - -Changes in synapse v0.9.3-rc1 (2015-06-23) -========================================== - -General: - -* Fix a memory leak in the notifier. (SYN-412) -* Improve performance of room initial sync. (SYN-418) -* General improvements to logging. -* Remove ``access_token`` query params from ``INFO`` level logging. - -Configuration: - -* Add support for specifying and configuring multiple listeners. (SYN-389) - -Application services: - -* Fix bug where synapse failed to send user queries to application services. - -Changes in synapse v0.9.2-r2 (2015-06-15) -========================================= - -Fix packaging so that schema delta python files get included in the package. - -Changes in synapse v0.9.2 (2015-06-12) -====================================== - -General: - -* Use ultrajson for json (de)serialisation when a canonical encoding is not - required. Ultrajson is significantly faster than simplejson in certain - circumstances. -* Use connection pools for outgoing HTTP connections. -* Process thumbnails on separate threads. - -Configuration: - -* Add option, ``gzip_responses``, to disable HTTP response compression. - -Federation: - -* Improve resilience of backfill by ensuring we fetch any missing auth events. -* Improve performance of backfill and joining remote rooms by removing - unnecessary computations. This included handling events we'd previously - handled as well as attempting to compute the current state for outliers. - - -Changes in synapse v0.9.1 (2015-05-26) -====================================== - -General: - -* Add support for backfilling when a client paginates. This allows servers to - request history for a room from remote servers when a client tries to - paginate history the server does not have - SYN-36 -* Fix bug where you couldn't disable non-default pushrules - SYN-378 -* Fix ``register_new_user`` script - SYN-359 -* Improve performance of fetching events from the database, this improves both - initialSync and sending of events. -* Improve performance of event streams, allowing synapse to handle more - simultaneous connected clients. - -Federation: - -* Fix bug with existing backfill implementation where it returned the wrong - selection of events in some circumstances. -* Improve performance of joining remote rooms. - -Configuration: - -* Add support for changing the bind host of the metrics listener via the - ``metrics_bind_host`` option. - - -Changes in synapse v0.9.0-r5 (2015-05-21) -========================================= - -* Add more database caches to reduce amount of work done for each pusher. This - radically reduces CPU usage when multiple pushers are set up in the same room. - -Changes in synapse v0.9.0 (2015-05-07) -====================================== - -General: - -* Add support for using a PostgreSQL database instead of SQLite. See - `docs/postgres.rst`_ for details. -* Add password change and reset APIs. See `Registration`_ in the spec. -* Fix memory leak due to not releasing stale notifiers - SYN-339. -* Fix race in caches that occasionally caused some presence updates to be - dropped - SYN-369. -* Check server name has not changed on restart. -* Add a sample systemd unit file and a logger configuration in - contrib/systemd. Contributed Ivan Shapovalov. - -Federation: - -* Add key distribution mechanisms for fetching public keys of unavailable - remote home servers. See `Retrieving Server Keys`_ in the spec. - -Configuration: - -* Add support for multiple config files. -* Add support for dictionaries in config files. -* Remove support for specifying config options on the command line, except - for: - - * ``--daemonize`` - Daemonize the home server. - * ``--manhole`` - Turn on the twisted telnet manhole service on the given - port. - * ``--database-path`` - The path to a sqlite database to use. - * ``--verbose`` - The verbosity level. - * ``--log-file`` - File to log to. - * ``--log-config`` - Python logging config file. - * ``--enable-registration`` - Enable registration for new users. - -Application services: - -* Reliably retry sending of events from Synapse to application services, as per - `Application Services`_ spec. -* Application services can no longer register via the ``/register`` API, - instead their configuration should be saved to a file and listed in the - synapse ``app_service_config_files`` config option. The AS configuration file - has the same format as the old ``/register`` request. - See `docs/application_services.rst`_ for more information. - -.. _`docs/postgres.rst`: docs/postgres.rst -.. _`docs/application_services.rst`: docs/application_services.rst -.. _`Registration`: https://github.com/matrix-org/matrix-doc/blob/master/specification/10_client_server_api.rst#registration -.. _`Retrieving Server Keys`: https://github.com/matrix-org/matrix-doc/blob/6f2698/specification/30_server_server_api.rst#retrieving-server-keys -.. _`Application Services`: https://github.com/matrix-org/matrix-doc/blob/0c6bd9/specification/25_application_service_api.rst#home-server---application-service-api - -Changes in synapse v0.8.1 (2015-03-18) -====================================== - -* Disable registration by default. New users can be added using the command - ``register_new_matrix_user`` or by enabling registration in the config. -* Add metrics to synapse. To enable metrics use config options - ``enable_metrics`` and ``metrics_port``. -* Fix bug where banning only kicked the user. - -Changes in synapse v0.8.0 (2015-03-06) -====================================== - -General: - -* Add support for registration fallback. This is a page hosted on the server - which allows a user to register for an account, regardless of what client - they are using (e.g. mobile devices). - -* Added new default push rules and made them configurable by clients: - - * Suppress all notice messages. - * Notify when invited to a new room. - * Notify for messages that don't match any rule. - * Notify on incoming call. - -Federation: - -* Added per host server side rate-limiting of incoming federation requests. -* Added a ``/get_missing_events/`` API to federation to reduce number of - ``/events/`` requests. - -Configuration: - -* Added configuration option to disable registration: - ``disable_registration``. -* Added configuration option to change soft limit of number of open file - descriptors: ``soft_file_limit``. -* Make ``tls_private_key_path`` optional when running with ``no_tls``. - -Application services: - -* Application services can now poll on the CS API ``/events`` for their events, - by providing their application service ``access_token``. -* Added exclusive namespace support to application services API. - - -Changes in synapse v0.7.1 (2015-02-19) -====================================== - -* Initial alpha implementation of parts of the Application Services API. - Including: - - - AS Registration / Unregistration - - User Query API - - Room Alias Query API - - Push transport for receiving events. - - User/Alias namespace admin control - -* Add cache when fetching events from remote servers to stop repeatedly - fetching events with bad signatures. -* Respect the per remote server retry scheme when fetching both events and - server keys to reduce the number of times we send requests to dead servers. -* Inform remote servers when the local server fails to handle a received event. -* Turn off python bytecode generation due to problems experienced when - upgrading from previous versions. - -Changes in synapse v0.7.0 (2015-02-12) -====================================== - -* Add initial implementation of the query auth federation API, allowing - servers to agree on whether an event should be allowed or rejected. -* Persist events we have rejected from federation, fixing the bug where - servers would keep requesting the same events. -* Various federation performance improvements, including: - - - Add in memory caches on queries such as: - - * Computing the state of a room at a point in time, used for - authorization on federation requests. - * Fetching events from the database. - * User's room membership, used for authorizing presence updates. - - - Upgraded JSON library to improve parsing and serialisation speeds. - -* Add default avatars to new user accounts using pydenticon library. -* Correctly time out federation requests. -* Retry federation requests against different servers. -* Add support for push and push rules. -* Add alpha versions of proposed new CSv2 APIs, including ``/sync`` API. - -Changes in synapse 0.6.1 (2015-01-07) -===================================== - -* Major optimizations to improve performance of initial sync and event sending - in large rooms (by up to 10x) -* Media repository now includes a Content-Length header on media downloads. -* Improve quality of thumbnails by changing resizing algorithm. - -Changes in synapse 0.6.0 (2014-12-16) -===================================== - -* Add new API for media upload and download that supports thumbnailing. -* Replicate media uploads over multiple homeservers so media is always served - to clients from their local homeserver. This obsoletes the - --content-addr parameter and confusion over accessing content directly - from remote homeservers. -* Implement exponential backoff when retrying federation requests when - sending to remote homeservers which are offline. -* Implement typing notifications. -* Fix bugs where we sent events with invalid signatures due to bugs where - we incorrectly persisted events. -* Improve performance of database queries involving retrieving events. - -Changes in synapse 0.5.4a (2014-12-13) -====================================== - -* Fix bug while generating the error message when a file path specified in - the config doesn't exist. - -Changes in synapse 0.5.4 (2014-12-03) -===================================== - -* Fix presence bug where some rooms did not display presence updates for - remote users. -* Do not log SQL timing log lines when started with "-v" -* Fix potential memory leak. - -Changes in synapse 0.5.3c (2014-12-02) -====================================== - -* Change the default value for the `content_addr` option to use the HTTP - listener, as by default the HTTPS listener will be using a self-signed - certificate. - -Changes in synapse 0.5.3 (2014-11-27) -===================================== - -* Fix bug that caused joining a remote room to fail if a single event was not - signed correctly. -* Fix bug which caused servers to continuously try and fetch events from other - servers. - -Changes in synapse 0.5.2 (2014-11-26) -===================================== - -Fix major bug that caused rooms to disappear from peoples initial sync. - -Changes in synapse 0.5.1 (2014-11-26) -===================================== -See UPGRADES.rst for specific instructions on how to upgrade. - - * Fix bug where we served up an Event that did not match its signatures. - * Fix regression where we no longer correctly handled the case where a - homeserver receives an event for a room it doesn't recognise (but is in.) - -Changes in synapse 0.5.0 (2014-11-19) -===================================== -This release includes changes to the federation protocol and client-server API -that is not backwards compatible. - -This release also changes the internal database schemas and so requires servers to -drop their current history. See UPGRADES.rst for details. - -Homeserver: - * Add authentication and authorization to the federation protocol. Events are - now signed by their originating homeservers. - * Implement the new authorization model for rooms. - * Split out web client into a seperate repository: matrix-angular-sdk. - * Change the structure of PDUs. - * Fix bug where user could not join rooms via an alias containing 4-byte - UTF-8 characters. - * Merge concept of PDUs and Events internally. - * Improve logging by adding request ids to log lines. - * Implement a very basic room initial sync API. - * Implement the new invite/join federation APIs. - -Webclient: - * The webclient has been moved to a seperate repository. - -Changes in synapse 0.4.2 (2014-10-31) -===================================== - -Homeserver: - * Fix bugs where we did not notify users of correct presence updates. - * Fix bug where we did not handle sub second event stream timeouts. - -Webclient: - * Add ability to click on messages to see JSON. - * Add ability to redact messages. - * Add ability to view and edit all room state JSON. - * Handle incoming redactions. - * Improve feedback on errors. - * Fix bugs in mobile CSS. - * Fix bugs with desktop notifications. - -Changes in synapse 0.4.1 (2014-10-17) -===================================== -Webclient: - * Fix bug with display of timestamps. - -Changes in synpase 0.4.0 (2014-10-17) -===================================== -This release includes changes to the federation protocol and client-server API -that is not backwards compatible. - -The Matrix specification has been moved to a separate git repository: -http://github.com/matrix-org/matrix-doc - -You will also need an updated syutil and config. See UPGRADES.rst. - -Homeserver: - * Sign federation transactions to assert strong identity over federation. - * Rename timestamp keys in PDUs and events from 'ts' and 'hsob_ts' to 'origin_server_ts'. - - -Changes in synapse 0.3.4 (2014-09-25) -===================================== -This version adds support for using a TURN server. See docs/turn-howto.rst on -how to set one up. - -Homeserver: - * Add support for redaction of messages. - * Fix bug where inviting a user on a remote home server could take up to - 20-30s. - * Implement a get current room state API. - * Add support specifying and retrieving turn server configuration. - -Webclient: - * Add button to send messages to users from the home page. - * Add support for using TURN for VoIP calls. - * Show display name change messages. - * Fix bug where the client didn't get the state of a newly joined room - until after it has been refreshed. - * Fix bugs with tab complete. - * Fix bug where holding down the down arrow caused chrome to chew 100% CPU. - * Fix bug where desktop notifications occasionally used "Undefined" as the - display name. - * Fix more places where we sometimes saw room IDs incorrectly. - * Fix bug which caused lag when entering text in the text box. - -Changes in synapse 0.3.3 (2014-09-22) -===================================== - -Homeserver: - * Fix bug where you continued to get events for rooms you had left. - -Webclient: - * Add support for video calls with basic UI. - * Fix bug where one to one chats were named after your display name rather - than the other person's. - * Fix bug which caused lag when typing in the textarea. - * Refuse to run on browsers we know won't work. - * Trigger pagination when joining new rooms. - * Fix bug where we sometimes didn't display invitations in recents. - * Automatically join room when accepting a VoIP call. - * Disable outgoing and reject incoming calls on browsers we don't support - VoIP in. - * Don't display desktop notifications for messages in the room you are - non-idle and speaking in. - -Changes in synapse 0.3.2 (2014-09-18) -===================================== - -Webclient: - * Fix bug where an empty "bing words" list in old accounts didn't send - notifications when it should have done. - -Changes in synapse 0.3.1 (2014-09-18) -===================================== -This is a release to hotfix v0.3.0 to fix two regressions. - -Webclient: - * Fix a regression where we sometimes displayed duplicate events. - * Fix a regression where we didn't immediately remove rooms you were - banned in from the recents list. - -Changes in synapse 0.3.0 (2014-09-18) -===================================== -See UPGRADE for information about changes to the client server API, including -breaking backwards compatibility with VoIP calls and registration API. - -Homeserver: - * When a user changes their displayname or avatar the server will now update - all their join states to reflect this. - * The server now adds "age" key to events to indicate how old they are. This - is clock independent, so at no point does any server or webclient have to - assume their clock is in sync with everyone else. - * Fix bug where we didn't correctly pull in missing PDUs. - * Fix bug where prev_content key wasn't always returned. - * Add support for password resets. - -Webclient: - * Improve page content loading. - * Join/parts now trigger desktop notifications. - * Always show room aliases in the UI if one is present. - * No longer show user-count in the recents side panel. - * Add up & down arrow support to the text box for message sending to step - through your sent history. - * Don't display notifications for our own messages. - * Emotes are now formatted correctly in desktop notifications. - * The recents list now differentiates between public & private rooms. - * Fix bug where when switching between rooms the pagination flickered before - the view jumped to the bottom of the screen. - * Add bing word support. - -Registration API: - * The registration API has been overhauled to function like the login API. In - practice, this means registration requests must now include the following: - 'type':'m.login.password'. See UPGRADE for more information on this. - * The 'user_id' key has been renamed to 'user' to better match the login API. - * There is an additional login type: 'm.login.email.identity'. - * The command client and web client have been updated to reflect these changes. - -Changes in synapse 0.2.3 (2014-09-12) -===================================== - -Homeserver: - * Fix bug where we stopped sending events to remote home servers if a - user from that home server left, even if there were some still in the - room. - * Fix bugs in the state conflict resolution where it was incorrectly - rejecting events. - -Webclient: - * Display room names and topics. - * Allow setting/editing of room names and topics. - * Display information about rooms on the main page. - * Handle ban and kick events in real time. - * VoIP UI and reliability improvements. - * Add glare support for VoIP. - * Improvements to initial startup speed. - * Don't display duplicate join events. - * Local echo of messages. - * Differentiate sending and sent of local echo. - * Various minor bug fixes. - -Changes in synapse 0.2.2 (2014-09-06) -===================================== - -Homeserver: - * When the server returns state events it now also includes the previous - content. - * Add support for inviting people when creating a new room. - * Make the homeserver inform the room via `m.room.aliases` when a new alias - is added for a room. - * Validate `m.room.power_level` events. - -Webclient: - * Add support for captchas on registration. - * Handle `m.room.aliases` events. - * Asynchronously send messages and show a local echo. - * Inform the UI when a message failed to send. - * Only autoscroll on receiving a new message if the user was already at the - bottom of the screen. - * Add support for ban/kick reasons. - -Changes in synapse 0.2.1 (2014-09-03) -===================================== - -Homeserver: - * Added support for signing up with a third party id. - * Add synctl scripts. - * Added rate limiting. - * Add option to change the external address the content repo uses. - * Presence bug fixes. - -Webclient: - * Added support for signing up with a third party id. - * Added support for banning and kicking users. - * Added support for displaying and setting ops. - * Added support for room names. - * Fix bugs with room membership event display. - -Changes in synapse 0.2.0 (2014-09-02) -===================================== -This update changes many configuration options, updates the -database schema and mandates SSL for server-server connections. - -Homeserver: - * Require SSL for server-server connections. - * Add SSL listener for client-server connections. - * Add ability to use config files. - * Add support for kicking/banning and power levels. - * Allow setting of room names and topics on creation. - * Change presence to include last seen time of the user. - * Change url path prefix to /_matrix/... - * Bug fixes to presence. - -Webclient: - * Reskin the CSS for registration and login. - * Various improvements to rooms CSS. - * Support changes in client-server API. - * Bug fixes to VOIP UI. - * Various bug fixes to handling of changes to room member list. - -Changes in synapse 0.1.2 (2014-08-29) -===================================== - -Webclient: - * Add basic call state UI for VoIP calls. - -Changes in synapse 0.1.1 (2014-08-29) -===================================== - -Homeserver: - * Fix bug that caused the event stream to not notify some clients about - changes. - -Changes in synapse 0.1.0 (2014-08-29) -===================================== -Presence has been reenabled in this release. - -Homeserver: - * Update client to server API, including: - - Use a more consistent url scheme. - - Provide more useful information in the initial sync api. - * Change the presence handling to be much more efficient. - * Change the presence server to server API to not require explicit polling of - all users who share a room with a user. - * Fix races in the event streaming logic. - -Webclient: - * Update to use new client to server API. - * Add basic VOIP support. - * Add idle timers that change your status to away. - * Add recent rooms column when viewing a room. - * Various network efficiency improvements. - * Add basic mobile browser support. - * Add a settings page. - -Changes in synapse 0.0.1 (2014-08-22) -===================================== -Presence has been disabled in this release due to a bug that caused the -homeserver to spam other remote homeservers. - -Homeserver: - * Completely change the database schema to support generic event types. - * Improve presence reliability. - * Improve reliability of joining remote rooms. - * Fix bug where room join events were duplicated. - * Improve initial sync API to return more information to the client. - * Stop generating fake messages for room membership events. - -Webclient: - * Add tab completion of names. - * Add ability to upload and send images. - * Add profile pages. - * Improve CSS layout of room. - * Disambiguate identical display names. - * Don't get remote users display names and avatars individually. - * Use the new initial sync API to reduce number of round trips to the homeserver. - * Change url scheme to use room aliases instead of room ids where known. - * Increase longpoll timeout. - -Changes in synapse 0.0.0 (2014-08-13) -===================================== - - * Initial alpha release diff --git a/CONTRIBUTING.rst b/CONTRIBUTING.rst index 2a88647ca..6c295cfbf 100644 --- a/CONTRIBUTING.rst +++ b/CONTRIBUTING.rst @@ -30,8 +30,12 @@ use github's pull request workflow to review the contribution, and either ask you to make any refinements needed or merge it and make them ourselves. The changes will then land on master when we next do a release. -We use Jenkins for continuous integration (http://matrix.org/jenkins), and -typically all pull requests get automatically tested Jenkins: if your change breaks the build, Jenkins will yell about it in #matrix-dev:matrix.org so please lurk there and keep an eye open. +We use `Jenkins `_ and +`Travis `_ for continuous +integration. All pull requests to synapse get automatically tested by Travis; +the Jenkins builds require an adminstrator to start them. If your change +breaks the build, this will be shown in github, so please keep an eye on the +pull request for feedback. Code style ~~~~~~~~~~ @@ -44,6 +48,26 @@ Please ensure your changes match the cosmetic style of the existing project, and **never** mix cosmetic and functional changes in the same commit, as it makes it horribly hard to review otherwise. +Changelog +~~~~~~~~~ + +All changes, even minor ones, need a corresponding changelog +entry. These are managed by Towncrier +(https://github.com/hawkowl/towncrier). + +To create a changelog entry, make a new file in the ``changelog.d`` +file named in the format of ``issuenumberOrPR.type``. The type can be +one of ``feature``, ``bugfix``, ``removal`` (also used for +deprecations), or ``misc`` (for internal-only changes). The content of +the file is your changelog entry, which can contain RestructuredText +formatting. A note of contributors is welcomed in changelogs for +non-misc changes (the content of misc changes is not displayed). + +For example, a fix for a bug reported in #1234 would have its +changelog entry in ``changelog.d/1234.bugfix``, and contain content +like "The security levels of Florbs are now validated when +recieved over federation. Contributed by Jane Matrix". + Attribution ~~~~~~~~~~~ @@ -106,13 +130,17 @@ If you agree to this for your contribution, then all that's needed is to include the line in your commit or pull request comment:: Signed-off-by: Your Name - -...using your real name; unfortunately pseudonyms and anonymous contributions -can't be accepted. Git makes this trivial - just use the -s flag when you do -``git commit``, having first set ``user.name`` and ``user.email`` git configs -(which you should have done anyway :) + +We accept contributions under a legally identifiable name, such as +your name on government documentation or common-law names (names +claimed by legitimate usage or repute). Unfortunately, we cannot +accept anonymous contributions at this time. + +Git allows you to add this signoff automatically when using the ``-s`` +flag to ``git commit``, which uses the name and email set in your +``user.name`` and ``user.email`` git configs. Conclusion ~~~~~~~~~~ -That's it! Matrix is a very open and collaborative project as you might expect given our obsession with open communication. If we're going to successfully matrix together all the fragmented communication technologies out there we are reliant on contributions and collaboration from the community to do so. So please get involved - and we hope you have as much fun hacking on Matrix as we do! \ No newline at end of file +That's it! Matrix is a very open and collaborative project as you might expect given our obsession with open communication. If we're going to successfully matrix together all the fragmented communication technologies out there we are reliant on contributions and collaboration from the community to do so. So please get involved - and we hope you have as much fun hacking on Matrix as we do! diff --git a/Dockerfile b/Dockerfile new file mode 100644 index 000000000..565341fee --- /dev/null +++ b/Dockerfile @@ -0,0 +1,19 @@ +FROM docker.io/python:2-alpine3.7 + +RUN apk add --no-cache --virtual .nacl_deps su-exec build-base libffi-dev zlib-dev libressl-dev libjpeg-turbo-dev linux-headers postgresql-dev libxslt-dev + +COPY . /synapse + +# A wheel cache may be provided in ./cache for faster build +RUN cd /synapse \ + && pip install --upgrade pip setuptools psycopg2 lxml \ + && mkdir -p /synapse/cache \ + && pip install -f /synapse/cache --upgrade --process-dependency-links . \ + && mv /synapse/contrib/docker/start.py /synapse/contrib/docker/conf / \ + && rm -rf setup.py setup.cfg synapse + +VOLUME ["/data"] + +EXPOSE 8008/tcp 8448/tcp + +ENTRYPOINT ["/start.py"] diff --git a/MANIFEST.in b/MANIFEST.in index 981698143..7076b608d 100644 --- a/MANIFEST.in +++ b/MANIFEST.in @@ -2,6 +2,7 @@ include synctl include LICENSE include VERSION include *.rst +include *.md include demo/README include demo/demo.tls.dh include demo/*.py @@ -25,6 +26,12 @@ recursive-include synapse/static *.js exclude jenkins.sh exclude jenkins*.sh exclude jenkins* +exclude Dockerfile +exclude .dockerignore recursive-exclude jenkins *.sh +include pyproject.toml +recursive-include changelog.d * + +prune .github prune demo/etc diff --git a/README.rst b/README.rst index 4491b4518..84b8cf572 100644 --- a/README.rst +++ b/README.rst @@ -71,7 +71,7 @@ We'd like to invite you to join #matrix:matrix.org (via https://matrix.org/docs/projects/try-matrix-now.html), run a homeserver, take a look at the `Matrix spec `_, and experiment with the `APIs `_ and `Client SDKs -`_. +`_. Thanks for using Matrix! @@ -157,8 +157,9 @@ if you prefer. In case of problems, please see the _`Troubleshooting` section below. -Alternatively, Silvio Fricke has contributed a Dockerfile to automate the -above in Docker at https://registry.hub.docker.com/u/silviof/docker-matrix/. +There is an offical synapse image available at https://hub.docker.com/r/matrixdotorg/synapse/tags/ which can be used with the docker-compose file available at `contrib/docker`. Further information on this including configuration options is available in `contrib/docker/README.md`. + +Alternatively, Andreas Peters (previously Silvio Fricke) has contributed a Dockerfile to automate a synapse server in a single Docker image, at https://hub.docker.com/r/avhost/docker-matrix/tags/ Also, Martin Giess has created an auto-deployment process with vagrant/ansible, tested with VirtualBox/AWS/DigitalOcean - see https://github.com/EMnify/matrix-synapse-auto-deploy @@ -200,11 +201,11 @@ different. See `the spec`__ for more information on key management.) .. __: `key_management`_ The default configuration exposes two HTTP ports: 8008 and 8448. Port 8008 is -configured without TLS; it is not recommended this be exposed outside your -local network. Port 8448 is configured to use TLS with a self-signed -certificate. This is fine for testing with but, to avoid your clients -complaining about the certificate, you will almost certainly want to use -another certificate for production purposes. (Note that a self-signed +configured without TLS; it should be behind a reverse proxy for TLS/SSL +termination on port 443 which in turn should be used for clients. Port 8448 +is configured to use TLS with a self-signed certificate. If you would like +to do initial test with a client without having to setup a reverse proxy, +you can temporarly use another certificate. (Note that a self-signed certificate is fine for `Federation`_). You can do so by changing ``tls_certificate_path``, ``tls_private_key_path`` and ``tls_dh_params_path`` in ``homeserver.yaml``; alternatively, you can use a reverse-proxy, but be sure @@ -282,11 +283,17 @@ Connecting to Synapse from a client The easiest way to try out your new Synapse installation is by connecting to it from a web client. The easiest option is probably the one at -http://riot.im/app. You will need to specify a "Custom server" when you log on -or register: set this to ``https://localhost:8448`` - remember to specify the -port (``:8448``) unless you changed the configuration. (Leave the identity +https://riot.im/app. You will need to specify a "Custom server" when you log on +or register: set this to ``https://domain.tld`` if you setup a reverse proxy +following the recommended setup, or ``https://localhost:8448`` - remember to specify the +port (``:8448``) if not ``:443`` unless you changed the configuration. (Leave the identity server as the default - see `Identity servers`_.) +If using port 8448 you will run into errors until you accept the self-signed +certificate. You can easily do this by going to ``https://localhost:8448`` +directly with your browser and accept the presented certificate. You can then +go back in your web client and proceed further. + If all goes well you should at least be able to log in, create a room, and start sending messages. @@ -322,7 +329,7 @@ Security Note ============= Matrix serves raw user generated data in some APIs - specifically the `content -repository endpoints `_. +repository endpoints `_. Whilst we have tried to mitigate against possible XSS attacks (e.g. https://github.com/matrix-org/synapse/pull/1021) we recommend running @@ -341,13 +348,17 @@ Platform-Specific Instructions Debian ------ -Matrix provides official Debian packages via apt from http://matrix.org/packages/debian/. +Matrix provides official Debian packages via apt from https://matrix.org/packages/debian/. Note that these packages do not include a client - choose one from https://matrix.org/docs/projects/try-matrix-now.html (or build your own with one of our SDKs :) Fedora ------ +Synapse is in the Fedora repositories as ``matrix-synapse``:: + + sudo dnf install matrix-synapse + Oleg Girko provides Fedora RPMs at https://obs.infoserver.lv/project/monitor/matrix-synapse @@ -513,7 +524,7 @@ Troubleshooting Running ----------------------- If synapse fails with ``missing "sodium.h"`` crypto errors, you may need -to manually upgrade PyNaCL, as synapse uses NaCl (http://nacl.cr.yp.to/) for +to manually upgrade PyNaCL, as synapse uses NaCl (https://nacl.cr.yp.to/) for encryption and digital signatures. Unfortunately PyNACL currently has a few issues (https://github.com/pyca/pynacl/issues/53) and @@ -593,8 +604,9 @@ you to run your server on a machine that might not have the same name as your domain name. For example, you might want to run your server at ``synapse.example.com``, but have your Matrix user-ids look like ``@user:example.com``. (A SRV record also allows you to change the port from -the default 8448. However, if you are thinking of using a reverse-proxy, be -sure to read `Reverse-proxying the federation port`_ first.) +the default 8448. However, if you are thinking of using a reverse-proxy on the +federation port, which is not recommended, be sure to read +`Reverse-proxying the federation port`_ first.) To use a SRV record, first create your SRV record and publish it in DNS. This should have the format ``_matrix._tcp. IN SRV 10 0 @@ -603,6 +615,9 @@ should have the format ``_matrix._tcp. IN SRV 10 0 $ dig -t srv _matrix._tcp.example.com _matrix._tcp.example.com. 3600 IN SRV 10 0 8448 synapse.example.com. +Note that the server hostname cannot be an alias (CNAME record): it has to point +directly to the server hosting the synapse instance. + You can then configure your homeserver to use ```` as the domain in its user-ids, by setting ``server_name``:: @@ -625,6 +640,11 @@ largest boxes pause for thought.) Troubleshooting --------------- + +You can use the federation tester to check if your homeserver is all set: +``https://matrix.org/federationtester/api/report?server_name=`` +If any of the attributes under "checks" is false, federation won't work. + The typical failure mode with federation is that when you try to join a room, it is rejected with "401: Unauthorized". Generally this means that other servers in the room couldn't access yours. (Joining a room over federation is a @@ -652,8 +672,8 @@ useful just for development purposes. See ``_. Using PostgreSQL ================ -As of Synapse 0.9, `PostgreSQL `_ is supported as an -alternative to the `SQLite `_ database that Synapse has +As of Synapse 0.9, `PostgreSQL `_ is supported as an +alternative to the `SQLite `_ database that Synapse has traditionally used for convenience and simplicity. The advantages of Postgres include: @@ -674,10 +694,10 @@ For information on how to install and use PostgreSQL, please see Using a reverse proxy with Synapse ================================== -It is possible to put a reverse proxy such as +It is recommended to put a reverse proxy such as `nginx `_, `Apache `_ or -`HAProxy `_ in front of Synapse. One advantage of +`HAProxy `_ in front of Synapse. One advantage of doing so is that it means that you can expose the default https port (443) to Matrix clients without needing to run Synapse with root privileges. @@ -692,9 +712,9 @@ federation port has a number of pitfalls. It is possible, but be sure to read `Reverse-proxying the federation port`_. The recommended setup is therefore to configure your reverse-proxy on port 443 -for client connections, but to also expose port 8448 for server-server -connections. All the Matrix endpoints begin ``/_matrix``, so an example nginx -configuration might look like:: +to port 8008 of synapse for client connections, but to also directly expose port +8448 for server-server connections. All the Matrix endpoints begin ``/_matrix``, +so an example nginx configuration might look like:: server { listen 443 ssl; @@ -816,7 +836,9 @@ spidering 'internal' URLs on your network. At the very least we recommend that your loopback and RFC1918 IP addresses are blacklisted. This also requires the optional lxml and netaddr python dependencies to be -installed. +installed. This in turn requires the libxml2 library to be available - on +Debian/Ubuntu this means ``apt-get install libxml2-dev``, or equivalent for +your OS. Password reset @@ -876,6 +898,17 @@ This should end with a 'PASSED' result:: PASSED (successes=143) +Running the Integration Tests +============================= + +Synapse is accompanied by `SyTest `_, +a Matrix homeserver integration testing suite, which uses HTTP requests to +access the API as a Matrix client would. It is able to run Synapse directly from +the source tree, so installation of the server is not required. + +Testing with SyTest is recommended for verifying that changes related to the +Client-Server API are functioning correctly. See the `installation instructions +`_ for details. Building Internal API Documentation =================================== diff --git a/UPGRADE.rst b/UPGRADE.rst index 62b22e910..f6bb1070b 100644 --- a/UPGRADE.rst +++ b/UPGRADE.rst @@ -5,39 +5,60 @@ Before upgrading check if any special steps are required to upgrade from the what you currently have installed to current version of synapse. The extra instructions that may be required are listed later in this document. -If synapse was installed in a virtualenv then active that virtualenv before -upgrading. If synapse is installed in a virtualenv in ``~/.synapse/`` then run: +1. If synapse was installed in a virtualenv then active that virtualenv before + upgrading. If synapse is installed in a virtualenv in ``~/.synapse/`` then + run: + + .. code:: bash + + source ~/.synapse/bin/activate + +2. If synapse was installed using pip then upgrade to the latest version by + running: + + .. code:: bash + + pip install --upgrade --process-dependency-links https://github.com/matrix-org/synapse/tarball/master + + # restart synapse + synctl restart + + + If synapse was installed using git then upgrade to the latest version by + running: + + .. code:: bash + + # Pull the latest version of the master branch. + git pull + # Update the versions of synapse's python dependencies. + python synapse/python_dependencies.py | xargs pip install --upgrade + + # restart synapse + ./synctl restart + + +To check whether your update was sucessful, you can check the Server header +returned by the Client-Server API: .. code:: bash - source ~/.synapse/bin/activate + # replace with the hostname of your synapse homeserver. + # You may need to specify a port (eg, :8448) if your server is not + # configured on port 443. + curl -kv https:///_matrix/client/versions 2>&1 | grep "Server:" -If synapse was installed using pip then upgrade to the latest version by -running: +Upgrading to $NEXT_VERSION +==================== -.. code:: bash - - pip install --upgrade --process-dependency-links https://github.com/matrix-org/synapse/tarball/master - -If synapse was installed using git then upgrade to the latest version by -running: - -.. code:: bash - - # Pull the latest version of the master branch. - git pull - # Update the versions of synapse's python dependencies. - python synapse/python_dependencies.py | xargs -n1 pip install --upgrade - -To check whether your update was sucessfull, run: - -.. code:: bash - - # replace your.server.domain with ther domain of your synapse homeserver - curl https:///_matrix/federation/v1/version - -So for the Matrix.org HS server the URL would be: https://matrix.org/_matrix/federation/v1/version. +This release expands the anonymous usage stats sent if the opt-in +``report_stats`` configuration is set to ``true``. We now capture RSS memory +and cpu use at a very coarse level. This requires administrators to install +the optional ``psutil`` python module. +We would appreciate it if you could assist by ensuring this module is available +and ``report_stats`` is enabled. This will let us see if performance changes to +synapse are having an impact to the general community. Upgrading to v0.15.0 ==================== @@ -77,7 +98,7 @@ It has been replaced by specifying a list of application service registrations i ``homeserver.yaml``:: app_service_config_files: ["registration-01.yaml", "registration-02.yaml"] - + Where ``registration-01.yaml`` looks like:: url: # e.g. "https://my.application.service.com" @@ -166,7 +187,7 @@ This release completely changes the database schema and so requires upgrading it before starting the new version of the homeserver. The script "database-prepare-for-0.5.0.sh" should be used to upgrade the -database. This will save all user information, such as logins and profiles, +database. This will save all user information, such as logins and profiles, but will otherwise purge the database. This includes messages, which rooms the home server was a member of and room alias mappings. @@ -175,18 +196,18 @@ file and ask for help in #matrix:matrix.org. The upgrade process is, unfortunately, non trivial and requires human intervention to resolve any resulting conflicts during the upgrade process. -Before running the command the homeserver should be first completely +Before running the command the homeserver should be first completely shutdown. To run it, simply specify the location of the database, e.g.: ./scripts/database-prepare-for-0.5.0.sh "homeserver.db" -Once this has successfully completed it will be safe to restart the -homeserver. You may notice that the homeserver takes a few seconds longer to +Once this has successfully completed it will be safe to restart the +homeserver. You may notice that the homeserver takes a few seconds longer to restart than usual as it reinitializes the database. On startup of the new version, users can either rejoin remote rooms using room aliases or by being reinvited. Alternatively, if any other homeserver sends a -message to a room that the homeserver was previously in the local HS will +message to a room that the homeserver was previously in the local HS will automatically rejoin the room. Upgrading to v0.4.0 @@ -245,7 +266,7 @@ automatically generate default config use:: --config-path homeserver.config \ --generate-config -This config can be edited if desired, for example to specify a different SSL +This config can be edited if desired, for example to specify a different SSL certificate to use. Once done you can run the home server using:: $ python synapse/app/homeserver.py --config-path homeserver.config @@ -266,20 +287,20 @@ This release completely changes the database schema and so requires upgrading it before starting the new version of the homeserver. The script "database-prepare-for-0.0.1.sh" should be used to upgrade the -database. This will save all user information, such as logins and profiles, +database. This will save all user information, such as logins and profiles, but will otherwise purge the database. This includes messages, which rooms the home server was a member of and room alias mappings. -Before running the command the homeserver should be first completely +Before running the command the homeserver should be first completely shutdown. To run it, simply specify the location of the database, e.g.: ./scripts/database-prepare-for-0.0.1.sh "homeserver.db" -Once this has successfully completed it will be safe to restart the -homeserver. You may notice that the homeserver takes a few seconds longer to +Once this has successfully completed it will be safe to restart the +homeserver. You may notice that the homeserver takes a few seconds longer to restart than usual as it reinitializes the database. On startup of the new version, users can either rejoin remote rooms using room aliases or by being reinvited. Alternatively, if any other homeserver sends a -message to a room that the homeserver was previously in the local HS will +message to a room that the homeserver was previously in the local HS will automatically rejoin the room. diff --git a/changelog.d/.gitignore b/changelog.d/.gitignore new file mode 100644 index 000000000..f935021a8 --- /dev/null +++ b/changelog.d/.gitignore @@ -0,0 +1 @@ +!.gitignore diff --git a/changelog.d/3350.misc b/changelog.d/3350.misc new file mode 100644 index 000000000..3713cd6d6 --- /dev/null +++ b/changelog.d/3350.misc @@ -0,0 +1 @@ +Remove redundant checks on who_forgot_in_room \ No newline at end of file diff --git a/changelog.d/3367.misc b/changelog.d/3367.misc new file mode 100644 index 000000000..1f21ddea4 --- /dev/null +++ b/changelog.d/3367.misc @@ -0,0 +1 @@ +Remove unnecessary event re-signing hacks \ No newline at end of file diff --git a/tests/metrics/__init__.py b/changelog.d/3415.misc similarity index 100% rename from tests/metrics/__init__.py rename to changelog.d/3415.misc diff --git a/changelog.d/3460.misc b/changelog.d/3460.misc new file mode 100644 index 000000000..e69de29bb diff --git a/changelog.d/3514.bugfix b/changelog.d/3514.bugfix new file mode 100644 index 000000000..460fe24ac --- /dev/null +++ b/changelog.d/3514.bugfix @@ -0,0 +1 @@ +Don't generate TURN credentials if no TURN config options are set diff --git a/changelog.d/3520.bugfix b/changelog.d/3520.bugfix new file mode 100644 index 000000000..9278cb370 --- /dev/null +++ b/changelog.d/3520.bugfix @@ -0,0 +1 @@ +Correctly announce deleted devices over federation diff --git a/changelog.d/3548.bugfix b/changelog.d/3548.bugfix new file mode 100644 index 000000000..38dc3b123 --- /dev/null +++ b/changelog.d/3548.bugfix @@ -0,0 +1 @@ +Catch failures saving metrics captured by Measure, and instead log the faulty metrics information for further analysis. diff --git a/changelog.d/3552.misc b/changelog.d/3552.misc new file mode 100644 index 000000000..709c3282b --- /dev/null +++ b/changelog.d/3552.misc @@ -0,0 +1 @@ +Release notes are now in the Markdown format. diff --git a/changelog.d/3553.feature b/changelog.d/3553.feature new file mode 100644 index 000000000..77a294cb9 --- /dev/null +++ b/changelog.d/3553.feature @@ -0,0 +1 @@ +Add metrics to track resource usage by background processes diff --git a/changelog.d/3554.feature b/changelog.d/3554.feature new file mode 100644 index 000000000..b00397872 --- /dev/null +++ b/changelog.d/3554.feature @@ -0,0 +1 @@ +Add `code` label to `synapse_http_server_response_time_seconds` prometheus metric diff --git a/changelog.d/3555.feature b/changelog.d/3555.feature new file mode 100644 index 000000000..ea4a85e0a --- /dev/null +++ b/changelog.d/3555.feature @@ -0,0 +1 @@ +Add support for client_reader to handle more APIs diff --git a/changelog.d/3556.feature b/changelog.d/3556.feature new file mode 100644 index 000000000..77a294cb9 --- /dev/null +++ b/changelog.d/3556.feature @@ -0,0 +1 @@ +Add metrics to track resource usage by background processes diff --git a/changelog.d/3559.misc b/changelog.d/3559.misc new file mode 100644 index 000000000..26df859e4 --- /dev/null +++ b/changelog.d/3559.misc @@ -0,0 +1 @@ +add config for pep8 diff --git a/changelog.d/3562.misc b/changelog.d/3562.misc new file mode 100644 index 000000000..e69de29bb diff --git a/changelog.d/3570.bugfix b/changelog.d/3570.bugfix new file mode 100644 index 000000000..cec5158a9 --- /dev/null +++ b/changelog.d/3570.bugfix @@ -0,0 +1 @@ +Fix potential stack overflow and deadlock under heavy load \ No newline at end of file diff --git a/changelog.d/3571.misc b/changelog.d/3571.misc new file mode 100644 index 000000000..8908324e6 --- /dev/null +++ b/changelog.d/3571.misc @@ -0,0 +1 @@ +Merge Linearizer and Limiter diff --git a/changelog.d/3572.misc b/changelog.d/3572.misc new file mode 100644 index 000000000..8908324e6 --- /dev/null +++ b/changelog.d/3572.misc @@ -0,0 +1 @@ +Merge Linearizer and Limiter diff --git a/changelog.d/3577.misc b/changelog.d/3577.misc new file mode 100644 index 000000000..e69de29bb diff --git a/changelog.d/3579.misc b/changelog.d/3579.misc new file mode 100644 index 000000000..2374dc0c4 --- /dev/null +++ b/changelog.d/3579.misc @@ -0,0 +1 @@ +Lazily load state on master process when using workers to reduce DB consumption diff --git a/changelog.d/3581.misc b/changelog.d/3581.misc new file mode 100644 index 000000000..2374dc0c4 --- /dev/null +++ b/changelog.d/3581.misc @@ -0,0 +1 @@ +Lazily load state on master process when using workers to reduce DB consumption diff --git a/changelog.d/3582.misc b/changelog.d/3582.misc new file mode 100644 index 000000000..2374dc0c4 --- /dev/null +++ b/changelog.d/3582.misc @@ -0,0 +1 @@ +Lazily load state on master process when using workers to reduce DB consumption diff --git a/changelog.d/3584.misc b/changelog.d/3584.misc new file mode 100644 index 000000000..2374dc0c4 --- /dev/null +++ b/changelog.d/3584.misc @@ -0,0 +1 @@ +Lazily load state on master process when using workers to reduce DB consumption diff --git a/changelog.d/3586.misc b/changelog.d/3586.misc new file mode 100644 index 000000000..e853e2481 --- /dev/null +++ b/changelog.d/3586.misc @@ -0,0 +1 @@ +Fixes and optimisations for resolve_state_groups diff --git a/changelog.d/3587.misc b/changelog.d/3587.misc new file mode 100644 index 000000000..75a347991 --- /dev/null +++ b/changelog.d/3587.misc @@ -0,0 +1 @@ +Improve logging for exceptions when handling PDUs \ No newline at end of file diff --git a/changelog.d/3590.misc b/changelog.d/3590.misc new file mode 100644 index 000000000..0f1688fd0 --- /dev/null +++ b/changelog.d/3590.misc @@ -0,0 +1 @@ +Add some measure blocks to persist_events diff --git a/changelog.d/3591.misc b/changelog.d/3591.misc new file mode 100644 index 000000000..f0137766a --- /dev/null +++ b/changelog.d/3591.misc @@ -0,0 +1 @@ +Fix some random logcontext leaks. \ No newline at end of file diff --git a/changelog.d/3592.misc b/changelog.d/3592.misc new file mode 100644 index 000000000..60129569c --- /dev/null +++ b/changelog.d/3592.misc @@ -0,0 +1 @@ +Speed up calculating state deltas in persist_event loop diff --git a/changelog.d/3595.misc b/changelog.d/3595.misc new file mode 100644 index 000000000..85903504c --- /dev/null +++ b/changelog.d/3595.misc @@ -0,0 +1 @@ +Attempt to reduce amount of state pulled out of DB during persist_events diff --git a/contrib/README.rst b/contrib/README.rst new file mode 100644 index 000000000..c296c5562 --- /dev/null +++ b/contrib/README.rst @@ -0,0 +1,10 @@ +Community Contributions +======================= + +Everything in this directory are projects submitted by the community that may be useful +to others. As such, the project maintainers cannot guarantee support, stability +or backwards compatibility of these projects. + +Files in this directory should *not* be relied on directly, as they may not +continue to work or exist in future. If you wish to use any of these files then +they should be copied to avoid them breaking from underneath you. diff --git a/contrib/docker/README.md b/contrib/docker/README.md new file mode 100644 index 000000000..61592109c --- /dev/null +++ b/contrib/docker/README.md @@ -0,0 +1,153 @@ +# Synapse Docker + +The `matrixdotorg/synapse` Docker image will run Synapse as a single process. It does not provide a +database server or a TURN server, you should run these separately. + +If you run a Postgres server, you should simply include it in the same Compose +project or set the proper environment variables and the image will automatically +use that server. + +## Build + +Build the docker image with the `docker build` command from the root of the synapse repository. + +``` +docker build -t docker.io/matrixdotorg/synapse . +``` + +The `-t` option sets the image tag. Official images are tagged `matrixdotorg/synapse:` where `` is the same as the release tag in the synapse git repository. + +You may have a local Python wheel cache available, in which case copy the relevant packages in the ``cache/`` directory at the root of the project. + +## Run + +This image is designed to run either with an automatically generated configuration +file or with a custom configuration that requires manual edition. + +### Automated configuration + +It is recommended that you use Docker Compose to run your containers, including +this image and a Postgres server. A sample ``docker-compose.yml`` is provided, +including example labels for reverse proxying and other artifacts. + +Read the section about environment variables and set at least mandatory variables, +then run the server: + +``` +docker-compose up -d +``` + +If secrets are not specified in the environment variables, they will be generated +as part of the startup. Please ensure these secrets are kept between launches of the +Docker container, as their loss may require users to log in again. + +### Manual configuration + +A sample ``docker-compose.yml`` is provided, including example labels for +reverse proxying and other artifacts. The docker-compose file is an example, +please comment/uncomment sections that are not suitable for your usecase. + +Specify a ``SYNAPSE_CONFIG_PATH``, preferably to a persistent path, +to use manual configuration. To generate a fresh ``homeserver.yaml``, simply run: + +``` +docker-compose run --rm -e SYNAPSE_SERVER_NAME=my.matrix.host synapse generate +``` + +Then, customize your configuration and run the server: + +``` +docker-compose up -d +``` + +### Without Compose + +If you do not wish to use Compose, you may still run this image using plain +Docker commands. Note that the following is just a guideline and you may need +to add parameters to the docker run command to account for the network situation +with your postgres database. + +``` +docker run \ + -d \ + --name synapse \ + -v ${DATA_PATH}:/data \ + -e SYNAPSE_SERVER_NAME=my.matrix.host \ + -e SYNAPSE_REPORT_STATS=yes \ + docker.io/matrixdotorg/synapse:latest +``` + +## Volumes + +The image expects a single volume, located at ``/data``, that will hold: + +* temporary files during uploads; +* uploaded media and thumbnails; +* the SQLite database if you do not configure postgres; +* the appservices configuration. + +You are free to use separate volumes depending on storage endpoints at your +disposal. For instance, ``/data/media`` coud be stored on a large but low +performance hdd storage while other files could be stored on high performance +endpoints. + +In order to setup an application service, simply create an ``appservices`` +directory in the data volume and write the application service Yaml +configuration file there. Multiple application services are supported. + +## Environment + +Unless you specify a custom path for the configuration file, a very generic +file will be generated, based on the following environment settings. +These are a good starting point for setting up your own deployment. + +Global settings: + +* ``UID``, the user id Synapse will run as [default 991] +* ``GID``, the group id Synapse will run as [default 991] +* ``SYNAPSE_CONFIG_PATH``, path to a custom config file + +If ``SYNAPSE_CONFIG_PATH`` is set, you should generate a configuration file +then customize it manually. No other environment variable is required. + +Otherwise, a dynamic configuration file will be used. The following environment +variables are available for configuration: + +* ``SYNAPSE_SERVER_NAME`` (mandatory), the current server public hostname. +* ``SYNAPSE_REPORT_STATS``, (mandatory, ``yes`` or ``no``), enable anonymous + statistics reporting back to the Matrix project which helps us to get funding. +* ``SYNAPSE_NO_TLS``, set this variable to disable TLS in Synapse (use this if + you run your own TLS-capable reverse proxy). +* ``SYNAPSE_ENABLE_REGISTRATION``, set this variable to enable registration on + the Synapse instance. +* ``SYNAPSE_ALLOW_GUEST``, set this variable to allow guest joining this server. +* ``SYNAPSE_EVENT_CACHE_SIZE``, the event cache size [default `10K`]. +* ``SYNAPSE_CACHE_FACTOR``, the cache factor [default `0.5`]. +* ``SYNAPSE_RECAPTCHA_PUBLIC_KEY``, set this variable to the recaptcha public + key in order to enable recaptcha upon registration. +* ``SYNAPSE_RECAPTCHA_PRIVATE_KEY``, set this variable to the recaptcha private + key in order to enable recaptcha upon registration. +* ``SYNAPSE_TURN_URIS``, set this variable to the coma-separated list of TURN + uris to enable TURN for this homeserver. +* ``SYNAPSE_TURN_SECRET``, set this to the TURN shared secret if required. + +Shared secrets, that will be initialized to random values if not set: + +* ``SYNAPSE_REGISTRATION_SHARED_SECRET``, secret for registrering users if + registration is disable. +* ``SYNAPSE_MACAROON_SECRET_KEY`` secret for signing access tokens + to the server. + +Database specific values (will use SQLite if not set): + +* `POSTGRES_DB` - The database name for the synapse postgres database. [default: `synapse`] +* `POSTGRES_HOST` - The host of the postgres database if you wish to use postgresql instead of sqlite3. [default: `db` which is useful when using a container on the same docker network in a compose file where the postgres service is called `db`] +* `POSTGRES_PASSWORD` - The password for the synapse postgres database. **If this is set then postgres will be used instead of sqlite3.** [default: none] **NOTE**: You are highly encouraged to use postgresql! Please use the compose file to make it easier to deploy. +* `POSTGRES_USER` - The user for the synapse postgres database. [default: `matrix`] + +Mail server specific values (will not send emails if not set): + +* ``SYNAPSE_SMTP_HOST``, hostname to the mail server. +* ``SYNAPSE_SMTP_PORT``, TCP port for accessing the mail server [default ``25``]. +* ``SYNAPSE_SMTP_USER``, username for authenticating against the mail server if any. +* ``SYNAPSE_SMTP_PASSWORD``, password for authenticating against the mail server if any. diff --git a/contrib/docker/conf/homeserver.yaml b/contrib/docker/conf/homeserver.yaml new file mode 100644 index 000000000..6bc25bb45 --- /dev/null +++ b/contrib/docker/conf/homeserver.yaml @@ -0,0 +1,219 @@ +# vim:ft=yaml + +## TLS ## + +tls_certificate_path: "/data/{{ SYNAPSE_SERVER_NAME }}.tls.crt" +tls_private_key_path: "/data/{{ SYNAPSE_SERVER_NAME }}.tls.key" +tls_dh_params_path: "/data/{{ SYNAPSE_SERVER_NAME }}.tls.dh" +no_tls: {{ "True" if SYNAPSE_NO_TLS else "False" }} +tls_fingerprints: [] + +## Server ## + +server_name: "{{ SYNAPSE_SERVER_NAME }}" +pid_file: /homeserver.pid +web_client: False +soft_file_limit: 0 + +## Ports ## + +listeners: + {% if not SYNAPSE_NO_TLS %} + - + port: 8448 + bind_addresses: ['0.0.0.0'] + type: http + tls: true + x_forwarded: false + resources: + - names: [client] + compress: true + - names: [federation] # Federation APIs + compress: false + {% endif %} + + - port: 8008 + tls: false + bind_addresses: ['0.0.0.0'] + type: http + x_forwarded: false + + resources: + - names: [client] + compress: true + - names: [federation] + compress: false + +## Database ## + +{% if POSTGRES_PASSWORD %} +database: + name: "psycopg2" + args: + user: "{{ POSTGRES_USER or "synapse" }}" + password: "{{ POSTGRES_PASSWORD }}" + database: "{{ POSTGRES_DB or "synapse" }}" + host: "{{ POSTGRES_HOST or "db" }}" + port: "{{ POSTGRES_PORT or "5432" }}" + cp_min: 5 + cp_max: 10 +{% else %} +database: + name: "sqlite3" + args: + database: "/data/homeserver.db" +{% endif %} + +## Performance ## + +event_cache_size: "{{ SYNAPSE_EVENT_CACHE_SIZE or "10K" }}" +verbose: 0 +log_file: "/data/homeserver.log" +log_config: "/compiled/log.config" + +## Ratelimiting ## + +rc_messages_per_second: 0.2 +rc_message_burst_count: 10.0 +federation_rc_window_size: 1000 +federation_rc_sleep_limit: 10 +federation_rc_sleep_delay: 500 +federation_rc_reject_limit: 50 +federation_rc_concurrent: 3 + +## Files ## + +media_store_path: "/data/media" +uploads_path: "/data/uploads" +max_upload_size: "10M" +max_image_pixels: "32M" +dynamic_thumbnails: false + +# List of thumbnail to precalculate when an image is uploaded. +thumbnail_sizes: +- width: 32 + height: 32 + method: crop +- width: 96 + height: 96 + method: crop +- width: 320 + height: 240 + method: scale +- width: 640 + height: 480 + method: scale +- width: 800 + height: 600 + method: scale + +url_preview_enabled: False +max_spider_size: "10M" + +## Captcha ## + +{% if SYNAPSE_RECAPTCHA_PUBLIC_KEY %} +recaptcha_public_key: "{{ SYNAPSE_RECAPTCHA_PUBLIC_KEY }}" +recaptcha_private_key: "{{ SYNAPSE_RECAPTCHA_PRIVATE_KEY }}" +enable_registration_captcha: True +recaptcha_siteverify_api: "https://www.google.com/recaptcha/api/siteverify" +{% else %} +recaptcha_public_key: "YOUR_PUBLIC_KEY" +recaptcha_private_key: "YOUR_PRIVATE_KEY" +enable_registration_captcha: False +recaptcha_siteverify_api: "https://www.google.com/recaptcha/api/siteverify" +{% endif %} + +## Turn ## + +{% if SYNAPSE_TURN_URIS %} +turn_uris: +{% for uri in SYNAPSE_TURN_URIS.split(',') %} - "{{ uri }}" +{% endfor %} +turn_shared_secret: "{{ SYNAPSE_TURN_SECRET }}" +turn_user_lifetime: "1h" +turn_allow_guests: True +{% else %} +turn_uris: [] +turn_shared_secret: "YOUR_SHARED_SECRET" +turn_user_lifetime: "1h" +turn_allow_guests: True +{% endif %} + +## Registration ## + +enable_registration: {{ "True" if SYNAPSE_ENABLE_REGISTRATION else "False" }} +registration_shared_secret: "{{ SYNAPSE_REGISTRATION_SHARED_SECRET }}" +bcrypt_rounds: 12 +allow_guest_access: {{ "True" if SYNAPSE_ALLOW_GUEST else "False" }} +enable_group_creation: true + +# The list of identity servers trusted to verify third party +# identifiers by this server. +trusted_third_party_id_servers: + - matrix.org + - vector.im + - riot.im + +## Metrics ### + +{% if SYNAPSE_REPORT_STATS.lower() == "yes" %} +enable_metrics: True +report_stats: True +{% else %} +enable_metrics: False +report_stats: False +{% endif %} + +## API Configuration ## + +room_invite_state_types: + - "m.room.join_rules" + - "m.room.canonical_alias" + - "m.room.avatar" + - "m.room.name" + +{% if SYNAPSE_APPSERVICES %} +app_service_config_files: +{% for appservice in SYNAPSE_APPSERVICES %} - "{{ appservice }}" +{% endfor %} +{% else %} +app_service_config_files: [] +{% endif %} + +macaroon_secret_key: "{{ SYNAPSE_MACAROON_SECRET_KEY }}" +expire_access_token: False + +## Signing Keys ## + +signing_key_path: "/data/{{ SYNAPSE_SERVER_NAME }}.signing.key" +old_signing_keys: {} +key_refresh_interval: "1d" # 1 Day. + +# The trusted servers to download signing keys from. +perspectives: + servers: + "matrix.org": + verify_keys: + "ed25519:auto": + key: "Noi6WqcDj0QmPxCNQqgezwTlBKrfqehY1u2FyWP9uYw" + +password_config: + enabled: true + +{% if SYNAPSE_SMTP_HOST %} +email: + enable_notifs: false + smtp_host: "{{ SYNAPSE_SMTP_HOST }}" + smtp_port: {{ SYNAPSE_SMTP_PORT or "25" }} + smtp_user: "{{ SYNAPSE_SMTP_USER }}" + smtp_pass: "{{ SYNAPSE_SMTP_PASSWORD }}" + require_transport_security: False + notif_from: "{{ SYNAPSE_SMTP_FROM or "hostmaster@" + SYNAPSE_SERVER_NAME }}" + app_name: Matrix + template_dir: res/templates + notif_template_html: notif_mail.html + notif_template_text: notif_mail.txt + notif_for_new_users: True + riot_base_url: "https://{{ SYNAPSE_SERVER_NAME }}" +{% endif %} diff --git a/contrib/docker/conf/log.config b/contrib/docker/conf/log.config new file mode 100644 index 000000000..185199580 --- /dev/null +++ b/contrib/docker/conf/log.config @@ -0,0 +1,29 @@ +version: 1 + +formatters: + precise: + format: '%(asctime)s - %(name)s - %(lineno)d - %(levelname)s - %(request)s- %(message)s' + +filters: + context: + (): synapse.util.logcontext.LoggingContextFilter + request: "" + +handlers: + console: + class: logging.StreamHandler + formatter: precise + filters: [context] + +loggers: + synapse: + level: {{ SYNAPSE_LOG_LEVEL or "WARNING" }} + + synapse.storage.SQL: + # beware: increasing this to DEBUG will make synapse log sensitive + # information such as access tokens. + level: {{ SYNAPSE_LOG_LEVEL or "WARNING" }} + +root: + level: {{ SYNAPSE_LOG_LEVEL or "WARNING" }} + handlers: [console] diff --git a/contrib/docker/docker-compose.yml b/contrib/docker/docker-compose.yml new file mode 100644 index 000000000..0b531949e --- /dev/null +++ b/contrib/docker/docker-compose.yml @@ -0,0 +1,49 @@ +# This compose file is compatible with Compose itself, it might need some +# adjustments to run properly with stack. + +version: '3' + +services: + + synapse: + image: docker.io/matrixdotorg/synapse:latest + # Since snyapse does not retry to connect to the database, restart upon + # failure + restart: unless-stopped + # See the readme for a full documentation of the environment settings + environment: + - SYNAPSE_SERVER_NAME=my.matrix.host + - SYNAPSE_REPORT_STATS=no + - SYNAPSE_ENABLE_REGISTRATION=yes + - SYNAPSE_LOG_LEVEL=INFO + - POSTGRES_PASSWORD=changeme + volumes: + # You may either store all the files in a local folder + - ./files:/data + # .. or you may split this between different storage points + # - ./files:/data + # - /path/to/ssd:/data/uploads + # - /path/to/large_hdd:/data/media + depends_on: + - db + # In order to expose Synapse, remove one of the following, you might for + # instance expose the TLS port directly: + ports: + - 8448:8448/tcp + # ... or use a reverse proxy, here is an example for traefik: + labels: + - traefik.enable=true + - traefik.frontend.rule=Host:my.matrix.Host + - traefik.port=8448 + + db: + image: docker.io/postgres:10-alpine + # Change that password, of course! + environment: + - POSTGRES_USER=synapse + - POSTGRES_PASSWORD=changeme + volumes: + # You may store the database tables in a local folder.. + - ./schemas:/var/lib/postgresql/data + # .. or store them on some high performance storage for better results + # - /path/to/ssd/storage:/var/lib/postfesql/data diff --git a/contrib/docker/start.py b/contrib/docker/start.py new file mode 100755 index 000000000..90e8b9c51 --- /dev/null +++ b/contrib/docker/start.py @@ -0,0 +1,66 @@ +#!/usr/local/bin/python + +import jinja2 +import os +import sys +import subprocess +import glob + +# Utility functions +convert = lambda src, dst, environ: open(dst, "w").write(jinja2.Template(open(src).read()).render(**environ)) + +def check_arguments(environ, args): + for argument in args: + if argument not in environ: + print("Environment variable %s is mandatory, exiting." % argument) + sys.exit(2) + +def generate_secrets(environ, secrets): + for name, secret in secrets.items(): + if secret not in environ: + filename = "/data/%s.%s.key" % (environ["SYNAPSE_SERVER_NAME"], name) + if os.path.exists(filename): + with open(filename) as handle: value = handle.read() + else: + print("Generating a random secret for {}".format(name)) + value = os.urandom(32).encode("hex") + with open(filename, "w") as handle: handle.write(value) + environ[secret] = value + +# Prepare the configuration +mode = sys.argv[1] if len(sys.argv) > 1 else None +environ = os.environ.copy() +ownership = "{}:{}".format(environ.get("UID", 991), environ.get("GID", 991)) +args = ["python", "-m", "synapse.app.homeserver"] + +# In generate mode, generate a configuration, missing keys, then exit +if mode == "generate": + check_arguments(environ, ("SYNAPSE_SERVER_NAME", "SYNAPSE_REPORT_STATS", "SYNAPSE_CONFIG_PATH")) + args += [ + "--server-name", environ["SYNAPSE_SERVER_NAME"], + "--report-stats", environ["SYNAPSE_REPORT_STATS"], + "--config-path", environ["SYNAPSE_CONFIG_PATH"], + "--generate-config" + ] + os.execv("/usr/local/bin/python", args) + +# In normal mode, generate missing keys if any, then run synapse +else: + # Parse the configuration file + if "SYNAPSE_CONFIG_PATH" in environ: + args += ["--config-path", environ["SYNAPSE_CONFIG_PATH"]] + else: + check_arguments(environ, ("SYNAPSE_SERVER_NAME", "SYNAPSE_REPORT_STATS")) + generate_secrets(environ, { + "registration": "SYNAPSE_REGISTRATION_SHARED_SECRET", + "macaroon": "SYNAPSE_MACAROON_SECRET_KEY" + }) + environ["SYNAPSE_APPSERVICES"] = glob.glob("/data/appservices/*.yaml") + if not os.path.exists("/compiled"): os.mkdir("/compiled") + convert("/conf/homeserver.yaml", "/compiled/homeserver.yaml", environ) + convert("/conf/log.config", "/compiled/log.config", environ) + subprocess.check_output(["chown", "-R", ownership, "/data"]) + args += ["--config-path", "/compiled/homeserver.yaml"] + # Generate missing keys and start synapse + subprocess.check_output(args + ["--generate-keys"]) + os.execv("/sbin/su-exec", ["su-exec", ownership] + args) diff --git a/contrib/graph/graph3.py b/contrib/graph/graph3.py index 88d92c89d..7d3b4d7eb 100644 --- a/contrib/graph/graph3.py +++ b/contrib/graph/graph3.py @@ -22,6 +22,8 @@ import argparse from synapse.events import FrozenEvent from synapse.util.frozenutils import unfreeze +from six import string_types + def make_graph(file_name, room_id, file_prefix, limit): print "Reading lines" @@ -58,7 +60,7 @@ def make_graph(file_name, room_id, file_prefix, limit): for key, value in unfreeze(event.get_dict()["content"]).items(): if value is None: value = "" - elif isinstance(value, basestring): + elif isinstance(value, string_types): pass else: value = json.dumps(value) diff --git a/contrib/prometheus/README b/contrib/prometheus/README new file mode 100644 index 000000000..7b733172e --- /dev/null +++ b/contrib/prometheus/README @@ -0,0 +1,37 @@ +This directory contains some sample monitoring config for using the +'Prometheus' monitoring server against synapse. + +To use it, first install prometheus by following the instructions at + + http://prometheus.io/ + +### for Prometheus v1 +Add a new job to the main prometheus.conf file: + + job: { + name: "synapse" + + target_group: { + target: "http://SERVER.LOCATION.HERE:PORT/_synapse/metrics" + } + } + +### for Prometheus v2 +Add a new job to the main prometheus.yml file: + + - job_name: "synapse" + metrics_path: "/_synapse/metrics" + # when endpoint uses https: + scheme: "https" + + static_configs: + - targets: ['SERVER.LOCATION:PORT'] + +To use `synapse.rules` add + + rule_files: + - "/PATH/TO/synapse-v2.rules" + +Metrics are disabled by default when running synapse; they must be enabled +with the 'enable-metrics' option, either in the synapse config file or as a +command-line option. diff --git a/contrib/prometheus/consoles/synapse.html b/contrib/prometheus/consoles/synapse.html new file mode 100644 index 000000000..69aa87f85 --- /dev/null +++ b/contrib/prometheus/consoles/synapse.html @@ -0,0 +1,395 @@ +{{ template "head" . }} + +{{ template "prom_content_head" . }} +

System Resources

+ +

CPU

+
+ + +

Memory

+
+ + +

File descriptors

+
+ + +

Reactor

+ +

Total reactor time

+
+ + +

Average reactor tick time

+
+ + +

Pending calls per tick

+
+ + +

Storage

+ +

Queries

+
+ + +

Transactions

+
+ + +

Transaction execution time

+
+ + +

Database scheduling latency

+
+ + +

Cache hit ratio

+
+ + +

Cache size

+
+ + +

Requests

+ +

Requests by Servlet

+
+ +

 (without EventStreamRestServlet or SyncRestServlet)

+
+ + +

Average response times

+
+ + +

All responses by code

+
+ + +

Error responses by code

+
+ + + +

CPU Usage

+
+ + + +

DB Usage

+
+ + + +

Average event send times

+
+ + +

Federation

+ +

Sent Messages

+
+ + +

Received Messages

+
+ + +

Pending

+
+ + +

Clients

+ +

Notifiers

+
+ + +

Notified Events

+
+ + +{{ template "prom_content_tail" . }} + +{{ template "tail" }} diff --git a/contrib/prometheus/synapse-v1.rules b/contrib/prometheus/synapse-v1.rules new file mode 100644 index 000000000..4c900ba53 --- /dev/null +++ b/contrib/prometheus/synapse-v1.rules @@ -0,0 +1,21 @@ +synapse_federation_transaction_queue_pendingEdus:total = sum(synapse_federation_transaction_queue_pendingEdus or absent(synapse_federation_transaction_queue_pendingEdus)*0) +synapse_federation_transaction_queue_pendingPdus:total = sum(synapse_federation_transaction_queue_pendingPdus or absent(synapse_federation_transaction_queue_pendingPdus)*0) + +synapse_http_server_request_count:method{servlet=""} = sum(synapse_http_server_request_count) by (method) +synapse_http_server_request_count:servlet{method=""} = sum(synapse_http_server_request_count) by (servlet) + +synapse_http_server_request_count:total{servlet=""} = sum(synapse_http_server_request_count:by_method) by (servlet) + +synapse_cache:hit_ratio_5m = rate(synapse_util_caches_cache:hits[5m]) / rate(synapse_util_caches_cache:total[5m]) +synapse_cache:hit_ratio_30s = rate(synapse_util_caches_cache:hits[30s]) / rate(synapse_util_caches_cache:total[30s]) + +synapse_federation_client_sent{type="EDU"} = synapse_federation_client_sent_edus + 0 +synapse_federation_client_sent{type="PDU"} = synapse_federation_client_sent_pdu_destinations:count + 0 +synapse_federation_client_sent{type="Query"} = sum(synapse_federation_client_sent_queries) by (job) + +synapse_federation_server_received{type="EDU"} = synapse_federation_server_received_edus + 0 +synapse_federation_server_received{type="PDU"} = synapse_federation_server_received_pdus + 0 +synapse_federation_server_received{type="Query"} = sum(synapse_federation_server_received_queries) by (job) + +synapse_federation_transaction_queue_pending{type="EDU"} = synapse_federation_transaction_queue_pending_edus + 0 +synapse_federation_transaction_queue_pending{type="PDU"} = synapse_federation_transaction_queue_pending_pdus + 0 diff --git a/contrib/prometheus/synapse-v2.rules b/contrib/prometheus/synapse-v2.rules new file mode 100644 index 000000000..6ccca2daa --- /dev/null +++ b/contrib/prometheus/synapse-v2.rules @@ -0,0 +1,60 @@ +groups: +- name: synapse + rules: + - record: "synapse_federation_transaction_queue_pendingEdus:total" + expr: "sum(synapse_federation_transaction_queue_pendingEdus or absent(synapse_federation_transaction_queue_pendingEdus)*0)" + - record: "synapse_federation_transaction_queue_pendingPdus:total" + expr: "sum(synapse_federation_transaction_queue_pendingPdus or absent(synapse_federation_transaction_queue_pendingPdus)*0)" + - record: 'synapse_http_server_request_count:method' + labels: + servlet: "" + expr: "sum(synapse_http_server_request_count) by (method)" + - record: 'synapse_http_server_request_count:servlet' + labels: + method: "" + expr: 'sum(synapse_http_server_request_count) by (servlet)' + + - record: 'synapse_http_server_request_count:total' + labels: + servlet: "" + expr: 'sum(synapse_http_server_request_count:by_method) by (servlet)' + + - record: 'synapse_cache:hit_ratio_5m' + expr: 'rate(synapse_util_caches_cache:hits[5m]) / rate(synapse_util_caches_cache:total[5m])' + - record: 'synapse_cache:hit_ratio_30s' + expr: 'rate(synapse_util_caches_cache:hits[30s]) / rate(synapse_util_caches_cache:total[30s])' + + - record: 'synapse_federation_client_sent' + labels: + type: "EDU" + expr: 'synapse_federation_client_sent_edus + 0' + - record: 'synapse_federation_client_sent' + labels: + type: "PDU" + expr: 'synapse_federation_client_sent_pdu_destinations:count + 0' + - record: 'synapse_federation_client_sent' + labels: + type: "Query" + expr: 'sum(synapse_federation_client_sent_queries) by (job)' + + - record: 'synapse_federation_server_received' + labels: + type: "EDU" + expr: 'synapse_federation_server_received_edus + 0' + - record: 'synapse_federation_server_received' + labels: + type: "PDU" + expr: 'synapse_federation_server_received_pdus + 0' + - record: 'synapse_federation_server_received' + labels: + type: "Query" + expr: 'sum(synapse_federation_server_received_queries) by (job)' + + - record: 'synapse_federation_transaction_queue_pending' + labels: + type: "EDU" + expr: 'synapse_federation_transaction_queue_pending_edus + 0' + - record: 'synapse_federation_transaction_queue_pending' + labels: + type: "PDU" + expr: 'synapse_federation_transaction_queue_pending_pdus + 0' diff --git a/contrib/systemd/synapse.service b/contrib/systemd/synapse.service index 92d94b9d5..b81ce3915 100644 --- a/contrib/systemd/synapse.service +++ b/contrib/systemd/synapse.service @@ -2,6 +2,9 @@ # (e.g. https://www.archlinux.org/packages/community/any/matrix-synapse/ for ArchLinux) # rather than in a user home directory or similar under virtualenv. +# **NOTE:** This is an example service file that may change in the future. If you +# wish to use this please copy rather than symlink it. + [Unit] Description=Synapse Matrix homeserver @@ -9,9 +12,11 @@ Description=Synapse Matrix homeserver Type=simple User=synapse Group=synapse -EnvironmentFile=-/etc/sysconfig/synapse WorkingDirectory=/var/lib/synapse -ExecStart=/usr/bin/python2.7 -m synapse.app.homeserver --config-path=/etc/synapse/homeserver.yaml --log-config=/etc/synapse/log_config.yaml +ExecStart=/usr/bin/python2.7 -m synapse.app.homeserver --config-path=/etc/synapse/homeserver.yaml +ExecStop=/usr/bin/synctl stop /etc/synapse/homeserver.yaml +# EnvironmentFile=-/etc/sysconfig/synapse # Can be used to e.g. set SYNAPSE_CACHE_FACTOR [Install] WantedBy=multi-user.target + diff --git a/docs/admin_api/media_admin_api.md b/docs/admin_api/media_admin_api.md new file mode 100644 index 000000000..abdbc1ea8 --- /dev/null +++ b/docs/admin_api/media_admin_api.md @@ -0,0 +1,23 @@ +# List all media in a room + +This API gets a list of known media in a room. + +The API is: +``` +GET /_matrix/client/r0/admin/room//media +``` +including an `access_token` of a server admin. + +It returns a JSON body like the following: +``` +{ + "local": [ + "mxc://localhost/xwvutsrqponmlkjihgfedcba", + "mxc://localhost/abcdefghijklmnopqrstuvwx" + ], + "remote": [ + "mxc://matrix.org/xwvutsrqponmlkjihgfedcba", + "mxc://matrix.org/abcdefghijklmnopqrstuvwx" + ] +} +``` diff --git a/docs/admin_api/purge_history_api.rst b/docs/admin_api/purge_history_api.rst index 986efe40f..2da833c82 100644 --- a/docs/admin_api/purge_history_api.rst +++ b/docs/admin_api/purge_history_api.rst @@ -8,8 +8,56 @@ Depending on the amount of history being purged a call to the API may take several minutes or longer. During this period users will not be able to paginate further back in the room from the point being purged from. -The API is simply: +The API is: -``POST /_matrix/client/r0/admin/purge_history//`` +``POST /_matrix/client/r0/admin/purge_history/[/]`` including an ``access_token`` of a server admin. + +By default, events sent by local users are not deleted, as they may represent +the only copies of this content in existence. (Events sent by remote users are +deleted.) + +Room state data (such as joins, leaves, topic) is always preserved. + +To delete local message events as well, set ``delete_local_events`` in the body: + +.. code:: json + + { + "delete_local_events": true + } + +The caller must specify the point in the room to purge up to. This can be +specified by including an event_id in the URI, or by setting a +``purge_up_to_event_id`` or ``purge_up_to_ts`` in the request body. If an event +id is given, that event (and others at the same graph depth) will be retained. +If ``purge_up_to_ts`` is given, it should be a timestamp since the unix epoch, +in milliseconds. + +The API starts the purge running, and returns immediately with a JSON body with +a purge id: + +.. code:: json + + { + "purge_id": "" + } + +Purge status query +------------------ + +It is possible to poll for updates on recent purges with a second API; + +``GET /_matrix/client/r0/admin/purge_history_status/`` + +(again, with a suitable ``access_token``). This API returns a JSON body like +the following: + +.. code:: json + + { + "status": "active" + } + +The status will be one of ``active``, ``complete``, or ``failed``. diff --git a/docs/admin_api/register_api.rst b/docs/admin_api/register_api.rst new file mode 100644 index 000000000..209cd140f --- /dev/null +++ b/docs/admin_api/register_api.rst @@ -0,0 +1,63 @@ +Shared-Secret Registration +========================== + +This API allows for the creation of users in an administrative and +non-interactive way. This is generally used for bootstrapping a Synapse +instance with administrator accounts. + +To authenticate yourself to the server, you will need both the shared secret +(``registration_shared_secret`` in the homeserver configuration), and a +one-time nonce. If the registration shared secret is not configured, this API +is not enabled. + +To fetch the nonce, you need to request one from the API:: + + > GET /_matrix/client/r0/admin/register + + < {"nonce": "thisisanonce"} + +Once you have the nonce, you can make a ``POST`` to the same URL with a JSON +body containing the nonce, username, password, whether they are an admin +(optional, False by default), and a HMAC digest of the content. + +As an example:: + + > POST /_matrix/client/r0/admin/register + > { + "nonce": "thisisanonce", + "username": "pepper_roni", + "password": "pizza", + "admin": true, + "mac": "mac_digest_here" + } + + < { + "access_token": "token_here", + "user_id": "@pepper_roni@test", + "home_server": "test", + "device_id": "device_id_here" + } + +The MAC is the hex digest output of the HMAC-SHA1 algorithm, with the key being +the shared secret and the content being the nonce, user, password, and either +the string "admin" or "notadmin", each separated by NULs. For an example of +generation in Python:: + + import hmac, hashlib + + def generate_mac(nonce, user, password, admin=False): + + mac = hmac.new( + key=shared_secret, + digestmod=hashlib.sha1, + ) + + mac.update(nonce.encode('utf8')) + mac.update(b"\x00") + mac.update(user.encode('utf8')) + mac.update(b"\x00") + mac.update(password.encode('utf8')) + mac.update(b"\x00") + mac.update(b"admin" if admin else b"notadmin") + + return mac.hexdigest() diff --git a/docs/admin_api/user_admin_api.rst b/docs/admin_api/user_admin_api.rst index 1c9c5a6bd..d17121a18 100644 --- a/docs/admin_api/user_admin_api.rst +++ b/docs/admin_api/user_admin_api.rst @@ -44,13 +44,26 @@ Deactivate Account This API deactivates an account. It removes active access tokens, resets the password, and deletes third-party IDs (to prevent the user requesting a -password reset). +password reset). It can also mark the user as GDPR-erased (stopping their data +from distributed further, and deleting it entirely if there are no other +references to it). The api is:: POST /_matrix/client/r0/admin/deactivate/ -including an ``access_token`` of a server admin, and an empty request body. +with a body of: + +.. code:: json + + { + "erase": true + } + +including an ``access_token`` of a server admin. + +The erase parameter is optional and defaults to 'false'. +An empty body may be passed for backwards compatibility. Reset password diff --git a/docs/code_style.rst b/docs/code_style.rst index 8d73d17be..62800b5b3 100644 --- a/docs/code_style.rst +++ b/docs/code_style.rst @@ -1,52 +1,119 @@ -Basically, PEP8 +- Everything should comply with PEP8. Code should pass + ``pep8 --max-line-length=100`` without any warnings. -- NEVER tabs. 4 spaces to indent. -- Max line width: 79 chars (with flexibility to overflow by a "few chars" if +- **Indenting**: + + - NEVER tabs. 4 spaces to indent. + + - follow PEP8; either hanging indent or multiline-visual indent depending + on the size and shape of the arguments and what makes more sense to the + author. In other words, both this:: + + print("I am a fish %s" % "moo") + + and this:: + + print("I am a fish %s" % + "moo") + + and this:: + + print( + "I am a fish %s" % + "moo", + ) + + ...are valid, although given each one takes up 2x more vertical space than + the previous, it's up to the author's discretion as to which layout makes + most sense for their function invocation. (e.g. if they want to add + comments per-argument, or put expressions in the arguments, or group + related arguments together, or want to deliberately extend or preserve + vertical/horizontal space) + +- **Line length**: + + Max line length is 79 chars (with flexibility to overflow by a "few chars" if the overflowing content is not semantically significant and avoids an explosion of vertical whitespace). -- Use camel case for class and type names -- Use underscores for functions and variables. -- Use double quotes. -- Use parentheses instead of '\\' for line continuation where ever possible - (which is pretty much everywhere) -- There should be max a single new line between: + + Use parentheses instead of ``\`` for line continuation where ever possible + (which is pretty much everywhere). + +- **Naming**: + + - Use camel case for class and type names + - Use underscores for functions and variables. + +- Use double quotes ``"foo"`` rather than single quotes ``'foo'``. + +- **Blank lines**: + + - There should be max a single new line between: + - statements - functions in a class -- There should be two new lines between: + + - There should be two new lines between: + - definitions in a module (e.g., between different classes) -- There should be spaces where spaces should be and not where there shouldn't be: - - a single space after a comma - - a single space before and after for '=' when used as assignment - - no spaces before and after for '=' for default values and keyword arguments. -- Indenting must follow PEP8; either hanging indent or multiline-visual indent - depending on the size and shape of the arguments and what makes more sense to - the author. In other words, both this:: - print("I am a fish %s" % "moo") +- **Whitespace**: - and this:: + There should be spaces where spaces should be and not where there shouldn't + be: - print("I am a fish %s" % - "moo") + - a single space after a comma + - a single space before and after for '=' when used as assignment + - no spaces before and after for '=' for default values and keyword arguments. - and this:: +- **Comments**: should follow the `google code style + `_. + This is so that we can generate documentation with `sphinx + `_. See the + `examples + `_ + in the sphinx documentation. - print( - "I am a fish %s" % - "moo" - ) +- **Imports**: - ...are valid, although given each one takes up 2x more vertical space than - the previous, it's up to the author's discretion as to which layout makes most - sense for their function invocation. (e.g. if they want to add comments - per-argument, or put expressions in the arguments, or group related arguments - together, or want to deliberately extend or preserve vertical/horizontal - space) + - Prefer to import classes and functions than packages or modules. -Comments should follow the `google code style `_. -This is so that we can generate documentation with -`sphinx `_. See the -`examples `_ -in the sphinx documentation. + Example:: -Code should pass pep8 --max-line-length=100 without any warnings. + from synapse.types import UserID + ... + user_id = UserID(local, server) + + is preferred over:: + + from synapse import types + ... + user_id = types.UserID(local, server) + + (or any other variant). + + This goes against the advice in the Google style guide, but it means that + errors in the name are caught early (at import time). + + - Multiple imports from the same package can be combined onto one line:: + + from synapse.types import GroupID, RoomID, UserID + + An effort should be made to keep the individual imports in alphabetical + order. + + If the list becomes long, wrap it with parentheses and split it over + multiple lines. + + - As per `PEP-8 `_, + imports should be grouped in the following order, with a blank line between + each group: + + 1. standard library imports + 2. related third party imports + 3. local application/library specific imports + + - Imports within each group should be sorted alphabetically by module name. + + - Avoid wildcard imports (``from synapse.types import *``) and relative + imports (``from .types import UserID``). diff --git a/docs/consent_tracking.md b/docs/consent_tracking.md new file mode 100644 index 000000000..064eae82f --- /dev/null +++ b/docs/consent_tracking.md @@ -0,0 +1,160 @@ +Support in Synapse for tracking agreement to server terms and conditions +======================================================================== + +Synapse 0.30 introduces support for tracking whether users have agreed to the +terms and conditions set by the administrator of a server - and blocking access +to the server until they have. + +There are several parts to this functionality; each requires some specific +configuration in `homeserver.yaml` to be enabled. + +Note that various parts of the configuation and this document refer to the +"privacy policy": agreement with a privacy policy is one particular use of this +feature, but of course adminstrators can specify other terms and conditions +unrelated to "privacy" per se. + +Collecting policy agreement from a user +--------------------------------------- + +Synapse can be configured to serve the user a simple policy form with an +"accept" button. Clicking "Accept" records the user's acceptance in the +database and shows a success page. + +To enable this, first create templates for the policy and success pages. +These should be stored on the local filesystem. + +These templates use the [Jinja2](http://jinja.pocoo.org) templating language, +and [docs/privacy_policy_templates](privacy_policy_templates) gives +examples of the sort of thing that can be done. + +Note that the templates must be stored under a name giving the language of the +template - currently this must always be `en` (for "English"); +internationalisation support is intended for the future. + +The template for the policy itself should be versioned and named according to +the version: for example `1.0.html`. The version of the policy which the user +has agreed to is stored in the database. + +Once the templates are in place, make the following changes to `homeserver.yaml`: + + 1. Add a `user_consent` section, which should look like: + + ```yaml + user_consent: + template_dir: privacy_policy_templates + version: 1.0 + ``` + + `template_dir` points to the directory containing the policy + templates. `version` defines the version of the policy which will be served + to the user. In the example above, Synapse will serve + `privacy_policy_templates/en/1.0.html`. + + + 2. Add a `form_secret` setting at the top level: + + + ```yaml + form_secret: "" + ``` + + This should be set to an arbitrary secret string (try `pwgen -y 30` to + generate suitable secrets). + + More on what this is used for below. + + 3. Add `consent` wherever the `client` resource is currently enabled in the + `listeners` configuration. For example: + + ```yaml + listeners: + - port: 8008 + resources: + - names: + - client + - consent + ``` + + +Finally, ensure that `jinja2` is installed. If you are using a virtualenv, this +should be a matter of `pip install Jinja2`. On debian, try `apt-get install +python-jinja2`. + +Once this is complete, and the server has been restarted, try visiting +`https:///_matrix/consent`. If correctly configured, this should give +an error "Missing string query parameter 'u'". It is now possible to manually +construct URIs where users can give their consent. + +### Constructing the consent URI + +It may be useful to manually construct the "consent URI" for a given user - for +instance, in order to send them an email asking them to consent. To do this, +take the base `https:///_matrix/consent` URL and add the following +query parameters: + + * `u`: the user id of the user. This can either be a full MXID + (`@user:server.com`) or just the localpart (`user`). + + * `h`: hex-encoded HMAC-SHA256 of `u` using the `form_secret` as a key. It is + possible to calculate this on the commandline with something like: + + ```bash + echo -n '' | openssl sha256 -hmac '' + ``` + + This should result in a URI which looks something like: + `https:///_matrix/consent?u=&h=68a152465a4d...`. + + +Sending users a server notice asking them to agree to the policy +---------------------------------------------------------------- + +It is possible to configure Synapse to send a [server +notice](server_notices.md) to anybody who has not yet agreed to the current +version of the policy. To do so: + + * ensure that the consent resource is configured, as in the previous section + + * ensure that server notices are configured, as in [server_notices.md](server_notices.md). + + * Add `server_notice_content` under `user_consent` in `homeserver.yaml`. For + example: + + ```yaml + user_consent: + server_notice_content: + msgtype: m.text + body: >- + Please give your consent to the privacy policy at %(consent_uri)s. + ``` + + Synapse automatically replaces the placeholder `%(consent_uri)s` with the + consent uri for that user. + + * ensure that `public_baseurl` is set in `homeserver.yaml`, and gives the base + URI that clients use to connect to the server. (It is used to construct + `consent_uri` in the server notice.) + + +Blocking users from using the server until they agree to the policy +------------------------------------------------------------------- + +Synapse can be configured to block any attempts to join rooms or send messages +until the user has given their agreement to the policy. (Joining the server +notices room is exempted from this). + +To enable this, add `block_events_error` under `user_consent`. For example: + +```yaml +user_consent: + block_events_error: >- + You can't send any messages until you consent to the privacy policy at + %(consent_uri)s. +``` + +Synapse automatically replaces the placeholder `%(consent_uri)s` with the +consent uri for that user. + +ensure that `public_baseurl` is set in `homeserver.yaml`, and gives the base +URI that clients use to connect to the server. (It is used to construct +`consent_uri` in the error.) diff --git a/docs/log_contexts.rst b/docs/log_contexts.rst index eb1784e70..82ac4f91e 100644 --- a/docs/log_contexts.rst +++ b/docs/log_contexts.rst @@ -279,9 +279,9 @@ Obviously that option means that the operations done in that might be fixed by setting a different logcontext via a ``with LoggingContext(...)`` in ``background_operation``). -The second option is to use ``logcontext.preserve_fn``, which wraps a function -so that it doesn't reset the logcontext even when it returns an incomplete -deferred, and adds a callback to the returned deferred to reset the +The second option is to use ``logcontext.run_in_background``, which wraps a +function so that it doesn't reset the logcontext even when it returns an +incomplete deferred, and adds a callback to the returned deferred to reset the logcontext. In other words, it turns a function that follows the Synapse rules about logcontexts and Deferreds into one which behaves more like an external function — the opposite operation to that described in the previous section. @@ -293,15 +293,11 @@ It can be used like this: def do_request_handling(): yield foreground_operation() - logcontext.preserve_fn(background_operation)() + logcontext.run_in_background(background_operation) # this will now be logged against the request context logger.debug("Request handling complete") -XXX: I think ``preserve_context_over_fn`` is supposed to do the first option, -but the fact that it does ``preserve_context_over_deferred`` on its results -means that its use is fraught with difficulty. - Passing synapse deferreds into third-party functions ---------------------------------------------------- diff --git a/docs/manhole.md b/docs/manhole.md new file mode 100644 index 000000000..7375f5ad4 --- /dev/null +++ b/docs/manhole.md @@ -0,0 +1,43 @@ +Using the synapse manhole +========================= + +The "manhole" allows server administrators to access a Python shell on a running +Synapse installation. This is a very powerful mechanism for administration and +debugging. + +To enable it, first uncomment the `manhole` listener configuration in +`homeserver.yaml`: + +```yaml +listeners: + - port: 9000 + bind_addresses: ['::1', '127.0.0.1'] + type: manhole +``` + +(`bind_addresses` in the above is important: it ensures that access to the +manhole is only possible for local users). + +Note that this will give administrative access to synapse to **all users** with +shell access to the server. It should therefore **not** be enabled in +environments where untrusted users have shell access. + +Then restart synapse, and point an ssh client at port 9000 on localhost, using +the username `matrix`: + +```bash +ssh -p9000 matrix@localhost +``` + +The password is `rabbithole`. + +This gives a Python REPL in which `hs` gives access to the +`synapse.server.HomeServer` object - which in turn gives access to many other +parts of the process. + +As a simple example, retrieving an event from the database: + +``` +>>> hs.get_datastore().get_event('$1416420717069yeQaw:matrix.org') +> +``` diff --git a/docs/metrics-howto.rst b/docs/metrics-howto.rst index 143cd0f42..5bbb5a4f3 100644 --- a/docs/metrics-howto.rst +++ b/docs/metrics-howto.rst @@ -1,25 +1,47 @@ How to monitor Synapse metrics using Prometheus =============================================== -1. Install prometheus: +1. Install Prometheus: Follow instructions at http://prometheus.io/docs/introduction/install/ -2. Enable synapse metrics: +2. Enable Synapse metrics: - Simply setting a (local) port number will enable it. Pick a port. - prometheus itself defaults to 9090, so starting just above that for - locally monitored services seems reasonable. E.g. 9092: + There are two methods of enabling metrics in Synapse. - Add to homeserver.yaml:: + The first serves the metrics as a part of the usual web server and can be + enabled by adding the "metrics" resource to the existing listener as such:: - metrics_port: 9092 + resources: + - names: + - client + - metrics - Also ensure that ``enable_metrics`` is set to ``True``. - - Restart synapse. + This provides a simple way of adding metrics to your Synapse installation, + and serves under ``/_synapse/metrics``. If you do not wish your metrics be + publicly exposed, you will need to either filter it out at your load + balancer, or use the second method. -3. Add a prometheus target for synapse. + The second method runs the metrics server on a different port, in a + different thread to Synapse. This can make it more resilient to heavy load + meaning metrics cannot be retrieved, and can be exposed to just internal + networks easier. The served metrics are available over HTTP only, and will + be available at ``/``. + + Add a new listener to homeserver.yaml:: + + listeners: + - type: metrics + port: 9000 + bind_addresses: + - '0.0.0.0' + + For both options, you will need to ensure that ``enable_metrics`` is set to + ``True``. + + Restart Synapse. + +3. Add a Prometheus target for Synapse. It needs to set the ``metrics_path`` to a non-default value (under ``scrape_configs``):: @@ -28,10 +50,100 @@ How to monitor Synapse metrics using Prometheus static_configs: - targets: ["my.server.here:9092"] - If your prometheus is older than 1.5.2, you will need to replace + If your prometheus is older than 1.5.2, you will need to replace ``static_configs`` in the above with ``target_groups``. - - Restart prometheus. + + Restart Prometheus. + + +Removal of deprecated metrics & time based counters becoming histograms in 0.31.0 +--------------------------------------------------------------------------------- + +The duplicated metrics deprecated in Synapse 0.27.0 have been removed. + +All time duration-based metrics have been changed to be seconds. This affects: + ++----------------------------------+ +| msec -> sec metrics | ++==================================+ +| python_gc_time | ++----------------------------------+ +| python_twisted_reactor_tick_time | ++----------------------------------+ +| synapse_storage_query_time | ++----------------------------------+ +| synapse_storage_schedule_time | ++----------------------------------+ +| synapse_storage_transaction_time | ++----------------------------------+ + +Several metrics have been changed to be histograms, which sort entries into +buckets and allow better analysis. The following metrics are now histograms: + ++-------------------------------------------+ +| Altered metrics | ++===========================================+ +| python_gc_time | ++-------------------------------------------+ +| python_twisted_reactor_pending_calls | ++-------------------------------------------+ +| python_twisted_reactor_tick_time | ++-------------------------------------------+ +| synapse_http_server_response_time_seconds | ++-------------------------------------------+ +| synapse_storage_query_time | ++-------------------------------------------+ +| synapse_storage_schedule_time | ++-------------------------------------------+ +| synapse_storage_transaction_time | ++-------------------------------------------+ + + +Block and response metrics renamed for 0.27.0 +--------------------------------------------- + +Synapse 0.27.0 begins the process of rationalising the duplicate ``*:count`` +metrics reported for the resource tracking for code blocks and HTTP requests. + +At the same time, the corresponding ``*:total`` metrics are being renamed, as +the ``:total`` suffix no longer makes sense in the absence of a corresponding +``:count`` metric. + +To enable a graceful migration path, this release just adds new names for the +metrics being renamed. A future release will remove the old ones. + +The following table shows the new metrics, and the old metrics which they are +replacing. + +==================================================== =================================================== +New name Old name +==================================================== =================================================== +synapse_util_metrics_block_count synapse_util_metrics_block_timer:count +synapse_util_metrics_block_count synapse_util_metrics_block_ru_utime:count +synapse_util_metrics_block_count synapse_util_metrics_block_ru_stime:count +synapse_util_metrics_block_count synapse_util_metrics_block_db_txn_count:count +synapse_util_metrics_block_count synapse_util_metrics_block_db_txn_duration:count + +synapse_util_metrics_block_time_seconds synapse_util_metrics_block_timer:total +synapse_util_metrics_block_ru_utime_seconds synapse_util_metrics_block_ru_utime:total +synapse_util_metrics_block_ru_stime_seconds synapse_util_metrics_block_ru_stime:total +synapse_util_metrics_block_db_txn_count synapse_util_metrics_block_db_txn_count:total +synapse_util_metrics_block_db_txn_duration_seconds synapse_util_metrics_block_db_txn_duration:total + +synapse_http_server_response_count synapse_http_server_requests +synapse_http_server_response_count synapse_http_server_response_time:count +synapse_http_server_response_count synapse_http_server_response_ru_utime:count +synapse_http_server_response_count synapse_http_server_response_ru_stime:count +synapse_http_server_response_count synapse_http_server_response_db_txn_count:count +synapse_http_server_response_count synapse_http_server_response_db_txn_duration:count + +synapse_http_server_response_time_seconds synapse_http_server_response_time:total +synapse_http_server_response_ru_utime_seconds synapse_http_server_response_ru_utime:total +synapse_http_server_response_ru_stime_seconds synapse_http_server_response_ru_stime:total +synapse_http_server_response_db_txn_count synapse_http_server_response_db_txn_count:total +synapse_http_server_response_db_txn_duration_seconds synapse_http_server_response_db_txn_duration:total +==================================================== =================================================== + Standard Metric Names --------------------- @@ -42,7 +154,7 @@ have been changed to seconds, from miliseconds. ================================== ============================= New name Old name ----------------------------------- ----------------------------- +================================== ============================= process_cpu_user_seconds_total process_resource_utime / 1000 process_cpu_system_seconds_total process_resource_stime / 1000 process_open_fds (no 'type' label) process_fds @@ -52,8 +164,8 @@ The python-specific counts of garbage collector performance have been renamed. =========================== ====================== New name Old name ---------------------------- ---------------------- -python_gc_time reactor_gc_time +=========================== ====================== +python_gc_time reactor_gc_time python_gc_unreachable_total reactor_gc_unreachable python_gc_counts reactor_gc_counts =========================== ====================== @@ -62,7 +174,7 @@ The twisted-specific reactor metrics have been renamed. ==================================== ===================== New name Old name ------------------------------------- --------------------- +==================================== ===================== python_twisted_reactor_pending_calls reactor_pending_calls python_twisted_reactor_tick_time reactor_tick_time ==================================== ===================== diff --git a/docs/password_auth_providers.rst b/docs/password_auth_providers.rst new file mode 100644 index 000000000..d8a7b61cd --- /dev/null +++ b/docs/password_auth_providers.rst @@ -0,0 +1,99 @@ +Password auth provider modules +============================== + +Password auth providers offer a way for server administrators to integrate +their Synapse installation with an existing authentication system. + +A password auth provider is a Python class which is dynamically loaded into +Synapse, and provides a number of methods by which it can integrate with the +authentication system. + +This document serves as a reference for those looking to implement their own +password auth providers. + +Required methods +---------------- + +Password auth provider classes must provide the following methods: + +*class* ``SomeProvider.parse_config``\(*config*) + + This method is passed the ``config`` object for this module from the + homeserver configuration file. + + It should perform any appropriate sanity checks on the provided + configuration, and return an object which is then passed into ``__init__``. + +*class* ``SomeProvider``\(*config*, *account_handler*) + + The constructor is passed the config object returned by ``parse_config``, + and a ``synapse.module_api.ModuleApi`` object which allows the + password provider to check if accounts exist and/or create new ones. + +Optional methods +---------------- + +Password auth provider classes may optionally provide the following methods. + +*class* ``SomeProvider.get_db_schema_files``\() + + This method, if implemented, should return an Iterable of ``(name, + stream)`` pairs of database schema files. Each file is applied in turn at + initialisation, and a record is then made in the database so that it is + not re-applied on the next start. + +``someprovider.get_supported_login_types``\() + + This method, if implemented, should return a ``dict`` mapping from a login + type identifier (such as ``m.login.password``) to an iterable giving the + fields which must be provided by the user in the submission to the + ``/login`` api. These fields are passed in the ``login_dict`` dictionary + to ``check_auth``. + + For example, if a password auth provider wants to implement a custom login + type of ``com.example.custom_login``, where the client is expected to pass + the fields ``secret1`` and ``secret2``, the provider should implement this + method and return the following dict:: + + {"com.example.custom_login": ("secret1", "secret2")} + +``someprovider.check_auth``\(*username*, *login_type*, *login_dict*) + + This method is the one that does the real work. If implemented, it will be + called for each login attempt where the login type matches one of the keys + returned by ``get_supported_login_types``. + + It is passed the (possibly UNqualified) ``user`` provided by the client, + the login type, and a dictionary of login secrets passed by the client. + + The method should return a Twisted ``Deferred`` object, which resolves to + the canonical ``@localpart:domain`` user id if authentication is successful, + and ``None`` if not. + + Alternatively, the ``Deferred`` can resolve to a ``(str, func)`` tuple, in + which case the second field is a callback which will be called with the + result from the ``/login`` call (including ``access_token``, ``device_id``, + etc.) + +``someprovider.check_password``\(*user_id*, *password*) + + This method provides a simpler interface than ``get_supported_login_types`` + and ``check_auth`` for password auth providers that just want to provide a + mechanism for validating ``m.login.password`` logins. + + Iif implemented, it will be called to check logins with an + ``m.login.password`` login type. It is passed a qualified + ``@localpart:domain`` user id, and the password provided by the user. + + The method should return a Twisted ``Deferred`` object, which resolves to + ``True`` if authentication is successful, and ``False`` if not. + +``someprovider.on_logged_out``\(*user_id*, *device_id*, *access_token*) + + This method, if implemented, is called when a user logs out. It is passed + the qualified user ID, the ID of the deactivated device (if any: access + tokens are occasionally created without an associated device ID), and the + (now deactivated) access token. + + It may return a Twisted ``Deferred`` object; the logout request will wait + for the deferred to complete but the result is ignored. diff --git a/docs/postgres.rst b/docs/postgres.rst index b592801e9..237754229 100644 --- a/docs/postgres.rst +++ b/docs/postgres.rst @@ -1,19 +1,27 @@ Using Postgres -------------- +Postgres version 9.4 or later is known to work. + Set up database =============== -The PostgreSQL database used *must* have the correct encoding set, otherwise +Assuming your PostgreSQL database user is called ``postgres``, create a user +``synapse_user`` with:: + + su - postgres + createuser --pwprompt synapse_user + +The PostgreSQL database used *must* have the correct encoding set, otherwise it would not be able to store UTF8 strings. To create a database with the correct encoding use, e.g.:: - CREATE DATABASE synapse - ENCODING 'UTF8' - LC_COLLATE='C' - LC_CTYPE='C' - template=template0 - OWNER synapse_user; + CREATE DATABASE synapse + ENCODING 'UTF8' + LC_COLLATE='C' + LC_CTYPE='C' + template=template0 + OWNER synapse_user; This would create an appropriate database named ``synapse`` owned by the ``synapse_user`` user (which must already exist). @@ -44,8 +52,8 @@ As with Debian/Ubuntu, postgres support depends on the postgres python connector Synapse config ============== -When you are ready to start using PostgreSQL, add the following line to your -config file:: +When you are ready to start using PostgreSQL, edit the ``database`` section in +your config file to match the following lines:: database: name: psycopg2 @@ -94,9 +102,12 @@ complete, restart synapse. For instance:: cp homeserver.db homeserver.db.snapshot ./synctl start -Assuming your new config file (as described in the section *Synapse config*) -is named ``homeserver-postgres.yaml`` and the SQLite snapshot is at -``homeserver.db.snapshot`` then simply run:: +Copy the old config file into a new config file:: + + cp homeserver.yaml homeserver-postgres.yaml + +Edit the database section as described in the section *Synapse config* above +and with the SQLite snapshot located at ``homeserver.db.snapshot`` simply run:: synapse_port_db --sqlite-database homeserver.db.snapshot \ --postgres-config homeserver-postgres.yaml @@ -115,6 +126,11 @@ run:: --postgres-config homeserver-postgres.yaml Once that has completed, change the synapse config to point at the PostgreSQL -database configuration file ``homeserver-postgres.yaml`` (i.e. rename it to -``homeserver.yaml``) and restart synapse. Synapse should now be running against -PostgreSQL. +database configuration file ``homeserver-postgres.yaml``:: + + ./synctl stop + mv homeserver.yaml homeserver-old-sqlite.yaml + mv homeserver-postgres.yaml homeserver.yaml + ./synctl start + +Synapse should now be running against PostgreSQL. diff --git a/docs/privacy_policy_templates/en/1.0.html b/docs/privacy_policy_templates/en/1.0.html new file mode 100644 index 000000000..55c5e4b61 --- /dev/null +++ b/docs/privacy_policy_templates/en/1.0.html @@ -0,0 +1,23 @@ + + + + Matrix.org Privacy policy + + + {% if has_consented %} +

+ Your base already belong to us. +

+ {% else %} +

+ All your base are belong to us. +

+
+ + + + +
+ {% endif %} + + diff --git a/docs/privacy_policy_templates/en/success.html b/docs/privacy_policy_templates/en/success.html new file mode 100644 index 000000000..d55e90c94 --- /dev/null +++ b/docs/privacy_policy_templates/en/success.html @@ -0,0 +1,11 @@ + + + + Matrix.org Privacy policy + + +

+ Sweet. +

+ + diff --git a/docs/server_notices.md b/docs/server_notices.md new file mode 100644 index 000000000..58f877631 --- /dev/null +++ b/docs/server_notices.md @@ -0,0 +1,74 @@ +Server Notices +============== + +'Server Notices' are a new feature introduced in Synapse 0.30. They provide a +channel whereby server administrators can send messages to users on the server. + +They are used as part of communication of the server polices(see +[consent_tracking.md](consent_tracking.md)), however the intention is that +they may also find a use for features such as "Message of the day". + +This is a feature specific to Synapse, but it uses standard Matrix +communication mechanisms, so should work with any Matrix client. + +User experience +--------------- + +When the user is first sent a server notice, they will get an invitation to a +room (typically called 'Server Notices', though this is configurable in +`homeserver.yaml`). They will be **unable to reject** this invitation - +attempts to do so will receive an error. + +Once they accept the invitation, they will see the notice message in the room +history; it will appear to have come from the 'server notices user' (see +below). + +The user is prevented from sending any messages in this room by the power +levels. + +Having joined the room, the user can leave the room if they want. Subsequent +server notices will then cause a new room to be created. + +Synapse configuration +--------------------- + +Server notices come from a specific user id on the server. Server +administrators are free to choose the user id - something like `server` is +suggested, meaning the notices will come from +`@server:`. Once the Server Notices user is configured, that +user id becomes a special, privileged user, so administrators should ensure +that **it is not already allocated**. + +In order to support server notices, it is necessary to add some configuration +to the `homeserver.yaml` file. In particular, you should add a `server_notices` +section, which should look like this: + +```yaml +server_notices: + system_mxid_localpart: server + system_mxid_display_name: "Server Notices" + system_mxid_avatar_url: "mxc://server.com/oumMVlgDnLYFaPVkExemNVVZ" + room_name: "Server Notices" +``` + +The only compulsory setting is `system_mxid_localpart`, which defines the user +id of the Server Notices user, as above. `room_name` defines the name of the +room which will be created. + +`system_mxid_display_name` and `system_mxid_avatar_url` can be used to set the +displayname and avatar of the Server Notices user. + +Sending notices +--------------- + +As of the current version of synapse, there is no convenient interface for +sending notices (other than the automated ones sent as part of consent +tracking). + +In the meantime, it is possible to test this feature using the manhole. Having +gone into the manhole as described in [manhole.md](manhole.md), a notice can be +sent with something like: + +``` +>>> hs.get_server_notices_manager().send_notice('@user:server.com', {'msgtype':'m.text', 'body':'foo'}) +``` diff --git a/docs/sphinx/conf.py b/docs/sphinx/conf.py index 15c19834f..0b15bd891 100644 --- a/docs/sphinx/conf.py +++ b/docs/sphinx/conf.py @@ -50,7 +50,7 @@ master_doc = 'index' # General information about the project. project = u'Synapse' -copyright = u'2014, TNG' +copyright = u'Copyright 2014-2017 OpenMarket Ltd, 2017 Vector Creations Ltd, 2017 New Vector Ltd' # The version info for the project you're documenting, acts as replacement for # |version| and |release|, also used in various other places throughout the diff --git a/docs/url_previews.rst b/docs/url_previews.md similarity index 99% rename from docs/url_previews.rst rename to docs/url_previews.md index 634d9d907..665554e16 100644 --- a/docs/url_previews.rst +++ b/docs/url_previews.md @@ -56,6 +56,7 @@ As a first cut, let's do #2 and have the receiver hit the API to calculate its o API --- +``` GET /_matrix/media/r0/preview_url?url=http://wherever.com 200 OK { @@ -66,6 +67,7 @@ GET /_matrix/media/r0/preview_url?url=http://wherever.com "og:description" : "“Synapse 0.12 is out! Lots of polishing, performance &amp; bugfixes: /sync API, /r0 prefix, fulltext search, 3PID invites https://t.co/5alhXLLEGP”" "og:site_name" : "Twitter" } +``` * Downloads the URL * If HTML, just stores it in RAM and parses it for OG meta tags diff --git a/docs/user_directory.md b/docs/user_directory.md new file mode 100644 index 000000000..4c8ee44f3 --- /dev/null +++ b/docs/user_directory.md @@ -0,0 +1,17 @@ +User Directory API Implementation +================================= + +The user directory is currently maintained based on the 'visible' users +on this particular server - i.e. ones which your account shares a room with, or +who are present in a publicly viewable room present on the server. + +The directory info is stored in various tables, which can (typically after +DB corruption) get stale or out of sync. If this happens, for now the +quickest solution to fix it is: + +``` +UPDATE user_directory_stream_pos SET stream_id = NULL; +``` + +and restart the synapse, which should then start a background task to +flush the current tables and regenerate the directory. diff --git a/docs/workers.rst b/docs/workers.rst index 2d3df9159..c5b37c3de 100644 --- a/docs/workers.rst +++ b/docs/workers.rst @@ -1,11 +1,15 @@ Scaling synapse via workers ---------------------------- +=========================== Synapse has experimental support for splitting out functionality into multiple separate python processes, helping greatly with scalability. These processes are called 'workers', and are (eventually) intended to scale horizontally independently. +All of the below is highly experimental and subject to change as Synapse evolves, +but documenting it here to help folks needing highly scalable Synapses similar +to the one running matrix.org! + All processes continue to share the same database instance, and as such, workers only work with postgres based synapse deployments (sharing a single sqlite across multiple processes is a recipe for disaster, plus you should be using @@ -16,37 +20,62 @@ TCP protocol called 'replication' - analogous to MySQL or Postgres style database replication; feeding a stream of relevant data to the workers so they can be kept in sync with the main synapse process and database state. -To enable workers, you need to add a replication listener to the master synapse, e.g.:: +Configuration +------------- + +To make effective use of the workers, you will need to configure an HTTP +reverse-proxy such as nginx or haproxy, which will direct incoming requests to +the correct worker, or to the main synapse instance. Note that this includes +requests made to the federation port. The caveats regarding running a +reverse-proxy on the federation port still apply (see +https://github.com/matrix-org/synapse/blob/master/README.rst#reverse-proxying-the-federation-port). + +To enable workers, you need to add two replication listeners to the master +synapse, e.g.:: listeners: + # The TCP replication port - port: 9092 bind_address: '127.0.0.1' type: replication + # The HTTP replication port + - port: 9093 + bind_address: '127.0.0.1' + type: http + resources: + - names: [replication] -Under **no circumstances** should this replication API listener be exposed to the -public internet; it currently implements no authentication whatsoever and is +Under **no circumstances** should these replication API listeners be exposed to +the public internet; it currently implements no authentication whatsoever and is unencrypted. -You then create a set of configs for the various worker processes. These should be -worker configuration files should be stored in a dedicated subdirectory, to allow -synctl to manipulate them. +(Roughly, the TCP port is used for streaming data from the master to the +workers, and the HTTP port for the workers to send data to the main +synapse process.) -The current available worker applications are: - * synapse.app.pusher - handles sending push notifications to sygnal and email - * synapse.app.synchrotron - handles /sync endpoints. can scales horizontally through multiple instances. - * synapse.app.appservice - handles output traffic to Application Services - * synapse.app.federation_reader - handles receiving federation traffic (including public_rooms API) - * synapse.app.media_repository - handles the media repository. - * synapse.app.client_reader - handles client API endpoints like /publicRooms +You then create a set of configs for the various worker processes. These +should be worker configuration files, and should be stored in a dedicated +subdirectory, to allow synctl to manipulate them. An additional configuration +for the master synapse process will need to be created because the process will +not be started automatically. That configuration should look like this:: + + worker_app: synapse.app.homeserver + daemonize: true Each worker configuration file inherits the configuration of the main homeserver configuration file. You can then override configuration specific to that worker, e.g. the HTTP listener that it provides (if any); logging configuration; etc. You should minimise the number of overrides though to maintain a usable config. -You must specify the type of worker application (worker_app) and the replication -endpoint that it's talking to on the main synapse process (worker_replication_host -and worker_replication_port). +You must specify the type of worker application (``worker_app``). The currently +available worker applications are listed below. You must also specify the +replication endpoints that it's talking to on the main synapse process. +``worker_replication_host`` should specify the host of the main synapse, +``worker_replication_port`` should point to the TCP replication listener port and +``worker_replication_http_port`` should point to the HTTP replication port. + +Currently, only the ``event_creator`` worker requires specifying +``worker_replication_http_port``. For instance:: @@ -55,6 +84,7 @@ For instance:: # The replication listener on the synapse to talk to. worker_replication_host: 127.0.0.1 worker_replication_port: 9092 + worker_replication_http_port: 9093 worker_listeners: - type: http @@ -68,11 +98,11 @@ For instance:: worker_log_config: /home/matrix/synapse/config/synchrotron_log_config.yaml ...is a full configuration for a synchrotron worker instance, which will expose a -plain HTTP /sync endpoint on port 8083 separately from the /sync endpoint provided +plain HTTP ``/sync`` endpoint on port 8083 separately from the ``/sync`` endpoint provided by the main synapse. -Obviously you should configure your loadbalancer to route the /sync endpoint to -the synchrotron instance(s) in this instance. +Obviously you should configure your reverse-proxy to route the relevant +endpoints to the worker (``localhost:8083`` in the above example). Finally, to actually run your worker-based synapse, you must pass synctl the -a commandline option to tell it to operate on all the worker configurations found @@ -89,6 +119,131 @@ To manipulate a specific worker, you pass the -w option to synctl:: synctl -w $CONFIG/workers/synchrotron.yaml restart -All of the above is highly experimental and subject to change as Synapse evolves, -but documenting it here to help folks needing highly scalable Synapses similar -to the one running matrix.org! + +Available worker applications +----------------------------- + +``synapse.app.pusher`` +~~~~~~~~~~~~~~~~~~~~~~ + +Handles sending push notifications to sygnal and email. Doesn't handle any +REST endpoints itself, but you should set ``start_pushers: False`` in the +shared configuration file to stop the main synapse sending these notifications. + +Note this worker cannot be load-balanced: only one instance should be active. + +``synapse.app.synchrotron`` +~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +The synchrotron handles ``sync`` requests from clients. In particular, it can +handle REST endpoints matching the following regular expressions:: + + ^/_matrix/client/(v2_alpha|r0)/sync$ + ^/_matrix/client/(api/v1|v2_alpha|r0)/events$ + ^/_matrix/client/(api/v1|r0)/initialSync$ + ^/_matrix/client/(api/v1|r0)/rooms/[^/]+/initialSync$ + +The above endpoints should all be routed to the synchrotron worker by the +reverse-proxy configuration. + +It is possible to run multiple instances of the synchrotron to scale +horizontally. In this case the reverse-proxy should be configured to +load-balance across the instances, though it will be more efficient if all +requests from a particular user are routed to a single instance. Extracting +a userid from the access token is currently left as an exercise for the reader. + +``synapse.app.appservice`` +~~~~~~~~~~~~~~~~~~~~~~~~~~ + +Handles sending output traffic to Application Services. Doesn't handle any +REST endpoints itself, but you should set ``notify_appservices: False`` in the +shared configuration file to stop the main synapse sending these notifications. + +Note this worker cannot be load-balanced: only one instance should be active. + +``synapse.app.federation_reader`` +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +Handles a subset of federation endpoints. In particular, it can handle REST +endpoints matching the following regular expressions:: + + ^/_matrix/federation/v1/event/ + ^/_matrix/federation/v1/state/ + ^/_matrix/federation/v1/state_ids/ + ^/_matrix/federation/v1/backfill/ + ^/_matrix/federation/v1/get_missing_events/ + ^/_matrix/federation/v1/publicRooms + +The above endpoints should all be routed to the federation_reader worker by the +reverse-proxy configuration. + +``synapse.app.federation_sender`` +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +Handles sending federation traffic to other servers. Doesn't handle any +REST endpoints itself, but you should set ``send_federation: False`` in the +shared configuration file to stop the main synapse sending this traffic. + +Note this worker cannot be load-balanced: only one instance should be active. + +``synapse.app.media_repository`` +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +Handles the media repository. It can handle all endpoints starting with:: + + /_matrix/media/ + +You should also set ``enable_media_repo: False`` in the shared configuration +file to stop the main synapse running background jobs related to managing the +media repository. + +Note this worker cannot be load-balanced: only one instance should be active. + +``synapse.app.client_reader`` +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +Handles client API endpoints. It can handle REST endpoints matching the +following regular expressions:: + + ^/_matrix/client/(api/v1|r0|unstable)/publicRooms$ + ^/_matrix/client/(api/v1|r0|unstable)/rooms/.*/joined_members$ + ^/_matrix/client/(api/v1|r0|unstable)/rooms/.*/context/.*$ + ^/_matrix/client/(api/v1|r0|unstable)/rooms/.*/members$ + ^/_matrix/client/(api/v1|r0|unstable)/rooms/.*/state$ + +``synapse.app.user_dir`` +~~~~~~~~~~~~~~~~~~~~~~~~ + +Handles searches in the user directory. It can handle REST endpoints matching +the following regular expressions:: + + ^/_matrix/client/(api/v1|r0|unstable)/user_directory/search$ + +``synapse.app.frontend_proxy`` +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +Proxies some frequently-requested client endpoints to add caching and remove +load from the main synapse. It can handle REST endpoints matching the following +regular expressions:: + + ^/_matrix/client/(api/v1|r0|unstable)/keys/upload + +It will proxy any requests it cannot handle to the main synapse instance. It +must therefore be configured with the location of the main instance, via +the ``worker_main_http_uri`` setting in the frontend_proxy worker configuration +file. For example:: + + worker_main_http_uri: http://127.0.0.1:8008 + + +``synapse.app.event_creator`` +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +Handles some event creation. It can handle REST endpoints matching:: + + ^/_matrix/client/(api/v1|r0|unstable)/rooms/.*/send + ^/_matrix/client/(api/v1|r0|unstable)/rooms/.*/(join|invite|leave|ban|unban|kick)$ + ^/_matrix/client/(api/v1|r0|unstable)/join/ + +It will create events locally and then send them on to the main synapse +instance to be persisted and handled. diff --git a/jenkins-dendron-haproxy-postgres.sh b/jenkins-dendron-haproxy-postgres.sh index d64b2d2c9..07979bf8b 100755 --- a/jenkins-dendron-haproxy-postgres.sh +++ b/jenkins-dendron-haproxy-postgres.sh @@ -17,6 +17,7 @@ export HAPROXY_BIN=/home/haproxy/haproxy-1.6.11/haproxy ./sytest/jenkins/prep_sytest_for_postgres.sh ./sytest/jenkins/install_and_run.sh \ + --python $WORKSPACE/.tox/py27/bin/python \ --synapse-directory $WORKSPACE \ --dendron $WORKSPACE/dendron/bin/dendron \ --haproxy \ diff --git a/jenkins-dendron-postgres.sh b/jenkins-dendron-postgres.sh index 37ae746f4..3b932fe34 100755 --- a/jenkins-dendron-postgres.sh +++ b/jenkins-dendron-postgres.sh @@ -15,5 +15,6 @@ export SYNAPSE_CACHE_FACTOR=1 ./sytest/jenkins/prep_sytest_for_postgres.sh ./sytest/jenkins/install_and_run.sh \ + --python $WORKSPACE/.tox/py27/bin/python \ --synapse-directory $WORKSPACE \ --dendron $WORKSPACE/dendron/bin/dendron \ diff --git a/jenkins-postgres.sh b/jenkins-postgres.sh index f2ca8ccdf..1afb73639 100755 --- a/jenkins-postgres.sh +++ b/jenkins-postgres.sh @@ -14,4 +14,5 @@ export SYNAPSE_CACHE_FACTOR=1 ./sytest/jenkins/prep_sytest_for_postgres.sh ./sytest/jenkins/install_and_run.sh \ + --python $WORKSPACE/.tox/py27/bin/python \ --synapse-directory $WORKSPACE \ diff --git a/jenkins-sqlite.sh b/jenkins-sqlite.sh index 84613d979..baf4713a0 100755 --- a/jenkins-sqlite.sh +++ b/jenkins-sqlite.sh @@ -12,4 +12,5 @@ export SYNAPSE_CACHE_FACTOR=1 ./jenkins/clone.sh sytest https://github.com/matrix-org/sytest.git ./sytest/jenkins/install_and_run.sh \ + --python $WORKSPACE/.tox/py27/bin/python \ --synapse-directory $WORKSPACE \ diff --git a/jenkins/prepare_synapse.sh b/jenkins/prepare_synapse.sh index ffcb1cfab..a30179f2a 100755 --- a/jenkins/prepare_synapse.sh +++ b/jenkins/prepare_synapse.sh @@ -1,5 +1,7 @@ #! /bin/bash +set -eux + cd "`dirname $0`/.." TOX_DIR=$WORKSPACE/.tox @@ -14,7 +16,20 @@ fi tox -e py27 --notest -v TOX_BIN=$TOX_DIR/py27/bin -$TOX_BIN/pip install setuptools + +# cryptography 2.2 requires setuptools >= 18.5. +# +# older versions of virtualenv (?) give us a virtualenv with the same version +# of setuptools as is installed on the system python (and tox runs virtualenv +# under python3, so we get the version of setuptools that is installed on that). +# +# anyway, make sure that we have a recent enough setuptools. +$TOX_BIN/pip install 'setuptools>=18.5' + +# we also need a semi-recent version of pip, because old ones fail to install +# the "enum34" dependency of cryptography. +$TOX_BIN/pip install 'pip>=10' + { python synapse/python_dependencies.py echo lxml psycopg2 } | xargs $TOX_BIN/pip install diff --git a/pyproject.toml b/pyproject.toml new file mode 100644 index 000000000..f4d6f0c6b --- /dev/null +++ b/pyproject.toml @@ -0,0 +1,30 @@ +[tool.towncrier] + package = "synapse" + filename = "CHANGES.md" + directory = "changelog.d" + issue_format = "[\\#{issue}](https://github.com/matrix-org/synapse/issues/{issue}>)" + + [[tool.towncrier.type]] + directory = "feature" + name = "Features" + showcontent = true + + [[tool.towncrier.type]] + directory = "bugfix" + name = "Bugfixes" + showcontent = true + + [[tool.towncrier.type]] + directory = "doc" + name = "Improved Documentation" + showcontent = true + + [[tool.towncrier.type]] + directory = "removal" + name = "Deprecations and Removals" + showcontent = true + + [[tool.towncrier.type]] + directory = "misc" + name = "Internal Changes" + showcontent = true diff --git a/scripts-dev/federation_client.py b/scripts-dev/federation_client.py old mode 100644 new mode 100755 index d1ab42d3a..d2acc7654 --- a/scripts-dev/federation_client.py +++ b/scripts-dev/federation_client.py @@ -1,10 +1,38 @@ +#!/usr/bin/env python +# +# Copyright 2015, 2016 OpenMarket Ltd +# Copyright 2017 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import print_function + +import argparse +from urlparse import urlparse, urlunparse + import nacl.signing import json import base64 import requests import sys -import srvlookup +from requests.adapters import HTTPAdapter +import srvlookup +import yaml + +# uncomment the following to enable debug logging of http requests +#from httplib import HTTPConnection +#HTTPConnection.debuglevel = 1 def encode_base64(input_bytes): """Encode bytes as a base64 string without any padding.""" @@ -93,25 +121,24 @@ def read_signing_keys(stream): return keys -def lookup(destination, path): - if ":" in destination: - return "https://%s%s" % (destination, path) - else: - try: - srv = srvlookup.lookup("matrix", "tcp", destination)[0] - return "https://%s:%d%s" % (srv.host, srv.port, path) - except: - return "https://%s:%d%s" % (destination, 8448, path) +def request_json(method, origin_name, origin_key, destination, path, content): + if method is None: + if content is None: + method = "GET" + else: + method = "POST" -def get_json(origin_name, origin_key, destination, path): - request_json = { - "method": "GET", + json_to_sign = { + "method": method, "uri": path, "origin": origin_name, "destination": destination, } - signed_json = sign_json(request_json, origin_key, origin_name) + if content is not None: + json_to_sign["content"] = json.loads(content) + + signed_json = sign_json(json_to_sign, origin_key, origin_name) authorization_headers = [] @@ -120,30 +147,137 @@ def get_json(origin_name, origin_key, destination, path): origin_name, key, sig, ) authorization_headers.append(bytes(header)) - sys.stderr.write(header) - sys.stderr.write("\n") + print ("Authorization: %s" % header, file=sys.stderr) - result = requests.get( - lookup(destination, path), - headers={"Authorization": authorization_headers[0]}, + dest = "matrix://%s%s" % (destination, path) + print ("Requesting %s" % dest, file=sys.stderr) + + s = requests.Session() + s.mount("matrix://", MatrixConnectionAdapter()) + + result = s.request( + method=method, + url=dest, + headers={ + "Host": destination, + "Authorization": authorization_headers[0] + }, verify=False, + data=content, ) sys.stderr.write("Status Code: %d\n" % (result.status_code,)) return result.json() def main(): - origin_name, keyfile, destination, path = sys.argv[1:] + parser = argparse.ArgumentParser( + description= + "Signs and sends a federation request to a matrix homeserver", + ) - with open(keyfile) as f: + parser.add_argument( + "-N", "--server-name", + help="Name to give as the local homeserver. If unspecified, will be " + "read from the config file.", + ) + + parser.add_argument( + "-k", "--signing-key-path", + help="Path to the file containing the private ed25519 key to sign the " + "request with.", + ) + + parser.add_argument( + "-c", "--config", + default="homeserver.yaml", + help="Path to server config file. Ignored if --server-name and " + "--signing-key-path are both given.", + ) + + parser.add_argument( + "-d", "--destination", + default="matrix.org", + help="name of the remote homeserver. We will do SRV lookups and " + "connect appropriately.", + ) + + parser.add_argument( + "-X", "--method", + help="HTTP method to use for the request. Defaults to GET if --data is" + "unspecified, POST if it is." + ) + + parser.add_argument( + "--body", + help="Data to send as the body of the HTTP request" + ) + + parser.add_argument( + "path", + help="request path. We will add '/_matrix/federation/v1/' to this." + ) + + args = parser.parse_args() + + if not args.server_name or not args.signing_key_path: + read_args_from_config(args) + + with open(args.signing_key_path) as f: key = read_signing_keys(f)[0] - result = get_json( - origin_name, key, destination, "/_matrix/federation/v1/" + path + result = request_json( + args.method, + args.server_name, key, args.destination, + "/_matrix/federation/v1/" + args.path, + content=args.body, ) json.dump(result, sys.stdout) - print "" + print ("") + + +def read_args_from_config(args): + with open(args.config, 'r') as fh: + config = yaml.safe_load(fh) + if not args.server_name: + args.server_name = config['server_name'] + if not args.signing_key_path: + args.signing_key_path = config['signing_key_path'] + + +class MatrixConnectionAdapter(HTTPAdapter): + @staticmethod + def lookup(s): + if s[-1] == ']': + # ipv6 literal (with no port) + return s, 8448 + + if ":" in s: + out = s.rsplit(":",1) + try: + port = int(out[1]) + except ValueError: + raise ValueError("Invalid host:port '%s'" % s) + return out[0], port + + try: + srv = srvlookup.lookup("matrix", "tcp", s)[0] + return srv.host, srv.port + except: + return s, 8448 + + def get_connection(self, url, proxies=None): + parsed = urlparse(url) + + (host, port) = self.lookup(parsed.netloc) + netloc = "%s:%d" % (host, port) + print("Connecting to %s" % (netloc,), file=sys.stderr) + url = urlunparse(( + "https", netloc, parsed.path, parsed.params, parsed.query, + parsed.fragment, + )) + return super(MatrixConnectionAdapter, self).get_connection(url, proxies) + if __name__ == "__main__": main() diff --git a/scripts-dev/nuke-room-from-db.sh b/scripts-dev/nuke-room-from-db.sh index 1201d176c..c62928afd 100755 --- a/scripts-dev/nuke-room-from-db.sh +++ b/scripts-dev/nuke-room-from-db.sh @@ -6,9 +6,19 @@ ## Do not run it lightly. +set -e + +if [ "$1" == "-h" ] || [ "$1" == "" ]; then + echo "Call with ROOM_ID as first option and then pipe it into the database. So for instance you might run" + echo " nuke-room-from-db.sh | sqlite3 homeserver.db" + echo "or" + echo " nuke-room-from-db.sh | psql --dbname=synapse" + exit +fi + ROOMID="$1" -sqlite3 homeserver.db <| + +This can be extracted from postgres with:: + + psql --tuples-only -A -c "select media_origin, filesystem_id from + matrix.remote_media_cache where ..." + +To use, pipe the above into:: + + PYTHON_PATH=. ./scripts/move_remote_media_to_new_store.py +""" + +from __future__ import print_function + +import argparse +import logging + +import sys + +import os + +import shutil + +from synapse.rest.media.v1.filepath import MediaFilePaths + +logger = logging.getLogger() + + +def main(src_repo, dest_repo): + src_paths = MediaFilePaths(src_repo) + dest_paths = MediaFilePaths(dest_repo) + for line in sys.stdin: + line = line.strip() + parts = line.split('|') + if len(parts) != 2: + print("Unable to parse input line %s" % line, file=sys.stderr) + exit(1) + + move_media(parts[0], parts[1], src_paths, dest_paths) + + +def move_media(origin_server, file_id, src_paths, dest_paths): + """Move the given file, and any thumbnails, to the dest repo + + Args: + origin_server (str): + file_id (str): + src_paths (MediaFilePaths): + dest_paths (MediaFilePaths): + """ + logger.info("%s/%s", origin_server, file_id) + + # check that the original exists + original_file = src_paths.remote_media_filepath(origin_server, file_id) + if not os.path.exists(original_file): + logger.warn( + "Original for %s/%s (%s) does not exist", + origin_server, file_id, original_file, + ) + else: + mkdir_and_move( + original_file, + dest_paths.remote_media_filepath(origin_server, file_id), + ) + + # now look for thumbnails + original_thumb_dir = src_paths.remote_media_thumbnail_dir( + origin_server, file_id, + ) + if not os.path.exists(original_thumb_dir): + return + + mkdir_and_move( + original_thumb_dir, + dest_paths.remote_media_thumbnail_dir(origin_server, file_id) + ) + + +def mkdir_and_move(original_file, dest_file): + dirname = os.path.dirname(dest_file) + if not os.path.exists(dirname): + logger.debug("mkdir %s", dirname) + os.makedirs(dirname) + logger.debug("mv %s %s", original_file, dest_file) + shutil.move(original_file, dest_file) + + +if __name__ == "__main__": + parser = argparse.ArgumentParser( + description=__doc__, + formatter_class = argparse.RawDescriptionHelpFormatter, + ) + parser.add_argument( + "-v", action='store_true', help='enable debug logging') + parser.add_argument( + "src_repo", + help="Path to source content repo", + ) + parser.add_argument( + "dest_repo", + help="Path to source content repo", + ) + args = parser.parse_args() + + logging_config = { + "level": logging.DEBUG if args.v else logging.INFO, + "format": "%(asctime)s - %(name)s - %(lineno)d - %(levelname)s - %(message)s" + } + logging.basicConfig(**logging_config) + + main(args.src_repo, args.dest_repo) diff --git a/scripts/register_new_matrix_user b/scripts/register_new_matrix_user index 12ed20d62..8c3d42935 100755 --- a/scripts/register_new_matrix_user +++ b/scripts/register_new_matrix_user @@ -26,11 +26,37 @@ import yaml def request_registration(user, password, server_location, shared_secret, admin=False): + req = urllib2.Request( + "%s/_matrix/client/r0/admin/register" % (server_location,), + headers={'Content-Type': 'application/json'} + ) + + try: + if sys.version_info[:3] >= (2, 7, 9): + # As of version 2.7.9, urllib2 now checks SSL certs + import ssl + f = urllib2.urlopen(req, context=ssl.SSLContext(ssl.PROTOCOL_SSLv23)) + else: + f = urllib2.urlopen(req) + body = f.read() + f.close() + nonce = json.loads(body)["nonce"] + except urllib2.HTTPError as e: + print "ERROR! Received %d %s" % (e.code, e.reason,) + if 400 <= e.code < 500: + if e.info().type == "application/json": + resp = json.load(e) + if "error" in resp: + print resp["error"] + sys.exit(1) + mac = hmac.new( key=shared_secret, digestmod=hashlib.sha1, ) + mac.update(nonce) + mac.update("\x00") mac.update(user) mac.update("\x00") mac.update(password) @@ -40,10 +66,10 @@ def request_registration(user, password, server_location, shared_secret, admin=F mac = mac.hexdigest() data = { - "user": user, + "nonce": nonce, + "username": user, "password": password, "mac": mac, - "type": "org.matrix.login.shared_secret", "admin": admin, } @@ -52,7 +78,7 @@ def request_registration(user, password, server_location, shared_secret, admin=F print "Sending registration request..." req = urllib2.Request( - "%s/_matrix/client/api/v1/register" % (server_location,), + "%s/_matrix/client/r0/admin/register" % (server_location,), data=json.dumps(data), headers={'Content-Type': 'application/json'} ) diff --git a/scripts/synapse_port_db b/scripts/synapse_port_db index 7d158a46a..b9b828c15 100755 --- a/scripts/synapse_port_db +++ b/scripts/synapse_port_db @@ -1,6 +1,7 @@ #!/usr/bin/env python # -*- coding: utf-8 -*- # Copyright 2015, 2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -29,6 +30,8 @@ import time import traceback import yaml +from six import string_types + logger = logging.getLogger("synapse_port_db") @@ -42,6 +45,14 @@ BOOLEAN_COLUMNS = { "public_room_list_stream": ["visibility"], "device_lists_outbound_pokes": ["sent"], "users_who_share_rooms": ["share_private"], + "groups": ["is_public"], + "group_rooms": ["is_public"], + "group_users": ["is_public", "is_admin"], + "group_summary_rooms": ["is_public"], + "group_room_categories": ["is_public"], + "group_summary_users": ["is_public"], + "group_roles": ["is_public"], + "local_group_membership": ["is_publicised", "is_admin"], } @@ -112,6 +123,7 @@ class Store(object): _simple_update_one = SQLBaseStore.__dict__["_simple_update_one"] _simple_update_one_txn = SQLBaseStore.__dict__["_simple_update_one_txn"] + _simple_update_txn = SQLBaseStore.__dict__["_simple_update_txn"] def runInteraction(self, desc, func, *args, **kwargs): def r(conn): @@ -241,6 +253,12 @@ class Porter(object): @defer.inlineCallbacks def handle_table(self, table, postgres_size, table_size, forward_chunk, backward_chunk): + logger.info( + "Table %s: %i/%i (rows %i-%i) already ported", + table, postgres_size, table_size, + backward_chunk+1, forward_chunk-1, + ) + if not table_size: return @@ -252,6 +270,25 @@ class Porter(object): ) return + if table in ( + "user_directory", "user_directory_search", "users_who_share_rooms", + "users_in_pubic_room", + ): + # We don't port these tables, as they're a faff and we can regenreate + # them anyway. + self.progress.update(table, table_size) # Mark table as done + return + + if table == "user_directory_stream_pos": + # We need to make sure there is a single row, `(X, null), as that is + # what synapse expects to be there. + yield self.postgres_store._simple_insert( + table=table, + values={"stream_id": None}, + ) + self.progress.update(table, table_size) # Mark table as done + return + forward_select = ( "SELECT rowid, * FROM %s WHERE rowid >= ? ORDER BY rowid LIMIT ?" % (table,) @@ -299,7 +336,7 @@ class Porter(object): backward_chunk = min(row[0] for row in brows) - 1 rows = frows + brows - self._convert_rows(table, headers, rows) + rows = self._convert_rows(table, headers, rows) def insert(txn): self.postgres_store.insert_many_txn( @@ -357,10 +394,13 @@ class Porter(object): " VALUES (?,?,?,?,to_tsvector('english', ?),?,?)" ) - rows_dict = [ - dict(zip(headers, row)) - for row in rows - ] + rows_dict = [] + for row in rows: + d = dict(zip(headers, row)) + if "\0" in d['value']: + logger.warn('dropping search row %s', d) + else: + rows_dict.append(d) txn.executemany(sql, [ ( @@ -436,31 +476,10 @@ class Porter(object): self.progress.set_state("Preparing PostgreSQL") self.setup_db(postgres_config, postgres_engine) - # Step 2. Get tables. - self.progress.set_state("Fetching tables") - sqlite_tables = yield self.sqlite_store._simple_select_onecol( - table="sqlite_master", - keyvalues={ - "type": "table", - }, - retcol="name", - ) - - postgres_tables = yield self.postgres_store._simple_select_onecol( - table="information_schema.tables", - keyvalues={}, - retcol="distinct table_name", - ) - - tables = set(sqlite_tables) & set(postgres_tables) - - self.progress.set_state("Creating tables") - - logger.info("Found %d tables", len(tables)) - + self.progress.set_state("Creating port tables") def create_port_table(txn): txn.execute( - "CREATE TABLE port_from_sqlite3 (" + "CREATE TABLE IF NOT EXISTS port_from_sqlite3 (" " table_name varchar(100) NOT NULL UNIQUE," " forward_rowid bigint NOT NULL," " backward_rowid bigint NOT NULL" @@ -486,18 +505,33 @@ class Porter(object): "alter_table", alter_table ) except Exception as e: - logger.info("Failed to create port table: %s", e) + pass - try: - yield self.postgres_store.runInteraction( - "create_port_table", create_port_table - ) - except Exception as e: - logger.info("Failed to create port table: %s", e) + yield self.postgres_store.runInteraction( + "create_port_table", create_port_table + ) - self.progress.set_state("Setting up") + # Step 2. Get tables. + self.progress.set_state("Fetching tables") + sqlite_tables = yield self.sqlite_store._simple_select_onecol( + table="sqlite_master", + keyvalues={ + "type": "table", + }, + retcol="name", + ) - # Set up tables. + postgres_tables = yield self.postgres_store._simple_select_onecol( + table="information_schema.tables", + keyvalues={}, + retcol="distinct table_name", + ) + + tables = set(sqlite_tables) & set(postgres_tables) + logger.info("Found %d tables", len(tables)) + + # Step 3. Figure out what still needs copying + self.progress.set_state("Checking on port progress") setup_res = yield defer.gatherResults( [ self.setup_table(table) @@ -508,7 +542,8 @@ class Porter(object): consumeErrors=True, ) - # Process tables. + # Step 4. Do the copying. + self.progress.set_state("Copying to postgres") yield defer.gatherResults( [ self.handle_table(*res) @@ -517,6 +552,9 @@ class Porter(object): consumeErrors=True, ) + # Step 5. Do final post-processing + yield self._setup_state_group_id_seq() + self.progress.done() except: global end_error_exec_info @@ -532,17 +570,29 @@ class Porter(object): i for i, h in enumerate(headers) if h in bool_col_names ] + class BadValueException(Exception): + pass + def conv(j, col): if j in bool_cols: return bool(col) + elif isinstance(col, string_types) and "\0" in col: + logger.warn("DROPPING ROW: NUL value in table %s col %s: %r", table, headers[j], col) + raise BadValueException(); return col + outrows = [] for i, row in enumerate(rows): - rows[i] = tuple( - conv(j, col) - for j, col in enumerate(row) - if j > 0 - ) + try: + outrows.append(tuple( + conv(j, col) + for j, col in enumerate(row) + if j > 0 + )) + except BadValueException: + pass + + return outrows @defer.inlineCallbacks def _setup_sent_transactions(self): @@ -570,7 +620,7 @@ class Porter(object): "select", r, ) - self._convert_rows("sent_transactions", headers, rows) + rows = self._convert_rows("sent_transactions", headers, rows) inserted_rows = len(rows) if inserted_rows: @@ -664,6 +714,16 @@ class Porter(object): defer.returnValue((done, remaining + done)) + def _setup_state_group_id_seq(self): + def r(txn): + txn.execute("SELECT MAX(id) FROM state_groups") + next_id = txn.fetchone()[0]+1 + txn.execute( + "ALTER SEQUENCE state_group_id_seq RESTART WITH %s", + (next_id,), + ) + return self.postgres_store.runInteraction("setup_state_group_id_seq", r) + ############################################## ###### The following is simply UI stuff ###### diff --git a/scripts/sync_room_to_group.pl b/scripts/sync_room_to_group.pl new file mode 100755 index 000000000..f0c2dfadf --- /dev/null +++ b/scripts/sync_room_to_group.pl @@ -0,0 +1,45 @@ +#!/usr/bin/env perl + +use strict; +use warnings; + +use JSON::XS; +use LWP::UserAgent; +use URI::Escape; + +if (@ARGV < 4) { + die "usage: $0 \n"; +} + +my ($hs, $access_token, $room_id, $group_id) = @ARGV; +my $ua = LWP::UserAgent->new(); +$ua->timeout(10); + +if ($room_id =~ /^#/) { + $room_id = uri_escape($room_id); + $room_id = decode_json($ua->get("${hs}/_matrix/client/r0/directory/room/${room_id}?access_token=${access_token}")->decoded_content)->{room_id}; +} + +my $room_users = [ keys %{decode_json($ua->get("${hs}/_matrix/client/r0/rooms/${room_id}/joined_members?access_token=${access_token}")->decoded_content)->{joined}} ]; +my $group_users = [ + (map { $_->{user_id} } @{decode_json($ua->get("${hs}/_matrix/client/unstable/groups/${group_id}/users?access_token=${access_token}" )->decoded_content)->{chunk}}), + (map { $_->{user_id} } @{decode_json($ua->get("${hs}/_matrix/client/unstable/groups/${group_id}/invited_users?access_token=${access_token}" )->decoded_content)->{chunk}}), +]; + +die "refusing to sync from empty room" unless (@$room_users); +die "refusing to sync to empty group" unless (@$group_users); + +my $diff = {}; +foreach my $user (@$room_users) { $diff->{$user}++ } +foreach my $user (@$group_users) { $diff->{$user}-- } + +foreach my $user (keys %$diff) { + if ($diff->{$user} == 1) { + warn "inviting $user"; + print STDERR $ua->put("${hs}/_matrix/client/unstable/groups/${group_id}/admin/users/invite/${user}?access_token=${access_token}", Content=>'{}')->status_line."\n"; + } + elsif ($diff->{$user} == -1) { + warn "removing $user"; + print STDERR $ua->put("${hs}/_matrix/client/unstable/groups/${group_id}/admin/users/remove/${user}?access_token=${access_token}", Content=>'{}')->status_line."\n"; + } +} diff --git a/setup.cfg b/setup.cfg index da8eafbb3..c2620be6c 100644 --- a/setup.cfg +++ b/setup.cfg @@ -14,7 +14,26 @@ ignore = pylint.cfg tox.ini -[flake8] +[pep8] max-line-length = 90 -# W503 requires that binary operators be at the end, not start, of lines. Erik doesn't like it. -ignore = W503 +# W503 requires that binary operators be at the end, not start, of lines. Erik +# doesn't like it. E203 is contrary to PEP8. +ignore = W503,E203 + +[flake8] +# note that flake8 inherits the "ignore" settings from "pep8" (because it uses +# pep8 to do those checks), but not the "max-line-length" setting +max-line-length = 90 + +[isort] +line_length = 89 +not_skip = __init__.py +sections=FUTURE,STDLIB,COMPAT,THIRDPARTY,TWISTED,FIRSTPARTY,TESTS,LOCALFOLDER +default_section=THIRDPARTY +known_first_party = synapse +known_tests=tests +known_compat = mock,six +known_twisted=twisted,OpenSSL +multi_line_output=3 +include_trailing_comma=true +combine_as_imports=true diff --git a/synapse/__init__.py b/synapse/__init__.py index 6b0a76639..5c0f2f83a 100644 --- a/synapse/__init__.py +++ b/synapse/__init__.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2014-2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -16,4 +17,4 @@ """ This is a reference implementation of a Matrix home server. """ -__version__ = "0.22.0-rc1" +__version__ = "0.33.0" diff --git a/synapse/api/auth.py b/synapse/api/auth.py index f8266d1c8..073229b4c 100644 --- a/synapse/api/auth.py +++ b/synapse/api/auth.py @@ -15,15 +15,19 @@ import logging +from six import itervalues + import pymacaroons +from netaddr import IPAddress + from twisted.internet import defer import synapse.types from synapse import event_auth -from synapse.api.constants import EventTypes, Membership, JoinRules +from synapse.api.constants import EventTypes, JoinRules, Membership from synapse.api.errors import AuthError, Codes from synapse.types import UserID -from synapse.util.caches import register_cache, CACHE_SIZE_FACTOR +from synapse.util.caches import CACHE_SIZE_FACTOR, register_cache from synapse.util.caches.lrucache import LruCache from synapse.util.metrics import Measure @@ -57,16 +61,17 @@ class Auth(object): self.TOKEN_NOT_FOUND_HTTP_STATUS = 401 self.token_cache = LruCache(CACHE_SIZE_FACTOR * 10000) - register_cache("token_cache", self.token_cache) + register_cache("cache", "token_cache", self.token_cache) @defer.inlineCallbacks def check_from_context(self, event, context, do_sig_check=True): + prev_state_ids = yield context.get_prev_state_ids(self.store) auth_events_ids = yield self.compute_auth_events( - event, context.prev_state_ids, for_verification=True, + event, prev_state_ids, for_verification=True, ) auth_events = yield self.store.get_events(auth_events_ids) auth_events = { - (e.type, e.state_key): e for e in auth_events.values() + (e.type, e.state_key): e for e in itervalues(auth_events) } self.check(event, auth_events=auth_events, do_sig_check=do_sig_check) @@ -189,7 +194,7 @@ class Auth(object): synapse.types.create_requester(user_id, app_service=app_service) ) - access_token = get_access_token_from_request( + access_token = self.get_access_token_from_request( request, self.TOKEN_NOT_FOUND_HTTP_STATUS ) @@ -204,12 +209,12 @@ class Auth(object): ip_addr = self.hs.get_ip_from_request(request) user_agent = request.requestHeaders.getRawHeaders( - "User-Agent", - default=[""] + b"User-Agent", + default=[b""] )[0] if user and access_token and ip_addr: self.store.insert_client_ip( - user=user, + user_id=user.to_string(), access_token=access_token, ip=ip_addr, user_agent=user_agent, @@ -235,13 +240,18 @@ class Auth(object): @defer.inlineCallbacks def _get_appservice_user_id(self, request): app_service = self.store.get_app_service_by_token( - get_access_token_from_request( + self.get_access_token_from_request( request, self.TOKEN_NOT_FOUND_HTTP_STATUS ) ) if app_service is None: defer.returnValue((None, None)) + if app_service.ip_range_whitelist: + ip_address = IPAddress(self.hs.get_ip_from_request(request)) + if ip_address not in app_service.ip_range_whitelist: + defer.returnValue((None, None)) + if "user_id" not in request.args: defer.returnValue((app_service.sender, app_service)) @@ -270,7 +280,11 @@ class Auth(object): rights (str): The operation being performed; the access token must allow this. Returns: - dict : dict that includes the user and the ID of their access token. + Deferred[dict]: dict that includes: + `user` (UserID) + `is_guest` (bool) + `token_id` (int|None): access token id. May be None if guest + `device_id` (str|None): device corresponding to access token Raises: AuthError if no user by that token exists or the token is invalid. """ @@ -482,7 +496,7 @@ class Auth(object): def _look_up_user_by_access_token(self, token): ret = yield self.store.get_user_by_access_token(token) if not ret: - logger.warn("Unrecognised access token - not in store: %s" % (token,)) + logger.warn("Unrecognised access token - not in store.") raise AuthError( self.TOKEN_NOT_FOUND_HTTP_STATUS, "Unrecognised access token.", errcode=Codes.UNKNOWN_TOKEN @@ -500,12 +514,12 @@ class Auth(object): def get_appservice_by_req(self, request): try: - token = get_access_token_from_request( + token = self.get_access_token_from_request( request, self.TOKEN_NOT_FOUND_HTTP_STATUS ) service = self.store.get_app_service_by_token(token) if not service: - logger.warn("Unrecognised appservice access token: %s" % (token,)) + logger.warn("Unrecognised appservice access token.") raise AuthError( self.TOKEN_NOT_FOUND_HTTP_STATUS, "Unrecognised access token.", @@ -519,11 +533,20 @@ class Auth(object): ) def is_server_admin(self, user): + """ Check if the given user is a local server admin. + + Args: + user (str): mxid of user to check + + Returns: + bool: True if the user is an admin + """ return self.store.is_server_admin(user) @defer.inlineCallbacks def add_auth_events(self, builder, context): - auth_ids = yield self.compute_auth_events(builder, context.prev_state_ids) + prev_state_ids = yield context.get_prev_state_ids(self.store) + auth_ids = yield self.compute_auth_events(builder, prev_state_ids) auth_events_entries = yield self.store.add_event_hashes( auth_ids @@ -641,7 +664,7 @@ class Auth(object): auth_events[(EventTypes.PowerLevels, "")] = power_level_event send_level = event_auth.get_send_level( - EventTypes.Aliases, "", auth_events + EventTypes.Aliases, "", power_level_event, ) user_level = event_auth.get_user_power_level(user_id, auth_events) @@ -652,67 +675,101 @@ class Auth(object): " edit its room list entry" ) + @staticmethod + def has_access_token(request): + """Checks if the request has an access_token. -def has_access_token(request): - """Checks if the request has an access_token. + Returns: + bool: False if no access_token was given, True otherwise. + """ + query_params = request.args.get("access_token") + auth_headers = request.requestHeaders.getRawHeaders(b"Authorization") + return bool(query_params) or bool(auth_headers) - Returns: - bool: False if no access_token was given, True otherwise. - """ - query_params = request.args.get("access_token") - auth_headers = request.requestHeaders.getRawHeaders("Authorization") - return bool(query_params) or bool(auth_headers) + @staticmethod + def get_access_token_from_request(request, token_not_found_http_status=401): + """Extracts the access_token from the request. + Args: + request: The http request. + token_not_found_http_status(int): The HTTP status code to set in the + AuthError if the token isn't found. This is used in some of the + legacy APIs to change the status code to 403 from the default of + 401 since some of the old clients depended on auth errors returning + 403. + Returns: + str: The access_token + Raises: + AuthError: If there isn't an access_token in the request. + """ -def get_access_token_from_request(request, token_not_found_http_status=401): - """Extracts the access_token from the request. - - Args: - request: The http request. - token_not_found_http_status(int): The HTTP status code to set in the - AuthError if the token isn't found. This is used in some of the - legacy APIs to change the status code to 403 from the default of - 401 since some of the old clients depended on auth errors returning - 403. - Returns: - str: The access_token - Raises: - AuthError: If there isn't an access_token in the request. - """ - - auth_headers = request.requestHeaders.getRawHeaders("Authorization") - query_params = request.args.get("access_token") - if auth_headers: - # Try the get the access_token from a "Authorization: Bearer" - # header - if query_params is not None: - raise AuthError( - token_not_found_http_status, - "Mixing Authorization headers and access_token query parameters.", - errcode=Codes.MISSING_TOKEN, - ) - if len(auth_headers) > 1: - raise AuthError( - token_not_found_http_status, - "Too many Authorization headers.", - errcode=Codes.MISSING_TOKEN, - ) - parts = auth_headers[0].split(" ") - if parts[0] == "Bearer" and len(parts) == 2: - return parts[1] + auth_headers = request.requestHeaders.getRawHeaders(b"Authorization") + query_params = request.args.get(b"access_token") + if auth_headers: + # Try the get the access_token from a "Authorization: Bearer" + # header + if query_params is not None: + raise AuthError( + token_not_found_http_status, + "Mixing Authorization headers and access_token query parameters.", + errcode=Codes.MISSING_TOKEN, + ) + if len(auth_headers) > 1: + raise AuthError( + token_not_found_http_status, + "Too many Authorization headers.", + errcode=Codes.MISSING_TOKEN, + ) + parts = auth_headers[0].split(" ") + if parts[0] == "Bearer" and len(parts) == 2: + return parts[1] + else: + raise AuthError( + token_not_found_http_status, + "Invalid Authorization header.", + errcode=Codes.MISSING_TOKEN, + ) else: - raise AuthError( - token_not_found_http_status, - "Invalid Authorization header.", - errcode=Codes.MISSING_TOKEN, - ) - else: - # Try to get the access_token from the query params. - if not query_params: - raise AuthError( - token_not_found_http_status, - "Missing access token.", - errcode=Codes.MISSING_TOKEN - ) + # Try to get the access_token from the query params. + if not query_params: + raise AuthError( + token_not_found_http_status, + "Missing access token.", + errcode=Codes.MISSING_TOKEN + ) - return query_params[0] + return query_params[0] + + @defer.inlineCallbacks + def check_in_room_or_world_readable(self, room_id, user_id): + """Checks that the user is or was in the room or the room is world + readable. If it isn't then an exception is raised. + + Returns: + Deferred[tuple[str, str|None]]: Resolves to the current membership of + the user in the room and the membership event ID of the user. If + the user is not in the room and never has been, then + `(Membership.JOIN, None)` is returned. + """ + + try: + # check_user_was_in_room will return the most recent membership + # event for the user if: + # * The user is a non-guest user, and was ever in the room + # * The user is a guest user, and has joined the room + # else it will throw. + member_event = yield self.check_user_was_in_room(room_id, user_id) + defer.returnValue((member_event.membership, member_event.event_id)) + except AuthError: + visibility = yield self.state.get_current_state( + room_id, EventTypes.RoomHistoryVisibility, "" + ) + if ( + visibility and + visibility.content["history_visibility"] == "world_readable" + ): + defer.returnValue((Membership.JOIN, None)) + return + raise AuthError( + 403, "Guest access not allowed", errcode=Codes.GUEST_ACCESS_FORBIDDEN + ) diff --git a/synapse/api/constants.py b/synapse/api/constants.py index 489efb7f8..4df930c8d 100644 --- a/synapse/api/constants.py +++ b/synapse/api/constants.py @@ -16,6 +16,9 @@ """Contains constants from the specification.""" +# the "depth" field on events is limited to 2**63 - 1 +MAX_DEPTH = 2**63 - 1 + class Membership(object): @@ -73,6 +76,8 @@ class EventTypes(object): Topic = "m.room.topic" Name = "m.room.name" + ServerACL = "m.room.server_acl" + class RejectedReason(object): AUTH_ERROR = "auth_error" diff --git a/synapse/api/errors.py b/synapse/api/errors.py index d0dfa959d..6074df292 100644 --- a/synapse/api/errors.py +++ b/synapse/api/errors.py @@ -15,9 +15,13 @@ """Contains exceptions and error codes.""" -import json import logging +from six import iteritems +from six.moves import http_client + +from canonicaljson import json + logger = logging.getLogger(__name__) @@ -46,8 +50,11 @@ class Codes(object): THREEPID_AUTH_FAILED = "M_THREEPID_AUTH_FAILED" THREEPID_IN_USE = "M_THREEPID_IN_USE" THREEPID_NOT_FOUND = "M_THREEPID_NOT_FOUND" + THREEPID_DENIED = "M_THREEPID_DENIED" INVALID_USERNAME = "M_INVALID_USERNAME" SERVER_NOT_TRUSTED = "M_SERVER_NOT_TRUSTED" + CONSENT_NOT_GIVEN = "M_CONSENT_NOT_GIVEN" + CANNOT_LEAVE_SERVER_NOTICE_ROOM = "M_CANNOT_LEAVE_SERVER_NOTICE_ROOM" class CodeMessageException(RuntimeError): @@ -135,11 +142,79 @@ class SynapseError(CodeMessageException): return res +class ConsentNotGivenError(SynapseError): + """The error returned to the client when the user has not consented to the + privacy policy. + """ + def __init__(self, msg, consent_uri): + """Constructs a ConsentNotGivenError + + Args: + msg (str): The human-readable error message + consent_url (str): The URL where the user can give their consent + """ + super(ConsentNotGivenError, self).__init__( + code=http_client.FORBIDDEN, + msg=msg, + errcode=Codes.CONSENT_NOT_GIVEN + ) + self._consent_uri = consent_uri + + def error_dict(self): + return cs_error( + self.msg, + self.errcode, + consent_uri=self._consent_uri + ) + + class RegistrationError(SynapseError): """An error raised when a registration event fails.""" pass +class FederationDeniedError(SynapseError): + """An error raised when the server tries to federate with a server which + is not on its federation whitelist. + + Attributes: + destination (str): The destination which has been denied + """ + + def __init__(self, destination): + """Raised by federation client or server to indicate that we are + are deliberately not attempting to contact a given server because it is + not on our federation whitelist. + + Args: + destination (str): the domain in question + """ + + self.destination = destination + + super(FederationDeniedError, self).__init__( + code=403, + msg="Federation denied with %s." % (self.destination,), + errcode=Codes.FORBIDDEN, + ) + + +class InteractiveAuthIncompleteError(Exception): + """An error raised when UI auth is not yet complete + + (This indicates we should return a 401 with 'result' as the body) + + Attributes: + result (dict): the server response to the request, which should be + passed back to the client + """ + def __init__(self, result): + super(InteractiveAuthIncompleteError, self).__init__( + "Interactive auth not yet complete", + ) + self.result = result + + class UnrecognizedRequestError(SynapseError): """An error indicating we don't understand the request you're trying to make""" def __init__(self, *args, **kwargs): @@ -247,13 +322,13 @@ def cs_error(msg, code=Codes.UNKNOWN, **kwargs): Args: msg (str): The error message. - code (int): The error code. + code (str): The error code. kwargs : Additional keys to add to the response. Returns: A dict representing the error response JSON. """ err = {"error": msg, "errcode": code} - for key, value in kwargs.iteritems(): + for key, value in iteritems(kwargs): err[key] = value return err diff --git a/synapse/api/filtering.py b/synapse/api/filtering.py index 83206348e..25346baa8 100644 --- a/synapse/api/filtering.py +++ b/synapse/api/filtering.py @@ -12,14 +12,15 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. -from synapse.api.errors import SynapseError -from synapse.storage.presence import UserPresenceState -from synapse.types import UserID, RoomID +import jsonschema +from canonicaljson import json +from jsonschema import FormatChecker + from twisted.internet import defer -import ujson as json -import jsonschema -from jsonschema import FormatChecker +from synapse.api.errors import SynapseError +from synapse.storage.presence import UserPresenceState +from synapse.types import RoomID, UserID FILTER_SCHEMA = { "additionalProperties": False, @@ -411,7 +412,7 @@ class Filter(object): return room_ids def filter(self, events): - return filter(self.check, events) + return list(filter(self.check, events)) def limit(self): return self.filter_json.get("limit", 10) diff --git a/synapse/api/urls.py b/synapse/api/urls.py index 91a33a340..71347912f 100644 --- a/synapse/api/urls.py +++ b/synapse/api/urls.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2014-2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -14,6 +15,12 @@ # limitations under the License. """Contains the URL paths to prefix various aspects of the server with. """ +import hmac +from hashlib import sha256 + +from six.moves.urllib.parse import urlencode + +from synapse.config import ConfigError CLIENT_PREFIX = "/_matrix/client/api/v1" CLIENT_V2_ALPHA_PREFIX = "/_matrix/client/v2_alpha" @@ -25,3 +32,46 @@ SERVER_KEY_PREFIX = "/_matrix/key/v1" SERVER_KEY_V2_PREFIX = "/_matrix/key/v2" MEDIA_PREFIX = "/_matrix/media/r0" LEGACY_MEDIA_PREFIX = "/_matrix/media/v1" + + +class ConsentURIBuilder(object): + def __init__(self, hs_config): + """ + Args: + hs_config (synapse.config.homeserver.HomeServerConfig): + """ + if hs_config.form_secret is None: + raise ConfigError( + "form_secret not set in config", + ) + if hs_config.public_baseurl is None: + raise ConfigError( + "public_baseurl not set in config", + ) + + self._hmac_secret = hs_config.form_secret.encode("utf-8") + self._public_baseurl = hs_config.public_baseurl + + def build_user_consent_uri(self, user_id): + """Build a URI which we can give to the user to do their privacy + policy consent + + Args: + user_id (str): mxid or username of user + + Returns + (str) the URI where the user can do consent + """ + mac = hmac.new( + key=self._hmac_secret, + msg=user_id, + digestmod=sha256, + ).hexdigest() + consent_uri = "%s_matrix/consent?%s" % ( + self._public_baseurl, + urlencode({ + "u": user_id, + "h": mac + }), + ) + return consent_uri diff --git a/synapse/app/__init__.py b/synapse/app/__init__.py index 9c2b62759..3b6b9368b 100644 --- a/synapse/app/__init__.py +++ b/synapse/app/__init__.py @@ -14,9 +14,11 @@ # limitations under the License. import sys + +from synapse import python_dependencies # noqa: E402 + sys.dont_write_bytecode = True -from synapse import python_dependencies # noqa: E402 try: python_dependencies.check_requirements() diff --git a/synapse/app/_base.py b/synapse/app/_base.py new file mode 100644 index 000000000..391bd14c5 --- /dev/null +++ b/synapse/app/_base.py @@ -0,0 +1,194 @@ +# -*- coding: utf-8 -*- +# Copyright 2017 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import gc +import logging +import sys + +from daemonize import Daemonize + +from twisted.internet import error, reactor + +from synapse.util import PreserveLoggingContext +from synapse.util.rlimit import change_resource_limit + +try: + import affinity +except Exception: + affinity = None + + +logger = logging.getLogger(__name__) + + +def start_worker_reactor(appname, config): + """ Run the reactor in the main process + + Daemonizes if necessary, and then configures some resources, before starting + the reactor. Pulls configuration from the 'worker' settings in 'config'. + + Args: + appname (str): application name which will be sent to syslog + config (synapse.config.Config): config object + """ + + logger = logging.getLogger(config.worker_app) + + start_reactor( + appname, + config.soft_file_limit, + config.gc_thresholds, + config.worker_pid_file, + config.worker_daemonize, + config.worker_cpu_affinity, + logger, + ) + + +def start_reactor( + appname, + soft_file_limit, + gc_thresholds, + pid_file, + daemonize, + cpu_affinity, + logger, +): + """ Run the reactor in the main process + + Daemonizes if necessary, and then configures some resources, before starting + the reactor + + Args: + appname (str): application name which will be sent to syslog + soft_file_limit (int): + gc_thresholds: + pid_file (str): name of pid file to write to if daemonize is True + daemonize (bool): true to run the reactor in a background process + cpu_affinity (int|None): cpu affinity mask + logger (logging.Logger): logger instance to pass to Daemonize + """ + + def run(): + # make sure that we run the reactor with the sentinel log context, + # otherwise other PreserveLoggingContext instances will get confused + # and complain when they see the logcontext arbitrarily swapping + # between the sentinel and `run` logcontexts. + with PreserveLoggingContext(): + logger.info("Running") + if cpu_affinity is not None: + if not affinity: + quit_with_error( + "Missing package 'affinity' required for cpu_affinity\n" + "option\n\n" + "Install by running:\n\n" + " pip install affinity\n\n" + ) + logger.info("Setting CPU affinity to %s" % cpu_affinity) + affinity.set_process_affinity_mask(0, cpu_affinity) + change_resource_limit(soft_file_limit) + if gc_thresholds: + gc.set_threshold(*gc_thresholds) + reactor.run() + + if daemonize: + daemon = Daemonize( + app=appname, + pid=pid_file, + action=run, + auto_close_fds=False, + verbose=True, + logger=logger, + ) + daemon.start() + else: + run() + + +def quit_with_error(error_string): + message_lines = error_string.split("\n") + line_length = max([len(l) for l in message_lines if len(l) < 80]) + 2 + sys.stderr.write("*" * line_length + '\n') + for line in message_lines: + sys.stderr.write(" %s\n" % (line.rstrip(),)) + sys.stderr.write("*" * line_length + '\n') + sys.exit(1) + + +def listen_metrics(bind_addresses, port): + """ + Start Prometheus metrics server. + """ + from synapse.metrics import RegistryProxy + from prometheus_client import start_http_server + + for host in bind_addresses: + reactor.callInThread(start_http_server, int(port), + addr=host, registry=RegistryProxy) + logger.info("Metrics now reporting on %s:%d", host, port) + + +def listen_tcp(bind_addresses, port, factory, backlog=50): + """ + Create a TCP socket for a port and several addresses + """ + for address in bind_addresses: + try: + reactor.listenTCP( + port, + factory, + backlog, + address + ) + except error.CannotListenError as e: + check_bind_error(e, address, bind_addresses) + + +def listen_ssl(bind_addresses, port, factory, context_factory, backlog=50): + """ + Create an SSL socket for a port and several addresses + """ + for address in bind_addresses: + try: + reactor.listenSSL( + port, + factory, + context_factory, + backlog, + address + ) + except error.CannotListenError as e: + check_bind_error(e, address, bind_addresses) + + +def check_bind_error(e, address, bind_addresses): + """ + This method checks an exception occurred while binding on 0.0.0.0. + If :: is specified in the bind addresses a warning is shown. + The exception is still raised otherwise. + + Binding on both 0.0.0.0 and :: causes an exception on Linux and macOS + because :: binds on both IPv4 and IPv6 (as per RFC 3493). + When binding on 0.0.0.0 after :: this can safely be ignored. + + Args: + e (Exception): Exception that was caught. + address (str): Address on which binding was attempted. + bind_addresses (list): Addresses on which the service listens. + """ + if address == '0.0.0.0' and '::' in bind_addresses: + logger.warn('Failed to listen on 0.0.0.0, continuing because listening on [::]') + else: + raise e diff --git a/synapse/app/appservice.py b/synapse/app/appservice.py index 9a476efa6..9a37384fb 100644 --- a/synapse/app/appservice.py +++ b/synapse/app/appservice.py @@ -13,38 +13,33 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. +import logging +import sys + +from twisted.internet import defer, reactor +from twisted.web.resource import NoResource import synapse - -from synapse.server import HomeServer +from synapse import events +from synapse.app import _base from synapse.config._base import ConfigError -from synapse.config.logger import setup_logging from synapse.config.homeserver import HomeServerConfig +from synapse.config.logger import setup_logging from synapse.http.site import SynapseSite -from synapse.metrics.resource import MetricsResource, METRICS_PREFIX +from synapse.metrics import RegistryProxy +from synapse.metrics.resource import METRICS_PREFIX, MetricsResource +from synapse.replication.slave.storage.appservice import SlavedApplicationServiceStore from synapse.replication.slave.storage.directory import DirectoryStore from synapse.replication.slave.storage.events import SlavedEventStore -from synapse.replication.slave.storage.appservice import SlavedApplicationServiceStore from synapse.replication.slave.storage.registration import SlavedRegistrationStore from synapse.replication.tcp.client import ReplicationClientHandler +from synapse.server import HomeServer from synapse.storage.engines import create_engine from synapse.util.httpresourcetree import create_resource_tree -from synapse.util.logcontext import LoggingContext, PreserveLoggingContext, preserve_fn +from synapse.util.logcontext import LoggingContext, run_in_background from synapse.util.manhole import manhole -from synapse.util.rlimit import change_resource_limit from synapse.util.versionstring import get_version_string -from synapse import events - -from twisted.internet import reactor -from twisted.web.resource import Resource - -from daemonize import Daemonize - -import sys -import logging -import gc - logger = logging.getLogger("synapse.app.appservice") @@ -56,19 +51,6 @@ class AppserviceSlaveStore( class AppserviceServer(HomeServer): - def get_db_conn(self, run_new_connection=True): - # Any param beginning with cp_ is a parameter for adbapi, and should - # not be passed to the database engine. - db_params = { - k: v for k, v in self.db_config.get("args", {}).items() - if not k.startswith("cp_") - } - db_conn = self.database_engine.module.connect(**db_params) - - if run_new_connection: - self.database_engine.on_new_connection(db_conn) - return db_conn - def setup(self): logger.info("Setting up.") self.datastore = AppserviceSlaveStore(self.get_db_conn(), self) @@ -82,21 +64,21 @@ class AppserviceServer(HomeServer): for res in listener_config["resources"]: for name in res["names"]: if name == "metrics": - resources[METRICS_PREFIX] = MetricsResource(self) + resources[METRICS_PREFIX] = MetricsResource(RegistryProxy) - root_resource = create_resource_tree(resources, Resource()) + root_resource = create_resource_tree(resources, NoResource()) - for address in bind_addresses: - reactor.listenTCP( - port, - SynapseSite( - "synapse.access.http.%s" % (site_tag,), - site_tag, - listener_config, - root_resource, - ), - interface=address + _base.listen_tcp( + bind_addresses, + port, + SynapseSite( + "synapse.access.http.%s" % (site_tag,), + site_tag, + listener_config, + root_resource, + self.version_string, ) + ) logger.info("Synapse appservice now listening on port %d", port) @@ -105,18 +87,22 @@ class AppserviceServer(HomeServer): if listener["type"] == "http": self._listen_http(listener) elif listener["type"] == "manhole": - bind_addresses = listener["bind_addresses"] - - for address in bind_addresses: - reactor.listenTCP( - listener["port"], - manhole( - username="matrix", - password="rabbithole", - globals={"hs": self}, - ), - interface=address + _base.listen_tcp( + listener["bind_addresses"], + listener["port"], + manhole( + username="matrix", + password="rabbithole", + globals={"hs": self}, ) + ) + elif listener["type"] == "metrics": + if not self.get_config().enable_metrics: + logger.warn(("Metrics listener configured, but " + "enable_metrics is not True!")) + else: + _base.listen_metrics(listener["bind_addresses"], + listener["port"]) else: logger.warn("Unrecognized listener type: %s", listener["type"]) @@ -136,9 +122,14 @@ class ASReplicationHandler(ReplicationClientHandler): if stream_name == "events": max_stream_id = self.store.get_room_max_stream_ordering() - preserve_fn( - self.appservice_handler.notify_interested_services - )(max_stream_id) + run_in_background(self._notify_app_services, max_stream_id) + + @defer.inlineCallbacks + def _notify_app_services(self, room_stream_id): + try: + yield self.appservice_handler.notify_interested_services(room_stream_id) + except Exception: + logger.exception("Error notifying application services of event") def start(config_options): @@ -181,36 +172,13 @@ def start(config_options): ps.setup() ps.start_listening(config.worker_listeners) - def run(): - # make sure that we run the reactor with the sentinel log context, - # otherwise other PreserveLoggingContext instances will get confused - # and complain when they see the logcontext arbitrarily swapping - # between the sentinel and `run` logcontexts. - with PreserveLoggingContext(): - logger.info("Running") - change_resource_limit(config.soft_file_limit) - if config.gc_thresholds: - gc.set_threshold(*config.gc_thresholds) - reactor.run() - def start(): ps.get_datastore().start_profiling() ps.get_state_handler().start_caching() reactor.callWhenRunning(start) - if config.worker_daemonize: - daemon = Daemonize( - app="synapse-appservice", - pid=config.worker_pid_file, - action=run, - auto_close_fds=False, - verbose=True, - logger=logger, - ) - daemon.start() - else: - run() + _base.start_worker_reactor("synapse-appservice", config) if __name__ == '__main__': diff --git a/synapse/app/client_reader.py b/synapse/app/client_reader.py index 09bc1935f..398bb3660 100644 --- a/synapse/app/client_reader.py +++ b/synapse/app/client_reader.py @@ -13,46 +13,46 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. +import logging +import sys + +from twisted.internet import reactor +from twisted.web.resource import NoResource import synapse - +from synapse import events +from synapse.app import _base from synapse.config._base import ConfigError from synapse.config.homeserver import HomeServerConfig from synapse.config.logger import setup_logging -from synapse.http.site import SynapseSite +from synapse.crypto import context_factory from synapse.http.server import JsonResource -from synapse.metrics.resource import MetricsResource, METRICS_PREFIX +from synapse.http.site import SynapseSite +from synapse.metrics import RegistryProxy +from synapse.metrics.resource import METRICS_PREFIX, MetricsResource from synapse.replication.slave.storage._base import BaseSlavedStore from synapse.replication.slave.storage.appservice import SlavedApplicationServiceStore from synapse.replication.slave.storage.client_ips import SlavedClientIpStore +from synapse.replication.slave.storage.directory import DirectoryStore from synapse.replication.slave.storage.events import SlavedEventStore from synapse.replication.slave.storage.keys import SlavedKeyStore -from synapse.replication.slave.storage.room import RoomStore -from synapse.replication.slave.storage.directory import DirectoryStore from synapse.replication.slave.storage.registration import SlavedRegistrationStore +from synapse.replication.slave.storage.room import RoomStore from synapse.replication.slave.storage.transactions import TransactionStore from synapse.replication.tcp.client import ReplicationClientHandler -from synapse.rest.client.v1.room import PublicRoomListRestServlet +from synapse.rest.client.v1.room import ( + JoinedRoomMemberListRestServlet, + PublicRoomListRestServlet, + RoomEventContextServlet, + RoomMemberListRestServlet, + RoomStateRestServlet, +) from synapse.server import HomeServer from synapse.storage.engines import create_engine from synapse.util.httpresourcetree import create_resource_tree -from synapse.util.logcontext import LoggingContext, PreserveLoggingContext +from synapse.util.logcontext import LoggingContext from synapse.util.manhole import manhole -from synapse.util.rlimit import change_resource_limit from synapse.util.versionstring import get_version_string -from synapse.crypto import context_factory - -from synapse import events - - -from twisted.internet import reactor -from twisted.web.resource import Resource - -from daemonize import Daemonize - -import sys -import logging -import gc logger = logging.getLogger("synapse.app.client_reader") @@ -72,19 +72,6 @@ class ClientReaderSlavedStore( class ClientReaderServer(HomeServer): - def get_db_conn(self, run_new_connection=True): - # Any param beginning with cp_ is a parameter for adbapi, and should - # not be passed to the database engine. - db_params = { - k: v for k, v in self.db_config.get("args", {}).items() - if not k.startswith("cp_") - } - db_conn = self.database_engine.module.connect(**db_params) - - if run_new_connection: - self.database_engine.on_new_connection(db_conn) - return db_conn - def setup(self): logger.info("Setting up.") self.datastore = ClientReaderSlavedStore(self.get_db_conn(), self) @@ -98,10 +85,16 @@ class ClientReaderServer(HomeServer): for res in listener_config["resources"]: for name in res["names"]: if name == "metrics": - resources[METRICS_PREFIX] = MetricsResource(self) + resources[METRICS_PREFIX] = MetricsResource(RegistryProxy) elif name == "client": resource = JsonResource(self, canonical_json=False) + PublicRoomListRestServlet(self).register(resource) + RoomMemberListRestServlet(self).register(resource) + JoinedRoomMemberListRestServlet(self).register(resource) + RoomStateRestServlet(self).register(resource) + RoomEventContextServlet(self).register(resource) + resources.update({ "/_matrix/client/r0": resource, "/_matrix/client/unstable": resource, @@ -109,19 +102,19 @@ class ClientReaderServer(HomeServer): "/_matrix/client/api/v1": resource, }) - root_resource = create_resource_tree(resources, Resource()) + root_resource = create_resource_tree(resources, NoResource()) - for address in bind_addresses: - reactor.listenTCP( - port, - SynapseSite( - "synapse.access.http.%s" % (site_tag,), - site_tag, - listener_config, - root_resource, - ), - interface=address + _base.listen_tcp( + bind_addresses, + port, + SynapseSite( + "synapse.access.http.%s" % (site_tag,), + site_tag, + listener_config, + root_resource, + self.version_string, ) + ) logger.info("Synapse client reader now listening on port %d", port) @@ -130,18 +123,22 @@ class ClientReaderServer(HomeServer): if listener["type"] == "http": self._listen_http(listener) elif listener["type"] == "manhole": - bind_addresses = listener["bind_addresses"] - - for address in bind_addresses: - reactor.listenTCP( - listener["port"], - manhole( - username="matrix", - password="rabbithole", - globals={"hs": self}, - ), - interface=address + _base.listen_tcp( + listener["bind_addresses"], + listener["port"], + manhole( + username="matrix", + password="rabbithole", + globals={"hs": self}, ) + ) + elif listener["type"] == "metrics": + if not self.get_config().enable_metrics: + logger.warn(("Metrics listener configured, but " + "enable_metrics is not True!")) + else: + _base.listen_metrics(listener["bind_addresses"], + listener["port"]) else: logger.warn("Unrecognized listener type: %s", listener["type"]) @@ -180,39 +177,15 @@ def start(config_options): ) ss.setup() - ss.get_handlers() ss.start_listening(config.worker_listeners) - def run(): - # make sure that we run the reactor with the sentinel log context, - # otherwise other PreserveLoggingContext instances will get confused - # and complain when they see the logcontext arbitrarily swapping - # between the sentinel and `run` logcontexts. - with PreserveLoggingContext(): - logger.info("Running") - change_resource_limit(config.soft_file_limit) - if config.gc_thresholds: - gc.set_threshold(*config.gc_thresholds) - reactor.run() - def start(): ss.get_state_handler().start_caching() ss.get_datastore().start_profiling() reactor.callWhenRunning(start) - if config.worker_daemonize: - daemon = Daemonize( - app="synapse-client-reader", - pid=config.worker_pid_file, - action=run, - auto_close_fds=False, - verbose=True, - logger=logger, - ) - daemon.start() - else: - run() + _base.start_worker_reactor("synapse-client-reader", config) if __name__ == '__main__': diff --git a/synapse/app/event_creator.py b/synapse/app/event_creator.py new file mode 100644 index 000000000..374f11564 --- /dev/null +++ b/synapse/app/event_creator.py @@ -0,0 +1,201 @@ +#!/usr/bin/env python +# -*- coding: utf-8 -*- +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import logging +import sys + +from twisted.internet import reactor +from twisted.web.resource import NoResource + +import synapse +from synapse import events +from synapse.app import _base +from synapse.config._base import ConfigError +from synapse.config.homeserver import HomeServerConfig +from synapse.config.logger import setup_logging +from synapse.crypto import context_factory +from synapse.http.server import JsonResource +from synapse.http.site import SynapseSite +from synapse.metrics import RegistryProxy +from synapse.metrics.resource import METRICS_PREFIX, MetricsResource +from synapse.replication.slave.storage._base import BaseSlavedStore +from synapse.replication.slave.storage.account_data import SlavedAccountDataStore +from synapse.replication.slave.storage.appservice import SlavedApplicationServiceStore +from synapse.replication.slave.storage.client_ips import SlavedClientIpStore +from synapse.replication.slave.storage.devices import SlavedDeviceStore +from synapse.replication.slave.storage.directory import DirectoryStore +from synapse.replication.slave.storage.events import SlavedEventStore +from synapse.replication.slave.storage.profile import SlavedProfileStore +from synapse.replication.slave.storage.push_rule import SlavedPushRuleStore +from synapse.replication.slave.storage.pushers import SlavedPusherStore +from synapse.replication.slave.storage.receipts import SlavedReceiptsStore +from synapse.replication.slave.storage.registration import SlavedRegistrationStore +from synapse.replication.slave.storage.room import RoomStore +from synapse.replication.slave.storage.transactions import TransactionStore +from synapse.replication.tcp.client import ReplicationClientHandler +from synapse.rest.client.v1.room import ( + JoinRoomAliasServlet, + RoomMembershipRestServlet, + RoomSendEventRestServlet, + RoomStateEventRestServlet, +) +from synapse.server import HomeServer +from synapse.storage.engines import create_engine +from synapse.util.httpresourcetree import create_resource_tree +from synapse.util.logcontext import LoggingContext +from synapse.util.manhole import manhole +from synapse.util.versionstring import get_version_string + +logger = logging.getLogger("synapse.app.event_creator") + + +class EventCreatorSlavedStore( + DirectoryStore, + TransactionStore, + SlavedProfileStore, + SlavedAccountDataStore, + SlavedPusherStore, + SlavedReceiptsStore, + SlavedPushRuleStore, + SlavedDeviceStore, + SlavedClientIpStore, + SlavedApplicationServiceStore, + SlavedEventStore, + SlavedRegistrationStore, + RoomStore, + BaseSlavedStore, +): + pass + + +class EventCreatorServer(HomeServer): + def setup(self): + logger.info("Setting up.") + self.datastore = EventCreatorSlavedStore(self.get_db_conn(), self) + logger.info("Finished setting up.") + + def _listen_http(self, listener_config): + port = listener_config["port"] + bind_addresses = listener_config["bind_addresses"] + site_tag = listener_config.get("tag", port) + resources = {} + for res in listener_config["resources"]: + for name in res["names"]: + if name == "metrics": + resources[METRICS_PREFIX] = MetricsResource(RegistryProxy) + elif name == "client": + resource = JsonResource(self, canonical_json=False) + RoomSendEventRestServlet(self).register(resource) + RoomMembershipRestServlet(self).register(resource) + RoomStateEventRestServlet(self).register(resource) + JoinRoomAliasServlet(self).register(resource) + resources.update({ + "/_matrix/client/r0": resource, + "/_matrix/client/unstable": resource, + "/_matrix/client/v2_alpha": resource, + "/_matrix/client/api/v1": resource, + }) + + root_resource = create_resource_tree(resources, NoResource()) + + _base.listen_tcp( + bind_addresses, + port, + SynapseSite( + "synapse.access.http.%s" % (site_tag,), + site_tag, + listener_config, + root_resource, + self.version_string, + ) + ) + + logger.info("Synapse event creator now listening on port %d", port) + + def start_listening(self, listeners): + for listener in listeners: + if listener["type"] == "http": + self._listen_http(listener) + elif listener["type"] == "manhole": + _base.listen_tcp( + listener["bind_addresses"], + listener["port"], + manhole( + username="matrix", + password="rabbithole", + globals={"hs": self}, + ) + ) + elif listener["type"] == "metrics": + if not self.get_config().enable_metrics: + logger.warn(("Metrics listener configured, but " + "enable_metrics is not True!")) + else: + _base.listen_metrics(listener["bind_addresses"], + listener["port"]) + else: + logger.warn("Unrecognized listener type: %s", listener["type"]) + + self.get_tcp_replication().start_replication(self) + + def build_tcp_replication(self): + return ReplicationClientHandler(self.get_datastore()) + + +def start(config_options): + try: + config = HomeServerConfig.load_config( + "Synapse event creator", config_options + ) + except ConfigError as e: + sys.stderr.write("\n" + e.message + "\n") + sys.exit(1) + + assert config.worker_app == "synapse.app.event_creator" + + assert config.worker_replication_http_port is not None + + setup_logging(config, use_worker_options=True) + + events.USE_FROZEN_DICTS = config.use_frozen_dicts + + database_engine = create_engine(config.database_config) + + tls_server_context_factory = context_factory.ServerContextFactory(config) + + ss = EventCreatorServer( + config.server_name, + db_config=config.database_config, + tls_server_context_factory=tls_server_context_factory, + config=config, + version_string="Synapse/" + get_version_string(synapse), + database_engine=database_engine, + ) + + ss.setup() + ss.start_listening(config.worker_listeners) + + def start(): + ss.get_state_handler().start_caching() + ss.get_datastore().start_profiling() + + reactor.callWhenRunning(start) + + _base.start_worker_reactor("synapse-event-creator", config) + + +if __name__ == '__main__': + with LoggingContext("main"): + start(sys.argv[1:]) diff --git a/synapse/app/federation_reader.py b/synapse/app/federation_reader.py index eb392e1c9..7af00b8bc 100644 --- a/synapse/app/federation_reader.py +++ b/synapse/app/federation_reader.py @@ -13,43 +13,37 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. +import logging +import sys + +from twisted.internet import reactor +from twisted.web.resource import NoResource import synapse - +from synapse import events +from synapse.api.urls import FEDERATION_PREFIX +from synapse.app import _base from synapse.config._base import ConfigError from synapse.config.homeserver import HomeServerConfig from synapse.config.logger import setup_logging +from synapse.crypto import context_factory +from synapse.federation.transport.server import TransportLayerServer from synapse.http.site import SynapseSite -from synapse.metrics.resource import MetricsResource, METRICS_PREFIX +from synapse.metrics import RegistryProxy +from synapse.metrics.resource import METRICS_PREFIX, MetricsResource from synapse.replication.slave.storage._base import BaseSlavedStore +from synapse.replication.slave.storage.directory import DirectoryStore from synapse.replication.slave.storage.events import SlavedEventStore from synapse.replication.slave.storage.keys import SlavedKeyStore from synapse.replication.slave.storage.room import RoomStore from synapse.replication.slave.storage.transactions import TransactionStore -from synapse.replication.slave.storage.directory import DirectoryStore from synapse.replication.tcp.client import ReplicationClientHandler from synapse.server import HomeServer from synapse.storage.engines import create_engine from synapse.util.httpresourcetree import create_resource_tree -from synapse.util.logcontext import LoggingContext, PreserveLoggingContext +from synapse.util.logcontext import LoggingContext from synapse.util.manhole import manhole -from synapse.util.rlimit import change_resource_limit from synapse.util.versionstring import get_version_string -from synapse.api.urls import FEDERATION_PREFIX -from synapse.federation.transport.server import TransportLayerServer -from synapse.crypto import context_factory - -from synapse import events - - -from twisted.internet import reactor -from twisted.web.resource import Resource - -from daemonize import Daemonize - -import sys -import logging -import gc logger = logging.getLogger("synapse.app.federation_reader") @@ -66,19 +60,6 @@ class FederationReaderSlavedStore( class FederationReaderServer(HomeServer): - def get_db_conn(self, run_new_connection=True): - # Any param beginning with cp_ is a parameter for adbapi, and should - # not be passed to the database engine. - db_params = { - k: v for k, v in self.db_config.get("args", {}).items() - if not k.startswith("cp_") - } - db_conn = self.database_engine.module.connect(**db_params) - - if run_new_connection: - self.database_engine.on_new_connection(db_conn) - return db_conn - def setup(self): logger.info("Setting up.") self.datastore = FederationReaderSlavedStore(self.get_db_conn(), self) @@ -92,25 +73,25 @@ class FederationReaderServer(HomeServer): for res in listener_config["resources"]: for name in res["names"]: if name == "metrics": - resources[METRICS_PREFIX] = MetricsResource(self) + resources[METRICS_PREFIX] = MetricsResource(RegistryProxy) elif name == "federation": resources.update({ FEDERATION_PREFIX: TransportLayerServer(self), }) - root_resource = create_resource_tree(resources, Resource()) + root_resource = create_resource_tree(resources, NoResource()) - for address in bind_addresses: - reactor.listenTCP( - port, - SynapseSite( - "synapse.access.http.%s" % (site_tag,), - site_tag, - listener_config, - root_resource, - ), - interface=address + _base.listen_tcp( + bind_addresses, + port, + SynapseSite( + "synapse.access.http.%s" % (site_tag,), + site_tag, + listener_config, + root_resource, + self.version_string, ) + ) logger.info("Synapse federation reader now listening on port %d", port) @@ -119,18 +100,22 @@ class FederationReaderServer(HomeServer): if listener["type"] == "http": self._listen_http(listener) elif listener["type"] == "manhole": - bind_addresses = listener["bind_addresses"] - - for address in bind_addresses: - reactor.listenTCP( - listener["port"], - manhole( - username="matrix", - password="rabbithole", - globals={"hs": self}, - ), - interface=address + _base.listen_tcp( + listener["bind_addresses"], + listener["port"], + manhole( + username="matrix", + password="rabbithole", + globals={"hs": self}, ) + ) + elif listener["type"] == "metrics": + if not self.get_config().enable_metrics: + logger.warn(("Metrics listener configured, but " + "enable_metrics is not True!")) + else: + _base.listen_metrics(listener["bind_addresses"], + listener["port"]) else: logger.warn("Unrecognized listener type: %s", listener["type"]) @@ -169,39 +154,15 @@ def start(config_options): ) ss.setup() - ss.get_handlers() ss.start_listening(config.worker_listeners) - def run(): - # make sure that we run the reactor with the sentinel log context, - # otherwise other PreserveLoggingContext instances will get confused - # and complain when they see the logcontext arbitrarily swapping - # between the sentinel and `run` logcontexts. - with PreserveLoggingContext(): - logger.info("Running") - change_resource_limit(config.soft_file_limit) - if config.gc_thresholds: - gc.set_threshold(*config.gc_thresholds) - reactor.run() - def start(): ss.get_state_handler().start_caching() ss.get_datastore().start_profiling() reactor.callWhenRunning(start) - if config.worker_daemonize: - daemon = Daemonize( - app="synapse-federation-reader", - pid=config.worker_pid_file, - action=run, - auto_close_fds=False, - verbose=True, - logger=logger, - ) - daemon.start() - else: - run() + _base.start_worker_reactor("synapse-federation-reader", config) if __name__ == '__main__': diff --git a/synapse/app/federation_sender.py b/synapse/app/federation_sender.py index 03327dc47..18469013f 100644 --- a/synapse/app/federation_sender.py +++ b/synapse/app/federation_sender.py @@ -13,44 +13,39 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. +import logging +import sys + +from twisted.internet import defer, reactor +from twisted.web.resource import NoResource import synapse - -from synapse.server import HomeServer +from synapse import events +from synapse.app import _base from synapse.config._base import ConfigError -from synapse.config.logger import setup_logging from synapse.config.homeserver import HomeServerConfig +from synapse.config.logger import setup_logging from synapse.crypto import context_factory -from synapse.http.site import SynapseSite from synapse.federation import send_queue -from synapse.metrics.resource import MetricsResource, METRICS_PREFIX +from synapse.http.site import SynapseSite +from synapse.metrics import RegistryProxy +from synapse.metrics.resource import METRICS_PREFIX, MetricsResource from synapse.replication.slave.storage.deviceinbox import SlavedDeviceInboxStore +from synapse.replication.slave.storage.devices import SlavedDeviceStore from synapse.replication.slave.storage.events import SlavedEventStore +from synapse.replication.slave.storage.presence import SlavedPresenceStore from synapse.replication.slave.storage.receipts import SlavedReceiptsStore from synapse.replication.slave.storage.registration import SlavedRegistrationStore -from synapse.replication.slave.storage.presence import SlavedPresenceStore from synapse.replication.slave.storage.transactions import TransactionStore -from synapse.replication.slave.storage.devices import SlavedDeviceStore from synapse.replication.tcp.client import ReplicationClientHandler +from synapse.server import HomeServer from synapse.storage.engines import create_engine from synapse.util.async import Linearizer from synapse.util.httpresourcetree import create_resource_tree -from synapse.util.logcontext import LoggingContext, PreserveLoggingContext, preserve_fn +from synapse.util.logcontext import LoggingContext, run_in_background from synapse.util.manhole import manhole -from synapse.util.rlimit import change_resource_limit from synapse.util.versionstring import get_version_string -from synapse import events - -from twisted.internet import reactor, defer -from twisted.web.resource import Resource - -from daemonize import Daemonize - -import sys -import logging -import gc - logger = logging.getLogger("synapse.app.federation_sender") @@ -83,19 +78,6 @@ class FederationSenderSlaveStore( class FederationSenderServer(HomeServer): - def get_db_conn(self, run_new_connection=True): - # Any param beginning with cp_ is a parameter for adbapi, and should - # not be passed to the database engine. - db_params = { - k: v for k, v in self.db_config.get("args", {}).items() - if not k.startswith("cp_") - } - db_conn = self.database_engine.module.connect(**db_params) - - if run_new_connection: - self.database_engine.on_new_connection(db_conn) - return db_conn - def setup(self): logger.info("Setting up.") self.datastore = FederationSenderSlaveStore(self.get_db_conn(), self) @@ -109,21 +91,21 @@ class FederationSenderServer(HomeServer): for res in listener_config["resources"]: for name in res["names"]: if name == "metrics": - resources[METRICS_PREFIX] = MetricsResource(self) + resources[METRICS_PREFIX] = MetricsResource(RegistryProxy) - root_resource = create_resource_tree(resources, Resource()) + root_resource = create_resource_tree(resources, NoResource()) - for address in bind_addresses: - reactor.listenTCP( - port, - SynapseSite( - "synapse.access.http.%s" % (site_tag,), - site_tag, - listener_config, - root_resource, - ), - interface=address + _base.listen_tcp( + bind_addresses, + port, + SynapseSite( + "synapse.access.http.%s" % (site_tag,), + site_tag, + listener_config, + root_resource, + self.version_string, ) + ) logger.info("Synapse federation_sender now listening on port %d", port) @@ -132,18 +114,22 @@ class FederationSenderServer(HomeServer): if listener["type"] == "http": self._listen_http(listener) elif listener["type"] == "manhole": - bind_addresses = listener["bind_addresses"] - - for address in bind_addresses: - reactor.listenTCP( - listener["port"], - manhole( - username="matrix", - password="rabbithole", - globals={"hs": self}, - ), - interface=address + _base.listen_tcp( + listener["bind_addresses"], + listener["port"], + manhole( + username="matrix", + password="rabbithole", + globals={"hs": self}, ) + ) + elif listener["type"] == "metrics": + if not self.get_config().enable_metrics: + logger.warn(("Metrics listener configured, but " + "enable_metrics is not True!")) + else: + _base.listen_metrics(listener["bind_addresses"], + listener["port"]) else: logger.warn("Unrecognized listener type: %s", listener["type"]) @@ -213,36 +199,12 @@ def start(config_options): ps.setup() ps.start_listening(config.worker_listeners) - def run(): - # make sure that we run the reactor with the sentinel log context, - # otherwise other PreserveLoggingContext instances will get confused - # and complain when they see the logcontext arbitrarily swapping - # between the sentinel and `run` logcontexts. - with PreserveLoggingContext(): - logger.info("Running") - change_resource_limit(config.soft_file_limit) - if config.gc_thresholds: - gc.set_threshold(*config.gc_thresholds) - reactor.run() - def start(): ps.get_datastore().start_profiling() ps.get_state_handler().start_caching() reactor.callWhenRunning(start) - - if config.worker_daemonize: - daemon = Daemonize( - app="synapse-federation-sender", - pid=config.worker_pid_file, - action=run, - auto_close_fds=False, - verbose=True, - logger=logger, - ) - daemon.start() - else: - run() + _base.start_worker_reactor("synapse-federation-sender", config) class FederationSenderHandler(object): @@ -277,7 +239,7 @@ class FederationSenderHandler(object): # presence, typing, etc. if stream_name == "federation": send_queue.process_rows_for_federation(self.federation_sender, rows) - preserve_fn(self.update_token)(token) + run_in_background(self.update_token, token) # We also need to poke the federation sender when new events happen elif stream_name == "events": @@ -285,19 +247,22 @@ class FederationSenderHandler(object): @defer.inlineCallbacks def update_token(self, token): - self.federation_position = token + try: + self.federation_position = token - # We linearize here to ensure we don't have races updating the token - with (yield self._fed_position_linearizer.queue(None)): - if self._last_ack < self.federation_position: - yield self.store.update_federation_out_pos( - "federation", self.federation_position - ) + # We linearize here to ensure we don't have races updating the token + with (yield self._fed_position_linearizer.queue(None)): + if self._last_ack < self.federation_position: + yield self.store.update_federation_out_pos( + "federation", self.federation_position + ) - # We ACK this token over replication so that the master can drop - # its in memory queues - self.replication_client.send_federation_ack(self.federation_position) - self._last_ack = self.federation_position + # We ACK this token over replication so that the master can drop + # its in memory queues + self.replication_client.send_federation_ack(self.federation_position) + self._last_ack = self.federation_position + except Exception: + logger.exception("Error updating federation stream position") if __name__ == '__main__': diff --git a/synapse/app/frontend_proxy.py b/synapse/app/frontend_proxy.py new file mode 100644 index 000000000..b5f78f464 --- /dev/null +++ b/synapse/app/frontend_proxy.py @@ -0,0 +1,235 @@ +#!/usr/bin/env python +# -*- coding: utf-8 -*- +# Copyright 2016 OpenMarket Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import logging +import sys + +from twisted.internet import defer, reactor +from twisted.web.resource import NoResource + +import synapse +from synapse import events +from synapse.api.errors import SynapseError +from synapse.app import _base +from synapse.config._base import ConfigError +from synapse.config.homeserver import HomeServerConfig +from synapse.config.logger import setup_logging +from synapse.crypto import context_factory +from synapse.http.server import JsonResource +from synapse.http.servlet import RestServlet, parse_json_object_from_request +from synapse.http.site import SynapseSite +from synapse.metrics import RegistryProxy +from synapse.metrics.resource import METRICS_PREFIX, MetricsResource +from synapse.replication.slave.storage._base import BaseSlavedStore +from synapse.replication.slave.storage.appservice import SlavedApplicationServiceStore +from synapse.replication.slave.storage.client_ips import SlavedClientIpStore +from synapse.replication.slave.storage.devices import SlavedDeviceStore +from synapse.replication.slave.storage.registration import SlavedRegistrationStore +from synapse.replication.tcp.client import ReplicationClientHandler +from synapse.rest.client.v2_alpha._base import client_v2_patterns +from synapse.server import HomeServer +from synapse.storage.engines import create_engine +from synapse.util.httpresourcetree import create_resource_tree +from synapse.util.logcontext import LoggingContext +from synapse.util.manhole import manhole +from synapse.util.versionstring import get_version_string + +logger = logging.getLogger("synapse.app.frontend_proxy") + + +class KeyUploadServlet(RestServlet): + PATTERNS = client_v2_patterns("/keys/upload(/(?P[^/]+))?$") + + def __init__(self, hs): + """ + Args: + hs (synapse.server.HomeServer): server + """ + super(KeyUploadServlet, self).__init__() + self.auth = hs.get_auth() + self.store = hs.get_datastore() + self.http_client = hs.get_simple_http_client() + self.main_uri = hs.config.worker_main_http_uri + + @defer.inlineCallbacks + def on_POST(self, request, device_id): + requester = yield self.auth.get_user_by_req(request, allow_guest=True) + user_id = requester.user.to_string() + body = parse_json_object_from_request(request) + + if device_id is not None: + # passing the device_id here is deprecated; however, we allow it + # for now for compatibility with older clients. + if (requester.device_id is not None and + device_id != requester.device_id): + logger.warning("Client uploading keys for a different device " + "(logged in as %s, uploading for %s)", + requester.device_id, device_id) + else: + device_id = requester.device_id + + if device_id is None: + raise SynapseError( + 400, + "To upload keys, you must pass device_id when authenticating" + ) + + if body: + # They're actually trying to upload something, proxy to main synapse. + # Pass through the auth headers, if any, in case the access token + # is there. + auth_headers = request.requestHeaders.getRawHeaders(b"Authorization", []) + headers = { + "Authorization": auth_headers, + } + result = yield self.http_client.post_json_get_json( + self.main_uri + request.uri, + body, + headers=headers, + ) + + defer.returnValue((200, result)) + else: + # Just interested in counts. + result = yield self.store.count_e2e_one_time_keys(user_id, device_id) + defer.returnValue((200, {"one_time_key_counts": result})) + + +class FrontendProxySlavedStore( + SlavedDeviceStore, + SlavedClientIpStore, + SlavedApplicationServiceStore, + SlavedRegistrationStore, + BaseSlavedStore, +): + pass + + +class FrontendProxyServer(HomeServer): + def setup(self): + logger.info("Setting up.") + self.datastore = FrontendProxySlavedStore(self.get_db_conn(), self) + logger.info("Finished setting up.") + + def _listen_http(self, listener_config): + port = listener_config["port"] + bind_addresses = listener_config["bind_addresses"] + site_tag = listener_config.get("tag", port) + resources = {} + for res in listener_config["resources"]: + for name in res["names"]: + if name == "metrics": + resources[METRICS_PREFIX] = MetricsResource(RegistryProxy) + elif name == "client": + resource = JsonResource(self, canonical_json=False) + KeyUploadServlet(self).register(resource) + resources.update({ + "/_matrix/client/r0": resource, + "/_matrix/client/unstable": resource, + "/_matrix/client/v2_alpha": resource, + "/_matrix/client/api/v1": resource, + }) + + root_resource = create_resource_tree(resources, NoResource()) + + _base.listen_tcp( + bind_addresses, + port, + SynapseSite( + "synapse.access.http.%s" % (site_tag,), + site_tag, + listener_config, + root_resource, + self.version_string, + ) + ) + + logger.info("Synapse client reader now listening on port %d", port) + + def start_listening(self, listeners): + for listener in listeners: + if listener["type"] == "http": + self._listen_http(listener) + elif listener["type"] == "manhole": + _base.listen_tcp( + listener["bind_addresses"], + listener["port"], + manhole( + username="matrix", + password="rabbithole", + globals={"hs": self}, + ) + ) + elif listener["type"] == "metrics": + if not self.get_config().enable_metrics: + logger.warn(("Metrics listener configured, but " + "enable_metrics is not True!")) + else: + _base.listen_metrics(listener["bind_addresses"], + listener["port"]) + else: + logger.warn("Unrecognized listener type: %s", listener["type"]) + + self.get_tcp_replication().start_replication(self) + + def build_tcp_replication(self): + return ReplicationClientHandler(self.get_datastore()) + + +def start(config_options): + try: + config = HomeServerConfig.load_config( + "Synapse frontend proxy", config_options + ) + except ConfigError as e: + sys.stderr.write("\n" + e.message + "\n") + sys.exit(1) + + assert config.worker_app == "synapse.app.frontend_proxy" + + assert config.worker_main_http_uri is not None + + setup_logging(config, use_worker_options=True) + + events.USE_FROZEN_DICTS = config.use_frozen_dicts + + database_engine = create_engine(config.database_config) + + tls_server_context_factory = context_factory.ServerContextFactory(config) + + ss = FrontendProxyServer( + config.server_name, + db_config=config.database_config, + tls_server_context_factory=tls_server_context_factory, + config=config, + version_string="Synapse/" + get_version_string(synapse), + database_engine=database_engine, + ) + + ss.setup() + ss.start_listening(config.worker_listeners) + + def start(): + ss.get_state_handler().start_caching() + ss.get_datastore().start_profiling() + + reactor.callWhenRunning(start) + + _base.start_worker_reactor("synapse-frontend-proxy", config) + + +if __name__ == '__main__': + with LoggingContext("main"): + start(sys.argv[1:]) diff --git a/synapse/app/homeserver.py b/synapse/app/homeserver.py index 081e7cce5..2ad1beb8d 100755 --- a/synapse/app/homeserver.py +++ b/synapse/app/homeserver.py @@ -13,61 +13,62 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. - -import synapse - import gc import logging import os import sys -import synapse.config.logger -from synapse.config._base import ConfigError +from six import iteritems -from synapse.python_dependencies import ( - check_requirements, CONDITIONAL_REQUIREMENTS -) - -from synapse.rest import ClientRestResource -from synapse.storage.engines import create_engine, IncorrectDatabaseSetup -from synapse.storage import are_all_users_on_domain -from synapse.storage.prepare_database import UpgradeDatabaseException, prepare_database - -from synapse.server import HomeServer - -from twisted.internet import reactor, defer from twisted.application import service -from twisted.web.resource import Resource, EncodingResourceWrapper -from twisted.web.static import File +from twisted.internet import defer, reactor +from twisted.web.resource import EncodingResourceWrapper, NoResource from twisted.web.server import GzipEncoderFactory -from synapse.http.server import RootRedirect -from synapse.rest.media.v0.content_repository import ContentRepoResource -from synapse.rest.media.v1.media_repository import MediaRepositoryResource -from synapse.rest.key.v1.server_key_resource import LocalKey -from synapse.rest.key.v2 import KeyApiV2Resource +from twisted.web.static import File + +import synapse +import synapse.config.logger +from synapse import events from synapse.api.urls import ( - FEDERATION_PREFIX, WEB_CLIENT_PREFIX, CONTENT_REPO_PREFIX, - SERVER_KEY_PREFIX, LEGACY_MEDIA_PREFIX, MEDIA_PREFIX, STATIC_PREFIX, + CONTENT_REPO_PREFIX, + FEDERATION_PREFIX, + LEGACY_MEDIA_PREFIX, + MEDIA_PREFIX, + SERVER_KEY_PREFIX, SERVER_KEY_V2_PREFIX, + STATIC_PREFIX, + WEB_CLIENT_PREFIX, ) +from synapse.app import _base +from synapse.app._base import listen_ssl, listen_tcp, quit_with_error +from synapse.config._base import ConfigError from synapse.config.homeserver import HomeServerConfig from synapse.crypto import context_factory -from synapse.util.logcontext import LoggingContext, PreserveLoggingContext -from synapse.metrics import register_memory_metrics -from synapse.metrics.resource import MetricsResource, METRICS_PREFIX -from synapse.replication.tcp.resource import ReplicationStreamProtocolFactory from synapse.federation.transport.server import TransportLayerServer - +from synapse.http.additional_resource import AdditionalResource +from synapse.http.server import RootRedirect +from synapse.http.site import SynapseSite +from synapse.metrics import RegistryProxy +from synapse.metrics.resource import METRICS_PREFIX, MetricsResource +from synapse.module_api import ModuleApi +from synapse.python_dependencies import CONDITIONAL_REQUIREMENTS, check_requirements +from synapse.replication.http import REPLICATION_PREFIX, ReplicationRestResource +from synapse.replication.tcp.resource import ReplicationStreamProtocolFactory +from synapse.rest import ClientRestResource +from synapse.rest.key.v1.server_key_resource import LocalKey +from synapse.rest.key.v2 import KeyApiV2Resource +from synapse.rest.media.v0.content_repository import ContentRepoResource +from synapse.server import HomeServer +from synapse.storage import are_all_users_on_domain +from synapse.storage.engines import IncorrectDatabaseSetup, create_engine +from synapse.storage.prepare_database import UpgradeDatabaseException, prepare_database +from synapse.util.caches import CACHE_SIZE_FACTOR +from synapse.util.httpresourcetree import create_resource_tree +from synapse.util.logcontext import LoggingContext +from synapse.util.manhole import manhole +from synapse.util.module_loader import load_module from synapse.util.rlimit import change_resource_limit from synapse.util.versionstring import get_version_string -from synapse.util.httpresourcetree import create_resource_tree -from synapse.util.manhole import manhole - -from synapse.http.site import SynapseSite - -from synapse import events - -from daemonize import Daemonize logger = logging.getLogger("synapse.app.homeserver") @@ -119,87 +120,132 @@ class SynapseHomeServer(HomeServer): resources = {} for res in listener_config["resources"]: for name in res["names"]: - if name == "client": - client_resource = ClientRestResource(self) - if res["compress"]: - client_resource = gz_wrap(client_resource) + resources.update(self._configure_named_resource( + name, res.get("compress", False), + )) - resources.update({ - "/_matrix/client/api/v1": client_resource, - "/_matrix/client/r0": client_resource, - "/_matrix/client/unstable": client_resource, - "/_matrix/client/v2_alpha": client_resource, - "/_matrix/client/versions": client_resource, - }) - - if name == "federation": - resources.update({ - FEDERATION_PREFIX: TransportLayerServer(self), - }) - - if name in ["static", "client"]: - resources.update({ - STATIC_PREFIX: File( - os.path.join(os.path.dirname(synapse.__file__), "static") - ), - }) - - if name in ["media", "federation", "client"]: - media_repo = MediaRepositoryResource(self) - resources.update({ - MEDIA_PREFIX: media_repo, - LEGACY_MEDIA_PREFIX: media_repo, - CONTENT_REPO_PREFIX: ContentRepoResource( - self, self.config.uploads_path - ), - }) - - if name in ["keys", "federation"]: - resources.update({ - SERVER_KEY_PREFIX: LocalKey(self), - SERVER_KEY_V2_PREFIX: KeyApiV2Resource(self), - }) - - if name == "webclient": - resources[WEB_CLIENT_PREFIX] = build_resource_for_web_client(self) - - if name == "metrics" and self.get_config().enable_metrics: - resources[METRICS_PREFIX] = MetricsResource(self) + additional_resources = listener_config.get("additional_resources", {}) + logger.debug("Configuring additional resources: %r", + additional_resources) + module_api = ModuleApi(self, self.get_auth_handler()) + for path, resmodule in additional_resources.items(): + handler_cls, config = load_module(resmodule) + handler = handler_cls(config, module_api) + resources[path] = AdditionalResource(self, handler.handle_request) if WEB_CLIENT_PREFIX in resources: root_resource = RootRedirect(WEB_CLIENT_PREFIX) else: - root_resource = Resource() + root_resource = NoResource() root_resource = create_resource_tree(resources, root_resource) if tls: - for address in bind_addresses: - reactor.listenSSL( - port, - SynapseSite( - "synapse.access.https.%s" % (site_tag,), - site_tag, - listener_config, - root_resource, - ), - self.tls_server_context_factory, - interface=address - ) + listen_ssl( + bind_addresses, + port, + SynapseSite( + "synapse.access.https.%s" % (site_tag,), + site_tag, + listener_config, + root_resource, + self.version_string, + ), + self.tls_server_context_factory, + ) + else: - for address in bind_addresses: - reactor.listenTCP( - port, - SynapseSite( - "synapse.access.http.%s" % (site_tag,), - site_tag, - listener_config, - root_resource, - ), - interface=address + listen_tcp( + bind_addresses, + port, + SynapseSite( + "synapse.access.http.%s" % (site_tag,), + site_tag, + listener_config, + root_resource, + self.version_string, ) + ) logger.info("Synapse now listening on port %d", port) + def _configure_named_resource(self, name, compress=False): + """Build a resource map for a named resource + + Args: + name (str): named resource: one of "client", "federation", etc + compress (bool): whether to enable gzip compression for this + resource + + Returns: + dict[str, Resource]: map from path to HTTP resource + """ + resources = {} + if name == "client": + client_resource = ClientRestResource(self) + if compress: + client_resource = gz_wrap(client_resource) + + resources.update({ + "/_matrix/client/api/v1": client_resource, + "/_matrix/client/r0": client_resource, + "/_matrix/client/unstable": client_resource, + "/_matrix/client/v2_alpha": client_resource, + "/_matrix/client/versions": client_resource, + }) + + if name == "consent": + from synapse.rest.consent.consent_resource import ConsentResource + consent_resource = ConsentResource(self) + if compress: + consent_resource = gz_wrap(consent_resource) + resources.update({ + "/_matrix/consent": consent_resource, + }) + + if name == "federation": + resources.update({ + FEDERATION_PREFIX: TransportLayerServer(self), + }) + + if name in ["static", "client"]: + resources.update({ + STATIC_PREFIX: File( + os.path.join(os.path.dirname(synapse.__file__), "static") + ), + }) + + if name in ["media", "federation", "client"]: + if self.get_config().enable_media_repo: + media_repo = self.get_media_repository_resource() + resources.update({ + MEDIA_PREFIX: media_repo, + LEGACY_MEDIA_PREFIX: media_repo, + CONTENT_REPO_PREFIX: ContentRepoResource( + self, self.config.uploads_path + ), + }) + elif name == "media": + raise ConfigError( + "'media' resource conflicts with enable_media_repo=False", + ) + + if name in ["keys", "federation"]: + resources.update({ + SERVER_KEY_PREFIX: LocalKey(self), + SERVER_KEY_V2_PREFIX: KeyApiV2Resource(self), + }) + + if name == "webclient": + resources[WEB_CLIENT_PREFIX] = build_resource_for_web_client(self) + + if name == "metrics" and self.get_config().enable_metrics: + resources[METRICS_PREFIX] = MetricsResource(RegistryProxy) + + if name == "replication": + resources[REPLICATION_PREFIX] = ReplicationRestResource(self) + + return resources + def start_listening(self): config = self.get_config() @@ -207,18 +253,15 @@ class SynapseHomeServer(HomeServer): if listener["type"] == "http": self._listener_http(config, listener) elif listener["type"] == "manhole": - bind_addresses = listener["bind_addresses"] - - for address in bind_addresses: - reactor.listenTCP( - listener["port"], - manhole( - username="matrix", - password="rabbithole", - globals={"hs": self}, - ), - interface=address + listen_tcp( + listener["bind_addresses"], + listener["port"], + manhole( + username="matrix", + password="rabbithole", + globals={"hs": self}, ) + ) elif listener["type"] == "replication": bind_addresses = listener["bind_addresses"] for address in bind_addresses: @@ -229,6 +272,13 @@ class SynapseHomeServer(HomeServer): reactor.addSystemEventTrigger( "before", "shutdown", server_listener.stopListening, ) + elif listener["type"] == "metrics": + if not self.get_config().enable_metrics: + logger.warn(("Metrics listener configured, but " + "enable_metrics is not True!")) + else: + _base.listen_metrics(listener["bind_addresses"], + listener["port"]) else: logger.warn("Unrecognized listener type: %s", listener["type"]) @@ -248,29 +298,6 @@ class SynapseHomeServer(HomeServer): except IncorrectDatabaseSetup as e: quit_with_error(e.message) - def get_db_conn(self, run_new_connection=True): - # Any param beginning with cp_ is a parameter for adbapi, and should - # not be passed to the database engine. - db_params = { - k: v for k, v in self.db_config.get("args", {}).items() - if not k.startswith("cp_") - } - db_conn = self.database_engine.module.connect(**db_params) - - if run_new_connection: - self.database_engine.on_new_connection(db_conn) - return db_conn - - -def quit_with_error(error_string): - message_lines = error_string.split("\n") - line_length = max([len(l) for l in message_lines if len(l) < 80]) + 2 - sys.stderr.write("*" * line_length + '\n') - for line in message_lines: - sys.stderr.write(" %s\n" % (line.rstrip(),)) - sys.stderr.write("*" * line_length + '\n') - sys.exit(1) - def setup(config_options): """ @@ -300,11 +327,6 @@ def setup(config_options): # check any extra requirements we have now we have a config check_requirements(config) - version_string = "Synapse/" + get_version_string(synapse) - - logger.info("Server hostname: %s", config.server_name) - logger.info("Server version: %s", version_string) - events.USE_FROZEN_DICTS = config.use_frozen_dicts tls_server_context_factory = context_factory.ServerContextFactory(config) @@ -317,7 +339,7 @@ def setup(config_options): db_config=config.database_config, tls_server_context_factory=tls_server_context_factory, config=config, - version_string=version_string, + version_string="Synapse/" + get_version_string(synapse), database_engine=database_engine, ) @@ -349,9 +371,7 @@ def setup(config_options): hs.get_state_handler().start_caching() hs.get_datastore().start_profiling() hs.get_datastore().start_doing_background_updates() - hs.get_replication_layer().start_get_pdu_cache() - - register_memory_metrics(hs) + hs.get_federation_client().start_get_pdu_cache() reactor.callWhenRunning(start) @@ -403,6 +423,10 @@ def run(hs): stats = {} + # Contains the list of processes we will be monitoring + # currently either 0 or 1 + stats_process = [] + @defer.inlineCallbacks def phone_stats_home(): logger.info("Gathering stats for reporting") @@ -419,6 +443,10 @@ def run(hs): total_nonbridged_users = yield hs.get_datastore().count_nonbridged_users() stats["total_nonbridged_users"] = total_nonbridged_users + daily_user_type_results = yield hs.get_datastore().count_daily_user_type() + for name, count in iteritems(daily_user_type_results): + stats["daily_user_type_" + name] = count + room_count = yield hs.get_datastore().get_room_count() stats["total_room_count"] = room_count @@ -426,8 +454,21 @@ def run(hs): stats["daily_active_rooms"] = yield hs.get_datastore().count_daily_active_rooms() stats["daily_messages"] = yield hs.get_datastore().count_daily_messages() + r30_results = yield hs.get_datastore().count_r30_users() + for name, count in iteritems(r30_results): + stats["r30_users_" + name] = count + daily_sent_messages = yield hs.get_datastore().count_daily_sent_messages() stats["daily_sent_messages"] = daily_sent_messages + stats["cache_factor"] = CACHE_SIZE_FACTOR + stats["event_cache_size"] = hs.config.event_cache_size + + if len(stats_process) > 0: + stats["memory_rss"] = 0 + stats["cpu_average"] = 0 + for process in stats_process: + stats["memory_rss"] += process.memory_info().rss + stats["cpu_average"] += int(process.cpu_percent(interval=None)) logger.info("Reporting stats to matrix.org: %s" % (stats,)) try: @@ -438,45 +479,56 @@ def run(hs): except Exception as e: logger.warn("Error reporting stats: %s", e) + def performance_stats_init(): + try: + import psutil + process = psutil.Process() + # Ensure we can fetch both, and make the initial request for cpu_percent + # so the next request will use this as the initial point. + process.memory_info().rss + process.cpu_percent(interval=None) + logger.info("report_stats can use psutil") + stats_process.append(process) + except (ImportError, AttributeError): + logger.warn( + "report_stats enabled but psutil is not installed or incorrect version." + " Disabling reporting of memory/cpu stats." + " Ensuring psutil is available will help matrix.org track performance" + " changes across releases." + ) + + def generate_user_daily_visit_stats(): + hs.get_datastore().generate_user_daily_visits() + + # Rather than update on per session basis, batch up the requests. + # If you increase the loop period, the accuracy of user_daily_visits + # table will decrease + clock.looping_call(generate_user_daily_visit_stats, 5 * 60 * 1000) + if hs.config.report_stats: logger.info("Scheduling stats reporting for 3 hour intervals") clock.looping_call(phone_stats_home, 3 * 60 * 60 * 1000) + # We need to defer this init for the cases that we daemonize + # otherwise the process ID we get is that of the non-daemon process + clock.call_later(0, performance_stats_init) + # We wait 5 minutes to send the first set of stats as the server can # be quite busy the first few minutes clock.call_later(5 * 60, phone_stats_home) - def in_thread(): - # Uncomment to enable tracing of log context changes. - # sys.settrace(logcontext_tracer) + if hs.config.daemonize and hs.config.print_pidfile: + print (hs.config.pid_file) - # make sure that we run the reactor with the sentinel log context, - # otherwise other PreserveLoggingContext instances will get confused - # and complain when they see the logcontext arbitrarily swapping - # between the sentinel and `run` logcontexts. - with PreserveLoggingContext(): - change_resource_limit(hs.config.soft_file_limit) - if hs.config.gc_thresholds: - gc.set_threshold(*hs.config.gc_thresholds) - reactor.run() - - if hs.config.daemonize: - - if hs.config.print_pidfile: - print (hs.config.pid_file) - - daemon = Daemonize( - app="synapse-homeserver", - pid=hs.config.pid_file, - action=lambda: in_thread(), - auto_close_fds=False, - verbose=True, - logger=logger, - ) - - daemon.start() - else: - in_thread() + _base.start_reactor( + "synapse-homeserver", + hs.config.soft_file_limit, + hs.config.gc_thresholds, + hs.config.pid_file, + hs.config.daemonize, + hs.config.cpu_affinity, + logger, + ) def main(): diff --git a/synapse/app/media_repository.py b/synapse/app/media_repository.py index f57ec784f..749bbf37d 100644 --- a/synapse/app/media_repository.py +++ b/synapse/app/media_repository.py @@ -13,14 +13,23 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. +import logging +import sys + +from twisted.internet import reactor +from twisted.web.resource import NoResource import synapse - +from synapse import events +from synapse.api.urls import CONTENT_REPO_PREFIX, LEGACY_MEDIA_PREFIX, MEDIA_PREFIX +from synapse.app import _base from synapse.config._base import ConfigError from synapse.config.homeserver import HomeServerConfig from synapse.config.logger import setup_logging +from synapse.crypto import context_factory from synapse.http.site import SynapseSite -from synapse.metrics.resource import MetricsResource, METRICS_PREFIX +from synapse.metrics import RegistryProxy +from synapse.metrics.resource import METRICS_PREFIX, MetricsResource from synapse.replication.slave.storage._base import BaseSlavedStore from synapse.replication.slave.storage.appservice import SlavedApplicationServiceStore from synapse.replication.slave.storage.client_ips import SlavedClientIpStore @@ -28,31 +37,13 @@ from synapse.replication.slave.storage.registration import SlavedRegistrationSto from synapse.replication.slave.storage.transactions import TransactionStore from synapse.replication.tcp.client import ReplicationClientHandler from synapse.rest.media.v0.content_repository import ContentRepoResource -from synapse.rest.media.v1.media_repository import MediaRepositoryResource from synapse.server import HomeServer from synapse.storage.engines import create_engine from synapse.storage.media_repository import MediaRepositoryStore from synapse.util.httpresourcetree import create_resource_tree -from synapse.util.logcontext import LoggingContext, PreserveLoggingContext +from synapse.util.logcontext import LoggingContext from synapse.util.manhole import manhole -from synapse.util.rlimit import change_resource_limit from synapse.util.versionstring import get_version_string -from synapse.api.urls import ( - CONTENT_REPO_PREFIX, LEGACY_MEDIA_PREFIX, MEDIA_PREFIX -) -from synapse.crypto import context_factory - -from synapse import events - - -from twisted.internet import reactor -from twisted.web.resource import Resource - -from daemonize import Daemonize - -import sys -import logging -import gc logger = logging.getLogger("synapse.app.media_repository") @@ -69,19 +60,6 @@ class MediaRepositorySlavedStore( class MediaRepositoryServer(HomeServer): - def get_db_conn(self, run_new_connection=True): - # Any param beginning with cp_ is a parameter for adbapi, and should - # not be passed to the database engine. - db_params = { - k: v for k, v in self.db_config.get("args", {}).items() - if not k.startswith("cp_") - } - db_conn = self.database_engine.module.connect(**db_params) - - if run_new_connection: - self.database_engine.on_new_connection(db_conn) - return db_conn - def setup(self): logger.info("Setting up.") self.datastore = MediaRepositorySlavedStore(self.get_db_conn(), self) @@ -95,9 +73,9 @@ class MediaRepositoryServer(HomeServer): for res in listener_config["resources"]: for name in res["names"]: if name == "metrics": - resources[METRICS_PREFIX] = MetricsResource(self) + resources[METRICS_PREFIX] = MetricsResource(RegistryProxy) elif name == "media": - media_repo = MediaRepositoryResource(self) + media_repo = self.get_media_repository_resource() resources.update({ MEDIA_PREFIX: media_repo, LEGACY_MEDIA_PREFIX: media_repo, @@ -106,19 +84,19 @@ class MediaRepositoryServer(HomeServer): ), }) - root_resource = create_resource_tree(resources, Resource()) + root_resource = create_resource_tree(resources, NoResource()) - for address in bind_addresses: - reactor.listenTCP( - port, - SynapseSite( - "synapse.access.http.%s" % (site_tag,), - site_tag, - listener_config, - root_resource, - ), - interface=address + _base.listen_tcp( + bind_addresses, + port, + SynapseSite( + "synapse.access.http.%s" % (site_tag,), + site_tag, + listener_config, + root_resource, + self.version_string, ) + ) logger.info("Synapse media repository now listening on port %d", port) @@ -127,18 +105,22 @@ class MediaRepositoryServer(HomeServer): if listener["type"] == "http": self._listen_http(listener) elif listener["type"] == "manhole": - bind_addresses = listener["bind_addresses"] - - for address in bind_addresses: - reactor.listenTCP( - listener["port"], - manhole( - username="matrix", - password="rabbithole", - globals={"hs": self}, - ), - interface=address + _base.listen_tcp( + listener["bind_addresses"], + listener["port"], + manhole( + username="matrix", + password="rabbithole", + globals={"hs": self}, ) + ) + elif listener["type"] == "metrics": + if not self.get_config().enable_metrics: + logger.warn(("Metrics listener configured, but " + "enable_metrics is not True!")) + else: + _base.listen_metrics(listener["bind_addresses"], + listener["port"]) else: logger.warn("Unrecognized listener type: %s", listener["type"]) @@ -159,6 +141,13 @@ def start(config_options): assert config.worker_app == "synapse.app.media_repository" + if config.enable_media_repo: + _base.quit_with_error( + "enable_media_repo must be disabled in the main synapse process\n" + "before the media repo can be run in a separate worker.\n" + "Please add ``enable_media_repo: false`` to the main config\n" + ) + setup_logging(config, use_worker_options=True) events.USE_FROZEN_DICTS = config.use_frozen_dicts @@ -177,39 +166,15 @@ def start(config_options): ) ss.setup() - ss.get_handlers() ss.start_listening(config.worker_listeners) - def run(): - # make sure that we run the reactor with the sentinel log context, - # otherwise other PreserveLoggingContext instances will get confused - # and complain when they see the logcontext arbitrarily swapping - # between the sentinel and `run` logcontexts. - with PreserveLoggingContext(): - logger.info("Running") - change_resource_limit(config.soft_file_limit) - if config.gc_thresholds: - gc.set_threshold(*config.gc_thresholds) - reactor.run() - def start(): ss.get_state_handler().start_caching() ss.get_datastore().start_profiling() reactor.callWhenRunning(start) - if config.worker_daemonize: - daemon = Daemonize( - app="synapse-media-repository", - pid=config.worker_pid_file, - action=run, - auto_close_fds=False, - verbose=True, - logger=logger, - ) - daemon.start() - else: - run() + _base.start_worker_reactor("synapse-media-repository", config) if __name__ == '__main__': diff --git a/synapse/app/pusher.py b/synapse/app/pusher.py index f9114acfc..9295a51d5 100644 --- a/synapse/app/pusher.py +++ b/synapse/app/pusher.py @@ -13,41 +13,34 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. +import logging +import sys + +from twisted.internet import defer, reactor +from twisted.web.resource import NoResource import synapse - -from synapse.server import HomeServer +from synapse import events +from synapse.app import _base from synapse.config._base import ConfigError -from synapse.config.logger import setup_logging from synapse.config.homeserver import HomeServerConfig +from synapse.config.logger import setup_logging from synapse.http.site import SynapseSite -from synapse.metrics.resource import MetricsResource, METRICS_PREFIX -from synapse.storage.roommember import RoomMemberStore +from synapse.metrics import RegistryProxy +from synapse.metrics.resource import METRICS_PREFIX, MetricsResource +from synapse.replication.slave.storage.account_data import SlavedAccountDataStore from synapse.replication.slave.storage.events import SlavedEventStore from synapse.replication.slave.storage.pushers import SlavedPusherStore from synapse.replication.slave.storage.receipts import SlavedReceiptsStore -from synapse.replication.slave.storage.account_data import SlavedAccountDataStore from synapse.replication.tcp.client import ReplicationClientHandler -from synapse.storage.engines import create_engine +from synapse.server import HomeServer from synapse.storage import DataStore +from synapse.storage.engines import create_engine from synapse.util.httpresourcetree import create_resource_tree -from synapse.util.logcontext import LoggingContext, preserve_fn, \ - PreserveLoggingContext +from synapse.util.logcontext import LoggingContext, run_in_background from synapse.util.manhole import manhole -from synapse.util.rlimit import change_resource_limit from synapse.util.versionstring import get_version_string -from synapse import events - -from twisted.internet import reactor, defer -from twisted.web.resource import Resource - -from daemonize import Daemonize - -import sys -import logging -import gc - logger = logging.getLogger("synapse.app.pusher") @@ -83,25 +76,8 @@ class PusherSlaveStore( DataStore.get_profile_displayname.__func__ ) - who_forgot_in_room = ( - RoomMemberStore.__dict__["who_forgot_in_room"] - ) - class PusherServer(HomeServer): - def get_db_conn(self, run_new_connection=True): - # Any param beginning with cp_ is a parameter for adbapi, and should - # not be passed to the database engine. - db_params = { - k: v for k, v in self.db_config.get("args", {}).items() - if not k.startswith("cp_") - } - db_conn = self.database_engine.module.connect(**db_params) - - if run_new_connection: - self.database_engine.on_new_connection(db_conn) - return db_conn - def setup(self): logger.info("Setting up.") self.datastore = PusherSlaveStore(self.get_db_conn(), self) @@ -118,21 +94,21 @@ class PusherServer(HomeServer): for res in listener_config["resources"]: for name in res["names"]: if name == "metrics": - resources[METRICS_PREFIX] = MetricsResource(self) + resources[METRICS_PREFIX] = MetricsResource(RegistryProxy) - root_resource = create_resource_tree(resources, Resource()) + root_resource = create_resource_tree(resources, NoResource()) - for address in bind_addresses: - reactor.listenTCP( - port, - SynapseSite( - "synapse.access.http.%s" % (site_tag,), - site_tag, - listener_config, - root_resource, - ), - interface=address + _base.listen_tcp( + bind_addresses, + port, + SynapseSite( + "synapse.access.http.%s" % (site_tag,), + site_tag, + listener_config, + root_resource, + self.version_string, ) + ) logger.info("Synapse pusher now listening on port %d", port) @@ -141,18 +117,22 @@ class PusherServer(HomeServer): if listener["type"] == "http": self._listen_http(listener) elif listener["type"] == "manhole": - bind_addresses = listener["bind_addresses"] - - for address in bind_addresses: - reactor.listenTCP( - listener["port"], - manhole( - username="matrix", - password="rabbithole", - globals={"hs": self}, - ), - interface=address + _base.listen_tcp( + listener["bind_addresses"], + listener["port"], + manhole( + username="matrix", + password="rabbithole", + globals={"hs": self}, ) + ) + elif listener["type"] == "metrics": + if not self.get_config().enable_metrics: + logger.warn(("Metrics listener configured, but " + "enable_metrics is not True!")) + else: + _base.listen_metrics(listener["bind_addresses"], + listener["port"]) else: logger.warn("Unrecognized listener type: %s", listener["type"]) @@ -170,24 +150,27 @@ class PusherReplicationHandler(ReplicationClientHandler): def on_rdata(self, stream_name, token, rows): super(PusherReplicationHandler, self).on_rdata(stream_name, token, rows) - preserve_fn(self.poke_pushers)(stream_name, token, rows) + run_in_background(self.poke_pushers, stream_name, token, rows) @defer.inlineCallbacks def poke_pushers(self, stream_name, token, rows): - if stream_name == "pushers": - for row in rows: - if row.deleted: - yield self.stop_pusher(row.user_id, row.app_id, row.pushkey) - else: - yield self.start_pusher(row.user_id, row.app_id, row.pushkey) - elif stream_name == "events": - yield self.pusher_pool.on_new_notifications( - token, token, - ) - elif stream_name == "receipts": - yield self.pusher_pool.on_new_receipts( - token, token, set(row.room_id for row in rows) - ) + try: + if stream_name == "pushers": + for row in rows: + if row.deleted: + yield self.stop_pusher(row.user_id, row.app_id, row.pushkey) + else: + yield self.start_pusher(row.user_id, row.app_id, row.pushkey) + elif stream_name == "events": + yield self.pusher_pool.on_new_notifications( + token, token, + ) + elif stream_name == "receipts": + yield self.pusher_pool.on_new_receipts( + token, token, set(row.room_id for row in rows) + ) + except Exception: + logger.exception("Error poking pushers") def stop_pusher(self, user_id, app_id, pushkey): key = "%s:%s" % (app_id, pushkey) @@ -244,18 +227,6 @@ def start(config_options): ps.setup() ps.start_listening(config.worker_listeners) - def run(): - # make sure that we run the reactor with the sentinel log context, - # otherwise other PreserveLoggingContext instances will get confused - # and complain when they see the logcontext arbitrarily swapping - # between the sentinel and `run` logcontexts. - with PreserveLoggingContext(): - logger.info("Running") - change_resource_limit(config.soft_file_limit) - if config.gc_thresholds: - gc.set_threshold(*config.gc_thresholds) - reactor.run() - def start(): ps.get_pusherpool().start() ps.get_datastore().start_profiling() @@ -263,18 +234,7 @@ def start(config_options): reactor.callWhenRunning(start) - if config.worker_daemonize: - daemon = Daemonize( - app="synapse-pusher", - pid=config.worker_pid_file, - action=run, - auto_close_fds=False, - verbose=True, - logger=logger, - ) - daemon.start() - else: - run() + _base.start_worker_reactor("synapse-pusher", config) if __name__ == '__main__': diff --git a/synapse/app/synchrotron.py b/synapse/app/synchrotron.py index 4bdd99a96..26b9ec85f 100644 --- a/synapse/app/synchrotron.py +++ b/synapse/app/synchrotron.py @@ -13,78 +13,74 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. +import contextlib +import logging +import sys + +from six import iteritems + +from twisted.internet import defer, reactor +from twisted.web.resource import NoResource import synapse - from synapse.api.constants import EventTypes +from synapse.app import _base from synapse.config._base import ConfigError from synapse.config.homeserver import HomeServerConfig from synapse.config.logger import setup_logging from synapse.handlers.presence import PresenceHandler, get_interested_parties -from synapse.http.site import SynapseSite from synapse.http.server import JsonResource -from synapse.metrics.resource import MetricsResource, METRICS_PREFIX -from synapse.rest.client.v2_alpha import sync -from synapse.rest.client.v1 import events -from synapse.rest.client.v1.room import RoomInitialSyncRestServlet -from synapse.rest.client.v1.initial_sync import InitialSyncRestServlet +from synapse.http.site import SynapseSite +from synapse.metrics import RegistryProxy +from synapse.metrics.resource import METRICS_PREFIX, MetricsResource from synapse.replication.slave.storage._base import BaseSlavedStore -from synapse.replication.slave.storage.client_ips import SlavedClientIpStore -from synapse.replication.slave.storage.events import SlavedEventStore -from synapse.replication.slave.storage.receipts import SlavedReceiptsStore from synapse.replication.slave.storage.account_data import SlavedAccountDataStore from synapse.replication.slave.storage.appservice import SlavedApplicationServiceStore -from synapse.replication.slave.storage.registration import SlavedRegistrationStore -from synapse.replication.slave.storage.filtering import SlavedFilteringStore -from synapse.replication.slave.storage.push_rule import SlavedPushRuleStore -from synapse.replication.slave.storage.presence import SlavedPresenceStore +from synapse.replication.slave.storage.client_ips import SlavedClientIpStore from synapse.replication.slave.storage.deviceinbox import SlavedDeviceInboxStore from synapse.replication.slave.storage.devices import SlavedDeviceStore +from synapse.replication.slave.storage.events import SlavedEventStore +from synapse.replication.slave.storage.filtering import SlavedFilteringStore +from synapse.replication.slave.storage.groups import SlavedGroupServerStore +from synapse.replication.slave.storage.presence import SlavedPresenceStore +from synapse.replication.slave.storage.push_rule import SlavedPushRuleStore +from synapse.replication.slave.storage.receipts import SlavedReceiptsStore +from synapse.replication.slave.storage.registration import SlavedRegistrationStore from synapse.replication.slave.storage.room import RoomStore from synapse.replication.tcp.client import ReplicationClientHandler +from synapse.rest.client.v1 import events +from synapse.rest.client.v1.initial_sync import InitialSyncRestServlet +from synapse.rest.client.v1.room import RoomInitialSyncRestServlet +from synapse.rest.client.v2_alpha import sync from synapse.server import HomeServer from synapse.storage.engines import create_engine from synapse.storage.presence import UserPresenceState from synapse.storage.roommember import RoomMemberStore from synapse.util.httpresourcetree import create_resource_tree -from synapse.util.logcontext import LoggingContext, PreserveLoggingContext, preserve_fn +from synapse.util.logcontext import LoggingContext, run_in_background from synapse.util.manhole import manhole -from synapse.util.rlimit import change_resource_limit from synapse.util.stringutils import random_string from synapse.util.versionstring import get_version_string -from twisted.internet import reactor, defer -from twisted.web.resource import Resource - -from daemonize import Daemonize - -import sys -import logging -import contextlib -import gc - logger = logging.getLogger("synapse.app.synchrotron") class SynchrotronSlavedStore( - SlavedPushRuleStore, - SlavedEventStore, SlavedReceiptsStore, SlavedAccountDataStore, SlavedApplicationServiceStore, SlavedRegistrationStore, SlavedFilteringStore, SlavedPresenceStore, + SlavedGroupServerStore, SlavedDeviceInboxStore, SlavedDeviceStore, + SlavedPushRuleStore, + SlavedEventStore, SlavedClientIpStore, RoomStore, BaseSlavedStore, ): - who_forgot_in_room = ( - RoomMemberStore.__dict__["who_forgot_in_room"] - ) - did_forget = ( RoomMemberStore.__dict__["did_forget"] ) @@ -219,7 +215,7 @@ class SynchrotronPresence(object): def get_currently_syncing_users(self): return [ - user_id for user_id, count in self.user_to_num_current_syncs.iteritems() + user_id for user_id, count in iteritems(self.user_to_num_current_syncs) if count > 0 ] @@ -250,19 +246,6 @@ class SynchrotronApplicationService(object): class SynchrotronServer(HomeServer): - def get_db_conn(self, run_new_connection=True): - # Any param beginning with cp_ is a parameter for adbapi, and should - # not be passed to the database engine. - db_params = { - k: v for k, v in self.db_config.get("args", {}).items() - if not k.startswith("cp_") - } - db_conn = self.database_engine.module.connect(**db_params) - - if run_new_connection: - self.database_engine.on_new_connection(db_conn) - return db_conn - def setup(self): logger.info("Setting up.") self.datastore = SynchrotronSlavedStore(self.get_db_conn(), self) @@ -276,7 +259,7 @@ class SynchrotronServer(HomeServer): for res in listener_config["resources"]: for name in res["names"]: if name == "metrics": - resources[METRICS_PREFIX] = MetricsResource(self) + resources[METRICS_PREFIX] = MetricsResource(RegistryProxy) elif name == "client": resource = JsonResource(self, canonical_json=False) sync.register_servlets(self, resource) @@ -290,19 +273,19 @@ class SynchrotronServer(HomeServer): "/_matrix/client/api/v1": resource, }) - root_resource = create_resource_tree(resources, Resource()) + root_resource = create_resource_tree(resources, NoResource()) - for address in bind_addresses: - reactor.listenTCP( - port, - SynapseSite( - "synapse.access.http.%s" % (site_tag,), - site_tag, - listener_config, - root_resource, - ), - interface=address + _base.listen_tcp( + bind_addresses, + port, + SynapseSite( + "synapse.access.http.%s" % (site_tag,), + site_tag, + listener_config, + root_resource, + self.version_string, ) + ) logger.info("Synapse synchrotron now listening on port %d", port) @@ -311,18 +294,22 @@ class SynchrotronServer(HomeServer): if listener["type"] == "http": self._listen_http(listener) elif listener["type"] == "manhole": - bind_addresses = listener["bind_addresses"] - - for address in bind_addresses: - reactor.listenTCP( - listener["port"], - manhole( - username="matrix", - password="rabbithole", - globals={"hs": self}, - ), - interface=address + _base.listen_tcp( + listener["bind_addresses"], + listener["port"], + manhole( + username="matrix", + password="rabbithole", + globals={"hs": self}, ) + ) + elif listener["type"] == "metrics": + if not self.get_config().enable_metrics: + logger.warn(("Metrics listener configured, but " + "enable_metrics is not True!")) + else: + _base.listen_metrics(listener["bind_addresses"], + listener["port"]) else: logger.warn("Unrecognized listener type: %s", listener["type"]) @@ -344,15 +331,13 @@ class SyncReplicationHandler(ReplicationClientHandler): self.store = hs.get_datastore() self.typing_handler = hs.get_typing_handler() + # NB this is a SynchrotronPresence, not a normal PresenceHandler self.presence_handler = hs.get_presence_handler() self.notifier = hs.get_notifier() - self.presence_handler.sync_callback = self.send_user_sync - def on_rdata(self, stream_name, token, rows): super(SyncReplicationHandler, self).on_rdata(stream_name, token, rows) - - preserve_fn(self.process_and_notify)(stream_name, token, rows) + run_in_background(self.process_and_notify, stream_name, token, rows) def get_streams_to_replicate(self): args = super(SyncReplicationHandler, self).get_streams_to_replicate() @@ -364,51 +349,58 @@ class SyncReplicationHandler(ReplicationClientHandler): @defer.inlineCallbacks def process_and_notify(self, stream_name, token, rows): - if stream_name == "events": - # We shouldn't get multiple rows per token for events stream, so - # we don't need to optimise this for multiple rows. - for row in rows: - event = yield self.store.get_event(row.event_id) - extra_users = () - if event.type == EventTypes.Member: - extra_users = (event.state_key,) - max_token = self.store.get_room_max_stream_ordering() - self.notifier.on_new_room_event( - event, token, max_token, extra_users - ) - elif stream_name == "push_rules": - self.notifier.on_new_event( - "push_rules_key", token, users=[row.user_id for row in rows], - ) - elif stream_name in ("account_data", "tag_account_data",): - self.notifier.on_new_event( - "account_data_key", token, users=[row.user_id for row in rows], - ) - elif stream_name == "receipts": - self.notifier.on_new_event( - "receipt_key", token, rooms=[row.room_id for row in rows], - ) - elif stream_name == "typing": - self.typing_handler.process_replication_rows(token, rows) - self.notifier.on_new_event( - "typing_key", token, rooms=[row.room_id for row in rows], - ) - elif stream_name == "to_device": - entities = [row.entity for row in rows if row.entity.startswith("@")] - if entities: + try: + if stream_name == "events": + # We shouldn't get multiple rows per token for events stream, so + # we don't need to optimise this for multiple rows. + for row in rows: + event = yield self.store.get_event(row.event_id) + extra_users = () + if event.type == EventTypes.Member: + extra_users = (event.state_key,) + max_token = self.store.get_room_max_stream_ordering() + self.notifier.on_new_room_event( + event, token, max_token, extra_users + ) + elif stream_name == "push_rules": self.notifier.on_new_event( - "to_device_key", token, users=entities, + "push_rules_key", token, users=[row.user_id for row in rows], ) - elif stream_name == "device_lists": - all_room_ids = set() - for row in rows: - room_ids = yield self.store.get_rooms_for_user(row.user_id) - all_room_ids.update(room_ids) - self.notifier.on_new_event( - "device_list_key", token, rooms=all_room_ids, - ) - elif stream_name == "presence": - yield self.presence_handler.process_replication_rows(token, rows) + elif stream_name in ("account_data", "tag_account_data",): + self.notifier.on_new_event( + "account_data_key", token, users=[row.user_id for row in rows], + ) + elif stream_name == "receipts": + self.notifier.on_new_event( + "receipt_key", token, rooms=[row.room_id for row in rows], + ) + elif stream_name == "typing": + self.typing_handler.process_replication_rows(token, rows) + self.notifier.on_new_event( + "typing_key", token, rooms=[row.room_id for row in rows], + ) + elif stream_name == "to_device": + entities = [row.entity for row in rows if row.entity.startswith("@")] + if entities: + self.notifier.on_new_event( + "to_device_key", token, users=entities, + ) + elif stream_name == "device_lists": + all_room_ids = set() + for row in rows: + room_ids = yield self.store.get_rooms_for_user(row.user_id) + all_room_ids.update(room_ids) + self.notifier.on_new_event( + "device_list_key", token, rooms=all_room_ids, + ) + elif stream_name == "presence": + yield self.presence_handler.process_replication_rows(token, rows) + elif stream_name == "receipts": + self.notifier.on_new_event( + "groups_key", token, users=[row.user_id for row in rows], + ) + except Exception: + logger.exception("Error processing replication") def start(config_options): @@ -440,36 +432,13 @@ def start(config_options): ss.setup() ss.start_listening(config.worker_listeners) - def run(): - # make sure that we run the reactor with the sentinel log context, - # otherwise other PreserveLoggingContext instances will get confused - # and complain when they see the logcontext arbitrarily swapping - # between the sentinel and `run` logcontexts. - with PreserveLoggingContext(): - logger.info("Running") - change_resource_limit(config.soft_file_limit) - if config.gc_thresholds: - gc.set_threshold(*config.gc_thresholds) - reactor.run() - def start(): ss.get_datastore().start_profiling() ss.get_state_handler().start_caching() reactor.callWhenRunning(start) - if config.worker_daemonize: - daemon = Daemonize( - app="synapse-synchrotron", - pid=config.worker_pid_file, - action=run, - auto_close_fds=False, - verbose=True, - logger=logger, - ) - daemon.start() - else: - run() + _base.start_worker_reactor("synapse-synchrotron", config) if __name__ == '__main__': diff --git a/synapse/app/synctl.py b/synapse/app/synctl.py index 3bd7ef7bb..d658f967b 100755 --- a/synapse/app/synctl.py +++ b/synapse/app/synctl.py @@ -16,16 +16,19 @@ import argparse import collections +import errno import glob import os import os.path import signal import subprocess import sys -import yaml -import errno import time +from six import iteritems + +import yaml + SYNAPSE = [sys.executable, "-B", "-m", "synapse.app.homeserver"] GREEN = "\x1b[1;32m" @@ -38,7 +41,7 @@ def pid_running(pid): try: os.kill(pid, 0) return True - except OSError, err: + except OSError as err: if err.errno == errno.EPERM: return True return False @@ -98,7 +101,7 @@ def stop(pidfile, app): try: os.kill(pid, signal.SIGTERM) write("stopped %s" % (app,), colour=GREEN) - except OSError, err: + except OSError as err: if err.errno == errno.ESRCH: write("%s not running" % (app,), colour=YELLOW) elif err.errno == errno.EPERM: @@ -171,6 +174,10 @@ def main(): if cache_factor: os.environ["SYNAPSE_CACHE_FACTOR"] = str(cache_factor) + cache_factors = config.get("synctl_cache_factors", {}) + for cache_name, factor in iteritems(cache_factors): + os.environ["SYNAPSE_CACHE_FACTOR_" + cache_name.upper()] = str(factor) + worker_configfiles = [] if options.worker: start_stop_synapse = False @@ -184,6 +191,9 @@ def main(): worker_configfiles.append(worker_configfile) if options.all_processes: + # To start the main synapse with -a you need to add a worker file + # with worker_app == "synapse.app.homeserver" + start_stop_synapse = False worker_configdir = options.all_processes if not os.path.isdir(worker_configdir): write( @@ -200,11 +210,29 @@ def main(): with open(worker_configfile) as stream: worker_config = yaml.load(stream) worker_app = worker_config["worker_app"] - worker_pidfile = worker_config["worker_pid_file"] - worker_daemonize = worker_config["worker_daemonize"] - assert worker_daemonize, "In config %r: expected '%s' to be True" % ( - worker_configfile, "worker_daemonize") - worker_cache_factor = worker_config.get("synctl_cache_factor") + if worker_app == "synapse.app.homeserver": + # We need to special case all of this to pick up options that may + # be set in the main config file or in this worker config file. + worker_pidfile = ( + worker_config.get("pid_file") + or pidfile + ) + worker_cache_factor = worker_config.get("synctl_cache_factor") or cache_factor + daemonize = worker_config.get("daemonize") or config.get("daemonize") + assert daemonize, "Main process must have daemonize set to true" + + # The master process doesn't support using worker_* config. + for key in worker_config: + if key == "worker_app": # But we allow worker_app + continue + assert not key.startswith("worker_"), \ + "Main process cannot use worker_* config" + else: + worker_pidfile = worker_config["worker_pid_file"] + worker_daemonize = worker_config["worker_daemonize"] + assert worker_daemonize, "In config %r: expected '%s' to be True" % ( + worker_configfile, "worker_daemonize") + worker_cache_factor = worker_config.get("synctl_cache_factor") workers.append(Worker( worker_app, worker_configfile, worker_pidfile, worker_cache_factor, )) @@ -231,6 +259,7 @@ def main(): for running_pid in running_pids: while pid_running(running_pid): time.sleep(0.2) + write("All processes exited; now restarting...") if action == "start" or action == "restart": if start_stop_synapse: diff --git a/synapse/app/user_dir.py b/synapse/app/user_dir.py index 8c6300db9..637a89530 100644 --- a/synapse/app/user_dir.py +++ b/synapse/app/user_dir.py @@ -14,16 +14,23 @@ # See the License for the specific language governing permissions and # limitations under the License. -import synapse +import logging +import sys -from synapse.server import HomeServer +from twisted.internet import defer, reactor +from twisted.web.resource import NoResource + +import synapse +from synapse import events +from synapse.app import _base from synapse.config._base import ConfigError -from synapse.config.logger import setup_logging from synapse.config.homeserver import HomeServerConfig +from synapse.config.logger import setup_logging from synapse.crypto import context_factory -from synapse.http.site import SynapseSite from synapse.http.server import JsonResource -from synapse.metrics.resource import MetricsResource, METRICS_PREFIX +from synapse.http.site import SynapseSite +from synapse.metrics import RegistryProxy +from synapse.metrics.resource import METRICS_PREFIX, MetricsResource from synapse.replication.slave.storage._base import BaseSlavedStore from synapse.replication.slave.storage.appservice import SlavedApplicationServiceStore from synapse.replication.slave.storage.client_ips import SlavedClientIpStore @@ -31,25 +38,14 @@ from synapse.replication.slave.storage.events import SlavedEventStore from synapse.replication.slave.storage.registration import SlavedRegistrationStore from synapse.replication.tcp.client import ReplicationClientHandler from synapse.rest.client.v2_alpha import user_directory +from synapse.server import HomeServer from synapse.storage.engines import create_engine from synapse.storage.user_directory import UserDirectoryStore -from synapse.util.httpresourcetree import create_resource_tree -from synapse.util.logcontext import LoggingContext, PreserveLoggingContext, preserve_fn -from synapse.util.manhole import manhole -from synapse.util.rlimit import change_resource_limit -from synapse.util.versionstring import get_version_string from synapse.util.caches.stream_change_cache import StreamChangeCache - -from synapse import events - -from twisted.internet import reactor -from twisted.web.resource import Resource - -from daemonize import Daemonize - -import sys -import logging -import gc +from synapse.util.httpresourcetree import create_resource_tree +from synapse.util.logcontext import LoggingContext, run_in_background +from synapse.util.manhole import manhole +from synapse.util.versionstring import get_version_string logger = logging.getLogger("synapse.app.user_dir") @@ -98,19 +94,6 @@ class UserDirectorySlaveStore( class UserDirectoryServer(HomeServer): - def get_db_conn(self, run_new_connection=True): - # Any param beginning with cp_ is a parameter for adbapi, and should - # not be passed to the database engine. - db_params = { - k: v for k, v in self.db_config.get("args", {}).items() - if not k.startswith("cp_") - } - db_conn = self.database_engine.module.connect(**db_params) - - if run_new_connection: - self.database_engine.on_new_connection(db_conn) - return db_conn - def setup(self): logger.info("Setting up.") self.datastore = UserDirectorySlaveStore(self.get_db_conn(), self) @@ -124,7 +107,7 @@ class UserDirectoryServer(HomeServer): for res in listener_config["resources"]: for name in res["names"]: if name == "metrics": - resources[METRICS_PREFIX] = MetricsResource(self) + resources[METRICS_PREFIX] = MetricsResource(RegistryProxy) elif name == "client": resource = JsonResource(self, canonical_json=False) user_directory.register_servlets(self, resource) @@ -135,19 +118,19 @@ class UserDirectoryServer(HomeServer): "/_matrix/client/api/v1": resource, }) - root_resource = create_resource_tree(resources, Resource()) + root_resource = create_resource_tree(resources, NoResource()) - for address in bind_addresses: - reactor.listenTCP( - port, - SynapseSite( - "synapse.access.http.%s" % (site_tag,), - site_tag, - listener_config, - root_resource, - ), - interface=address + _base.listen_tcp( + bind_addresses, + port, + SynapseSite( + "synapse.access.http.%s" % (site_tag,), + site_tag, + listener_config, + root_resource, + self.version_string, ) + ) logger.info("Synapse user_dir now listening on port %d", port) @@ -156,18 +139,22 @@ class UserDirectoryServer(HomeServer): if listener["type"] == "http": self._listen_http(listener) elif listener["type"] == "manhole": - bind_addresses = listener["bind_addresses"] - - for address in bind_addresses: - reactor.listenTCP( - listener["port"], - manhole( - username="matrix", - password="rabbithole", - globals={"hs": self}, - ), - interface=address + _base.listen_tcp( + listener["bind_addresses"], + listener["port"], + manhole( + username="matrix", + password="rabbithole", + globals={"hs": self}, ) + ) + elif listener["type"] == "metrics": + if not self.get_config().enable_metrics: + logger.warn(("Metrics listener configured, but " + "enable_metrics is not True!")) + else: + _base.listen_metrics(listener["bind_addresses"], + listener["port"]) else: logger.warn("Unrecognized listener type: %s", listener["type"]) @@ -187,7 +174,14 @@ class UserDirectoryReplicationHandler(ReplicationClientHandler): stream_name, token, rows ) if stream_name == "current_state_deltas": - preserve_fn(self.user_directory.notify_new_event)() + run_in_background(self._notify_directory) + + @defer.inlineCallbacks + def _notify_directory(self): + try: + yield self.user_directory.notify_new_event() + except Exception: + logger.exception("Error notifiying user directory of state update") def start(config_options): @@ -233,36 +227,13 @@ def start(config_options): ps.setup() ps.start_listening(config.worker_listeners) - def run(): - # make sure that we run the reactor with the sentinel log context, - # otherwise other PreserveLoggingContext instances will get confused - # and complain when they see the logcontext arbitrarily swapping - # between the sentinel and `run` logcontexts. - with PreserveLoggingContext(): - logger.info("Running") - change_resource_limit(config.soft_file_limit) - if config.gc_thresholds: - gc.set_threshold(*config.gc_thresholds) - reactor.run() - def start(): ps.get_datastore().start_profiling() ps.get_state_handler().start_caching() reactor.callWhenRunning(start) - if config.worker_daemonize: - daemon = Daemonize( - app="synapse-user-dir", - pid=config.worker_pid_file, - action=run, - auto_close_fds=False, - verbose=True, - logger=logger, - ) - daemon.start() - else: - run() + _base.start_worker_reactor("synapse-user-dir", config) if __name__ == '__main__': diff --git a/synapse/appservice/__init__.py b/synapse/appservice/__init__.py index b98900731..57ed8a3ca 100644 --- a/synapse/appservice/__init__.py +++ b/synapse/appservice/__init__.py @@ -12,13 +12,16 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. -from synapse.api.constants import EventTypes -from synapse.util.caches.descriptors import cachedInlineCallbacks +import logging +import re + +from six import string_types from twisted.internet import defer -import logging -import re +from synapse.api.constants import EventTypes +from synapse.types import GroupID, get_domain_from_id +from synapse.util.caches.descriptors import cachedInlineCallbacks logger = logging.getLogger(__name__) @@ -81,14 +84,17 @@ class ApplicationService(object): # values. NS_LIST = [NS_USERS, NS_ALIASES, NS_ROOMS] - def __init__(self, token, url=None, namespaces=None, hs_token=None, - sender=None, id=None, protocols=None, rate_limited=True): + def __init__(self, token, hostname, url=None, namespaces=None, hs_token=None, + sender=None, id=None, protocols=None, rate_limited=True, + ip_range_whitelist=None): self.token = token self.url = url self.hs_token = hs_token self.sender = sender + self.server_name = hostname self.namespaces = self._check_namespaces(namespaces) self.id = id + self.ip_range_whitelist = ip_range_whitelist if "|" in self.id: raise Exception("application service ID cannot contain '|' character") @@ -125,8 +131,26 @@ class ApplicationService(object): raise ValueError( "Expected bool for 'exclusive' in ns '%s'" % ns ) + group_id = regex_obj.get("group_id") + if group_id: + if not isinstance(group_id, str): + raise ValueError( + "Expected string for 'group_id' in ns '%s'" % ns + ) + try: + GroupID.from_string(group_id) + except Exception: + raise ValueError( + "Expected valid group ID for 'group_id' in ns '%s'" % ns + ) + + if get_domain_from_id(group_id) != self.server_name: + raise ValueError( + "Expected 'group_id' to be this host in ns '%s'" % ns + ) + regex = regex_obj.get("regex") - if isinstance(regex, basestring): + if isinstance(regex, string_types): regex_obj["regex"] = re.compile(regex) # Pre-compile regex else: raise ValueError( @@ -251,8 +275,27 @@ class ApplicationService(object): if regex_obj["exclusive"] ] + def get_groups_for_user(self, user_id): + """Get the groups that this user is associated with by this AS + + Args: + user_id (str): The ID of the user. + + Returns: + iterable[str]: an iterable that yields group_id strings. + """ + return ( + regex_obj["group_id"] + for regex_obj in self.namespaces[ApplicationService.NS_USERS] + if "group_id" in regex_obj and regex_obj["regex"].match(user_id) + ) + def is_rate_limited(self): return self.rate_limited def __str__(self): - return "ApplicationService: %s" % (self.__dict__,) + # copy dictionary and redact token fields so they don't get logged + dict_copy = self.__dict__.copy() + dict_copy["token"] = "" + dict_copy["hs_token"] = "" + return "ApplicationService: %s" % (dict_copy,) diff --git a/synapse/appservice/api.py b/synapse/appservice/api.py index 6893610e7..6980e5890 100644 --- a/synapse/appservice/api.py +++ b/synapse/appservice/api.py @@ -12,20 +12,39 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. +import logging +import urllib + +from prometheus_client import Counter + from twisted.internet import defer from synapse.api.constants import ThirdPartyEntityKind from synapse.api.errors import CodeMessageException -from synapse.http.client import SimpleHttpClient from synapse.events.utils import serialize_event -from synapse.util.caches.response_cache import ResponseCache +from synapse.http.client import SimpleHttpClient from synapse.types import ThirdPartyInstanceID - -import logging -import urllib +from synapse.util.caches.response_cache import ResponseCache logger = logging.getLogger(__name__) +sent_transactions_counter = Counter( + "synapse_appservice_api_sent_transactions", + "Number of /transactions/ requests sent", + ["service"] +) + +failed_transactions_counter = Counter( + "synapse_appservice_api_failed_transactions", + "Number of /transactions/ requests that failed to send", + ["service"] +) + +sent_events_counter = Counter( + "synapse_appservice_api_sent_events", + "Number of events sent to the AS", + ["service"] +) HOUR_IN_MS = 60 * 60 * 1000 @@ -72,7 +91,8 @@ class ApplicationServiceApi(SimpleHttpClient): super(ApplicationServiceApi, self).__init__(hs) self.clock = hs.get_clock() - self.protocol_meta_cache = ResponseCache(hs, timeout_ms=HOUR_IN_MS) + self.protocol_meta_cache = ResponseCache(hs, "as_protocol_meta", + timeout_ms=HOUR_IN_MS) @defer.inlineCallbacks def query_user(self, service, user_id): @@ -192,9 +212,7 @@ class ApplicationServiceApi(SimpleHttpClient): defer.returnValue(None) key = (service.id, protocol) - return self.protocol_meta_cache.get(key) or ( - self.protocol_meta_cache.set(key, _get()) - ) + return self.protocol_meta_cache.wrap(key, _get) @defer.inlineCallbacks def push_bulk(self, service, events, txn_id=None): @@ -220,12 +238,15 @@ class ApplicationServiceApi(SimpleHttpClient): args={ "access_token": service.hs_token }) + sent_transactions_counter.labels(service.id).inc() + sent_events_counter.labels(service.id).inc(len(events)) defer.returnValue(True) return except CodeMessageException as e: logger.warning("push_bulk to %s received %s", uri, e.code) except Exception as ex: logger.warning("push_bulk to %s threw exception %s", uri, ex) + failed_transactions_counter.labels(service.id).inc() defer.returnValue(False) def _serialize(self, events): diff --git a/synapse/appservice/scheduler.py b/synapse/appservice/scheduler.py index 68a9de17b..243081479 100644 --- a/synapse/appservice/scheduler.py +++ b/synapse/appservice/scheduler.py @@ -48,14 +48,14 @@ UP & quit +---------- YES SUCCESS This is all tied together by the AppServiceScheduler which DIs the required components. """ +import logging + from twisted.internet import defer from synapse.appservice import ApplicationServiceState -from synapse.util.logcontext import preserve_fn +from synapse.util.logcontext import run_in_background from synapse.util.metrics import Measure -import logging - logger = logging.getLogger(__name__) @@ -106,7 +106,7 @@ class _ServiceQueuer(object): def enqueue(self, service, event): # if this service isn't being sent something self.queued_events.setdefault(service.id, []).append(event) - preserve_fn(self._send_request)(service) + run_in_background(self._send_request, service) @defer.inlineCallbacks def _send_request(self, service): @@ -123,7 +123,7 @@ class _ServiceQueuer(object): with Measure(self.clock, "servicequeuer.send"): try: yield self.txn_ctrl.send(service, events) - except: + except Exception: logger.exception("AS request failed") finally: self.requests_in_flight.discard(service.id) @@ -152,10 +152,10 @@ class _TransactionController(object): if sent: yield txn.complete(self.store) else: - preserve_fn(self._start_recoverer)(service) - except Exception as e: - logger.exception(e) - preserve_fn(self._start_recoverer)(service) + run_in_background(self._start_recoverer, service) + except Exception: + logger.exception("Error creating appservice transaction") + run_in_background(self._start_recoverer, service) @defer.inlineCallbacks def on_recovered(self, recoverer): @@ -176,17 +176,20 @@ class _TransactionController(object): @defer.inlineCallbacks def _start_recoverer(self, service): - yield self.store.set_appservice_state( - service, - ApplicationServiceState.DOWN - ) - logger.info( - "Application service falling behind. Starting recoverer. AS ID %s", - service.id - ) - recoverer = self.recoverer_fn(service, self.on_recovered) - self.add_recoverers([recoverer]) - recoverer.recover() + try: + yield self.store.set_appservice_state( + service, + ApplicationServiceState.DOWN + ) + logger.info( + "Application service falling behind. Starting recoverer. AS ID %s", + service.id + ) + recoverer = self.recoverer_fn(service, self.on_recovered) + self.add_recoverers([recoverer]) + recoverer.recover() + except Exception: + logger.exception("Error starting AS recoverer") @defer.inlineCallbacks def _is_service_up(self, service): diff --git a/synapse/config/__init__.py b/synapse/config/__init__.py index bfebb0f64..f2a5a41e9 100644 --- a/synapse/config/__init__.py +++ b/synapse/config/__init__.py @@ -12,3 +12,9 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. + +from ._base import ConfigError + +# export ConfigError if somebody does import * +# this is largely a fudge to stop PEP8 moaning about the import +__all__ = ["ConfigError"] diff --git a/synapse/config/_base.py b/synapse/config/_base.py index 1ab5593c6..3d2e90dd5 100644 --- a/synapse/config/_base.py +++ b/synapse/config/_base.py @@ -16,9 +16,12 @@ import argparse import errno import os -import yaml from textwrap import dedent +from six import integer_types + +import yaml + class ConfigError(Exception): pass @@ -49,7 +52,7 @@ Missing mandatory `server_name` config option. class Config(object): @staticmethod def parse_size(value): - if isinstance(value, int) or isinstance(value, long): + if isinstance(value, integer_types): return value sizes = {"K": 1024, "M": 1024 * 1024} size = 1 @@ -61,7 +64,7 @@ class Config(object): @staticmethod def parse_duration(value): - if isinstance(value, int) or isinstance(value, long): + if isinstance(value, integer_types): return value second = 1000 minute = 60 * second @@ -81,22 +84,38 @@ class Config(object): def abspath(file_path): return os.path.abspath(file_path) if file_path else file_path + @classmethod + def path_exists(cls, file_path): + """Check if a file exists + + Unlike os.path.exists, this throws an exception if there is an error + checking if the file exists (for example, if there is a perms error on + the parent dir). + + Returns: + bool: True if the file exists; False if not. + """ + try: + os.stat(file_path) + return True + except OSError as e: + if e.errno != errno.ENOENT: + raise e + return False + @classmethod def check_file(cls, file_path, config_name): if file_path is None: raise ConfigError( "Missing config for %s." - " You must specify a path for the config file. You can " - "do this with the -c or --config-path option. " - "Adding --generate-config along with --server-name " - " will generate a config file at the given path." % (config_name,) ) - if not os.path.exists(file_path): + try: + os.stat(file_path) + except OSError as e: raise ConfigError( - "File %s config for %s doesn't exist." - " Try running again with --generate-config" - % (file_path, config_name,) + "Error accessing file '%s' (config for %s): %s" + % (file_path, config_name, e.strerror) ) return cls.abspath(file_path) @@ -248,7 +267,7 @@ class Config(object): " -c CONFIG-FILE\"" ) (config_path,) = config_files - if not os.path.exists(config_path): + if not cls.path_exists(config_path): if config_args.keys_directory: config_dir_path = config_args.keys_directory else: @@ -261,33 +280,33 @@ class Config(object): "Must specify a server_name to a generate config for." " Pass -H server.name." ) - if not os.path.exists(config_dir_path): + if not cls.path_exists(config_dir_path): os.makedirs(config_dir_path) - with open(config_path, "wb") as config_file: - config_bytes, config = obj.generate_config( + with open(config_path, "w") as config_file: + config_str, config = obj.generate_config( config_dir_path=config_dir_path, server_name=server_name, report_stats=(config_args.report_stats == "yes"), is_generating_file=True ) obj.invoke_all("generate_files", config) - config_file.write(config_bytes) - print ( + config_file.write(config_str) + print(( "A config file has been generated in %r for server name" " %r with corresponding SSL keys and self-signed" " certificates. Please review this file and customise it" " to your needs." - ) % (config_path, server_name) - print ( + ) % (config_path, server_name)) + print( "If this server name is incorrect, you will need to" " regenerate the SSL certificates" ) return else: - print ( + print(( "Config file %r already exists. Generating any missing key" " files." - ) % (config_path,) + ) % (config_path,)) generate_keys = True parser = argparse.ArgumentParser( diff --git a/synapse/config/api.py b/synapse/config/api.py index 20ba33226..403d96ba7 100644 --- a/synapse/config/api.py +++ b/synapse/config/api.py @@ -12,10 +12,10 @@ # See the License for the specific language governing permissions and # limitations under the License. -from ._base import Config - from synapse.api.constants import EventTypes +from ._base import Config + class ApiConfig(Config): diff --git a/synapse/config/appservice.py b/synapse/config/appservice.py index 82c50b824..3b161d708 100644 --- a/synapse/config/appservice.py +++ b/synapse/config/appservice.py @@ -12,14 +12,18 @@ # See the License for the specific language governing permissions and # limitations under the License. -from ._base import Config, ConfigError +import logging + +from six import string_types +from six.moves.urllib import parse as urlparse + +import yaml +from netaddr import IPSet from synapse.appservice import ApplicationService from synapse.types import UserID -import urllib -import yaml -import logging +from ._base import Config, ConfigError logger = logging.getLogger(__name__) @@ -89,21 +93,21 @@ def _load_appservice(hostname, as_info, config_filename): "id", "as_token", "hs_token", "sender_localpart" ] for field in required_string_fields: - if not isinstance(as_info.get(field), basestring): + if not isinstance(as_info.get(field), string_types): raise KeyError("Required string field: '%s' (%s)" % ( field, config_filename, )) # 'url' must either be a string or explicitly null, not missing # to avoid accidentally turning off push for ASes. - if (not isinstance(as_info.get("url"), basestring) and + if (not isinstance(as_info.get("url"), string_types) and as_info.get("url", "") is not None): raise KeyError( "Required string field or explicit null: 'url' (%s)" % (config_filename,) ) localpart = as_info["sender_localpart"] - if urllib.quote(localpart) != localpart: + if urlparse.quote(localpart) != localpart: raise ValueError( "sender_localpart needs characters which are not URL encoded." ) @@ -128,7 +132,7 @@ def _load_appservice(hostname, as_info, config_filename): "Expected namespace entry in %s to be an object," " but got %s", ns, regex_obj ) - if not isinstance(regex_obj.get("regex"), basestring): + if not isinstance(regex_obj.get("regex"), string_types): raise ValueError( "Missing/bad type 'regex' key in %s", regex_obj ) @@ -152,13 +156,22 @@ def _load_appservice(hostname, as_info, config_filename): " will not receive events or queries.", config_filename, ) + + ip_range_whitelist = None + if as_info.get('ip_range_whitelist'): + ip_range_whitelist = IPSet( + as_info.get('ip_range_whitelist') + ) + return ApplicationService( token=as_info["as_token"], + hostname=hostname, url=as_info["url"], namespaces=as_info["namespaces"], hs_token=as_info["hs_token"], sender=user_id, id=as_info["id"], protocols=protocols, - rate_limited=rate_limited + rate_limited=rate_limited, + ip_range_whitelist=ip_range_whitelist, ) diff --git a/synapse/config/cas.py b/synapse/config/cas.py index 938f6f25f..8109e5f95 100644 --- a/synapse/config/cas.py +++ b/synapse/config/cas.py @@ -41,7 +41,7 @@ class CasConfig(Config): #cas_config: # enabled: true # server_url: "https://cas-server.com" - # service_url: "https://homesever.domain.com:8448" + # service_url: "https://homeserver.domain.com:8448" # #required_attributes: # # name: value """ diff --git a/synapse/config/consent_config.py b/synapse/config/consent_config.py new file mode 100644 index 000000000..e22c731aa --- /dev/null +++ b/synapse/config/consent_config.py @@ -0,0 +1,88 @@ +# -*- coding: utf-8 -*- +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from ._base import Config + +DEFAULT_CONFIG = """\ +# User Consent configuration +# +# for detailed instructions, see +# https://github.com/matrix-org/synapse/blob/master/docs/consent_tracking.md +# +# Parts of this section are required if enabling the 'consent' resource under +# 'listeners', in particular 'template_dir' and 'version'. +# +# 'template_dir' gives the location of the templates for the HTML forms. +# This directory should contain one subdirectory per language (eg, 'en', 'fr'), +# and each language directory should contain the policy document (named as +# '.html') and a success page (success.html). +# +# 'version' specifies the 'current' version of the policy document. It defines +# the version to be served by the consent resource if there is no 'v' +# parameter. +# +# 'server_notice_content', if enabled, will send a user a "Server Notice" +# asking them to consent to the privacy policy. The 'server_notices' section +# must also be configured for this to work. Notices will *not* be sent to +# guest users unless 'send_server_notice_to_guests' is set to true. +# +# 'block_events_error', if set, will block any attempts to send events +# until the user consents to the privacy policy. The value of the setting is +# used as the text of the error. +# +# user_consent: +# template_dir: res/templates/privacy +# version: 1.0 +# server_notice_content: +# msgtype: m.text +# body: >- +# To continue using this homeserver you must review and agree to the +# terms and conditions at %(consent_uri)s +# send_server_notice_to_guests: True +# block_events_error: >- +# To continue using this homeserver you must review and agree to the +# terms and conditions at %(consent_uri)s +# +""" + + +class ConsentConfig(Config): + def __init__(self): + super(ConsentConfig, self).__init__() + + self.user_consent_version = None + self.user_consent_template_dir = None + self.user_consent_server_notice_content = None + self.user_consent_server_notice_to_guests = False + self.block_events_without_consent_error = None + + def read_config(self, config): + consent_config = config.get("user_consent") + if consent_config is None: + return + self.user_consent_version = str(consent_config["version"]) + self.user_consent_template_dir = consent_config["template_dir"] + self.user_consent_server_notice_content = consent_config.get( + "server_notice_content", + ) + self.block_events_without_consent_error = consent_config.get( + "block_events_error", + ) + self.user_consent_server_notice_to_guests = bool(consent_config.get( + "send_server_notice_to_guests", False, + )) + + def default_config(self, **kwargs): + return DEFAULT_CONFIG diff --git a/synapse/config/groups.py b/synapse/config/groups.py new file mode 100644 index 000000000..997fa2881 --- /dev/null +++ b/synapse/config/groups.py @@ -0,0 +1,32 @@ +# -*- coding: utf-8 -*- +# Copyright 2017 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from ._base import Config + + +class GroupsConfig(Config): + def read_config(self, config): + self.enable_group_creation = config.get("enable_group_creation", False) + self.group_creation_prefix = config.get("group_creation_prefix", "") + + def default_config(self, **kwargs): + return """\ + # Whether to allow non server admins to create groups on this server + enable_group_creation: false + + # If enabled, non server admins can only create groups with local parts + # starting with this prefix + # group_creation_prefix: "unofficial/" + """ diff --git a/synapse/config/homeserver.py b/synapse/config/homeserver.py index b22cacf8d..2fd9c48ab 100644 --- a/synapse/config/homeserver.py +++ b/synapse/config/homeserver.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2014-2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -12,28 +13,32 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. - -from .tls import TlsConfig -from .server import ServerConfig -from .logger import LoggingConfig -from .database import DatabaseConfig -from .ratelimiting import RatelimitConfig -from .repository import ContentRepositoryConfig -from .captcha import CaptchaConfig -from .voip import VoipConfig -from .registration import RegistrationConfig -from .metrics import MetricsConfig from .api import ApiConfig from .appservice import AppServiceConfig -from .key import KeyConfig -from .saml2 import SAML2Config +from .captcha import CaptchaConfig from .cas import CasConfig -from .password import PasswordConfig -from .jwt import JWTConfig -from .password_auth_providers import PasswordAuthProviderConfig +from .consent_config import ConsentConfig +from .database import DatabaseConfig from .emailconfig import EmailConfig -from .workers import WorkerConfig +from .groups import GroupsConfig +from .jwt import JWTConfig +from .key import KeyConfig +from .logger import LoggingConfig +from .metrics import MetricsConfig +from .password import PasswordConfig +from .password_auth_providers import PasswordAuthProviderConfig from .push import PushConfig +from .ratelimiting import RatelimitConfig +from .registration import RegistrationConfig +from .repository import ContentRepositoryConfig +from .saml2 import SAML2Config +from .server import ServerConfig +from .server_notices_config import ServerNoticesConfig +from .spam_checker import SpamCheckerConfig +from .tls import TlsConfig +from .user_directory import UserDirectoryConfig +from .voip import VoipConfig +from .workers import WorkerConfig class HomeServerConfig(TlsConfig, ServerConfig, DatabaseConfig, LoggingConfig, @@ -41,12 +46,16 @@ class HomeServerConfig(TlsConfig, ServerConfig, DatabaseConfig, LoggingConfig, VoipConfig, RegistrationConfig, MetricsConfig, ApiConfig, AppServiceConfig, KeyConfig, SAML2Config, CasConfig, JWTConfig, PasswordConfig, EmailConfig, - WorkerConfig, PasswordAuthProviderConfig, PushConfig,): + WorkerConfig, PasswordAuthProviderConfig, PushConfig, + SpamCheckerConfig, GroupsConfig, UserDirectoryConfig, + ConsentConfig, + ServerNoticesConfig, + ): pass if __name__ == '__main__': import sys sys.stdout.write( - HomeServerConfig().generate_config(sys.argv[1], sys.argv[2])[0] + HomeServerConfig().generate_config(sys.argv[1], sys.argv[2], True)[0] ) diff --git a/synapse/config/jwt.py b/synapse/config/jwt.py index 47f145c58..51e7f7e00 100644 --- a/synapse/config/jwt.py +++ b/synapse/config/jwt.py @@ -15,7 +15,6 @@ from ._base import Config, ConfigError - MISSING_JWT = ( """Missing jwt library. This is required for jwt login. diff --git a/synapse/config/key.py b/synapse/config/key.py index 6ee643793..279c47bb4 100644 --- a/synapse/config/key.py +++ b/synapse/config/key.py @@ -13,21 +13,24 @@ # See the License for the specific language governing permissions and # limitations under the License. -from ._base import Config, ConfigError - -from synapse.util.stringutils import random_string -from signedjson.key import ( - generate_signing_key, is_signing_algorithm_supported, - decode_signing_key_base64, decode_verify_key_bytes, - read_signing_keys, write_signing_keys, NACL_ED25519 -) -from unpaddedbase64 import decode_base64 -from synapse.util.stringutils import random_string_with_symbols - -import os import hashlib import logging +import os +from signedjson.key import ( + NACL_ED25519, + decode_signing_key_base64, + decode_verify_key_bytes, + generate_signing_key, + is_signing_algorithm_supported, + read_signing_keys, + write_signing_keys, +) +from unpaddedbase64 import decode_base64 + +from synapse.util.stringutils import random_string, random_string_with_symbols + +from ._base import Config, ConfigError logger = logging.getLogger(__name__) @@ -59,14 +62,20 @@ class KeyConfig(Config): self.expire_access_token = config.get("expire_access_token", False) + # a secret which is used to calculate HMACs for form values, to stop + # falsification of values + self.form_secret = config.get("form_secret", None) + def default_config(self, config_dir_path, server_name, is_generating_file=False, **kwargs): base_key_name = os.path.join(config_dir_path, server_name) if is_generating_file: macaroon_secret_key = random_string_with_symbols(50) + form_secret = '"%s"' % random_string_with_symbols(50) else: macaroon_secret_key = None + form_secret = 'null' return """\ macaroon_secret_key: "%(macaroon_secret_key)s" @@ -74,6 +83,10 @@ class KeyConfig(Config): # Used to enable access token expiration. expire_access_token: False + # a secret which is used to calculate HMACs for form values, to stop + # falsification of values + form_secret: %(form_secret)s + ## Signing Keys ## # Path to the signing key to sign messages with @@ -118,10 +131,9 @@ class KeyConfig(Config): signing_keys = self.read_file(signing_key_path, "signing_key") try: return read_signing_keys(signing_keys.splitlines(True)) - except Exception: + except Exception as e: raise ConfigError( - "Error reading signing_key." - " Try running again with --generate-config" + "Error reading signing_key: %s" % (str(e)) ) def read_old_signing_keys(self, old_signing_keys): @@ -141,7 +153,8 @@ class KeyConfig(Config): def generate_files(self, config): signing_key_path = config["signing_key_path"] - if not os.path.exists(signing_key_path): + + if not self.path_exists(signing_key_path): with open(signing_key_path, "w") as signing_key_file: key_id = "a_" + random_string(4) write_signing_keys( diff --git a/synapse/config/logger.py b/synapse/config/logger.py index 2dbeafa9d..a87b11a1d 100644 --- a/synapse/config/logger.py +++ b/synapse/config/logger.py @@ -12,43 +12,48 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. - -from ._base import Config -from synapse.util.logcontext import LoggingContextFilter -from twisted.logger import globalLogBeginner, STDLibLogObserver import logging import logging.config -import yaml -from string import Template import os import signal +import sys +from string import Template +import yaml + +from twisted.logger import STDLibLogObserver, globalLogBeginner + +import synapse +from synapse.util.logcontext import LoggingContextFilter +from synapse.util.versionstring import get_version_string + +from ._base import Config DEFAULT_LOG_CONFIG = Template(""" version: 1 formatters: - precise: - format: '%(asctime)s - %(name)s - %(lineno)d - %(levelname)s - %(request)s\ -- %(message)s' + precise: + format: '%(asctime)s - %(name)s - %(lineno)d - %(levelname)s - \ +%(request)s - %(message)s' filters: - context: - (): synapse.util.logcontext.LoggingContextFilter - request: "" + context: + (): synapse.util.logcontext.LoggingContextFilter + request: "" handlers: - file: - class: logging.handlers.RotatingFileHandler - formatter: precise - filename: ${log_file} - maxBytes: 104857600 - backupCount: 10 - filters: [context] - console: - class: logging.StreamHandler - formatter: precise - filters: [context] + file: + class: logging.handlers.RotatingFileHandler + formatter: precise + filename: ${log_file} + maxBytes: 104857600 + backupCount: 10 + filters: [context] + console: + class: logging.StreamHandler + formatter: precise + filters: [context] loggers: synapse: @@ -74,17 +79,10 @@ class LoggingConfig(Config): self.log_file = self.abspath(config.get("log_file")) def default_config(self, config_dir_path, server_name, **kwargs): - log_file = self.abspath("homeserver.log") log_config = self.abspath( os.path.join(config_dir_path, server_name + ".log.config") ) return """ - # Logging verbosity level. Ignored if log_config is specified. - verbose: 0 - - # File to write logging to. Ignored if log_config is specified. - log_file: "%(log_file)s" - # A yaml python logging config file log_config: "%(log_config)s" """ % locals() @@ -123,9 +121,10 @@ class LoggingConfig(Config): def generate_files(self, config): log_config = config.get("log_config") if log_config and not os.path.exists(log_config): - with open(log_config, "wb") as log_config_file: + log_file = self.abspath("homeserver.log") + with open(log_config, "w") as log_config_file: log_config_file.write( - DEFAULT_LOG_CONFIG.substitute(log_file=config["log_file"]) + DEFAULT_LOG_CONFIG.substitute(log_file=log_file) ) @@ -148,8 +147,11 @@ def setup_logging(config, use_worker_options=False): "%(asctime)s - %(name)s - %(lineno)d - %(levelname)s - %(request)s" " - %(message)s" ) - if log_config is None: + if log_config is None: + # We don't have a logfile, so fall back to the 'verbosity' param from + # the config or cmdline. (Note that we generate a log config for new + # installs, so this will be an unusual case) level = logging.INFO level_for_storage = logging.INFO if config.verbosity: @@ -157,11 +159,10 @@ def setup_logging(config, use_worker_options=False): if config.verbosity > 1: level_for_storage = logging.DEBUG - # FIXME: we need a logging.WARN for a -q quiet option logger = logging.getLogger('') logger.setLevel(level) - logging.getLogger('synapse.storage').setLevel(level_for_storage) + logging.getLogger('synapse.storage.SQL').setLevel(level_for_storage) formatter = logging.Formatter(log_format) if log_file: @@ -176,6 +177,10 @@ def setup_logging(config, use_worker_options=False): logger.info("Opened new log file due to SIGHUP") else: handler = logging.StreamHandler() + + def sighup(signum, stack): + pass + handler.setFormatter(formatter) handler.addFilter(LoggingContextFilter(request="")) @@ -202,6 +207,15 @@ def setup_logging(config, use_worker_options=False): if getattr(signal, "SIGHUP"): signal.signal(signal.SIGHUP, sighup) + # make sure that the first thing we log is a thing we can grep backwards + # for + logging.warn("***** STARTING SERVER *****") + logging.warn( + "Server %s version %s", + sys.argv[0], get_version_string(synapse), + ) + logging.info("Server hostname: %s", config.server_name) + # It's critical to point twisted's internal logging somewhere, otherwise it # stacks up and leaks kup to 64K object; # see: https://twistedmatrix.com/trac/ticket/8164 diff --git a/synapse/config/password_auth_providers.py b/synapse/config/password_auth_providers.py index 83762d089..f4066abc2 100644 --- a/synapse/config/password_auth_providers.py +++ b/synapse/config/password_auth_providers.py @@ -13,44 +13,41 @@ # See the License for the specific language governing permissions and # limitations under the License. -from ._base import Config, ConfigError +from synapse.util.module_loader import load_module -import importlib +from ._base import Config + +LDAP_PROVIDER = 'ldap_auth_provider.LdapAuthProvider' class PasswordAuthProviderConfig(Config): def read_config(self, config): self.password_providers = [] + providers = [] # We want to be backwards compatible with the old `ldap_config` # param. ldap_config = config.get("ldap_config", {}) - self.ldap_enabled = ldap_config.get("enabled", False) - if self.ldap_enabled: - from ldap_auth_provider import LdapAuthProvider - parsed_config = LdapAuthProvider.parse_config(ldap_config) - self.password_providers.append((LdapAuthProvider, parsed_config)) + if ldap_config.get("enabled", False): + providers.append({ + 'module': LDAP_PROVIDER, + 'config': ldap_config, + }) - providers = config.get("password_providers", []) + providers.extend(config.get("password_providers", [])) for provider in providers: + mod_name = provider['module'] + # This is for backwards compat when the ldap auth provider resided # in this package. - if provider['module'] == "synapse.util.ldap_auth_provider.LdapAuthProvider": - from ldap_auth_provider import LdapAuthProvider - provider_class = LdapAuthProvider - else: - # We need to import the module, and then pick the class out of - # that, so we split based on the last dot. - module, clz = provider['module'].rsplit(".", 1) - module = importlib.import_module(module) - provider_class = getattr(module, clz) + if mod_name == "synapse.util.ldap_auth_provider.LdapAuthProvider": + mod_name = LDAP_PROVIDER + + (provider_class, provider_config) = load_module({ + "module": mod_name, + "config": provider['config'], + }) - try: - provider_config = provider_class.parse_config(provider["config"]) - except Exception as e: - raise ConfigError( - "Failed to parse config for %r: %r" % (provider['module'], e) - ) self.password_providers.append((provider_class, provider_config)) def default_config(self, **kwargs): diff --git a/synapse/config/push.py b/synapse/config/push.py index 9c68318b4..b7e0d46af 100644 --- a/synapse/config/push.py +++ b/synapse/config/push.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2015, 2016 OpenMarket Ltd +# Copyright 2017 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -18,28 +19,43 @@ from ._base import Config class PushConfig(Config): def read_config(self, config): - self.push_redact_content = False + push_config = config.get("push", {}) + self.push_include_content = push_config.get("include_content", True) + # There was a a 'redact_content' setting but mistakenly read from the + # 'email'section'. Check for the flag in the 'push' section, and log, + # but do not honour it to avoid nasty surprises when people upgrade. + if push_config.get("redact_content") is not None: + print( + "The push.redact_content content option has never worked. " + "Please set push.include_content if you want this behaviour" + ) + + # Now check for the one in the 'email' section and honour it, + # with a warning. push_config = config.get("email", {}) - self.push_redact_content = push_config.get("redact_content", False) + redact_content = push_config.get("redact_content") + if redact_content is not None: + print( + "The 'email.redact_content' option is deprecated: " + "please set push.include_content instead" + ) + self.push_include_content = not redact_content def default_config(self, config_dir_path, server_name, **kwargs): return """ - # Control how push messages are sent to google/apple to notifications. - # Normally every message said in a room with one or more people using - # mobile devices will be posted to a push server hosted by matrix.org - # which is registered with google and apple in order to allow push - # notifications to be sent to these mobile devices. - # - # Setting redact_content to true will make the push messages contain no - # message content which will provide increased privacy. This is a - # temporary solution pending improvements to Android and iPhone apps - # to get content from the app rather than the notification. - # + # Clients requesting push notifications can either have the body of + # the message sent in the notification poke along with other details + # like the sender, or just the event ID and room ID (`event_id_only`). + # If clients choose the former, this option controls whether the + # notification request includes the content of the event (other details + # like the sender are still included). For `event_id_only` push, it + # has no effect. + # For modern android devices the notification content will still appear # because it is loaded by the app. iPhone, however will send a # notification saying only that a message arrived and who it came from. # #push: - # redact_content: false + # include_content: true """ diff --git a/synapse/config/registration.py b/synapse/config/registration.py index f7e03c4cd..0fb964eb6 100644 --- a/synapse/config/registration.py +++ b/synapse/config/registration.py @@ -13,11 +13,11 @@ # See the License for the specific language governing permissions and # limitations under the License. -from ._base import Config +from distutils.util import strtobool from synapse.util.stringutils import random_string_with_symbols -from distutils.util import strtobool +from ._base import Config class RegistrationConfig(Config): @@ -31,6 +31,8 @@ class RegistrationConfig(Config): strtobool(str(config["disable_registration"])) ) + self.registrations_require_3pid = config.get("registrations_require_3pid", []) + self.allowed_local_3pids = config.get("allowed_local_3pids", []) self.registration_shared_secret = config.get("registration_shared_secret") self.bcrypt_rounds = config.get("bcrypt_rounds", 12) @@ -41,6 +43,8 @@ class RegistrationConfig(Config): self.allow_guest_access and config.get("invite_3pid_guest", False) ) + self.auto_join_rooms = config.get("auto_join_rooms", []) + def default_config(self, **kwargs): registration_shared_secret = random_string_with_symbols(50) @@ -50,13 +54,32 @@ class RegistrationConfig(Config): # Enable registration for new users. enable_registration: False + # The user must provide all of the below types of 3PID when registering. + # + # registrations_require_3pid: + # - email + # - msisdn + + # Mandate that users are only allowed to associate certain formats of + # 3PIDs with accounts on this server. + # + # allowed_local_3pids: + # - medium: email + # pattern: ".*@matrix\\.org" + # - medium: email + # pattern: ".*@vector\\.im" + # - medium: msisdn + # pattern: "\\+44" + # If set, allows registration by anyone who also has the shared # secret, even if registration is otherwise disabled. registration_shared_secret: "%(registration_shared_secret)s" # Set the number of bcrypt rounds used to generate password hash. # Larger numbers increase the work factor needed to generate the hash. - # The default number of rounds is 12. + # The default number is 12 (which equates to 2^12 rounds). + # N.B. that increasing this will exponentially increase the time required + # to register or login - e.g. 24 => 2^24 rounds which will take >20 mins. bcrypt_rounds: 12 # Allows users to register as guests without a password/email/etc, and @@ -70,6 +93,11 @@ class RegistrationConfig(Config): - matrix.org - vector.im - riot.im + + # Users who register on this homeserver will automatically be joined + # to these rooms + #auto_join_rooms: + # - "#example:example.com" """ % locals() def add_arguments(self, parser): diff --git a/synapse/config/repository.py b/synapse/config/repository.py index 2c6f57168..fc909c1fa 100644 --- a/synapse/config/repository.py +++ b/synapse/config/repository.py @@ -13,9 +13,11 @@ # See the License for the specific language governing permissions and # limitations under the License. -from ._base import Config, ConfigError from collections import namedtuple +from synapse.util.module_loader import load_module + +from ._base import Config, ConfigError MISSING_NETADDR = ( "Missing netaddr library. This is required for URL preview API." @@ -36,6 +38,14 @@ ThumbnailRequirement = namedtuple( "ThumbnailRequirement", ["width", "height", "method", "media_type"] ) +MediaStorageProviderConfig = namedtuple( + "MediaStorageProviderConfig", ( + "store_local", # Whether to store newly uploaded local files + "store_remote", # Whether to store newly downloaded remote files + "store_synchronous", # Whether to wait for successful storage for local uploads + ), +) + def parse_thumbnail_requirements(thumbnail_sizes): """ Takes a list of dictionaries with "width", "height", and "method" keys @@ -70,7 +80,64 @@ class ContentRepositoryConfig(Config): self.max_upload_size = self.parse_size(config["max_upload_size"]) self.max_image_pixels = self.parse_size(config["max_image_pixels"]) self.max_spider_size = self.parse_size(config["max_spider_size"]) + self.media_store_path = self.ensure_directory(config["media_store_path"]) + + backup_media_store_path = config.get("backup_media_store_path") + + synchronous_backup_media_store = config.get( + "synchronous_backup_media_store", False + ) + + storage_providers = config.get("media_storage_providers", []) + + if backup_media_store_path: + if storage_providers: + raise ConfigError( + "Cannot use both 'backup_media_store_path' and 'storage_providers'" + ) + + storage_providers = [{ + "module": "file_system", + "store_local": True, + "store_synchronous": synchronous_backup_media_store, + "store_remote": True, + "config": { + "directory": backup_media_store_path, + } + }] + + # This is a list of config that can be used to create the storage + # providers. The entries are tuples of (Class, class_config, + # MediaStorageProviderConfig), where Class is the class of the provider, + # the class_config the config to pass to it, and + # MediaStorageProviderConfig are options for StorageProviderWrapper. + # + # We don't create the storage providers here as not all workers need + # them to be started. + self.media_storage_providers = [] + + for provider_config in storage_providers: + # We special case the module "file_system" so as not to need to + # expose FileStorageProviderBackend + if provider_config["module"] == "file_system": + provider_config["module"] = ( + "synapse.rest.media.v1.storage_provider" + ".FileStorageProviderBackend" + ) + + provider_class, parsed_config = load_module(provider_config) + + wrapper_config = MediaStorageProviderConfig( + provider_config.get("store_local", False), + provider_config.get("store_remote", False), + provider_config.get("store_synchronous", False), + ) + + self.media_storage_providers.append( + (provider_class, parsed_config, wrapper_config,) + ) + self.uploads_path = self.ensure_directory(config["uploads_path"]) self.dynamic_thumbnails = config["dynamic_thumbnails"] self.thumbnail_requirements = parse_thumbnail_requirements( @@ -115,6 +182,20 @@ class ContentRepositoryConfig(Config): # Directory where uploaded images and attachments are stored. media_store_path: "%(media_store)s" + # Media storage providers allow media to be stored in different + # locations. + # media_storage_providers: + # - module: file_system + # # Whether to write new local files. + # store_local: false + # # Whether to write new remote media + # store_remote: false + # # Whether to block upload requests waiting for write to this + # # provider to complete + # store_synchronous: false + # config: + # directory: /mnt/some/other/directory + # Directory where in-progress uploads are stored. uploads_path: "%(uploads_path)s" @@ -169,6 +250,9 @@ class ContentRepositoryConfig(Config): # - '192.168.0.0/16' # - '100.64.0.0/10' # - '169.254.0.0/16' + # - '::1/128' + # - 'fe80::/64' + # - 'fc00::/7' # # List of IP address CIDR ranges that the URL preview spider is allowed # to access even if they are specified in url_preview_ip_range_blacklist. diff --git a/synapse/config/server.py b/synapse/config/server.py index 28b4e5f50..18102656b 100644 --- a/synapse/config/server.py +++ b/synapse/config/server.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2014-2016 OpenMarket Ltd +# Copyright 2017 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -13,13 +14,25 @@ # See the License for the specific language governing permissions and # limitations under the License. +import logging + +from synapse.http.endpoint import parse_and_validate_server_name + from ._base import Config, ConfigError +logger = logging.Logger(__name__) + class ServerConfig(Config): def read_config(self, config): self.server_name = config["server_name"] + + try: + parse_and_validate_server_name(self.server_name) + except ValueError as e: + raise ConfigError(str(e)) + self.pid_file = self.abspath(config.get("pid_file")) self.web_client = config["web_client"] self.web_client_location = config.get("web_client_location", None) @@ -29,6 +42,7 @@ class ServerConfig(Config): self.user_agent_suffix = config.get("user_agent_suffix") self.use_frozen_dicts = config.get("use_frozen_dicts", False) self.public_baseurl = config.get("public_baseurl") + self.cpu_affinity = config.get("cpu_affinity") # Whether to send federation traffic out in this process. This only # applies to some federation traffic, and so shouldn't be used to @@ -39,8 +53,31 @@ class ServerConfig(Config): # false only if we are updating the user directory in a worker self.update_user_directory = config.get("update_user_directory", True) + # whether to enable the media repository endpoints. This should be set + # to false if the media repository is running as a separate endpoint; + # doing so ensures that we will not run cache cleanup jobs on the + # master, potentially causing inconsistency. + self.enable_media_repo = config.get("enable_media_repo", True) + self.filter_timeline_limit = config.get("filter_timeline_limit", -1) + # Whether we should block invites sent to users on this server + # (other than those sent by local server admins) + self.block_non_admin_invites = config.get( + "block_non_admin_invites", False, + ) + + # FIXME: federation_domain_whitelist needs sytests + self.federation_domain_whitelist = None + federation_domain_whitelist = config.get( + "federation_domain_whitelist", None + ) + # turn the whitelist into a hash for speed of lookup + if federation_domain_whitelist is not None: + self.federation_domain_whitelist = {} + for domain in federation_domain_whitelist: + self.federation_domain_whitelist[domain] = True + if self.public_baseurl is not None: if self.public_baseurl[-1] != '/': self.public_baseurl += '/' @@ -113,6 +150,12 @@ class ServerConfig(Config): metrics_port = config.get("metrics_port") if metrics_port: + logger.warn( + ("The metrics_port configuration option is deprecated in Synapse 0.31 " + "in favour of a listener. Please see " + "http://github.com/matrix-org/synapse/blob/master/docs/metrics-howto.rst" + " on how to configure the new listener.")) + self.listeners.append({ "port": metrics_port, "bind_addresses": [config.get("metrics_bind_host", "127.0.0.1")], @@ -127,8 +170,8 @@ class ServerConfig(Config): }) def default_config(self, server_name, **kwargs): - if ":" in server_name: - bind_port = int(server_name.split(":")[1]) + _, bind_port = parse_and_validate_server_name(server_name) + if bind_port is not None: unsecure_port = bind_port - 400 else: bind_port = 8448 @@ -147,6 +190,27 @@ class ServerConfig(Config): # When running as a daemon, the file to store the pid in pid_file: %(pid_file)s + # CPU affinity mask. Setting this restricts the CPUs on which the + # process will be scheduled. It is represented as a bitmask, with the + # lowest order bit corresponding to the first logical CPU and the + # highest order bit corresponding to the last logical CPU. Not all CPUs + # may exist on a given system but a mask may specify more CPUs than are + # present. + # + # For example: + # 0x00000001 is processor #0, + # 0x00000003 is processors #0 and #1, + # 0xFFFFFFFF is all processors (#0 through #31). + # + # Pinning a Python process to a single CPU is desirable, because Python + # is inherently single-threaded due to the GIL, and can suffer a + # 30-40%% slowdown due to cache blow-out and thread context switching + # if the scheduler happens to schedule the underlying threads across + # different cores. See + # https://www.mirantis.com/blog/improve-performance-python-programs-restricting-single-cpu/. + # + # cpu_affinity: 0xFFFFFFFF + # Whether to serve a web client from the HTTP/HTTPS root resource. web_client: True @@ -171,6 +235,21 @@ class ServerConfig(Config): # and sync operations. The default value is -1, means no upper limit. # filter_timeline_limit: 5000 + # Whether room invites to users on this server should be blocked + # (except those sent by local server admins). The default is False. + # block_non_admin_invites: True + + # Restrict federation to the following whitelist of domains. + # N.B. we recommend also firewalling your federation listener to limit + # inbound federation traffic as early as possible, rather than relying + # purely on this application-layer restriction. If not specified, the + # default is to whitelist everything. + # + # federation_domain_whitelist: + # - lon.example.com + # - nyc.example.com + # - syd.example.com + # List of ports that Synapse should listen on, their purpose and their # configuration. listeners: @@ -181,13 +260,12 @@ class ServerConfig(Config): port: %(bind_port)s # Local addresses to listen on. - # This will listen on all IPv4 addresses by default. + # On Linux and Mac OS, `::` will listen on all IPv4 and IPv6 + # addresses by default. For most other OSes, this will only listen + # on IPv6. bind_addresses: + - '::' - '0.0.0.0' - # Uncomment to listen on all IPv6 interfaces - # N.B: On at least Linux this will also listen on all IPv4 - # addresses, so you will need to comment out the line above. - # - '::' # This is a 'http' listener, allows us to specify 'resources'. type: http @@ -214,11 +292,18 @@ class ServerConfig(Config): - names: [federation] # Federation APIs compress: false + # optional list of additional endpoints which can be loaded via + # dynamic modules + # additional_resources: + # "/_matrix/my/custom/endpoint": + # module: my_module.CustomRequestHandler + # config: {} + # Unsecure HTTP listener, # For when matrix traffic passes through loadbalancer that unwraps TLS. - port: %(unsecure_port)s tls: false - bind_addresses: ['0.0.0.0'] + bind_addresses: ['::', '0.0.0.0'] type: http x_forwarded: false @@ -232,7 +317,7 @@ class ServerConfig(Config): # Turn on the twisted ssh manhole service on localhost on the given # port. # - port: 9000 - # bind_address: 127.0.0.1 + # bind_addresses: ['::1', '127.0.0.1'] # type: manhole """ % locals() @@ -270,7 +355,7 @@ def read_gc_thresholds(thresholds): return ( int(thresholds[0]), int(thresholds[1]), int(thresholds[2]), ) - except: + except Exception: raise ConfigError( "Value of `gc_threshold` must be a list of three integers if set" ) diff --git a/synapse/config/server_notices_config.py b/synapse/config/server_notices_config.py new file mode 100644 index 000000000..3c39850ac --- /dev/null +++ b/synapse/config/server_notices_config.py @@ -0,0 +1,87 @@ +# -*- coding: utf-8 -*- +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +from synapse.types import UserID + +from ._base import Config + +DEFAULT_CONFIG = """\ +# Server Notices room configuration +# +# Uncomment this section to enable a room which can be used to send notices +# from the server to users. It is a special room which cannot be left; notices +# come from a special "notices" user id. +# +# If you uncomment this section, you *must* define the system_mxid_localpart +# setting, which defines the id of the user which will be used to send the +# notices. +# +# It's also possible to override the room name, the display name of the +# "notices" user, and the avatar for the user. +# +# server_notices: +# system_mxid_localpart: notices +# system_mxid_display_name: "Server Notices" +# system_mxid_avatar_url: "mxc://server.com/oumMVlgDnLYFaPVkExemNVVZ" +# room_name: "Server Notices" +""" + + +class ServerNoticesConfig(Config): + """Configuration for the server notices room. + + Attributes: + server_notices_mxid (str|None): + The MXID to use for server notices. + None if server notices are not enabled. + + server_notices_mxid_display_name (str|None): + The display name to use for the server notices user. + None if server notices are not enabled. + + server_notices_mxid_avatar_url (str|None): + The display name to use for the server notices user. + None if server notices are not enabled. + + server_notices_room_name (str|None): + The name to use for the server notices room. + None if server notices are not enabled. + """ + def __init__(self): + super(ServerNoticesConfig, self).__init__() + self.server_notices_mxid = None + self.server_notices_mxid_display_name = None + self.server_notices_mxid_avatar_url = None + self.server_notices_room_name = None + + def read_config(self, config): + c = config.get("server_notices") + if c is None: + return + + mxid_localpart = c['system_mxid_localpart'] + self.server_notices_mxid = UserID( + mxid_localpart, self.server_name, + ).to_string() + self.server_notices_mxid_display_name = c.get( + 'system_mxid_display_name', None, + ) + self.server_notices_mxid_avatar_url = c.get( + 'system_mxid_avatar_url', None, + ) + # todo: i18n + self.server_notices_room_name = c.get('room_name', "Server Notices") + + def default_config(self, **kwargs): + return DEFAULT_CONFIG diff --git a/synapse/config/spam_checker.py b/synapse/config/spam_checker.py new file mode 100644 index 000000000..3fec42bdb --- /dev/null +++ b/synapse/config/spam_checker.py @@ -0,0 +1,35 @@ +# -*- coding: utf-8 -*- +# Copyright 2017 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from synapse.util.module_loader import load_module + +from ._base import Config + + +class SpamCheckerConfig(Config): + def read_config(self, config): + self.spam_checker = None + + provider = config.get("spam_checker", None) + if provider is not None: + self.spam_checker = load_module(provider) + + def default_config(self, **kwargs): + return """\ + # spam_checker: + # module: "my_custom_project.SuperSpamChecker" + # config: + # example_option: 'things' + """ diff --git a/synapse/config/tls.py b/synapse/config/tls.py index e081840a8..fef1ea99c 100644 --- a/synapse/config/tls.py +++ b/synapse/config/tls.py @@ -13,14 +13,15 @@ # See the License for the specific language governing permissions and # limitations under the License. -from ._base import Config +import os +import subprocess +from hashlib import sha256 + +from unpaddedbase64 import encode_base64 from OpenSSL import crypto -import subprocess -import os -from hashlib import sha256 -from unpaddedbase64 import encode_base64 +from ._base import Config GENERATE_DH_PARAMS = False @@ -96,7 +97,7 @@ class TlsConfig(Config): # certificates returned by this server match one of the fingerprints. # # Synapse automatically adds the fingerprint of its own certificate - # to the list. So if federation traffic is handle directly by synapse + # to the list. So if federation traffic is handled directly by synapse # then no modification to the list is required. # # If synapse is run behind a load balancer that handles the TLS then it @@ -109,6 +110,12 @@ class TlsConfig(Config): # key. It may be necessary to publish the fingerprints of a new # certificate and wait until the "valid_until_ts" of the previous key # responses have passed before deploying it. + # + # You can calculate a fingerprint from a given TLS listener via: + # openssl s_client -connect $host:$port < /dev/null 2> /dev/null | + # openssl x509 -outform DER | openssl sha256 -binary | base64 | tr -d '=' + # or by checking matrix.org/federationtester/api/report?server_name=$host + # tls_fingerprints: [] # tls_fingerprints: [{"sha256": ""}] """ % locals() @@ -126,8 +133,8 @@ class TlsConfig(Config): tls_private_key_path = config["tls_private_key_path"] tls_dh_params_path = config["tls_dh_params_path"] - if not os.path.exists(tls_private_key_path): - with open(tls_private_key_path, "w") as private_key_file: + if not self.path_exists(tls_private_key_path): + with open(tls_private_key_path, "wb") as private_key_file: tls_private_key = crypto.PKey() tls_private_key.generate_key(crypto.TYPE_RSA, 2048) private_key_pem = crypto.dump_privatekey( @@ -141,8 +148,8 @@ class TlsConfig(Config): crypto.FILETYPE_PEM, private_key_pem ) - if not os.path.exists(tls_certificate_path): - with open(tls_certificate_path, "w") as certificate_file: + if not self.path_exists(tls_certificate_path): + with open(tls_certificate_path, "wb") as certificate_file: cert = crypto.X509() subject = cert.get_subject() subject.CN = config["server_name"] @@ -159,7 +166,7 @@ class TlsConfig(Config): certificate_file.write(cert_pem) - if not os.path.exists(tls_dh_params_path): + if not self.path_exists(tls_dh_params_path): if GENERATE_DH_PARAMS: subprocess.check_call([ "openssl", "dhparam", diff --git a/synapse/config/user_directory.py b/synapse/config/user_directory.py new file mode 100644 index 000000000..38e894784 --- /dev/null +++ b/synapse/config/user_directory.py @@ -0,0 +1,44 @@ +# -*- coding: utf-8 -*- +# Copyright 2017 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from ._base import Config + + +class UserDirectoryConfig(Config): + """User Directory Configuration + Configuration for the behaviour of the /user_directory API + """ + + def read_config(self, config): + self.user_directory_search_all_users = False + user_directory_config = config.get("user_directory", None) + if user_directory_config: + self.user_directory_search_all_users = ( + user_directory_config.get("search_all_users", False) + ) + + def default_config(self, config_dir_path, server_name, **kwargs): + return """ + # User Directory configuration + # + # 'search_all_users' defines whether to search all users visible to your HS + # when searching the user directory, rather than limiting to users visible + # in public rooms. Defaults to false. If you set it True, you'll have to run + # UPDATE user_directory_stream_pos SET stream_id = NULL; + # on your database to tell it to rebuild the user_directory search indexes. + # + #user_directory: + # search_all_users: false + """ diff --git a/synapse/config/voip.py b/synapse/config/voip.py index 3a4e16fa9..d07bd24ff 100644 --- a/synapse/config/voip.py +++ b/synapse/config/voip.py @@ -30,10 +30,10 @@ class VoipConfig(Config): ## Turn ## # The public URIs of the TURN server to give to clients - turn_uris: [] + #turn_uris: [] # The shared secret used to compute passwords for the TURN server - turn_shared_secret: "YOUR_SHARED_SECRET" + #turn_shared_secret: "YOUR_SHARED_SECRET" # The Username and password if the TURN server needs them and # does not use a token diff --git a/synapse/config/workers.py b/synapse/config/workers.py index ea48d931a..80baf0ce0 100644 --- a/synapse/config/workers.py +++ b/synapse/config/workers.py @@ -23,15 +23,31 @@ class WorkerConfig(Config): def read_config(self, config): self.worker_app = config.get("worker_app") + + # Canonicalise worker_app so that master always has None + if self.worker_app == "synapse.app.homeserver": + self.worker_app = None + self.worker_listeners = config.get("worker_listeners") self.worker_daemonize = config.get("worker_daemonize") self.worker_pid_file = config.get("worker_pid_file") self.worker_log_file = config.get("worker_log_file") self.worker_log_config = config.get("worker_log_config") + + # The host used to connect to the main synapse self.worker_replication_host = config.get("worker_replication_host", None) + + # The port on the main synapse for TCP replication self.worker_replication_port = config.get("worker_replication_port", None) + + # The port on the main synapse for HTTP replication endpoint + self.worker_replication_http_port = config.get("worker_replication_http_port") + self.worker_name = config.get("worker_name", self.worker_app) + self.worker_main_http_uri = config.get("worker_main_http_uri", None) + self.worker_cpu_affinity = config.get("worker_cpu_affinity") + if self.worker_listeners: for listener in self.worker_listeners: bind_address = listener.pop("bind_address", None) diff --git a/synapse/crypto/context_factory.py b/synapse/crypto/context_factory.py index aad4752fe..a1e1d0d33 100644 --- a/synapse/crypto/context_factory.py +++ b/synapse/crypto/context_factory.py @@ -12,12 +12,12 @@ # See the License for the specific language governing permissions and # limitations under the License. -from twisted.internet import ssl -from OpenSSL import SSL -from twisted.internet._sslverify import _OpenSSLECCurve, _defaultCurveName - import logging +from OpenSSL import SSL, crypto +from twisted.internet import ssl +from twisted.internet._sslverify import _defaultCurveName + logger = logging.getLogger(__name__) @@ -32,9 +32,10 @@ class ServerContextFactory(ssl.ContextFactory): @staticmethod def configure_context(context, config): try: - _ecCurve = _OpenSSLECCurve(_defaultCurveName) - _ecCurve.addECKeyToContext(context) - except: + _ecCurve = crypto.get_elliptic_curve(_defaultCurveName) + context.set_tmp_ecdh(_ecCurve) + + except Exception: logger.exception("Failed to enable elliptic curve for TLS") context.set_options(SSL.OP_NO_SSLv2 | SSL.OP_NO_SSLv3) context.use_certificate_chain_file(config.tls_certificate_file) diff --git a/synapse/crypto/event_signing.py b/synapse/crypto/event_signing.py index ec7711ba7..8774b2896 100644 --- a/synapse/crypto/event_signing.py +++ b/synapse/crypto/event_signing.py @@ -15,16 +15,16 @@ # limitations under the License. -from synapse.api.errors import SynapseError, Codes -from synapse.events.utils import prune_event - -from canonicaljson import encode_canonical_json -from unpaddedbase64 import encode_base64, decode_base64 -from signedjson.sign import sign_json - import hashlib import logging +from canonicaljson import encode_canonical_json +from signedjson.sign import sign_json +from unpaddedbase64 import decode_base64, encode_base64 + +from synapse.api.errors import Codes, SynapseError +from synapse.events.utils import prune_event + logger = logging.getLogger(__name__) @@ -32,18 +32,25 @@ def check_event_content_hash(event, hash_algorithm=hashlib.sha256): """Check whether the hash for this PDU matches the contents""" name, expected_hash = compute_content_hash(event, hash_algorithm) logger.debug("Expecting hash: %s", encode_base64(expected_hash)) - if name not in event.hashes: + + # some malformed events lack a 'hashes'. Protect against it being missing + # or a weird type by basically treating it the same as an unhashed event. + hashes = event.get("hashes") + if not isinstance(hashes, dict): + raise SynapseError(400, "Malformed 'hashes'", Codes.UNAUTHORIZED) + + if name not in hashes: raise SynapseError( 400, "Algorithm %s not in hashes %s" % ( - name, list(event.hashes), + name, list(hashes), ), Codes.UNAUTHORIZED, ) - message_hash_base64 = event.hashes[name] + message_hash_base64 = hashes[name] try: message_hash_bytes = decode_base64(message_hash_base64) - except: + except Exception: raise SynapseError( 400, "Invalid base64: %s" % (message_hash_base64,), diff --git a/synapse/crypto/keyclient.py b/synapse/crypto/keyclient.py index c2bd64d6c..668b4f517 100644 --- a/synapse/crypto/keyclient.py +++ b/synapse/crypto/keyclient.py @@ -13,17 +13,16 @@ # See the License for the specific language governing permissions and # limitations under the License. - -from twisted.web.http import HTTPClient -from twisted.internet.protocol import Factory -from twisted.internet import defer, reactor -from synapse.http.endpoint import matrix_federation_endpoint -from synapse.util.logcontext import ( - preserve_context_over_fn, preserve_context_over_deferred -) -import simplejson as json import logging +from canonicaljson import json + +from twisted.internet import defer, reactor +from twisted.internet.protocol import Factory +from twisted.web.http import HTTPClient + +from synapse.http.endpoint import matrix_federation_endpoint +from synapse.util import logcontext logger = logging.getLogger(__name__) @@ -43,14 +42,10 @@ def fetch_server_key(server_name, ssl_context_factory, path=KEY_API_V1): for i in range(5): try: - protocol = yield preserve_context_over_fn( - endpoint.connect, factory - ) - server_response, server_certificate = yield preserve_context_over_deferred( - protocol.remote_key - ) - defer.returnValue((server_response, server_certificate)) - return + with logcontext.PreserveLoggingContext(): + protocol = yield endpoint.connect(factory) + server_response, server_certificate = yield protocol.remote_key + defer.returnValue((server_response, server_certificate)) except SynapseKeyClientError as e: logger.exception("Error getting key for %r" % (server_name,)) if e.status.startswith("4"): diff --git a/synapse/crypto/keyring.py b/synapse/crypto/keyring.py index 1bb27edc0..e95b9fb43 100644 --- a/synapse/crypto/keyring.py +++ b/synapse/crypto/keyring.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2014-2016 OpenMarket Ltd +# Copyright 2017 New Vector Ltd. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -13,33 +14,37 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.crypto.keyclient import fetch_server_key -from synapse.api.errors import SynapseError, Codes -from synapse.util import unwrapFirstError -from synapse.util.async import ObservableDeferred -from synapse.util.logcontext import ( - preserve_context_over_deferred, preserve_context_over_fn, PreserveLoggingContext, - preserve_fn -) -from synapse.util.metrics import Measure +import hashlib +import logging +import urllib +from collections import namedtuple -from twisted.internet import defer - -from signedjson.sign import ( - verify_signed_json, signature_ids, sign_json, encode_canonical_json -) from signedjson.key import ( - is_signing_algorithm_supported, decode_verify_key_bytes + decode_verify_key_bytes, + encode_verify_key_base64, + is_signing_algorithm_supported, +) +from signedjson.sign import ( + SignatureVerifyException, + encode_canonical_json, + sign_json, + signature_ids, + verify_signed_json, ) from unpaddedbase64 import decode_base64, encode_base64 from OpenSSL import crypto +from twisted.internet import defer -from collections import namedtuple -import urllib -import hashlib -import logging - +from synapse.api.errors import Codes, SynapseError +from synapse.crypto.keyclient import fetch_server_key +from synapse.util import logcontext, unwrapFirstError +from synapse.util.logcontext import ( + PreserveLoggingContext, + preserve_fn, + run_in_background, +) +from synapse.util.metrics import Measure logger = logging.getLogger(__name__) @@ -55,9 +60,10 @@ Attributes: key_ids(set(str)): The set of key_ids to that could be used to verify the JSON object json_object(dict): The JSON object to verify. - deferred(twisted.internet.defer.Deferred): + deferred(Deferred[str, str, nacl.signing.VerifyKey]): A deferred (server_name, key_id, verify_key) tuple that resolves when - a verify key has been fetched + a verify key has been fetched. The deferreds' callbacks are run with no + logcontext. """ @@ -74,23 +80,32 @@ class Keyring(object): self.perspective_servers = self.config.perspectives self.hs = hs + # map from server name to Deferred. Has an entry for each server with + # an ongoing key download; the Deferred completes once the download + # completes. + # + # These are regular, logcontext-agnostic Deferreds. self.key_downloads = {} def verify_json_for_server(self, server_name, json_object): - return self.verify_json_objects_for_server( - [(server_name, json_object)] - )[0] + return logcontext.make_deferred_yieldable( + self.verify_json_objects_for_server( + [(server_name, json_object)] + )[0] + ) def verify_json_objects_for_server(self, server_and_json): - """Bulk verfies signatures of json objects, bulk fetching keys as + """Bulk verifies signatures of json objects, bulk fetching keys as necessary. Args: server_and_json (list): List of pairs of (server_name, json_object) Returns: - list of deferreds indicating success or failure to verify each - json object's signature for the given server_name. + List: for each input pair, a deferred indicating success + or failure to verify each json object's signature for the given + server_name. The deferreds run their callbacks in the sentinel + logcontext. """ verify_requests = [] @@ -117,73 +132,60 @@ class Keyring(object): verify_requests.append(verify_request) - @defer.inlineCallbacks - def handle_key_deferred(verify_request): - server_name = verify_request.server_name - try: - _, key_id, verify_key = yield verify_request.deferred - except IOError as e: - logger.warn( - "Got IOError when downloading keys for %s: %s %s", - server_name, type(e).__name__, str(e.message), - ) - raise SynapseError( - 502, - "Error downloading keys for %s" % (server_name,), - Codes.UNAUTHORIZED, - ) - except Exception as e: - logger.exception( - "Got Exception when downloading keys for %s: %s %s", - server_name, type(e).__name__, str(e.message), - ) - raise SynapseError( - 401, - "No key for %s with id %s" % (server_name, key_ids), - Codes.UNAUTHORIZED, - ) + run_in_background(self._start_key_lookups, verify_requests) - json_object = verify_request.json_object + # Pass those keys to handle_key_deferred so that the json object + # signatures can be verified + handle = preserve_fn(_handle_key_deferred) + return [ + handle(rq) for rq in verify_requests + ] - logger.debug("Got key %s %s:%s for server %s, verifying" % ( - key_id, verify_key.alg, verify_key.version, server_name, - )) - try: - verify_signed_json(json_object, server_name, verify_key) - except: - raise SynapseError( - 401, - "Invalid signature for server %s with key %s:%s" % ( - server_name, verify_key.alg, verify_key.version - ), - Codes.UNAUTHORIZED, - ) + @defer.inlineCallbacks + def _start_key_lookups(self, verify_requests): + """Sets off the key fetches for each verify request - server_to_deferred = { - server_name: defer.Deferred() - for server_name, _ in server_and_json - } + Once each fetch completes, verify_request.deferred will be resolved. - with PreserveLoggingContext(): + Args: + verify_requests (List[VerifyKeyRequest]): + """ + + try: + # create a deferred for each server we're going to look up the keys + # for; we'll resolve them once we have completed our lookups. + # These will be passed into wait_for_previous_lookups to block + # any other lookups until we have finished. + # The deferreds are called with no logcontext. + server_to_deferred = { + rq.server_name: defer.Deferred() + for rq in verify_requests + } # We want to wait for any previous lookups to complete before # proceeding. - wait_on_deferred = self.wait_for_previous_lookups( - [server_name for server_name, _ in server_and_json], + yield self.wait_for_previous_lookups( + [rq.server_name for rq in verify_requests], server_to_deferred, ) # Actually start fetching keys. - wait_on_deferred.addBoth( - lambda _: self.get_server_verify_keys(verify_requests) - ) + self._get_server_verify_keys(verify_requests) # When we've finished fetching all the keys for a given server_name, # resolve the deferred passed to `wait_for_previous_lookups` so that # any lookups waiting will proceed. + # + # map from server name to a set of request ids server_to_request_ids = {} - def remove_deferreds(res, server_name, verify_request): + for verify_request in verify_requests: + server_name = verify_request.server_name + request_id = id(verify_request) + server_to_request_ids.setdefault(server_name, set()).add(request_id) + + def remove_deferreds(res, verify_request): + server_name = verify_request.server_name request_id = id(verify_request) server_to_request_ids[server_name].discard(request_id) if not server_to_request_ids[server_name]: @@ -193,17 +195,11 @@ class Keyring(object): return res for verify_request in verify_requests: - server_name = verify_request.server_name - request_id = id(verify_request) - server_to_request_ids.setdefault(server_name, set()).add(request_id) - deferred.addBoth(remove_deferreds, server_name, verify_request) - - # Pass those keys to handle_key_deferred so that the json object - # signatures can be verified - return [ - preserve_context_over_fn(handle_key_deferred, verify_request) - for verify_request in verify_requests - ] + verify_request.deferred.addBoth( + remove_deferreds, verify_request, + ) + except Exception: + logger.exception("Error starting key lookups") @defer.inlineCallbacks def wait_for_previous_lookups(self, server_names, server_to_deferred): @@ -212,7 +208,13 @@ class Keyring(object): Args: server_names (list): list of server_names we want to lookup server_to_deferred (dict): server_name to deferred which gets - resolved once we've finished looking up keys for that server + resolved once we've finished looking up keys for that server. + The Deferreds should be regular twisted ones which call their + callbacks with no logcontext. + + Returns: a Deferred which resolves once all key lookups for the given + servers have completed. Follows the synapse rules of logcontext + preservation. """ while True: wait_on = [ @@ -226,17 +228,15 @@ class Keyring(object): else: break + def rm(r, server_name_): + self.key_downloads.pop(server_name_, None) + return r + for server_name, deferred in server_to_deferred.items(): - d = ObservableDeferred(preserve_context_over_deferred(deferred)) - self.key_downloads[server_name] = d + self.key_downloads[server_name] = deferred + deferred.addBoth(rm, server_name) - def rm(r, server_name): - self.key_downloads.pop(server_name, None) - return r - - d.addBoth(rm, server_name) - - def get_server_verify_keys(self, verify_requests): + def _get_server_verify_keys(self, verify_requests): """Tries to find at least one key for each verify request For each verify_request, verify_request.deferred is called back with @@ -305,21 +305,23 @@ class Keyring(object): if not missing_keys: break - for verify_request in requests_missing_keys.values(): - verify_request.deferred.errback(SynapseError( - 401, - "No key for %s with id %s" % ( - verify_request.server_name, verify_request.key_ids, - ), - Codes.UNAUTHORIZED, - )) + with PreserveLoggingContext(): + for verify_request in requests_missing_keys: + verify_request.deferred.errback(SynapseError( + 401, + "No key for %s with id %s" % ( + verify_request.server_name, verify_request.key_ids, + ), + Codes.UNAUTHORIZED, + )) def on_err(err): - for verify_request in verify_requests: - if not verify_request.deferred.called: - verify_request.deferred.errback(err) + with PreserveLoggingContext(): + for verify_request in verify_requests: + if not verify_request.deferred.called: + verify_request.deferred.errback(err) - do_iterations().addErrback(on_err) + run_in_background(do_iterations).addErrback(on_err) @defer.inlineCallbacks def get_keys_from_store(self, server_name_and_key_ids): @@ -333,15 +335,16 @@ class Keyring(object): Deferred: resolves to dict[str, dict[str, VerifyKey]]: map from server_name -> key_id -> VerifyKey """ - res = yield preserve_context_over_deferred(defer.gatherResults( + res = yield logcontext.make_deferred_yieldable(defer.gatherResults( [ - preserve_fn(self.store.get_server_verify_keys)( - server_name, key_ids + run_in_background( + self.store.get_server_verify_keys, + server_name, key_ids, ).addCallback(lambda ks, server: (server, ks), server_name) for server_name, key_ids in server_name_and_key_ids ], consumeErrors=True, - )).addErrback(unwrapFirstError) + ).addErrback(unwrapFirstError)) defer.returnValue(dict(res)) @@ -358,17 +361,17 @@ class Keyring(object): logger.exception( "Unable to get key from %r: %s %s", perspective_name, - type(e).__name__, str(e.message), + type(e).__name__, str(e), ) defer.returnValue({}) - results = yield preserve_context_over_deferred(defer.gatherResults( + results = yield logcontext.make_deferred_yieldable(defer.gatherResults( [ - preserve_fn(get_key)(p_name, p_keys) + run_in_background(get_key, p_name, p_keys) for p_name, p_keys in self.perspective_servers.items() ], consumeErrors=True, - )).addErrback(unwrapFirstError) + ).addErrback(unwrapFirstError)) union_of_keys = {} for result in results: @@ -390,7 +393,7 @@ class Keyring(object): logger.info( "Unable to get key %r for %r directly: %s %s", key_ids, server_name, - type(e).__name__, str(e.message), + type(e).__name__, str(e), ) if not keys: @@ -402,13 +405,13 @@ class Keyring(object): defer.returnValue(keys) - results = yield preserve_context_over_deferred(defer.gatherResults( + results = yield logcontext.make_deferred_yieldable(defer.gatherResults( [ - preserve_fn(get_key)(server_name, key_ids) + run_in_background(get_key, server_name, key_ids) for server_name, key_ids in server_name_and_key_ids ], consumeErrors=True, - )).addErrback(unwrapFirstError) + ).addErrback(unwrapFirstError)) merged = {} for result in results: @@ -485,9 +488,10 @@ class Keyring(object): for server_name, response_keys in processed_response.items(): keys.setdefault(server_name, {}).update(response_keys) - yield preserve_context_over_deferred(defer.gatherResults( + yield logcontext.make_deferred_yieldable(defer.gatherResults( [ - preserve_fn(self.store_keys)( + run_in_background( + self.store_keys, server_name=server_name, from_server=perspective_name, verify_keys=response_keys, @@ -495,7 +499,7 @@ class Keyring(object): for server_name, response_keys in keys.items() ], consumeErrors=True - )).addErrback(unwrapFirstError) + ).addErrback(unwrapFirstError)) defer.returnValue(keys) @@ -543,9 +547,10 @@ class Keyring(object): keys.update(response_keys) - yield preserve_context_over_deferred(defer.gatherResults( + yield logcontext.make_deferred_yieldable(defer.gatherResults( [ - preserve_fn(self.store_keys)( + run_in_background( + self.store_keys, server_name=key_server_name, from_server=server_name, verify_keys=verify_keys, @@ -553,7 +558,7 @@ class Keyring(object): for key_server_name, verify_keys in keys.items() ], consumeErrors=True - )).addErrback(unwrapFirstError) + ).addErrback(unwrapFirstError)) defer.returnValue(keys) @@ -619,9 +624,10 @@ class Keyring(object): response_keys.update(verify_keys) response_keys.update(old_verify_keys) - yield preserve_context_over_deferred(defer.gatherResults( + yield logcontext.make_deferred_yieldable(defer.gatherResults( [ - preserve_fn(self.store.store_server_keys_json)( + run_in_background( + self.store.store_server_keys_json, server_name=server_name, key_id=key_id, from_server=server_name, @@ -632,7 +638,7 @@ class Keyring(object): for key_id in updated_key_ids ], consumeErrors=True, - )).addErrback(unwrapFirstError) + ).addErrback(unwrapFirstError)) results[server_name] = response_keys @@ -710,7 +716,6 @@ class Keyring(object): defer.returnValue(verify_keys) - @defer.inlineCallbacks def store_keys(self, server_name, from_server, verify_keys): """Store a collection of verify keys for a given server Args: @@ -721,12 +726,74 @@ class Keyring(object): A deferred that completes when the keys are stored. """ # TODO(markjh): Store whether the keys have expired. - yield preserve_context_over_deferred(defer.gatherResults( + return logcontext.make_deferred_yieldable(defer.gatherResults( [ - preserve_fn(self.store.store_server_verify_key)( + run_in_background( + self.store.store_server_verify_key, server_name, server_name, key.time_added, key ) for key_id, key in verify_keys.items() ], consumeErrors=True, - )).addErrback(unwrapFirstError) + ).addErrback(unwrapFirstError)) + + +@defer.inlineCallbacks +def _handle_key_deferred(verify_request): + """Waits for the key to become available, and then performs a verification + + Args: + verify_request (VerifyKeyRequest): + + Returns: + Deferred[None] + + Raises: + SynapseError if there was a problem performing the verification + """ + server_name = verify_request.server_name + try: + with PreserveLoggingContext(): + _, key_id, verify_key = yield verify_request.deferred + except IOError as e: + logger.warn( + "Got IOError when downloading keys for %s: %s %s", + server_name, type(e).__name__, str(e), + ) + raise SynapseError( + 502, + "Error downloading keys for %s" % (server_name,), + Codes.UNAUTHORIZED, + ) + except Exception as e: + logger.exception( + "Got Exception when downloading keys for %s: %s %s", + server_name, type(e).__name__, str(e), + ) + raise SynapseError( + 401, + "No key for %s with id %s" % (server_name, verify_request.key_ids), + Codes.UNAUTHORIZED, + ) + + json_object = verify_request.json_object + + logger.debug("Got key %s %s:%s for server %s, verifying" % ( + key_id, verify_key.alg, verify_key.version, server_name, + )) + try: + verify_signed_json(json_object, server_name, verify_key) + except SignatureVerifyException as e: + logger.debug( + "Error verifying signature for %s:%s:%s with key %s: %s", + server_name, verify_key.alg, verify_key.version, + encode_verify_key_base64(verify_key), + str(e), + ) + raise SynapseError( + 401, + "Invalid signature for server %s with key %s:%s: %s" % ( + server_name, verify_key.alg, verify_key.version, str(e), + ), + Codes.UNAUTHORIZED, + ) diff --git a/synapse/event_auth.py b/synapse/event_auth.py index 4096c606f..b32f64e72 100644 --- a/synapse/event_auth.py +++ b/synapse/event_auth.py @@ -17,11 +17,11 @@ import logging from canonicaljson import encode_canonical_json from signedjson.key import decode_verify_key_bytes -from signedjson.sign import verify_signed_json, SignatureVerifyException +from signedjson.sign import SignatureVerifyException, verify_signed_json from unpaddedbase64 import decode_base64 -from synapse.api.constants import EventTypes, Membership, JoinRules -from synapse.api.errors import AuthError, SynapseError, EventSizeError +from synapse.api.constants import EventTypes, JoinRules, Membership +from synapse.api.errors import AuthError, EventSizeError, SynapseError from synapse.types import UserID, get_domain_from_id logger = logging.getLogger(__name__) @@ -34,9 +34,11 @@ def check(event, auth_events, do_sig_check=True, do_size_check=True): event: the event being checked. auth_events (dict: event-key -> event): the existing room state. + Raises: + AuthError if the checks fail Returns: - True if the auth checks pass. + if the auth checks pass. """ if do_size_check: _check_size_limits(event) @@ -71,9 +73,10 @@ def check(event, auth_events, do_sig_check=True, do_size_check=True): # Oh, we don't know what the state of the room was, so we # are trusting that this is allowed (at least for now) logger.warn("Trusting event: %s", event.event_id) - return True + return if event.type == EventTypes.Create: + sender_domain = get_domain_from_id(event.sender) room_id_domain = get_domain_from_id(event.room_id) if room_id_domain != sender_domain: raise AuthError( @@ -81,7 +84,8 @@ def check(event, auth_events, do_sig_check=True, do_size_check=True): "Creation event's room_id domain does not match sender's" ) # FIXME - return True + logger.debug("Allowing! %s", event) + return creation_event = auth_events.get((EventTypes.Create, ""), None) @@ -118,7 +122,8 @@ def check(event, auth_events, do_sig_check=True, do_size_check=True): 403, "Alias event's state_key does not match sender's domain" ) - return True + logger.debug("Allowing! %s", event) + return if logger.isEnabledFor(logging.DEBUG): logger.debug( @@ -127,14 +132,9 @@ def check(event, auth_events, do_sig_check=True, do_size_check=True): ) if event.type == EventTypes.Member: - allowed = _is_membership_change_allowed( - event, auth_events - ) - if allowed: - logger.debug("Allowing! %s", event) - else: - logger.debug("Denying! %s", event) - return allowed + _is_membership_change_allowed(event, auth_events) + logger.debug("Allowing! %s", event) + return _check_event_sender_in_room(event, auth_events) @@ -153,7 +153,8 @@ def check(event, auth_events, do_sig_check=True, do_size_check=True): ) ) else: - return True + logger.debug("Allowing! %s", event) + return _can_send_event(event, auth_events) @@ -200,7 +201,7 @@ def _is_membership_change_allowed(event, auth_events): create = auth_events.get(key) if create and event.prev_events[0][0] == create.event_id: if create.content["creator"] == event.state_key: - return True + return target_user_id = event.state_key @@ -265,13 +266,13 @@ def _is_membership_change_allowed(event, auth_events): raise AuthError( 403, "%s is banned from the room" % (target_user_id,) ) - return True + return if Membership.JOIN != membership: if (caller_invited and Membership.LEAVE == membership and target_user_id == event.user_id): - return True + return if not caller_in_room: # caller isn't joined raise AuthError( @@ -319,7 +320,7 @@ def _is_membership_change_allowed(event, auth_events): # TODO (erikj): Implement kicks. if target_banned and user_level < ban_level: raise AuthError( - 403, "You cannot unban user &s." % (target_user_id,) + 403, "You cannot unban user %s." % (target_user_id,) ) elif target_user_id != event.user_id: kick_level = _get_named_level(auth_events, "kick", 50) @@ -334,8 +335,6 @@ def _is_membership_change_allowed(event, auth_events): else: raise AuthError(500, "Unknown membership %s" % membership) - return True - def _check_event_sender_in_room(event, auth_events): key = (EventTypes.Member, event.user_id, ) @@ -355,35 +354,46 @@ def _check_joined_room(member, user_id, room_id): )) -def get_send_level(etype, state_key, auth_events): - key = (EventTypes.PowerLevels, "", ) - send_level_event = auth_events.get(key) - send_level = None - if send_level_event: - send_level = send_level_event.content.get("events", {}).get( - etype - ) - if send_level is None: - if state_key is not None: - send_level = send_level_event.content.get( - "state_default", 50 - ) - else: - send_level = send_level_event.content.get( - "events_default", 0 - ) +def get_send_level(etype, state_key, power_levels_event): + """Get the power level required to send an event of a given type - if send_level: - send_level = int(send_level) + The federation spec [1] refers to this as "Required Power Level". + + https://matrix.org/docs/spec/server_server/unstable.html#definitions + + Args: + etype (str): type of event + state_key (str|None): state_key of state event, or None if it is not + a state event. + power_levels_event (synapse.events.EventBase|None): power levels event + in force at this point in the room + Returns: + int: power level required to send this event. + """ + + if power_levels_event: + power_levels_content = power_levels_event.content else: - send_level = 0 + power_levels_content = {} - return send_level + # see if we have a custom level for this event type + send_level = power_levels_content.get("events", {}).get(etype) + + # otherwise, fall back to the state_default/events_default. + if send_level is None: + if state_key is not None: + send_level = power_levels_content.get("state_default", 50) + else: + send_level = power_levels_content.get("events_default", 0) + + return int(send_level) def _can_send_event(event, auth_events): + power_levels_event = _get_power_level_event(auth_events) + send_level = get_send_level( - event.type, event.get("state_key", None), auth_events + event.type, event.get("state_key"), power_levels_event, ) user_level = get_user_power_level(event.user_id, auth_events) @@ -443,12 +453,12 @@ def _check_power_levels(event, auth_events): for k, v in user_list.items(): try: UserID.from_string(k) - except: + except Exception: raise SynapseError(400, "Not a valid user_id: %s" % (k,)) try: int(v) - except: + except Exception: raise SynapseError(400, "Not a valid power level: %s" % (v,)) key = (event.type, event.state_key, ) @@ -470,15 +480,15 @@ def _check_power_levels(event, auth_events): ("invite", None), ] - old_list = current_state.content.get("users") - for user in set(old_list.keys() + user_list.keys()): + old_list = current_state.content.get("users", {}) + for user in set(list(old_list) + list(user_list)): levels_to_check.append( (user, "users") ) - old_list = current_state.content.get("events") - new_list = event.content.get("events") - for ev_id in set(old_list.keys() + new_list.keys()): + old_list = current_state.content.get("events", {}) + new_list = event.content.get("events", {}) + for ev_id in set(list(old_list) + list(new_list)): levels_to_check.append( (ev_id, "events") ) @@ -515,7 +525,11 @@ def _check_power_levels(event, auth_events): "to your own" ) - if old_level > user_level or new_level > user_level: + # Check if the old and new levels are greater than the user level + # (if defined) + old_level_too_big = old_level is not None and old_level > user_level + new_level_too_big = new_level is not None and new_level > user_level + if old_level_too_big or new_level_too_big: raise AuthError( 403, "You don't have permission to add ops level greater " @@ -524,13 +538,22 @@ def _check_power_levels(event, auth_events): def _get_power_level_event(auth_events): - key = (EventTypes.PowerLevels, "", ) - return auth_events.get(key) + return auth_events.get((EventTypes.PowerLevels, "")) def get_user_power_level(user_id, auth_events): - power_level_event = _get_power_level_event(auth_events) + """Get a user's power level + Args: + user_id (str): user's id to look up in power_levels + auth_events (dict[(str, str), synapse.events.EventBase]): + state in force at this point in the room (or rather, a subset of + it including at least the create event and power levels event. + + Returns: + int: the user's power level in this room. + """ + power_level_event = _get_power_level_event(auth_events) if power_level_event: level = power_level_event.content.get("users", {}).get(user_id) if not level: @@ -541,6 +564,11 @@ def get_user_power_level(user_id, auth_events): else: return int(level) else: + # if there is no power levels event, the creator gets 100 and everyone + # else gets 0. + + # some things which call this don't pass the create event: hack around + # that. key = (EventTypes.Create, "", ) create_event = auth_events.get(key) if (create_event is not None and diff --git a/synapse/events/__init__.py b/synapse/events/__init__.py index e673e96cc..51f9084b9 100644 --- a/synapse/events/__init__.py +++ b/synapse/events/__init__.py @@ -13,9 +13,8 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.util.frozenutils import freeze from synapse.util.caches import intern_dict - +from synapse.util.frozenutils import freeze # Whether we should use frozen_dict in FrozenEvent. Using frozen_dicts prevents # bugs where we accidentally share e.g. signature dicts. However, converting @@ -47,14 +46,26 @@ class _EventInternalMetadata(object): def _event_dict_property(key): + # We want to be able to use hasattr with the event dict properties. + # However, (on python3) hasattr expects AttributeError to be raised. Hence, + # we need to transform the KeyError into an AttributeError def getter(self): - return self._event_dict[key] + try: + return self._event_dict[key] + except KeyError: + raise AttributeError(key) def setter(self, v): - self._event_dict[key] = v + try: + self._event_dict[key] = v + except KeyError: + raise AttributeError(key) def delete(self): - del self._event_dict[key] + try: + del self._event_dict[key] + except KeyError: + raise AttributeError(key) return property( getter, @@ -134,7 +145,7 @@ class EventBase(object): return field in self._event_dict def items(self): - return self._event_dict.items() + return list(self._event_dict.items()) class FrozenEvent(EventBase): diff --git a/synapse/events/builder.py b/synapse/events/builder.py index 365fd96bd..e662eaef1 100644 --- a/synapse/events/builder.py +++ b/synapse/events/builder.py @@ -13,13 +13,12 @@ # See the License for the specific language governing permissions and # limitations under the License. -from . import EventBase, FrozenEvent, _event_dict_property +import copy from synapse.types import EventID - from synapse.util.stringutils import random_string -import copy +from . import EventBase, FrozenEvent, _event_dict_property class EventBuilder(EventBase): @@ -55,7 +54,7 @@ class EventBuilderFactory(object): local_part = str(int(self.clock.time())) + i + random_string(5) - e_id = EventID.create(local_part, self.hostname) + e_id = EventID(local_part, self.hostname) return e_id.to_string() diff --git a/synapse/events/snapshot.py b/synapse/events/snapshot.py index e9a732ff0..368b5f6ae 100644 --- a/synapse/events/snapshot.py +++ b/synapse/events/snapshot.py @@ -13,19 +13,21 @@ # See the License for the specific language governing permissions and # limitations under the License. +from six import iteritems + +from frozendict import frozendict + +from twisted.internet import defer + +from synapse.util.logcontext import make_deferred_yieldable, run_in_background + class EventContext(object): """ Attributes: - current_state_ids (dict[(str, str), str]): - The current state map including the current event. - (type, state_key) -> event_id - - prev_state_ids (dict[(str, str), str]): - The current state map excluding the current event. - (type, state_key) -> event_id - - state_group (int): state group id + state_group (int|None): state group id, if the state has been stored + as a state group. This is usually only None if e.g. the event is + an outlier. rejected (bool|str): A rejection reason if the event was rejected, else False @@ -39,35 +41,250 @@ class EventContext(object): prev_state_events (?): XXX: is this ever set to anything other than the empty list? + + _current_state_ids (dict[(str, str), str]|None): + The current state map including the current event. None if outlier + or we haven't fetched the state from DB yet. + (type, state_key) -> event_id + + _prev_state_ids (dict[(str, str), str]|None): + The current state map excluding the current event. None if outlier + or we haven't fetched the state from DB yet. + (type, state_key) -> event_id + + _fetching_state_deferred (Deferred|None): Resolves when *_state_ids have + been calculated. None if we haven't started calculating yet + + _event_type (str): The type of the event the context is associated with. + Only set when state has not been fetched yet. + + _event_state_key (str|None): The state_key of the event the context is + associated with. Only set when state has not been fetched yet. + + _prev_state_id (str|None): If the event associated with the context is + a state event, then `_prev_state_id` is the event_id of the state + that was replaced. + Only set when state has not been fetched yet. """ __slots__ = [ - "current_state_ids", - "prev_state_ids", "state_group", "rejected", - "push_actions", "prev_group", "delta_ids", "prev_state_events", "app_service", + "_current_state_ids", + "_prev_state_ids", + "_prev_state_id", + "_event_type", + "_event_state_key", + "_fetching_state_deferred", ] def __init__(self): - # The current state including the current event - self.current_state_ids = None - # The current state excluding the current event - self.prev_state_ids = None - self.state_group = None - + self.prev_state_events = [] self.rejected = False - self.push_actions = [] + self.app_service = None + + @staticmethod + def with_state(state_group, current_state_ids, prev_state_ids, + prev_group=None, delta_ids=None): + context = EventContext() + + # The current state including the current event + context._current_state_ids = current_state_ids + # The current state excluding the current event + context._prev_state_ids = prev_state_ids + context.state_group = state_group + + context._prev_state_id = None + context._event_type = None + context._event_state_key = None + context._fetching_state_deferred = defer.succeed(None) # A previously persisted state group and a delta between that # and this state. - self.prev_group = None - self.delta_ids = None + context.prev_group = prev_group + context.delta_ids = delta_ids - self.prev_state_events = None + return context - self.app_service = None + @defer.inlineCallbacks + def serialize(self, event, store): + """Converts self to a type that can be serialized as JSON, and then + deserialized by `deserialize` + + Args: + event (FrozenEvent): The event that this context relates to + + Returns: + dict + """ + + # We don't serialize the full state dicts, instead they get pulled out + # of the DB on the other side. However, the other side can't figure out + # the prev_state_ids, so if we're a state event we include the event + # id that we replaced in the state. + if event.is_state(): + prev_state_ids = yield self.get_prev_state_ids(store) + prev_state_id = prev_state_ids.get((event.type, event.state_key)) + else: + prev_state_id = None + + defer.returnValue({ + "prev_state_id": prev_state_id, + "event_type": event.type, + "event_state_key": event.state_key if event.is_state() else None, + "state_group": self.state_group, + "rejected": self.rejected, + "prev_group": self.prev_group, + "delta_ids": _encode_state_dict(self.delta_ids), + "prev_state_events": self.prev_state_events, + "app_service_id": self.app_service.id if self.app_service else None + }) + + @staticmethod + def deserialize(store, input): + """Converts a dict that was produced by `serialize` back into a + EventContext. + + Args: + store (DataStore): Used to convert AS ID to AS object + input (dict): A dict produced by `serialize` + + Returns: + EventContext + """ + context = EventContext() + + # We use the state_group and prev_state_id stuff to pull the + # current_state_ids out of the DB and construct prev_state_ids. + context._prev_state_id = input["prev_state_id"] + context._event_type = input["event_type"] + context._event_state_key = input["event_state_key"] + + context._current_state_ids = None + context._prev_state_ids = None + context._fetching_state_deferred = None + + context.state_group = input["state_group"] + context.prev_group = input["prev_group"] + context.delta_ids = _decode_state_dict(input["delta_ids"]) + + context.rejected = input["rejected"] + context.prev_state_events = input["prev_state_events"] + + app_service_id = input["app_service_id"] + if app_service_id: + context.app_service = store.get_app_service_by_id(app_service_id) + + return context + + @defer.inlineCallbacks + def get_current_state_ids(self, store): + """Gets the current state IDs + + Returns: + Deferred[dict[(str, str), str]|None]: Returns None if state_group + is None, which happens when the associated event is an outlier. + """ + + if not self._fetching_state_deferred: + self._fetching_state_deferred = run_in_background( + self._fill_out_state, store, + ) + + yield make_deferred_yieldable(self._fetching_state_deferred) + + defer.returnValue(self._current_state_ids) + + @defer.inlineCallbacks + def get_prev_state_ids(self, store): + """Gets the prev state IDs + + Returns: + Deferred[dict[(str, str), str]|None]: Returns None if state_group + is None, which happens when the associated event is an outlier. + """ + + if not self._fetching_state_deferred: + self._fetching_state_deferred = run_in_background( + self._fill_out_state, store, + ) + + yield make_deferred_yieldable(self._fetching_state_deferred) + + defer.returnValue(self._prev_state_ids) + + def get_cached_current_state_ids(self): + """Gets the current state IDs if we have them already cached. + + Returns: + dict[(str, str), str]|None: Returns None if we haven't cached the + state or if state_group is None, which happens when the associated + event is an outlier. + """ + + return self._current_state_ids + + @defer.inlineCallbacks + def _fill_out_state(self, store): + """Called to populate the _current_state_ids and _prev_state_ids + attributes by loading from the database. + """ + if self.state_group is None: + return + + self._current_state_ids = yield store.get_state_ids_for_group( + self.state_group, + ) + if self._prev_state_id and self._event_state_key is not None: + self._prev_state_ids = dict(self._current_state_ids) + + key = (self._event_type, self._event_state_key) + self._prev_state_ids[key] = self._prev_state_id + else: + self._prev_state_ids = self._current_state_ids + + @defer.inlineCallbacks + def update_state(self, state_group, prev_state_ids, current_state_ids, + prev_group, delta_ids): + """Replace the state in the context + """ + + # We need to make sure we wait for any ongoing fetching of state + # to complete so that the updated state doesn't get clobbered + if self._fetching_state_deferred: + yield make_deferred_yieldable(self._fetching_state_deferred) + + self.state_group = state_group + self._prev_state_ids = prev_state_ids + self.prev_group = prev_group + self._current_state_ids = current_state_ids + self.delta_ids = delta_ids + + # We need to ensure that that we've marked as having fetched the state + self._fetching_state_deferred = defer.succeed(None) + + +def _encode_state_dict(state_dict): + """Since dicts of (type, state_key) -> event_id cannot be serialized in + JSON we need to convert them to a form that can. + """ + if state_dict is None: + return None + + return [ + (etype, state_key, v) + for (etype, state_key), v in iteritems(state_dict) + ] + + +def _decode_state_dict(input): + """Decodes a state dict encoded using `_encode_state_dict` above + """ + if input is None: + return None + + return frozendict({(etype, state_key,): v for etype, state_key, v in input}) diff --git a/synapse/events/spamcheck.py b/synapse/events/spamcheck.py new file mode 100644 index 000000000..633e068eb --- /dev/null +++ b/synapse/events/spamcheck.py @@ -0,0 +1,113 @@ +# -*- coding: utf-8 -*- +# Copyright 2017 New Vector Ltd. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + + +class SpamChecker(object): + def __init__(self, hs): + self.spam_checker = None + + module = None + config = None + try: + module, config = hs.config.spam_checker + except Exception: + pass + + if module is not None: + self.spam_checker = module(config=config) + + def check_event_for_spam(self, event): + """Checks if a given event is considered "spammy" by this server. + + If the server considers an event spammy, then it will be rejected if + sent by a local user. If it is sent by a user on another server, then + users receive a blank event. + + Args: + event (synapse.events.EventBase): the event to be checked + + Returns: + bool: True if the event is spammy. + """ + if self.spam_checker is None: + return False + + return self.spam_checker.check_event_for_spam(event) + + def user_may_invite(self, inviter_userid, invitee_userid, room_id): + """Checks if a given user may send an invite + + If this method returns false, the invite will be rejected. + + Args: + userid (string): The sender's user ID + + Returns: + bool: True if the user may send an invite, otherwise False + """ + if self.spam_checker is None: + return True + + return self.spam_checker.user_may_invite(inviter_userid, invitee_userid, room_id) + + def user_may_create_room(self, userid): + """Checks if a given user may create a room + + If this method returns false, the creation request will be rejected. + + Args: + userid (string): The sender's user ID + + Returns: + bool: True if the user may create a room, otherwise False + """ + if self.spam_checker is None: + return True + + return self.spam_checker.user_may_create_room(userid) + + def user_may_create_room_alias(self, userid, room_alias): + """Checks if a given user may create a room alias + + If this method returns false, the association request will be rejected. + + Args: + userid (string): The sender's user ID + room_alias (string): The alias to be created + + Returns: + bool: True if the user may create a room alias, otherwise False + """ + if self.spam_checker is None: + return True + + return self.spam_checker.user_may_create_room_alias(userid, room_alias) + + def user_may_publish_room(self, userid, room_id): + """Checks if a given user may publish a room to the directory + + If this method returns false, the publish request will be rejected. + + Args: + userid (string): The sender's user ID + room_id (string): The ID of the room that would be published + + Returns: + bool: True if the user may publish the room, otherwise False + """ + if self.spam_checker is None: + return True + + return self.spam_checker.user_may_publish_room(userid, room_id) diff --git a/synapse/events/utils.py b/synapse/events/utils.py index 824f4a42e..652941ca0 100644 --- a/synapse/events/utils.py +++ b/synapse/events/utils.py @@ -13,12 +13,15 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.api.constants import EventTypes -from . import EventBase +import re + +from six import string_types from frozendict import frozendict -import re +from synapse.api.constants import EventTypes + +from . import EventBase # Split strings on "." but not "\." This uses a negative lookbehind assertion for '\' # (? MAX_DEPTH: + raise SynapseError(400, "Depth too large", Codes.BAD_JSON) + + event = FrozenEvent( + pdu_json + ) + + event.internal_metadata.outlier = outlier + + return event diff --git a/synapse/federation/federation_client.py b/synapse/federation/federation_client.py index 861441708..62d7ed13c 100644 --- a/synapse/federation/federation_client.py +++ b/synapse/federation/federation_client.py @@ -14,36 +14,35 @@ # limitations under the License. -from twisted.internet import defer - -from .federation_base import FederationBase -from synapse.api.constants import Membership - -from synapse.api.errors import ( - CodeMessageException, HttpResponseException, SynapseError, -) -from synapse.util import unwrapFirstError -from synapse.util.caches.expiringcache import ExpiringCache -from synapse.util.logutils import log_function -from synapse.util.logcontext import preserve_fn, preserve_context_over_deferred -from synapse.events import FrozenEvent, builder -import synapse.metrics - -from synapse.util.retryutils import NotRetryingDestination - import copy import itertools import logging import random +from six.moves import range + +from prometheus_client import Counter + +from twisted.internet import defer + +from synapse.api.constants import Membership +from synapse.api.errors import ( + CodeMessageException, + FederationDeniedError, + HttpResponseException, + SynapseError, +) +from synapse.events import builder +from synapse.federation.federation_base import FederationBase, event_from_pdu_json +from synapse.util import logcontext, unwrapFirstError +from synapse.util.caches.expiringcache import ExpiringCache +from synapse.util.logcontext import make_deferred_yieldable, run_in_background +from synapse.util.logutils import log_function +from synapse.util.retryutils import NotRetryingDestination logger = logging.getLogger(__name__) - -# synapse.federation.federation_client is a silly name -metrics = synapse.metrics.get_metrics_for("synapse.federation.client") - -sent_queries_counter = metrics.register_counter("sent_queries", labels=["type"]) +sent_queries_counter = Counter("synapse_federation_client_sent_queries", "", ["type"]) PDU_RETRY_TIME_MS = 1 * 60 * 1000 @@ -58,6 +57,7 @@ class FederationClient(FederationBase): self._clear_tried_cache, 60 * 1000, ) self.state = hs.get_state_handler() + self.transport_layer = hs.get_federation_transport_client() def _clear_tried_cache(self): """Clear pdu_destination_tried cache""" @@ -105,7 +105,7 @@ class FederationClient(FederationBase): a Deferred which will eventually yield a JSON object from the response """ - sent_queries_counter.inc(query_type) + sent_queries_counter.labels(query_type).inc() return self.transport_layer.make_query( destination, query_type, args, retry_on_dns_fail=retry_on_dns_fail, @@ -124,7 +124,7 @@ class FederationClient(FederationBase): a Deferred which will eventually yield a JSON object from the response """ - sent_queries_counter.inc("client_device_keys") + sent_queries_counter.labels("client_device_keys").inc() return self.transport_layer.query_client_keys( destination, content, timeout ) @@ -134,7 +134,7 @@ class FederationClient(FederationBase): """Query the device keys for a list of user ids hosted on a remote server. """ - sent_queries_counter.inc("user_devices") + sent_queries_counter.labels("user_devices").inc() return self.transport_layer.query_user_devices( destination, user_id, timeout ) @@ -151,7 +151,7 @@ class FederationClient(FederationBase): a Deferred which will eventually yield a JSON object from the response """ - sent_queries_counter.inc("client_one_time_keys") + sent_queries_counter.labels("client_one_time_keys").inc() return self.transport_layer.claim_client_keys( destination, content, timeout ) @@ -184,15 +184,15 @@ class FederationClient(FederationBase): logger.debug("backfill transaction_data=%s", repr(transaction_data)) pdus = [ - self.event_from_pdu_json(p, outlier=False) + event_from_pdu_json(p, outlier=False) for p in transaction_data["pdus"] ] # FIXME: We should handle signature failures more gracefully. - pdus[:] = yield preserve_context_over_deferred(defer.gatherResults( + pdus[:] = yield logcontext.make_deferred_yieldable(defer.gatherResults( self._check_sigs_and_hashes(pdus), consumeErrors=True, - )).addErrback(unwrapFirstError) + ).addErrback(unwrapFirstError)) defer.returnValue(pdus) @@ -244,7 +244,7 @@ class FederationClient(FederationBase): logger.debug("transaction_data %r", transaction_data) pdu_list = [ - self.event_from_pdu_json(p, outlier=outlier) + event_from_pdu_json(p, outlier=outlier) for p in transaction_data["pdus"] ] @@ -252,7 +252,7 @@ class FederationClient(FederationBase): pdu = pdu_list[0] # Check signatures are correct. - signed_pdu = yield self._check_sigs_and_hashes([pdu])[0] + signed_pdu = yield self._check_sigs_and_hash(pdu) break @@ -266,6 +266,9 @@ class FederationClient(FederationBase): except NotRetryingDestination as e: logger.info(e.message) continue + except FederationDeniedError as e: + logger.info(e.message) + continue except Exception as e: pdu_attempts[destination] = now @@ -336,11 +339,11 @@ class FederationClient(FederationBase): ) pdus = [ - self.event_from_pdu_json(p, outlier=True) for p in result["pdus"] + event_from_pdu_json(p, outlier=True) for p in result["pdus"] ] auth_chain = [ - self.event_from_pdu_json(p, outlier=True) + event_from_pdu_json(p, outlier=True) for p in result.get("auth_chain", []) ] @@ -388,9 +391,9 @@ class FederationClient(FederationBase): """ if return_local: seen_events = yield self.store.get_events(event_ids, allow_rejected=True) - signed_events = seen_events.values() + signed_events = list(seen_events.values()) else: - seen_events = yield self.store.have_events(event_ids) + seen_events = yield self.store.have_seen_events(event_ids) signed_events = [] failed_to_fetch = set() @@ -409,18 +412,19 @@ class FederationClient(FederationBase): batch_size = 20 missing_events = list(missing_events) - for i in xrange(0, len(missing_events), batch_size): + for i in range(0, len(missing_events), batch_size): batch = set(missing_events[i:i + batch_size]) deferreds = [ - preserve_fn(self.get_pdu)( + run_in_background( + self.get_pdu, destinations=random_server_list(), event_id=e_id, ) for e_id in batch ] - res = yield preserve_context_over_deferred( + res = yield make_deferred_yieldable( defer.DeferredList(deferreds, consumeErrors=True) ) for success, result in res: @@ -441,7 +445,7 @@ class FederationClient(FederationBase): ) auth_chain = [ - self.event_from_pdu_json(p, outlier=True) + event_from_pdu_json(p, outlier=True) for p in res["auth_chain"] ] @@ -570,12 +574,12 @@ class FederationClient(FederationBase): logger.debug("Got content: %s", content) state = [ - self.event_from_pdu_json(p, outlier=True) + event_from_pdu_json(p, outlier=True) for p in content.get("state", []) ] auth_chain = [ - self.event_from_pdu_json(p, outlier=True) + event_from_pdu_json(p, outlier=True) for p in content.get("auth_chain", []) ] @@ -585,7 +589,7 @@ class FederationClient(FederationBase): } valid_pdus = yield self._check_sigs_and_hash_and_fetch( - destination, pdus.values(), + destination, list(pdus.values()), outlier=True, ) @@ -650,7 +654,7 @@ class FederationClient(FederationBase): logger.debug("Got response to send_invite: %s", pdu_dict) - pdu = self.event_from_pdu_json(pdu_dict) + pdu = event_from_pdu_json(pdu_dict) # Check signatures are correct. pdu = yield self._check_sigs_and_hash(pdu) @@ -740,7 +744,7 @@ class FederationClient(FederationBase): ) auth_chain = [ - self.event_from_pdu_json(e) + event_from_pdu_json(e) for e in content["auth_chain"] ] @@ -788,7 +792,7 @@ class FederationClient(FederationBase): ) events = [ - self.event_from_pdu_json(e) + event_from_pdu_json(e) for e in content.get("events", []) ] @@ -805,15 +809,6 @@ class FederationClient(FederationBase): defer.returnValue(signed_events) - def event_from_pdu_json(self, pdu_json, outlier=False): - event = FrozenEvent( - pdu_json - ) - - event.internal_metadata.outlier = outlier - - return event - @defer.inlineCallbacks def forward_third_party_invite(self, destinations, room_id, event_dict): for destination in destinations: diff --git a/synapse/federation/federation_server.py b/synapse/federation/federation_server.py index 51e3fdea0..e501251b6 100644 --- a/synapse/federation/federation_server.py +++ b/synapse/federation/federation_server.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2015, 2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -12,92 +13,72 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. +import logging +import re +import six +from six import iteritems + +from canonicaljson import json +from prometheus_client import Counter from twisted.internet import defer +from twisted.internet.abstract import isIPAddress +from twisted.python import failure -from .federation_base import FederationBase -from .units import Transaction, Edu - -from synapse.util.async import Linearizer -from synapse.util.logutils import log_function -from synapse.util.caches.response_cache import ResponseCache -from synapse.events import FrozenEvent -from synapse.types import get_domain_from_id -import synapse.metrics - -from synapse.api.errors import AuthError, FederationError, SynapseError - +from synapse.api.constants import EventTypes +from synapse.api.errors import AuthError, FederationError, NotFoundError, SynapseError from synapse.crypto.event_signing import compute_event_signature +from synapse.federation.federation_base import FederationBase, event_from_pdu_json +from synapse.federation.persistence import TransactionActions +from synapse.federation.units import Edu, Transaction +from synapse.http.endpoint import parse_server_name +from synapse.types import get_domain_from_id +from synapse.util import async +from synapse.util.caches.response_cache import ResponseCache +from synapse.util.logutils import log_function -import simplejson as json -import logging - +# when processing incoming transactions, we try to handle multiple rooms in +# parallel, up to this limit. +TRANSACTION_CONCURRENCY_LIMIT = 10 logger = logging.getLogger(__name__) -# synapse.federation.federation_server is a silly name -metrics = synapse.metrics.get_metrics_for("synapse.federation.server") +received_pdus_counter = Counter("synapse_federation_server_received_pdus", "") -received_pdus_counter = metrics.register_counter("received_pdus") +received_edus_counter = Counter("synapse_federation_server_received_edus", "") -received_edus_counter = metrics.register_counter("received_edus") - -received_queries_counter = metrics.register_counter("received_queries", labels=["type"]) +received_queries_counter = Counter( + "synapse_federation_server_received_queries", "", ["type"] +) class FederationServer(FederationBase): + def __init__(self, hs): super(FederationServer, self).__init__(hs) self.auth = hs.get_auth() + self.handler = hs.get_handlers().federation_handler - self._server_linearizer = Linearizer("fed_server") + self._server_linearizer = async.Linearizer("fed_server") + self._transaction_linearizer = async.Linearizer("fed_txn_handler") + + self.transaction_actions = TransactionActions(self.store) + + self.registry = hs.get_federation_registry() # We cache responses to state queries, as they take a while and often # come in waves. - self._state_resp_cache = ResponseCache(hs, timeout_ms=30000) - - def set_handler(self, handler): - """Sets the handler that the replication layer will use to communicate - receipt of new PDUs from other home servers. The required methods are - documented on :py:class:`.ReplicationHandler`. - """ - self.handler = handler - - def register_edu_handler(self, edu_type, handler): - if edu_type in self.edu_handlers: - raise KeyError("Already have an EDU handler for %s" % (edu_type,)) - - self.edu_handlers[edu_type] = handler - - def register_query_handler(self, query_type, handler): - """Sets the handler callable that will be used to handle an incoming - federation Query of the given type. - - Args: - query_type (str): Category name of the query, which should match - the string used by make_query. - handler (callable): Invoked to handle incoming queries of this type - - handler is invoked as: - result = handler(args) - - where 'args' is a dict mapping strings to strings of the query - arguments. It should return a Deferred that will eventually yield an - object to encode as JSON. - """ - if query_type in self.query_handlers: - raise KeyError( - "Already have a Query handler for %s" % (query_type,) - ) - - self.query_handlers[query_type] = handler + self._state_resp_cache = ResponseCache(hs, "state_resp", timeout_ms=30000) @defer.inlineCallbacks @log_function def on_backfill_request(self, origin, room_id, versions, limit): with (yield self._server_linearizer.queue((origin, room_id))): + origin_host, _ = parse_server_name(origin) + yield self.check_server_matches_acl(origin_host, room_id) + pdus = yield self.handler.on_backfill_request( origin, room_id, versions, limit ) @@ -109,25 +90,41 @@ class FederationServer(FederationBase): @defer.inlineCallbacks @log_function def on_incoming_transaction(self, transaction_data): + # keep this as early as possible to make the calculated origin ts as + # accurate as possible. + request_time = self._clock.time_msec() + transaction = Transaction(**transaction_data) - received_pdus_counter.inc_by(len(transaction.pdus)) - - for p in transaction.pdus: - if "unsigned" in p: - unsigned = p["unsigned"] - if "age" in unsigned: - p["age"] = unsigned["age"] - if "age" in p: - p["age_ts"] = int(self._clock.time_msec()) - int(p["age"]) - del p["age"] - - pdu_list = [ - self.event_from_pdu_json(p) for p in transaction.pdus - ] + if not transaction.transaction_id: + raise Exception("Transaction missing transaction_id") + if not transaction.origin: + raise Exception("Transaction missing origin") logger.debug("[%s] Got transaction", transaction.transaction_id) + # use a linearizer to ensure that we don't process the same transaction + # multiple times in parallel. + with (yield self._transaction_linearizer.queue( + (transaction.origin, transaction.transaction_id), + )): + result = yield self._handle_incoming_transaction( + transaction, request_time, + ) + + defer.returnValue(result) + + @defer.inlineCallbacks + def _handle_incoming_transaction(self, transaction, request_time): + """ Process an incoming transaction and return the HTTP response + + Args: + transaction (Transaction): incoming transaction + request_time (int): timestamp that the HTTP request arrived at + + Returns: + Deferred[(int, object)]: http response code and body + """ response = yield self.transaction_actions.have_responded(transaction) if response: @@ -140,42 +137,67 @@ class FederationServer(FederationBase): logger.debug("[%s] Transaction is new", transaction.transaction_id) - results = [] + received_pdus_counter.inc(len(transaction.pdus)) - for pdu in pdu_list: - # check that it's actually being sent from a valid destination to - # workaround bug #1753 in 0.18.5 and 0.18.6 - if transaction.origin != get_domain_from_id(pdu.event_id): - # We continue to accept join events from any server; this is - # necessary for the federation join dance to work correctly. - # (When we join over federation, the "helper" server is - # responsible for sending out the join event, rather than the - # origin. See bug #1893). - if not ( - pdu.type == 'm.room.member' and - pdu.content and - pdu.content.get("membership", None) == 'join' - ): - logger.info( - "Discarding PDU %s from invalid origin %s", - pdu.event_id, transaction.origin - ) - continue - else: - logger.info( - "Accepting join PDU %s from %s", - pdu.event_id, transaction.origin - ) + origin_host, _ = parse_server_name(transaction.origin) + pdus_by_room = {} + + for p in transaction.pdus: + if "unsigned" in p: + unsigned = p["unsigned"] + if "age" in unsigned: + p["age"] = unsigned["age"] + if "age" in p: + p["age_ts"] = request_time - int(p["age"]) + del p["age"] + + event = event_from_pdu_json(p) + room_id = event.room_id + pdus_by_room.setdefault(room_id, []).append(event) + + pdu_results = {} + + # we can process different rooms in parallel (which is useful if they + # require callouts to other servers to fetch missing events), but + # impose a limit to avoid going too crazy with ram/cpu. + + @defer.inlineCallbacks + def process_pdus_for_room(room_id): + logger.debug("Processing PDUs for %s", room_id) try: - yield self._handle_received_pdu(transaction.origin, pdu) - results.append({}) - except FederationError as e: - self.send_failure(e, transaction.origin) - results.append({"error": str(e)}) - except Exception as e: - results.append({"error": str(e)}) - logger.exception("Failed to handle PDU") + yield self.check_server_matches_acl(origin_host, room_id) + except AuthError as e: + logger.warn( + "Ignoring PDUs for room %s from banned server", room_id, + ) + for pdu in pdus_by_room[room_id]: + event_id = pdu.event_id + pdu_results[event_id] = e.error_dict() + return + + for pdu in pdus_by_room[room_id]: + event_id = pdu.event_id + try: + yield self._handle_received_pdu( + transaction.origin, pdu + ) + pdu_results[event_id] = {} + except FederationError as e: + logger.warn("Error handling PDU %s: %s", event_id, e) + pdu_results[event_id] = {"error": str(e)} + except Exception as e: + f = failure.Failure() + pdu_results[event_id] = {"error": str(e)} + logger.error( + "Failed to handle PDU %s: %s", + event_id, f.getTraceback().rstrip(), + ) + + yield async.concurrently_execute( + process_pdus_for_room, pdus_by_room.keys(), + TRANSACTION_CONCURRENCY_LIMIT, + ) if hasattr(transaction, "edus"): for edu in (Edu(**x) for x in transaction.edus): @@ -185,17 +207,16 @@ class FederationServer(FederationBase): edu.content ) - for failure in getattr(transaction, "pdu_failures", []): - logger.info("Got failure %r", failure) - - logger.debug("Returning: %s", str(results)) + pdu_failures = getattr(transaction, "pdu_failures", []) + for fail in pdu_failures: + logger.info("Got failure %r", fail) response = { - "pdus": dict(zip( - (p.event_id for p in pdu_list), results - )), + "pdus": pdu_results, } + logger.debug("Returning: %s", str(response)) + yield self.transaction_actions.set_response( transaction, 200, response @@ -205,16 +226,7 @@ class FederationServer(FederationBase): @defer.inlineCallbacks def received_edu(self, origin, edu_type, content): received_edus_counter.inc() - - if edu_type in self.edu_handlers: - try: - yield self.edu_handlers[edu_type](origin, content) - except SynapseError as e: - logger.info("Failed to handle edu %r: %r", edu_type, e) - except Exception as e: - logger.exception("Failed to handle edu %r", edu_type) - else: - logger.warn("Received EDU of type %s with no handler", edu_type) + yield self.registry.on_edu(edu_type, origin, content) @defer.inlineCallbacks @log_function @@ -222,19 +234,24 @@ class FederationServer(FederationBase): if not event_id: raise NotImplementedError("Specify an event") + origin_host, _ = parse_server_name(origin) + yield self.check_server_matches_acl(origin_host, room_id) + in_room = yield self.auth.check_host_in_room(room_id, origin) if not in_room: raise AuthError(403, "Host not in room.") - result = self._state_resp_cache.get((room_id, event_id)) - if not result: - with (yield self._server_linearizer.queue((origin, room_id))): - resp = yield self._state_resp_cache.set( - (room_id, event_id), - self._on_context_state_request_compute(room_id, event_id) - ) - else: - resp = yield result + # we grab the linearizer to protect ourselves from servers which hammer + # us. In theory we might already have the response to this query + # in the cache so we could return it without waiting for the linearizer + # - but that's non-trivial to get right, and anyway somewhat defeats + # the point of the linearizer. + with (yield self._server_linearizer.queue((origin, room_id))): + resp = yield self._state_resp_cache.wrap( + (room_id, event_id), + self._on_context_state_request_compute, + room_id, event_id, + ) defer.returnValue((200, resp)) @@ -243,6 +260,9 @@ class FederationServer(FederationBase): if not event_id: raise NotImplementedError("Specify an event") + origin_host, _ = parse_server_name(origin) + yield self.check_server_matches_acl(origin_host, room_id) + in_room = yield self.auth.check_host_in_room(room_id, origin) if not in_room: raise AuthError(403, "Host not in room.") @@ -286,7 +306,7 @@ class FederationServer(FederationBase): @defer.inlineCallbacks @log_function def on_pdu_request(self, origin, event_id): - pdu = yield self._get_persisted_pdu(origin, event_id) + pdu = yield self.handler.get_persisted_pdu(origin, event_id) if pdu: defer.returnValue( @@ -302,25 +322,23 @@ class FederationServer(FederationBase): @defer.inlineCallbacks def on_query_request(self, query_type, args): - received_queries_counter.inc(query_type) - - if query_type in self.query_handlers: - response = yield self.query_handlers[query_type](args) - defer.returnValue((200, response)) - else: - defer.returnValue( - (404, "No handler for Query type '%s'" % (query_type,)) - ) + received_queries_counter.labels(query_type).inc() + resp = yield self.registry.on_query(query_type, args) + defer.returnValue((200, resp)) @defer.inlineCallbacks - def on_make_join_request(self, room_id, user_id): + def on_make_join_request(self, origin, room_id, user_id): + origin_host, _ = parse_server_name(origin) + yield self.check_server_matches_acl(origin_host, room_id) pdu = yield self.handler.on_make_join_request(room_id, user_id) time_now = self._clock.time_msec() defer.returnValue({"event": pdu.get_pdu_json(time_now)}) @defer.inlineCallbacks def on_invite_request(self, origin, content): - pdu = self.event_from_pdu_json(content) + pdu = event_from_pdu_json(content) + origin_host, _ = parse_server_name(origin) + yield self.check_server_matches_acl(origin_host, pdu.room_id) ret_pdu = yield self.handler.on_invite_request(origin, pdu) time_now = self._clock.time_msec() defer.returnValue((200, {"event": ret_pdu.get_pdu_json(time_now)})) @@ -328,7 +346,11 @@ class FederationServer(FederationBase): @defer.inlineCallbacks def on_send_join_request(self, origin, content): logger.debug("on_send_join_request: content: %s", content) - pdu = self.event_from_pdu_json(content) + pdu = event_from_pdu_json(content) + + origin_host, _ = parse_server_name(origin) + yield self.check_server_matches_acl(origin_host, pdu.room_id) + logger.debug("on_send_join_request: pdu sigs: %s", pdu.signatures) res_pdus = yield self.handler.on_send_join_request(origin, pdu) time_now = self._clock.time_msec() @@ -340,7 +362,9 @@ class FederationServer(FederationBase): })) @defer.inlineCallbacks - def on_make_leave_request(self, room_id, user_id): + def on_make_leave_request(self, origin, room_id, user_id): + origin_host, _ = parse_server_name(origin) + yield self.check_server_matches_acl(origin_host, room_id) pdu = yield self.handler.on_make_leave_request(room_id, user_id) time_now = self._clock.time_msec() defer.returnValue({"event": pdu.get_pdu_json(time_now)}) @@ -348,7 +372,11 @@ class FederationServer(FederationBase): @defer.inlineCallbacks def on_send_leave_request(self, origin, content): logger.debug("on_send_leave_request: content: %s", content) - pdu = self.event_from_pdu_json(content) + pdu = event_from_pdu_json(content) + + origin_host, _ = parse_server_name(origin) + yield self.check_server_matches_acl(origin_host, pdu.room_id) + logger.debug("on_send_leave_request: pdu sigs: %s", pdu.signatures) yield self.handler.on_send_leave_request(origin, pdu) defer.returnValue((200, {})) @@ -356,6 +384,9 @@ class FederationServer(FederationBase): @defer.inlineCallbacks def on_event_auth(self, origin, room_id, event_id): with (yield self._server_linearizer.queue((origin, room_id))): + origin_host, _ = parse_server_name(origin) + yield self.check_server_matches_acl(origin_host, room_id) + time_now = self._clock.time_msec() auth_pdus = yield self.handler.on_event_auth(event_id) res = { @@ -384,8 +415,11 @@ class FederationServer(FederationBase): Deferred: Results in `dict` with the same format as `content` """ with (yield self._server_linearizer.queue((origin, room_id))): + origin_host, _ = parse_server_name(origin) + yield self.check_server_matches_acl(origin_host, room_id) + auth_chain = [ - self.event_from_pdu_json(e) + event_from_pdu_json(e) for e in content["auth_chain"] ] @@ -444,9 +478,9 @@ class FederationServer(FederationBase): "Claimed one-time-keys: %s", ",".join(( "%s for %s:%s" % (key_id, user_id, device_id) - for user_id, user_keys in json_result.iteritems() - for device_id, device_keys in user_keys.iteritems() - for key_id, _ in device_keys.iteritems() + for user_id, user_keys in iteritems(json_result) + for device_id, device_keys in iteritems(user_keys) + for key_id, _ in iteritems(device_keys) )), ) @@ -457,6 +491,9 @@ class FederationServer(FederationBase): def on_get_missing_events(self, origin, room_id, earliest_events, latest_events, limit, min_depth): with (yield self._server_linearizer.queue((origin, room_id))): + origin_host, _ = parse_server_name(origin) + yield self.check_server_matches_acl(origin_host, room_id) + logger.info( "on_get_missing_events: earliest_events: %r, latest_events: %r," " limit: %d, min_depth: %d", @@ -485,17 +522,6 @@ class FederationServer(FederationBase): ts_now_ms = self._clock.time_msec() return self.store.get_user_id_for_open_id_token(token, ts_now_ms) - @log_function - def _get_persisted_pdu(self, origin, event_id, do_auth=True): - """ Get a PDU from the database with given origin and id. - - Returns: - Deferred: Results in a `Pdu`. - """ - return self.handler.get_persisted_pdu( - origin, event_id, do_auth=do_auth - ) - def _transaction_from_pdus(self, pdu_list): """Returns a new Transaction containing the given PDUs suitable for transmission. @@ -513,13 +539,57 @@ class FederationServer(FederationBase): def _handle_received_pdu(self, origin, pdu): """ Process a PDU received in a federation /send/ transaction. + If the event is invalid, then this method throws a FederationError. + (The error will then be logged and sent back to the sender (which + probably won't do anything with it), and other events in the + transaction will be processed as normal). + + It is likely that we'll then receive other events which refer to + this rejected_event in their prev_events, etc. When that happens, + we'll attempt to fetch the rejected event again, which will presumably + fail, so those second-generation events will also get rejected. + + Eventually, we get to the point where there are more than 10 events + between any new events and the original rejected event. Since we + only try to backfill 10 events deep on received pdu, we then accept the + new event, possibly introducing a discontinuity in the DAG, with new + forward extremities, so normal service is approximately returned, + until we try to backfill across the discontinuity. + Args: origin (str): server which sent the pdu pdu (FrozenEvent): received pdu Returns (Deferred): completes with None - Raises: FederationError if the signatures / hash do not match - """ + + Raises: FederationError if the signatures / hash do not match, or + if the event was unacceptable for any other reason (eg, too large, + too many prev_events, couldn't find the prev_events) + """ + # check that it's actually being sent from a valid destination to + # workaround bug #1753 in 0.18.5 and 0.18.6 + if origin != get_domain_from_id(pdu.event_id): + # We continue to accept join events from any server; this is + # necessary for the federation join dance to work correctly. + # (When we join over federation, the "helper" server is + # responsible for sending out the join event, rather than the + # origin. See bug #1893). + if not ( + pdu.type == 'm.room.member' and + pdu.content and + pdu.content.get("membership", None) == 'join' + ): + logger.info( + "Discarding PDU %s from invalid origin %s", + pdu.event_id, origin + ) + return + else: + logger.info( + "Accepting join PDU %s from %s", + pdu.event_id, origin + ) + # Check signature. try: pdu = yield self._check_sigs_and_hash(pdu) @@ -531,20 +601,13 @@ class FederationServer(FederationBase): affected=pdu.event_id, ) - yield self.handler.on_receive_pdu(origin, pdu, get_missing=True) + yield self.handler.on_receive_pdu( + origin, pdu, get_missing=True, sent_to_us_directly=True, + ) def __str__(self): return "" % self.server_name - def event_from_pdu_json(self, pdu_json, outlier=False): - event = FrozenEvent( - pdu_json - ) - - event.internal_metadata.outlier = outlier - - return event - @defer.inlineCallbacks def exchange_third_party_invite( self, @@ -567,3 +630,161 @@ class FederationServer(FederationBase): origin, room_id, event_dict ) defer.returnValue(ret) + + @defer.inlineCallbacks + def check_server_matches_acl(self, server_name, room_id): + """Check if the given server is allowed by the server ACLs in the room + + Args: + server_name (str): name of server, *without any port part* + room_id (str): ID of the room to check + + Raises: + AuthError if the server does not match the ACL + """ + state_ids = yield self.store.get_current_state_ids(room_id) + acl_event_id = state_ids.get((EventTypes.ServerACL, "")) + + if not acl_event_id: + return + + acl_event = yield self.store.get_event(acl_event_id) + if server_matches_acl_event(server_name, acl_event): + return + + raise AuthError(code=403, msg="Server is banned from room") + + +def server_matches_acl_event(server_name, acl_event): + """Check if the given server is allowed by the ACL event + + Args: + server_name (str): name of server, without any port part + acl_event (EventBase): m.room.server_acl event + + Returns: + bool: True if this server is allowed by the ACLs + """ + logger.debug("Checking %s against acl %s", server_name, acl_event.content) + + # first of all, check if literal IPs are blocked, and if so, whether the + # server name is a literal IP + allow_ip_literals = acl_event.content.get("allow_ip_literals", True) + if not isinstance(allow_ip_literals, bool): + logger.warn("Ignorning non-bool allow_ip_literals flag") + allow_ip_literals = True + if not allow_ip_literals: + # check for ipv6 literals. These start with '['. + if server_name[0] == '[': + return False + + # check for ipv4 literals. We can just lift the routine from twisted. + if isIPAddress(server_name): + return False + + # next, check the deny list + deny = acl_event.content.get("deny", []) + if not isinstance(deny, (list, tuple)): + logger.warn("Ignorning non-list deny ACL %s", deny) + deny = [] + for e in deny: + if _acl_entry_matches(server_name, e): + # logger.info("%s matched deny rule %s", server_name, e) + return False + + # then the allow list. + allow = acl_event.content.get("allow", []) + if not isinstance(allow, (list, tuple)): + logger.warn("Ignorning non-list allow ACL %s", allow) + allow = [] + for e in allow: + if _acl_entry_matches(server_name, e): + # logger.info("%s matched allow rule %s", server_name, e) + return True + + # everything else should be rejected. + # logger.info("%s fell through", server_name) + return False + + +def _acl_entry_matches(server_name, acl_entry): + if not isinstance(acl_entry, six.string_types): + logger.warn("Ignoring non-str ACL entry '%s' (is %s)", acl_entry, type(acl_entry)) + return False + regex = _glob_to_regex(acl_entry) + return regex.match(server_name) + + +def _glob_to_regex(glob): + res = '' + for c in glob: + if c == '*': + res = res + '.*' + elif c == '?': + res = res + '.' + else: + res = res + re.escape(c) + return re.compile(res + "\\Z", re.IGNORECASE) + + +class FederationHandlerRegistry(object): + """Allows classes to register themselves as handlers for a given EDU or + query type for incoming federation traffic. + """ + def __init__(self): + self.edu_handlers = {} + self.query_handlers = {} + + def register_edu_handler(self, edu_type, handler): + """Sets the handler callable that will be used to handle an incoming + federation EDU of the given type. + + Args: + edu_type (str): The type of the incoming EDU to register handler for + handler (Callable[[str, dict]]): A callable invoked on incoming EDU + of the given type. The arguments are the origin server name and + the EDU contents. + """ + if edu_type in self.edu_handlers: + raise KeyError("Already have an EDU handler for %s" % (edu_type,)) + + self.edu_handlers[edu_type] = handler + + def register_query_handler(self, query_type, handler): + """Sets the handler callable that will be used to handle an incoming + federation query of the given type. + + Args: + query_type (str): Category name of the query, which should match + the string used by make_query. + handler (Callable[[dict], Deferred[dict]]): Invoked to handle + incoming queries of this type. The return will be yielded + on and the result used as the response to the query request. + """ + if query_type in self.query_handlers: + raise KeyError( + "Already have a Query handler for %s" % (query_type,) + ) + + self.query_handlers[query_type] = handler + + @defer.inlineCallbacks + def on_edu(self, edu_type, origin, content): + handler = self.edu_handlers.get(edu_type) + if not handler: + logger.warn("No handler registered for EDU type %s", edu_type) + + try: + yield handler(origin, content) + except SynapseError as e: + logger.info("Failed to handle edu %r: %r", edu_type, e) + except Exception as e: + logger.exception("Failed to handle edu %r", edu_type) + + def on_query(self, query_type, args): + handler = self.query_handlers.get(query_type) + if not handler: + logger.warn("No handler registered for query type %s", query_type) + raise NotFoundError("No handler for Query type '%s'" % (query_type,)) + + return handler(args) diff --git a/synapse/federation/persistence.py b/synapse/federation/persistence.py index 84dc60667..9146215c2 100644 --- a/synapse/federation/persistence.py +++ b/synapse/federation/persistence.py @@ -19,13 +19,12 @@ package. These actions are mostly only used by the :py:mod:`.replication` module. """ +import logging + from twisted.internet import defer from synapse.util.logutils import log_function -import logging - - logger = logging.getLogger(__name__) diff --git a/synapse/federation/replication.py b/synapse/federation/replication.py deleted file mode 100644 index 62d865ec4..000000000 --- a/synapse/federation/replication.py +++ /dev/null @@ -1,73 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2014-2016 OpenMarket Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -"""This layer is responsible for replicating with remote home servers using -a given transport. -""" - -from .federation_client import FederationClient -from .federation_server import FederationServer - -from .persistence import TransactionActions - -import logging - - -logger = logging.getLogger(__name__) - - -class ReplicationLayer(FederationClient, FederationServer): - """This layer is responsible for replicating with remote home servers over - the given transport. I.e., does the sending and receiving of PDUs to - remote home servers. - - The layer communicates with the rest of the server via a registered - ReplicationHandler. - - In more detail, the layer: - * Receives incoming data and processes it into transactions and pdus. - * Fetches any PDUs it thinks it might have missed. - * Keeps the current state for contexts up to date by applying the - suitable conflict resolution. - * Sends outgoing pdus wrapped in transactions. - * Fills out the references to previous pdus/transactions appropriately - for outgoing data. - """ - - def __init__(self, hs, transport_layer): - self.server_name = hs.hostname - - self.keyring = hs.get_keyring() - - self.transport_layer = transport_layer - - self.federation_client = self - - self.store = hs.get_datastore() - - self.handler = None - self.edu_handlers = {} - self.query_handlers = {} - - self._clock = hs.get_clock() - - self.transaction_actions = TransactionActions(self.store) - - self.hs = hs - - super(ReplicationLayer, self).__init__(hs) - - def __str__(self): - return "" % self.server_name diff --git a/synapse/federation/send_queue.py b/synapse/federation/send_queue.py index 93e5acebc..5157c3860 100644 --- a/synapse/federation/send_queue.py +++ b/synapse/federation/send_queue.py @@ -29,23 +29,22 @@ dead worker doesn't cause the queues to grow limitlessly. Events are replicated via a separate events stream. """ -from .units import Edu - -from synapse.storage.presence import UserPresenceState -from synapse.util.metrics import Measure -import synapse.metrics - -from blist import sorteddict +import logging from collections import namedtuple -import logging +from six import iteritems, itervalues + +from sortedcontainers import SortedDict + +from synapse.metrics import LaterGauge +from synapse.storage.presence import UserPresenceState +from synapse.util.metrics import Measure + +from .units import Edu logger = logging.getLogger(__name__) -metrics = synapse.metrics.get_metrics_for(__name__) - - class FederationRemoteSendQueue(object): """A drop in replacement for TransactionQueue""" @@ -56,29 +55,27 @@ class FederationRemoteSendQueue(object): self.is_mine_id = hs.is_mine_id self.presence_map = {} # Pending presence map user_id -> UserPresenceState - self.presence_changed = sorteddict() # Stream position -> user_id + self.presence_changed = SortedDict() # Stream position -> user_id self.keyed_edu = {} # (destination, key) -> EDU - self.keyed_edu_changed = sorteddict() # stream position -> (destination, key) + self.keyed_edu_changed = SortedDict() # stream position -> (destination, key) - self.edus = sorteddict() # stream position -> Edu + self.edus = SortedDict() # stream position -> Edu - self.failures = sorteddict() # stream position -> (destination, Failure) + self.failures = SortedDict() # stream position -> (destination, Failure) - self.device_messages = sorteddict() # stream position -> destination + self.device_messages = SortedDict() # stream position -> destination self.pos = 1 - self.pos_time = sorteddict() + self.pos_time = SortedDict() # EVERYTHING IS SAD. In particular, python only makes new scopes when # we make a new function, so we need to make a new function so the inner # lambda binds to the queue rather than to the name of the queue which # changes. ARGH. def register(name, queue): - metrics.register_callback( - queue_name + "_size", - lambda: len(queue), - ) + LaterGauge("synapse_federation_send_queue_%s_size" % (queue_name,), + "", [], lambda: len(queue)) for queue_name in [ "presence_map", "presence_changed", "keyed_edu", "keyed_edu_changed", @@ -101,7 +98,7 @@ class FederationRemoteSendQueue(object): now = self.clock.time_msec() keys = self.pos_time.keys() - time = keys.bisect_left(now - FIVE_MINUTES_AGO) + time = self.pos_time.bisect_left(now - FIVE_MINUTES_AGO) if not keys[:time]: return @@ -116,13 +113,13 @@ class FederationRemoteSendQueue(object): with Measure(self.clock, "send_queue._clear"): # Delete things out of presence maps keys = self.presence_changed.keys() - i = keys.bisect_left(position_to_delete) + i = self.presence_changed.bisect_left(position_to_delete) for key in keys[:i]: del self.presence_changed[key] user_ids = set( user_id - for uids in self.presence_changed.itervalues() + for uids in itervalues(self.presence_changed) for user_id in uids ) @@ -134,7 +131,7 @@ class FederationRemoteSendQueue(object): # Delete things out of keyed edus keys = self.keyed_edu_changed.keys() - i = keys.bisect_left(position_to_delete) + i = self.keyed_edu_changed.bisect_left(position_to_delete) for key in keys[:i]: del self.keyed_edu_changed[key] @@ -148,19 +145,19 @@ class FederationRemoteSendQueue(object): # Delete things out of edu map keys = self.edus.keys() - i = keys.bisect_left(position_to_delete) + i = self.edus.bisect_left(position_to_delete) for key in keys[:i]: del self.edus[key] # Delete things out of failure map keys = self.failures.keys() - i = keys.bisect_left(position_to_delete) + i = self.failures.bisect_left(position_to_delete) for key in keys[:i]: del self.failures[key] # Delete things out of device map keys = self.device_messages.keys() - i = keys.bisect_left(position_to_delete) + i = self.device_messages.bisect_left(position_to_delete) for key in keys[:i]: del self.device_messages[key] @@ -200,7 +197,7 @@ class FederationRemoteSendQueue(object): # We only want to send presence for our own users, so lets always just # filter here just in case. - local_states = filter(lambda s: self.is_mine_id(s.user_id), states) + local_states = list(filter(lambda s: self.is_mine_id(s.user_id), states)) self.presence_map.update({state.user_id: state for state in local_states}) self.presence_changed[pos] = [state.user_id for state in local_states] @@ -253,13 +250,12 @@ class FederationRemoteSendQueue(object): self._clear_queue_before_pos(federation_ack) # Fetch changed presence - keys = self.presence_changed.keys() - i = keys.bisect_right(from_token) - j = keys.bisect_right(to_token) + 1 + i = self.presence_changed.bisect_right(from_token) + j = self.presence_changed.bisect_right(to_token) + 1 dest_user_ids = [ (pos, user_id) - for pos in keys[i:j] - for user_id in self.presence_changed[pos] + for pos, user_id_list in self.presence_changed.items()[i:j] + for user_id in user_id_list ] for (key, user_id) in dest_user_ids: @@ -268,34 +264,31 @@ class FederationRemoteSendQueue(object): ))) # Fetch changes keyed edus - keys = self.keyed_edu_changed.keys() - i = keys.bisect_right(from_token) - j = keys.bisect_right(to_token) + 1 + i = self.keyed_edu_changed.bisect_right(from_token) + j = self.keyed_edu_changed.bisect_right(to_token) + 1 # We purposefully clobber based on the key here, python dict comprehensions # always use the last value, so this will correctly point to the last # stream position. - keyed_edus = {self.keyed_edu_changed[k]: k for k in keys[i:j]} + keyed_edus = {v: k for k, v in self.keyed_edu_changed.items()[i:j]} - for ((destination, edu_key), pos) in keyed_edus.iteritems(): + for ((destination, edu_key), pos) in iteritems(keyed_edus): rows.append((pos, KeyedEduRow( key=edu_key, edu=self.keyed_edu[(destination, edu_key)], ))) # Fetch changed edus - keys = self.edus.keys() - i = keys.bisect_right(from_token) - j = keys.bisect_right(to_token) + 1 - edus = ((k, self.edus[k]) for k in keys[i:j]) + i = self.edus.bisect_right(from_token) + j = self.edus.bisect_right(to_token) + 1 + edus = self.edus.items()[i:j] for (pos, edu) in edus: rows.append((pos, EduRow(edu))) # Fetch changed failures - keys = self.failures.keys() - i = keys.bisect_right(from_token) - j = keys.bisect_right(to_token) + 1 - failures = ((k, self.failures[k]) for k in keys[i:j]) + i = self.failures.bisect_right(from_token) + j = self.failures.bisect_right(to_token) + 1 + failures = self.failures.items()[i:j] for (pos, (destination, failure)) in failures: rows.append((pos, FailureRow( @@ -304,12 +297,11 @@ class FederationRemoteSendQueue(object): ))) # Fetch changed device messages - keys = self.device_messages.keys() - i = keys.bisect_right(from_token) - j = keys.bisect_right(to_token) + 1 - device_messages = {self.device_messages[k]: k for k in keys[i:j]} + i = self.device_messages.bisect_right(from_token) + j = self.device_messages.bisect_right(to_token) + 1 + device_messages = {v: k for k, v in self.device_messages.items()[i:j]} - for (destination, pos) in device_messages.iteritems(): + for (destination, pos) in iteritems(device_messages): rows.append((pos, DeviceRow( destination=destination, ))) @@ -528,19 +520,19 @@ def process_rows_for_federation(transaction_queue, rows): if buff.presence: transaction_queue.send_presence(buff.presence) - for destination, edu_map in buff.keyed_edus.iteritems(): + for destination, edu_map in iteritems(buff.keyed_edus): for key, edu in edu_map.items(): transaction_queue.send_edu( edu.destination, edu.edu_type, edu.content, key=key, ) - for destination, edu_list in buff.edus.iteritems(): + for destination, edu_list in iteritems(buff.edus): for edu in edu_list: transaction_queue.send_edu( edu.destination, edu.edu_type, edu.content, key=None, ) - for destination, failure_list in buff.failures.iteritems(): + for destination, failure_list in iteritems(buff.failures): for failure in failure_list: transaction_queue.send_failure(destination, failure) diff --git a/synapse/federation/transaction_queue.py b/synapse/federation/transaction_queue.py index 003eaba89..6996d6b69 100644 --- a/synapse/federation/transaction_queue.py +++ b/synapse/federation/transaction_queue.py @@ -13,34 +13,39 @@ # See the License for the specific language governing permissions and # limitations under the License. import datetime +import logging + +from six import itervalues + +from prometheus_client import Counter from twisted.internet import defer -from .persistence import TransactionActions -from .units import Transaction, Edu - -from synapse.api.errors import HttpResponseException -from synapse.util.async import run_on_reactor -from synapse.util.logcontext import preserve_context_over_fn, preserve_fn -from synapse.util.retryutils import NotRetryingDestination, get_retry_limiter -from synapse.util.metrics import measure_func -from synapse.handlers.presence import format_user_presence_state, get_interested_remotes import synapse.metrics +from synapse.api.errors import FederationDeniedError, HttpResponseException +from synapse.handlers.presence import format_user_presence_state, get_interested_remotes +from synapse.metrics import ( + LaterGauge, + events_processed_counter, + sent_edus_counter, + sent_transactions_counter, +) +from synapse.metrics.background_process_metrics import run_as_background_process +from synapse.util import logcontext +from synapse.util.metrics import measure_func +from synapse.util.retryutils import NotRetryingDestination, get_retry_limiter -import logging - +from .persistence import TransactionActions +from .units import Edu, Transaction logger = logging.getLogger(__name__) -metrics = synapse.metrics.get_metrics_for(__name__) - -client_metrics = synapse.metrics.get_metrics_for("synapse.federation.client") -sent_pdus_destination_dist = client_metrics.register_distribution( - "sent_pdu_destinations" +sent_pdus_destination_dist_count = Counter( + "synapse_federation_client_sent_pdu_destinations:count", "" +) +sent_pdus_destination_dist_total = Counter( + "synapse_federation_client_sent_pdu_destinations:total", "" ) -sent_edus_counter = client_metrics.register_counter("sent_edus") - -sent_transactions_counter = client_metrics.register_counter("sent_transactions") class TransactionQueue(object): @@ -67,8 +72,10 @@ class TransactionQueue(object): # done self.pending_transactions = {} - metrics.register_callback( - "pending_destinations", + LaterGauge( + "synapse_federation_transaction_queue_pending_destinations", + "", + [], lambda: len(self.pending_transactions), ) @@ -92,12 +99,16 @@ class TransactionQueue(object): # Map of destination -> (edu_type, key) -> Edu self.pending_edus_keyed_by_dest = edus_keyed = {} - metrics.register_callback( - "pending_pdus", + LaterGauge( + "synapse_federation_transaction_queue_pending_pdus", + "", + [], lambda: sum(map(len, pdus.values())), ) - metrics.register_callback( - "pending_edus", + LaterGauge( + "synapse_federation_transaction_queue_pending_edus", + "", + [], lambda: ( sum(map(len, edus.values())) + sum(map(len, presence.values())) @@ -146,7 +157,6 @@ class TransactionQueue(object): else: return not destination.startswith("localhost") - @defer.inlineCallbacks def notify_new_events(self, current_id): """This gets called when we have some new events we might want to send out to other servers. @@ -156,12 +166,20 @@ class TransactionQueue(object): if self._is_processing: return + # fire off a processing loop in the background + run_as_background_process( + "process_event_queue_for_federation", + self._process_event_queue_loop, + ) + + @defer.inlineCallbacks + def _process_event_queue_loop(self): try: self._is_processing = True while True: last_token = yield self.store.get_federation_out_pos("events") next_token, events = yield self.store.get_all_new_events_stream( - last_token, self._last_poked_id, limit=20, + last_token, self._last_poked_id, limit=100, ) logger.debug("Handling %s -> %s", last_token, next_token) @@ -169,24 +187,33 @@ class TransactionQueue(object): if not events and next_token >= self._last_poked_id: break - for event in events: + @defer.inlineCallbacks + def handle_event(event): # Only send events for this server. send_on_behalf_of = event.internal_metadata.get_send_on_behalf_of() is_mine = self.is_mine_id(event.event_id) if not is_mine and send_on_behalf_of is None: - continue + return + + try: + # Get the state from before the event. + # We need to make sure that this is the state from before + # the event and not from after it. + # Otherwise if the last member on a server in a room is + # banned then it won't receive the event because it won't + # be in the room after the ban. + destinations = yield self.state.get_current_hosts_in_room( + event.room_id, latest_event_ids=[ + prev_id for prev_id, _ in event.prev_events + ], + ) + except Exception: + logger.exception( + "Failed to calculate hosts in room for event: %s", + event.event_id, + ) + return - # Get the state from before the event. - # We need to make sure that this is the state from before - # the event and not from after it. - # Otherwise if the last member on a server in a room is - # banned then it won't receive the event because it won't - # be in the room after the ban. - destinations = yield self.state.get_current_hosts_in_room( - event.room_id, latest_event_ids=[ - prev_id for prev_id, _ in event.prev_events - ], - ) destinations = set(destinations) if send_on_behalf_of is not None: @@ -199,10 +226,41 @@ class TransactionQueue(object): self._send_pdu(event, destinations) + @defer.inlineCallbacks + def handle_room_events(events): + for event in events: + yield handle_event(event) + + events_by_room = {} + for event in events: + events_by_room.setdefault(event.room_id, []).append(event) + + yield logcontext.make_deferred_yieldable(defer.gatherResults( + [ + logcontext.run_in_background(handle_room_events, evs) + for evs in itervalues(events_by_room) + ], + consumeErrors=True + )) + yield self.store.update_federation_out_pos( "events", next_token ) + if events: + now = self.clock.time_msec() + ts = yield self.store.get_received_ts(events[-1].event_id) + + synapse.metrics.event_processing_lag.labels( + "federation_sender").set(now - ts) + synapse.metrics.event_processing_last_ts.labels( + "federation_sender").set(ts) + + events_processed_counter.inc(len(events)) + + synapse.metrics.event_processing_positions.labels( + "federation_sender").set(next_token) + finally: self._is_processing = False @@ -224,18 +282,17 @@ class TransactionQueue(object): if not destinations: return - sent_pdus_destination_dist.inc_by(len(destinations)) + sent_pdus_destination_dist_total.inc(len(destinations)) + sent_pdus_destination_dist_count.inc() for destination in destinations: self.pending_pdus_by_dest.setdefault(destination, []).append( (pdu, order) ) - preserve_context_over_fn( - self._attempt_new_transaction, destination - ) + self._attempt_new_transaction(destination) - @preserve_fn # the caller should not yield on this + @logcontext.preserve_fn # the caller should not yield on this @defer.inlineCallbacks def send_presence(self, states): """Send the new presence states to the appropriate destinations. @@ -273,7 +330,9 @@ class TransactionQueue(object): if not states_map: break - yield self._process_presence_inner(states_map.values()) + yield self._process_presence_inner(list(states_map.values())) + except Exception: + logger.exception("Error sending presence states to servers") finally: self._processing_pending_presence = False @@ -299,7 +358,7 @@ class TransactionQueue(object): state.user_id: state for state in states }) - preserve_fn(self._attempt_new_transaction)(destination) + self._attempt_new_transaction(destination) def send_edu(self, destination, edu_type, content, key=None): edu = Edu( @@ -321,9 +380,7 @@ class TransactionQueue(object): else: self.pending_edus_by_dest.setdefault(destination, []).append(edu) - preserve_context_over_fn( - self._attempt_new_transaction, destination - ) + self._attempt_new_transaction(destination) def send_failure(self, failure, destination): if destination == self.server_name or destination == "localhost": @@ -336,9 +393,7 @@ class TransactionQueue(object): destination, [] ).append(failure) - preserve_context_over_fn( - self._attempt_new_transaction, destination - ) + self._attempt_new_transaction(destination) def send_device_messages(self, destination): if destination == self.server_name or destination == "localhost": @@ -347,15 +402,24 @@ class TransactionQueue(object): if not self.can_send_to(destination): return - preserve_context_over_fn( - self._attempt_new_transaction, destination - ) + self._attempt_new_transaction(destination) def get_current_token(self): return 0 - @defer.inlineCallbacks def _attempt_new_transaction(self, destination): + """Try to start a new transaction to this destination + + If there is already a transaction in progress to this destination, + returns immediately. Otherwise kicks off the process of sending a + transaction in the background. + + Args: + destination (str): + + Returns: + None + """ # list of (pending_pdu, deferred, order) if destination in self.pending_transactions: # XXX: pending_transactions can get stuck on by a never-ending @@ -368,6 +432,16 @@ class TransactionQueue(object): ) return + logger.debug("TX [%s] Starting transaction loop", destination) + + run_as_background_process( + "federation_transaction_transmission_loop", + self._transaction_transmission_loop, + destination, + ) + + @defer.inlineCallbacks + def _transaction_transmission_loop(self, destination): pending_pdus = [] try: self.pending_transactions[destination] = 1 @@ -377,9 +451,6 @@ class TransactionQueue(object): # hence why we throw the result away. yield get_retry_limiter(destination, self.clock, self.store) - # XXX: what's this for? - yield run_on_reactor() - pending_pdus = [] while True: device_message_edus, device_stream_id, dev_list_id = ( @@ -464,6 +535,8 @@ class TransactionQueue(object): (e.retry_last_ts + e.retry_interval) / 1000.0 ), ) + except FederationDeniedError as e: + logger.info(e) except Exception as e: logger.warn( "TX [%s] Failed to send transaction: %s", diff --git a/synapse/federation/transport/client.py b/synapse/federation/transport/client.py index 52b2a717d..4529d454a 100644 --- a/synapse/federation/transport/client.py +++ b/synapse/federation/transport/client.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2014-2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -13,15 +14,15 @@ # See the License for the specific language governing permissions and # limitations under the License. -from twisted.internet import defer -from synapse.api.constants import Membership +import logging +import urllib +from twisted.internet import defer + +from synapse.api.constants import Membership from synapse.api.urls import FEDERATION_PREFIX as PREFIX from synapse.util.logutils import log_function -import logging - - logger = logging.getLogger(__name__) @@ -49,7 +50,7 @@ class TransportLayerClient(object): logger.debug("get_room_state dest=%s, room=%s", destination, room_id) - path = PREFIX + "/state/%s/" % room_id + path = _create_path(PREFIX, "/state/%s/", room_id) return self.client.get_json( destination, path=path, args={"event_id": event_id}, ) @@ -71,7 +72,7 @@ class TransportLayerClient(object): logger.debug("get_room_state_ids dest=%s, room=%s", destination, room_id) - path = PREFIX + "/state_ids/%s/" % room_id + path = _create_path(PREFIX, "/state_ids/%s/", room_id) return self.client.get_json( destination, path=path, args={"event_id": event_id}, ) @@ -93,7 +94,7 @@ class TransportLayerClient(object): logger.debug("get_pdu dest=%s, event_id=%s", destination, event_id) - path = PREFIX + "/event/%s/" % (event_id, ) + path = _create_path(PREFIX, "/event/%s/", event_id) return self.client.get_json(destination, path=path, timeout=timeout) @log_function @@ -119,7 +120,7 @@ class TransportLayerClient(object): # TODO: raise? return - path = PREFIX + "/backfill/%s/" % (room_id,) + path = _create_path(PREFIX, "/backfill/%s/", room_id) args = { "v": event_tuples, @@ -157,9 +158,11 @@ class TransportLayerClient(object): # generated by the json_data_callback. json_data = transaction.get_dict() + path = _create_path(PREFIX, "/send/%s/", transaction.transaction_id) + response = yield self.client.put_json( transaction.destination, - path=PREFIX + "/send/%s/" % transaction.transaction_id, + path=path, data=json_data, json_data_callback=json_data_callback, long_retries=True, @@ -177,7 +180,7 @@ class TransportLayerClient(object): @log_function def make_query(self, destination, query_type, args, retry_on_dns_fail, ignore_backoff=False): - path = PREFIX + "/query/%s" % query_type + path = _create_path(PREFIX, "/query/%s", query_type) content = yield self.client.get_json( destination=destination, @@ -212,6 +215,9 @@ class TransportLayerClient(object): Fails with ``NotRetryingDestination`` if we are not yet ready to retry this server. + + Fails with ``FederationDeniedError`` if the remote destination + is not in our federation whitelist """ valid_memberships = {Membership.JOIN, Membership.LEAVE} if membership not in valid_memberships: @@ -219,7 +225,7 @@ class TransportLayerClient(object): "make_membership_event called with membership='%s', must be one of %s" % (membership, ",".join(valid_memberships)) ) - path = PREFIX + "/make_%s/%s/%s" % (membership, room_id, user_id) + path = _create_path(PREFIX, "/make_%s/%s/%s", membership, room_id, user_id) ignore_backoff = False retry_on_dns_fail = False @@ -245,7 +251,7 @@ class TransportLayerClient(object): @defer.inlineCallbacks @log_function def send_join(self, destination, room_id, event_id, content): - path = PREFIX + "/send_join/%s/%s" % (room_id, event_id) + path = _create_path(PREFIX, "/send_join/%s/%s", room_id, event_id) response = yield self.client.put_json( destination=destination, @@ -258,7 +264,7 @@ class TransportLayerClient(object): @defer.inlineCallbacks @log_function def send_leave(self, destination, room_id, event_id, content): - path = PREFIX + "/send_leave/%s/%s" % (room_id, event_id) + path = _create_path(PREFIX, "/send_leave/%s/%s", room_id, event_id) response = yield self.client.put_json( destination=destination, @@ -277,7 +283,7 @@ class TransportLayerClient(object): @defer.inlineCallbacks @log_function def send_invite(self, destination, room_id, event_id, content): - path = PREFIX + "/invite/%s/%s" % (room_id, event_id) + path = _create_path(PREFIX, "/invite/%s/%s", room_id, event_id) response = yield self.client.put_json( destination=destination, @@ -319,7 +325,7 @@ class TransportLayerClient(object): @defer.inlineCallbacks @log_function def exchange_third_party_invite(self, destination, room_id, event_dict): - path = PREFIX + "/exchange_third_party_invite/%s" % (room_id,) + path = _create_path(PREFIX, "/exchange_third_party_invite/%s", room_id,) response = yield self.client.put_json( destination=destination, @@ -332,7 +338,7 @@ class TransportLayerClient(object): @defer.inlineCallbacks @log_function def get_event_auth(self, destination, room_id, event_id): - path = PREFIX + "/event_auth/%s/%s" % (room_id, event_id) + path = _create_path(PREFIX, "/event_auth/%s/%s", room_id, event_id) content = yield self.client.get_json( destination=destination, @@ -344,7 +350,7 @@ class TransportLayerClient(object): @defer.inlineCallbacks @log_function def send_query_auth(self, destination, room_id, event_id, content): - path = PREFIX + "/query_auth/%s/%s" % (room_id, event_id) + path = _create_path(PREFIX, "/query_auth/%s/%s", room_id, event_id) content = yield self.client.post_json( destination=destination, @@ -406,7 +412,7 @@ class TransportLayerClient(object): Returns: A dict containg the device keys. """ - path = PREFIX + "/user/devices/" + user_id + path = _create_path(PREFIX, "/user/devices/%s", user_id) content = yield self.client.get_json( destination=destination, @@ -456,7 +462,7 @@ class TransportLayerClient(object): @log_function def get_missing_events(self, destination, room_id, earliest_events, latest_events, limit, min_depth, timeout): - path = PREFIX + "/get_missing_events/%s" % (room_id,) + path = _create_path(PREFIX, "/get_missing_events/%s", room_id,) content = yield self.client.post_json( destination=destination, @@ -471,3 +477,475 @@ class TransportLayerClient(object): ) defer.returnValue(content) + + @log_function + def get_group_profile(self, destination, group_id, requester_user_id): + """Get a group profile + """ + path = _create_path(PREFIX, "/groups/%s/profile", group_id,) + + return self.client.get_json( + destination=destination, + path=path, + args={"requester_user_id": requester_user_id}, + ignore_backoff=True, + ) + + @log_function + def update_group_profile(self, destination, group_id, requester_user_id, content): + """Update a remote group profile + + Args: + destination (str) + group_id (str) + requester_user_id (str) + content (dict): The new profile of the group + """ + path = _create_path(PREFIX, "/groups/%s/profile", group_id,) + + return self.client.post_json( + destination=destination, + path=path, + args={"requester_user_id": requester_user_id}, + data=content, + ignore_backoff=True, + ) + + @log_function + def get_group_summary(self, destination, group_id, requester_user_id): + """Get a group summary + """ + path = _create_path(PREFIX, "/groups/%s/summary", group_id,) + + return self.client.get_json( + destination=destination, + path=path, + args={"requester_user_id": requester_user_id}, + ignore_backoff=True, + ) + + @log_function + def get_rooms_in_group(self, destination, group_id, requester_user_id): + """Get all rooms in a group + """ + path = _create_path(PREFIX, "/groups/%s/rooms", group_id,) + + return self.client.get_json( + destination=destination, + path=path, + args={"requester_user_id": requester_user_id}, + ignore_backoff=True, + ) + + def add_room_to_group(self, destination, group_id, requester_user_id, room_id, + content): + """Add a room to a group + """ + path = _create_path(PREFIX, "/groups/%s/room/%s", group_id, room_id,) + + return self.client.post_json( + destination=destination, + path=path, + args={"requester_user_id": requester_user_id}, + data=content, + ignore_backoff=True, + ) + + def update_room_in_group(self, destination, group_id, requester_user_id, room_id, + config_key, content): + """Update room in group + """ + path = _create_path( + PREFIX, "/groups/%s/room/%s/config/%s", + group_id, room_id, config_key, + ) + + return self.client.post_json( + destination=destination, + path=path, + args={"requester_user_id": requester_user_id}, + data=content, + ignore_backoff=True, + ) + + def remove_room_from_group(self, destination, group_id, requester_user_id, room_id): + """Remove a room from a group + """ + path = _create_path(PREFIX, "/groups/%s/room/%s", group_id, room_id,) + + return self.client.delete_json( + destination=destination, + path=path, + args={"requester_user_id": requester_user_id}, + ignore_backoff=True, + ) + + @log_function + def get_users_in_group(self, destination, group_id, requester_user_id): + """Get users in a group + """ + path = _create_path(PREFIX, "/groups/%s/users", group_id,) + + return self.client.get_json( + destination=destination, + path=path, + args={"requester_user_id": requester_user_id}, + ignore_backoff=True, + ) + + @log_function + def get_invited_users_in_group(self, destination, group_id, requester_user_id): + """Get users that have been invited to a group + """ + path = _create_path(PREFIX, "/groups/%s/invited_users", group_id,) + + return self.client.get_json( + destination=destination, + path=path, + args={"requester_user_id": requester_user_id}, + ignore_backoff=True, + ) + + @log_function + def accept_group_invite(self, destination, group_id, user_id, content): + """Accept a group invite + """ + path = _create_path( + PREFIX, "/groups/%s/users/%s/accept_invite", + group_id, user_id, + ) + + return self.client.post_json( + destination=destination, + path=path, + data=content, + ignore_backoff=True, + ) + + @log_function + def join_group(self, destination, group_id, user_id, content): + """Attempts to join a group + """ + path = _create_path(PREFIX, "/groups/%s/users/%s/join", group_id, user_id) + + return self.client.post_json( + destination=destination, + path=path, + data=content, + ignore_backoff=True, + ) + + @log_function + def invite_to_group(self, destination, group_id, user_id, requester_user_id, content): + """Invite a user to a group + """ + path = _create_path(PREFIX, "/groups/%s/users/%s/invite", group_id, user_id) + + return self.client.post_json( + destination=destination, + path=path, + args={"requester_user_id": requester_user_id}, + data=content, + ignore_backoff=True, + ) + + @log_function + def invite_to_group_notification(self, destination, group_id, user_id, content): + """Sent by group server to inform a user's server that they have been + invited. + """ + + path = _create_path(PREFIX, "/groups/local/%s/users/%s/invite", group_id, user_id) + + return self.client.post_json( + destination=destination, + path=path, + data=content, + ignore_backoff=True, + ) + + @log_function + def remove_user_from_group(self, destination, group_id, requester_user_id, + user_id, content): + """Remove a user fron a group + """ + path = _create_path(PREFIX, "/groups/%s/users/%s/remove", group_id, user_id) + + return self.client.post_json( + destination=destination, + path=path, + args={"requester_user_id": requester_user_id}, + data=content, + ignore_backoff=True, + ) + + @log_function + def remove_user_from_group_notification(self, destination, group_id, user_id, + content): + """Sent by group server to inform a user's server that they have been + kicked from the group. + """ + + path = _create_path(PREFIX, "/groups/local/%s/users/%s/remove", group_id, user_id) + + return self.client.post_json( + destination=destination, + path=path, + data=content, + ignore_backoff=True, + ) + + @log_function + def renew_group_attestation(self, destination, group_id, user_id, content): + """Sent by either a group server or a user's server to periodically update + the attestations + """ + + path = _create_path(PREFIX, "/groups/%s/renew_attestation/%s", group_id, user_id) + + return self.client.post_json( + destination=destination, + path=path, + data=content, + ignore_backoff=True, + ) + + @log_function + def update_group_summary_room(self, destination, group_id, user_id, room_id, + category_id, content): + """Update a room entry in a group summary + """ + if category_id: + path = _create_path( + PREFIX, "/groups/%s/summary/categories/%s/rooms/%s", + group_id, category_id, room_id, + ) + else: + path = _create_path(PREFIX, "/groups/%s/summary/rooms/%s", group_id, room_id,) + + return self.client.post_json( + destination=destination, + path=path, + args={"requester_user_id": user_id}, + data=content, + ignore_backoff=True, + ) + + @log_function + def delete_group_summary_room(self, destination, group_id, user_id, room_id, + category_id): + """Delete a room entry in a group summary + """ + if category_id: + path = _create_path( + PREFIX + "/groups/%s/summary/categories/%s/rooms/%s", + group_id, category_id, room_id, + ) + else: + path = _create_path(PREFIX, "/groups/%s/summary/rooms/%s", group_id, room_id,) + + return self.client.delete_json( + destination=destination, + path=path, + args={"requester_user_id": user_id}, + ignore_backoff=True, + ) + + @log_function + def get_group_categories(self, destination, group_id, requester_user_id): + """Get all categories in a group + """ + path = _create_path(PREFIX, "/groups/%s/categories", group_id,) + + return self.client.get_json( + destination=destination, + path=path, + args={"requester_user_id": requester_user_id}, + ignore_backoff=True, + ) + + @log_function + def get_group_category(self, destination, group_id, requester_user_id, category_id): + """Get category info in a group + """ + path = _create_path(PREFIX, "/groups/%s/categories/%s", group_id, category_id,) + + return self.client.get_json( + destination=destination, + path=path, + args={"requester_user_id": requester_user_id}, + ignore_backoff=True, + ) + + @log_function + def update_group_category(self, destination, group_id, requester_user_id, category_id, + content): + """Update a category in a group + """ + path = _create_path(PREFIX, "/groups/%s/categories/%s", group_id, category_id,) + + return self.client.post_json( + destination=destination, + path=path, + args={"requester_user_id": requester_user_id}, + data=content, + ignore_backoff=True, + ) + + @log_function + def delete_group_category(self, destination, group_id, requester_user_id, + category_id): + """Delete a category in a group + """ + path = _create_path(PREFIX, "/groups/%s/categories/%s", group_id, category_id,) + + return self.client.delete_json( + destination=destination, + path=path, + args={"requester_user_id": requester_user_id}, + ignore_backoff=True, + ) + + @log_function + def get_group_roles(self, destination, group_id, requester_user_id): + """Get all roles in a group + """ + path = _create_path(PREFIX, "/groups/%s/roles", group_id,) + + return self.client.get_json( + destination=destination, + path=path, + args={"requester_user_id": requester_user_id}, + ignore_backoff=True, + ) + + @log_function + def get_group_role(self, destination, group_id, requester_user_id, role_id): + """Get a roles info + """ + path = _create_path(PREFIX, "/groups/%s/roles/%s", group_id, role_id,) + + return self.client.get_json( + destination=destination, + path=path, + args={"requester_user_id": requester_user_id}, + ignore_backoff=True, + ) + + @log_function + def update_group_role(self, destination, group_id, requester_user_id, role_id, + content): + """Update a role in a group + """ + path = _create_path(PREFIX, "/groups/%s/roles/%s", group_id, role_id,) + + return self.client.post_json( + destination=destination, + path=path, + args={"requester_user_id": requester_user_id}, + data=content, + ignore_backoff=True, + ) + + @log_function + def delete_group_role(self, destination, group_id, requester_user_id, role_id): + """Delete a role in a group + """ + path = _create_path(PREFIX, "/groups/%s/roles/%s", group_id, role_id,) + + return self.client.delete_json( + destination=destination, + path=path, + args={"requester_user_id": requester_user_id}, + ignore_backoff=True, + ) + + @log_function + def update_group_summary_user(self, destination, group_id, requester_user_id, + user_id, role_id, content): + """Update a users entry in a group + """ + if role_id: + path = _create_path( + PREFIX, "/groups/%s/summary/roles/%s/users/%s", + group_id, role_id, user_id, + ) + else: + path = _create_path(PREFIX, "/groups/%s/summary/users/%s", group_id, user_id,) + + return self.client.post_json( + destination=destination, + path=path, + args={"requester_user_id": requester_user_id}, + data=content, + ignore_backoff=True, + ) + + @log_function + def set_group_join_policy(self, destination, group_id, requester_user_id, + content): + """Sets the join policy for a group + """ + path = _create_path(PREFIX, "/groups/%s/settings/m.join_policy", group_id,) + + return self.client.put_json( + destination=destination, + path=path, + args={"requester_user_id": requester_user_id}, + data=content, + ignore_backoff=True, + ) + + @log_function + def delete_group_summary_user(self, destination, group_id, requester_user_id, + user_id, role_id): + """Delete a users entry in a group + """ + if role_id: + path = _create_path( + PREFIX, "/groups/%s/summary/roles/%s/users/%s", + group_id, role_id, user_id, + ) + else: + path = _create_path(PREFIX, "/groups/%s/summary/users/%s", group_id, user_id,) + + return self.client.delete_json( + destination=destination, + path=path, + args={"requester_user_id": requester_user_id}, + ignore_backoff=True, + ) + + def bulk_get_publicised_groups(self, destination, user_ids): + """Get the groups a list of users are publicising + """ + + path = PREFIX + "/get_groups_publicised" + + content = {"user_ids": user_ids} + + return self.client.post_json( + destination=destination, + path=path, + data=content, + ignore_backoff=True, + ) + + +def _create_path(prefix, path, *args): + """Creates a path from the prefix, path template and args. Ensures that + all args are url encoded. + + Example: + + _create_path(PREFIX, "/event/%s/", event_id) + + Args: + prefix (str) + path (str): String template for the path + args: ([str]): Args to insert into path. Each arg will be url encoded + + Returns: + str + """ + return prefix + path % tuple(urllib.quote(arg, "") for arg in args) diff --git a/synapse/federation/transport/server.py b/synapse/federation/transport/server.py index a78f01e44..c9beca27c 100644 --- a/synapse/federation/transport/server.py +++ b/synapse/federation/transport/server.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2014-2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -13,25 +14,27 @@ # See the License for the specific language governing permissions and # limitations under the License. -from twisted.internet import defer - -from synapse.api.urls import FEDERATION_PREFIX as PREFIX -from synapse.api.errors import Codes, SynapseError -from synapse.http.server import JsonResource -from synapse.http.servlet import ( - parse_json_object_from_request, parse_integer_from_args, parse_string_from_args, - parse_boolean_from_args, -) -from synapse.util.ratelimitutils import FederationRateLimiter -from synapse.util.versionstring import get_version_string -from synapse.util.logcontext import preserve_fn -from synapse.types import ThirdPartyInstanceID - import functools import logging import re -import synapse +from twisted.internet import defer + +import synapse +from synapse.api.errors import Codes, FederationDeniedError, SynapseError +from synapse.api.urls import FEDERATION_PREFIX as PREFIX +from synapse.http.endpoint import parse_and_validate_server_name +from synapse.http.server import JsonResource +from synapse.http.servlet import ( + parse_boolean_from_args, + parse_integer_from_args, + parse_json_object_from_request, + parse_string_from_args, +) +from synapse.types import ThirdPartyInstanceID, get_domain_from_id +from synapse.util.logcontext import run_in_background +from synapse.util.ratelimitutils import FederationRateLimiter +from synapse.util.versionstring import get_version_string logger = logging.getLogger(__name__) @@ -81,6 +84,7 @@ class Authenticator(object): self.keyring = hs.get_keyring() self.server_name = hs.hostname self.store = hs.get_datastore() + self.federation_domain_whitelist = hs.config.federation_domain_whitelist # A method just so we can pass 'self' as the authenticator to the Servlets @defer.inlineCallbacks @@ -97,26 +101,6 @@ class Authenticator(object): origin = None - def parse_auth_header(header_str): - try: - params = auth.split(" ")[1].split(",") - param_dict = dict(kv.split("=") for kv in params) - - def strip_quotes(value): - if value.startswith("\""): - return value[1:-1] - else: - return value - - origin = strip_quotes(param_dict["origin"]) - key = strip_quotes(param_dict["key"]) - sig = strip_quotes(param_dict["sig"]) - return (origin, key, sig) - except: - raise AuthenticationError( - 400, "Malformed Authorization header", Codes.UNAUTHORIZED - ) - auth_headers = request.requestHeaders.getRawHeaders(b"Authorization") if not auth_headers: @@ -125,11 +109,17 @@ class Authenticator(object): ) for auth in auth_headers: - if auth.startswith("X-Matrix"): - (origin, key, sig) = parse_auth_header(auth) + if auth.startswith(b"X-Matrix"): + (origin, key, sig) = _parse_auth_header(auth) json_request["origin"] = origin json_request["signatures"].setdefault(origin, {})[key] = sig + if ( + self.federation_domain_whitelist is not None and + origin not in self.federation_domain_whitelist + ): + raise FederationDeniedError(origin) + if not json_request["signatures"]: raise NoAuthenticationError( 401, "Missing Authorization headers", Codes.UNAUTHORIZED, @@ -144,11 +134,60 @@ class Authenticator(object): # alive retry_timings = yield self.store.get_destination_retry_timings(origin) if retry_timings and retry_timings["retry_last_ts"]: - logger.info("Marking origin %r as up", origin) - preserve_fn(self.store.set_destination_retry_timings)(origin, 0, 0) + run_in_background(self._reset_retry_timings, origin) defer.returnValue(origin) + @defer.inlineCallbacks + def _reset_retry_timings(self, origin): + try: + logger.info("Marking origin %r as up", origin) + yield self.store.set_destination_retry_timings(origin, 0, 0) + except Exception: + logger.exception("Error resetting retry timings on %s", origin) + + +def _parse_auth_header(header_bytes): + """Parse an X-Matrix auth header + + Args: + header_bytes (bytes): header value + + Returns: + Tuple[str, str, str]: origin, key id, signature. + + Raises: + AuthenticationError if the header could not be parsed + """ + try: + header_str = header_bytes.decode('utf-8') + params = header_str.split(" ")[1].split(",") + param_dict = dict(kv.split("=") for kv in params) + + def strip_quotes(value): + if value.startswith(b"\""): + return value[1:-1] + else: + return value + + origin = strip_quotes(param_dict["origin"]) + + # ensure that the origin is a valid server name + parse_and_validate_server_name(origin) + + key = strip_quotes(param_dict["key"]) + sig = strip_quotes(param_dict["sig"]) + return origin, key, sig + except Exception as e: + logger.warn( + "Error parsing auth header '%s': %s", + header_bytes.decode('ascii', 'replace'), + e, + ) + raise AuthenticationError( + 400, "Malformed Authorization header", Codes.UNAUTHORIZED, + ) + class BaseFederationServlet(object): REQUIRE_AUTH = True @@ -177,7 +216,7 @@ class BaseFederationServlet(object): if self.REQUIRE_AUTH: logger.exception("authenticate_request failed") raise - except: + except Exception: logger.exception("authenticate_request failed") raise @@ -270,7 +309,7 @@ class FederationSendServlet(BaseFederationServlet): code, response = yield self.handler.on_incoming_transaction( transaction_data ) - except: + except Exception: logger.exception("on_incoming_transaction failed") raise @@ -347,7 +386,9 @@ class FederationMakeJoinServlet(BaseFederationServlet): @defer.inlineCallbacks def on_GET(self, origin, content, query, context, user_id): - content = yield self.handler.on_make_join_request(context, user_id) + content = yield self.handler.on_make_join_request( + origin, context, user_id, + ) defer.returnValue((200, content)) @@ -356,7 +397,9 @@ class FederationMakeLeaveServlet(BaseFederationServlet): @defer.inlineCallbacks def on_GET(self, origin, content, query, context, user_id): - content = yield self.handler.on_make_leave_request(context, user_id) + content = yield self.handler.on_make_leave_request( + origin, context, user_id, + ) defer.returnValue((200, content)) @@ -609,6 +652,549 @@ class FederationVersionServlet(BaseFederationServlet): })) +class FederationGroupsProfileServlet(BaseFederationServlet): + """Get/set the basic profile of a group on behalf of a user + """ + PATH = "/groups/(?P[^/]*)/profile$" + + @defer.inlineCallbacks + def on_GET(self, origin, content, query, group_id): + requester_user_id = parse_string_from_args(query, "requester_user_id") + if get_domain_from_id(requester_user_id) != origin: + raise SynapseError(403, "requester_user_id doesn't match origin") + + new_content = yield self.handler.get_group_profile( + group_id, requester_user_id + ) + + defer.returnValue((200, new_content)) + + @defer.inlineCallbacks + def on_POST(self, origin, content, query, group_id): + requester_user_id = parse_string_from_args(query, "requester_user_id") + if get_domain_from_id(requester_user_id) != origin: + raise SynapseError(403, "requester_user_id doesn't match origin") + + new_content = yield self.handler.update_group_profile( + group_id, requester_user_id, content + ) + + defer.returnValue((200, new_content)) + + +class FederationGroupsSummaryServlet(BaseFederationServlet): + PATH = "/groups/(?P[^/]*)/summary$" + + @defer.inlineCallbacks + def on_GET(self, origin, content, query, group_id): + requester_user_id = parse_string_from_args(query, "requester_user_id") + if get_domain_from_id(requester_user_id) != origin: + raise SynapseError(403, "requester_user_id doesn't match origin") + + new_content = yield self.handler.get_group_summary( + group_id, requester_user_id + ) + + defer.returnValue((200, new_content)) + + +class FederationGroupsRoomsServlet(BaseFederationServlet): + """Get the rooms in a group on behalf of a user + """ + PATH = "/groups/(?P[^/]*)/rooms$" + + @defer.inlineCallbacks + def on_GET(self, origin, content, query, group_id): + requester_user_id = parse_string_from_args(query, "requester_user_id") + if get_domain_from_id(requester_user_id) != origin: + raise SynapseError(403, "requester_user_id doesn't match origin") + + new_content = yield self.handler.get_rooms_in_group( + group_id, requester_user_id + ) + + defer.returnValue((200, new_content)) + + +class FederationGroupsAddRoomsServlet(BaseFederationServlet): + """Add/remove room from group + """ + PATH = "/groups/(?P[^/]*)/room/(?P[^/]*)$" + + @defer.inlineCallbacks + def on_POST(self, origin, content, query, group_id, room_id): + requester_user_id = parse_string_from_args(query, "requester_user_id") + if get_domain_from_id(requester_user_id) != origin: + raise SynapseError(403, "requester_user_id doesn't match origin") + + new_content = yield self.handler.add_room_to_group( + group_id, requester_user_id, room_id, content + ) + + defer.returnValue((200, new_content)) + + @defer.inlineCallbacks + def on_DELETE(self, origin, content, query, group_id, room_id): + requester_user_id = parse_string_from_args(query, "requester_user_id") + if get_domain_from_id(requester_user_id) != origin: + raise SynapseError(403, "requester_user_id doesn't match origin") + + new_content = yield self.handler.remove_room_from_group( + group_id, requester_user_id, room_id, + ) + + defer.returnValue((200, new_content)) + + +class FederationGroupsAddRoomsConfigServlet(BaseFederationServlet): + """Update room config in group + """ + PATH = ( + "/groups/(?P[^/]*)/room/(?P[^/]*)" + "/config/(?P[^/]*)$" + ) + + @defer.inlineCallbacks + def on_POST(self, origin, content, query, group_id, room_id, config_key): + requester_user_id = parse_string_from_args(query, "requester_user_id") + if get_domain_from_id(requester_user_id) != origin: + raise SynapseError(403, "requester_user_id doesn't match origin") + + result = yield self.groups_handler.update_room_in_group( + group_id, requester_user_id, room_id, config_key, content, + ) + + defer.returnValue((200, result)) + + +class FederationGroupsUsersServlet(BaseFederationServlet): + """Get the users in a group on behalf of a user + """ + PATH = "/groups/(?P[^/]*)/users$" + + @defer.inlineCallbacks + def on_GET(self, origin, content, query, group_id): + requester_user_id = parse_string_from_args(query, "requester_user_id") + if get_domain_from_id(requester_user_id) != origin: + raise SynapseError(403, "requester_user_id doesn't match origin") + + new_content = yield self.handler.get_users_in_group( + group_id, requester_user_id + ) + + defer.returnValue((200, new_content)) + + +class FederationGroupsInvitedUsersServlet(BaseFederationServlet): + """Get the users that have been invited to a group + """ + PATH = "/groups/(?P[^/]*)/invited_users$" + + @defer.inlineCallbacks + def on_GET(self, origin, content, query, group_id): + requester_user_id = parse_string_from_args(query, "requester_user_id") + if get_domain_from_id(requester_user_id) != origin: + raise SynapseError(403, "requester_user_id doesn't match origin") + + new_content = yield self.handler.get_invited_users_in_group( + group_id, requester_user_id + ) + + defer.returnValue((200, new_content)) + + +class FederationGroupsInviteServlet(BaseFederationServlet): + """Ask a group server to invite someone to the group + """ + PATH = "/groups/(?P[^/]*)/users/(?P[^/]*)/invite$" + + @defer.inlineCallbacks + def on_POST(self, origin, content, query, group_id, user_id): + requester_user_id = parse_string_from_args(query, "requester_user_id") + if get_domain_from_id(requester_user_id) != origin: + raise SynapseError(403, "requester_user_id doesn't match origin") + + new_content = yield self.handler.invite_to_group( + group_id, user_id, requester_user_id, content, + ) + + defer.returnValue((200, new_content)) + + +class FederationGroupsAcceptInviteServlet(BaseFederationServlet): + """Accept an invitation from the group server + """ + PATH = "/groups/(?P[^/]*)/users/(?P[^/]*)/accept_invite$" + + @defer.inlineCallbacks + def on_POST(self, origin, content, query, group_id, user_id): + if get_domain_from_id(user_id) != origin: + raise SynapseError(403, "user_id doesn't match origin") + + new_content = yield self.handler.accept_invite( + group_id, user_id, content, + ) + + defer.returnValue((200, new_content)) + + +class FederationGroupsJoinServlet(BaseFederationServlet): + """Attempt to join a group + """ + PATH = "/groups/(?P[^/]*)/users/(?P[^/]*)/join$" + + @defer.inlineCallbacks + def on_POST(self, origin, content, query, group_id, user_id): + if get_domain_from_id(user_id) != origin: + raise SynapseError(403, "user_id doesn't match origin") + + new_content = yield self.handler.join_group( + group_id, user_id, content, + ) + + defer.returnValue((200, new_content)) + + +class FederationGroupsRemoveUserServlet(BaseFederationServlet): + """Leave or kick a user from the group + """ + PATH = "/groups/(?P[^/]*)/users/(?P[^/]*)/remove$" + + @defer.inlineCallbacks + def on_POST(self, origin, content, query, group_id, user_id): + requester_user_id = parse_string_from_args(query, "requester_user_id") + if get_domain_from_id(requester_user_id) != origin: + raise SynapseError(403, "requester_user_id doesn't match origin") + + new_content = yield self.handler.remove_user_from_group( + group_id, user_id, requester_user_id, content, + ) + + defer.returnValue((200, new_content)) + + +class FederationGroupsLocalInviteServlet(BaseFederationServlet): + """A group server has invited a local user + """ + PATH = "/groups/local/(?P[^/]*)/users/(?P[^/]*)/invite$" + + @defer.inlineCallbacks + def on_POST(self, origin, content, query, group_id, user_id): + if get_domain_from_id(group_id) != origin: + raise SynapseError(403, "group_id doesn't match origin") + + new_content = yield self.handler.on_invite( + group_id, user_id, content, + ) + + defer.returnValue((200, new_content)) + + +class FederationGroupsRemoveLocalUserServlet(BaseFederationServlet): + """A group server has removed a local user + """ + PATH = "/groups/local/(?P[^/]*)/users/(?P[^/]*)/remove$" + + @defer.inlineCallbacks + def on_POST(self, origin, content, query, group_id, user_id): + if get_domain_from_id(group_id) != origin: + raise SynapseError(403, "user_id doesn't match origin") + + new_content = yield self.handler.user_removed_from_group( + group_id, user_id, content, + ) + + defer.returnValue((200, new_content)) + + +class FederationGroupsRenewAttestaionServlet(BaseFederationServlet): + """A group or user's server renews their attestation + """ + PATH = "/groups/(?P[^/]*)/renew_attestation/(?P[^/]*)$" + + @defer.inlineCallbacks + def on_POST(self, origin, content, query, group_id, user_id): + # We don't need to check auth here as we check the attestation signatures + + new_content = yield self.handler.on_renew_attestation( + group_id, user_id, content + ) + + defer.returnValue((200, new_content)) + + +class FederationGroupsSummaryRoomsServlet(BaseFederationServlet): + """Add/remove a room from the group summary, with optional category. + + Matches both: + - /groups/:group/summary/rooms/:room_id + - /groups/:group/summary/categories/:category/rooms/:room_id + """ + PATH = ( + "/groups/(?P[^/]*)/summary" + "(/categories/(?P[^/]+))?" + "/rooms/(?P[^/]*)$" + ) + + @defer.inlineCallbacks + def on_POST(self, origin, content, query, group_id, category_id, room_id): + requester_user_id = parse_string_from_args(query, "requester_user_id") + if get_domain_from_id(requester_user_id) != origin: + raise SynapseError(403, "requester_user_id doesn't match origin") + + if category_id == "": + raise SynapseError(400, "category_id cannot be empty string") + + resp = yield self.handler.update_group_summary_room( + group_id, requester_user_id, + room_id=room_id, + category_id=category_id, + content=content, + ) + + defer.returnValue((200, resp)) + + @defer.inlineCallbacks + def on_DELETE(self, origin, content, query, group_id, category_id, room_id): + requester_user_id = parse_string_from_args(query, "requester_user_id") + if get_domain_from_id(requester_user_id) != origin: + raise SynapseError(403, "requester_user_id doesn't match origin") + + if category_id == "": + raise SynapseError(400, "category_id cannot be empty string") + + resp = yield self.handler.delete_group_summary_room( + group_id, requester_user_id, + room_id=room_id, + category_id=category_id, + ) + + defer.returnValue((200, resp)) + + +class FederationGroupsCategoriesServlet(BaseFederationServlet): + """Get all categories for a group + """ + PATH = ( + "/groups/(?P[^/]*)/categories/$" + ) + + @defer.inlineCallbacks + def on_GET(self, origin, content, query, group_id): + requester_user_id = parse_string_from_args(query, "requester_user_id") + if get_domain_from_id(requester_user_id) != origin: + raise SynapseError(403, "requester_user_id doesn't match origin") + + resp = yield self.handler.get_group_categories( + group_id, requester_user_id, + ) + + defer.returnValue((200, resp)) + + +class FederationGroupsCategoryServlet(BaseFederationServlet): + """Add/remove/get a category in a group + """ + PATH = ( + "/groups/(?P[^/]*)/categories/(?P[^/]+)$" + ) + + @defer.inlineCallbacks + def on_GET(self, origin, content, query, group_id, category_id): + requester_user_id = parse_string_from_args(query, "requester_user_id") + if get_domain_from_id(requester_user_id) != origin: + raise SynapseError(403, "requester_user_id doesn't match origin") + + resp = yield self.handler.get_group_category( + group_id, requester_user_id, category_id + ) + + defer.returnValue((200, resp)) + + @defer.inlineCallbacks + def on_POST(self, origin, content, query, group_id, category_id): + requester_user_id = parse_string_from_args(query, "requester_user_id") + if get_domain_from_id(requester_user_id) != origin: + raise SynapseError(403, "requester_user_id doesn't match origin") + + if category_id == "": + raise SynapseError(400, "category_id cannot be empty string") + + resp = yield self.handler.upsert_group_category( + group_id, requester_user_id, category_id, content, + ) + + defer.returnValue((200, resp)) + + @defer.inlineCallbacks + def on_DELETE(self, origin, content, query, group_id, category_id): + requester_user_id = parse_string_from_args(query, "requester_user_id") + if get_domain_from_id(requester_user_id) != origin: + raise SynapseError(403, "requester_user_id doesn't match origin") + + if category_id == "": + raise SynapseError(400, "category_id cannot be empty string") + + resp = yield self.handler.delete_group_category( + group_id, requester_user_id, category_id, + ) + + defer.returnValue((200, resp)) + + +class FederationGroupsRolesServlet(BaseFederationServlet): + """Get roles in a group + """ + PATH = ( + "/groups/(?P[^/]*)/roles/$" + ) + + @defer.inlineCallbacks + def on_GET(self, origin, content, query, group_id): + requester_user_id = parse_string_from_args(query, "requester_user_id") + if get_domain_from_id(requester_user_id) != origin: + raise SynapseError(403, "requester_user_id doesn't match origin") + + resp = yield self.handler.get_group_roles( + group_id, requester_user_id, + ) + + defer.returnValue((200, resp)) + + +class FederationGroupsRoleServlet(BaseFederationServlet): + """Add/remove/get a role in a group + """ + PATH = ( + "/groups/(?P[^/]*)/roles/(?P[^/]+)$" + ) + + @defer.inlineCallbacks + def on_GET(self, origin, content, query, group_id, role_id): + requester_user_id = parse_string_from_args(query, "requester_user_id") + if get_domain_from_id(requester_user_id) != origin: + raise SynapseError(403, "requester_user_id doesn't match origin") + + resp = yield self.handler.get_group_role( + group_id, requester_user_id, role_id + ) + + defer.returnValue((200, resp)) + + @defer.inlineCallbacks + def on_POST(self, origin, content, query, group_id, role_id): + requester_user_id = parse_string_from_args(query, "requester_user_id") + if get_domain_from_id(requester_user_id) != origin: + raise SynapseError(403, "requester_user_id doesn't match origin") + + if role_id == "": + raise SynapseError(400, "role_id cannot be empty string") + + resp = yield self.handler.update_group_role( + group_id, requester_user_id, role_id, content, + ) + + defer.returnValue((200, resp)) + + @defer.inlineCallbacks + def on_DELETE(self, origin, content, query, group_id, role_id): + requester_user_id = parse_string_from_args(query, "requester_user_id") + if get_domain_from_id(requester_user_id) != origin: + raise SynapseError(403, "requester_user_id doesn't match origin") + + if role_id == "": + raise SynapseError(400, "role_id cannot be empty string") + + resp = yield self.handler.delete_group_role( + group_id, requester_user_id, role_id, + ) + + defer.returnValue((200, resp)) + + +class FederationGroupsSummaryUsersServlet(BaseFederationServlet): + """Add/remove a user from the group summary, with optional role. + + Matches both: + - /groups/:group/summary/users/:user_id + - /groups/:group/summary/roles/:role/users/:user_id + """ + PATH = ( + "/groups/(?P[^/]*)/summary" + "(/roles/(?P[^/]+))?" + "/users/(?P[^/]*)$" + ) + + @defer.inlineCallbacks + def on_POST(self, origin, content, query, group_id, role_id, user_id): + requester_user_id = parse_string_from_args(query, "requester_user_id") + if get_domain_from_id(requester_user_id) != origin: + raise SynapseError(403, "requester_user_id doesn't match origin") + + if role_id == "": + raise SynapseError(400, "role_id cannot be empty string") + + resp = yield self.handler.update_group_summary_user( + group_id, requester_user_id, + user_id=user_id, + role_id=role_id, + content=content, + ) + + defer.returnValue((200, resp)) + + @defer.inlineCallbacks + def on_DELETE(self, origin, content, query, group_id, role_id, user_id): + requester_user_id = parse_string_from_args(query, "requester_user_id") + if get_domain_from_id(requester_user_id) != origin: + raise SynapseError(403, "requester_user_id doesn't match origin") + + if role_id == "": + raise SynapseError(400, "role_id cannot be empty string") + + resp = yield self.handler.delete_group_summary_user( + group_id, requester_user_id, + user_id=user_id, + role_id=role_id, + ) + + defer.returnValue((200, resp)) + + +class FederationGroupsBulkPublicisedServlet(BaseFederationServlet): + """Get roles in a group + """ + PATH = ( + "/get_groups_publicised$" + ) + + @defer.inlineCallbacks + def on_POST(self, origin, content, query): + resp = yield self.handler.bulk_get_publicised_groups( + content["user_ids"], proxy=False, + ) + + defer.returnValue((200, resp)) + + +class FederationGroupsSettingJoinPolicyServlet(BaseFederationServlet): + """Sets whether a group is joinable without an invite or knock + """ + PATH = "/groups/(?P[^/]*)/settings/m.join_policy$" + + @defer.inlineCallbacks + def on_PUT(self, origin, content, query, group_id): + requester_user_id = parse_string_from_args(query, "requester_user_id") + if get_domain_from_id(requester_user_id) != origin: + raise SynapseError(403, "requester_user_id doesn't match origin") + + new_content = yield self.handler.set_group_join_policy( + group_id, requester_user_id, content + ) + + defer.returnValue((200, new_content)) + + FEDERATION_SERVLET_CLASSES = ( FederationSendServlet, FederationPullServlet, @@ -635,15 +1221,49 @@ FEDERATION_SERVLET_CLASSES = ( FederationVersionServlet, ) + ROOM_LIST_CLASSES = ( PublicRoomList, ) +GROUP_SERVER_SERVLET_CLASSES = ( + FederationGroupsProfileServlet, + FederationGroupsSummaryServlet, + FederationGroupsRoomsServlet, + FederationGroupsUsersServlet, + FederationGroupsInvitedUsersServlet, + FederationGroupsInviteServlet, + FederationGroupsAcceptInviteServlet, + FederationGroupsJoinServlet, + FederationGroupsRemoveUserServlet, + FederationGroupsSummaryRoomsServlet, + FederationGroupsCategoriesServlet, + FederationGroupsCategoryServlet, + FederationGroupsRolesServlet, + FederationGroupsRoleServlet, + FederationGroupsSummaryUsersServlet, + FederationGroupsAddRoomsServlet, + FederationGroupsAddRoomsConfigServlet, + FederationGroupsSettingJoinPolicyServlet, +) + + +GROUP_LOCAL_SERVLET_CLASSES = ( + FederationGroupsLocalInviteServlet, + FederationGroupsRemoveLocalUserServlet, + FederationGroupsBulkPublicisedServlet, +) + + +GROUP_ATTESTATION_SERVLET_CLASSES = ( + FederationGroupsRenewAttestaionServlet, +) + def register_servlets(hs, resource, authenticator, ratelimiter): for servletclass in FEDERATION_SERVLET_CLASSES: servletclass( - handler=hs.get_replication_layer(), + handler=hs.get_federation_server(), authenticator=authenticator, ratelimiter=ratelimiter, server_name=hs.hostname, @@ -656,3 +1276,27 @@ def register_servlets(hs, resource, authenticator, ratelimiter): ratelimiter=ratelimiter, server_name=hs.hostname, ).register(resource) + + for servletclass in GROUP_SERVER_SERVLET_CLASSES: + servletclass( + handler=hs.get_groups_server_handler(), + authenticator=authenticator, + ratelimiter=ratelimiter, + server_name=hs.hostname, + ).register(resource) + + for servletclass in GROUP_LOCAL_SERVLET_CLASSES: + servletclass( + handler=hs.get_groups_local_handler(), + authenticator=authenticator, + ratelimiter=ratelimiter, + server_name=hs.hostname, + ).register(resource) + + for servletclass in GROUP_ATTESTATION_SERVLET_CLASSES: + servletclass( + handler=hs.get_groups_attestation_renewer(), + authenticator=authenticator, + ratelimiter=ratelimiter, + server_name=hs.hostname, + ).register(resource) diff --git a/synapse/federation/units.py b/synapse/federation/units.py index 3f645acc4..bb1b3b13f 100644 --- a/synapse/federation/units.py +++ b/synapse/federation/units.py @@ -17,10 +17,9 @@ server protocol. """ -from synapse.util.jsonobject import JsonEncodedObject - import logging +from synapse.util.jsonobject import JsonEncodedObject logger = logging.getLogger(__name__) @@ -74,8 +73,6 @@ class Transaction(JsonEncodedObject): "previous_ids", "pdus", "edus", - "transaction_id", - "destination", "pdu_failures", ] diff --git a/synapse/groups/__init__.py b/synapse/groups/__init__.py new file mode 100644 index 000000000..e69de29bb diff --git a/synapse/groups/attestations.py b/synapse/groups/attestations.py new file mode 100644 index 000000000..47452700a --- /dev/null +++ b/synapse/groups/attestations.py @@ -0,0 +1,198 @@ +# -*- coding: utf-8 -*- +# Copyright 2017 Vector Creations Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +"""Attestations ensure that users and groups can't lie about their memberships. + +When a user joins a group the HS and GS swap attestations, which allow them +both to independently prove to third parties their membership.These +attestations have a validity period so need to be periodically renewed. + +If a user leaves (or gets kicked out of) a group, either side can still use +their attestation to "prove" their membership, until the attestation expires. +Therefore attestations shouldn't be relied on to prove membership in important +cases, but can for less important situtations, e.g. showing a users membership +of groups on their profile, showing flairs, etc. + +An attestation is a signed blob of json that looks like: + + { + "user_id": "@foo:a.example.com", + "group_id": "+bar:b.example.com", + "valid_until_ms": 1507994728530, + "signatures":{"matrix.org":{"ed25519:auto":"..."}} + } +""" + +import logging +import random + +from signedjson.sign import sign_json + +from twisted.internet import defer + +from synapse.api.errors import SynapseError +from synapse.types import get_domain_from_id +from synapse.util.logcontext import run_in_background + +logger = logging.getLogger(__name__) + + +# Default validity duration for new attestations we create +DEFAULT_ATTESTATION_LENGTH_MS = 3 * 24 * 60 * 60 * 1000 + +# We add some jitter to the validity duration of attestations so that if we +# add lots of users at once we don't need to renew them all at once. +# The jitter is a multiplier picked randomly between the first and second number +DEFAULT_ATTESTATION_JITTER = (0.9, 1.3) + +# Start trying to update our attestations when they come this close to expiring +UPDATE_ATTESTATION_TIME_MS = 1 * 24 * 60 * 60 * 1000 + + +class GroupAttestationSigning(object): + """Creates and verifies group attestations. + """ + def __init__(self, hs): + self.keyring = hs.get_keyring() + self.clock = hs.get_clock() + self.server_name = hs.hostname + self.signing_key = hs.config.signing_key[0] + + @defer.inlineCallbacks + def verify_attestation(self, attestation, group_id, user_id, server_name=None): + """Verifies that the given attestation matches the given parameters. + + An optional server_name can be supplied to explicitly set which server's + signature is expected. Otherwise assumes that either the group_id or user_id + is local and uses the other's server as the one to check. + """ + + if not server_name: + if get_domain_from_id(group_id) == self.server_name: + server_name = get_domain_from_id(user_id) + elif get_domain_from_id(user_id) == self.server_name: + server_name = get_domain_from_id(group_id) + else: + raise Exception("Expected either group_id or user_id to be local") + + if user_id != attestation["user_id"]: + raise SynapseError(400, "Attestation has incorrect user_id") + + if group_id != attestation["group_id"]: + raise SynapseError(400, "Attestation has incorrect group_id") + valid_until_ms = attestation["valid_until_ms"] + + # TODO: We also want to check that *new* attestations that people give + # us to store are valid for at least a little while. + if valid_until_ms < self.clock.time_msec(): + raise SynapseError(400, "Attestation expired") + + yield self.keyring.verify_json_for_server(server_name, attestation) + + def create_attestation(self, group_id, user_id): + """Create an attestation for the group_id and user_id with default + validity length. + """ + validity_period = DEFAULT_ATTESTATION_LENGTH_MS + validity_period *= random.uniform(*DEFAULT_ATTESTATION_JITTER) + valid_until_ms = int(self.clock.time_msec() + validity_period) + + return sign_json({ + "group_id": group_id, + "user_id": user_id, + "valid_until_ms": valid_until_ms, + }, self.server_name, self.signing_key) + + +class GroupAttestionRenewer(object): + """Responsible for sending and receiving attestation updates. + """ + + def __init__(self, hs): + self.clock = hs.get_clock() + self.store = hs.get_datastore() + self.assestations = hs.get_groups_attestation_signing() + self.transport_client = hs.get_federation_transport_client() + self.is_mine_id = hs.is_mine_id + self.attestations = hs.get_groups_attestation_signing() + + self._renew_attestations_loop = self.clock.looping_call( + self._renew_attestations, 30 * 60 * 1000, + ) + + @defer.inlineCallbacks + def on_renew_attestation(self, group_id, user_id, content): + """When a remote updates an attestation + """ + attestation = content["attestation"] + + if not self.is_mine_id(group_id) and not self.is_mine_id(user_id): + raise SynapseError(400, "Neither user not group are on this server") + + yield self.attestations.verify_attestation( + attestation, + user_id=user_id, + group_id=group_id, + ) + + yield self.store.update_remote_attestion(group_id, user_id, attestation) + + defer.returnValue({}) + + @defer.inlineCallbacks + def _renew_attestations(self): + """Called periodically to check if we need to update any of our attestations + """ + + now = self.clock.time_msec() + + rows = yield self.store.get_attestations_need_renewals( + now + UPDATE_ATTESTATION_TIME_MS + ) + + @defer.inlineCallbacks + def _renew_attestation(group_id, user_id): + try: + if not self.is_mine_id(group_id): + destination = get_domain_from_id(group_id) + elif not self.is_mine_id(user_id): + destination = get_domain_from_id(user_id) + else: + logger.warn( + "Incorrectly trying to do attestations for user: %r in %r", + user_id, group_id, + ) + yield self.store.remove_attestation_renewal(group_id, user_id) + return + + attestation = self.attestations.create_attestation(group_id, user_id) + + yield self.transport_client.renew_group_attestation( + destination, group_id, user_id, + content={"attestation": attestation}, + ) + + yield self.store.update_attestation_renewal( + group_id, user_id, attestation + ) + except Exception: + logger.exception("Error renewing attestation of %r in %r", + user_id, group_id) + + for row in rows: + group_id = row["group_id"] + user_id = row["user_id"] + + run_in_background(_renew_attestation, group_id, user_id) diff --git a/synapse/groups/groups_server.py b/synapse/groups/groups_server.py new file mode 100644 index 000000000..633c865ed --- /dev/null +++ b/synapse/groups/groups_server.py @@ -0,0 +1,953 @@ +# -*- coding: utf-8 -*- +# Copyright 2017 Vector Creations Ltd +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging + +from six import string_types + +from twisted.internet import defer + +from synapse.api.errors import SynapseError +from synapse.types import GroupID, RoomID, UserID, get_domain_from_id + +logger = logging.getLogger(__name__) + + +# TODO: Allow users to "knock" or simpkly join depending on rules +# TODO: Federation admin APIs +# TODO: is_priveged flag to users and is_public to users and rooms +# TODO: Audit log for admins (profile updates, membership changes, users who tried +# to join but were rejected, etc) +# TODO: Flairs + + +class GroupsServerHandler(object): + def __init__(self, hs): + self.hs = hs + self.store = hs.get_datastore() + self.room_list_handler = hs.get_room_list_handler() + self.auth = hs.get_auth() + self.clock = hs.get_clock() + self.keyring = hs.get_keyring() + self.is_mine_id = hs.is_mine_id + self.signing_key = hs.config.signing_key[0] + self.server_name = hs.hostname + self.attestations = hs.get_groups_attestation_signing() + self.transport_client = hs.get_federation_transport_client() + self.profile_handler = hs.get_profile_handler() + + # Ensure attestations get renewed + hs.get_groups_attestation_renewer() + + @defer.inlineCallbacks + def check_group_is_ours(self, group_id, requester_user_id, + and_exists=False, and_is_admin=None): + """Check that the group is ours, and optionally if it exists. + + If group does exist then return group. + + Args: + group_id (str) + and_exists (bool): whether to also check if group exists + and_is_admin (str): whether to also check if given str is a user_id + that is an admin + """ + if not self.is_mine_id(group_id): + raise SynapseError(400, "Group not on this server") + + group = yield self.store.get_group(group_id) + if and_exists and not group: + raise SynapseError(404, "Unknown group") + + is_user_in_group = yield self.store.is_user_in_group(requester_user_id, group_id) + if group and not is_user_in_group and not group["is_public"]: + raise SynapseError(404, "Unknown group") + + if and_is_admin: + is_admin = yield self.store.is_user_admin_in_group(group_id, and_is_admin) + if not is_admin: + raise SynapseError(403, "User is not admin in group") + + defer.returnValue(group) + + @defer.inlineCallbacks + def get_group_summary(self, group_id, requester_user_id): + """Get the summary for a group as seen by requester_user_id. + + The group summary consists of the profile of the room, and a curated + list of users and rooms. These list *may* be organised by role/category. + The roles/categories are ordered, and so are the users/rooms within them. + + A user/room may appear in multiple roles/categories. + """ + yield self.check_group_is_ours(group_id, requester_user_id, and_exists=True) + + is_user_in_group = yield self.store.is_user_in_group(requester_user_id, group_id) + + profile = yield self.get_group_profile(group_id, requester_user_id) + + users, roles = yield self.store.get_users_for_summary_by_role( + group_id, include_private=is_user_in_group, + ) + + # TODO: Add profiles to users + + rooms, categories = yield self.store.get_rooms_for_summary_by_category( + group_id, include_private=is_user_in_group, + ) + + for room_entry in rooms: + room_id = room_entry["room_id"] + joined_users = yield self.store.get_users_in_room(room_id) + entry = yield self.room_list_handler.generate_room_entry( + room_id, len(joined_users), + with_alias=False, allow_private=True, + ) + entry = dict(entry) # so we don't change whats cached + entry.pop("room_id", None) + + room_entry["profile"] = entry + + rooms.sort(key=lambda e: e.get("order", 0)) + + for entry in users: + user_id = entry["user_id"] + + if not self.is_mine_id(requester_user_id): + attestation = yield self.store.get_remote_attestation(group_id, user_id) + if not attestation: + continue + + entry["attestation"] = attestation + else: + entry["attestation"] = self.attestations.create_attestation( + group_id, user_id, + ) + + user_profile = yield self.profile_handler.get_profile_from_cache(user_id) + entry.update(user_profile) + + users.sort(key=lambda e: e.get("order", 0)) + + membership_info = yield self.store.get_users_membership_info_in_group( + group_id, requester_user_id, + ) + + defer.returnValue({ + "profile": profile, + "users_section": { + "users": users, + "roles": roles, + "total_user_count_estimate": 0, # TODO + }, + "rooms_section": { + "rooms": rooms, + "categories": categories, + "total_room_count_estimate": 0, # TODO + }, + "user": membership_info, + }) + + @defer.inlineCallbacks + def update_group_summary_room(self, group_id, requester_user_id, + room_id, category_id, content): + """Add/update a room to the group summary + """ + yield self.check_group_is_ours( + group_id, + requester_user_id, + and_exists=True, + and_is_admin=requester_user_id, + ) + + RoomID.from_string(room_id) # Ensure valid room id + + order = content.get("order", None) + + is_public = _parse_visibility_from_contents(content) + + yield self.store.add_room_to_summary( + group_id=group_id, + room_id=room_id, + category_id=category_id, + order=order, + is_public=is_public, + ) + + defer.returnValue({}) + + @defer.inlineCallbacks + def delete_group_summary_room(self, group_id, requester_user_id, + room_id, category_id): + """Remove a room from the summary + """ + yield self.check_group_is_ours( + group_id, + requester_user_id, + and_exists=True, + and_is_admin=requester_user_id, + ) + + yield self.store.remove_room_from_summary( + group_id=group_id, + room_id=room_id, + category_id=category_id, + ) + + defer.returnValue({}) + + @defer.inlineCallbacks + def set_group_join_policy(self, group_id, requester_user_id, content): + """Sets the group join policy. + + Currently supported policies are: + - "invite": an invite must be received and accepted in order to join. + - "open": anyone can join. + """ + yield self.check_group_is_ours( + group_id, requester_user_id, and_exists=True, and_is_admin=requester_user_id + ) + + join_policy = _parse_join_policy_from_contents(content) + if join_policy is None: + raise SynapseError( + 400, "No value specified for 'm.join_policy'" + ) + + yield self.store.set_group_join_policy(group_id, join_policy=join_policy) + + defer.returnValue({}) + + @defer.inlineCallbacks + def get_group_categories(self, group_id, requester_user_id): + """Get all categories in a group (as seen by user) + """ + yield self.check_group_is_ours(group_id, requester_user_id, and_exists=True) + + categories = yield self.store.get_group_categories( + group_id=group_id, + ) + defer.returnValue({"categories": categories}) + + @defer.inlineCallbacks + def get_group_category(self, group_id, requester_user_id, category_id): + """Get a specific category in a group (as seen by user) + """ + yield self.check_group_is_ours(group_id, requester_user_id, and_exists=True) + + res = yield self.store.get_group_category( + group_id=group_id, + category_id=category_id, + ) + + defer.returnValue(res) + + @defer.inlineCallbacks + def update_group_category(self, group_id, requester_user_id, category_id, content): + """Add/Update a group category + """ + yield self.check_group_is_ours( + group_id, + requester_user_id, + and_exists=True, + and_is_admin=requester_user_id, + ) + + is_public = _parse_visibility_from_contents(content) + profile = content.get("profile") + + yield self.store.upsert_group_category( + group_id=group_id, + category_id=category_id, + is_public=is_public, + profile=profile, + ) + + defer.returnValue({}) + + @defer.inlineCallbacks + def delete_group_category(self, group_id, requester_user_id, category_id): + """Delete a group category + """ + yield self.check_group_is_ours( + group_id, + requester_user_id, + and_exists=True, + and_is_admin=requester_user_id + ) + + yield self.store.remove_group_category( + group_id=group_id, + category_id=category_id, + ) + + defer.returnValue({}) + + @defer.inlineCallbacks + def get_group_roles(self, group_id, requester_user_id): + """Get all roles in a group (as seen by user) + """ + yield self.check_group_is_ours(group_id, requester_user_id, and_exists=True) + + roles = yield self.store.get_group_roles( + group_id=group_id, + ) + defer.returnValue({"roles": roles}) + + @defer.inlineCallbacks + def get_group_role(self, group_id, requester_user_id, role_id): + """Get a specific role in a group (as seen by user) + """ + yield self.check_group_is_ours(group_id, requester_user_id, and_exists=True) + + res = yield self.store.get_group_role( + group_id=group_id, + role_id=role_id, + ) + defer.returnValue(res) + + @defer.inlineCallbacks + def update_group_role(self, group_id, requester_user_id, role_id, content): + """Add/update a role in a group + """ + yield self.check_group_is_ours( + group_id, + requester_user_id, + and_exists=True, + and_is_admin=requester_user_id, + ) + + is_public = _parse_visibility_from_contents(content) + + profile = content.get("profile") + + yield self.store.upsert_group_role( + group_id=group_id, + role_id=role_id, + is_public=is_public, + profile=profile, + ) + + defer.returnValue({}) + + @defer.inlineCallbacks + def delete_group_role(self, group_id, requester_user_id, role_id): + """Remove role from group + """ + yield self.check_group_is_ours( + group_id, + requester_user_id, + and_exists=True, + and_is_admin=requester_user_id, + ) + + yield self.store.remove_group_role( + group_id=group_id, + role_id=role_id, + ) + + defer.returnValue({}) + + @defer.inlineCallbacks + def update_group_summary_user(self, group_id, requester_user_id, user_id, role_id, + content): + """Add/update a users entry in the group summary + """ + yield self.check_group_is_ours( + group_id, requester_user_id, and_exists=True, and_is_admin=requester_user_id, + ) + + order = content.get("order", None) + + is_public = _parse_visibility_from_contents(content) + + yield self.store.add_user_to_summary( + group_id=group_id, + user_id=user_id, + role_id=role_id, + order=order, + is_public=is_public, + ) + + defer.returnValue({}) + + @defer.inlineCallbacks + def delete_group_summary_user(self, group_id, requester_user_id, user_id, role_id): + """Remove a user from the group summary + """ + yield self.check_group_is_ours( + group_id, requester_user_id, and_exists=True, and_is_admin=requester_user_id, + ) + + yield self.store.remove_user_from_summary( + group_id=group_id, + user_id=user_id, + role_id=role_id, + ) + + defer.returnValue({}) + + @defer.inlineCallbacks + def get_group_profile(self, group_id, requester_user_id): + """Get the group profile as seen by requester_user_id + """ + + yield self.check_group_is_ours(group_id, requester_user_id) + + group = yield self.store.get_group(group_id) + + if group: + cols = [ + "name", "short_description", "long_description", + "avatar_url", "is_public", + ] + group_description = {key: group[key] for key in cols} + group_description["is_openly_joinable"] = group["join_policy"] == "open" + + defer.returnValue(group_description) + else: + raise SynapseError(404, "Unknown group") + + @defer.inlineCallbacks + def update_group_profile(self, group_id, requester_user_id, content): + """Update the group profile + """ + yield self.check_group_is_ours( + group_id, requester_user_id, and_exists=True, and_is_admin=requester_user_id, + ) + + profile = {} + for keyname in ("name", "avatar_url", "short_description", + "long_description"): + if keyname in content: + value = content[keyname] + if not isinstance(value, string_types): + raise SynapseError(400, "%r value is not a string" % (keyname,)) + profile[keyname] = value + + yield self.store.update_group_profile(group_id, profile) + + @defer.inlineCallbacks + def get_users_in_group(self, group_id, requester_user_id): + """Get the users in group as seen by requester_user_id. + + The ordering is arbitrary at the moment + """ + + yield self.check_group_is_ours(group_id, requester_user_id, and_exists=True) + + is_user_in_group = yield self.store.is_user_in_group(requester_user_id, group_id) + + user_results = yield self.store.get_users_in_group( + group_id, include_private=is_user_in_group, + ) + + chunk = [] + for user_result in user_results: + g_user_id = user_result["user_id"] + is_public = user_result["is_public"] + is_privileged = user_result["is_admin"] + + entry = {"user_id": g_user_id} + + profile = yield self.profile_handler.get_profile_from_cache(g_user_id) + entry.update(profile) + + entry["is_public"] = bool(is_public) + entry["is_privileged"] = bool(is_privileged) + + if not self.is_mine_id(g_user_id): + attestation = yield self.store.get_remote_attestation(group_id, g_user_id) + if not attestation: + continue + + entry["attestation"] = attestation + else: + entry["attestation"] = self.attestations.create_attestation( + group_id, g_user_id, + ) + + chunk.append(entry) + + # TODO: If admin add lists of users whose attestations have timed out + + defer.returnValue({ + "chunk": chunk, + "total_user_count_estimate": len(user_results), + }) + + @defer.inlineCallbacks + def get_invited_users_in_group(self, group_id, requester_user_id): + """Get the users that have been invited to a group as seen by requester_user_id. + + The ordering is arbitrary at the moment + """ + + yield self.check_group_is_ours(group_id, requester_user_id, and_exists=True) + + is_user_in_group = yield self.store.is_user_in_group(requester_user_id, group_id) + + if not is_user_in_group: + raise SynapseError(403, "User not in group") + + invited_users = yield self.store.get_invited_users_in_group(group_id) + + user_profiles = [] + + for user_id in invited_users: + user_profile = { + "user_id": user_id + } + try: + profile = yield self.profile_handler.get_profile_from_cache(user_id) + user_profile.update(profile) + except Exception as e: + logger.warn("Error getting profile for %s: %s", user_id, e) + user_profiles.append(user_profile) + + defer.returnValue({ + "chunk": user_profiles, + "total_user_count_estimate": len(invited_users), + }) + + @defer.inlineCallbacks + def get_rooms_in_group(self, group_id, requester_user_id): + """Get the rooms in group as seen by requester_user_id + + This returns rooms in order of decreasing number of joined users + """ + + yield self.check_group_is_ours(group_id, requester_user_id, and_exists=True) + + is_user_in_group = yield self.store.is_user_in_group(requester_user_id, group_id) + + room_results = yield self.store.get_rooms_in_group( + group_id, include_private=is_user_in_group, + ) + + chunk = [] + for room_result in room_results: + room_id = room_result["room_id"] + + joined_users = yield self.store.get_users_in_room(room_id) + entry = yield self.room_list_handler.generate_room_entry( + room_id, len(joined_users), + with_alias=False, allow_private=True, + ) + + if not entry: + continue + + entry["is_public"] = bool(room_result["is_public"]) + + chunk.append(entry) + + chunk.sort(key=lambda e: -e["num_joined_members"]) + + defer.returnValue({ + "chunk": chunk, + "total_room_count_estimate": len(room_results), + }) + + @defer.inlineCallbacks + def add_room_to_group(self, group_id, requester_user_id, room_id, content): + """Add room to group + """ + RoomID.from_string(room_id) # Ensure valid room id + + yield self.check_group_is_ours( + group_id, requester_user_id, and_exists=True, and_is_admin=requester_user_id + ) + + is_public = _parse_visibility_from_contents(content) + + yield self.store.add_room_to_group(group_id, room_id, is_public=is_public) + + defer.returnValue({}) + + @defer.inlineCallbacks + def update_room_in_group(self, group_id, requester_user_id, room_id, config_key, + content): + """Update room in group + """ + RoomID.from_string(room_id) # Ensure valid room id + + yield self.check_group_is_ours( + group_id, requester_user_id, and_exists=True, and_is_admin=requester_user_id + ) + + if config_key == "m.visibility": + is_public = _parse_visibility_dict(content) + + yield self.store.update_room_in_group_visibility( + group_id, room_id, + is_public=is_public, + ) + else: + raise SynapseError(400, "Uknown config option") + + defer.returnValue({}) + + @defer.inlineCallbacks + def remove_room_from_group(self, group_id, requester_user_id, room_id): + """Remove room from group + """ + yield self.check_group_is_ours( + group_id, requester_user_id, and_exists=True, and_is_admin=requester_user_id + ) + + yield self.store.remove_room_from_group(group_id, room_id) + + defer.returnValue({}) + + @defer.inlineCallbacks + def invite_to_group(self, group_id, user_id, requester_user_id, content): + """Invite user to group + """ + + group = yield self.check_group_is_ours( + group_id, requester_user_id, and_exists=True, and_is_admin=requester_user_id + ) + + # TODO: Check if user knocked + # TODO: Check if user is already invited + + content = { + "profile": { + "name": group["name"], + "avatar_url": group["avatar_url"], + }, + "inviter": requester_user_id, + } + + if self.hs.is_mine_id(user_id): + groups_local = self.hs.get_groups_local_handler() + res = yield groups_local.on_invite(group_id, user_id, content) + local_attestation = None + else: + local_attestation = self.attestations.create_attestation(group_id, user_id) + content.update({ + "attestation": local_attestation, + }) + + res = yield self.transport_client.invite_to_group_notification( + get_domain_from_id(user_id), group_id, user_id, content + ) + + user_profile = res.get("user_profile", {}) + yield self.store.add_remote_profile_cache( + user_id, + displayname=user_profile.get("displayname"), + avatar_url=user_profile.get("avatar_url"), + ) + + if res["state"] == "join": + if not self.hs.is_mine_id(user_id): + remote_attestation = res["attestation"] + + yield self.attestations.verify_attestation( + remote_attestation, + user_id=user_id, + group_id=group_id, + ) + else: + remote_attestation = None + + yield self.store.add_user_to_group( + group_id, user_id, + is_admin=False, + is_public=False, # TODO + local_attestation=local_attestation, + remote_attestation=remote_attestation, + ) + elif res["state"] == "invite": + yield self.store.add_group_invite( + group_id, user_id, + ) + defer.returnValue({ + "state": "invite" + }) + elif res["state"] == "reject": + defer.returnValue({ + "state": "reject" + }) + else: + raise SynapseError(502, "Unknown state returned by HS") + + @defer.inlineCallbacks + def _add_user(self, group_id, user_id, content): + """Add a user to a group based on a content dict. + + See accept_invite, join_group. + """ + if not self.hs.is_mine_id(user_id): + local_attestation = self.attestations.create_attestation( + group_id, user_id, + ) + + remote_attestation = content["attestation"] + + yield self.attestations.verify_attestation( + remote_attestation, + user_id=user_id, + group_id=group_id, + ) + else: + local_attestation = None + remote_attestation = None + + is_public = _parse_visibility_from_contents(content) + + yield self.store.add_user_to_group( + group_id, user_id, + is_admin=False, + is_public=is_public, + local_attestation=local_attestation, + remote_attestation=remote_attestation, + ) + + defer.returnValue(local_attestation) + + @defer.inlineCallbacks + def accept_invite(self, group_id, requester_user_id, content): + """User tries to accept an invite to the group. + + This is different from them asking to join, and so should error if no + invite exists (and they're not a member of the group) + """ + + yield self.check_group_is_ours(group_id, requester_user_id, and_exists=True) + + is_invited = yield self.store.is_user_invited_to_local_group( + group_id, requester_user_id, + ) + if not is_invited: + raise SynapseError(403, "User not invited to group") + + local_attestation = yield self._add_user(group_id, requester_user_id, content) + + defer.returnValue({ + "state": "join", + "attestation": local_attestation, + }) + + @defer.inlineCallbacks + def join_group(self, group_id, requester_user_id, content): + """User tries to join the group. + + This will error if the group requires an invite/knock to join + """ + + group_info = yield self.check_group_is_ours( + group_id, requester_user_id, and_exists=True + ) + if group_info['join_policy'] != "open": + raise SynapseError(403, "Group is not publicly joinable") + + local_attestation = yield self._add_user(group_id, requester_user_id, content) + + defer.returnValue({ + "state": "join", + "attestation": local_attestation, + }) + + @defer.inlineCallbacks + def knock(self, group_id, requester_user_id, content): + """A user requests becoming a member of the group + """ + yield self.check_group_is_ours(group_id, requester_user_id, and_exists=True) + + raise NotImplementedError() + + @defer.inlineCallbacks + def accept_knock(self, group_id, requester_user_id, content): + """Accept a users knock to the room. + + Errors if the user hasn't knocked, rather than inviting them. + """ + + yield self.check_group_is_ours(group_id, requester_user_id, and_exists=True) + + raise NotImplementedError() + + @defer.inlineCallbacks + def remove_user_from_group(self, group_id, user_id, requester_user_id, content): + """Remove a user from the group; either a user is leaving or an admin + kicked them. + """ + + yield self.check_group_is_ours(group_id, requester_user_id, and_exists=True) + + is_kick = False + if requester_user_id != user_id: + is_admin = yield self.store.is_user_admin_in_group( + group_id, requester_user_id + ) + if not is_admin: + raise SynapseError(403, "User is not admin in group") + + is_kick = True + + yield self.store.remove_user_from_group( + group_id, user_id, + ) + + if is_kick: + if self.hs.is_mine_id(user_id): + groups_local = self.hs.get_groups_local_handler() + yield groups_local.user_removed_from_group(group_id, user_id, {}) + else: + yield self.transport_client.remove_user_from_group_notification( + get_domain_from_id(user_id), group_id, user_id, {} + ) + + if not self.hs.is_mine_id(user_id): + yield self.store.maybe_delete_remote_profile_cache(user_id) + + defer.returnValue({}) + + @defer.inlineCallbacks + def create_group(self, group_id, requester_user_id, content): + group = yield self.check_group_is_ours(group_id, requester_user_id) + + logger.info("Attempting to create group with ID: %r", group_id) + + # parsing the id into a GroupID validates it. + group_id_obj = GroupID.from_string(group_id) + + if group: + raise SynapseError(400, "Group already exists") + + is_admin = yield self.auth.is_server_admin(UserID.from_string(requester_user_id)) + if not is_admin: + if not self.hs.config.enable_group_creation: + raise SynapseError( + 403, "Only a server admin can create groups on this server", + ) + localpart = group_id_obj.localpart + if not localpart.startswith(self.hs.config.group_creation_prefix): + raise SynapseError( + 400, + "Can only create groups with prefix %r on this server" % ( + self.hs.config.group_creation_prefix, + ), + ) + + profile = content.get("profile", {}) + name = profile.get("name") + avatar_url = profile.get("avatar_url") + short_description = profile.get("short_description") + long_description = profile.get("long_description") + user_profile = content.get("user_profile", {}) + + yield self.store.create_group( + group_id, + requester_user_id, + name=name, + avatar_url=avatar_url, + short_description=short_description, + long_description=long_description, + ) + + if not self.hs.is_mine_id(requester_user_id): + remote_attestation = content["attestation"] + + yield self.attestations.verify_attestation( + remote_attestation, + user_id=requester_user_id, + group_id=group_id, + ) + + local_attestation = self.attestations.create_attestation( + group_id, + requester_user_id, + ) + else: + local_attestation = None + remote_attestation = None + + yield self.store.add_user_to_group( + group_id, requester_user_id, + is_admin=True, + is_public=True, # TODO + local_attestation=local_attestation, + remote_attestation=remote_attestation, + ) + + if not self.hs.is_mine_id(requester_user_id): + yield self.store.add_remote_profile_cache( + requester_user_id, + displayname=user_profile.get("displayname"), + avatar_url=user_profile.get("avatar_url"), + ) + + defer.returnValue({ + "group_id": group_id, + }) + + +def _parse_join_policy_from_contents(content): + """Given a content for a request, return the specified join policy or None + """ + + join_policy_dict = content.get("m.join_policy") + if join_policy_dict: + return _parse_join_policy_dict(join_policy_dict) + else: + return None + + +def _parse_join_policy_dict(join_policy_dict): + """Given a dict for the "m.join_policy" config return the join policy specified + """ + join_policy_type = join_policy_dict.get("type") + if not join_policy_type: + return "invite" + + if join_policy_type not in ("invite", "open"): + raise SynapseError( + 400, "Synapse only supports 'invite'/'open' join rule" + ) + return join_policy_type + + +def _parse_visibility_from_contents(content): + """Given a content for a request parse out whether the entity should be + public or not + """ + + visibility = content.get("m.visibility") + if visibility: + return _parse_visibility_dict(visibility) + else: + is_public = True + + return is_public + + +def _parse_visibility_dict(visibility): + """Given a dict for the "m.visibility" config return if the entity should + be public or not + """ + vis_type = visibility.get("type") + if not vis_type: + return True + + if vis_type not in ("public", "private"): + raise SynapseError( + 400, "Synapse only supports 'public'/'private' visibility" + ) + return vis_type == "public" diff --git a/synapse/handlers/__init__.py b/synapse/handlers/__init__.py index 5ad408f54..413425fed 100644 --- a/synapse/handlers/__init__.py +++ b/synapse/handlers/__init__.py @@ -13,17 +13,11 @@ # See the License for the specific language governing permissions and # limitations under the License. -from .register import RegistrationHandler -from .room import ( - RoomCreationHandler, RoomContextHandler, -) -from .room_member import RoomMemberHandler -from .message import MessageHandler -from .federation import FederationHandler -from .profile import ProfileHandler -from .directory import DirectoryHandler from .admin import AdminHandler +from .directory import DirectoryHandler +from .federation import FederationHandler from .identity import IdentityHandler +from .register import RegistrationHandler from .search import SearchHandler @@ -48,13 +42,8 @@ class Handlers(object): def __init__(self, hs): self.registration_handler = RegistrationHandler(hs) - self.message_handler = MessageHandler(hs) - self.room_creation_handler = RoomCreationHandler(hs) - self.room_member_handler = RoomMemberHandler(hs) self.federation_handler = FederationHandler(hs) - self.profile_handler = ProfileHandler(hs) self.directory_handler = DirectoryHandler(hs) self.admin_handler = AdminHandler(hs) self.identity_handler = IdentityHandler(hs) self.search_handler = SearchHandler(hs) - self.room_context_handler = RoomContextHandler(hs) diff --git a/synapse/handlers/_base.py b/synapse/handlers/_base.py index faa5609c0..704181d2d 100644 --- a/synapse/handlers/_base.py +++ b/synapse/handlers/_base.py @@ -18,11 +18,10 @@ import logging from twisted.internet import defer import synapse.types -from synapse.api.constants import Membership, EventTypes +from synapse.api.constants import EventTypes, Membership from synapse.api.errors import LimitExceededError from synapse.types import UserID - logger = logging.getLogger(__name__) @@ -113,15 +112,16 @@ class BaseHandler(object): guest_access = event.content.get("guest_access", "forbidden") if guest_access != "can_join": if context: + current_state_ids = yield context.get_current_state_ids(self.store) current_state = yield self.store.get_events( - context.current_state_ids.values() + list(current_state_ids.values()) ) else: current_state = yield self.state_handler.get_current_state( event.room_id ) - current_state = current_state.values() + current_state = list(current_state.values()) logger.info("maybe_kick_guest_users %r", current_state) yield self.kick_guest_users(current_state) @@ -158,7 +158,7 @@ class BaseHandler(object): # homeserver. requester = synapse.types.create_requester( target_user, is_guest=True) - handler = self.hs.get_handlers().room_member_handler + handler = self.hs.get_room_member_handler() yield handler.update_membership( requester, target_user, diff --git a/synapse/handlers/admin.py b/synapse/handlers/admin.py index f36b358b4..5d629126f 100644 --- a/synapse/handlers/admin.py +++ b/synapse/handlers/admin.py @@ -13,12 +13,12 @@ # See the License for the specific language governing permissions and # limitations under the License. +import logging + from twisted.internet import defer from ._base import BaseHandler -import logging - logger = logging.getLogger(__name__) diff --git a/synapse/handlers/appservice.py b/synapse/handlers/appservice.py index 05af54d31..ee41aed69 100644 --- a/synapse/handlers/appservice.py +++ b/synapse/handlers/appservice.py @@ -13,16 +13,24 @@ # See the License for the specific language governing permissions and # limitations under the License. -from twisted.internet import defer - -from synapse.api.constants import EventTypes -from synapse.util.metrics import Measure -from synapse.util.logcontext import preserve_fn, preserve_context_over_deferred - import logging +from six import itervalues + +from prometheus_client import Counter + +from twisted.internet import defer + +import synapse +from synapse.api.constants import EventTypes +from synapse.metrics.background_process_metrics import run_as_background_process +from synapse.util.logcontext import make_deferred_yieldable, run_in_background +from synapse.util.metrics import Measure + logger = logging.getLogger(__name__) +events_processed_counter = Counter("synapse_handlers_appservice_events_processed", "") + def log_failure(failure): logger.error( @@ -70,21 +78,25 @@ class ApplicationServicesHandler(object): with Measure(self.clock, "notify_interested_services"): self.is_processing = True try: - upper_bound = self.current_max limit = 100 while True: upper_bound, events = yield self.store.get_new_events_for_appservice( - upper_bound, limit + self.current_max, limit ) if not events: break + events_by_room = {} for event in events: + events_by_room.setdefault(event.room_id, []).append(event) + + @defer.inlineCallbacks + def handle_event(event): # Gather interested services services = yield self._get_services_for_event(event) if len(services) == 0: - continue # no services need notifying + return # no services need notifying # Do we know this user exists? If not, poke the user # query API for all services which match that user regex. @@ -95,19 +107,39 @@ class ApplicationServicesHandler(object): yield self._check_user_exists(event.state_key) if not self.started_scheduler: - self.scheduler.start().addErrback(log_failure) + def start_scheduler(): + return self.scheduler.start().addErrback(log_failure) + run_as_background_process("as_scheduler", start_scheduler) self.started_scheduler = True # Fork off pushes to these services for service in services: - preserve_fn(self.scheduler.submit_event_for_as)( - service, event - ) + self.scheduler.submit_event_for_as(service, event) + + @defer.inlineCallbacks + def handle_room_events(events): + for event in events: + yield handle_event(event) + + yield make_deferred_yieldable(defer.gatherResults([ + run_in_background(handle_room_events, evs) + for evs in itervalues(events_by_room) + ], consumeErrors=True)) yield self.store.set_appservice_last_pos(upper_bound) - if len(events) < limit: - break + now = self.clock.time_msec() + ts = yield self.store.get_received_ts(events[-1].event_id) + + synapse.metrics.event_processing_positions.labels( + "appservice_sender").set(upper_bound) + + events_processed_counter.inc(len(events)) + + synapse.metrics.event_processing_lag.labels( + "appservice_sender").set(now - ts) + synapse.metrics.event_processing_last_ts.labels( + "appservice_sender").set(ts) finally: self.is_processing = False @@ -163,8 +195,11 @@ class ApplicationServicesHandler(object): def query_3pe(self, kind, protocol, fields): services = yield self._get_services_for_3pn(protocol) - results = yield preserve_context_over_deferred(defer.DeferredList([ - preserve_fn(self.appservice_api.query_3pe)(service, kind, protocol, fields) + results = yield make_deferred_yieldable(defer.DeferredList([ + run_in_background( + self.appservice_api.query_3pe, + service, kind, protocol, fields, + ) for service in services ], consumeErrors=True)) @@ -225,11 +260,15 @@ class ApplicationServicesHandler(object): event based on the service regex. """ services = self.store.get_app_services() - interested_list = [ - s for s in services if ( - yield s.is_interested(event, self.store) - ) - ] + + # we can't use a list comprehension here. Since python 3, list + # comprehensions use a generator internally. This means you can't yield + # inside of a list comprehension anymore. + interested_list = [] + for s in services: + if (yield s.is_interested(event, self.store)): + interested_list.append(s) + defer.returnValue(interested_list) def _get_services_for_user(self, user_id): diff --git a/synapse/handlers/auth.py b/synapse/handlers/auth.py index b00446bec..402e44cde 100644 --- a/synapse/handlers/auth.py +++ b/synapse/handlers/auth.py @@ -14,24 +14,32 @@ # See the License for the specific language governing permissions and # limitations under the License. -from twisted.internet import defer - -from ._base import BaseHandler -from synapse.api.constants import LoginType -from synapse.types import UserID -from synapse.api.errors import AuthError, LoginError, Codes, StoreError, SynapseError -from synapse.util.async import run_on_reactor -from synapse.util.caches.expiringcache import ExpiringCache - -from twisted.web.client import PartialDownloadError - import logging + +import attr import bcrypt import pymacaroons -import simplejson +from canonicaljson import json + +from twisted.internet import defer, threads +from twisted.web.client import PartialDownloadError import synapse.util.stringutils as stringutils +from synapse.api.constants import LoginType +from synapse.api.errors import ( + AuthError, + Codes, + InteractiveAuthIncompleteError, + LoginError, + StoreError, + SynapseError, +) +from synapse.module_api import ModuleApi +from synapse.types import UserID +from synapse.util.caches.expiringcache import ExpiringCache +from synapse.util.logcontext import make_deferred_yieldable +from ._base import BaseHandler logger = logging.getLogger(__name__) @@ -46,7 +54,6 @@ class AuthHandler(BaseHandler): """ super(AuthHandler, self).__init__(hs) self.checkers = { - LoginType.PASSWORD: self._check_password_auth, LoginType.RECAPTCHA: self._check_recaptcha, LoginType.EMAIL_IDENTITY: self._check_email_identity, LoginType.MSISDN: self._check_msisdn, @@ -63,10 +70,7 @@ class AuthHandler(BaseHandler): reset_expiry_on_get=True, ) - account_handler = _AccountHandler( - hs, check_user_exists=self.check_user_exists - ) - + account_handler = ModuleApi(hs, self) self.password_providers = [ module(config=config, account_handler=account_handler) for module, config in hs.config.password_providers @@ -75,39 +79,120 @@ class AuthHandler(BaseHandler): logger.info("Extra password_providers: %r", self.password_providers) self.hs = hs # FIXME better possibility to access registrationHandler later? - self.device_handler = hs.get_device_handler() self.macaroon_gen = hs.get_macaroon_generator() + self._password_enabled = hs.config.password_enabled + + # we keep this as a list despite the O(N^2) implication so that we can + # keep PASSWORD first and avoid confusing clients which pick the first + # type in the list. (NB that the spec doesn't require us to do so and + # clients which favour types that they don't understand over those that + # they do are technically broken) + login_types = [] + if self._password_enabled: + login_types.append(LoginType.PASSWORD) + for provider in self.password_providers: + if hasattr(provider, "get_supported_login_types"): + for t in provider.get_supported_login_types().keys(): + if t not in login_types: + login_types.append(t) + self._supported_login_types = login_types + + @defer.inlineCallbacks + def validate_user_via_ui_auth(self, requester, request_body, clientip): + """ + Checks that the user is who they claim to be, via a UI auth. + + This is used for things like device deletion and password reset where + the user already has a valid access token, but we want to double-check + that it isn't stolen by re-authenticating them. + + Args: + requester (Requester): The user, as given by the access token + + request_body (dict): The body of the request sent by the client + + clientip (str): The IP address of the client. + + Returns: + defer.Deferred[dict]: the parameters for this request (which may + have been given only in a previous call). + + Raises: + InteractiveAuthIncompleteError if the client has not yet completed + any of the permitted login flows + + AuthError if the client has completed a login flow, and it gives + a different user to `requester` + """ + + # build a list of supported flows + flows = [ + [login_type] for login_type in self._supported_login_types + ] + + result, params, _ = yield self.check_auth( + flows, request_body, clientip, + ) + + # find the completed login type + for login_type in self._supported_login_types: + if login_type not in result: + continue + + user_id = result[login_type] + break + else: + # this can't happen + raise Exception( + "check_auth returned True but no successful login type", + ) + + # check that the UI auth matched the access token + if user_id != requester.user.to_string(): + raise AuthError(403, "Invalid auth") + + defer.returnValue(params) @defer.inlineCallbacks def check_auth(self, flows, clientdict, clientip): """ Takes a dictionary sent by the client in the login / registration - protocol and handles the login flow. + protocol and handles the User-Interactive Auth flow. As a side effect, this function fills in the 'creds' key on the user's session with a map, which maps each auth-type (str) to the relevant identity authenticated by that auth-type (mostly str, but for captcha, bool). + If no auth flows have been completed successfully, raises an + InteractiveAuthIncompleteError. To handle this, you can use + synapse.rest.client.v2_alpha._base.interactive_auth_handler as a + decorator. + Args: flows (list): A list of login flows. Each flow is an ordered list of strings representing auth-types. At least one full flow must be completed in order for auth to be successful. + clientdict: The dictionary from the client root level, not the 'auth' key: this method prompts for auth if none is sent. + clientip (str): The IP address of the client. + Returns: - A tuple of (authed, dict, dict, session_id) where authed is true if - the client has successfully completed an auth flow. If it is true - the first dict contains the authenticated credentials of each stage. + defer.Deferred[dict, dict, str]: a deferred tuple of + (creds, params, session_id). - If authed is false, the first dictionary is the server response to - the login request and should be passed back to the client. + 'creds' contains the authenticated credentials of each stage. - In either case, the second dict contains the parameters for this - request (which may have been given only in a previous call). + 'params' contains the parameters for this request (which may + have been given only in a previous call). - session_id is the ID of this session, either passed in by the client - or assigned by the call to check_auth + 'session_id' is the ID of this session, either passed in by the + client or assigned by this call + + Raises: + InteractiveAuthIncompleteError if the client has not yet completed + all the stages in any of the permitted flows. """ authdict = None @@ -135,11 +220,8 @@ class AuthHandler(BaseHandler): clientdict = session['clientdict'] if not authdict: - defer.returnValue( - ( - False, self._auth_dict_for_flows(flows, session), - clientdict, session['id'] - ) + raise InteractiveAuthIncompleteError( + self._auth_dict_for_flows(flows, session), ) if 'creds' not in session: @@ -150,14 +232,12 @@ class AuthHandler(BaseHandler): errordict = {} if 'type' in authdict: login_type = authdict['type'] - if login_type not in self.checkers: - raise LoginError(400, "", Codes.UNRECOGNIZED) try: - result = yield self.checkers[login_type](authdict, clientip) + result = yield self._check_auth_dict(authdict, clientip) if result: creds[login_type] = result self._save_session(session) - except LoginError, e: + except LoginError as e: if login_type == LoginType.EMAIL_IDENTITY: # riot used to have a bug where it would request a new # validation token (thus sending a new email) each time it @@ -166,14 +246,14 @@ class AuthHandler(BaseHandler): # # Grandfather in the old behaviour for now to avoid # breaking old riot deployments. - raise e + raise # this step failed. Merge the error dict into the response # so that the client can have another go. errordict = e.error_dict() for f in flows: - if len(set(f) - set(creds.keys())) == 0: + if len(set(f) - set(creds)) == 0: # it's very useful to know what args are stored, but this can # include the password in the case of registering, so only log # the keys (confusingly, clientdict may contain a password @@ -181,14 +261,16 @@ class AuthHandler(BaseHandler): # and is not sensitive). logger.info( "Auth completed with creds: %r. Client dict has keys: %r", - creds, clientdict.keys() + creds, list(clientdict) ) - defer.returnValue((True, creds, clientdict, session['id'])) + defer.returnValue((creds, clientdict, session['id'])) ret = self._auth_dict_for_flows(flows, session) - ret['completed'] = creds.keys() + ret['completed'] = list(creds) ret.update(errordict) - defer.returnValue((False, ret, clientdict, session['id'])) + raise InteractiveAuthIncompleteError( + ret, + ) @defer.inlineCallbacks def add_oob_auth(self, stagetype, authdict, clientip): @@ -260,16 +342,37 @@ class AuthHandler(BaseHandler): sess = self._get_session_info(session_id) return sess.setdefault('serverdict', {}).get(key, default) - def _check_password_auth(self, authdict, _): - if "user" not in authdict or "password" not in authdict: - raise LoginError(400, "", Codes.MISSING_PARAM) + @defer.inlineCallbacks + def _check_auth_dict(self, authdict, clientip): + """Attempt to validate the auth dict provided by a client - user_id = authdict["user"] - password = authdict["password"] - if not user_id.startswith('@'): - user_id = UserID.create(user_id, self.hs.hostname).to_string() + Args: + authdict (object): auth dict provided by the client + clientip (str): IP address of the client - return self._check_password(user_id, password) + Returns: + Deferred: result of the stage verification. + + Raises: + StoreError if there was a problem accessing the database + SynapseError if there was a problem with the request + LoginError if there was an authentication problem. + """ + login_type = authdict['type'] + checker = self.checkers.get(login_type) + if checker is not None: + res = yield checker(authdict, clientip) + defer.returnValue(res) + + # build a v1-login-style dict out of the authdict and fall back to the + # v1 code + user_id = authdict.get("user") + + if user_id is None: + raise SynapseError(400, "", Codes.MISSING_PARAM) + + (canonical_id, callback) = yield self.validate_login(user_id, authdict) + defer.returnValue(canonical_id) @defer.inlineCallbacks def _check_recaptcha(self, authdict, clientip): @@ -303,7 +406,7 @@ class AuthHandler(BaseHandler): except PartialDownloadError as pde: # Twisted is silly data = pde.response - resp_body = simplejson.loads(data) + resp_body = json.loads(data) if 'success' in resp_body: # Note that we do NOT check the hostname here: we explicitly @@ -324,15 +427,11 @@ class AuthHandler(BaseHandler): def _check_msisdn(self, authdict, _): return self._check_threepid('msisdn', authdict) - @defer.inlineCallbacks def _check_dummy_auth(self, authdict, _): - yield run_on_reactor() - defer.returnValue(True) + return defer.succeed(True) @defer.inlineCallbacks def _check_threepid(self, medium, authdict): - yield run_on_reactor() - if 'threepid_creds' not in authdict: raise LoginError(400, "Missing threepid_creds", Codes.MISSING_PARAM) @@ -398,26 +497,8 @@ class AuthHandler(BaseHandler): return self.sessions[session_id] - def validate_password_login(self, user_id, password): - """ - Authenticates the user with their username and password. - - Used only by the v1 login API. - - Args: - user_id (str): complete @user:id - password (str): Password - Returns: - defer.Deferred: (str) canonical user id - Raises: - StoreError if there was a problem accessing the database - LoginError if there was an authentication problem. - """ - return self._check_password(user_id, password) - @defer.inlineCallbacks - def get_access_token_for_user_id(self, user_id, device_id=None, - initial_display_name=None): + def get_access_token_for_user_id(self, user_id, device_id=None): """ Creates a new access token for the user with the given user ID. @@ -431,13 +512,10 @@ class AuthHandler(BaseHandler): device_id (str|None): the device ID to associate with the tokens. None to leave the tokens unassociated with a device (deprecated: we should always have a device ID) - initial_display_name (str): display name to associate with the - device if it needs re-registering Returns: The access token for the user's session. Raises: StoreError if there was a problem storing the token. - LoginError if there was an authentication problem. """ logger.info("Logging in user %s on device %s", user_id, device_id) access_token = yield self.issue_access_token(user_id, device_id) @@ -447,9 +525,11 @@ class AuthHandler(BaseHandler): # really don't want is active access_tokens without a record of the # device, so we double-check it here. if device_id is not None: - yield self.device_handler.check_device_registered( - user_id, device_id, initial_display_name - ) + try: + yield self.store.get_device(user_id, device_id) + except StoreError: + yield self.store.delete_access_token(access_token) + raise StoreError(400, "Login raced against device deletion") defer.returnValue(access_token) @@ -501,29 +581,115 @@ class AuthHandler(BaseHandler): ) defer.returnValue(result) - @defer.inlineCallbacks - def _check_password(self, user_id, password): - """Authenticate a user against the LDAP and local databases. + def get_supported_login_types(self): + """Get a the login types supported for the /login API - user_id is checked case insensitively against the local database, but - will throw if there are multiple inexact matches. + By default this is just 'm.login.password' (unless password_enabled is + False in the config file), but password auth providers can provide + other login types. + + Returns: + Iterable[str]: login types + """ + return self._supported_login_types + + @defer.inlineCallbacks + def validate_login(self, username, login_submission): + """Authenticates the user for the /login API + + Also used by the user-interactive auth flow to validate + m.login.password auth types. Args: - user_id (str): complete @user:id + username (str): username supplied by the user + login_submission (dict): the whole of the login submission + (including 'type' and other relevant fields) Returns: - (str) the canonical_user_id + Deferred[str, func]: canonical user id, and optional callback + to be called once the access token and device id are issued Raises: - LoginError if login fails + StoreError if there was a problem accessing the database + SynapseError if there was a problem with the request + LoginError if there was an authentication problem. """ + + if username.startswith('@'): + qualified_user_id = username + else: + qualified_user_id = UserID( + username, self.hs.hostname + ).to_string() + + login_type = login_submission.get("type") + known_login_type = False + + # special case to check for "password" for the check_password interface + # for the auth providers + password = login_submission.get("password") + if login_type == LoginType.PASSWORD: + if not self._password_enabled: + raise SynapseError(400, "Password login has been disabled.") + if not password: + raise SynapseError(400, "Missing parameter: password") + for provider in self.password_providers: - is_valid = yield provider.check_password(user_id, password) - if is_valid: - defer.returnValue(user_id) + if (hasattr(provider, "check_password") + and login_type == LoginType.PASSWORD): + known_login_type = True + is_valid = yield provider.check_password( + qualified_user_id, password, + ) + if is_valid: + defer.returnValue((qualified_user_id, None)) - canonical_user_id = yield self._check_local_password(user_id, password) + if (not hasattr(provider, "get_supported_login_types") + or not hasattr(provider, "check_auth")): + # this password provider doesn't understand custom login types + continue - if canonical_user_id: - defer.returnValue(canonical_user_id) + supported_login_types = provider.get_supported_login_types() + if login_type not in supported_login_types: + # this password provider doesn't understand this login type + continue + + known_login_type = True + login_fields = supported_login_types[login_type] + + missing_fields = [] + login_dict = {} + for f in login_fields: + if f not in login_submission: + missing_fields.append(f) + else: + login_dict[f] = login_submission[f] + if missing_fields: + raise SynapseError( + 400, "Missing parameters for login type %s: %s" % ( + login_type, + missing_fields, + ), + ) + + result = yield provider.check_auth( + username, login_type, login_dict, + ) + if result: + if isinstance(result, str): + result = (result, None) + defer.returnValue(result) + + if login_type == LoginType.PASSWORD: + known_login_type = True + + canonical_user_id = yield self._check_local_password( + qualified_user_id, password, + ) + + if canonical_user_id: + defer.returnValue((canonical_user_id, None)) + + if not known_login_type: + raise SynapseError(400, "Unknown login type %s" % login_type) # unknown username or invalid password. We raise a 403 here, but note # that if we're doing user-interactive login, it turns all LoginErrors @@ -549,7 +715,7 @@ class AuthHandler(BaseHandler): if not lookupres: defer.returnValue(None) (user_id, password_hash) = lookupres - result = self.validate_hash(password, password_hash) + result = yield self.validate_hash(password, password_hash) if not result: logger.warn("Failed password login for user %s", user_id) defer.returnValue(None) @@ -573,22 +739,65 @@ class AuthHandler(BaseHandler): raise AuthError(403, "Invalid token", errcode=Codes.FORBIDDEN) @defer.inlineCallbacks - def set_password(self, user_id, newpassword, requester=None): - password_hash = self.hash(newpassword) + def delete_access_token(self, access_token): + """Invalidate a single access token - except_access_token_id = requester.access_token_id if requester else None + Args: + access_token (str): access token to be deleted - try: - yield self.store.user_set_password_hash(user_id, password_hash) - except StoreError as e: - if e.code == 404: - raise SynapseError(404, "Unknown user", Codes.NOT_FOUND) - raise e - yield self.store.user_delete_access_tokens( - user_id, except_access_token_id + Returns: + Deferred + """ + user_info = yield self.auth.get_user_by_access_token(access_token) + yield self.store.delete_access_token(access_token) + + # see if any of our auth providers want to know about this + for provider in self.password_providers: + if hasattr(provider, "on_logged_out"): + yield provider.on_logged_out( + user_id=str(user_info["user"]), + device_id=user_info["device_id"], + access_token=access_token, + ) + + # delete pushers associated with this access token + if user_info["token_id"] is not None: + yield self.hs.get_pusherpool().remove_pushers_by_access_token( + str(user_info["user"]), (user_info["token_id"], ) + ) + + @defer.inlineCallbacks + def delete_access_tokens_for_user(self, user_id, except_token_id=None, + device_id=None): + """Invalidate access tokens belonging to a user + + Args: + user_id (str): ID of user the tokens belong to + except_token_id (str|None): access_token ID which should *not* be + deleted + device_id (str|None): ID of device the tokens are associated with. + If None, tokens associated with any device (or no device) will + be deleted + Returns: + Deferred + """ + tokens_and_devices = yield self.store.user_delete_access_tokens( + user_id, except_token_id=except_token_id, device_id=device_id, ) - yield self.hs.get_pusherpool().remove_pushers_by_user( - user_id, except_access_token_id + + # see if any of our auth providers want to know about this + for provider in self.password_providers: + if hasattr(provider, "on_logged_out"): + for token, token_id, device_id in tokens_and_devices: + yield provider.on_logged_out( + user_id=user_id, + device_id=device_id, + access_token=token, + ) + + # delete pushers associated with the access tokens + yield self.hs.get_pusherpool().remove_pushers_by_access_token( + user_id, (token_id for _, token_id, _ in tokens_and_devices), ) @defer.inlineCallbacks @@ -616,6 +825,15 @@ class AuthHandler(BaseHandler): if medium == 'email': address = address.lower() + identity_handler = self.hs.get_handlers().identity_handler + yield identity_handler.unbind_threepid( + user_id, + { + 'medium': medium, + 'address': address, + }, + ) + ret = yield self.store.user_delete_threepid( user_id, medium, address, ) @@ -634,10 +852,17 @@ class AuthHandler(BaseHandler): password (str): Password to hash. Returns: - Hashed password (str). + Deferred(str): Hashed password. """ - return bcrypt.hashpw(password.encode('utf8') + self.hs.config.password_pepper, - bcrypt.gensalt(self.bcrypt_rounds)) + def _do_hash(): + return bcrypt.hashpw(password.encode('utf8') + self.hs.config.password_pepper, + bcrypt.gensalt(self.bcrypt_rounds)) + + return make_deferred_yieldable( + threads.deferToThreadPool( + self.hs.get_reactor(), self.hs.get_reactor().getThreadPool(), _do_hash + ), + ) def validate_hash(self, password, stored_hash): """Validates that self.hash(password) == stored_hash. @@ -647,20 +872,31 @@ class AuthHandler(BaseHandler): stored_hash (str): Expected hash value. Returns: - Whether self.hash(password) == stored_hash (bool). + Deferred(bool): Whether self.hash(password) == stored_hash. """ + + def _do_validate_hash(): + return bcrypt.checkpw( + password.encode('utf8') + self.hs.config.password_pepper, + stored_hash.encode('utf8') + ) + if stored_hash: - return bcrypt.hashpw(password.encode('utf8') + self.hs.config.password_pepper, - stored_hash.encode('utf8')) == stored_hash + return make_deferred_yieldable( + threads.deferToThreadPool( + self.hs.get_reactor(), + self.hs.get_reactor().getThreadPool(), + _do_validate_hash, + ), + ) else: - return False + return defer.succeed(False) -class MacaroonGeneartor(object): - def __init__(self, hs): - self.clock = hs.get_clock() - self.server_name = hs.config.server_name - self.macaroon_secret_key = hs.config.macaroon_secret_key +@attr.s +class MacaroonGenerator(object): + + hs = attr.ib() def generate_access_token(self, user_id, extra_caveats=None): extra_caveats = extra_caveats or [] @@ -678,7 +914,7 @@ class MacaroonGeneartor(object): def generate_short_term_login_token(self, user_id, duration_in_ms=(2 * 60 * 1000)): macaroon = self._generate_base_macaroon(user_id) macaroon.add_first_party_caveat("type = login") - now = self.clock.time_msec() + now = self.hs.get_clock().time_msec() expiry = now + duration_in_ms macaroon.add_first_party_caveat("time < %d" % (expiry,)) return macaroon.serialize() @@ -690,36 +926,9 @@ class MacaroonGeneartor(object): def _generate_base_macaroon(self, user_id): macaroon = pymacaroons.Macaroon( - location=self.server_name, + location=self.hs.config.server_name, identifier="key", - key=self.macaroon_secret_key) + key=self.hs.config.macaroon_secret_key) macaroon.add_first_party_caveat("gen = 1") macaroon.add_first_party_caveat("user_id = %s" % (user_id,)) return macaroon - - -class _AccountHandler(object): - """A proxy object that gets passed to password auth providers so they - can register new users etc if necessary. - """ - def __init__(self, hs, check_user_exists): - self.hs = hs - - self._check_user_exists = check_user_exists - - def check_user_exists(self, user_id): - """Check if user exissts. - - Returns: - Deferred(bool) - """ - return self._check_user_exists(user_id) - - def register(self, localpart): - """Registers a new user with given localpart - - Returns: - Deferred: a 2-tuple of (user_id, access_token) - """ - reg = self.hs.get_handlers().registration_handler - return reg.register(localpart=localpart) diff --git a/synapse/handlers/deactivate_account.py b/synapse/handlers/deactivate_account.py new file mode 100644 index 000000000..b3c5a9ee6 --- /dev/null +++ b/synapse/handlers/deactivate_account.py @@ -0,0 +1,163 @@ +# -*- coding: utf-8 -*- +# Copyright 2017, 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import logging + +from twisted.internet import defer + +from synapse.api.errors import SynapseError +from synapse.types import UserID, create_requester +from synapse.util.logcontext import run_in_background + +from ._base import BaseHandler + +logger = logging.getLogger(__name__) + + +class DeactivateAccountHandler(BaseHandler): + """Handler which deals with deactivating user accounts.""" + def __init__(self, hs): + super(DeactivateAccountHandler, self).__init__(hs) + self._auth_handler = hs.get_auth_handler() + self._device_handler = hs.get_device_handler() + self._room_member_handler = hs.get_room_member_handler() + self._identity_handler = hs.get_handlers().identity_handler + self.user_directory_handler = hs.get_user_directory_handler() + + # Flag that indicates whether the process to part users from rooms is running + self._user_parter_running = False + + # Start the user parter loop so it can resume parting users from rooms where + # it left off (if it has work left to do). + hs.get_reactor().callWhenRunning(self._start_user_parting) + + @defer.inlineCallbacks + def deactivate_account(self, user_id, erase_data): + """Deactivate a user's account + + Args: + user_id (str): ID of user to be deactivated + erase_data (bool): whether to GDPR-erase the user's data + + Returns: + Deferred + """ + # FIXME: Theoretically there is a race here wherein user resets + # password using threepid. + + # delete threepids first. We remove these from the IS so if this fails, + # leave the user still active so they can try again. + # Ideally we would prevent password resets and then do this in the + # background thread. + threepids = yield self.store.user_get_threepids(user_id) + for threepid in threepids: + try: + yield self._identity_handler.unbind_threepid( + user_id, + { + 'medium': threepid['medium'], + 'address': threepid['address'], + }, + ) + except Exception: + # Do we want this to be a fatal error or should we carry on? + logger.exception("Failed to remove threepid from ID server") + raise SynapseError(400, "Failed to remove threepid from ID server") + yield self.store.user_delete_threepid( + user_id, threepid['medium'], threepid['address'], + ) + + # delete any devices belonging to the user, which will also + # delete corresponding access tokens. + yield self._device_handler.delete_all_devices_for_user(user_id) + # then delete any remaining access tokens which weren't associated with + # a device. + yield self._auth_handler.delete_access_tokens_for_user(user_id) + + yield self.store.user_set_password_hash(user_id, None) + + # Add the user to a table of users pending deactivation (ie. + # removal from all the rooms they're a member of) + yield self.store.add_user_pending_deactivation(user_id) + + # delete from user directory + yield self.user_directory_handler.handle_user_deactivated(user_id) + + # Mark the user as erased, if they asked for that + if erase_data: + logger.info("Marking %s as erased", user_id) + yield self.store.mark_user_erased(user_id) + + # Now start the process that goes through that list and + # parts users from rooms (if it isn't already running) + self._start_user_parting() + + def _start_user_parting(self): + """ + Start the process that goes through the table of users + pending deactivation, if it isn't already running. + + Returns: + None + """ + if not self._user_parter_running: + run_in_background(self._user_parter_loop) + + @defer.inlineCallbacks + def _user_parter_loop(self): + """Loop that parts deactivated users from rooms + + Returns: + None + """ + self._user_parter_running = True + logger.info("Starting user parter") + try: + while True: + user_id = yield self.store.get_user_pending_deactivation() + if user_id is None: + break + logger.info("User parter parting %r", user_id) + yield self._part_user(user_id) + yield self.store.del_user_pending_deactivation(user_id) + logger.info("User parter finished parting %r", user_id) + logger.info("User parter finished: stopping") + finally: + self._user_parter_running = False + + @defer.inlineCallbacks + def _part_user(self, user_id): + """Causes the given user_id to leave all the rooms they're joined to + + Returns: + None + """ + user = UserID.from_string(user_id) + + rooms_for_user = yield self.store.get_rooms_for_user(user_id) + for room_id in rooms_for_user: + logger.info("User parter parting %r from %r", user_id, room_id) + try: + yield self._room_member_handler.update_membership( + create_requester(user), + user, + room_id, + "leave", + ratelimit=False, + ) + except Exception: + logger.exception( + "Failed to part user %r from room %r: ignoring and continuing", + user_id, room_id, + ) diff --git a/synapse/handlers/device.py b/synapse/handlers/device.py index ed60d494f..2d44f15da 100644 --- a/synapse/handlers/device.py +++ b/synapse/handlers/device.py @@ -12,18 +12,23 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. +import logging + +from six import iteritems, itervalues + +from twisted.internet import defer + from synapse.api import errors from synapse.api.constants import EventTypes +from synapse.api.errors import FederationDeniedError +from synapse.types import RoomStreamToken, get_domain_from_id from synapse.util import stringutils from synapse.util.async import Linearizer from synapse.util.caches.expiringcache import ExpiringCache -from synapse.util.retryutils import NotRetryingDestination from synapse.util.metrics import measure_func -from synapse.types import get_domain_from_id, RoomStreamToken -from twisted.internet import defer -from ._base import BaseHandler +from synapse.util.retryutils import NotRetryingDestination -import logging +from ._base import BaseHandler logger = logging.getLogger(__name__) @@ -34,15 +39,17 @@ class DeviceHandler(BaseHandler): self.hs = hs self.state = hs.get_state_handler() + self._auth_handler = hs.get_auth_handler() self.federation_sender = hs.get_federation_sender() - self.federation = hs.get_replication_layer() self._edu_updater = DeviceListEduUpdater(hs, self) - self.federation.register_edu_handler( + federation_registry = hs.get_federation_registry() + + federation_registry.register_edu_handler( "m.device_list_update", self._edu_updater.incoming_device_list_update, ) - self.federation.register_query_handler( + federation_registry.register_query_handler( "user_devices", self.on_federation_query_user_devices, ) @@ -109,7 +116,7 @@ class DeviceHandler(BaseHandler): user_id, device_id=None ) - devices = device_map.values() + devices = list(device_map.values()) for device in devices: _update_device_from_client_ips(device, ips) @@ -152,16 +159,15 @@ class DeviceHandler(BaseHandler): try: yield self.store.delete_device(user_id, device_id) - except errors.StoreError, e: + except errors.StoreError as e: if e.code == 404: # no match pass else: raise - yield self.store.user_delete_access_tokens( + yield self._auth_handler.delete_access_tokens_for_user( user_id, device_id=device_id, - delete_refresh_tokens=True, ) yield self.store.delete_e2e_keys_by_device( @@ -170,13 +176,31 @@ class DeviceHandler(BaseHandler): yield self.notify_device_update(user_id, [device_id]) + @defer.inlineCallbacks + def delete_all_devices_for_user(self, user_id, except_device_id=None): + """Delete all of the user's devices + + Args: + user_id (str): + except_device_id (str|None): optional device id which should not + be deleted + + Returns: + defer.Deferred: + """ + device_map = yield self.store.get_devices_by_user(user_id) + device_ids = list(device_map) + if except_device_id is not None: + device_ids = [d for d in device_ids if d != except_device_id] + yield self.delete_devices(user_id, device_ids) + @defer.inlineCallbacks def delete_devices(self, user_id, device_ids): """ Delete several devices Args: user_id (str): - device_ids (str): The list of device IDs to delete + device_ids (List[str]): The list of device IDs to delete Returns: defer.Deferred: @@ -184,7 +208,7 @@ class DeviceHandler(BaseHandler): try: yield self.store.delete_devices(user_id, device_ids) - except errors.StoreError, e: + except errors.StoreError as e: if e.code == 404: # no match pass @@ -194,9 +218,8 @@ class DeviceHandler(BaseHandler): # Delete access tokens and e2e keys for each device. Not optimised as it is not # considered as part of a critical path. for device_id in device_ids: - yield self.store.user_delete_access_tokens( + yield self._auth_handler.delete_access_tokens_for_user( user_id, device_id=device_id, - delete_refresh_tokens=True, ) yield self.store.delete_e2e_keys_by_device( user_id=user_id, device_id=device_id @@ -224,7 +247,7 @@ class DeviceHandler(BaseHandler): new_display_name=content.get("display_name") ) yield self.notify_device_update(user_id, [device_id]) - except errors.StoreError, e: + except errors.StoreError as e: if e.code == 404: raise errors.NotFoundError() else: @@ -270,6 +293,8 @@ class DeviceHandler(BaseHandler): user_id (str) from_token (StreamToken) """ + now_token = yield self.hs.get_event_sources().get_current_token() + room_ids = yield self.store.get_rooms_for_user(user_id) # First we check if any devices have changed @@ -280,11 +305,30 @@ class DeviceHandler(BaseHandler): # Then work out if any users have since joined rooms_changed = self.store.get_rooms_that_changed(room_ids, from_token.room_key) + member_events = yield self.store.get_membership_changes_for_user( + user_id, from_token.room_key, now_token.room_key + ) + rooms_changed.update(event.room_id for event in member_events) + stream_ordering = RoomStreamToken.parse_stream_token( - from_token.room_key).stream + from_token.room_key + ).stream possibly_changed = set(changed) + possibly_left = set() for room_id in rooms_changed: + current_state_ids = yield self.store.get_current_state_ids(room_id) + + # The user may have left the room + # TODO: Check if they actually did or if we were just invited. + if room_id not in room_ids: + for key, event_id in iteritems(current_state_ids): + etype, state_key = key + if etype != EventTypes.Member: + continue + possibly_left.add(state_key) + continue + # Fetch the current state at the time. try: event_ids = yield self.store.get_forward_extremeties_for_room( @@ -295,44 +339,69 @@ class DeviceHandler(BaseHandler): # ordering: treat it the same as a new room event_ids = [] - current_state_ids = yield self.store.get_current_state_ids(room_id) - # special-case for an empty prev state: include all members # in the changed list if not event_ids: - for key, event_id in current_state_ids.iteritems(): + for key, event_id in iteritems(current_state_ids): etype, state_key = key if etype != EventTypes.Member: continue possibly_changed.add(state_key) continue + current_member_id = current_state_ids.get((EventTypes.Member, user_id)) + if not current_member_id: + continue + # mapping from event_id -> state_dict prev_state_ids = yield self.store.get_state_ids_for_events(event_ids) + # Check if we've joined the room? If so we just blindly add all the users to + # the "possibly changed" users. + for state_dict in itervalues(prev_state_ids): + member_event = state_dict.get((EventTypes.Member, user_id), None) + if not member_event or member_event != current_member_id: + for key, event_id in iteritems(current_state_ids): + etype, state_key = key + if etype != EventTypes.Member: + continue + possibly_changed.add(state_key) + break + # If there has been any change in membership, include them in the # possibly changed list. We'll check if they are joined below, # and we're not toooo worried about spuriously adding users. - for key, event_id in current_state_ids.iteritems(): + for key, event_id in iteritems(current_state_ids): etype, state_key = key if etype != EventTypes.Member: continue # check if this member has changed since any of the extremities # at the stream_ordering, and add them to the list if so. - for state_dict in prev_state_ids.values(): + for state_dict in itervalues(prev_state_ids): prev_event_id = state_dict.get(key, None) if not prev_event_id or prev_event_id != event_id: - possibly_changed.add(state_key) + if state_key != user_id: + possibly_changed.add(state_key) break - users_who_share_room = yield self.store.get_users_who_share_room_with_user( - user_id - ) + if possibly_changed or possibly_left: + users_who_share_room = yield self.store.get_users_who_share_room_with_user( + user_id + ) - # Take the intersection of the users whose devices may have changed - # and those that actually still share a room with the user - defer.returnValue(users_who_share_room & possibly_changed) + # Take the intersection of the users whose devices may have changed + # and those that actually still share a room with the user + possibly_joined = possibly_changed & users_who_share_room + possibly_left = (possibly_changed | possibly_left) - users_who_share_room + else: + possibly_joined = [] + possibly_left = [] + + defer.returnValue({ + "changed": list(possibly_joined), + "left": list(possibly_left), + }) @defer.inlineCallbacks def on_federation_query_user_devices(self, user_id): @@ -366,7 +435,7 @@ class DeviceListEduUpdater(object): def __init__(self, hs, device_handler): self.store = hs.get_datastore() - self.federation = hs.get_replication_layer() + self.federation = hs.get_federation_client() self.clock = hs.get_clock() self.device_handler = device_handler @@ -450,6 +519,9 @@ class DeviceListEduUpdater(object): # This makes it more likely that the device lists will # eventually become consistent. return + except FederationDeniedError as e: + logger.info(e) + return except Exception: # TODO: Remember that we are now out of sync and try again # later @@ -467,7 +539,7 @@ class DeviceListEduUpdater(object): yield self.device_handler.notify_device_update(user_id, device_ids) else: # Simply update the single device, since we know that is the only - # change (becuase of the single prev_id matching the current cache) + # change (because of the single prev_id matching the current cache) for device_id, stream_id, prev_ids, content in pending_updates: yield self.store.update_remote_device_list_cache_entry( user_id, device_id, content, stream_id, diff --git a/synapse/handlers/devicemessage.py b/synapse/handlers/devicemessage.py index f7fad15c6..2e2e5261d 100644 --- a/synapse/handlers/devicemessage.py +++ b/synapse/handlers/devicemessage.py @@ -17,10 +17,10 @@ import logging from twisted.internet import defer -from synapse.types import get_domain_from_id +from synapse.api.errors import SynapseError +from synapse.types import UserID, get_domain_from_id from synapse.util.stringutils import random_string - logger = logging.getLogger(__name__) @@ -33,10 +33,10 @@ class DeviceMessageHandler(object): """ self.store = hs.get_datastore() self.notifier = hs.get_notifier() - self.is_mine_id = hs.is_mine_id + self.is_mine = hs.is_mine self.federation = hs.get_federation_sender() - hs.get_replication_layer().register_edu_handler( + hs.get_federation_registry().register_edu_handler( "m.direct_to_device", self.on_direct_to_device_edu ) @@ -52,6 +52,12 @@ class DeviceMessageHandler(object): message_type = content["type"] message_id = content["message_id"] for user_id, by_device in content["messages"].items(): + # we use UserID.from_string to catch invalid user ids + if not self.is_mine(UserID.from_string(user_id)): + logger.warning("Request for keys for non-local user %s", + user_id) + raise SynapseError(400, "Not a user here") + messages_by_device = { device_id: { "content": message_content, @@ -77,7 +83,8 @@ class DeviceMessageHandler(object): local_messages = {} remote_messages = {} for user_id, by_device in messages.items(): - if self.is_mine_id(user_id): + # we use UserID.from_string to catch invalid user ids + if self.is_mine(UserID.from_string(user_id)): messages_by_device = { device_id: { "content": message_content, diff --git a/synapse/handlers/directory.py b/synapse/handlers/directory.py index 943554ce9..ef866da1b 100644 --- a/synapse/handlers/directory.py +++ b/synapse/handlers/directory.py @@ -14,16 +14,17 @@ # limitations under the License. -from twisted.internet import defer -from ._base import BaseHandler - -from synapse.api.errors import SynapseError, Codes, CodeMessageException, AuthError -from synapse.api.constants import EventTypes -from synapse.types import RoomAlias, UserID, get_domain_from_id - import logging import string +from twisted.internet import defer + +from synapse.api.constants import EventTypes +from synapse.api.errors import AuthError, CodeMessageException, Codes, SynapseError +from synapse.types import RoomAlias, UserID, get_domain_from_id + +from ._base import BaseHandler + logger = logging.getLogger(__name__) @@ -34,12 +35,15 @@ class DirectoryHandler(BaseHandler): self.state = hs.get_state_handler() self.appservice_handler = hs.get_application_service_handler() + self.event_creation_handler = hs.get_event_creation_handler() - self.federation = hs.get_replication_layer() - self.federation.register_query_handler( + self.federation = hs.get_federation_client() + hs.get_federation_registry().register_query_handler( "directory", self.on_directory_query ) + self.spam_checker = hs.get_spam_checker() + @defer.inlineCallbacks def _create_association(self, room_alias, room_id, servers=None, creator=None): # general association creation for both human users and app services @@ -73,6 +77,11 @@ class DirectoryHandler(BaseHandler): # association creation for human users # TODO(erikj): Do user auth. + if not self.spam_checker.user_may_create_room_alias(user_id, room_alias): + raise SynapseError( + 403, "This user is not permitted to create this alias", + ) + can_create = yield self.can_modify_alias( room_alias, user_id=user_id @@ -242,8 +251,7 @@ class DirectoryHandler(BaseHandler): def send_room_alias_update_event(self, requester, user_id, room_id): aliases = yield self.store.get_aliases_for_room(room_id) - msg_handler = self.hs.get_handlers().message_handler - yield msg_handler.create_and_send_nonmember_event( + yield self.event_creation_handler.create_and_send_nonmember_event( requester, { "type": EventTypes.Aliases, @@ -265,8 +273,7 @@ class DirectoryHandler(BaseHandler): if not alias_event or alias_event.content.get("alias", "") != alias_str: return - msg_handler = self.hs.get_handlers().message_handler - yield msg_handler.create_and_send_nonmember_event( + yield self.event_creation_handler.create_and_send_nonmember_event( requester, { "type": EventTypes.CanonicalAlias, @@ -327,6 +334,14 @@ class DirectoryHandler(BaseHandler): room_id (str) visibility (str): "public" or "private" """ + if not self.spam_checker.user_may_publish_room( + requester.user.to_string(), room_id + ): + raise AuthError( + 403, + "This user is not permitted to publish rooms to the room list" + ) + if requester.is_guest: raise AuthError(403, "Guests cannot edit the published room list") diff --git a/synapse/handlers/e2e_keys.py b/synapse/handlers/e2e_keys.py index 668a90e49..5816bf8b4 100644 --- a/synapse/handlers/e2e_keys.py +++ b/synapse/handlers/e2e_keys.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -13,15 +14,17 @@ # See the License for the specific language governing permissions and # limitations under the License. -import ujson as json import logging -from canonicaljson import encode_canonical_json +from six import iteritems + +from canonicaljson import encode_canonical_json, json + from twisted.internet import defer -from synapse.api.errors import SynapseError, CodeMessageException -from synapse.types import get_domain_from_id -from synapse.util.logcontext import preserve_fn, make_deferred_yieldable +from synapse.api.errors import CodeMessageException, FederationDeniedError, SynapseError +from synapse.types import UserID, get_domain_from_id +from synapse.util.logcontext import make_deferred_yieldable, run_in_background from synapse.util.retryutils import NotRetryingDestination logger = logging.getLogger(__name__) @@ -30,15 +33,15 @@ logger = logging.getLogger(__name__) class E2eKeysHandler(object): def __init__(self, hs): self.store = hs.get_datastore() - self.federation = hs.get_replication_layer() + self.federation = hs.get_federation_client() self.device_handler = hs.get_device_handler() - self.is_mine_id = hs.is_mine_id + self.is_mine = hs.is_mine self.clock = hs.get_clock() # doesn't really work as part of the generic query API, because the # query request requires an object POST, but we abuse the # "query handler" interface. - self.federation.register_query_handler( + hs.get_federation_registry().register_query_handler( "client_keys", self.on_federation_query_client_keys ) @@ -70,12 +73,13 @@ class E2eKeysHandler(object): remote_queries = {} for user_id, device_ids in device_keys_query.items(): - if self.is_mine_id(user_id): + # we use UserID.from_string to catch invalid user ids + if self.is_mine(UserID.from_string(user_id)): local_query[user_id] = device_ids else: remote_queries[user_id] = device_ids - # Firt get local devices. + # First get local devices. failures = {} results = {} if local_query: @@ -88,7 +92,7 @@ class E2eKeysHandler(object): remote_queries_not_in_cache = {} if remote_queries: query_list = [] - for user_id, device_ids in remote_queries.iteritems(): + for user_id, device_ids in iteritems(remote_queries): if device_ids: query_list.extend((user_id, device_id) for device_id in device_ids) else: @@ -99,9 +103,9 @@ class E2eKeysHandler(object): query_list ) ) - for user_id, devices in remote_results.iteritems(): + for user_id, devices in iteritems(remote_results): user_devices = results.setdefault(user_id, {}) - for device_id, device in devices.iteritems(): + for device_id, device in iteritems(devices): keys = device.get("keys", None) device_display_name = device.get("device_display_name", None) if keys: @@ -131,24 +135,13 @@ class E2eKeysHandler(object): if user_id in destination_query: results[user_id] = keys - except CodeMessageException as e: - failures[destination] = { - "status": e.code, "message": e.message - } - except NotRetryingDestination as e: - failures[destination] = { - "status": 503, "message": "Not ready for retry", - } except Exception as e: - # include ConnectionRefused and other errors - failures[destination] = { - "status": 503, "message": e.message - } + failures[destination] = _exception_to_failure(e) yield make_deferred_yieldable(defer.gatherResults([ - preserve_fn(do_remote_query)(destination) + run_in_background(do_remote_query, destination) for destination in remote_queries_not_in_cache - ])) + ], consumeErrors=True)) defer.returnValue({ "device_keys": results, "failures": failures, @@ -170,7 +163,8 @@ class E2eKeysHandler(object): result_dict = {} for user_id, device_ids in query.items(): - if not self.is_mine_id(user_id): + # we use UserID.from_string to catch invalid user ids + if not self.is_mine(UserID.from_string(user_id)): logger.warning("Request for keys for non-local user %s", user_id) raise SynapseError(400, "Not a user here") @@ -213,7 +207,8 @@ class E2eKeysHandler(object): remote_queries = {} for user_id, device_keys in query.get("one_time_keys", {}).items(): - if self.is_mine_id(user_id): + # we use UserID.from_string to catch invalid user ids + if self.is_mine(UserID.from_string(user_id)): for device_id, algorithm in device_keys.items(): local_query.append((user_id, device_id, algorithm)) else: @@ -243,32 +238,21 @@ class E2eKeysHandler(object): for user_id, keys in remote_result["one_time_keys"].items(): if user_id in device_keys: json_result[user_id] = keys - except CodeMessageException as e: - failures[destination] = { - "status": e.code, "message": e.message - } - except NotRetryingDestination as e: - failures[destination] = { - "status": 503, "message": "Not ready for retry", - } except Exception as e: - # include ConnectionRefused and other errors - failures[destination] = { - "status": 503, "message": e.message - } + failures[destination] = _exception_to_failure(e) yield make_deferred_yieldable(defer.gatherResults([ - preserve_fn(claim_client_keys)(destination) + run_in_background(claim_client_keys, destination) for destination in remote_queries - ])) + ], consumeErrors=True)) logger.info( "Claimed one-time-keys: %s", ",".join(( "%s for %s:%s" % (key_id, user_id, device_id) - for user_id, user_keys in json_result.iteritems() - for device_id, device_keys in user_keys.iteritems() - for key_id, _ in device_keys.iteritems() + for user_id, user_keys in iteritems(json_result) + for device_id, device_keys in iteritems(user_keys) + for key_id, _ in iteritems(device_keys) )), ) @@ -353,6 +337,31 @@ class E2eKeysHandler(object): ) +def _exception_to_failure(e): + if isinstance(e, CodeMessageException): + return { + "status": e.code, "message": e.message, + } + + if isinstance(e, NotRetryingDestination): + return { + "status": 503, "message": "Not ready for retry", + } + + if isinstance(e, FederationDeniedError): + return { + "status": 403, "message": "Federation Denied", + } + + # include ConnectionRefused and other errors + # + # Note that some Exceptions (notably twisted's ResponseFailed etc) don't + # give a string for e.message, which json then fails to serialize. + return { + "status": 503, "message": str(e.message), + } + + def _one_time_keys_match(old_key_json, new_key): old_key = json.loads(old_key_json) diff --git a/synapse/handlers/events.py b/synapse/handlers/events.py index d3685fb12..c3f2d7fef 100644 --- a/synapse/handlers/events.py +++ b/synapse/handlers/events.py @@ -13,19 +13,18 @@ # See the License for the specific language governing permissions and # limitations under the License. -from twisted.internet import defer - -from synapse.util.logutils import log_function -from synapse.types import UserID -from synapse.events.utils import serialize_event -from synapse.api.constants import Membership, EventTypes -from synapse.events import EventBase - -from ._base import BaseHandler - import logging import random +from twisted.internet import defer + +from synapse.api.constants import EventTypes, Membership +from synapse.events import EventBase +from synapse.events.utils import serialize_event +from synapse.types import UserID +from synapse.util.logutils import log_function + +from ._base import BaseHandler logger = logging.getLogger(__name__) @@ -48,6 +47,7 @@ class EventStreamHandler(BaseHandler): self.notifier = hs.get_notifier() self.state = hs.get_state_handler() + self._server_notices_sender = hs.get_server_notices_sender() @defer.inlineCallbacks @log_function @@ -58,6 +58,10 @@ class EventStreamHandler(BaseHandler): If `only_keys` is not None, events from keys will be sent down. """ + + # send any outstanding server notices to the user. + yield self._server_notices_sender.on_user_syncing(auth_user_id) + auth_user = UserID.from_string(auth_user_id) presence_handler = self.hs.get_presence_handler() diff --git a/synapse/handlers/federation.py b/synapse/handlers/federation.py index 483cb8eac..145c1a21d 100644 --- a/synapse/handlers/federation.py +++ b/synapse/handlers/federation.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2014-2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -14,41 +15,46 @@ # limitations under the License. """Contains handlers for federation events.""" -import synapse.util.logcontext + +import itertools +import logging +import sys + +import six +from six import iteritems, itervalues +from six.moves import http_client, zip + from signedjson.key import decode_verify_key_bytes from signedjson.sign import verify_signed_json from unpaddedbase64 import decode_base64 -from ._base import BaseHandler - -from synapse.api.errors import ( - AuthError, FederationError, StoreError, CodeMessageException, SynapseError, -) -from synapse.api.constants import EventTypes, Membership, RejectedReason -from synapse.events.validator import EventValidator -from synapse.util import unwrapFirstError -from synapse.util.logcontext import ( - preserve_fn, preserve_context_over_deferred -) -from synapse.util.metrics import measure_func -from synapse.util.logutils import log_function -from synapse.util.async import run_on_reactor, Linearizer -from synapse.util.frozenutils import unfreeze -from synapse.crypto.event_signing import ( - compute_event_signature, add_hashes_and_signatures, -) -from synapse.types import UserID, get_domain_from_id - -from synapse.events.utils import prune_event - -from synapse.util.retryutils import NotRetryingDestination - -from synapse.util.distributor import user_joined_room - from twisted.internet import defer -import itertools -import logging +from synapse.api.constants import EventTypes, Membership, RejectedReason +from synapse.api.errors import ( + AuthError, + CodeMessageException, + FederationDeniedError, + FederationError, + StoreError, + SynapseError, +) +from synapse.crypto.event_signing import ( + add_hashes_and_signatures, + compute_event_signature, +) +from synapse.events.validator import EventValidator +from synapse.state import resolve_events_with_factory +from synapse.types import UserID, get_domain_from_id +from synapse.util import logcontext, unwrapFirstError +from synapse.util.async import Linearizer +from synapse.util.distributor import user_joined_room +from synapse.util.frozenutils import unfreeze +from synapse.util.logutils import log_function +from synapse.util.retryutils import NotRetryingDestination +from synapse.visibility import filter_events_for_server + +from ._base import BaseHandler logger = logging.getLogger(__name__) @@ -70,14 +76,16 @@ class FederationHandler(BaseHandler): self.hs = hs self.store = hs.get_datastore() - self.replication_layer = hs.get_replication_layer() + self.replication_layer = hs.get_federation_client() self.state_handler = hs.get_state_handler() self.server_name = hs.hostname self.keyring = hs.get_keyring() self.action_generator = hs.get_action_generator() self.is_mine_id = hs.is_mine_id - - self.replication_layer.set_handler(self) + self.pusher_pool = hs.get_pusherpool() + self.spam_checker = hs.get_spam_checker() + self.event_creation_handler = hs.get_event_creation_handler() + self._server_notices_mxid = hs.config.server_notices_mxid # When joining a room we need to queue any events for that room up self.room_queues = {} @@ -85,7 +93,9 @@ class FederationHandler(BaseHandler): @defer.inlineCallbacks @log_function - def on_receive_pdu(self, origin, pdu, get_missing=True): + def on_receive_pdu( + self, origin, pdu, get_missing=True, sent_to_us_directly=False, + ): """ Process a PDU received via a federation /send/ transaction, or via backfill of missing prev_events @@ -99,8 +109,10 @@ class FederationHandler(BaseHandler): """ # We reprocess pdus when we have seen them only as outliers - existing = yield self.get_persisted_pdu( - origin, pdu.event_id, do_auth=False + existing = yield self.store.get_event( + pdu.event_id, + allow_none=True, + allow_rejected=True, ) # FIXME: Currently we fetch an event again when we already have it @@ -116,6 +128,19 @@ class FederationHandler(BaseHandler): logger.debug("Already seen pdu %s", pdu.event_id) return + # do some initial sanity-checking of the event. In particular, make + # sure it doesn't have hundreds of prev_events or auth_events, which + # could cause a huge state resolution or cascade of event fetches. + try: + self._sanity_check_event(pdu) + except SynapseError as err: + raise FederationError( + "ERROR", + err.code, + err.msg, + affected=pdu.event_id, + ) + # If we are currently in the process of joining this room, then we # queue up events for later processing. if pdu.room_id in self.room_queues: @@ -124,15 +149,30 @@ class FederationHandler(BaseHandler): self.room_queues[pdu.room_id].append((pdu, origin)) return - state = None - - auth_chain = [] - - have_seen = yield self.store.have_events( - [ev for ev, _ in pdu.prev_events] + # If we're no longer in the room just ditch the event entirely. This + # is probably an old server that has come back and thinks we're still + # in the room (or we've been rejoined to the room by a state reset). + # + # If we were never in the room then maybe our database got vaped and + # we should check if we *are* in fact in the room. If we are then we + # can magically rejoin the room. + is_in_room = yield self.auth.check_host_in_room( + pdu.room_id, + self.server_name ) + if not is_in_room: + was_in_room = yield self.store.was_host_joined( + pdu.room_id, self.server_name, + ) + if was_in_room: + logger.info( + "Ignoring PDU %s for room %s from %s as we've left the room!", + pdu.event_id, pdu.room_id, origin, + ) + defer.returnValue(None) - fetch_state = False + state = None + auth_chain = [] # Get missing pdus if necessary. if not pdu.internal_metadata.is_outlier(): @@ -147,7 +187,7 @@ class FederationHandler(BaseHandler): ) prevs = {e_id for e_id, _ in pdu.prev_events} - seen = set(have_seen.keys()) + seen = yield self.store.have_seen_events(prevs) if min_depth and pdu.depth < min_depth: # This is so that we don't notify the user about this @@ -175,8 +215,7 @@ class FederationHandler(BaseHandler): # Update the set of things we've seen after trying to # fetch the missing stuff - have_seen = yield self.store.have_events(prevs) - seen = set(have_seen.iterkeys()) + seen = yield self.store.have_seen_events(prevs) if not prevs - seen: logger.info( @@ -189,26 +228,60 @@ class FederationHandler(BaseHandler): list(prevs - seen)[:5], ) - if prevs - seen: - logger.info( - "Still missing %d events for room %r: %r...", - len(prevs - seen), pdu.room_id, list(prevs - seen)[:5] + if sent_to_us_directly and prevs - seen: + # If they have sent it to us directly, and the server + # isn't telling us about the auth events that it's + # made a message referencing, we explode + raise FederationError( + "ERROR", + 403, + ( + "Your server isn't divulging details about prev_events " + "referenced in this event." + ), + affected=pdu.event_id, ) - fetch_state = True + elif prevs - seen: + # Calculate the state of the previous events, and + # de-conflict them to find the current state. + state_groups = [] + auth_chains = set() + try: + # Get the state of the events we know about + ours = yield self.store.get_state_groups(pdu.room_id, list(seen)) + state_groups.append(ours) - if fetch_state: - # We need to get the state at this event, since we haven't - # processed all the prev events. - logger.debug( - "_handle_new_pdu getting state for %s", - pdu.room_id - ) - try: - state, auth_chain = yield self.replication_layer.get_state_for_room( - origin, pdu.room_id, pdu.event_id, - ) - except: - logger.exception("Failed to get state for event: %s", pdu.event_id) + # Ask the remote server for the states we don't + # know about + for p in prevs - seen: + state, got_auth_chain = ( + yield self.replication_layer.get_state_for_room( + origin, pdu.room_id, p + ) + ) + auth_chains.update(got_auth_chain) + state_group = {(x.type, x.state_key): x.event_id for x in state} + state_groups.append(state_group) + + # Resolve any conflicting state + def fetch(ev_ids): + return self.store.get_events( + ev_ids, get_prev_content=False, check_redacted=False + ) + + state_map = yield resolve_events_with_factory( + state_groups, {pdu.event_id: pdu}, fetch + ) + + state = (yield self.store.get_events(state_map.values())).values() + auth_chain = list(auth_chains) + except Exception: + raise FederationError( + "ERROR", + 403, + "We can't get valid state history.", + affected=pdu.event_id, + ) yield self._process_received_pdu( origin, @@ -227,8 +300,7 @@ class FederationHandler(BaseHandler): min_depth (int): Minimum depth of events to return. """ # We recalculate seen, since it may have changed. - have_seen = yield self.store.have_events(prevs) - seen = set(have_seen.keys()) + seen = yield self.store.have_seen_events(prevs) if not prevs - seen: return @@ -287,11 +359,17 @@ class FederationHandler(BaseHandler): for e in missing_events: logger.info("Handling found event %s", e.event_id) - yield self.on_receive_pdu( - origin, - e, - get_missing=False - ) + try: + yield self.on_receive_pdu( + origin, + e, + get_missing=False + ) + except FederationError as e: + if e.code == 403: + logger.warn("Event %s failed history check.") + else: + raise @log_function @defer.inlineCallbacks @@ -340,9 +418,7 @@ class FederationHandler(BaseHandler): if auth_chain: event_ids |= {e.event_id for e in auth_chain} - seen_ids = set( - (yield self.store.have_events(event_ids)).keys() - ) + seen_ids = yield self.store.have_seen_events(event_ids) if state and auth_chain is not None: # If we have any state or auth_chain given to us by the replication @@ -410,7 +486,10 @@ class FederationHandler(BaseHandler): # joined the room. Don't bother if the user is just # changing their profile info. newly_joined = True - prev_state_id = context.prev_state_ids.get( + + prev_state_ids = yield context.get_prev_state_ids(self.store) + + prev_state_id = prev_state_ids.get( (event.type, event.state_key) ) if prev_state_id: @@ -424,91 +503,21 @@ class FederationHandler(BaseHandler): user = UserID.from_string(event.state_key) yield user_joined_room(self.distributor, user, event.room_id) - @measure_func("_filter_events_for_server") - @defer.inlineCallbacks - def _filter_events_for_server(self, server_name, room_id, events): - event_to_state_ids = yield self.store.get_state_ids_for_events( - frozenset(e.event_id for e in events), - types=( - (EventTypes.RoomHistoryVisibility, ""), - (EventTypes.Member, None), - ) - ) - - # We only want to pull out member events that correspond to the - # server's domain. - - def check_match(id): - try: - return server_name == get_domain_from_id(id) - except: - return False - - # Parses mapping `event_id -> (type, state_key) -> state event_id` - # to get all state ids that we're interested in. - event_map = yield self.store.get_events([ - e_id - for key_to_eid in event_to_state_ids.values() - for key, e_id in key_to_eid.items() - if key[0] != EventTypes.Member or check_match(key[1]) - ]) - - event_to_state = { - e_id: { - key: event_map[inner_e_id] - for key, inner_e_id in key_to_eid.items() - if inner_e_id in event_map - } - for e_id, key_to_eid in event_to_state_ids.items() - } - - def redact_disallowed(event, state): - if not state: - return event - - history = state.get((EventTypes.RoomHistoryVisibility, ''), None) - if history: - visibility = history.content.get("history_visibility", "shared") - if visibility in ["invited", "joined"]: - # We now loop through all state events looking for - # membership states for the requesting server to determine - # if the server is either in the room or has been invited - # into the room. - for ev in state.values(): - if ev.type != EventTypes.Member: - continue - try: - domain = get_domain_from_id(ev.state_key) - except: - continue - - if domain != server_name: - continue - - memtype = ev.membership - if memtype == Membership.JOIN: - return event - elif memtype == Membership.INVITE: - if visibility == "invited": - return event - else: - return prune_event(event) - - return event - - defer.returnValue([ - redact_disallowed(e, event_to_state[e.event_id]) - for e in events - ]) - @log_function @defer.inlineCallbacks def backfill(self, dest, room_id, limit, extremities): """ Trigger a backfill request to `dest` for the given `room_id` - This will attempt to get more events from the remote. This may return - be successfull and still return no events if the other side has no new - events to offer. + This will attempt to get more events from the remote. If the other side + has no new events to offer, this will return an empty list. + + As the events are received, we check their signatures, and also do some + sanity-checking on them. If any of the backfilled events are invalid, + this method throws a SynapseError. + + TODO: make this more useful to distinguish failures of the remote + server from invalid events (there is probably no point in trying to + re-fetch invalid events from every other HS in the room.) """ if dest == self.server_name: raise SynapseError(400, "Can't backfill from self.") @@ -520,6 +529,16 @@ class FederationHandler(BaseHandler): extremities=extremities, ) + # ideally we'd sanity check the events here for excess prev_events etc, + # but it's hard to reject events at this point without completely + # breaking backfill in the same way that it is currently broken by + # events whose signature we cannot verify (#3121). + # + # So for now we accept the events anyway. #3124 tracks this. + # + # for ev in events: + # self._sanity_check_event(ev) + # Don't bother processing events we already have. seen_events = yield self.store.have_events_in_timeline( set(e.event_id for e in events) @@ -590,9 +609,10 @@ class FederationHandler(BaseHandler): missing_auth - failed_to_fetch ) - results = yield preserve_context_over_deferred(defer.gatherResults( + results = yield logcontext.make_deferred_yieldable(defer.gatherResults( [ - preserve_fn(self.replication_layer.get_pdu)( + logcontext.run_in_background( + self.replication_layer.get_pdu, [dest], event_id, outlier=True, @@ -612,7 +632,7 @@ class FederationHandler(BaseHandler): failed_to_fetch = missing_auth - set(auth_events) - seen_events = yield self.store.have_events( + seen_events = yield self.store.have_seen_events( set(auth_events.keys()) | set(state_events.keys()) ) @@ -702,9 +722,19 @@ class FederationHandler(BaseHandler): curr_state = yield self.state_handler.get_current_state(room_id) def get_domains_from_state(state): + """Get joined domains from state + + Args: + state (dict[tuple, FrozenEvent]): State map from type/state + key to event. + + Returns: + list[tuple[str, int]]: Returns a list of servers with the + lowest depth of their joins. Sorted by lowest depth first. + """ joined_users = [ (state_key, int(event.depth)) - for (e_type, state_key), event in state.items() + for (e_type, state_key), event in iteritems(state) if e_type == EventTypes.Member and event.membership == Membership.JOIN ] @@ -718,7 +748,7 @@ class FederationHandler(BaseHandler): joined_domains[dom] = min(d, old_d) else: joined_domains[dom] = d - except: + except Exception: pass return sorted(joined_domains.items(), key=lambda d: d[1]) @@ -738,7 +768,7 @@ class FederationHandler(BaseHandler): yield self.backfill( dom, room_id, limit=100, - extremities=[e for e in extremities.keys()] + extremities=extremities, ) # If this succeeded then we probably already have the # appropriate stuff. @@ -762,6 +792,9 @@ class FederationHandler(BaseHandler): except NotRetryingDestination as e: logger.info(e.message) continue + except FederationDeniedError as e: + logger.info(e) + continue except Exception as e: logger.exception( "Failed to backfill from %s because %s", @@ -784,38 +817,76 @@ class FederationHandler(BaseHandler): event_ids = list(extremities.keys()) logger.debug("calling resolve_state_groups in _maybe_backfill") - states = yield preserve_context_over_deferred(defer.gatherResults([ - preserve_fn(self.state_handler.resolve_state_groups)(room_id, [e]) - for e in event_ids - ])) + resolve = logcontext.preserve_fn( + self.state_handler.resolve_state_groups_for_events + ) + states = yield logcontext.make_deferred_yieldable(defer.gatherResults( + [resolve(room_id, [e]) for e in event_ids], + consumeErrors=True, + )) + + # dict[str, dict[tuple, str]], a map from event_id to state map of + # event_ids. states = dict(zip(event_ids, [s.state for s in states])) state_map = yield self.store.get_events( - [e_id for ids in states.values() for e_id in ids], + [e_id for ids in itervalues(states) for e_id in itervalues(ids)], get_prev_content=False ) states = { key: { k: state_map[e_id] - for k, e_id in state_dict.items() + for k, e_id in iteritems(state_dict) if e_id in state_map - } for key, state_dict in states.items() + } for key, state_dict in iteritems(states) } for e_id, _ in sorted_extremeties_tuple: likely_domains = get_domains_from_state(states[e_id]) success = yield try_backfill([ - dom for dom in likely_domains + dom for dom, _ in likely_domains if dom not in tried_domains ]) if success: defer.returnValue(True) - tried_domains.update(likely_domains) + tried_domains.update(dom for dom, _ in likely_domains) defer.returnValue(False) + def _sanity_check_event(self, ev): + """ + Do some early sanity checks of a received event + + In particular, checks it doesn't have an excessive number of + prev_events or auth_events, which could cause a huge state resolution + or cascade of event fetches. + + Args: + ev (synapse.events.EventBase): event to be checked + + Returns: None + + Raises: + SynapseError if the event does not pass muster + """ + if len(ev.prev_events) > 20: + logger.warn("Rejecting event %s which has %i prev_events", + ev.event_id, len(ev.prev_events)) + raise SynapseError( + http_client.BAD_REQUEST, + "Too many prev_events", + ) + + if len(ev.auth_events) > 10: + logger.warn("Rejecting event %s which has %i auth_events", + ev.event_id, len(ev.auth_events)) + raise SynapseError( + http_client.BAD_REQUEST, + "Too many auth_events", + ) + @defer.inlineCallbacks def send_invite(self, target_host, event): """ Sends the invite to the remote server for signing. @@ -838,16 +909,6 @@ class FederationHandler(BaseHandler): [auth_id for auth_id, _ in event.auth_events], include_given=True ) - - for event in auth: - event.signatures.update( - compute_event_signature( - event, - self.hs.hostname, - self.hs.config.signing_key[0] - ) - ) - defer.returnValue([e for e in auth]) @log_function @@ -916,7 +977,7 @@ class FederationHandler(BaseHandler): room_creator_user_id="", is_public=False ) - except: + except Exception: # FIXME pass @@ -940,9 +1001,7 @@ class FederationHandler(BaseHandler): # lots of requests for missing prev_events which we do actually # have. Hence we fire off the deferred, but don't wait for it. - synapse.util.logcontext.preserve_fn(self._handle_queued_pdus)( - room_queue - ) + logcontext.run_in_background(self._handle_queued_pdus, room_queue) defer.returnValue(True) @@ -982,8 +1041,7 @@ class FederationHandler(BaseHandler): }) try: - message_handler = self.hs.get_handlers().message_handler - event, context = yield message_handler._create_new_client_event( + event, context = yield self.event_creation_handler.create_new_client_event( builder=builder, ) except AuthError as e: @@ -1051,13 +1109,15 @@ class FederationHandler(BaseHandler): user = UserID.from_string(event.state_key) yield user_joined_room(self.distributor, user, event.room_id) - state_ids = context.prev_state_ids.values() + prev_state_ids = yield context.get_prev_state_ids(self.store) + + state_ids = list(prev_state_ids.values()) auth_chain = yield self.store.get_auth_chain(state_ids) - state = yield self.store.get_events(context.prev_state_ids.values()) + state = yield self.store.get_events(list(prev_state_ids.values())) defer.returnValue({ - "state": state.values(), + "state": list(state.values()), "auth_chain": auth_chain, }) @@ -1069,10 +1129,23 @@ class FederationHandler(BaseHandler): """ event = pdu + if event.state_key is None: + raise SynapseError(400, "The invite event did not have a state key") + is_blocked = yield self.store.is_room_blocked(event.room_id) if is_blocked: raise SynapseError(403, "This room has been blocked on this server") + if self.hs.config.block_non_admin_invites: + raise SynapseError(403, "This server does not accept room invites") + + if not self.spam_checker.user_may_invite( + event.sender, event.state_key, event.room_id, + ): + raise SynapseError( + 403, "This user is not permitted to send invites to this server/user" + ) + membership = event.content.get("membership") if event.type != EventTypes.Member or membership != Membership.INVITE: raise SynapseError(400, "The event was not an m.room.member invite event") @@ -1081,12 +1154,16 @@ class FederationHandler(BaseHandler): if sender_domain != origin: raise SynapseError(400, "The invite event was not from the server sending it") - if event.state_key is None: - raise SynapseError(400, "The invite event did not have a state key") - if not self.is_mine_id(event.state_key): raise SynapseError(400, "The invite event must be for this server") + # block any attempts to invite the server notices mxid + if event.state_key == self._server_notices_mxid: + raise SynapseError( + http_client.FORBIDDEN, + "Cannot invite this user", + ) + event.internal_metadata.outlier = True event.internal_metadata.invite_from_remote = True @@ -1213,8 +1290,7 @@ class FederationHandler(BaseHandler): "state_key": user_id, }) - message_handler = self.hs.get_handlers().message_handler - event, context = yield message_handler._create_new_client_event( + event, context = yield self.event_creation_handler.create_new_client_event( builder=builder, ) @@ -1268,14 +1344,12 @@ class FederationHandler(BaseHandler): def get_state_for_pdu(self, room_id, event_id): """Returns the state at the event. i.e. not including said event. """ - yield run_on_reactor() - state_groups = yield self.store.get_state_groups( room_id, [event_id] ) if state_groups: - _, state = state_groups.items().pop() + _, state = list(iteritems(state_groups)).pop() results = { (e.type, e.state_key): e for e in state } @@ -1291,19 +1365,7 @@ class FederationHandler(BaseHandler): else: del results[(event.type, event.state_key)] - res = results.values() - for event in res: - # We sign these again because there was a bug where we - # incorrectly signed things the first time round - if self.is_mine_id(event.event_id): - event.signatures.update( - compute_event_signature( - event, - self.hs.hostname, - self.hs.config.signing_key[0] - ) - ) - + res = list(results.values()) defer.returnValue(res) else: defer.returnValue([]) @@ -1312,8 +1374,6 @@ class FederationHandler(BaseHandler): def get_state_ids_for_pdu(self, room_id, event_id): """Returns the state at the event. i.e. not including said event. """ - yield run_on_reactor() - state_groups = yield self.store.get_state_groups_ids( room_id, [event_id] ) @@ -1332,7 +1392,7 @@ class FederationHandler(BaseHandler): else: results.pop((event.type, event.state_key), None) - defer.returnValue(results.values()) + defer.returnValue(list(results.values())) else: defer.returnValue([]) @@ -1349,17 +1409,26 @@ class FederationHandler(BaseHandler): limit ) - events = yield self._filter_events_for_server(origin, room_id, events) + events = yield filter_events_for_server(self.store, origin, events) defer.returnValue(events) @defer.inlineCallbacks @log_function - def get_persisted_pdu(self, origin, event_id, do_auth=True): - """ Get a PDU from the database with given origin and id. + def get_persisted_pdu(self, origin, event_id): + """Get an event from the database for the given server. + + Args: + origin [str]: hostname of server which is requesting the event; we + will check that the server is allowed to see it. + event_id [str]: id of the event being requested Returns: - Deferred: Results in a `Pdu`. + Deferred[EventBase|None]: None if we know nothing about the event; + otherwise the (possibly-redacted) event. + + Raises: + AuthError if the server is not currently in the room """ event = yield self.store.get_event( event_id, @@ -1368,32 +1437,17 @@ class FederationHandler(BaseHandler): ) if event: - if self.is_mine_id(event.event_id): - # FIXME: This is a temporary work around where we occasionally - # return events slightly differently than when they were - # originally signed - event.signatures.update( - compute_event_signature( - event, - self.hs.hostname, - self.hs.config.signing_key[0] - ) - ) - - if do_auth: - in_room = yield self.auth.check_host_in_room( - event.room_id, - origin - ) - if not in_room: - raise AuthError(403, "Host not in room.") - - events = yield self._filter_events_for_server( - origin, event.room_id, [event] - ) - - event = events[0] + in_room = yield self.auth.check_host_in_room( + event.room_id, + origin + ) + if not in_room: + raise AuthError(403, "Host not in room.") + events = yield filter_events_for_server( + self.store, origin, [event], + ) + event = events[0] defer.returnValue(event) else: defer.returnValue(None) @@ -1412,22 +1466,33 @@ class FederationHandler(BaseHandler): auth_events=auth_events, ) - if not event.internal_metadata.is_outlier(): - yield self.action_generator.handle_push_actions_for_event( - event, context + try: + if not event.internal_metadata.is_outlier() and not backfilled: + yield self.action_generator.handle_push_actions_for_event( + event, context + ) + + event_stream_id, max_stream_id = yield self.store.persist_event( + event, + context=context, + backfilled=backfilled, + ) + except: # noqa: E722, as we reraise the exception this is fine. + tp, value, tb = sys.exc_info() + + logcontext.run_in_background( + self.store.remove_push_actions_from_staging, + event.event_id, ) - event_stream_id, max_stream_id = yield self.store.persist_event( - event, - context=context, - backfilled=backfilled, - ) + six.reraise(tp, value, tb) if not backfilled: # this intentionally does not yield: we don't care about the result # and don't need to wait for it. - preserve_fn(self.hs.get_pusherpool().on_new_notifications)( - event_stream_id, max_stream_id + logcontext.run_in_background( + self.pusher_pool.on_new_notifications, + event_stream_id, max_stream_id, ) defer.returnValue((context, event_stream_id, max_stream_id)) @@ -1439,22 +1504,23 @@ class FederationHandler(BaseHandler): a bunch of outliers, but not a chunk of individual events that depend on each other for state calculations. """ - contexts = yield preserve_context_over_deferred(defer.gatherResults( + contexts = yield logcontext.make_deferred_yieldable(defer.gatherResults( [ - preserve_fn(self._prep_event)( + logcontext.run_in_background( + self._prep_event, origin, ev_info["event"], state=ev_info.get("state"), auth_events=ev_info.get("auth_events"), ) for ev_info in event_infos - ] + ], consumeErrors=True, )) yield self.store.persist_events( [ (ev_info["event"], context) - for ev_info, context in itertools.izip(event_infos, contexts) + for ev_info, context in zip(event_infos, contexts) ], backfilled=backfilled, ) @@ -1574,8 +1640,9 @@ class FederationHandler(BaseHandler): ) if not auth_events: + prev_state_ids = yield context.get_prev_state_ids(self.store) auth_events_ids = yield self.auth.compute_auth_events( - event, context.prev_state_ids, for_verification=True, + event, prev_state_ids, for_verification=True, ) auth_events = yield self.store.get_events(auth_events_ids) auth_events = { @@ -1605,7 +1672,7 @@ class FederationHandler(BaseHandler): context.rejected = RejectedReason.AUTH_ERROR - if event.type == EventTypes.GuestAccess: + if event.type == EventTypes.GuestAccess and not context.rejected: yield self.maybe_kick_guest_users(event) defer.returnValue(context) @@ -1635,15 +1702,6 @@ class FederationHandler(BaseHandler): local_auth_chain, remote_auth_chain ) - for event in ret["auth_chain"]: - event.signatures.update( - compute_event_signature( - event, - self.hs.hostname, - self.hs.config.signing_key[0] - ) - ) - logger.debug("on_query_auth returning: %s", ret) defer.returnValue(ret) @@ -1669,11 +1727,26 @@ class FederationHandler(BaseHandler): min_depth=min_depth, ) + missing_events = yield filter_events_for_server( + self.store, origin, missing_events, + ) + defer.returnValue(missing_events) @defer.inlineCallbacks @log_function def do_auth(self, origin, event, context, auth_events): + """ + + Args: + origin (str): + event (synapse.events.FrozenEvent): + context (synapse.events.snapshot.EventContext): + auth_events (dict[(str, str)->str]): + + Returns: + defer.Deferred[None] + """ # Check if we have all the auth events. current_state = set(e.event_id for e in auth_events.values()) event_auth_events = set(e_id for e_id, _ in event.auth_events) @@ -1684,7 +1757,8 @@ class FederationHandler(BaseHandler): event_key = None if event_auth_events - current_state: - have_events = yield self.store.have_events( + # TODO: can we use store.have_seen_events here instead? + have_events = yield self.store.get_seen_events_with_rejections( event_auth_events - current_state ) else: @@ -1707,12 +1781,12 @@ class FederationHandler(BaseHandler): origin, event.room_id, event.event_id ) - seen_remotes = yield self.store.have_events( + seen_remotes = yield self.store.have_seen_events( [e.event_id for e in remote_auth_chain] ) for e in remote_auth_chain: - if e.event_id in seen_remotes.keys(): + if e.event_id in seen_remotes: continue if e.event_id == event.event_id: @@ -1739,11 +1813,11 @@ class FederationHandler(BaseHandler): except AuthError: pass - have_events = yield self.store.have_events( + have_events = yield self.store.get_seen_events_with_rejections( [e_id for e_id, _ in event.auth_events] ) seen_events = set(have_events.keys()) - except: + except Exception: # FIXME: logger.exception("Failed to get auth chain") @@ -1756,18 +1830,18 @@ class FederationHandler(BaseHandler): # Do auth conflict res. logger.info("Different auth: %s", different_auth) - different_events = yield preserve_context_over_deferred(defer.gatherResults( - [ - preserve_fn(self.store.get_event)( + different_events = yield logcontext.make_deferred_yieldable( + defer.gatherResults([ + logcontext.run_in_background( + self.store.get_event, d, allow_none=True, allow_rejected=False, ) for d in different_auth if d in have_events and not have_events[d] - ], - consumeErrors=True - )).addErrback(unwrapFirstError) + ], consumeErrors=True) + ).addErrback(unwrapFirstError) if different_events: local_view = dict(auth_events) @@ -1777,7 +1851,7 @@ class FederationHandler(BaseHandler): }) new_state = self.state_handler.resolve_events( - [local_view.values(), remote_view.values()], + [list(local_view.values()), list(remote_view.values())], event ) @@ -1786,16 +1860,9 @@ class FederationHandler(BaseHandler): current_state = set(e.event_id for e in auth_events.values()) different_auth = event_auth_events - current_state - context.current_state_ids = dict(context.current_state_ids) - context.current_state_ids.update({ - k: a.event_id for k, a in auth_events.items() - if k != event_key - }) - context.prev_state_ids = dict(context.prev_state_ids) - context.prev_state_ids.update({ - k: a.event_id for k, a in auth_events.items() - }) - context.state_group = self.store.get_next_state_group() + yield self._update_context_for_auth_events( + event, context, auth_events, event_key, + ) if different_auth and not event.internal_metadata.is_outlier(): logger.info("Different auth after resolution: %s", different_auth) @@ -1815,9 +1882,10 @@ class FederationHandler(BaseHandler): break if do_resolution: + prev_state_ids = yield context.get_prev_state_ids(self.store) # 1. Get what we think is the auth chain. auth_ids = yield self.auth.compute_auth_events( - event, context.prev_state_ids + event, prev_state_ids ) local_auth_chain = yield self.store.get_auth_chain( auth_ids, include_given=True @@ -1832,13 +1900,13 @@ class FederationHandler(BaseHandler): local_auth_chain, ) - seen_remotes = yield self.store.have_events( + seen_remotes = yield self.store.have_seen_events( [e.event_id for e in result["auth_chain"]] ) # 3. Process any remote auth chain events we haven't seen. for ev in result["auth_chain"]: - if ev.event_id in seen_remotes.keys(): + if ev.event_id in seen_remotes: continue if ev.event_id == event.event_id: @@ -1868,23 +1936,16 @@ class FederationHandler(BaseHandler): except AuthError: pass - except: + except Exception: # FIXME: logger.exception("Failed to query auth chain") # 4. Look at rejects and their proofs. # TODO. - context.current_state_ids = dict(context.current_state_ids) - context.current_state_ids.update({ - k: a.event_id for k, a in auth_events.items() - if k != event_key - }) - context.prev_state_ids = dict(context.prev_state_ids) - context.prev_state_ids.update({ - k: a.event_id for k, a in auth_events.items() - }) - context.state_group = self.store.get_next_state_group() + yield self._update_context_for_auth_events( + event, context, auth_events, event_key, + ) try: self.auth.check(event, auth_events=auth_events) @@ -1892,6 +1953,58 @@ class FederationHandler(BaseHandler): logger.warn("Failed auth resolution for %r because %s", event, e) raise e + @defer.inlineCallbacks + def _update_context_for_auth_events(self, event, context, auth_events, + event_key): + """Update the state_ids in an event context after auth event resolution, + storing the changes as a new state group. + + Args: + event (Event): The event we're handling the context for + + context (synapse.events.snapshot.EventContext): event context + to be updated + + auth_events (dict[(str, str)->str]): Events to update in the event + context. + + event_key ((str, str)): (type, state_key) for the current event. + this will not be included in the current_state in the context. + """ + state_updates = { + k: a.event_id for k, a in iteritems(auth_events) + if k != event_key + } + current_state_ids = yield context.get_current_state_ids(self.store) + current_state_ids = dict(current_state_ids) + + current_state_ids.update(state_updates) + + prev_state_ids = yield context.get_prev_state_ids(self.store) + prev_state_ids = dict(prev_state_ids) + + prev_state_ids.update({ + k: a.event_id for k, a in iteritems(auth_events) + }) + + # create a new state group as a delta from the existing one. + prev_group = context.state_group + state_group = yield self.store.store_state_group( + event.event_id, + event.room_id, + prev_group=prev_group, + delta_ids=state_updates, + current_state_ids=current_state_ids, + ) + + yield context.update_state( + state_group=state_group, + current_state_ids=current_state_ids, + prev_state_ids=prev_state_ids, + prev_group=prev_group, + delta_ids=state_updates, + ) + @defer.inlineCallbacks def construct_auth_difference(self, local_auth, remote_auth): """ Given a local and remote auth chain, find the differences. This @@ -1934,8 +2047,8 @@ class FederationHandler(BaseHandler): def get_next(it, opt=None): try: - return it.next() - except: + return next(it) + except Exception: return opt current_local = get_next(local_iter) @@ -2060,8 +2173,7 @@ class FederationHandler(BaseHandler): if (yield self.auth.check_host_in_room(room_id, self.hs.hostname)): builder = self.event_builder_factory.new(event_dict) EventValidator().validate_new(builder) - message_handler = self.hs.get_handlers().message_handler - event, context = yield message_handler._create_new_client_event( + event, context = yield self.event_creation_handler.create_new_client_event( builder=builder ) @@ -2076,7 +2188,7 @@ class FederationHandler(BaseHandler): raise e yield self._check_signature(event, context) - member_handler = self.hs.get_handlers().room_member_handler + member_handler = self.hs.get_room_member_handler() yield member_handler.send_membership_event(None, event, context) else: destinations = set(x.split(":", 1)[-1] for x in (sender_user_id, room_id)) @@ -2089,10 +2201,17 @@ class FederationHandler(BaseHandler): @defer.inlineCallbacks @log_function def on_exchange_third_party_invite_request(self, origin, room_id, event_dict): + """Handle an exchange_third_party_invite request from a remote server + + The remote server will call this when it wants to turn a 3pid invite + into a normal m.room.member invite. + + Returns: + Deferred: resolves (to None) + """ builder = self.event_builder_factory.new(event_dict) - message_handler = self.hs.get_handlers().message_handler - event, context = yield message_handler._create_new_client_event( + event, context = yield self.event_creation_handler.create_new_client_event( builder=builder, ) @@ -2107,10 +2226,13 @@ class FederationHandler(BaseHandler): raise e yield self._check_signature(event, context) + # XXX we send the invite here, but send_membership_event also sends it, + # so we end up making two requests. I think this is redundant. returned_invite = yield self.send_invite(origin, event) # TODO: Make sure the signatures actually are correct. event.signatures.update(returned_invite.signatures) - member_handler = self.hs.get_handlers().room_member_handler + + member_handler = self.hs.get_room_member_handler() yield member_handler.send_membership_event(None, event, context) @defer.inlineCallbacks @@ -2120,7 +2242,8 @@ class FederationHandler(BaseHandler): event.content["third_party_invite"]["signed"]["token"] ) original_invite = None - original_invite_id = context.prev_state_ids.get(key) + prev_state_ids = yield context.get_prev_state_ids(self.store) + original_invite_id = prev_state_ids.get(key) if original_invite_id: original_invite = yield self.store.get_event( original_invite_id, allow_none=True @@ -2139,8 +2262,9 @@ class FederationHandler(BaseHandler): builder = self.event_builder_factory.new(event_dict) EventValidator().validate_new(builder) - message_handler = self.hs.get_handlers().message_handler - event, context = yield message_handler._create_new_client_event(builder=builder) + event, context = yield self.event_creation_handler.create_new_client_event( + builder=builder, + ) defer.returnValue((event, context)) @defer.inlineCallbacks @@ -2161,7 +2285,8 @@ class FederationHandler(BaseHandler): signed = event.content["third_party_invite"]["signed"] token = signed["token"] - invite_event_id = context.prev_state_ids.get( + prev_state_ids = yield context.get_prev_state_ids(self.store) + invite_event_id = prev_state_ids.get( (EventTypes.ThirdPartyInvite, token,) ) diff --git a/synapse/handlers/groups_local.py b/synapse/handlers/groups_local.py new file mode 100644 index 000000000..53e5e2648 --- /dev/null +++ b/synapse/handlers/groups_local.py @@ -0,0 +1,473 @@ +# -*- coding: utf-8 -*- +# Copyright 2017 Vector Creations Ltd +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging + +from six import iteritems + +from twisted.internet import defer + +from synapse.api.errors import SynapseError +from synapse.types import get_domain_from_id + +logger = logging.getLogger(__name__) + + +def _create_rerouter(func_name): + """Returns a function that looks at the group id and calls the function + on federation or the local group server if the group is local + """ + def f(self, group_id, *args, **kwargs): + if self.is_mine_id(group_id): + return getattr(self.groups_server_handler, func_name)( + group_id, *args, **kwargs + ) + else: + destination = get_domain_from_id(group_id) + return getattr(self.transport_client, func_name)( + destination, group_id, *args, **kwargs + ) + return f + + +class GroupsLocalHandler(object): + def __init__(self, hs): + self.hs = hs + self.store = hs.get_datastore() + self.room_list_handler = hs.get_room_list_handler() + self.groups_server_handler = hs.get_groups_server_handler() + self.transport_client = hs.get_federation_transport_client() + self.auth = hs.get_auth() + self.clock = hs.get_clock() + self.keyring = hs.get_keyring() + self.is_mine_id = hs.is_mine_id + self.signing_key = hs.config.signing_key[0] + self.server_name = hs.hostname + self.notifier = hs.get_notifier() + self.attestations = hs.get_groups_attestation_signing() + + self.profile_handler = hs.get_profile_handler() + + # Ensure attestations get renewed + hs.get_groups_attestation_renewer() + + # The following functions merely route the query to the local groups server + # or federation depending on if the group is local or remote + + get_group_profile = _create_rerouter("get_group_profile") + update_group_profile = _create_rerouter("update_group_profile") + get_rooms_in_group = _create_rerouter("get_rooms_in_group") + + get_invited_users_in_group = _create_rerouter("get_invited_users_in_group") + + add_room_to_group = _create_rerouter("add_room_to_group") + update_room_in_group = _create_rerouter("update_room_in_group") + remove_room_from_group = _create_rerouter("remove_room_from_group") + + update_group_summary_room = _create_rerouter("update_group_summary_room") + delete_group_summary_room = _create_rerouter("delete_group_summary_room") + + update_group_category = _create_rerouter("update_group_category") + delete_group_category = _create_rerouter("delete_group_category") + get_group_category = _create_rerouter("get_group_category") + get_group_categories = _create_rerouter("get_group_categories") + + update_group_summary_user = _create_rerouter("update_group_summary_user") + delete_group_summary_user = _create_rerouter("delete_group_summary_user") + + update_group_role = _create_rerouter("update_group_role") + delete_group_role = _create_rerouter("delete_group_role") + get_group_role = _create_rerouter("get_group_role") + get_group_roles = _create_rerouter("get_group_roles") + + set_group_join_policy = _create_rerouter("set_group_join_policy") + + @defer.inlineCallbacks + def get_group_summary(self, group_id, requester_user_id): + """Get the group summary for a group. + + If the group is remote we check that the users have valid attestations. + """ + if self.is_mine_id(group_id): + res = yield self.groups_server_handler.get_group_summary( + group_id, requester_user_id + ) + else: + res = yield self.transport_client.get_group_summary( + get_domain_from_id(group_id), group_id, requester_user_id, + ) + + group_server_name = get_domain_from_id(group_id) + + # Loop through the users and validate the attestations. + chunk = res["users_section"]["users"] + valid_users = [] + for entry in chunk: + g_user_id = entry["user_id"] + attestation = entry.pop("attestation", {}) + try: + if get_domain_from_id(g_user_id) != group_server_name: + yield self.attestations.verify_attestation( + attestation, + group_id=group_id, + user_id=g_user_id, + server_name=get_domain_from_id(g_user_id), + ) + valid_users.append(entry) + except Exception as e: + logger.info("Failed to verify user is in group: %s", e) + + res["users_section"]["users"] = valid_users + + res["users_section"]["users"].sort(key=lambda e: e.get("order", 0)) + res["rooms_section"]["rooms"].sort(key=lambda e: e.get("order", 0)) + + # Add `is_publicised` flag to indicate whether the user has publicised their + # membership of the group on their profile + result = yield self.store.get_publicised_groups_for_user(requester_user_id) + is_publicised = group_id in result + + res.setdefault("user", {})["is_publicised"] = is_publicised + + defer.returnValue(res) + + @defer.inlineCallbacks + def create_group(self, group_id, user_id, content): + """Create a group + """ + + logger.info("Asking to create group with ID: %r", group_id) + + if self.is_mine_id(group_id): + res = yield self.groups_server_handler.create_group( + group_id, user_id, content + ) + local_attestation = None + remote_attestation = None + else: + local_attestation = self.attestations.create_attestation(group_id, user_id) + content["attestation"] = local_attestation + + content["user_profile"] = yield self.profile_handler.get_profile(user_id) + + res = yield self.transport_client.create_group( + get_domain_from_id(group_id), group_id, user_id, content, + ) + + remote_attestation = res["attestation"] + yield self.attestations.verify_attestation( + remote_attestation, + group_id=group_id, + user_id=user_id, + server_name=get_domain_from_id(group_id), + ) + + is_publicised = content.get("publicise", False) + token = yield self.store.register_user_group_membership( + group_id, user_id, + membership="join", + is_admin=True, + local_attestation=local_attestation, + remote_attestation=remote_attestation, + is_publicised=is_publicised, + ) + self.notifier.on_new_event( + "groups_key", token, users=[user_id], + ) + + defer.returnValue(res) + + @defer.inlineCallbacks + def get_users_in_group(self, group_id, requester_user_id): + """Get users in a group + """ + if self.is_mine_id(group_id): + res = yield self.groups_server_handler.get_users_in_group( + group_id, requester_user_id + ) + defer.returnValue(res) + + group_server_name = get_domain_from_id(group_id) + + res = yield self.transport_client.get_users_in_group( + get_domain_from_id(group_id), group_id, requester_user_id, + ) + + chunk = res["chunk"] + valid_entries = [] + for entry in chunk: + g_user_id = entry["user_id"] + attestation = entry.pop("attestation", {}) + try: + if get_domain_from_id(g_user_id) != group_server_name: + yield self.attestations.verify_attestation( + attestation, + group_id=group_id, + user_id=g_user_id, + server_name=get_domain_from_id(g_user_id), + ) + valid_entries.append(entry) + except Exception as e: + logger.info("Failed to verify user is in group: %s", e) + + res["chunk"] = valid_entries + + defer.returnValue(res) + + @defer.inlineCallbacks + def join_group(self, group_id, user_id, content): + """Request to join a group + """ + if self.is_mine_id(group_id): + yield self.groups_server_handler.join_group( + group_id, user_id, content + ) + local_attestation = None + remote_attestation = None + else: + local_attestation = self.attestations.create_attestation(group_id, user_id) + content["attestation"] = local_attestation + + res = yield self.transport_client.join_group( + get_domain_from_id(group_id), group_id, user_id, content, + ) + + remote_attestation = res["attestation"] + + yield self.attestations.verify_attestation( + remote_attestation, + group_id=group_id, + user_id=user_id, + server_name=get_domain_from_id(group_id), + ) + + # TODO: Check that the group is public and we're being added publically + is_publicised = content.get("publicise", False) + + token = yield self.store.register_user_group_membership( + group_id, user_id, + membership="join", + is_admin=False, + local_attestation=local_attestation, + remote_attestation=remote_attestation, + is_publicised=is_publicised, + ) + self.notifier.on_new_event( + "groups_key", token, users=[user_id], + ) + + defer.returnValue({}) + + @defer.inlineCallbacks + def accept_invite(self, group_id, user_id, content): + """Accept an invite to a group + """ + if self.is_mine_id(group_id): + yield self.groups_server_handler.accept_invite( + group_id, user_id, content + ) + local_attestation = None + remote_attestation = None + else: + local_attestation = self.attestations.create_attestation(group_id, user_id) + content["attestation"] = local_attestation + + res = yield self.transport_client.accept_group_invite( + get_domain_from_id(group_id), group_id, user_id, content, + ) + + remote_attestation = res["attestation"] + + yield self.attestations.verify_attestation( + remote_attestation, + group_id=group_id, + user_id=user_id, + server_name=get_domain_from_id(group_id), + ) + + # TODO: Check that the group is public and we're being added publically + is_publicised = content.get("publicise", False) + + token = yield self.store.register_user_group_membership( + group_id, user_id, + membership="join", + is_admin=False, + local_attestation=local_attestation, + remote_attestation=remote_attestation, + is_publicised=is_publicised, + ) + self.notifier.on_new_event( + "groups_key", token, users=[user_id], + ) + + defer.returnValue({}) + + @defer.inlineCallbacks + def invite(self, group_id, user_id, requester_user_id, config): + """Invite a user to a group + """ + content = { + "requester_user_id": requester_user_id, + "config": config, + } + if self.is_mine_id(group_id): + res = yield self.groups_server_handler.invite_to_group( + group_id, user_id, requester_user_id, content, + ) + else: + res = yield self.transport_client.invite_to_group( + get_domain_from_id(group_id), group_id, user_id, requester_user_id, + content, + ) + + defer.returnValue(res) + + @defer.inlineCallbacks + def on_invite(self, group_id, user_id, content): + """One of our users were invited to a group + """ + # TODO: Support auto join and rejection + + if not self.is_mine_id(user_id): + raise SynapseError(400, "User not on this server") + + local_profile = {} + if "profile" in content: + if "name" in content["profile"]: + local_profile["name"] = content["profile"]["name"] + if "avatar_url" in content["profile"]: + local_profile["avatar_url"] = content["profile"]["avatar_url"] + + token = yield self.store.register_user_group_membership( + group_id, user_id, + membership="invite", + content={"profile": local_profile, "inviter": content["inviter"]}, + ) + self.notifier.on_new_event( + "groups_key", token, users=[user_id], + ) + try: + user_profile = yield self.profile_handler.get_profile(user_id) + except Exception as e: + logger.warn("No profile for user %s: %s", user_id, e) + user_profile = {} + + defer.returnValue({"state": "invite", "user_profile": user_profile}) + + @defer.inlineCallbacks + def remove_user_from_group(self, group_id, user_id, requester_user_id, content): + """Remove a user from a group + """ + if user_id == requester_user_id: + token = yield self.store.register_user_group_membership( + group_id, user_id, + membership="leave", + ) + self.notifier.on_new_event( + "groups_key", token, users=[user_id], + ) + + # TODO: Should probably remember that we tried to leave so that we can + # retry if the group server is currently down. + + if self.is_mine_id(group_id): + res = yield self.groups_server_handler.remove_user_from_group( + group_id, user_id, requester_user_id, content, + ) + else: + content["requester_user_id"] = requester_user_id + res = yield self.transport_client.remove_user_from_group( + get_domain_from_id(group_id), group_id, requester_user_id, + user_id, content, + ) + + defer.returnValue(res) + + @defer.inlineCallbacks + def user_removed_from_group(self, group_id, user_id, content): + """One of our users was removed/kicked from a group + """ + # TODO: Check if user in group + token = yield self.store.register_user_group_membership( + group_id, user_id, + membership="leave", + ) + self.notifier.on_new_event( + "groups_key", token, users=[user_id], + ) + + @defer.inlineCallbacks + def get_joined_groups(self, user_id): + group_ids = yield self.store.get_joined_groups(user_id) + defer.returnValue({"groups": group_ids}) + + @defer.inlineCallbacks + def get_publicised_groups_for_user(self, user_id): + if self.hs.is_mine_id(user_id): + result = yield self.store.get_publicised_groups_for_user(user_id) + + # Check AS associated groups for this user - this depends on the + # RegExps in the AS registration file (under `users`) + for app_service in self.store.get_app_services(): + result.extend(app_service.get_groups_for_user(user_id)) + + defer.returnValue({"groups": result}) + else: + bulk_result = yield self.transport_client.bulk_get_publicised_groups( + get_domain_from_id(user_id), [user_id], + ) + result = bulk_result.get("users", {}).get(user_id) + # TODO: Verify attestations + defer.returnValue({"groups": result}) + + @defer.inlineCallbacks + def bulk_get_publicised_groups(self, user_ids, proxy=True): + destinations = {} + local_users = set() + + for user_id in user_ids: + if self.hs.is_mine_id(user_id): + local_users.add(user_id) + else: + destinations.setdefault( + get_domain_from_id(user_id), set() + ).add(user_id) + + if not proxy and destinations: + raise SynapseError(400, "Some user_ids are not local") + + results = {} + failed_results = [] + for destination, dest_user_ids in iteritems(destinations): + try: + r = yield self.transport_client.bulk_get_publicised_groups( + destination, list(dest_user_ids), + ) + results.update(r["users"]) + except Exception: + failed_results.extend(dest_user_ids) + + for uid in local_users: + results[uid] = yield self.store.get_publicised_groups_for_user( + uid + ) + + # Check AS associated groups for this user - this depends on the + # RegExps in the AS registration file (under `users`) + for app_service in self.store.get_app_services(): + results[uid].extend(app_service.get_groups_for_user(uid)) + + defer.returnValue({"users": results}) diff --git a/synapse/handlers/identity.py b/synapse/handlers/identity.py index 9efcdff1d..8c8aedb2b 100644 --- a/synapse/handlers/identity.py +++ b/synapse/handlers/identity.py @@ -1,6 +1,7 @@ # -*- coding: utf-8 -*- # Copyright 2015, 2016 OpenMarket Ltd # Copyright 2017 Vector Creations Ltd +# Copyright 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -15,17 +16,21 @@ # limitations under the License. """Utilities for interacting with Identity Servers""" + +import logging + +from canonicaljson import json + from twisted.internet import defer from synapse.api.errors import ( - MatrixCodeMessageException, CodeMessageException + CodeMessageException, + Codes, + MatrixCodeMessageException, + SynapseError, ) -from ._base import BaseHandler -from synapse.util.async import run_on_reactor -from synapse.api.errors import SynapseError, Codes -import json -import logging +from ._base import BaseHandler logger = logging.getLogger(__name__) @@ -36,6 +41,7 @@ class IdentityHandler(BaseHandler): super(IdentityHandler, self).__init__(hs) self.http_client = hs.get_simple_http_client() + self.federation_http_client = hs.get_http_client() self.trusted_id_servers = set(hs.config.trusted_third_party_id_servers) self.trust_any_id_server_just_for_testing_do_not_use = ( @@ -58,8 +64,6 @@ class IdentityHandler(BaseHandler): @defer.inlineCallbacks def threepid_from_creds(self, creds): - yield run_on_reactor() - if 'id_server' in creds: id_server = creds['id_server'] elif 'idServer' in creds: @@ -102,7 +106,6 @@ class IdentityHandler(BaseHandler): @defer.inlineCallbacks def bind_threepid(self, creds, mxid): - yield run_on_reactor() logger.debug("binding threepid %r to %s", creds, mxid) data = None @@ -137,9 +140,53 @@ class IdentityHandler(BaseHandler): defer.returnValue(data) @defer.inlineCallbacks - def requestEmailToken(self, id_server, email, client_secret, send_attempt, **kwargs): - yield run_on_reactor() + def unbind_threepid(self, mxid, threepid): + """ + Removes a binding from an identity server + Args: + mxid (str): Matrix user ID of binding to be removed + threepid (dict): Dict with medium & address of binding to be removed + Returns: + Deferred[bool]: True on success, otherwise False + """ + logger.debug("unbinding threepid %r from %s", threepid, mxid) + if not self.trusted_id_servers: + logger.warn("Can't unbind threepid: no trusted ID servers set in config") + defer.returnValue(False) + + # We don't track what ID server we added 3pids on (perhaps we ought to) + # but we assume that any of the servers in the trusted list are in the + # same ID server federation, so we can pick any one of them to send the + # deletion request to. + id_server = next(iter(self.trusted_id_servers)) + + url = "https://%s/_matrix/identity/api/v1/3pid/unbind" % (id_server,) + content = { + "mxid": mxid, + "threepid": threepid, + } + headers = {} + # we abuse the federation http client to sign the request, but we have to send it + # using the normal http client since we don't want the SRV lookup and want normal + # 'browser-like' HTTPS. + self.federation_http_client.sign_request( + destination=None, + method='POST', + url_bytes='/_matrix/identity/api/v1/3pid/unbind'.encode('ascii'), + headers_dict=headers, + content=content, + destination_is=id_server, + ) + yield self.http_client.post_json_get_json( + url, + content, + headers, + ) + defer.returnValue(True) + + @defer.inlineCallbacks + def requestEmailToken(self, id_server, email, client_secret, send_attempt, **kwargs): if not self._should_trust_id_server(id_server): raise SynapseError( 400, "Untrusted ID server '%s'" % id_server, @@ -174,8 +221,6 @@ class IdentityHandler(BaseHandler): self, id_server, country, phone_number, client_secret, send_attempt, **kwargs ): - yield run_on_reactor() - if not self._should_trust_id_server(id_server): raise SynapseError( 400, "Untrusted ID server '%s'" % id_server, diff --git a/synapse/handlers/initial_sync.py b/synapse/handlers/initial_sync.py index 10f5f35a6..40e7580a6 100644 --- a/synapse/handlers/initial_sync.py +++ b/synapse/handlers/initial_sync.py @@ -13,6 +13,8 @@ # See the License for the specific language governing permissions and # limitations under the License. +import logging + from twisted.internet import defer from synapse.api.constants import EventTypes, Membership @@ -21,20 +23,15 @@ from synapse.events.utils import serialize_event from synapse.events.validator import EventValidator from synapse.handlers.presence import format_user_presence_state from synapse.streams.config import PaginationConfig -from synapse.types import ( - UserID, StreamToken, -) +from synapse.types import StreamToken, UserID from synapse.util import unwrapFirstError from synapse.util.async import concurrently_execute from synapse.util.caches.snapshot_cache import SnapshotCache -from synapse.util.logcontext import preserve_fn, preserve_context_over_deferred +from synapse.util.logcontext import make_deferred_yieldable, run_in_background from synapse.visibility import filter_events_for_client from ._base import BaseHandler -import logging - - logger = logging.getLogger(__name__) @@ -151,22 +148,25 @@ class InitialSyncHandler(BaseHandler): try: if event.membership == Membership.JOIN: room_end_token = now_token.room_key - deferred_room_state = self.state_handler.get_current_state( - event.room_id + deferred_room_state = run_in_background( + self.state_handler.get_current_state, + event.room_id, ) elif event.membership == Membership.LEAVE: room_end_token = "s%d" % (event.stream_ordering,) - deferred_room_state = self.store.get_state_for_events( - [event.event_id], None + deferred_room_state = run_in_background( + self.store.get_state_for_events, + [event.event_id], None, ) deferred_room_state.addCallback( lambda states: states[event.event_id] ) - (messages, token), current_state = yield preserve_context_over_deferred( + (messages, token), current_state = yield make_deferred_yieldable( defer.gatherResults( [ - preserve_fn(self.store.get_recent_events_for_room)( + run_in_background( + self.store.get_recent_events_for_room, event.room_id, limit=limit, end_token=room_end_token, @@ -180,8 +180,8 @@ class InitialSyncHandler(BaseHandler): self.store, user_id, messages ) - start_token = now_token.copy_and_replace("room_key", token[0]) - end_token = now_token.copy_and_replace("room_key", token[1]) + start_token = now_token.copy_and_replace("room_key", token) + end_token = now_token.copy_and_replace("room_key", room_end_token) time_now = self.clock.time_msec() d["messages"] = { @@ -214,7 +214,7 @@ class InitialSyncHandler(BaseHandler): }) d["account_data"] = account_data_events - except: + except Exception: logger.exception("Failed to get snapshot") yield concurrently_execute(handle_room, room_list, 10) @@ -324,8 +324,8 @@ class InitialSyncHandler(BaseHandler): self.store, user_id, messages, is_peeking=is_peeking ) - start_token = StreamToken.START.copy_and_replace("room_key", token[0]) - end_token = StreamToken.START.copy_and_replace("room_key", token[1]) + start_token = StreamToken.START.copy_and_replace("room_key", token) + end_token = StreamToken.START.copy_and_replace("room_key", stream_token) time_now = self.clock.time_msec() @@ -389,25 +389,28 @@ class InitialSyncHandler(BaseHandler): receipts = [] defer.returnValue(receipts) - presence, receipts, (messages, token) = yield defer.gatherResults( - [ - preserve_fn(get_presence)(), - preserve_fn(get_receipts)(), - preserve_fn(self.store.get_recent_events_for_room)( - room_id, - limit=limit, - end_token=now_token.room_key, - ) - ], - consumeErrors=True, - ).addErrback(unwrapFirstError) + presence, receipts, (messages, token) = yield make_deferred_yieldable( + defer.gatherResults( + [ + run_in_background(get_presence), + run_in_background(get_receipts), + run_in_background( + self.store.get_recent_events_for_room, + room_id, + limit=limit, + end_token=now_token.room_key, + ) + ], + consumeErrors=True, + ).addErrback(unwrapFirstError), + ) messages = yield filter_events_for_client( self.store, user_id, messages, is_peeking=is_peeking, ) - start_token = now_token.copy_and_replace("room_key", token[0]) - end_token = now_token.copy_and_replace("room_key", token[1]) + start_token = now_token.copy_and_replace("room_key", token) + end_token = now_token time_now = self.clock.time_msec() diff --git a/synapse/handlers/message.py b/synapse/handlers/message.py index 24c9ffdb2..39d772477 100644 --- a/synapse/handlers/message.py +++ b/synapse/handlers/message.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2014 - 2016 OpenMarket Ltd +# Copyright 2017 - 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -12,173 +13,185 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. +import logging +import sys + +import six +from six import iteritems, itervalues, string_types + +from canonicaljson import encode_canonical_json, json from twisted.internet import defer +from twisted.internet.defer import succeed -from synapse.api.constants import EventTypes, Membership -from synapse.api.errors import AuthError, Codes, SynapseError +from synapse.api.constants import MAX_DEPTH, EventTypes, Membership +from synapse.api.errors import AuthError, Codes, ConsentNotGivenError, SynapseError +from synapse.api.urls import ConsentURIBuilder from synapse.crypto.event_signing import add_hashes_and_signatures from synapse.events.utils import serialize_event from synapse.events.validator import EventValidator -from synapse.types import ( - UserID, RoomAlias, RoomStreamToken, -) -from synapse.util.async import run_on_reactor, ReadWriteLock, Limiter -from synapse.util.logcontext import preserve_fn +from synapse.replication.http.send_event import send_event_to_master +from synapse.types import RoomAlias, UserID +from synapse.util.async import Linearizer +from synapse.util.frozenutils import frozendict_json_encoder +from synapse.util.logcontext import run_in_background from synapse.util.metrics import measure_func -from synapse.visibility import filter_events_for_client from ._base import BaseHandler -from canonicaljson import encode_canonical_json - -import logging -import random -import ujson - logger = logging.getLogger(__name__) -class MessageHandler(BaseHandler): +class MessageHandler(object): + """Contains some read only APIs to get state about a room + """ def __init__(self, hs): - super(MessageHandler, self).__init__(hs) + self.auth = hs.get_auth() + self.clock = hs.get_clock() + self.state = hs.get_state_handler() + self.store = hs.get_datastore() + + @defer.inlineCallbacks + def get_room_data(self, user_id=None, room_id=None, + event_type=None, state_key="", is_guest=False): + """ Get data from a room. + + Args: + event : The room path event + Returns: + The path data content. + Raises: + SynapseError if something went wrong. + """ + membership, membership_event_id = yield self.auth.check_in_room_or_world_readable( + room_id, user_id + ) + + if membership == Membership.JOIN: + data = yield self.state.get_current_state( + room_id, event_type, state_key + ) + elif membership == Membership.LEAVE: + key = (event_type, state_key) + room_state = yield self.store.get_state_for_events( + [membership_event_id], [key] + ) + data = room_state[membership_event_id].get(key) + + defer.returnValue(data) + + @defer.inlineCallbacks + def get_state_events(self, user_id, room_id, is_guest=False): + """Retrieve all state events for a given room. If the user is + joined to the room then return the current state. If the user has + left the room return the state events from when they left. + + Args: + user_id(str): The user requesting state events. + room_id(str): The room ID to get all state events from. + Returns: + A list of dicts representing state events. [{}, {}, {}] + """ + membership, membership_event_id = yield self.auth.check_in_room_or_world_readable( + room_id, user_id + ) + + if membership == Membership.JOIN: + room_state = yield self.state.get_current_state(room_id) + elif membership == Membership.LEAVE: + room_state = yield self.store.get_state_for_events( + [membership_event_id], None + ) + room_state = room_state[membership_event_id] + + now = self.clock.time_msec() + defer.returnValue( + [serialize_event(c, now) for c in room_state.values()] + ) + + @defer.inlineCallbacks + def get_joined_members(self, requester, room_id): + """Get all the joined members in the room and their profile information. + + If the user has left the room return the state events from when they left. + + Args: + requester(Requester): The user requesting state events. + room_id(str): The room ID to get all state events from. + Returns: + A dict of user_id to profile info + """ + user_id = requester.user.to_string() + if not requester.app_service: + # We check AS auth after fetching the room membership, as it + # requires us to pull out all joined members anyway. + membership, _ = yield self.auth.check_in_room_or_world_readable( + room_id, user_id + ) + if membership != Membership.JOIN: + raise NotImplementedError( + "Getting joined members after leaving is not implemented" + ) + + users_with_profile = yield self.state.get_current_user_in_room(room_id) + + # If this is an AS, double check that they are allowed to see the members. + # This can either be because the AS user is in the room or because there + # is a user in the room that the AS is "interested in" + if requester.app_service and user_id not in users_with_profile: + for uid in users_with_profile: + if requester.app_service.is_interested_in_user(uid): + break + else: + # Loop fell through, AS has no interested users in room + raise AuthError(403, "Appservice not in room") + + defer.returnValue({ + user_id: { + "avatar_url": profile.avatar_url, + "display_name": profile.display_name, + } + for user_id, profile in iteritems(users_with_profile) + }) + + +class EventCreationHandler(object): + def __init__(self, hs): self.hs = hs + self.auth = hs.get_auth() + self.store = hs.get_datastore() self.state = hs.get_state_handler() self.clock = hs.get_clock() self.validator = EventValidator() + self.profile_handler = hs.get_profile_handler() + self.event_builder_factory = hs.get_event_builder_factory() + self.server_name = hs.hostname + self.ratelimiter = hs.get_ratelimiter() + self.notifier = hs.get_notifier() + self.config = hs.config - self.pagination_lock = ReadWriteLock() + self.http_client = hs.get_simple_http_client() + + # This is only used to get at ratelimit function, and maybe_kick_guest_users + self.base_handler = BaseHandler(hs) + + self.pusher_pool = hs.get_pusherpool() # We arbitrarily limit concurrent event creation for a room to 5. # This is to stop us from diverging history *too* much. - self.limiter = Limiter(max_count=5) + self.limiter = Linearizer(max_count=5, name="room_event_creation_limit") self.action_generator = hs.get_action_generator() - @defer.inlineCallbacks - def purge_history(self, room_id, event_id): - event = yield self.store.get_event(event_id) + self.spam_checker = hs.get_spam_checker() - if event.room_id != room_id: - raise SynapseError(400, "Event is for wrong room.") - - depth = event.depth - - with (yield self.pagination_lock.write(room_id)): - yield self.store.delete_old_state(room_id, depth) - - @defer.inlineCallbacks - def get_messages(self, requester, room_id=None, pagin_config=None, - as_client_event=True, event_filter=None): - """Get messages in a room. - - Args: - requester (Requester): The user requesting messages. - room_id (str): The room they want messages from. - pagin_config (synapse.api.streams.PaginationConfig): The pagination - config rules to apply, if any. - as_client_event (bool): True to get events in client-server format. - event_filter (Filter): Filter to apply to results or None - Returns: - dict: Pagination API results - """ - user_id = requester.user.to_string() - - if pagin_config.from_token: - room_token = pagin_config.from_token.room_key - else: - pagin_config.from_token = ( - yield self.hs.get_event_sources().get_current_token_for_room( - room_id=room_id - ) - ) - room_token = pagin_config.from_token.room_key - - room_token = RoomStreamToken.parse(room_token) - - pagin_config.from_token = pagin_config.from_token.copy_and_replace( - "room_key", str(room_token) - ) - - source_config = pagin_config.get_source_config("room") - - with (yield self.pagination_lock.read(room_id)): - membership, member_event_id = yield self._check_in_room_or_world_readable( - room_id, user_id - ) - - if source_config.direction == 'b': - # if we're going backwards, we might need to backfill. This - # requires that we have a topo token. - if room_token.topological: - max_topo = room_token.topological - else: - max_topo = yield self.store.get_max_topological_token( - room_id, room_token.stream - ) - - if membership == Membership.LEAVE: - # If they have left the room then clamp the token to be before - # they left the room, to save the effort of loading from the - # database. - leave_token = yield self.store.get_topological_token_for_event( - member_event_id - ) - leave_token = RoomStreamToken.parse(leave_token) - if leave_token.topological < max_topo: - source_config.from_key = str(leave_token) - - yield self.hs.get_handlers().federation_handler.maybe_backfill( - room_id, max_topo - ) - - events, next_key = yield self.store.paginate_room_events( - room_id=room_id, - from_key=source_config.from_key, - to_key=source_config.to_key, - direction=source_config.direction, - limit=source_config.limit, - event_filter=event_filter, - ) - - next_token = pagin_config.from_token.copy_and_replace( - "room_key", next_key - ) - - if not events: - defer.returnValue({ - "chunk": [], - "start": pagin_config.from_token.to_string(), - "end": next_token.to_string(), - }) - - if event_filter: - events = event_filter.filter(events) - - events = yield filter_events_for_client( - self.store, - user_id, - events, - is_peeking=(member_event_id is None), - ) - - time_now = self.clock.time_msec() - - chunk = { - "chunk": [ - serialize_event(e, time_now, as_client_event) - for e in events - ], - "start": pagin_config.from_token.to_string(), - "end": next_token.to_string(), - } - - defer.returnValue(chunk) + if self.config.block_events_without_consent_error is not None: + self._consent_uri_builder = ConsentURIBuilder(self.config) @defer.inlineCallbacks def create_event(self, requester, event_dict, token_id=None, txn_id=None, - prev_event_ids=None): + prev_events_and_hashes=None): """ Given a dict from a client, create a new event. @@ -192,50 +205,143 @@ class MessageHandler(BaseHandler): event_dict (dict): An entire event token_id (str) txn_id (str) - prev_event_ids (list): The prev event ids to use when creating the event + + prev_events_and_hashes (list[(str, dict[str, str], int)]|None): + the forward extremities to use as the prev_events for the + new event. For each event, a tuple of (event_id, hashes, depth) + where *hashes* is a map from algorithm to hash. + + If None, they will be requested from the database. Returns: Tuple of created event (FrozenEvent), Context """ builder = self.event_builder_factory.new(event_dict) - with (yield self.limiter.queue(builder.room_id)): - self.validator.validate_new(builder) + self.validator.validate_new(builder) - if builder.type == EventTypes.Member: - membership = builder.content.get("membership", None) - target = UserID.from_string(builder.state_key) + if builder.type == EventTypes.Member: + membership = builder.content.get("membership", None) + target = UserID.from_string(builder.state_key) - if membership in {Membership.JOIN, Membership.INVITE}: - # If event doesn't include a display name, add one. - profile = self.hs.get_handlers().profile_handler - content = builder.content + if membership in {Membership.JOIN, Membership.INVITE}: + # If event doesn't include a display name, add one. + profile = self.profile_handler + content = builder.content - try: - if "displayname" not in content: - content["displayname"] = yield profile.get_displayname(target) - if "avatar_url" not in content: - content["avatar_url"] = yield profile.get_avatar_url(target) - except Exception as e: - logger.info( - "Failed to get profile information for %r: %s", - target, e - ) + try: + if "displayname" not in content: + content["displayname"] = yield profile.get_displayname(target) + if "avatar_url" not in content: + content["avatar_url"] = yield profile.get_avatar_url(target) + except Exception as e: + logger.info( + "Failed to get profile information for %r: %s", + target, e + ) - if token_id is not None: - builder.internal_metadata.token_id = token_id + is_exempt = yield self._is_exempt_from_privacy_policy(builder, requester) + if not is_exempt: + yield self.assert_accepted_privacy_policy(requester) - if txn_id is not None: - builder.internal_metadata.txn_id = txn_id + if token_id is not None: + builder.internal_metadata.token_id = token_id - event, context = yield self._create_new_client_event( - builder=builder, - requester=requester, - prev_event_ids=prev_event_ids, - ) + if txn_id is not None: + builder.internal_metadata.txn_id = txn_id + + event, context = yield self.create_new_client_event( + builder=builder, + requester=requester, + prev_events_and_hashes=prev_events_and_hashes, + ) defer.returnValue((event, context)) + def _is_exempt_from_privacy_policy(self, builder, requester): + """"Determine if an event to be sent is exempt from having to consent + to the privacy policy + + Args: + builder (synapse.events.builder.EventBuilder): event being created + requester (Requster): user requesting this event + + Returns: + Deferred[bool]: true if the event can be sent without the user + consenting + """ + # the only thing the user can do is join the server notices room. + if builder.type == EventTypes.Member: + membership = builder.content.get("membership", None) + if membership == Membership.JOIN: + return self._is_server_notices_room(builder.room_id) + elif membership == Membership.LEAVE: + # the user is always allowed to leave (but not kick people) + return builder.state_key == requester.user.to_string() + return succeed(False) + + @defer.inlineCallbacks + def _is_server_notices_room(self, room_id): + if self.config.server_notices_mxid is None: + defer.returnValue(False) + user_ids = yield self.store.get_users_in_room(room_id) + defer.returnValue(self.config.server_notices_mxid in user_ids) + + @defer.inlineCallbacks + def assert_accepted_privacy_policy(self, requester): + """Check if a user has accepted the privacy policy + + Called when the given user is about to do something that requires + privacy consent. We see if the user is exempt and otherwise check that + they have given consent. If they have not, a ConsentNotGiven error is + raised. + + Args: + requester (synapse.types.Requester): + The user making the request + + Returns: + Deferred[None]: returns normally if the user has consented or is + exempt + + Raises: + ConsentNotGivenError: if the user has not given consent yet + """ + if self.config.block_events_without_consent_error is None: + return + + # exempt AS users from needing consent + if requester.app_service is not None: + return + + user_id = requester.user.to_string() + + # exempt the system notices user + if ( + self.config.server_notices_mxid is not None and + user_id == self.config.server_notices_mxid + ): + return + + u = yield self.store.get_user_by_id(user_id) + assert u is not None + if u["appservice_id"] is not None: + # users registered by an appservice are exempt + return + if u["consent_version"] == self.config.user_consent_version: + return + + consent_uri = self._consent_uri_builder.build_user_consent_uri( + requester.user.localpart, + ) + msg = self.config.block_events_without_consent_error % { + 'consent_uri': consent_uri, + } + raise ConsentNotGivenError( + msg=msg, + consent_uri=consent_uri, + ) + @defer.inlineCallbacks def send_nonmember_event(self, requester, event, context, ratelimit=True): """ @@ -253,11 +359,6 @@ class MessageHandler(BaseHandler): "Tried to send member event through non-member codepath" ) - # We check here if we are currently being rate limited, so that we - # don't do unnecessary work. We check again just before we actually - # send the event. - yield self.ratelimit(requester, update=False) - user = UserID.from_string(event.sender) assert self.hs.is_mine(user), "User must be our own: %s" % (user,) @@ -274,12 +375,6 @@ class MessageHandler(BaseHandler): ratelimit=ratelimit, ) - if event.type == EventTypes.Message: - presence = self.hs.get_presence_handler() - # We don't want to block sending messages on any presence code. This - # matters as sometimes presence code can take a while. - preserve_fn(presence.bump_presence_active_time)(user) - @defer.inlineCallbacks def deduplicate_state_event(self, event, context): """ @@ -288,7 +383,8 @@ class MessageHandler(BaseHandler): If so, returns the version of the event in context. Otherwise, returns None. """ - prev_event_id = context.prev_state_ids.get((event.type, event.state_key)) + prev_state_ids = yield context.get_prev_state_ids(self.store) + prev_event_id = prev_state_ids.get((event.type, event.state_key)) prev_event = yield self.store.get_event(prev_event_id, allow_none=True) if not prev_event: return @@ -313,145 +409,85 @@ class MessageHandler(BaseHandler): See self.create_event and self.send_nonmember_event. """ - event, context = yield self.create_event( - requester, - event_dict, - token_id=requester.access_token_id, - txn_id=txn_id - ) - yield self.send_nonmember_event( - requester, - event, - context, - ratelimit=ratelimit, - ) + + # We limit the number of concurrent event sends in a room so that we + # don't fork the DAG too much. If we don't limit then we can end up in + # a situation where event persistence can't keep up, causing + # extremities to pile up, which in turn leads to state resolution + # taking longer. + with (yield self.limiter.queue(event_dict["room_id"])): + event, context = yield self.create_event( + requester, + event_dict, + token_id=requester.access_token_id, + txn_id=txn_id + ) + + spam_error = self.spam_checker.check_event_for_spam(event) + if spam_error: + if not isinstance(spam_error, string_types): + spam_error = "Spam is not permitted here" + raise SynapseError( + 403, spam_error, Codes.FORBIDDEN + ) + + yield self.send_nonmember_event( + requester, + event, + context, + ratelimit=ratelimit, + ) defer.returnValue(event) + @measure_func("create_new_client_event") @defer.inlineCallbacks - def get_room_data(self, user_id=None, room_id=None, - event_type=None, state_key="", is_guest=False): - """ Get data from a room. + def create_new_client_event(self, builder, requester=None, + prev_events_and_hashes=None): + """Create a new event for a local client Args: - event : The room path event + builder (EventBuilder): + + requester (synapse.types.Requester|None): + + prev_events_and_hashes (list[(str, dict[str, str], int)]|None): + the forward extremities to use as the prev_events for the + new event. For each event, a tuple of (event_id, hashes, depth) + where *hashes* is a map from algorithm to hash. + + If None, they will be requested from the database. + Returns: - The path data content. - Raises: - SynapseError if something went wrong. + Deferred[(synapse.events.EventBase, synapse.events.snapshot.EventContext)] """ - membership, membership_event_id = yield self._check_in_room_or_world_readable( - room_id, user_id - ) - if membership == Membership.JOIN: - data = yield self.state_handler.get_current_state( - room_id, event_type, state_key + if prev_events_and_hashes is not None: + assert len(prev_events_and_hashes) <= 10, \ + "Attempting to create an event with %i prev_events" % ( + len(prev_events_and_hashes), ) - elif membership == Membership.LEAVE: - key = (event_type, state_key) - room_state = yield self.store.get_state_for_events( - [membership_event_id], [key] - ) - data = room_state[membership_event_id].get(key) - - defer.returnValue(data) - - @defer.inlineCallbacks - def _check_in_room_or_world_readable(self, room_id, user_id): - try: - # check_user_was_in_room will return the most recent membership - # event for the user if: - # * The user is a non-guest user, and was ever in the room - # * The user is a guest user, and has joined the room - # else it will throw. - member_event = yield self.auth.check_user_was_in_room(room_id, user_id) - defer.returnValue((member_event.membership, member_event.event_id)) - return - except AuthError: - visibility = yield self.state_handler.get_current_state( - room_id, EventTypes.RoomHistoryVisibility, "" - ) - if ( - visibility and - visibility.content["history_visibility"] == "world_readable" - ): - defer.returnValue((Membership.JOIN, None)) - return - raise AuthError( - 403, "Guest access not allowed", errcode=Codes.GUEST_ACCESS_FORBIDDEN - ) - - @defer.inlineCallbacks - def get_state_events(self, user_id, room_id, is_guest=False): - """Retrieve all state events for a given room. If the user is - joined to the room then return the current state. If the user has - left the room return the state events from when they left. - - Args: - user_id(str): The user requesting state events. - room_id(str): The room ID to get all state events from. - Returns: - A list of dicts representing state events. [{}, {}, {}] - """ - membership, membership_event_id = yield self._check_in_room_or_world_readable( - room_id, user_id - ) - - if membership == Membership.JOIN: - room_state = yield self.state_handler.get_current_state(room_id) - elif membership == Membership.LEAVE: - room_state = yield self.store.get_state_for_events( - [membership_event_id], None - ) - room_state = room_state[membership_event_id] - - now = self.clock.time_msec() - defer.returnValue( - [serialize_event(c, now) for c in room_state.values()] - ) - - @measure_func("_create_new_client_event") - @defer.inlineCallbacks - def _create_new_client_event(self, builder, requester=None, prev_event_ids=None): - if prev_event_ids: - prev_events = yield self.store.add_event_hashes(prev_event_ids) - prev_max_depth = yield self.store.get_max_depth_of_events(prev_event_ids) - depth = prev_max_depth + 1 else: - latest_ret = yield self.store.get_latest_event_ids_and_hashes_in_room( - builder.room_id, - ) + prev_events_and_hashes = \ + yield self.store.get_prev_events_for_room(builder.room_id) - # We want to limit the max number of prev events we point to in our - # new event - if len(latest_ret) > 10: - # Sort by reverse depth, so we point to the most recent. - latest_ret.sort(key=lambda a: -a[2]) - new_latest_ret = latest_ret[:5] + if prev_events_and_hashes: + depth = max([d for _, _, d in prev_events_and_hashes]) + 1 + # we cap depth of generated events, to ensure that they are not + # rejected by other servers (and so that they can be persisted in + # the db) + depth = min(depth, MAX_DEPTH) + else: + depth = 1 - # We also randomly point to some of the older events, to make - # sure that we don't completely ignore the older events. - if latest_ret[5:]: - sample_size = min(5, len(latest_ret[5:])) - new_latest_ret.extend(random.sample(latest_ret[5:], sample_size)) - latest_ret = new_latest_ret - - if latest_ret: - depth = max([d for _, _, d in latest_ret]) + 1 - else: - depth = 1 - - prev_events = [ - (event_id, prev_hashes) - for event_id, prev_hashes, _ in latest_ret - ] + prev_events = [ + (event_id, prev_hashes) + for event_id, prev_hashes, _ in prev_events_and_hashes + ] builder.prev_events = prev_events builder.depth = depth - state_handler = self.state_handler - - context = yield state_handler.compute_event_context(builder) + context = yield self.state.compute_event_context(builder) if requester: context.app_service = requester.app_service @@ -470,8 +506,8 @@ class MessageHandler(BaseHandler): event = builder.build() logger.debug( - "Created event %s with state: %s", - event.event_id, context.prev_state_ids, + "Created event %s", + event.event_id, ) defer.returnValue( @@ -486,12 +522,21 @@ class MessageHandler(BaseHandler): event, context, ratelimit=True, - extra_users=[] + extra_users=[], ): - # We now need to go and hit out to wherever we need to hit out to. + """Processes a new event. This includes checking auth, persisting it, + notifying users, sending to remote servers, etc. - if ratelimit: - yield self.ratelimit(requester) + If called from a worker will hit out to the master process for final + processing. + + Args: + requester (Requester) + event (FrozenEvent) + context (EventContext) + ratelimit (bool) + extra_users (list(UserID)): Any extra users to notify about event + """ try: yield self.auth.check_from_context(event, context) @@ -501,13 +546,72 @@ class MessageHandler(BaseHandler): # Ensure that we can round trip before trying to persist in db try: - dump = ujson.dumps(event.content) - ujson.loads(dump) - except: + dump = frozendict_json_encoder.encode(event.content) + json.loads(dump) + except Exception: logger.exception("Failed to encode content: %r", event.content) raise - yield self.maybe_kick_guest_users(event, context) + yield self.action_generator.handle_push_actions_for_event( + event, context + ) + + try: + # If we're a worker we need to hit out to the master. + if self.config.worker_app: + yield send_event_to_master( + clock=self.hs.get_clock(), + store=self.store, + client=self.http_client, + host=self.config.worker_replication_host, + port=self.config.worker_replication_http_port, + requester=requester, + event=event, + context=context, + ratelimit=ratelimit, + extra_users=extra_users, + ) + return + + yield self.persist_and_notify_client_event( + requester, + event, + context, + ratelimit=ratelimit, + extra_users=extra_users, + ) + except: # noqa: E722, as we reraise the exception this is fine. + # Ensure that we actually remove the entries in the push actions + # staging area, if we calculated them. + tp, value, tb = sys.exc_info() + + run_in_background( + self.store.remove_push_actions_from_staging, + event.event_id, + ) + + six.reraise(tp, value, tb) + + @defer.inlineCallbacks + def persist_and_notify_client_event( + self, + requester, + event, + context, + ratelimit=True, + extra_users=[], + ): + """Called when we have fully built the event, have already + calculated the push actions for the event, and checked auth. + + This should only be run on master. + """ + assert not self.config.worker_app + + if ratelimit: + yield self.base_handler.ratelimit(requester) + + yield self.base_handler.maybe_kick_guest_users(event, context) if event.type == EventTypes.CanonicalAlias: # Check the alias is acually valid (at this time at least) @@ -535,9 +639,11 @@ class MessageHandler(BaseHandler): e.sender == event.sender ) + current_state_ids = yield context.get_current_state_ids(self.store) + state_to_include_ids = [ e_id - for k, e_id in context.current_state_ids.iteritems() + for k, e_id in iteritems(current_state_ids) if k[0] in self.hs.config.room_invite_state_types or k == (EventTypes.Member, event.sender) ] @@ -551,7 +657,7 @@ class MessageHandler(BaseHandler): "content": e.content, "sender": e.sender, } - for e in state_to_include.itervalues() + for e in itervalues(state_to_include) ] invitee = UserID.from_string(event.state_key) @@ -573,8 +679,9 @@ class MessageHandler(BaseHandler): ) if event.type == EventTypes.Redaction: + prev_state_ids = yield context.get_prev_state_ids(self.store) auth_events_ids = yield self.auth.compute_auth_events( - event, context.prev_state_ids, for_verification=True, + event, prev_state_ids, for_verification=True, ) auth_events = yield self.store.get_events(auth_events_ids) auth_events = { @@ -594,15 +701,13 @@ class MessageHandler(BaseHandler): "You don't have permission to redact events" ) - if event.type == EventTypes.Create and context.prev_state_ids: - raise AuthError( - 403, - "Changing the room create event is forbidden", - ) - - yield self.action_generator.handle_push_actions_for_event( - event, context - ) + if event.type == EventTypes.Create: + prev_state_ids = yield context.get_prev_state_ids(self.store) + if prev_state_ids: + raise AuthError( + 403, + "Changing the room create event is forbidden", + ) (event_stream_id, max_stream_id) = yield self.store.persist_event( event, context=context @@ -610,16 +715,31 @@ class MessageHandler(BaseHandler): # this intentionally does not yield: we don't care about the result # and don't need to wait for it. - preserve_fn(self.hs.get_pusherpool().on_new_notifications)( + run_in_background( + self.pusher_pool.on_new_notifications, event_stream_id, max_stream_id ) - @defer.inlineCallbacks def _notify(): - yield run_on_reactor() - self.notifier.on_new_room_event( - event, event_stream_id, max_stream_id, - extra_users=extra_users - ) + try: + self.notifier.on_new_room_event( + event, event_stream_id, max_stream_id, + extra_users=extra_users + ) + except Exception: + logger.exception("Error notifying about new room event") - preserve_fn(_notify)() + run_in_background(_notify) + + if event.type == EventTypes.Message: + # We don't want to block sending messages on any presence code. This + # matters as sometimes presence code can take a while. + run_in_background(self._bump_active_time, requester.user) + + @defer.inlineCallbacks + def _bump_active_time(self, user): + try: + presence = self.hs.get_presence_handler() + yield presence.bump_presence_active_time(user) + except Exception: + logger.exception("Error bumping presence active time") diff --git a/synapse/handlers/pagination.py b/synapse/handlers/pagination.py new file mode 100644 index 000000000..b2849783e --- /dev/null +++ b/synapse/handlers/pagination.py @@ -0,0 +1,265 @@ +# -*- coding: utf-8 -*- +# Copyright 2014 - 2016 OpenMarket Ltd +# Copyright 2017 - 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import logging + +from twisted.internet import defer +from twisted.python.failure import Failure + +from synapse.api.constants import Membership +from synapse.api.errors import SynapseError +from synapse.events.utils import serialize_event +from synapse.types import RoomStreamToken +from synapse.util.async import ReadWriteLock +from synapse.util.logcontext import run_in_background +from synapse.util.stringutils import random_string +from synapse.visibility import filter_events_for_client + +logger = logging.getLogger(__name__) + + +class PurgeStatus(object): + """Object tracking the status of a purge request + + This class contains information on the progress of a purge request, for + return by get_purge_status. + + Attributes: + status (int): Tracks whether this request has completed. One of + STATUS_{ACTIVE,COMPLETE,FAILED} + """ + + STATUS_ACTIVE = 0 + STATUS_COMPLETE = 1 + STATUS_FAILED = 2 + + STATUS_TEXT = { + STATUS_ACTIVE: "active", + STATUS_COMPLETE: "complete", + STATUS_FAILED: "failed", + } + + def __init__(self): + self.status = PurgeStatus.STATUS_ACTIVE + + def asdict(self): + return { + "status": PurgeStatus.STATUS_TEXT[self.status] + } + + +class PaginationHandler(object): + """Handles pagination and purge history requests. + + These are in the same handler due to the fact we need to block clients + paginating during a purge. + """ + + def __init__(self, hs): + self.hs = hs + self.auth = hs.get_auth() + self.store = hs.get_datastore() + self.clock = hs.get_clock() + + self.pagination_lock = ReadWriteLock() + self._purges_in_progress_by_room = set() + # map from purge id to PurgeStatus + self._purges_by_id = {} + + def start_purge_history(self, room_id, token, + delete_local_events=False): + """Start off a history purge on a room. + + Args: + room_id (str): The room to purge from + + token (str): topological token to delete events before + delete_local_events (bool): True to delete local events as well as + remote ones + + Returns: + str: unique ID for this purge transaction. + """ + if room_id in self._purges_in_progress_by_room: + raise SynapseError( + 400, + "History purge already in progress for %s" % (room_id, ), + ) + + purge_id = random_string(16) + + # we log the purge_id here so that it can be tied back to the + # request id in the log lines. + logger.info("[purge] starting purge_id %s", purge_id) + + self._purges_by_id[purge_id] = PurgeStatus() + run_in_background( + self._purge_history, + purge_id, room_id, token, delete_local_events, + ) + return purge_id + + @defer.inlineCallbacks + def _purge_history(self, purge_id, room_id, token, + delete_local_events): + """Carry out a history purge on a room. + + Args: + purge_id (str): The id for this purge + room_id (str): The room to purge from + token (str): topological token to delete events before + delete_local_events (bool): True to delete local events as well as + remote ones + + Returns: + Deferred + """ + self._purges_in_progress_by_room.add(room_id) + try: + with (yield self.pagination_lock.write(room_id)): + yield self.store.purge_history( + room_id, token, delete_local_events, + ) + logger.info("[purge] complete") + self._purges_by_id[purge_id].status = PurgeStatus.STATUS_COMPLETE + except Exception: + logger.error("[purge] failed: %s", Failure().getTraceback().rstrip()) + self._purges_by_id[purge_id].status = PurgeStatus.STATUS_FAILED + finally: + self._purges_in_progress_by_room.discard(room_id) + + # remove the purge from the list 24 hours after it completes + def clear_purge(): + del self._purges_by_id[purge_id] + self.hs.get_reactor().callLater(24 * 3600, clear_purge) + + def get_purge_status(self, purge_id): + """Get the current status of an active purge + + Args: + purge_id (str): purge_id returned by start_purge_history + + Returns: + PurgeStatus|None + """ + return self._purges_by_id.get(purge_id) + + @defer.inlineCallbacks + def get_messages(self, requester, room_id=None, pagin_config=None, + as_client_event=True, event_filter=None): + """Get messages in a room. + + Args: + requester (Requester): The user requesting messages. + room_id (str): The room they want messages from. + pagin_config (synapse.api.streams.PaginationConfig): The pagination + config rules to apply, if any. + as_client_event (bool): True to get events in client-server format. + event_filter (Filter): Filter to apply to results or None + Returns: + dict: Pagination API results + """ + user_id = requester.user.to_string() + + if pagin_config.from_token: + room_token = pagin_config.from_token.room_key + else: + pagin_config.from_token = ( + yield self.hs.get_event_sources().get_current_token_for_room( + room_id=room_id + ) + ) + room_token = pagin_config.from_token.room_key + + room_token = RoomStreamToken.parse(room_token) + + pagin_config.from_token = pagin_config.from_token.copy_and_replace( + "room_key", str(room_token) + ) + + source_config = pagin_config.get_source_config("room") + + with (yield self.pagination_lock.read(room_id)): + membership, member_event_id = yield self.auth.check_in_room_or_world_readable( + room_id, user_id + ) + + if source_config.direction == 'b': + # if we're going backwards, we might need to backfill. This + # requires that we have a topo token. + if room_token.topological: + max_topo = room_token.topological + else: + max_topo = yield self.store.get_max_topological_token( + room_id, room_token.stream + ) + + if membership == Membership.LEAVE: + # If they have left the room then clamp the token to be before + # they left the room, to save the effort of loading from the + # database. + leave_token = yield self.store.get_topological_token_for_event( + member_event_id + ) + leave_token = RoomStreamToken.parse(leave_token) + if leave_token.topological < max_topo: + source_config.from_key = str(leave_token) + + yield self.hs.get_handlers().federation_handler.maybe_backfill( + room_id, max_topo + ) + + events, next_key = yield self.store.paginate_room_events( + room_id=room_id, + from_key=source_config.from_key, + to_key=source_config.to_key, + direction=source_config.direction, + limit=source_config.limit, + event_filter=event_filter, + ) + + next_token = pagin_config.from_token.copy_and_replace( + "room_key", next_key + ) + + if not events: + defer.returnValue({ + "chunk": [], + "start": pagin_config.from_token.to_string(), + "end": next_token.to_string(), + }) + + if event_filter: + events = event_filter.filter(events) + + events = yield filter_events_for_client( + self.store, + user_id, + events, + is_peeking=(member_event_id is None), + ) + + time_now = self.clock.time_msec() + + chunk = { + "chunk": [ + serialize_event(e, time_now, as_client_event) + for e in events + ], + "start": pagin_config.from_token.to_string(), + "end": next_token.to_string(), + } + + defer.returnValue(chunk) diff --git a/synapse/handlers/presence.py b/synapse/handlers/presence.py index c7c0b0a1e..373283019 100644 --- a/synapse/handlers/presence.py +++ b/synapse/handlers/presence.py @@ -22,41 +22,44 @@ The methods that define policy are: - should_notify """ -from twisted.internet import defer, reactor +import logging from contextlib import contextmanager -from synapse.api.errors import SynapseError -from synapse.api.constants import PresenceState -from synapse.storage.presence import UserPresenceState +from six import iteritems, itervalues -from synapse.util.caches.descriptors import cachedInlineCallbacks +from prometheus_client import Counter + +from twisted.internet import defer + +from synapse.api.constants import PresenceState +from synapse.api.errors import SynapseError +from synapse.metrics import LaterGauge +from synapse.storage.presence import UserPresenceState +from synapse.types import UserID, get_domain_from_id from synapse.util.async import Linearizer -from synapse.util.logcontext import preserve_fn +from synapse.util.caches.descriptors import cachedInlineCallbacks +from synapse.util.logcontext import run_in_background from synapse.util.logutils import log_function from synapse.util.metrics import Measure from synapse.util.wheel_timer import WheelTimer -from synapse.types import UserID, get_domain_from_id -import synapse.metrics - -import logging - logger = logging.getLogger(__name__) -metrics = synapse.metrics.get_metrics_for(__name__) -notified_presence_counter = metrics.register_counter("notified_presence") -federation_presence_out_counter = metrics.register_counter("federation_presence_out") -presence_updates_counter = metrics.register_counter("presence_updates") -timers_fired_counter = metrics.register_counter("timers_fired") -federation_presence_counter = metrics.register_counter("federation_presence") -bump_active_time_counter = metrics.register_counter("bump_active_time") +notified_presence_counter = Counter("synapse_handler_presence_notified_presence", "") +federation_presence_out_counter = Counter( + "synapse_handler_presence_federation_presence_out", "") +presence_updates_counter = Counter("synapse_handler_presence_presence_updates", "") +timers_fired_counter = Counter("synapse_handler_presence_timers_fired", "") +federation_presence_counter = Counter("synapse_handler_presence_federation_presence", "") +bump_active_time_counter = Counter("synapse_handler_presence_bump_active_time", "") -get_updates_counter = metrics.register_counter("get_updates", labels=["type"]) +get_updates_counter = Counter("synapse_handler_presence_get_updates", "", ["type"]) -notify_reason_counter = metrics.register_counter("notify_reason", labels=["reason"]) -state_transition_counter = metrics.register_counter( - "state_transition", labels=["from", "to"] +notify_reason_counter = Counter( + "synapse_handler_presence_notify_reason", "", ["reason"]) +state_transition_counter = Counter( + "synapse_handler_presence_state_transition", "", ["from", "to"] ) @@ -87,35 +90,40 @@ assert LAST_ACTIVE_GRANULARITY < IDLE_TIMER class PresenceHandler(object): def __init__(self, hs): + """ + + Args: + hs (synapse.server.HomeServer): + """ self.is_mine = hs.is_mine self.is_mine_id = hs.is_mine_id self.clock = hs.get_clock() self.store = hs.get_datastore() self.wheel_timer = WheelTimer() self.notifier = hs.get_notifier() - self.replication = hs.get_replication_layer() self.federation = hs.get_federation_sender() - self.state = hs.get_state_handler() - self.replication.register_edu_handler( + federation_registry = hs.get_federation_registry() + + federation_registry.register_edu_handler( "m.presence", self.incoming_presence ) - self.replication.register_edu_handler( + federation_registry.register_edu_handler( "m.presence_invite", lambda origin, content: self.invite_presence( observed_user=UserID.from_string(content["observed_user"]), observer_user=UserID.from_string(content["observer_user"]), ) ) - self.replication.register_edu_handler( + federation_registry.register_edu_handler( "m.presence_accept", lambda origin, content: self.accept_presence( observed_user=UserID.from_string(content["observed_user"]), observer_user=UserID.from_string(content["observer_user"]), ) ) - self.replication.register_edu_handler( + federation_registry.register_edu_handler( "m.presence_deny", lambda origin, content: self.deny_presence( observed_user=UserID.from_string(content["observed_user"]), @@ -136,8 +144,9 @@ class PresenceHandler(object): for state in active_presence } - metrics.register_callback( - "user_to_current_state_size", lambda: len(self.user_to_current_state) + LaterGauge( + "synapse_handlers_presence_user_to_current_state_size", "", [], + lambda: len(self.user_to_current_state) ) now = self.clock.time_msec() @@ -169,7 +178,7 @@ class PresenceHandler(object): # have not yet been persisted self.unpersisted_users_changes = set() - reactor.addSystemEventTrigger("before", "shutdown", self._on_shutdown) + hs.get_reactor().addSystemEventTrigger("before", "shutdown", self._on_shutdown) self.serial_to_user = {} self._next_serial = 1 @@ -207,7 +216,8 @@ class PresenceHandler(object): 60 * 1000, ) - metrics.register_callback("wheel_timer_size", lambda: len(self.wheel_timer)) + LaterGauge("synapse_handlers_presence_wheel_timer_size", "", [], + lambda: len(self.wheel_timer)) @defer.inlineCallbacks def _on_shutdown(self): @@ -253,6 +263,14 @@ class PresenceHandler(object): logger.info("Finished _persist_unpersisted_changes") + @defer.inlineCallbacks + def _update_states_and_catch_exception(self, new_states): + try: + res = yield self._update_states(new_states) + defer.returnValue(res) + except Exception: + logger.exception("Error updating presence") + @defer.inlineCallbacks def _update_states(self, new_states): """Updates presence of users. Sets the appropriate timeouts. Pokes @@ -302,11 +320,11 @@ class PresenceHandler(object): # TODO: We should probably ensure there are no races hereafter - presence_updates_counter.inc_by(len(new_states)) + presence_updates_counter.inc(len(new_states)) if to_notify: - notified_presence_counter.inc_by(len(to_notify)) - yield self._persist_and_notify(to_notify.values()) + notified_presence_counter.inc(len(to_notify)) + yield self._persist_and_notify(list(to_notify.values())) self.unpersisted_users_changes |= set(s.user_id for s in new_states) self.unpersisted_users_changes -= set(to_notify.keys()) @@ -316,7 +334,7 @@ class PresenceHandler(object): if user_id not in to_notify } if to_federation_ping: - federation_presence_out_counter.inc_by(len(to_federation_ping)) + federation_presence_out_counter.inc(len(to_federation_ping)) self._push_to_remotes(to_federation_ping.values()) @@ -354,7 +372,7 @@ class PresenceHandler(object): for user_id in users_to_check ] - timers_fired_counter.inc_by(len(states)) + timers_fired_counter.inc(len(states)) changes = handle_timeouts( states, @@ -363,8 +381,8 @@ class PresenceHandler(object): now=now, ) - preserve_fn(self._update_states)(changes) - except: + run_in_background(self._update_states_and_catch_exception, changes) + except Exception: logger.exception("Exception in _handle_timeouts loop") @defer.inlineCallbacks @@ -421,20 +439,23 @@ class PresenceHandler(object): @defer.inlineCallbacks def _end(): - if affect_presence: + try: self.user_to_num_current_syncs[user_id] -= 1 prev_state = yield self.current_state_for_user(user_id) yield self._update_states([prev_state.copy_and_replace( last_user_sync_ts=self.clock.time_msec(), )]) + except Exception: + logger.exception("Error updating presence after sync") @contextmanager def _user_syncing(): try: yield finally: - preserve_fn(_end)() + if affect_presence: + run_in_background(_end) defer.returnValue(_user_syncing()) @@ -451,61 +472,6 @@ class PresenceHandler(object): syncing_user_ids.update(user_ids) return syncing_user_ids - @defer.inlineCallbacks - def update_external_syncs(self, process_id, syncing_user_ids): - """Update the syncing users for an external process - - Args: - process_id(str): An identifier for the process the users are - syncing against. This allows synapse to process updates - as user start and stop syncing against a given process. - syncing_user_ids(set(str)): The set of user_ids that are - currently syncing on that server. - """ - - # Grab the previous list of user_ids that were syncing on that process - prev_syncing_user_ids = ( - self.external_process_to_current_syncs.get(process_id, set()) - ) - # Grab the current presence state for both the users that are syncing - # now and the users that were syncing before this update. - prev_states = yield self.current_state_for_users( - syncing_user_ids | prev_syncing_user_ids - ) - updates = [] - time_now_ms = self.clock.time_msec() - - # For each new user that is syncing check if we need to mark them as - # being online. - for new_user_id in syncing_user_ids - prev_syncing_user_ids: - prev_state = prev_states[new_user_id] - if prev_state.state == PresenceState.OFFLINE: - updates.append(prev_state.copy_and_replace( - state=PresenceState.ONLINE, - last_active_ts=time_now_ms, - last_user_sync_ts=time_now_ms, - )) - else: - updates.append(prev_state.copy_and_replace( - last_user_sync_ts=time_now_ms, - )) - - # For each user that is still syncing or stopped syncing update the - # last sync time so that we will correctly apply the grace period when - # they stop syncing. - for old_user_id in prev_syncing_user_ids: - prev_state = prev_states[old_user_id] - updates.append(prev_state.copy_and_replace( - last_user_sync_ts=time_now_ms, - )) - - yield self._update_states(updates) - - # Update the last updated time for the process. We expire the entries - # if we don't receive an update in the given timeframe. - self.external_process_last_updated_ms[process_id] = self.clock.time_msec() - self.external_process_to_current_syncs[process_id] = syncing_user_ids - @defer.inlineCallbacks def update_external_syncs_row(self, process_id, user_id, is_syncing, sync_time_msec): """Update the syncing users for an external process as a delta. @@ -569,7 +535,7 @@ class PresenceHandler(object): prev_state.copy_and_replace( last_user_sync_ts=time_now_ms, ) - for prev_state in prev_states.itervalues() + for prev_state in itervalues(prev_states) ]) self.external_process_last_updated_ms.pop(process_id, None) @@ -592,14 +558,14 @@ class PresenceHandler(object): for user_id in user_ids } - missing = [user_id for user_id, state in states.iteritems() if not state] + missing = [user_id for user_id, state in iteritems(states) if not state] if missing: # There are things not in our in memory cache. Lets pull them out of # the database. res = yield self.store.get_presence_for_users(missing) states.update(res) - missing = [user_id for user_id, state in states.iteritems() if not state] + missing = [user_id for user_id, state in iteritems(states) if not state] if missing: new = { user_id: UserPresenceState.default(user_id) @@ -695,7 +661,7 @@ class PresenceHandler(object): updates.append(prev_state.copy_and_replace(**new_fields)) if updates: - federation_presence_counter.inc_by(len(updates)) + federation_presence_counter.inc(len(updates)) yield self._update_states(updates) @defer.inlineCallbacks @@ -720,7 +686,7 @@ class PresenceHandler(object): """ updates = yield self.current_state_for_users(target_user_ids) - updates = updates.values() + updates = list(updates.values()) for user_id in set(target_user_ids) - set(u.user_id for u in updates): updates.append(UserPresenceState.default(user_id)) @@ -786,11 +752,11 @@ class PresenceHandler(object): self._push_to_remotes([state]) else: user_ids = yield self.store.get_users_in_room(room_id) - user_ids = filter(self.is_mine_id, user_ids) + user_ids = list(filter(self.is_mine_id, user_ids)) states = yield self.current_state_for_users(user_ids) - self._push_to_remotes(states.values()) + self._push_to_remotes(list(states.values())) @defer.inlineCallbacks def get_presence_list(self, observer_user, accepted=None): @@ -970,28 +936,28 @@ def should_notify(old_state, new_state): return False if old_state.status_msg != new_state.status_msg: - notify_reason_counter.inc("status_msg_change") + notify_reason_counter.labels("status_msg_change").inc() return True if old_state.state != new_state.state: - notify_reason_counter.inc("state_change") - state_transition_counter.inc(old_state.state, new_state.state) + notify_reason_counter.labels("state_change").inc() + state_transition_counter.labels(old_state.state, new_state.state).inc() return True if old_state.state == PresenceState.ONLINE: if new_state.currently_active != old_state.currently_active: - notify_reason_counter.inc("current_active_change") + notify_reason_counter.labels("current_active_change").inc() return True if new_state.last_active_ts - old_state.last_active_ts > LAST_ACTIVE_GRANULARITY: # Only notify about last active bumps if we're not currently acive if not new_state.currently_active: - notify_reason_counter.inc("last_active_change_online") + notify_reason_counter.labels("last_active_change_online").inc() return True elif new_state.last_active_ts - old_state.last_active_ts > LAST_ACTIVE_GRANULARITY: # Always notify for a transition where last active gets bumped. - notify_reason_counter.inc("last_active_change_not_online") + notify_reason_counter.labels("last_active_change_not_online").inc() return True return False @@ -1065,14 +1031,14 @@ class PresenceEventSource(object): if changed is not None and len(changed) < 500: # For small deltas, its quicker to get all changes and then # work out if we share a room or they're in our presence list - get_updates_counter.inc("stream") + get_updates_counter.labels("stream").inc() for other_user_id in changed: if other_user_id in users_interested_in: user_ids_changed.add(other_user_id) else: # Too many possible updates. Find all users we can see and check # if any of them have changed. - get_updates_counter.inc("full") + get_updates_counter.labels("full").inc() if from_key: user_ids_changed = stream_change_cache.get_entities_changed( @@ -1084,10 +1050,10 @@ class PresenceEventSource(object): updates = yield presence.current_state_for_users(user_ids_changed) if include_offline: - defer.returnValue((updates.values(), max_token)) + defer.returnValue((list(updates.values()), max_token)) else: defer.returnValue(([ - s for s in updates.itervalues() + s for s in itervalues(updates) if s.state != PresenceState.OFFLINE ], max_token)) @@ -1145,7 +1111,7 @@ def handle_timeouts(user_states, is_mine_fn, syncing_user_ids, now): if new_state: changes[state.user_id] = new_state - return changes.values() + return list(changes.values()) def handle_timeout(state, is_mine, syncing_user_ids, now): @@ -1199,7 +1165,7 @@ def handle_timeout(state, is_mine, syncing_user_ids, now): ) changed = True else: - # We expect to be poked occaisonally by the other side. + # We expect to be poked occasionally by the other side. # This is to protect against forgetful/buggy servers, so that # no one gets stuck online forever. if now - state.last_federation_update_ts > FEDERATION_TIMEOUT: @@ -1344,11 +1310,11 @@ def get_interested_remotes(store, states, state_handler): # hosts in those rooms. room_ids_to_states, users_to_states = yield get_interested_parties(store, states) - for room_id, states in room_ids_to_states.iteritems(): + for room_id, states in iteritems(room_ids_to_states): hosts = yield state_handler.get_current_hosts_in_room(room_id) hosts_and_states.append((hosts, states)) - for user_id, states in users_to_states.iteritems(): + for user_id, states in iteritems(users_to_states): host = get_domain_from_id(user_id) hosts_and_states.append(([host], states)) diff --git a/synapse/handlers/profile.py b/synapse/handlers/profile.py index 7abee98de..859f6d2b2 100644 --- a/synapse/handlers/profile.py +++ b/synapse/handlers/profile.py @@ -17,25 +17,88 @@ import logging from twisted.internet import defer -import synapse.types -from synapse.api.errors import SynapseError, AuthError, CodeMessageException -from synapse.types import UserID -from ._base import BaseHandler +from synapse.api.errors import AuthError, CodeMessageException, SynapseError +from synapse.types import UserID, get_domain_from_id +from ._base import BaseHandler logger = logging.getLogger(__name__) class ProfileHandler(BaseHandler): + PROFILE_UPDATE_MS = 60 * 1000 + PROFILE_UPDATE_EVERY_MS = 24 * 60 * 60 * 1000 def __init__(self, hs): super(ProfileHandler, self).__init__(hs) - self.federation = hs.get_replication_layer() - self.federation.register_query_handler( + self.federation = hs.get_federation_client() + hs.get_federation_registry().register_query_handler( "profile", self.on_profile_query ) + self.user_directory_handler = hs.get_user_directory_handler() + + if hs.config.worker_app is None: + self.clock.looping_call( + self._update_remote_profile_cache, self.PROFILE_UPDATE_MS, + ) + + @defer.inlineCallbacks + def get_profile(self, user_id): + target_user = UserID.from_string(user_id) + if self.hs.is_mine(target_user): + displayname = yield self.store.get_profile_displayname( + target_user.localpart + ) + avatar_url = yield self.store.get_profile_avatar_url( + target_user.localpart + ) + + defer.returnValue({ + "displayname": displayname, + "avatar_url": avatar_url, + }) + else: + try: + result = yield self.federation.make_query( + destination=target_user.domain, + query_type="profile", + args={ + "user_id": user_id, + }, + ignore_backoff=True, + ) + defer.returnValue(result) + except CodeMessageException as e: + if e.code != 404: + logger.exception("Failed to get displayname") + + raise + + @defer.inlineCallbacks + def get_profile_from_cache(self, user_id): + """Get the profile information from our local cache. If the user is + ours then the profile information will always be corect. Otherwise, + it may be out of date/missing. + """ + target_user = UserID.from_string(user_id) + if self.hs.is_mine(target_user): + displayname = yield self.store.get_profile_displayname( + target_user.localpart + ) + avatar_url = yield self.store.get_profile_avatar_url( + target_user.localpart + ) + + defer.returnValue({ + "displayname": displayname, + "avatar_url": avatar_url, + }) + else: + profile = yield self.store.get_from_remote_profile_cache(user_id) + defer.returnValue(profile or {}) + @defer.inlineCallbacks def get_displayname(self, target_user): if self.hs.is_mine(target_user): @@ -60,7 +123,7 @@ class ProfileHandler(BaseHandler): logger.exception("Failed to get displayname") raise - except: + except Exception: logger.exception("Failed to get displayname") else: defer.returnValue(result["displayname"]) @@ -82,7 +145,13 @@ class ProfileHandler(BaseHandler): target_user.localpart, new_displayname ) - yield self._update_join_states(requester) + if self.hs.config.user_directory_search_all_users: + profile = yield self.store.get_profileinfo(target_user.localpart) + yield self.user_directory_handler.handle_local_profile_change( + target_user.to_string(), profile + ) + + yield self._update_join_states(requester, target_user) @defer.inlineCallbacks def get_avatar_url(self, target_user): @@ -107,7 +176,7 @@ class ProfileHandler(BaseHandler): if e.code != 404: logger.exception("Failed to get avatar_url") raise - except: + except Exception: logger.exception("Failed to get avatar_url") defer.returnValue(result["avatar_url"]) @@ -126,7 +195,13 @@ class ProfileHandler(BaseHandler): target_user.localpart, new_avatar_url ) - yield self._update_join_states(requester) + if self.hs.config.user_directory_search_all_users: + profile = yield self.store.get_profileinfo(target_user.localpart) + yield self.user_directory_handler.handle_local_profile_change( + target_user.to_string(), profile + ) + + yield self._update_join_states(requester, target_user) @defer.inlineCallbacks def on_profile_query(self, args): @@ -151,28 +226,24 @@ class ProfileHandler(BaseHandler): defer.returnValue(response) @defer.inlineCallbacks - def _update_join_states(self, requester): - user = requester.user - if not self.hs.is_mine(user): + def _update_join_states(self, requester, target_user): + if not self.hs.is_mine(target_user): return yield self.ratelimit(requester) room_ids = yield self.store.get_rooms_for_user( - user.to_string(), + target_user.to_string(), ) for room_id in room_ids: - handler = self.hs.get_handlers().room_member_handler + handler = self.hs.get_room_member_handler() try: - # Assume the user isn't a guest because we don't let guests set - # profile or avatar data. - # XXX why are we recreating `requester` here for each room? - # what was wrong with the `requester` we were passed? - requester = synapse.types.create_requester(user) + # Assume the target_user isn't a guest, + # because we don't let guests set profile or avatar data. yield handler.update_membership( requester, - user, + target_user, room_id, "join", # We treat a profile update like a join. ratelimit=False, # Try to hide that these events aren't atomic. @@ -182,3 +253,44 @@ class ProfileHandler(BaseHandler): "Failed to update join event for room %s - %s", room_id, str(e.message) ) + + def _update_remote_profile_cache(self): + """Called periodically to check profiles of remote users we haven't + checked in a while. + """ + entries = yield self.store.get_remote_profile_cache_entries_that_expire( + last_checked=self.clock.time_msec() - self.PROFILE_UPDATE_EVERY_MS + ) + + for user_id, displayname, avatar_url in entries: + is_subscribed = yield self.store.is_subscribed_remote_profile_for_user( + user_id, + ) + if not is_subscribed: + yield self.store.maybe_delete_remote_profile_cache(user_id) + continue + + try: + profile = yield self.federation.make_query( + destination=get_domain_from_id(user_id), + query_type="profile", + args={ + "user_id": user_id, + }, + ignore_backoff=True, + ) + except Exception: + logger.exception("Failed to get avatar_url") + + yield self.store.update_remote_profile_cache( + user_id, displayname, avatar_url + ) + continue + + new_name = profile.get("displayname") + new_avatar = profile.get("avatar_url") + + # We always hit update to update the last_check timestamp + yield self.store.update_remote_profile_cache( + user_id, new_name, new_avatar + ) diff --git a/synapse/handlers/read_marker.py b/synapse/handlers/read_marker.py index b5b0303d5..995460f82 100644 --- a/synapse/handlers/read_marker.py +++ b/synapse/handlers/read_marker.py @@ -13,13 +13,14 @@ # See the License for the specific language governing permissions and # limitations under the License. -from ._base import BaseHandler +import logging from twisted.internet import defer from synapse.util.async import Linearizer -import logging +from ._base import BaseHandler + logger = logging.getLogger(__name__) @@ -41,9 +42,9 @@ class ReadMarkerHandler(BaseHandler): """ with (yield self.read_marker_linearizer.queue((room_id, user_id))): - account_data = yield self.store.get_account_data_for_room(user_id, room_id) - - existing_read_marker = account_data.get("m.fully_read", None) + existing_read_marker = yield self.store.get_account_data_for_room_and_type( + user_id, room_id, "m.fully_read", + ) should_update = True diff --git a/synapse/handlers/receipts.py b/synapse/handlers/receipts.py index e1cd3a48e..cb905a390 100644 --- a/synapse/handlers/receipts.py +++ b/synapse/handlers/receipts.py @@ -12,16 +12,15 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. - -from ._base import BaseHandler +import logging from twisted.internet import defer -from synapse.util.logcontext import PreserveLoggingContext from synapse.types import get_domain_from_id +from synapse.util import logcontext +from synapse.util.logcontext import PreserveLoggingContext -import logging - +from ._base import BaseHandler logger = logging.getLogger(__name__) @@ -34,7 +33,7 @@ class ReceiptsHandler(BaseHandler): self.store = hs.get_datastore() self.hs = hs self.federation = hs.get_federation_sender() - hs.get_replication_layer().register_edu_handler( + hs.get_federation_registry().register_edu_handler( "m.receipt", self._received_remote_receipt ) self.clock = self.hs.get_clock() @@ -59,6 +58,8 @@ class ReceiptsHandler(BaseHandler): is_new = yield self._handle_new_receipts([receipt]) if is_new: + # fire off a process in the background to send the receipt to + # remote servers self._push_remotes([receipt]) @defer.inlineCallbacks @@ -126,42 +127,46 @@ class ReceiptsHandler(BaseHandler): defer.returnValue(True) + @logcontext.preserve_fn # caller should not yield on this @defer.inlineCallbacks def _push_remotes(self, receipts): """Given a list of receipts, works out which remote servers should be poked and pokes them. """ - # TODO: Some of this stuff should be coallesced. - for receipt in receipts: - room_id = receipt["room_id"] - receipt_type = receipt["receipt_type"] - user_id = receipt["user_id"] - event_ids = receipt["event_ids"] - data = receipt["data"] + try: + # TODO: Some of this stuff should be coallesced. + for receipt in receipts: + room_id = receipt["room_id"] + receipt_type = receipt["receipt_type"] + user_id = receipt["user_id"] + event_ids = receipt["event_ids"] + data = receipt["data"] - users = yield self.state.get_current_user_in_room(room_id) - remotedomains = set(get_domain_from_id(u) for u in users) - remotedomains = remotedomains.copy() - remotedomains.discard(self.server_name) + users = yield self.state.get_current_user_in_room(room_id) + remotedomains = set(get_domain_from_id(u) for u in users) + remotedomains = remotedomains.copy() + remotedomains.discard(self.server_name) - logger.debug("Sending receipt to: %r", remotedomains) + logger.debug("Sending receipt to: %r", remotedomains) - for domain in remotedomains: - self.federation.send_edu( - destination=domain, - edu_type="m.receipt", - content={ - room_id: { - receipt_type: { - user_id: { - "event_ids": event_ids, - "data": data, + for domain in remotedomains: + self.federation.send_edu( + destination=domain, + edu_type="m.receipt", + content={ + room_id: { + receipt_type: { + user_id: { + "event_ids": event_ids, + "data": data, + } } - } + }, }, - }, - key=(room_id, receipt_type, user_id), - ) + key=(room_id, receipt_type, user_id), + ) + except Exception: + logger.exception("Error pushing receipts to remote servers") @defer.inlineCallbacks def get_receipts_for_room(self, room_id, to_key): diff --git a/synapse/handlers/register.py b/synapse/handlers/register.py index ee3a2269a..7caff0cbc 100644 --- a/synapse/handlers/register.py +++ b/synapse/handlers/register.py @@ -15,16 +15,22 @@ """Contains functions for registering clients.""" import logging -import urllib from twisted.internet import defer +from synapse import types from synapse.api.errors import ( - AuthError, Codes, SynapseError, RegistrationError, InvalidCaptchaError + AuthError, + Codes, + InvalidCaptchaError, + RegistrationError, + SynapseError, ) from synapse.http.client import CaptchaServerHttpClient -from synapse.types import UserID -from synapse.util.async import run_on_reactor +from synapse.types import RoomAlias, RoomID, UserID, create_requester +from synapse.util.async import Linearizer +from synapse.util.threepids import check_3pid_allowed + from ._base import BaseHandler logger = logging.getLogger(__name__) @@ -33,24 +39,35 @@ logger = logging.getLogger(__name__) class RegistrationHandler(BaseHandler): def __init__(self, hs): + """ + + Args: + hs (synapse.server.HomeServer): + """ super(RegistrationHandler, self).__init__(hs) self.auth = hs.get_auth() + self._auth_handler = hs.get_auth_handler() + self.profile_handler = hs.get_profile_handler() + self.user_directory_handler = hs.get_user_directory_handler() self.captcha_client = CaptchaServerHttpClient(hs) self._next_generated_user_id = None self.macaroon_gen = hs.get_macaroon_generator() + self._generate_user_id_linearizer = Linearizer( + name="_generate_user_id_linearizer", + ) + self._server_notices_mxid = hs.config.server_notices_mxid + @defer.inlineCallbacks def check_username(self, localpart, guest_access_token=None, assigned_user_id=None): - yield run_on_reactor() - - if urllib.quote(localpart.encode('utf-8')) != localpart: + if types.contains_invalid_mxid_characters(localpart): raise SynapseError( 400, - "User ID can only contain characters a-z, 0-9, or '_-./'", + "User ID can only contain characters a-z, 0-9, or '=_-./'", Codes.INVALID_USERNAME ) @@ -80,7 +97,7 @@ class RegistrationHandler(BaseHandler): "A different user ID has already been registered for this session", ) - yield self.check_user_id_not_appservice_exclusive(user_id) + self.check_user_id_not_appservice_exclusive(user_id) users = yield self.store.get_users_by_id_case_insensitive(user_id) if users: @@ -127,10 +144,9 @@ class RegistrationHandler(BaseHandler): Raises: RegistrationError if there was a problem registering. """ - yield run_on_reactor() password_hash = None if password: - password_hash = self.auth_handler().hash(password) + password_hash = yield self.auth_handler().hash(password) if localpart: yield self.check_username(localpart, guest_access_token=guest_access_token) @@ -165,6 +181,13 @@ class RegistrationHandler(BaseHandler): ), admin=admin, ) + + if self.hs.config.user_directory_search_all_users: + profile = yield self.store.get_profileinfo(localpart) + yield self.user_directory_handler.handle_local_profile_change( + user_id, profile + ) + else: # autogen a sequential user ID attempts = 0 @@ -192,10 +215,17 @@ class RegistrationHandler(BaseHandler): token = None attempts += 1 + # auto-join the user to any rooms we're supposed to dump them into + fake_requester = create_requester(user_id) + for r in self.hs.config.auto_join_rooms: + try: + yield self._join_user_to_room(fake_requester, r) + except Exception as e: + logger.error("Failed to join new user to %r: %r", r, e) + # We used to generate default identicons here, but nowadays # we want clients to generate their own as part of their branding # rather than there being consistent matrix-wide ones, so we don't. - defer.returnValue((user_id, token)) @defer.inlineCallbacks @@ -253,11 +283,10 @@ class RegistrationHandler(BaseHandler): """ Registers email_id as SAML2 Based Auth. """ - if urllib.quote(localpart) != localpart: + if types.contains_invalid_mxid_characters(localpart): raise SynapseError( 400, - "User ID must only contain characters which do not" - " require URL encoding." + "User ID can only contain characters a-z, 0-9, or '=_-./'", ) user = UserID(localpart, self.hs.hostname) user_id = user.to_string() @@ -286,12 +315,12 @@ class RegistrationHandler(BaseHandler): """ for c in threepidCreds: - logger.info("validating theeepidcred sid %s on id server %s", + logger.info("validating threepidcred sid %s on id server %s", c['sid'], c['idServer']) try: identity_handler = self.hs.get_handlers().identity_handler threepid = yield identity_handler.threepid_from_creds(c) - except: + except Exception: logger.exception("Couldn't validate 3pid") raise RegistrationError(400, "Couldn't validate 3pid") @@ -300,6 +329,11 @@ class RegistrationHandler(BaseHandler): logger.info("got threepid with medium '%s' and address '%s'", threepid['medium'], threepid['address']) + if not check_3pid_allowed(self.hs, threepid['medium'], threepid['address']): + raise RegistrationError( + 403, "Third party identifier is not allowed" + ) + @defer.inlineCallbacks def bind_emails(self, user_id, threepidCreds): """Links emails with a user ID and informs an identity server. @@ -314,6 +348,14 @@ class RegistrationHandler(BaseHandler): yield identity_handler.bind_threepid(c, user_id) def check_user_id_not_appservice_exclusive(self, user_id, allowed_appservice=None): + # don't allow people to register the server notices mxid + if self._server_notices_mxid is not None: + if user_id == self._server_notices_mxid: + raise SynapseError( + 400, "This user ID is reserved.", + errcode=Codes.EXCLUSIVE + ) + # valid user IDs must not clash with any user ID namespaces claimed by # application services. services = self.store.get_app_services() @@ -332,9 +374,11 @@ class RegistrationHandler(BaseHandler): @defer.inlineCallbacks def _generate_user_id(self, reseed=False): if reseed or self._next_generated_user_id is None: - self._next_generated_user_id = ( - yield self.store.find_next_generated_user_id_localpart() - ) + with (yield self._generate_user_id_linearizer.queue(())): + if reseed or self._next_generated_user_id is None: + self._next_generated_user_id = ( + yield self.store.find_next_generated_user_id_localpart() + ) id = self._next_generated_user_id self._next_generated_user_id += 1 @@ -391,8 +435,6 @@ class RegistrationHandler(BaseHandler): Raises: RegistrationError if there was a problem registering. """ - yield run_on_reactor() - if localpart is None: raise SynapseError(400, "Request must include user id") @@ -418,13 +460,12 @@ class RegistrationHandler(BaseHandler): create_profile_with_localpart=user.localpart, ) else: - yield self.store.user_delete_access_tokens(user_id=user_id) + yield self._auth_handler.delete_access_tokens_for_user(user_id) yield self.store.add_access_token_to_user(user_id=user_id, token=token) if displayname is not None: logger.info("setting user display name: %s -> %s", user_id, displayname) - profile_handler = self.hs.get_handlers().profile_handler - yield profile_handler.set_displayname( + yield self.profile_handler.set_displayname( user, requester, displayname, by_admin=True, ) @@ -434,16 +475,59 @@ class RegistrationHandler(BaseHandler): return self.hs.get_auth_handler() @defer.inlineCallbacks - def guest_access_token_for(self, medium, address, inviter_user_id): + def get_or_register_3pid_guest(self, medium, address, inviter_user_id): + """Get a guest access token for a 3PID, creating a guest account if + one doesn't already exist. + + Args: + medium (str) + address (str) + inviter_user_id (str): The user ID who is trying to invite the + 3PID + + Returns: + Deferred[(str, str)]: A 2-tuple of `(user_id, access_token)` of the + 3PID guest account. + """ access_token = yield self.store.get_3pid_guest_access_token(medium, address) if access_token: - defer.returnValue(access_token) + user_info = yield self.auth.get_user_by_access_token( + access_token + ) - _, access_token = yield self.register( + defer.returnValue((user_info["user"].to_string(), access_token)) + + user_id, access_token = yield self.register( generate_token=True, make_guest=True ) access_token = yield self.store.save_or_get_3pid_guest_access_token( medium, address, access_token, inviter_user_id ) - defer.returnValue(access_token) + + defer.returnValue((user_id, access_token)) + + @defer.inlineCallbacks + def _join_user_to_room(self, requester, room_identifier): + room_id = None + room_member_handler = self.hs.get_room_member_handler() + if RoomID.is_valid(room_identifier): + room_id = room_identifier + elif RoomAlias.is_valid(room_identifier): + room_alias = RoomAlias.from_string(room_identifier) + room_id, remote_room_hosts = ( + yield room_member_handler.lookup_room_alias(room_alias) + ) + room_id = room_id.to_string() + else: + raise SynapseError(400, "%s was not legal room ID or room alias" % ( + room_identifier, + )) + + yield room_member_handler.update_membership( + requester=requester, + target=requester.user, + room_id=room_id, + remote_room_hosts=remote_room_hosts, + action="join", + ) diff --git a/synapse/handlers/room.py b/synapse/handlers/room.py index 5698d2808..6150b7e22 100644 --- a/synapse/handlers/room.py +++ b/synapse/handlers/room.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2014 - 2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -14,23 +15,20 @@ # limitations under the License. """Contains functions for performing events on rooms.""" -from twisted.internet import defer - -from ._base import BaseHandler - -from synapse.types import UserID, RoomAlias, RoomID, RoomStreamToken -from synapse.api.constants import ( - EventTypes, JoinRules, RoomCreationPreset -) -from synapse.api.errors import AuthError, StoreError, SynapseError -from synapse.util import stringutils -from synapse.visibility import filter_events_for_client - -from collections import OrderedDict - import logging import math import string +from collections import OrderedDict + +from twisted.internet import defer + +from synapse.api.constants import EventTypes, JoinRules, RoomCreationPreset +from synapse.api.errors import AuthError, Codes, StoreError, SynapseError +from synapse.types import RoomAlias, RoomID, RoomStreamToken, UserID +from synapse.util import stringutils +from synapse.visibility import filter_events_for_client + +from ._base import BaseHandler logger = logging.getLogger(__name__) @@ -60,21 +58,43 @@ class RoomCreationHandler(BaseHandler): }, } + def __init__(self, hs): + super(RoomCreationHandler, self).__init__(hs) + + self.spam_checker = hs.get_spam_checker() + self.event_creation_handler = hs.get_event_creation_handler() + @defer.inlineCallbacks - def create_room(self, requester, config, ratelimit=True): + def create_room(self, requester, config, ratelimit=True, + creator_join_profile=None): """ Creates a new room. Args: - requester (Requester): The user who requested the room creation. + requester (synapse.types.Requester): + The user who requested the room creation. config (dict) : A dict of configuration options. + ratelimit (bool): set to False to disable the rate limiter + + creator_join_profile (dict|None): + Set to override the displayname and avatar for the creating + user in this room. If unset, displayname and avatar will be + derived from the user's profile. If set, should contain the + values to go in the body of the 'join' event (typically + `avatar_url` and/or `displayname`. + Returns: - The new room ID. + Deferred[dict]: + a dict containing the keys `room_id` and, if an alias was + requested, `room_alias`. Raises: SynapseError if the room ID couldn't be stored, or something went horribly wrong. """ user_id = requester.user.to_string() + if not self.spam_checker.user_may_create_room(user_id): + raise SynapseError(403, "You are not permitted to create rooms") + if ratelimit: yield self.ratelimit(requester) @@ -83,7 +103,7 @@ class RoomCreationHandler(BaseHandler): if wchar in config["room_alias_name"]: raise SynapseError(400, "Invalid characters in room alias") - room_alias = RoomAlias.create( + room_alias = RoomAlias( config["room_alias_name"], self.hs.hostname, ) @@ -92,7 +112,11 @@ class RoomCreationHandler(BaseHandler): ) if mapping: - raise SynapseError(400, "Room alias already taken") + raise SynapseError( + 400, + "Room alias already taken", + Codes.ROOM_IN_USE + ) else: room_alias = None @@ -100,9 +124,13 @@ class RoomCreationHandler(BaseHandler): for i in invite_list: try: UserID.from_string(i) - except: + except Exception: raise SynapseError(400, "Invalid user_id: %s" % (i,)) + yield self.event_creation_handler.assert_accepted_privacy_policy( + requester, + ) + invite_3pid_list = config.get("invite_3pid", []) visibility = config.get("visibility", None) @@ -115,7 +143,7 @@ class RoomCreationHandler(BaseHandler): while attempts < 5: try: random_string = stringutils.random_string(18) - gen_room_id = RoomID.create( + gen_room_id = RoomID( random_string, self.hs.hostname, ) @@ -155,25 +183,24 @@ class RoomCreationHandler(BaseHandler): creation_content = config.get("creation_content", {}) - msg_handler = self.hs.get_handlers().message_handler - room_member_handler = self.hs.get_handlers().room_member_handler + room_member_handler = self.hs.get_room_member_handler() yield self._send_events_for_new_room( requester, room_id, - msg_handler, room_member_handler, preset_config=preset_config, invite_list=invite_list, initial_state=initial_state, creation_content=creation_content, room_alias=room_alias, - power_level_content_override=config.get("power_level_content_override", {}) + power_level_content_override=config.get("power_level_content_override", {}), + creator_join_profile=creator_join_profile, ) if "name" in config: name = config["name"] - yield msg_handler.create_and_send_nonmember_event( + yield self.event_creation_handler.create_and_send_nonmember_event( requester, { "type": EventTypes.Name, @@ -186,7 +213,7 @@ class RoomCreationHandler(BaseHandler): if "topic" in config: topic = config["topic"] - yield msg_handler.create_and_send_nonmember_event( + yield self.event_creation_handler.create_and_send_nonmember_event( requester, { "type": EventTypes.Topic, @@ -197,12 +224,12 @@ class RoomCreationHandler(BaseHandler): }, ratelimit=False) - content = {} - is_direct = config.get("is_direct", None) - if is_direct: - content["is_direct"] = is_direct - for invitee in invite_list: + content = {} + is_direct = config.get("is_direct", None) + if is_direct: + content["is_direct"] = is_direct + yield room_member_handler.update_membership( requester, UserID.from_string(invitee), @@ -216,7 +243,7 @@ class RoomCreationHandler(BaseHandler): id_server = invite_3pid["id_server"] address = invite_3pid["address"] medium = invite_3pid["medium"] - yield self.hs.get_handlers().room_member_handler.do_3pid_invite( + yield self.hs.get_room_member_handler().do_3pid_invite( room_id, requester.user, medium, @@ -241,7 +268,6 @@ class RoomCreationHandler(BaseHandler): self, creator, # A Requester object. room_id, - msg_handler, room_member_handler, preset_config, invite_list, @@ -249,6 +275,7 @@ class RoomCreationHandler(BaseHandler): creation_content, room_alias, power_level_content_override, + creator_join_profile, ): def create(etype, content, **kwargs): e = { @@ -264,7 +291,7 @@ class RoomCreationHandler(BaseHandler): @defer.inlineCallbacks def send(etype, content, **kwargs): event = create(etype, content, **kwargs) - yield msg_handler.create_and_send_nonmember_event( + yield self.event_creation_handler.create_and_send_nonmember_event( creator, event, ratelimit=False @@ -292,6 +319,7 @@ class RoomCreationHandler(BaseHandler): room_id, "join", ratelimit=False, + content=creator_join_profile, ) # We treat the power levels override specially as this needs to be one @@ -367,7 +395,11 @@ class RoomCreationHandler(BaseHandler): ) -class RoomContextHandler(BaseHandler): +class RoomContextHandler(object): + def __init__(self, hs): + self.hs = hs + self.store = hs.get_datastore() + @defer.inlineCallbacks def get_event_context(self, user, room_id, event_id, limit): """Retrieves events, pagination tokens and state around a given event @@ -428,7 +460,7 @@ class RoomContextHandler(BaseHandler): state = yield self.store.get_state_for_events( [last_event_id], None ) - results["state"] = state[last_event_id].values() + results["state"] = list(state[last_event_id].values()) results["start"] = now_token.copy_and_replace( "room_key", results["start"] @@ -468,12 +500,9 @@ class RoomEventSource(object): user.to_string() ) if app_service: - events, end_key = yield self.store.get_appservice_room_stream( - service=app_service, - from_key=from_key, - to_key=to_key, - limit=limit, - ) + # We no longer support AS users using /sync directly. + # See https://github.com/matrix-org/matrix-doc/issues/1144 + raise NotImplementedError() else: room_events = yield self.store.get_membership_changes_for_user( user.to_string(), from_key, to_key diff --git a/synapse/handlers/room_list.py b/synapse/handlers/room_list.py index 516cd9a6a..828229f5c 100644 --- a/synapse/handlers/room_list.py +++ b/synapse/handlers/room_list.py @@ -13,23 +13,24 @@ # See the License for the specific language governing permissions and # limitations under the License. +import logging +from collections import namedtuple + +from six import iteritems +from six.moves import range + +import msgpack +from unpaddedbase64 import decode_base64, encode_base64 + from twisted.internet import defer -from ._base import BaseHandler - -from synapse.api.constants import ( - EventTypes, JoinRules, -) +from synapse.api.constants import EventTypes, JoinRules +from synapse.types import ThirdPartyInstanceID from synapse.util.async import concurrently_execute from synapse.util.caches.descriptors import cachedInlineCallbacks from synapse.util.caches.response_cache import ResponseCache -from synapse.types import ThirdPartyInstanceID -from collections import namedtuple -from unpaddedbase64 import encode_base64, decode_base64 - -import logging -import msgpack +from ._base import BaseHandler logger = logging.getLogger(__name__) @@ -37,18 +38,19 @@ REMOTE_ROOM_LIST_POLL_INTERVAL = 60 * 1000 # This is used to indicate we should only return rooms published to the main list. -EMTPY_THIRD_PARTY_ID = ThirdPartyInstanceID(None, None) +EMPTY_THIRD_PARTY_ID = ThirdPartyInstanceID(None, None) class RoomListHandler(BaseHandler): def __init__(self, hs): super(RoomListHandler, self).__init__(hs) - self.response_cache = ResponseCache(hs) - self.remote_response_cache = ResponseCache(hs, timeout_ms=30 * 1000) + self.response_cache = ResponseCache(hs, "room_list") + self.remote_response_cache = ResponseCache(hs, "remote_room_list", + timeout_ms=30 * 1000) def get_local_public_room_list(self, limit=None, since_token=None, search_filter=None, - network_tuple=EMTPY_THIRD_PARTY_ID,): + network_tuple=EMPTY_THIRD_PARTY_ID,): """Generate a local public room list. There are multiple different lists: the main one plus one per third @@ -70,25 +72,22 @@ class RoomListHandler(BaseHandler): if search_filter: # We explicitly don't bother caching searches or requests for # appservice specific lists. + logger.info("Bypassing cache as search request.") return self._get_public_room_list( limit, since_token, search_filter, network_tuple=network_tuple, ) key = (limit, since_token, network_tuple) - result = self.response_cache.get(key) - if not result: - result = self.response_cache.set( - key, - self._get_public_room_list( - limit, since_token, network_tuple=network_tuple - ) - ) - return result + return self.response_cache.wrap( + key, + self._get_public_room_list, + limit, since_token, network_tuple=network_tuple, + ) @defer.inlineCallbacks def _get_public_room_list(self, limit=None, since_token=None, search_filter=None, - network_tuple=EMTPY_THIRD_PARTY_ID,): + network_tuple=EMPTY_THIRD_PARTY_ID,): if since_token and since_token != "END": since_token = RoomListNextBatch.from_token(since_token) else: @@ -149,6 +148,8 @@ class RoomListHandler(BaseHandler): # We want larger rooms to be first, hence negating num_joined_users rooms_to_order_value[room_id] = (-num_joined_users, room_id) + logger.info("Getting ordering for %i rooms since %s", + len(room_ids), stream_token) yield concurrently_execute(get_order_for_room, room_ids, 10) sorted_entries = sorted(rooms_to_order_value.items(), key=lambda e: e[1]) @@ -176,34 +177,43 @@ class RoomListHandler(BaseHandler): rooms_to_scan = rooms_to_scan[:since_token.current_limit] rooms_to_scan.reverse() - # Actually generate the entries. _append_room_entry_to_chunk will append to - # chunk but will stop if len(chunk) > limit - chunk = [] - if limit and not search_filter: + logger.info("After sorting and filtering, %i rooms remain", + len(rooms_to_scan)) + + # _append_room_entry_to_chunk will append to chunk but will stop if + # len(chunk) > limit + # + # Normally we will generate enough results on the first iteration here, + # but if there is a search filter, _append_room_entry_to_chunk may + # filter some results out, in which case we loop again. + # + # We don't want to scan over the entire range either as that + # would potentially waste a lot of work. + # + # XXX if there is no limit, we may end up DoSing the server with + # calls to get_current_state_ids for every single room on the + # server. Surely we should cap this somehow? + # + if limit: step = limit + 1 - for i in xrange(0, len(rooms_to_scan), step): - # We iterate here because the vast majority of cases we'll stop - # at first iteration, but occaisonally _append_room_entry_to_chunk - # won't append to the chunk and so we need to loop again. - # We don't want to scan over the entire range either as that - # would potentially waste a lot of work. - yield concurrently_execute( - lambda r: self._append_room_entry_to_chunk( - r, rooms_to_num_joined[r], - chunk, limit, search_filter - ), - rooms_to_scan[i:i + step], 10 - ) - if len(chunk) >= limit + 1: - break else: + # step cannot be zero + step = len(rooms_to_scan) if len(rooms_to_scan) != 0 else 1 + + chunk = [] + for i in range(0, len(rooms_to_scan), step): + batch = rooms_to_scan[i:i + step] + logger.info("Processing %i rooms for result", len(batch)) yield concurrently_execute( lambda r: self._append_room_entry_to_chunk( r, rooms_to_num_joined[r], chunk, limit, search_filter ), - rooms_to_scan, 5 + batch, 5, ) + logger.info("Now %i rooms in result", len(chunk)) + if len(chunk) >= limit + 1: + break chunk.sort(key=lambda e: (-e["num_joined_members"], e["room_id"])) @@ -276,13 +286,14 @@ class RoomListHandler(BaseHandler): # We've already got enough, so lets just drop it. return - result = yield self._generate_room_entry(room_id, num_joined_users) + result = yield self.generate_room_entry(room_id, num_joined_users) if result and _matches_room_entry(result, search_filter): chunk.append(result) @cachedInlineCallbacks(num_args=1, cache_context=True) - def _generate_room_entry(self, room_id, num_joined_users, cache_context): + def generate_room_entry(self, room_id, num_joined_users, cache_context, + with_alias=True, allow_private=False): """Returns the entry for a room """ result = { @@ -295,7 +306,7 @@ class RoomListHandler(BaseHandler): ) event_map = yield self.store.get_events([ - event_id for key, event_id in current_state_ids.iteritems() + event_id for key, event_id in iteritems(current_state_ids) if key[0] in ( EventTypes.JoinRules, EventTypes.Name, @@ -316,14 +327,15 @@ class RoomListHandler(BaseHandler): join_rules_event = current_state.get((EventTypes.JoinRules, "")) if join_rules_event: join_rule = join_rules_event.content.get("join_rule", None) - if join_rule and join_rule != JoinRules.PUBLIC: + if not allow_private and join_rule and join_rule != JoinRules.PUBLIC: defer.returnValue(None) - aliases = yield self.store.get_aliases_for_room( - room_id, on_invalidate=cache_context.invalidate - ) - if aliases: - result["aliases"] = aliases + if with_alias: + aliases = yield self.store.get_aliases_for_room( + room_id, on_invalidate=cache_context.invalidate + ) + if aliases: + result["aliases"] = aliases name_event = yield current_state.get((EventTypes.Name, "")) if name_event: @@ -391,7 +403,7 @@ class RoomListHandler(BaseHandler): def _get_remote_list_cached(self, server_name, limit=None, since_token=None, search_filter=None, include_all_networks=False, third_party_instance_id=None,): - repl_layer = self.hs.get_replication_layer() + repl_layer = self.hs.get_federation_client() if search_filter: # We can't cache when asking for search return repl_layer.get_public_rooms( @@ -404,18 +416,14 @@ class RoomListHandler(BaseHandler): server_name, limit, since_token, include_all_networks, third_party_instance_id, ) - result = self.remote_response_cache.get(key) - if not result: - result = self.remote_response_cache.set( - key, - repl_layer.get_public_rooms( - server_name, limit=limit, since_token=since_token, - search_filter=search_filter, - include_all_networks=include_all_networks, - third_party_instance_id=third_party_instance_id, - ) - ) - return result + return self.remote_response_cache.wrap( + key, + repl_layer.get_public_rooms, + server_name, limit=limit, since_token=since_token, + search_filter=search_filter, + include_all_networks=include_all_networks, + third_party_instance_id=third_party_instance_id, + ) class RoomListNextBatch(namedtuple("RoomListNextBatch", ( diff --git a/synapse/handlers/room_member.py b/synapse/handlers/room_member.py index 1b8dfa825..0d4a3f467 100644 --- a/synapse/handlers/room_member.py +++ b/synapse/handlers/room_member.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -13,63 +14,161 @@ # See the License for the specific language governing permissions and # limitations under the License. - +import abc import logging +from six.moves import http_client + from signedjson.key import decode_verify_key_bytes from signedjson.sign import verify_signed_json -from twisted.internet import defer from unpaddedbase64 import decode_base64 +from twisted.internet import defer + +import synapse.server import synapse.types -from synapse.api.constants import ( - EventTypes, Membership, -) -from synapse.api.errors import AuthError, SynapseError, Codes -from synapse.types import UserID, RoomID +from synapse.api.constants import EventTypes, Membership +from synapse.api.errors import AuthError, Codes, SynapseError +from synapse.types import RoomID, UserID from synapse.util.async import Linearizer -from synapse.util.distributor import user_left_room, user_joined_room -from ._base import BaseHandler +from synapse.util.distributor import user_joined_room, user_left_room logger = logging.getLogger(__name__) id_server_scheme = "https://" -class RoomMemberHandler(BaseHandler): +class RoomMemberHandler(object): # TODO(paul): This handler currently contains a messy conflation of # low-level API that works on UserID objects and so on, and REST-level # API that takes ID strings and returns pagination chunks. These concerns # ought to be separated out a lot better. + __metaclass__ = abc.ABCMeta + def __init__(self, hs): - super(RoomMemberHandler, self).__init__(hs) + """ + + Args: + hs (synapse.server.HomeServer): + """ + self.hs = hs + self.store = hs.get_datastore() + self.auth = hs.get_auth() + self.state_handler = hs.get_state_handler() + self.config = hs.config + self.simple_http_client = hs.get_simple_http_client() + + self.federation_handler = hs.get_handlers().federation_handler + self.directory_handler = hs.get_handlers().directory_handler + self.registration_handler = hs.get_handlers().registration_handler + self.profile_handler = hs.get_profile_handler() + self.event_creation_hander = hs.get_event_creation_handler() self.member_linearizer = Linearizer(name="member") self.clock = hs.get_clock() + self.spam_checker = hs.get_spam_checker() + self._server_notices_mxid = self.config.server_notices_mxid - self.distributor = hs.get_distributor() - self.distributor.declare("user_joined_room") - self.distributor.declare("user_left_room") + @abc.abstractmethod + def _remote_join(self, requester, remote_room_hosts, room_id, user, content): + """Try and join a room that this server is not in + + Args: + requester (Requester) + remote_room_hosts (list[str]): List of servers that can be used + to join via. + room_id (str): Room that we are trying to join + user (UserID): User who is trying to join + content (dict): A dict that should be used as the content of the + join event. + + Returns: + Deferred + """ + raise NotImplementedError() + + @abc.abstractmethod + def _remote_reject_invite(self, remote_room_hosts, room_id, target): + """Attempt to reject an invite for a room this server is not in. If we + fail to do so we locally mark the invite as rejected. + + Args: + requester (Requester) + remote_room_hosts (list[str]): List of servers to use to try and + reject invite + room_id (str) + target (UserID): The user rejecting the invite + + Returns: + Deferred[dict]: A dictionary to be returned to the client, may + include event_id etc, or nothing if we locally rejected + """ + raise NotImplementedError() + + @abc.abstractmethod + def get_or_register_3pid_guest(self, requester, medium, address, inviter_user_id): + """Get a guest access token for a 3PID, creating a guest account if + one doesn't already exist. + + Args: + requester (Requester) + medium (str) + address (str) + inviter_user_id (str): The user ID who is trying to invite the + 3PID + + Returns: + Deferred[(str, str)]: A 2-tuple of `(user_id, access_token)` of the + 3PID guest account. + """ + raise NotImplementedError() + + @abc.abstractmethod + def _user_joined_room(self, target, room_id): + """Notifies distributor on master process that the user has joined the + room. + + Args: + target (UserID) + room_id (str) + + Returns: + Deferred|None + """ + raise NotImplementedError() + + @abc.abstractmethod + def _user_left_room(self, target, room_id): + """Notifies distributor on master process that the user has left the + room. + + Args: + target (UserID) + room_id (str) + + Returns: + Deferred|None + """ + raise NotImplementedError() @defer.inlineCallbacks def _local_membership_update( self, requester, target, room_id, membership, - prev_event_ids, + prev_events_and_hashes, txn_id=None, ratelimit=True, content=None, ): if content is None: content = {} - msg_handler = self.hs.get_handlers().message_handler content["membership"] = membership if requester.is_guest: content["kind"] = "guest" - event, context = yield msg_handler.create_event( + event, context = yield self.event_creation_hander.create_event( requester, { "type": EventTypes.Member, @@ -83,16 +182,18 @@ class RoomMemberHandler(BaseHandler): }, token_id=requester.access_token_id, txn_id=txn_id, - prev_event_ids=prev_event_ids, + prev_events_and_hashes=prev_events_and_hashes, ) # Check if this event matches the previous membership event for the user. - duplicate = yield msg_handler.deduplicate_state_event(event, context) + duplicate = yield self.event_creation_hander.deduplicate_state_event( + event, context, + ) if duplicate is not None: # Discard the new event since this membership change is a no-op. defer.returnValue(duplicate) - yield msg_handler.handle_new_client_event( + yield self.event_creation_hander.handle_new_client_event( requester, event, context, @@ -100,7 +201,9 @@ class RoomMemberHandler(BaseHandler): ratelimit=ratelimit, ) - prev_member_event_id = context.prev_state_ids.get( + prev_state_ids = yield context.get_prev_state_ids(self.store) + + prev_member_event_id = prev_state_ids.get( (EventTypes.Member, target.to_string()), None ) @@ -114,32 +217,15 @@ class RoomMemberHandler(BaseHandler): prev_member_event = yield self.store.get_event(prev_member_event_id) newly_joined = prev_member_event.membership != Membership.JOIN if newly_joined: - yield user_joined_room(self.distributor, target, room_id) + yield self._user_joined_room(target, room_id) elif event.membership == Membership.LEAVE: if prev_member_event_id: prev_member_event = yield self.store.get_event(prev_member_event_id) if prev_member_event.membership == Membership.JOIN: - user_left_room(self.distributor, target, room_id) + yield self._user_left_room(target, room_id) defer.returnValue(event) - @defer.inlineCallbacks - def remote_join(self, remote_room_hosts, room_id, user, content): - if len(remote_room_hosts) == 0: - raise SynapseError(404, "No known servers") - - # We don't do an auth check if we are doing an invite - # join dance for now, since we're kinda implicitly checking - # that we are allowed to join when we decide whether or not we - # need to do the invite/join dance. - yield self.hs.get_handlers().federation_handler.do_invite_join( - remote_room_hosts, - room_id, - user.to_string(), - content, - ) - yield user_joined_room(self.distributor, user, room_id) - @defer.inlineCallbacks def update_membership( self, @@ -186,14 +272,19 @@ class RoomMemberHandler(BaseHandler): content_specified = bool(content) if content is None: content = {} + else: + # We do a copy here as we potentially change some keys + # later on. + content = dict(content) effective_membership_state = action if action in ["kick", "unban"]: effective_membership_state = "leave" + # if this is a join with a 3pid signature, we may need to turn a 3pid + # invite into a normal invite before we can handle the join. if third_party_signed is not None: - replication = self.hs.get_replication_layer() - yield replication.exchange_third_party_invite( + yield self.federation_handler.exchange_third_party_invite( third_party_signed["sender"], target.to_string(), room_id, @@ -208,7 +299,51 @@ class RoomMemberHandler(BaseHandler): if is_blocked: raise SynapseError(403, "This room has been blocked on this server") - latest_event_ids = yield self.store.get_latest_event_ids_in_room(room_id) + if effective_membership_state == Membership.INVITE: + # block any attempts to invite the server notices mxid + if target.to_string() == self._server_notices_mxid: + raise SynapseError( + http_client.FORBIDDEN, + "Cannot invite this user", + ) + + block_invite = False + + if (self._server_notices_mxid is not None and + requester.user.to_string() == self._server_notices_mxid): + # allow the server notices mxid to send invites + is_requester_admin = True + + else: + is_requester_admin = yield self.auth.is_server_admin( + requester.user, + ) + + if not is_requester_admin: + if self.config.block_non_admin_invites: + logger.info( + "Blocking invite: user is not admin and non-admin " + "invites disabled" + ) + block_invite = True + + if not self.spam_checker.user_may_invite( + requester.user.to_string(), target.to_string(), room_id, + ): + logger.info("Blocking invite due to spam checker") + block_invite = True + + if block_invite: + raise SynapseError( + 403, "Invites have been disabled on this server", + ) + + prev_events_and_hashes = yield self.store.get_prev_events_for_room( + room_id, + ) + latest_event_ids = ( + event_id for (event_id, _, _) in prev_events_and_hashes + ) current_state_ids = yield self.state_handler.get_current_state_ids( room_id, latest_event_ids=latest_event_ids, ) @@ -238,6 +373,20 @@ class RoomMemberHandler(BaseHandler): if same_sender and same_membership and same_content: defer.returnValue(old_state) + # we don't allow people to reject invites to the server notice + # room, but they can leave it once they are joined. + if ( + old_membership == Membership.INVITE and + effective_membership_state == Membership.LEAVE + ): + is_blocked = yield self._is_server_notice_room(room_id) + if is_blocked: + raise SynapseError( + http_client.FORBIDDEN, + "You cannot reject this invite", + errcode=Codes.CANNOT_LEAVE_SERVER_NOTICE_ROOM, + ) + is_host_in_room = yield self._is_host_in_room(current_state_ids) if effective_membership_state == Membership.JOIN: @@ -249,13 +398,13 @@ class RoomMemberHandler(BaseHandler): raise AuthError(403, "Guest access not allowed") if not is_host_in_room: - inviter = yield self.get_inviter(target.to_string(), room_id) + inviter = yield self._get_inviter(target.to_string(), room_id) if inviter and not self.hs.is_mine(inviter): remote_room_hosts.append(inviter.domain) content["membership"] = Membership.JOIN - profile = self.hs.get_handlers().profile_handler + profile = self.profile_handler if not content_specified: content["displayname"] = yield profile.get_displayname(target) content["avatar_url"] = yield profile.get_avatar_url(target) @@ -263,15 +412,15 @@ class RoomMemberHandler(BaseHandler): if requester.is_guest: content["kind"] = "guest" - ret = yield self.remote_join( - remote_room_hosts, room_id, target, content + ret = yield self._remote_join( + requester, remote_room_hosts, room_id, target, content ) defer.returnValue(ret) elif effective_membership_state == Membership.LEAVE: if not is_host_in_room: # perhaps we've been invited - inviter = yield self.get_inviter(target.to_string(), room_id) + inviter = yield self._get_inviter(target.to_string(), room_id) if not inviter: raise SynapseError(404, "Not a known room") @@ -285,28 +434,10 @@ class RoomMemberHandler(BaseHandler): else: # send the rejection to the inviter's HS. remote_room_hosts = remote_room_hosts + [inviter.domain] - fed_handler = self.hs.get_handlers().federation_handler - try: - ret = yield fed_handler.do_remotely_reject_invite( - remote_room_hosts, - room_id, - target.to_string(), - ) - defer.returnValue(ret) - except Exception as e: - # if we were unable to reject the exception, just mark - # it as rejected on our end and plough ahead. - # - # The 'except' clause is very broad, but we need to - # capture everything from DNS failures upwards - # - logger.warn("Failed to reject invite: %s", e) - - yield self.store.locally_reject_invite( - target.to_string(), room_id - ) - - defer.returnValue({}) + res = yield self._remote_reject_invite( + requester, remote_room_hosts, room_id, target, + ) + defer.returnValue(res) res = yield self._local_membership_update( requester=requester, @@ -315,7 +446,7 @@ class RoomMemberHandler(BaseHandler): membership=effective_membership_state, txn_id=txn_id, ratelimit=ratelimit, - prev_event_ids=latest_event_ids, + prev_events_and_hashes=prev_events_and_hashes, content=content, ) defer.returnValue(res) @@ -361,14 +492,16 @@ class RoomMemberHandler(BaseHandler): else: requester = synapse.types.create_requester(target_user) - message_handler = self.hs.get_handlers().message_handler - prev_event = yield message_handler.deduplicate_state_event(event, context) + prev_event = yield self.event_creation_hander.deduplicate_state_event( + event, context, + ) if prev_event is not None: return + prev_state_ids = yield context.get_prev_state_ids(self.store) if event.membership == Membership.JOIN: if requester.is_guest: - guest_can_join = yield self._can_guest_join(context.prev_state_ids) + guest_can_join = yield self._can_guest_join(prev_state_ids) if not guest_can_join: # This should be an auth check, but guests are a local concept, # so don't really fit into the general auth process. @@ -379,7 +512,7 @@ class RoomMemberHandler(BaseHandler): if is_blocked: raise SynapseError(403, "This room has been blocked on this server") - yield message_handler.handle_new_client_event( + yield self.event_creation_hander.handle_new_client_event( requester, event, context, @@ -387,7 +520,7 @@ class RoomMemberHandler(BaseHandler): ratelimit=ratelimit, ) - prev_member_event_id = context.prev_state_ids.get( + prev_member_event_id = prev_state_ids.get( (EventTypes.Member, event.state_key), None ) @@ -401,12 +534,12 @@ class RoomMemberHandler(BaseHandler): prev_member_event = yield self.store.get_event(prev_member_event_id) newly_joined = prev_member_event.membership != Membership.JOIN if newly_joined: - yield user_joined_room(self.distributor, target_user, room_id) + yield self._user_joined_room(target_user, room_id) elif event.membership == Membership.LEAVE: if prev_member_event_id: prev_member_event = yield self.store.get_event(prev_member_event_id) if prev_member_event.membership == Membership.JOIN: - user_left_room(self.distributor, target_user, room_id) + yield self._user_left_room(target_user, room_id) @defer.inlineCallbacks def _can_guest_join(self, current_state_ids): @@ -440,7 +573,7 @@ class RoomMemberHandler(BaseHandler): Raises: SynapseError if room alias could not be found. """ - directory_handler = self.hs.get_handlers().directory_handler + directory_handler = self.directory_handler mapping = yield directory_handler.get_association(room_alias) if not mapping: @@ -452,7 +585,7 @@ class RoomMemberHandler(BaseHandler): defer.returnValue((RoomID.from_string(room_id), servers)) @defer.inlineCallbacks - def get_inviter(self, user_id, room_id): + def _get_inviter(self, user_id, room_id): invite = yield self.store.get_invite_for_user_in_room( user_id=user_id, room_id=room_id, @@ -471,6 +604,16 @@ class RoomMemberHandler(BaseHandler): requester, txn_id ): + if self.config.block_non_admin_invites: + is_requester_admin = yield self.auth.is_server_admin( + requester.user, + ) + if not is_requester_admin: + raise SynapseError( + 403, "Invites have been disabled on this server", + Codes.FORBIDDEN, + ) + invitee = yield self._lookup_3pid( id_server, medium, address ) @@ -508,7 +651,7 @@ class RoomMemberHandler(BaseHandler): str: the matrix ID of the 3pid, or None if it is not recognized. """ try: - data = yield self.hs.get_simple_http_client().get_json( + data = yield self.simple_http_client.get_json( "%s%s/_matrix/identity/api/v1/lookup" % (id_server_scheme, id_server,), { "medium": medium, @@ -519,7 +662,7 @@ class RoomMemberHandler(BaseHandler): if "mxid" in data: if "signatures" not in data: raise AuthError(401, "No signatures on 3pid binding") - self.verify_any_signature(data, id_server) + yield self._verify_any_signature(data, id_server) defer.returnValue(data["mxid"]) except IOError as e: @@ -527,11 +670,11 @@ class RoomMemberHandler(BaseHandler): defer.returnValue(None) @defer.inlineCallbacks - def verify_any_signature(self, data, server_hostname): + def _verify_any_signature(self, data, server_hostname): if server_hostname not in data["signatures"]: raise AuthError(401, "No signature from server %s" % (server_hostname,)) for key_name, signature in data["signatures"][server_hostname].items(): - key_data = yield self.hs.get_simple_http_client().get_json( + key_data = yield self.simple_http_client.get_json( "%s%s/_matrix/identity/api/v1/pubkey/%s" % (id_server_scheme, server_hostname, key_name,), ) @@ -556,7 +699,7 @@ class RoomMemberHandler(BaseHandler): user, txn_id ): - room_state = yield self.hs.get_state_handler().get_current_state(room_id) + room_state = yield self.state_handler.get_current_state(room_id) inviter_display_name = "" inviter_avatar_url = "" @@ -591,6 +734,7 @@ class RoomMemberHandler(BaseHandler): token, public_keys, fallback_public_key, display_name = ( yield self._ask_id_server_for_third_party_invite( + requester=requester, id_server=id_server, medium=medium, address=address, @@ -605,8 +749,7 @@ class RoomMemberHandler(BaseHandler): ) ) - msg_handler = self.hs.get_handlers().message_handler - yield msg_handler.create_and_send_nonmember_event( + yield self.event_creation_hander.create_and_send_nonmember_event( requester, { "type": EventTypes.ThirdPartyInvite, @@ -628,6 +771,7 @@ class RoomMemberHandler(BaseHandler): @defer.inlineCallbacks def _ask_id_server_for_third_party_invite( self, + requester, id_server, medium, address, @@ -644,6 +788,7 @@ class RoomMemberHandler(BaseHandler): Asks an identity server for a third party invite. Args: + requester (Requester) id_server (str): hostname + optional port for the identity server. medium (str): The literal string "email". address (str): The third party address being invited. @@ -685,24 +830,20 @@ class RoomMemberHandler(BaseHandler): "sender_avatar_url": inviter_avatar_url, } - if self.hs.config.invite_3pid_guest: - registration_handler = self.hs.get_handlers().registration_handler - guest_access_token = yield registration_handler.guest_access_token_for( + if self.config.invite_3pid_guest: + guest_access_token, guest_user_id = yield self.get_or_register_3pid_guest( + requester=requester, medium=medium, address=address, inviter_user_id=inviter_user_id, ) - guest_user_info = yield self.hs.get_auth().get_user_by_access_token( - guest_access_token - ) - invite_config.update({ "guest_access_token": guest_access_token, - "guest_user_id": guest_user_info["user"].to_string(), + "guest_user_id": guest_user_id, }) - data = yield self.hs.get_simple_http_client().post_urlencoded_get_json( + data = yield self.simple_http_client.post_urlencoded_get_json( is_url, invite_config ) @@ -724,27 +865,6 @@ class RoomMemberHandler(BaseHandler): display_name = data["display_name"] defer.returnValue((token, public_keys, fallback_public_key, display_name)) - @defer.inlineCallbacks - def forget(self, user, room_id): - user_id = user.to_string() - - member = yield self.state_handler.get_current_state( - room_id=room_id, - event_type=EventTypes.Member, - state_key=user_id - ) - membership = member.membership if member else None - - if membership is not None and membership not in [ - Membership.LEAVE, Membership.BAN - ]: - raise SynapseError(400, "User %s in room %s" % ( - user_id, room_id - )) - - if membership: - yield self.store.forget(user_id, room_id) - @defer.inlineCallbacks def _is_host_in_room(self, current_state_ids): # Have we just created the room, and is this about to be the very @@ -766,3 +886,109 @@ class RoomMemberHandler(BaseHandler): defer.returnValue(True) defer.returnValue(False) + + @defer.inlineCallbacks + def _is_server_notice_room(self, room_id): + if self._server_notices_mxid is None: + defer.returnValue(False) + user_ids = yield self.store.get_users_in_room(room_id) + defer.returnValue(self._server_notices_mxid in user_ids) + + +class RoomMemberMasterHandler(RoomMemberHandler): + def __init__(self, hs): + super(RoomMemberMasterHandler, self).__init__(hs) + + self.distributor = hs.get_distributor() + self.distributor.declare("user_joined_room") + self.distributor.declare("user_left_room") + + @defer.inlineCallbacks + def _remote_join(self, requester, remote_room_hosts, room_id, user, content): + """Implements RoomMemberHandler._remote_join + """ + # filter ourselves out of remote_room_hosts: do_invite_join ignores it + # and if it is the only entry we'd like to return a 404 rather than a + # 500. + + remote_room_hosts = [ + host for host in remote_room_hosts if host != self.hs.hostname + ] + + if len(remote_room_hosts) == 0: + raise SynapseError(404, "No known servers") + + # We don't do an auth check if we are doing an invite + # join dance for now, since we're kinda implicitly checking + # that we are allowed to join when we decide whether or not we + # need to do the invite/join dance. + yield self.federation_handler.do_invite_join( + remote_room_hosts, + room_id, + user.to_string(), + content, + ) + yield self._user_joined_room(user, room_id) + + @defer.inlineCallbacks + def _remote_reject_invite(self, requester, remote_room_hosts, room_id, target): + """Implements RoomMemberHandler._remote_reject_invite + """ + fed_handler = self.federation_handler + try: + ret = yield fed_handler.do_remotely_reject_invite( + remote_room_hosts, + room_id, + target.to_string(), + ) + defer.returnValue(ret) + except Exception as e: + # if we were unable to reject the exception, just mark + # it as rejected on our end and plough ahead. + # + # The 'except' clause is very broad, but we need to + # capture everything from DNS failures upwards + # + logger.warn("Failed to reject invite: %s", e) + + yield self.store.locally_reject_invite( + target.to_string(), room_id + ) + defer.returnValue({}) + + def get_or_register_3pid_guest(self, requester, medium, address, inviter_user_id): + """Implements RoomMemberHandler.get_or_register_3pid_guest + """ + rg = self.registration_handler + return rg.get_or_register_3pid_guest(medium, address, inviter_user_id) + + def _user_joined_room(self, target, room_id): + """Implements RoomMemberHandler._user_joined_room + """ + return user_joined_room(self.distributor, target, room_id) + + def _user_left_room(self, target, room_id): + """Implements RoomMemberHandler._user_left_room + """ + return user_left_room(self.distributor, target, room_id) + + @defer.inlineCallbacks + def forget(self, user, room_id): + user_id = user.to_string() + + member = yield self.state_handler.get_current_state( + room_id=room_id, + event_type=EventTypes.Member, + state_key=user_id + ) + membership = member.membership if member else None + + if membership is not None and membership not in [ + Membership.LEAVE, Membership.BAN + ]: + raise SynapseError(400, "User %s in room %s" % ( + user_id, room_id + )) + + if membership: + yield self.store.forget(user_id, room_id) diff --git a/synapse/handlers/room_member_worker.py b/synapse/handlers/room_member_worker.py new file mode 100644 index 000000000..22d8b4b0d --- /dev/null +++ b/synapse/handlers/room_member_worker.py @@ -0,0 +1,103 @@ +# -*- coding: utf-8 -*- +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging + +from twisted.internet import defer + +from synapse.api.errors import SynapseError +from synapse.handlers.room_member import RoomMemberHandler +from synapse.replication.http.membership import ( + get_or_register_3pid_guest, + notify_user_membership_change, + remote_join, + remote_reject_invite, +) + +logger = logging.getLogger(__name__) + + +class RoomMemberWorkerHandler(RoomMemberHandler): + @defer.inlineCallbacks + def _remote_join(self, requester, remote_room_hosts, room_id, user, content): + """Implements RoomMemberHandler._remote_join + """ + if len(remote_room_hosts) == 0: + raise SynapseError(404, "No known servers") + + ret = yield remote_join( + self.simple_http_client, + host=self.config.worker_replication_host, + port=self.config.worker_replication_http_port, + requester=requester, + remote_room_hosts=remote_room_hosts, + room_id=room_id, + user_id=user.to_string(), + content=content, + ) + + yield self._user_joined_room(user, room_id) + + defer.returnValue(ret) + + def _remote_reject_invite(self, requester, remote_room_hosts, room_id, target): + """Implements RoomMemberHandler._remote_reject_invite + """ + return remote_reject_invite( + self.simple_http_client, + host=self.config.worker_replication_host, + port=self.config.worker_replication_http_port, + requester=requester, + remote_room_hosts=remote_room_hosts, + room_id=room_id, + user_id=target.to_string(), + ) + + def _user_joined_room(self, target, room_id): + """Implements RoomMemberHandler._user_joined_room + """ + return notify_user_membership_change( + self.simple_http_client, + host=self.config.worker_replication_host, + port=self.config.worker_replication_http_port, + user_id=target.to_string(), + room_id=room_id, + change="joined", + ) + + def _user_left_room(self, target, room_id): + """Implements RoomMemberHandler._user_left_room + """ + return notify_user_membership_change( + self.simple_http_client, + host=self.config.worker_replication_host, + port=self.config.worker_replication_http_port, + user_id=target.to_string(), + room_id=room_id, + change="left", + ) + + def get_or_register_3pid_guest(self, requester, medium, address, inviter_user_id): + """Implements RoomMemberHandler.get_or_register_3pid_guest + """ + return get_or_register_3pid_guest( + self.simple_http_client, + host=self.config.worker_replication_host, + port=self.config.worker_replication_http_port, + requester=requester, + medium=medium, + address=address, + inviter_user_id=inviter_user_id, + ) diff --git a/synapse/handlers/search.py b/synapse/handlers/search.py index df75d70fa..69ae9731d 100644 --- a/synapse/handlers/search.py +++ b/synapse/handlers/search.py @@ -13,21 +13,20 @@ # See the License for the specific language governing permissions and # limitations under the License. -from twisted.internet import defer - -from ._base import BaseHandler - -from synapse.api.constants import Membership, EventTypes -from synapse.api.filtering import Filter -from synapse.api.errors import SynapseError -from synapse.events.utils import serialize_event -from synapse.visibility import filter_events_for_client - -from unpaddedbase64 import decode_base64, encode_base64 - import itertools import logging +from unpaddedbase64 import decode_base64, encode_base64 + +from twisted.internet import defer + +from synapse.api.constants import EventTypes, Membership +from synapse.api.errors import SynapseError +from synapse.api.filtering import Filter +from synapse.events.utils import serialize_event +from synapse.visibility import filter_events_for_client + +from ._base import BaseHandler logger = logging.getLogger(__name__) @@ -61,9 +60,16 @@ class SearchHandler(BaseHandler): assert batch_group is not None assert batch_group_key is not None assert batch_token is not None - except: + except Exception: raise SynapseError(400, "Invalid batch") + logger.info( + "Search batch properties: %r, %r, %r", + batch_group, batch_group_key, batch_token, + ) + + logger.info("Search content: %s", content) + try: room_cat = content["search_categories"]["room_events"] @@ -271,6 +277,8 @@ class SearchHandler(BaseHandler): # We should never get here due to the guard earlier. raise NotImplementedError() + logger.info("Found %d events to return", len(allowed_events)) + # If client has asked for "context" for each event (i.e. some surrounding # events and state), fetch that if event_context is not None: @@ -282,6 +290,11 @@ class SearchHandler(BaseHandler): event.room_id, event.event_id, before_limit, after_limit ) + logger.info( + "Context for search returned %d and %d events", + len(res["events_before"]), len(res["events_after"]), + ) + res["events_before"] = yield filter_events_for_client( self.store, user.to_string(), res["events_before"] ) @@ -348,7 +361,7 @@ class SearchHandler(BaseHandler): rooms = set(e.room_id for e in allowed_events) for room_id in rooms: state = yield self.state_handler.get_current_state(room_id) - state_results[room_id] = state.values() + state_results[room_id] = list(state.values()) state_results.values() diff --git a/synapse/handlers/set_password.py b/synapse/handlers/set_password.py new file mode 100644 index 000000000..7ecdede4d --- /dev/null +++ b/synapse/handlers/set_password.py @@ -0,0 +1,57 @@ +# -*- coding: utf-8 -*- +# Copyright 2017 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import logging + +from twisted.internet import defer + +from synapse.api.errors import Codes, StoreError, SynapseError + +from ._base import BaseHandler + +logger = logging.getLogger(__name__) + + +class SetPasswordHandler(BaseHandler): + """Handler which deals with changing user account passwords""" + def __init__(self, hs): + super(SetPasswordHandler, self).__init__(hs) + self._auth_handler = hs.get_auth_handler() + self._device_handler = hs.get_device_handler() + + @defer.inlineCallbacks + def set_password(self, user_id, newpassword, requester=None): + password_hash = yield self._auth_handler.hash(newpassword) + + except_device_id = requester.device_id if requester else None + except_access_token_id = requester.access_token_id if requester else None + + try: + yield self.store.user_set_password_hash(user_id, password_hash) + except StoreError as e: + if e.code == 404: + raise SynapseError(404, "Unknown user", Codes.NOT_FOUND) + raise e + + # we want to log out all of the user's other sessions. First delete + # all his other devices. + yield self._device_handler.delete_all_devices_for_user( + user_id, except_device_id=except_device_id, + ) + + # and now delete any access tokens which weren't associated with + # devices (or were associated with this device). + yield self._auth_handler.delete_access_tokens_for_user( + user_id, except_token_id=except_access_token_id, + ) diff --git a/synapse/handlers/sync.py b/synapse/handlers/sync.py index 91c6c6be3..c24e35362 100644 --- a/synapse/handlers/sync.py +++ b/synapse/handlers/sync.py @@ -13,20 +13,22 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.api.constants import Membership, EventTypes -from synapse.util.async import concurrently_execute -from synapse.util.logcontext import LoggingContext -from synapse.util.metrics import Measure, measure_func -from synapse.util.caches.response_cache import ResponseCache -from synapse.push.clientformat import format_push_rules_for_user -from synapse.visibility import filter_events_for_client -from synapse.types import RoomStreamToken +import collections +import itertools +import logging + +from six import iteritems, itervalues from twisted.internet import defer -import collections -import logging -import itertools +from synapse.api.constants import EventTypes, Membership +from synapse.push.clientformat import format_push_rules_for_user +from synapse.types import RoomStreamToken +from synapse.util.async import concurrently_execute +from synapse.util.caches.response_cache import ResponseCache +from synapse.util.logcontext import LoggingContext +from synapse.util.metrics import Measure, measure_func +from synapse.visibility import filter_events_for_client logger = logging.getLogger(__name__) @@ -52,6 +54,7 @@ class TimelineBatch(collections.namedtuple("TimelineBatch", [ to tell if room needs to be part of the sync result. """ return bool(self.events) + __bool__ = __nonzero__ # python3 class JoinedSyncResult(collections.namedtuple("JoinedSyncResult", [ @@ -76,6 +79,7 @@ class JoinedSyncResult(collections.namedtuple("JoinedSyncResult", [ # nb the notification count does not, er, count: if there's nothing # else in the result, we don't need to send it. ) + __bool__ = __nonzero__ # python3 class ArchivedSyncResult(collections.namedtuple("ArchivedSyncResult", [ @@ -95,6 +99,7 @@ class ArchivedSyncResult(collections.namedtuple("ArchivedSyncResult", [ or self.state or self.account_data ) + __bool__ = __nonzero__ # python3 class InvitedSyncResult(collections.namedtuple("InvitedSyncResult", [ @@ -106,6 +111,30 @@ class InvitedSyncResult(collections.namedtuple("InvitedSyncResult", [ def __nonzero__(self): """Invited rooms should always be reported to the client""" return True + __bool__ = __nonzero__ # python3 + + +class GroupsSyncResult(collections.namedtuple("GroupsSyncResult", [ + "join", + "invite", + "leave", +])): + __slots__ = [] + + def __nonzero__(self): + return bool(self.join or self.invite or self.leave) + __bool__ = __nonzero__ # python3 + + +class DeviceLists(collections.namedtuple("DeviceLists", [ + "changed", # list of user_ids whose devices may have changed + "left", # list of user_ids whose devices we no longer track +])): + __slots__ = [] + + def __nonzero__(self): + return bool(self.changed or self.left) + __bool__ = __nonzero__ # python3 class SyncResult(collections.namedtuple("SyncResult", [ @@ -116,9 +145,10 @@ class SyncResult(collections.namedtuple("SyncResult", [ "invited", # InvitedSyncResult for each invited room. "archived", # ArchivedSyncResult for each archived room. "to_device", # List of direct messages for the device. - "device_lists", # List of user_ids whose devices have chanegd + "device_lists", # List of user_ids whose devices have changed "device_one_time_keys_count", # Dict of algorithm to count for one time keys # for this device + "groups", ])): __slots__ = [] @@ -134,8 +164,10 @@ class SyncResult(collections.namedtuple("SyncResult", [ self.archived or self.account_data or self.to_device or - self.device_lists + self.device_lists or + self.groups ) + __bool__ = __nonzero__ # python3 class SyncHandler(object): @@ -146,7 +178,7 @@ class SyncHandler(object): self.presence_handler = hs.get_presence_handler() self.event_sources = hs.get_event_sources() self.clock = hs.get_clock() - self.response_cache = ResponseCache(hs) + self.response_cache = ResponseCache(hs, "sync") self.state = hs.get_state_handler() def wait_for_sync_for_user(self, sync_config, since_token=None, timeout=0, @@ -157,15 +189,11 @@ class SyncHandler(object): Returns: A Deferred SyncResult. """ - result = self.response_cache.get(sync_config.request_key) - if not result: - result = self.response_cache.set( - sync_config.request_key, - self._wait_for_sync_for_user( - sync_config, since_token, timeout, full_state - ) - ) - return result + return self.response_cache.wrap( + sync_config.request_key, + self._wait_for_sync_for_user, + sync_config, since_token, timeout, full_state, + ) @defer.inlineCallbacks def _wait_for_sync_for_user(self, sync_config, since_token, timeout, @@ -212,10 +240,10 @@ class SyncHandler(object): defer.returnValue(rules) @defer.inlineCallbacks - def ephemeral_by_room(self, sync_config, now_token, since_token=None): + def ephemeral_by_room(self, sync_result_builder, now_token, since_token=None): """Get the ephemeral events for each room the user is in Args: - sync_config (SyncConfig): The flags, filters and user for the sync. + sync_result_builder(SyncResultBuilder) now_token (StreamToken): Where the server is currently up to. since_token (StreamToken): Where the server was when the client last synced. @@ -225,10 +253,12 @@ class SyncHandler(object): typing events for that room. """ + sync_config = sync_result_builder.sync_config + with Measure(self.clock, "ephemeral_by_room"): typing_key = since_token.typing_key if since_token else "0" - room_ids = yield self.store.get_rooms_for_user(sync_config.user.to_string()) + room_ids = sync_result_builder.joined_room_ids typing_source = self.event_sources.sources["typing"] typing, typing_key = yield typing_source.get_new_events( @@ -247,7 +277,7 @@ class SyncHandler(object): # result returned by the event source is poor form (it might cache # the object) room_id = event["room_id"] - event_copy = {k: v for (k, v) in event.iteritems() + event_copy = {k: v for (k, v) in iteritems(event) if k != "room_id"} ephemeral_by_room.setdefault(room_id, []).append(event_copy) @@ -266,7 +296,7 @@ class SyncHandler(object): for event in receipts: room_id = event["room_id"] # exclude room id, as above - event_copy = {k: v for (k, v) in event.iteritems() + event_copy = {k: v for (k, v) in iteritems(event) if k != "room_id"} ephemeral_by_room.setdefault(room_id, []).append(event_copy) @@ -290,10 +320,20 @@ class SyncHandler(object): if recents: recents = sync_config.filter_collection.filter_room_timeline(recents) + + # We check if there are any state events, if there are then we pass + # all current state events to the filter_events function. This is to + # ensure that we always include current state in the timeline + current_state_ids = frozenset() + if any(e.is_state() for e in recents): + current_state_ids = yield self.state.get_current_state_ids(room_id) + current_state_ids = frozenset(itervalues(current_state_ids)) + recents = yield filter_events_for_client( self.store, sync_config.user.to_string(), recents, + always_include_ids=current_state_ids, ) else: recents = [] @@ -316,19 +356,41 @@ class SyncHandler(object): since_key = since_token.room_key while limited and len(recents) < timeline_limit and max_repeat: - events, end_key = yield self.store.get_room_events_stream_for_room( - room_id, - limit=load_limit + 1, - from_key=since_key, - to_key=end_key, - ) + # If we have a since_key then we are trying to get any events + # that have happened since `since_key` up to `end_key`, so we + # can just use `get_room_events_stream_for_room`. + # Otherwise, we want to return the last N events in the room + # in toplogical ordering. + if since_key: + events, end_key = yield self.store.get_room_events_stream_for_room( + room_id, + limit=load_limit + 1, + from_key=since_key, + to_key=end_key, + ) + else: + events, end_key = yield self.store.get_recent_events_for_room( + room_id, + limit=load_limit + 1, + end_token=end_key, + ) loaded_recents = sync_config.filter_collection.filter_room_timeline( events ) + + # We check if there are any state events, if there are then we pass + # all current state events to the filter_events function. This is to + # ensure that we always include current state in the timeline + current_state_ids = frozenset() + if any(e.is_state() for e in loaded_recents): + current_state_ids = yield self.state.get_current_state_ids(room_id) + current_state_ids = frozenset(itervalues(current_state_ids)) + loaded_recents = yield filter_events_for_client( self.store, sync_config.user.to_string(), loaded_recents, + always_include_ids=current_state_ids, ) loaded_recents.extend(recents) recents = loaded_recents @@ -381,7 +443,11 @@ class SyncHandler(object): Returns: A Deferred map from ((type, state_key)->Event) """ - last_events, token = yield self.store.get_recent_events_for_room( + # FIXME this claims to get the state at a stream position, but + # get_recent_events_for_room operates by topo ordering. This therefore + # does not reliably give you the state at the given stream position. + # (https://github.com/matrix-org/synapse/issues/3305) + last_events, _ = yield self.store.get_recent_events_for_room( room_id, end_token=stream_position.room_key, limit=1, ) @@ -475,11 +541,11 @@ class SyncHandler(object): state = {} if state_ids: - state = yield self.store.get_events(state_ids.values()) + state = yield self.store.get_events(list(state_ids.values())) defer.returnValue({ (e.type, e.state_key): e - for e in sync_config.filter_collection.filter_room_state(state.values()) + for e in sync_config.filter_collection.filter_room_state(list(state.values())) }) @defer.inlineCallbacks @@ -522,10 +588,22 @@ class SyncHandler(object): # Always use the `now_token` in `SyncResultBuilder` now_token = yield self.event_sources.get_current_token() + user_id = sync_config.user.to_string() + app_service = self.store.get_app_service_by_user_id(user_id) + if app_service: + # We no longer support AS users using /sync directly. + # See https://github.com/matrix-org/matrix-doc/issues/1144 + raise NotImplementedError() + else: + joined_room_ids = yield self.get_rooms_for_user_at( + user_id, now_token.room_stream_id, + ) + sync_result_builder = SyncResultBuilder( sync_config, full_state, since_token=since_token, now_token=now_token, + joined_room_ids=joined_room_ids, ) account_data_by_room = yield self._generate_sync_entry_for_account_data( @@ -535,7 +613,8 @@ class SyncHandler(object): res = yield self._generate_sync_entry_for_rooms( sync_result_builder, account_data_by_room ) - newly_joined_rooms, newly_joined_users = res + newly_joined_rooms, newly_joined_users, _, _ = res + _, _, newly_left_rooms, newly_left_users = res block_all_presence_data = ( since_token is None and @@ -549,17 +628,22 @@ class SyncHandler(object): yield self._generate_sync_entry_for_to_device(sync_result_builder) device_lists = yield self._generate_sync_entry_for_device_list( - sync_result_builder + sync_result_builder, + newly_joined_rooms=newly_joined_rooms, + newly_joined_users=newly_joined_users, + newly_left_rooms=newly_left_rooms, + newly_left_users=newly_left_users, ) device_id = sync_config.device_id one_time_key_counts = {} if device_id: - user_id = sync_config.user.to_string() one_time_key_counts = yield self.store.count_e2e_one_time_keys( user_id, device_id ) + yield self._generate_sync_entry_for_groups(sync_result_builder) + defer.returnValue(SyncResult( presence=sync_result_builder.presence, account_data=sync_result_builder.account_data, @@ -568,31 +652,103 @@ class SyncHandler(object): archived=sync_result_builder.archived, to_device=sync_result_builder.to_device, device_lists=device_lists, + groups=sync_result_builder.groups, device_one_time_keys_count=one_time_key_counts, next_batch=sync_result_builder.now_token, )) + @measure_func("_generate_sync_entry_for_groups") + @defer.inlineCallbacks + def _generate_sync_entry_for_groups(self, sync_result_builder): + user_id = sync_result_builder.sync_config.user.to_string() + since_token = sync_result_builder.since_token + now_token = sync_result_builder.now_token + + if since_token and since_token.groups_key: + results = yield self.store.get_groups_changes_for_user( + user_id, since_token.groups_key, now_token.groups_key, + ) + else: + results = yield self.store.get_all_groups_for_user( + user_id, now_token.groups_key, + ) + + invited = {} + joined = {} + left = {} + for result in results: + membership = result["membership"] + group_id = result["group_id"] + gtype = result["type"] + content = result["content"] + + if membership == "join": + if gtype == "membership": + # TODO: Add profile + content.pop("membership", None) + joined[group_id] = content["content"] + else: + joined.setdefault(group_id, {})[gtype] = content + elif membership == "invite": + if gtype == "membership": + content.pop("membership", None) + invited[group_id] = content["content"] + else: + if gtype == "membership": + left[group_id] = content["content"] + + sync_result_builder.groups = GroupsSyncResult( + join=joined, + invite=invited, + leave=left, + ) + @measure_func("_generate_sync_entry_for_device_list") @defer.inlineCallbacks - def _generate_sync_entry_for_device_list(self, sync_result_builder): + def _generate_sync_entry_for_device_list(self, sync_result_builder, + newly_joined_rooms, newly_joined_users, + newly_left_rooms, newly_left_users): user_id = sync_result_builder.sync_config.user.to_string() since_token = sync_result_builder.since_token if since_token and since_token.device_list_key: - room_ids = yield self.store.get_rooms_for_user(user_id) - - user_ids_changed = set() changed = yield self.store.get_user_whose_devices_changed( since_token.device_list_key ) - for other_user_id in changed: - other_room_ids = yield self.store.get_rooms_for_user(other_user_id) - if room_ids.intersection(other_room_ids): - user_ids_changed.add(other_user_id) - defer.returnValue(user_ids_changed) + # TODO: Be more clever than this, i.e. remove users who we already + # share a room with? + for room_id in newly_joined_rooms: + joined_users = yield self.state.get_current_user_in_room(room_id) + newly_joined_users.update(joined_users) + + for room_id in newly_left_rooms: + left_users = yield self.state.get_current_user_in_room(room_id) + newly_left_users.update(left_users) + + # TODO: Check that these users are actually new, i.e. either they + # weren't in the previous sync *or* they left and rejoined. + changed.update(newly_joined_users) + + if not changed and not newly_left_users: + defer.returnValue(DeviceLists( + changed=[], + left=newly_left_users, + )) + + users_who_share_room = yield self.store.get_users_who_share_room_with_user( + user_id + ) + + defer.returnValue(DeviceLists( + changed=users_who_share_room & changed, + left=set(newly_left_users) - users_who_share_room, + )) else: - defer.returnValue([]) + defer.returnValue(DeviceLists( + changed=[], + left=[], + )) @defer.inlineCallbacks def _generate_sync_entry_for_to_device(self, sync_result_builder): @@ -738,7 +894,7 @@ class SyncHandler(object): presence.extend(states) # Deduplicate the presence entries so that there's at most one per user - presence = {p.user_id: p for p in presence}.values() + presence = list({p.user_id: p for p in presence}.values()) presence = sync_config.filter_collection.filter_presence( presence @@ -756,8 +912,8 @@ class SyncHandler(object): account_data_by_room(dict): Dictionary of per room account data Returns: - Deferred(tuple): Returns a 2-tuple of - `(newly_joined_rooms, newly_joined_users)` + Deferred(tuple): Returns a 4-tuple of + `(newly_joined_rooms, newly_joined_users, newly_left_rooms, newly_left_users)` """ user_id = sync_result_builder.sync_config.user.to_string() block_all_room_ephemeral = ( @@ -769,7 +925,7 @@ class SyncHandler(object): ephemeral_by_room = {} else: now_token, ephemeral_by_room = yield self.ephemeral_by_room( - sync_result_builder.sync_config, + sync_result_builder, now_token=sync_result_builder.now_token, since_token=sync_result_builder.since_token, ) @@ -788,7 +944,7 @@ class SyncHandler(object): ) if not tags_by_room: logger.debug("no-oping sync") - defer.returnValue(([], [])) + defer.returnValue(([], [], [], [])) ignored_account_data = yield self.store.get_global_account_data_by_type_for_user( "m.ignored_user_list", user_id=user_id, @@ -801,7 +957,7 @@ class SyncHandler(object): if since_token: res = yield self._get_rooms_changed(sync_result_builder, ignored_users) - room_entries, invited, newly_joined_rooms = res + room_entries, invited, newly_joined_rooms, newly_left_rooms = res tags_by_room = yield self.store.get_updated_tags( user_id, since_token.account_data_key, @@ -809,6 +965,7 @@ class SyncHandler(object): else: res = yield self._get_all_rooms(sync_result_builder, ignored_users) room_entries, invited, newly_joined_rooms = res + newly_left_rooms = [] tags_by_room = yield self.store.get_tags_for_user(user_id) @@ -829,17 +986,30 @@ class SyncHandler(object): # Now we want to get any newly joined users newly_joined_users = set() + newly_left_users = set() if since_token: for joined_sync in sync_result_builder.joined: it = itertools.chain( - joined_sync.timeline.events, joined_sync.state.values() + joined_sync.timeline.events, itervalues(joined_sync.state) ) for event in it: if event.type == EventTypes.Member: if event.membership == Membership.JOIN: newly_joined_users.add(event.state_key) + else: + prev_content = event.unsigned.get("prev_content", {}) + prev_membership = prev_content.get("membership", None) + if prev_membership == Membership.JOIN: + newly_left_users.add(event.state_key) - defer.returnValue((newly_joined_rooms, newly_joined_users)) + newly_left_users -= newly_joined_users + + defer.returnValue(( + newly_joined_rooms, + newly_joined_users, + newly_left_rooms, + newly_left_users, + )) @defer.inlineCallbacks def _have_rooms_changed(self, sync_result_builder): @@ -860,15 +1030,8 @@ class SyncHandler(object): if rooms_changed: defer.returnValue(True) - app_service = self.store.get_app_service_by_user_id(user_id) - if app_service: - rooms = yield self.store.get_app_service_rooms(app_service) - joined_room_ids = set(r.room_id for r in rooms) - else: - joined_room_ids = yield self.store.get_rooms_for_user(user_id) - stream_id = RoomStreamToken.parse_stream_token(since_token.room_key).stream - for room_id in joined_room_ids: + for room_id in sync_result_builder.joined_room_ids: if self.store.has_room_changed_since(room_id, stream_id): defer.returnValue(True) defer.returnValue(False) @@ -883,7 +1046,13 @@ class SyncHandler(object): Returns: Deferred(tuple): Returns a tuple of the form: - `([RoomSyncResultBuilder], [InvitedSyncResult], newly_joined_rooms)` + `(room_entries, invited_rooms, newly_joined_rooms, newly_left_rooms)` + + where: + room_entries is a list [RoomSyncResultBuilder] + invited_rooms is a list [InvitedSyncResult] + newly_joined rooms is a list[str] of room ids + newly_left_rooms is a list[str] of room ids """ user_id = sync_result_builder.sync_config.user.to_string() since_token = sync_result_builder.since_token @@ -892,13 +1061,6 @@ class SyncHandler(object): assert since_token - app_service = self.store.get_app_service_by_user_id(user_id) - if app_service: - rooms = yield self.store.get_app_service_rooms(app_service) - joined_room_ids = set(r.room_id for r in rooms) - else: - joined_room_ids = yield self.store.get_rooms_for_user(user_id) - # Get a list of membership change events that have happened. rooms_changed = yield self.store.get_membership_changes_for_user( user_id, since_token.room_key, now_token.room_key @@ -909,16 +1071,29 @@ class SyncHandler(object): mem_change_events_by_room_id.setdefault(event.room_id, []).append(event) newly_joined_rooms = [] + newly_left_rooms = [] room_entries = [] invited = [] - for room_id, events in mem_change_events_by_room_id.items(): + for room_id, events in iteritems(mem_change_events_by_room_id): non_joins = [e for e in events if e.membership != Membership.JOIN] has_join = len(non_joins) != len(events) # We want to figure out if we joined the room at some point since # the last sync (even if we have since left). This is to make sure # we do send down the room, and with full state, where necessary - if room_id in joined_room_ids or has_join: + + old_state_ids = None + if room_id in sync_result_builder.joined_room_ids and non_joins: + # Always include if the user (re)joined the room, especially + # important so that device list changes are calculated correctly. + # If there are non join member events, but we are still in the room, + # then the user must have left and joined + newly_joined_rooms.append(room_id) + + # User is in the room so we don't need to do the invite/leave checks + continue + + if room_id in sync_result_builder.joined_room_ids or has_join: old_state_ids = yield self.get_state_at(room_id, since_token) old_mem_ev_id = old_state_ids.get((EventTypes.Member, user_id), None) old_mem_ev = None @@ -929,12 +1104,33 @@ class SyncHandler(object): if not old_mem_ev or old_mem_ev.membership != Membership.JOIN: newly_joined_rooms.append(room_id) - if room_id in joined_room_ids: - continue + # If user is in the room then we don't need to do the invite/leave checks + if room_id in sync_result_builder.joined_room_ids: + continue if not non_joins: continue + # Check if we have left the room. This can either be because we were + # joined before *or* that we since joined and then left. + if events[-1].membership != Membership.JOIN: + if has_join: + newly_left_rooms.append(room_id) + else: + if not old_state_ids: + old_state_ids = yield self.get_state_at(room_id, since_token) + old_mem_ev_id = old_state_ids.get( + (EventTypes.Member, user_id), + None, + ) + old_mem_ev = None + if old_mem_ev_id: + old_mem_ev = yield self.store.get_event( + old_mem_ev_id, allow_none=True + ) + if old_mem_ev and old_mem_ev.membership == Membership.JOIN: + newly_left_rooms.append(room_id) + # Only bother if we're still currently invited should_invite = non_joins[-1].membership == Membership.INVITE if should_invite: @@ -976,7 +1172,7 @@ class SyncHandler(object): # Get all events for rooms we're currently joined to. room_to_events = yield self.store.get_room_events_stream_for_rooms( - room_ids=joined_room_ids, + room_ids=sync_result_builder.joined_room_ids, from_key=since_token.room_key, to_key=now_token.room_key, limit=timeline_limit + 1, @@ -984,7 +1180,7 @@ class SyncHandler(object): # We loop through all room ids, even if there are no new events, in case # there are non room events taht we need to notify about. - for room_id in joined_room_ids: + for room_id in sync_result_builder.joined_room_ids: room_entry = room_to_events.get(room_id, None) if room_entry: @@ -1012,7 +1208,7 @@ class SyncHandler(object): upto_token=since_token, )) - defer.returnValue((room_entries, invited, newly_joined_rooms)) + defer.returnValue((room_entries, invited, newly_joined_rooms, newly_left_rooms)) @defer.inlineCallbacks def _get_all_rooms(self, sync_result_builder, ignored_users): @@ -1192,6 +1388,54 @@ class SyncHandler(object): else: raise Exception("Unrecognized rtype: %r", room_builder.rtype) + @defer.inlineCallbacks + def get_rooms_for_user_at(self, user_id, stream_ordering): + """Get set of joined rooms for a user at the given stream ordering. + + The stream ordering *must* be recent, otherwise this may throw an + exception if older than a month. (This function is called with the + current token, which should be perfectly fine). + + Args: + user_id (str) + stream_ordering (int) + + ReturnValue: + Deferred[frozenset[str]]: Set of room_ids the user is in at given + stream_ordering. + """ + joined_rooms = yield self.store.get_rooms_for_user_with_stream_ordering( + user_id, + ) + + joined_room_ids = set() + + # We need to check that the stream ordering of the join for each room + # is before the stream_ordering asked for. This might not be the case + # if the user joins a room between us getting the current token and + # calling `get_rooms_for_user_with_stream_ordering`. + # If the membership's stream ordering is after the given stream + # ordering, we need to go and work out if the user was in the room + # before. + for room_id, membership_stream_ordering in joined_rooms: + if membership_stream_ordering <= stream_ordering: + joined_room_ids.add(room_id) + continue + + logger.info("User joined room after current token: %s", room_id) + + extrems = yield self.store.get_forward_extremeties_for_room( + room_id, stream_ordering, + ) + users_in_room = yield self.state.get_current_user_in_room( + room_id, extrems, + ) + if user_id in users_in_room: + joined_room_ids.add(room_id) + + joined_room_ids = frozenset(joined_room_ids) + defer.returnValue(joined_room_ids) + def _action_has_highlight(actions): for action in actions: @@ -1241,7 +1485,8 @@ def _calculate_state(timeline_contains, timeline_start, previous, current): class SyncResultBuilder(object): "Used to help build up a new SyncResult for a user" - def __init__(self, sync_config, full_state, since_token, now_token): + def __init__(self, sync_config, full_state, since_token, now_token, + joined_room_ids): """ Args: sync_config(SyncConfig) @@ -1253,6 +1498,7 @@ class SyncResultBuilder(object): self.full_state = full_state self.since_token = since_token self.now_token = now_token + self.joined_room_ids = joined_room_ids self.presence = [] self.account_data = [] @@ -1260,6 +1506,8 @@ class SyncResultBuilder(object): self.invited = [] self.archived = [] self.device = [] + self.groups = None + self.to_device = [] class RoomSyncResultBuilder(object): diff --git a/synapse/handlers/typing.py b/synapse/handlers/typing.py index 82dedbbc9..2d2d3d5a0 100644 --- a/synapse/handlers/typing.py +++ b/synapse/handlers/typing.py @@ -13,17 +13,16 @@ # See the License for the specific language governing permissions and # limitations under the License. +import logging +from collections import namedtuple + from twisted.internet import defer -from synapse.api.errors import SynapseError, AuthError -from synapse.util.logcontext import preserve_fn +from synapse.api.errors import AuthError, SynapseError +from synapse.types import UserID, get_domain_from_id +from synapse.util.logcontext import run_in_background from synapse.util.metrics import Measure from synapse.util.wheel_timer import WheelTimer -from synapse.types import UserID, get_domain_from_id - -import logging - -from collections import namedtuple logger = logging.getLogger(__name__) @@ -56,7 +55,7 @@ class TypingHandler(object): self.federation = hs.get_federation_sender() - hs.get_replication_layer().register_edu_handler("m.typing", self._recv_edu) + hs.get_federation_registry().register_edu_handler("m.typing", self._recv_edu) hs.get_distributor().observe("user_left_room", self.user_left_room) @@ -97,7 +96,8 @@ class TypingHandler(object): if self.hs.is_mine_id(member.user_id): last_fed_poke = self._member_last_federation_poke.get(member, None) if not last_fed_poke or last_fed_poke + FEDERATION_PING_INTERVAL <= now: - preserve_fn(self._push_remote)( + run_in_background( + self._push_remote, member=member, typing=True ) @@ -196,7 +196,7 @@ class TypingHandler(object): def _push_update(self, member, typing): if self.hs.is_mine_id(member.user_id): # Only send updates for changes to our own users. - preserve_fn(self._push_remote)(member, typing) + run_in_background(self._push_remote, member, typing) self._push_update_local( member=member, @@ -205,28 +205,31 @@ class TypingHandler(object): @defer.inlineCallbacks def _push_remote(self, member, typing): - users = yield self.state.get_current_user_in_room(member.room_id) - self._member_last_federation_poke[member] = self.clock.time_msec() + try: + users = yield self.state.get_current_user_in_room(member.room_id) + self._member_last_federation_poke[member] = self.clock.time_msec() - now = self.clock.time_msec() - self.wheel_timer.insert( - now=now, - obj=member, - then=now + FEDERATION_PING_INTERVAL, - ) + now = self.clock.time_msec() + self.wheel_timer.insert( + now=now, + obj=member, + then=now + FEDERATION_PING_INTERVAL, + ) - for domain in set(get_domain_from_id(u) for u in users): - if domain != self.server_name: - self.federation.send_edu( - destination=domain, - edu_type="m.typing", - content={ - "room_id": member.room_id, - "user_id": member.user_id, - "typing": typing, - }, - key=member, - ) + for domain in set(get_domain_from_id(u) for u in users): + if domain != self.server_name: + self.federation.send_edu( + destination=domain, + edu_type="m.typing", + content={ + "room_id": member.room_id, + "user_id": member.user_id, + "typing": typing, + }, + key=member, + ) + except Exception: + logger.exception("Error pushing typing notif to remotes") @defer.inlineCallbacks def _recv_edu(self, origin, content): diff --git a/synapse/handlers/user_directory.py b/synapse/handlers/user_directory.py index 2a49456bf..37dda6458 100644 --- a/synapse/handlers/user_directory.py +++ b/synapse/handlers/user_directory.py @@ -14,18 +14,20 @@ # limitations under the License. import logging + +from six import iteritems + from twisted.internet import defer from synapse.api.constants import EventTypes, JoinRules, Membership from synapse.storage.roommember import ProfileInfo +from synapse.types import get_localpart_from_id from synapse.util.metrics import Measure -from synapse.util.async import sleep - logger = logging.getLogger(__name__) -class UserDirectoyHandler(object): +class UserDirectoryHandler(object): """Handles querying of and keeping updated the user_directory. N.B.: ASSUMES IT IS THE ONLY THING THAT MODIFIES THE USER DIRECTORY @@ -41,9 +43,10 @@ class UserDirectoyHandler(object): one public room. """ - INITIAL_SLEEP_MS = 50 - INITIAL_SLEEP_COUNT = 100 - INITIAL_BATCH_SIZE = 100 + INITIAL_ROOM_SLEEP_MS = 50 + INITIAL_ROOM_SLEEP_COUNT = 100 + INITIAL_ROOM_BATCH_SIZE = 100 + INITIAL_USER_SLEEP_MS = 10 def __init__(self, hs): self.store = hs.get_datastore() @@ -53,6 +56,7 @@ class UserDirectoyHandler(object): self.notifier = hs.get_notifier() self.is_mine_id = hs.is_mine_id self.update_user_directory = hs.config.update_user_directory + self.search_all_users = hs.config.user_directory_search_all_users # When start up for the first time we need to populate the user_directory. # This is a set of user_id's we've inserted already @@ -110,6 +114,22 @@ class UserDirectoyHandler(object): finally: self._is_processing = False + @defer.inlineCallbacks + def handle_local_profile_change(self, user_id, profile): + """Called to update index of our local user profiles when they change + irrespective of any rooms the user may be in. + """ + yield self.store.update_profile_in_user_dir( + user_id, profile.display_name, profile.avatar_url, None, + ) + + @defer.inlineCallbacks + def handle_user_deactivated(self, user_id): + """Called when a user ID is deactivated + """ + yield self.store.remove_from_user_dir(user_id) + yield self.store.remove_from_user_in_public_room(user_id) + @defer.inlineCallbacks def _unsafe_process(self): # If self.pos is None then means we haven't fetched it from DB @@ -148,16 +168,30 @@ class UserDirectoyHandler(object): room_ids = yield self.store.get_all_rooms() logger.info("Doing initial update of user directory. %d rooms", len(room_ids)) - num_processed_rooms = 1 + num_processed_rooms = 0 for room_id in room_ids: - logger.info("Handling room %d/%d", num_processed_rooms, len(room_ids)) - yield self._handle_intial_room(room_id) + logger.info("Handling room %d/%d", num_processed_rooms + 1, len(room_ids)) + yield self._handle_initial_room(room_id) num_processed_rooms += 1 - yield sleep(self.INITIAL_SLEEP_MS / 1000.) + yield self.clock.sleep(self.INITIAL_ROOM_SLEEP_MS / 1000.) logger.info("Processed all rooms.") + if self.search_all_users: + num_processed_users = 0 + user_ids = yield self.store.get_all_local_users() + logger.info("Doing initial update of user directory. %d users", len(user_ids)) + for user_id in user_ids: + # We add profiles for all users even if they don't match the + # include pattern, just in case we want to change it in future + logger.info("Handling user %d/%d", num_processed_users + 1, len(user_ids)) + yield self._handle_local_user(user_id) + num_processed_users += 1 + yield self.clock.sleep(self.INITIAL_USER_SLEEP_MS / 1000.) + + logger.info("Processed all users") + self.initially_handled_users = None self.initially_handled_users_in_public = None self.initially_handled_users_share = None @@ -166,7 +200,7 @@ class UserDirectoyHandler(object): yield self.store.update_user_directory_stream_pos(new_pos) @defer.inlineCallbacks - def _handle_intial_room(self, room_id): + def _handle_initial_room(self, room_id): """Called when we initially fill out user_directory one room at a time """ is_in_room = yield self.store.is_host_joined(room_id, self.server_name) @@ -201,8 +235,8 @@ class UserDirectoyHandler(object): to_update = set() count = 0 for user_id in user_ids: - if count % self.INITIAL_SLEEP_COUNT == 0: - yield sleep(self.INITIAL_SLEEP_MS / 1000.) + if count % self.INITIAL_ROOM_SLEEP_COUNT == 0: + yield self.clock.sleep(self.INITIAL_ROOM_SLEEP_MS / 1000.) if not self.is_mine_id(user_id): count += 1 @@ -216,8 +250,8 @@ class UserDirectoyHandler(object): if user_id == other_user_id: continue - if count % self.INITIAL_SLEEP_COUNT == 0: - yield sleep(self.INITIAL_SLEEP_MS / 1000.) + if count % self.INITIAL_ROOM_SLEEP_COUNT == 0: + yield self.clock.sleep(self.INITIAL_ROOM_SLEEP_MS / 1000.) count += 1 user_set = (user_id, other_user_id) @@ -237,13 +271,13 @@ class UserDirectoyHandler(object): else: self.initially_handled_users_share_private_room.add(user_set) - if len(to_insert) > self.INITIAL_BATCH_SIZE: + if len(to_insert) > self.INITIAL_ROOM_BATCH_SIZE: yield self.store.add_users_who_share_room( room_id, not is_public, to_insert, ) to_insert.clear() - if len(to_update) > self.INITIAL_BATCH_SIZE: + if len(to_update) > self.INITIAL_ROOM_BATCH_SIZE: yield self.store.update_users_who_share_room( room_id, not is_public, to_update, ) @@ -377,22 +411,36 @@ class UserDirectoyHandler(object): if change: users_with_profile = yield self.state.get_current_user_in_room(room_id) - for user_id, profile in users_with_profile.iteritems(): + for user_id, profile in iteritems(users_with_profile): yield self._handle_new_user(room_id, user_id, profile) else: users = yield self.store.get_users_in_public_due_to_room(room_id) for user_id in users: yield self._handle_remove_user(room_id, user_id) + @defer.inlineCallbacks + def _handle_local_user(self, user_id): + """Adds a new local roomless user into the user_directory_search table. + Used to populate up the user index when we have an + user_directory_search_all_users specified. + """ + logger.debug("Adding new local user to dir, %r", user_id) + + profile = yield self.store.get_profileinfo(get_localpart_from_id(user_id)) + + row = yield self.store.get_user_in_directory(user_id) + if not row: + yield self.store.add_profiles_to_user_dir(None, {user_id: profile}) + @defer.inlineCallbacks def _handle_new_user(self, room_id, user_id, profile): """Called when we might need to add user to directory Args: - room_id (str): room_id that user joined or started being public that + room_id (str): room_id that user joined or started being public user_id (str) """ - logger.debug("Adding user to dir, %r", user_id) + logger.debug("Adding new user to dir, %r", user_id) row = yield self.store.get_user_in_directory(user_id) if not row: @@ -407,7 +455,7 @@ class UserDirectoyHandler(object): if not row: yield self.store.add_users_to_public_room(room_id, [user_id]) else: - logger.debug("Not adding user to public dir, %r", user_id) + logger.debug("Not adding new user to public dir, %r", user_id) # Now we update users who share rooms with users. We do this by getting # all the current users in the room and seeing which aren't already diff --git a/synapse/http/__init__.py b/synapse/http/__init__.py index bfebb0f64..58ef8d3ce 100644 --- a/synapse/http/__init__.py +++ b/synapse/http/__init__.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2014-2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -12,3 +13,37 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. +import re + +from twisted.internet.defer import CancelledError +from twisted.python import failure + +from synapse.api.errors import SynapseError + + +class RequestTimedOutError(SynapseError): + """Exception representing timeout of an outbound request""" + def __init__(self): + super(RequestTimedOutError, self).__init__(504, "Timed out") + + +def cancelled_to_request_timed_out_error(value, timeout): + """Turns CancelledErrors into RequestTimedOutErrors. + + For use with async.add_timeout_to_deferred + """ + if isinstance(value, failure.Failure): + value.trap(CancelledError) + raise RequestTimedOutError() + return value + + +ACCESS_TOKEN_RE = re.compile(br'(\?.*access(_|%5[Ff])token=)[^&]*(.*)$') + + +def redact_uri(uri): + """Strips access tokens from the uri replaces with """ + return ACCESS_TOKEN_RE.sub( + br'\1\3', + uri + ) diff --git a/synapse/http/additional_resource.py b/synapse/http/additional_resource.py new file mode 100644 index 000000000..0e10e3f8f --- /dev/null +++ b/synapse/http/additional_resource.py @@ -0,0 +1,55 @@ +# -*- coding: utf-8 -*- +# Copyright 2017 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from twisted.web.resource import Resource +from twisted.web.server import NOT_DONE_YET + +from synapse.http.server import wrap_json_request_handler + + +class AdditionalResource(Resource): + """Resource wrapper for additional_resources + + If the user has configured additional_resources, we need to wrap the + handler class with a Resource so that we can map it into the resource tree. + + This class is also where we wrap the request handler with logging, metrics, + and exception handling. + """ + def __init__(self, hs, handler): + """Initialise AdditionalResource + + The ``handler`` should return a deferred which completes when it has + done handling the request. It should write a response with + ``request.write()``, and call ``request.finish()``. + + Args: + hs (synapse.server.HomeServer): homeserver + handler ((twisted.web.server.Request) -> twisted.internet.defer.Deferred): + function to be called to handle the request. + """ + Resource.__init__(self) + self._handler = handler + + # required by the request_handler wrapper + self.clock = hs.get_clock() + + def render(self, request): + self._async_render(request) + return NOT_DONE_YET + + @wrap_json_request_handler + def _async_render(self, request): + return self._handler(request) diff --git a/synapse/http/client.py b/synapse/http/client.py index 9eba046bb..25b630788 100644 --- a/synapse/http/client.py +++ b/synapse/http/client.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2014-2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -12,49 +13,49 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. -from OpenSSL import SSL -from OpenSSL.SSL import VERIFY_NONE - -from synapse.api.errors import ( - CodeMessageException, MatrixCodeMessageException, SynapseError, Codes, -) -from synapse.util.logcontext import preserve_context_over_fn -from synapse.util import logcontext -import synapse.metrics -from synapse.http.endpoint import SpiderEndpoint - -from canonicaljson import encode_canonical_json - -from twisted.internet import defer, reactor, ssl, protocol, task -from twisted.internet.endpoints import HostnameEndpoint, wrapClientTLS -from twisted.web.client import ( - BrowserLikeRedirectAgent, ContentDecoderAgent, GzipDecoder, Agent, - readBody, PartialDownloadError, -) -from twisted.web.client import FileBodyProducer as TwistedFileBodyProducer -from twisted.web.http import PotentialDataLoss -from twisted.web.http_headers import Headers -from twisted.web._newclient import ResponseDone - -from StringIO import StringIO - -import simplejson as json import logging import urllib +from six import StringIO + +from canonicaljson import encode_canonical_json, json +from prometheus_client import Counter + +from OpenSSL import SSL +from OpenSSL.SSL import VERIFY_NONE +from twisted.internet import defer, protocol, reactor, ssl, task +from twisted.internet.endpoints import HostnameEndpoint, wrapClientTLS +from twisted.web._newclient import ResponseDone +from twisted.web.client import ( + Agent, + BrowserLikeRedirectAgent, + ContentDecoderAgent, + FileBodyProducer as TwistedFileBodyProducer, + GzipDecoder, + HTTPConnectionPool, + PartialDownloadError, + readBody, +) +from twisted.web.http import PotentialDataLoss +from twisted.web.http_headers import Headers + +from synapse.api.errors import ( + CodeMessageException, + Codes, + MatrixCodeMessageException, + SynapseError, +) +from synapse.http import cancelled_to_request_timed_out_error, redact_uri +from synapse.http.endpoint import SpiderEndpoint +from synapse.util.async import add_timeout_to_deferred +from synapse.util.caches import CACHE_SIZE_FACTOR +from synapse.util.logcontext import make_deferred_yieldable logger = logging.getLogger(__name__) -metrics = synapse.metrics.get_metrics_for(__name__) - -outgoing_requests_counter = metrics.register_counter( - "requests", - labels=["method"], -) -incoming_responses_counter = metrics.register_counter( - "responses", - labels=["method", "code"], -) +outgoing_requests_counter = Counter("synapse_http_client_requests", "", ["method"]) +incoming_responses_counter = Counter("synapse_http_client_responses", "", + ["method", "code"]) class SimpleHttpClient(object): @@ -64,13 +65,23 @@ class SimpleHttpClient(object): """ def __init__(self, hs): self.hs = hs + + pool = HTTPConnectionPool(reactor) + + # the pusher makes lots of concurrent SSL connections to sygnal, and + # tends to do so in batches, so we need to allow the pool to keep lots + # of idle connections around. + pool.maxPersistentPerHost = max((100 * CACHE_SIZE_FACTOR, 5)) + pool.cachedConnectionTimeout = 2 * 60 + # The default context factory in Twisted 14.0.0 (which we require) is # BrowserLikePolicyForHTTPS which will do regular cert validation # 'like a browser' self.agent = Agent( reactor, connectTimeout=15, - contextFactory=hs.get_http_client_context_factory() + contextFactory=hs.get_http_client_context_factory(), + pool=pool, ) self.user_agent = hs.version_string self.clock = hs.get_clock() @@ -81,76 +92,103 @@ class SimpleHttpClient(object): def request(self, method, uri, *args, **kwargs): # A small wrapper around self.agent.request() so we can easily attach # counters to it - outgoing_requests_counter.inc(method) + outgoing_requests_counter.labels(method).inc() - def send_request(): + # log request but strip `access_token` (AS requests for example include this) + logger.info("Sending request %s %s", method, redact_uri(uri)) + + try: request_deferred = self.agent.request( method, uri, *args, **kwargs ) - - return self.clock.time_bound_deferred( - request_deferred, - time_out=60, + add_timeout_to_deferred( + request_deferred, 60, self.hs.get_reactor(), + cancelled_to_request_timed_out_error, ) + response = yield make_deferred_yieldable(request_deferred) - logger.info("Sending request %s %s", method, uri) - - try: - with logcontext.PreserveLoggingContext(): - response = yield send_request() - - incoming_responses_counter.inc(method, response.code) + incoming_responses_counter.labels(method, response.code).inc() logger.info( "Received response to %s %s: %s", - method, uri, response.code + method, redact_uri(uri), response.code ) defer.returnValue(response) except Exception as e: - incoming_responses_counter.inc(method, "ERR") + incoming_responses_counter.labels(method, "ERR").inc() logger.info( "Error sending request to %s %s: %s %s", - method, uri, type(e).__name__, e.message + method, redact_uri(uri), type(e).__name__, e.message ) - raise e + raise @defer.inlineCallbacks - def post_urlencoded_get_json(self, uri, args={}): + def post_urlencoded_get_json(self, uri, args={}, headers=None): + """ + Args: + uri (str): + args (dict[str, str|List[str]]): query params + headers (dict[str, List[str]]|None): If not None, a map from + header name to a list of values for that header + + Returns: + Deferred[object]: parsed json + """ + # TODO: Do we ever want to log message contents? logger.debug("post_urlencoded_get_json args: %s", args) query_bytes = urllib.urlencode(encode_urlencode_args(args), True) + actual_headers = { + b"Content-Type": [b"application/x-www-form-urlencoded"], + b"User-Agent": [self.user_agent], + } + if headers: + actual_headers.update(headers) + response = yield self.request( "POST", uri.encode("ascii"), - headers=Headers({ - b"Content-Type": [b"application/x-www-form-urlencoded"], - b"User-Agent": [self.user_agent], - }), + headers=Headers(actual_headers), bodyProducer=FileBodyProducer(StringIO(query_bytes)) ) - body = yield preserve_context_over_fn(readBody, response) + body = yield make_deferred_yieldable(readBody(response)) defer.returnValue(json.loads(body)) @defer.inlineCallbacks - def post_json_get_json(self, uri, post_json): + def post_json_get_json(self, uri, post_json, headers=None): + """ + + Args: + uri (str): + post_json (object): + headers (dict[str, List[str]]|None): If not None, a map from + header name to a list of values for that header + + Returns: + Deferred[object]: parsed json + """ json_str = encode_canonical_json(post_json) logger.debug("HTTP POST %s -> %s", json_str, uri) + actual_headers = { + b"Content-Type": [b"application/json"], + b"User-Agent": [self.user_agent], + } + if headers: + actual_headers.update(headers) + response = yield self.request( "POST", uri.encode("ascii"), - headers=Headers({ - b"Content-Type": [b"application/json"], - b"User-Agent": [self.user_agent], - }), + headers=Headers(actual_headers), bodyProducer=FileBodyProducer(StringIO(json_str)) ) - body = yield preserve_context_over_fn(readBody, response) + body = yield make_deferred_yieldable(readBody(response)) if 200 <= response.code < 300: defer.returnValue(json.loads(body)) @@ -160,7 +198,7 @@ class SimpleHttpClient(object): defer.returnValue(json.loads(body)) @defer.inlineCallbacks - def get_json(self, uri, args={}): + def get_json(self, uri, args={}, headers=None): """ Gets some json from the given URI. Args: @@ -169,6 +207,8 @@ class SimpleHttpClient(object): None. **Note**: The value of each key is assumed to be an iterable and *not* a string. + headers (dict[str, List[str]]|None): If not None, a map from + header name to a list of values for that header Returns: Deferred: Succeeds when we get *any* 2xx HTTP response, with the HTTP body as JSON. @@ -177,13 +217,13 @@ class SimpleHttpClient(object): error message. """ try: - body = yield self.get_raw(uri, args) + body = yield self.get_raw(uri, args, headers=headers) defer.returnValue(json.loads(body)) except CodeMessageException as e: raise self._exceptionFromFailedRequest(e.code, e.msg) @defer.inlineCallbacks - def put_json(self, uri, json_body, args={}): + def put_json(self, uri, json_body, args={}, headers=None): """ Puts some json to the given URI. Args: @@ -193,6 +233,8 @@ class SimpleHttpClient(object): None. **Note**: The value of each key is assumed to be an iterable and *not* a string. + headers (dict[str, List[str]]|None): If not None, a map from + header name to a list of values for that header Returns: Deferred: Succeeds when we get *any* 2xx HTTP response, with the HTTP body as JSON. @@ -205,17 +247,21 @@ class SimpleHttpClient(object): json_str = encode_canonical_json(json_body) + actual_headers = { + b"Content-Type": [b"application/json"], + b"User-Agent": [self.user_agent], + } + if headers: + actual_headers.update(headers) + response = yield self.request( "PUT", uri.encode("ascii"), - headers=Headers({ - b"User-Agent": [self.user_agent], - "Content-Type": ["application/json"] - }), + headers=Headers(actual_headers), bodyProducer=FileBodyProducer(StringIO(json_str)) ) - body = yield preserve_context_over_fn(readBody, response) + body = yield make_deferred_yieldable(readBody(response)) if 200 <= response.code < 300: defer.returnValue(json.loads(body)) @@ -226,7 +272,7 @@ class SimpleHttpClient(object): raise CodeMessageException(response.code, body) @defer.inlineCallbacks - def get_raw(self, uri, args={}): + def get_raw(self, uri, args={}, headers=None): """ Gets raw text from the given URI. Args: @@ -235,6 +281,8 @@ class SimpleHttpClient(object): None. **Note**: The value of each key is assumed to be an iterable and *not* a string. + headers (dict[str, List[str]]|None): If not None, a map from + header name to a list of values for that header Returns: Deferred: Succeeds when we get *any* 2xx HTTP response, with the HTTP body at text. @@ -246,15 +294,19 @@ class SimpleHttpClient(object): query_bytes = urllib.urlencode(args, True) uri = "%s?%s" % (uri, query_bytes) + actual_headers = { + b"User-Agent": [self.user_agent], + } + if headers: + actual_headers.update(headers) + response = yield self.request( "GET", uri.encode("ascii"), - headers=Headers({ - b"User-Agent": [self.user_agent], - }) + headers=Headers(actual_headers), ) - body = yield preserve_context_over_fn(readBody, response) + body = yield make_deferred_yieldable(readBody(response)) if 200 <= response.code < 300: defer.returnValue(body) @@ -274,27 +326,33 @@ class SimpleHttpClient(object): # The two should be factored out. @defer.inlineCallbacks - def get_file(self, url, output_stream, max_size=None): + def get_file(self, url, output_stream, max_size=None, headers=None): """GETs a file from a given URL Args: url (str): The URL to GET output_stream (file): File to write the response body to. + headers (dict[str, List[str]]|None): If not None, a map from + header name to a list of values for that header Returns: A (int,dict,string,int) tuple of the file length, dict of the response headers, absolute URI of the response and HTTP response code. """ + actual_headers = { + b"User-Agent": [self.user_agent], + } + if headers: + actual_headers.update(headers) + response = yield self.request( "GET", url.encode("ascii"), - headers=Headers({ - b"User-Agent": [self.user_agent], - }) + headers=Headers(actual_headers), ) - headers = dict(response.headers.getAllRawHeaders()) + resp_headers = dict(response.headers.getAllRawHeaders()) - if 'Content-Length' in headers and headers['Content-Length'] > max_size: + if 'Content-Length' in resp_headers and resp_headers['Content-Length'] > max_size: logger.warn("Requested URL is too large > %r bytes" % (self.max_size,)) raise SynapseError( 502, @@ -315,10 +373,9 @@ class SimpleHttpClient(object): # straight back in again try: - length = yield preserve_context_over_fn( - _readBodyToFile, - response, output_stream, max_size - ) + length = yield make_deferred_yieldable(_readBodyToFile( + response, output_stream, max_size, + )) except Exception as e: logger.exception("Failed to download body") raise SynapseError( @@ -327,7 +384,9 @@ class SimpleHttpClient(object): Codes.UNKNOWN, ) - defer.returnValue((length, headers, response.request.absoluteURI, response.code)) + defer.returnValue( + (length, resp_headers, response.request.absoluteURI, response.code), + ) # XXX: FIXME: This is horribly copy-pasted from matrixfederationclient. @@ -395,7 +454,7 @@ class CaptchaServerHttpClient(SimpleHttpClient): ) try: - body = yield preserve_context_over_fn(readBody, response) + body = yield make_deferred_yieldable(readBody(response)) defer.returnValue(body) except PartialDownloadError as e: # twisted dislikes google's response, no content length. @@ -446,7 +505,7 @@ class SpiderHttpClient(SimpleHttpClient): reactor, SpiderEndpointFactory(hs) ) - ), [('gzip', GzipDecoder)] + ), [(b'gzip', GzipDecoder)] ) # We could look like Chrome: # self.user_agent = ("Mozilla/5.0 (%s) (KHTML, like Gecko) diff --git a/synapse/http/endpoint.py b/synapse/http/endpoint.py index d8923c9ab..d65daa72b 100644 --- a/synapse/http/endpoint.py +++ b/synapse/http/endpoint.py @@ -12,30 +12,97 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. - -from twisted.internet.endpoints import HostnameEndpoint, wrapClientTLS -from twisted.internet import defer, reactor -from twisted.internet.error import ConnectError -from twisted.names import client, dns -from twisted.names.error import DNSNameError, DomainError - import collections import logging import random +import re import time +from twisted.internet import defer +from twisted.internet.endpoints import HostnameEndpoint, wrapClientTLS +from twisted.internet.error import ConnectError +from twisted.names import client, dns +from twisted.names.error import DNSNameError, DomainError logger = logging.getLogger(__name__) SERVER_CACHE = {} - +# our record of an individual server which can be tried to reach a destination. +# +# "host" is the hostname acquired from the SRV record. Except when there's +# no SRV record, in which case it is the original hostname. _Server = collections.namedtuple( "_Server", "priority weight host port expires" ) +def parse_server_name(server_name): + """Split a server name into host/port parts. + + Args: + server_name (str): server name to parse + + Returns: + Tuple[str, int|None]: host/port parts. + + Raises: + ValueError if the server name could not be parsed. + """ + try: + if server_name[-1] == ']': + # ipv6 literal, hopefully + return server_name, None + + domain_port = server_name.rsplit(":", 1) + domain = domain_port[0] + port = int(domain_port[1]) if domain_port[1:] else None + return domain, port + except Exception: + raise ValueError("Invalid server name '%s'" % server_name) + + +VALID_HOST_REGEX = re.compile( + "\\A[0-9a-zA-Z.-]+\\Z", +) + + +def parse_and_validate_server_name(server_name): + """Split a server name into host/port parts and do some basic validation. + + Args: + server_name (str): server name to parse + + Returns: + Tuple[str, int|None]: host/port parts. + + Raises: + ValueError if the server name could not be parsed. + """ + host, port = parse_server_name(server_name) + + # these tests don't need to be bulletproof as we'll find out soon enough + # if somebody is giving us invalid data. What we *do* need is to be sure + # that nobody is sneaking IP literals in that look like hostnames, etc. + + # look for ipv6 literals + if host[0] == '[': + if host[-1] != ']': + raise ValueError("Mismatched [...] in server name '%s'" % ( + server_name, + )) + return host, port + + # otherwise it should only be alphanumerics. + if not VALID_HOST_REGEX.match(host): + raise ValueError("Server name '%s' contains invalid characters" % ( + server_name, + )) + + return host, port + + def matrix_federation_endpoint(reactor, destination, ssl_context_factory=None, timeout=None): """Construct an endpoint for the given matrix destination. @@ -48,9 +115,7 @@ def matrix_federation_endpoint(reactor, destination, ssl_context_factory=None, timeout (int): connection timeout in seconds """ - domain_port = destination.split(":") - domain = domain_port[0] - port = int(domain_port[1]) if domain_port[1:] else None + domain, port = parse_server_name(destination) endpoint_kw_args = {} @@ -72,21 +137,22 @@ def matrix_federation_endpoint(reactor, destination, ssl_context_factory=None, reactor, "matrix", domain, protocol="tcp", default_port=default_port, endpoint=transport_endpoint, endpoint_kw_args=endpoint_kw_args - )) + ), reactor) else: return _WrappingEndpointFac(transport_endpoint( reactor, domain, port, **endpoint_kw_args - )) + ), reactor) class _WrappingEndpointFac(object): - def __init__(self, endpoint_fac): + def __init__(self, endpoint_fac, reactor): self.endpoint_fac = endpoint_fac + self.reactor = reactor @defer.inlineCallbacks def connect(self, protocolFactory): conn = yield self.endpoint_fac.connect(protocolFactory) - conn = _WrappedConnection(conn) + conn = _WrappedConnection(conn, self.reactor) defer.returnValue(conn) @@ -96,9 +162,10 @@ class _WrappedConnection(object): """ __slots__ = ["conn", "last_request"] - def __init__(self, conn): + def __init__(self, conn, reactor): object.__setattr__(self, "conn", conn) object.__setattr__(self, "last_request", time.time()) + self._reactor = reactor def __getattr__(self, name): return getattr(self.conn, name) @@ -113,10 +180,15 @@ class _WrappedConnection(object): if time.time() - self.last_request >= 2.5 * 60: self.abort() # Abort the underlying TLS connection. The abort() method calls - # loseConnection() on the underlying TLS connection which tries to + # loseConnection() on the TLS connection which tries to # shutdown the connection cleanly. We call abortConnection() - # since that will promptly close the underlying TCP connection. - self.transport.abortConnection() + # since that will promptly close the TLS connection. + # + # In Twisted >18.4; the TLS connection will be None if it has closed + # which will make abortConnection() throw. Check that the TLS connection + # is not None before trying to close it. + if self.transport.getHandle() is not None: + self.transport.abortConnection() def request(self, request): self.last_request = time.time() @@ -124,14 +196,14 @@ class _WrappedConnection(object): # Time this connection out if we haven't send a request in the last # N minutes # TODO: Cancel the previous callLater? - reactor.callLater(3 * 60, self._time_things_out_maybe) + self._reactor.callLater(3 * 60, self._time_things_out_maybe) d = self.conn.request(request) def update_request_time(res): self.last_request = time.time() # TODO: Cancel the previous callLater? - reactor.callLater(3 * 60, self._time_things_out_maybe) + self._reactor.callLater(3 * 60, self._time_things_out_maybe) return res d.addCallback(update_request_time) @@ -219,9 +291,10 @@ class SRVClientEndpoint(object): return self.default_server else: raise ConnectError( - "Not server available for %s" % self.service_name + "No server available for %s" % self.service_name ) + # look for all servers with the same priority min_priority = self.servers[0].priority weight_indexes = list( (index, server.weight + 1) @@ -231,11 +304,22 @@ class SRVClientEndpoint(object): total_weight = sum(weight for index, weight in weight_indexes) target_weight = random.randint(0, total_weight) - for index, weight in weight_indexes: target_weight -= weight if target_weight <= 0: server = self.servers[index] + # XXX: this looks totally dubious: + # + # (a) we never reuse a server until we have been through + # all of the servers at the same priority, so if the + # weights are A: 100, B:1, we always do ABABAB instead of + # AAAA...AAAB (approximately). + # + # (b) After using all the servers at the lowest priority, + # we move onto the next priority. We should only use the + # second priority if servers at the top priority are + # unreachable. + # del self.servers[index] self.used_servers.append(server) return server @@ -272,7 +356,7 @@ def resolve_service(service_name, dns_client=client, cache=SERVER_CACHE, clock=t if (len(answers) == 1 and answers[0].type == dns.SRV and answers[0].payload - and answers[0].payload.target == dns.Name('.')): + and answers[0].payload.target == dns.Name(b'.')): raise ConnectError("Service %s unavailable" % service_name) for answer in answers: @@ -280,26 +364,14 @@ def resolve_service(service_name, dns_client=client, cache=SERVER_CACHE, clock=t continue payload = answer.payload - host = str(payload.target) - srv_ttl = answer.ttl - try: - answers, _, _ = yield dns_client.lookupAddress(host) - except DNSNameError: - continue - - for answer in answers: - if answer.type == dns.A and answer.payload: - ip = answer.payload.dottedQuad() - host_ttl = min(srv_ttl, answer.ttl) - - servers.append(_Server( - host=ip, - port=int(payload.port), - priority=int(payload.priority), - weight=int(payload.weight), - expires=int(clock.time()) + host_ttl, - )) + servers.append(_Server( + host=str(payload.target), + port=int(payload.port), + priority=int(payload.priority), + weight=int(payload.weight), + expires=int(clock.time()) + answer.ttl, + )) servers.sort() cache[service_name] = list(servers) diff --git a/synapse/http/matrixfederationclient.py b/synapse/http/matrixfederationclient.py index 747a791f8..bf1aa2950 100644 --- a/synapse/http/matrixfederationclient.py +++ b/synapse/http/matrixfederationclient.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2014-2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -12,48 +13,46 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. -import synapse.util.retryutils -from twisted.internet import defer, reactor, protocol -from twisted.internet.error import DNSLookupError -from twisted.web.client import readBody, HTTPConnectionPool, Agent -from twisted.web.http_headers import Headers -from twisted.web._newclient import ResponseDone - -from synapse.http.endpoint import matrix_federation_endpoint -from synapse.util.async import sleep -from synapse.util import logcontext -import synapse.metrics - -from canonicaljson import encode_canonical_json - -from synapse.api.errors import ( - SynapseError, Codes, HttpResponseException, -) - -from signedjson.sign import sign_json - import cgi -import simplejson as json import logging import random import sys import urllib -import urlparse +from six import string_types +from six.moves.urllib import parse as urlparse + +from canonicaljson import encode_canonical_json, json +from prometheus_client import Counter +from signedjson.sign import sign_json + +from twisted.internet import defer, protocol, reactor +from twisted.internet.error import DNSLookupError +from twisted.web._newclient import ResponseDone +from twisted.web.client import Agent, HTTPConnectionPool, readBody +from twisted.web.http_headers import Headers + +import synapse.metrics +import synapse.util.retryutils +from synapse.api.errors import ( + Codes, + FederationDeniedError, + HttpResponseException, + SynapseError, +) +from synapse.http import cancelled_to_request_timed_out_error +from synapse.http.endpoint import matrix_federation_endpoint +from synapse.util import logcontext +from synapse.util.async import add_timeout_to_deferred +from synapse.util.logcontext import make_deferred_yieldable logger = logging.getLogger(__name__) outbound_logger = logging.getLogger("synapse.http.outbound") -metrics = synapse.metrics.get_metrics_for(__name__) - -outgoing_requests_counter = metrics.register_counter( - "requests", - labels=["method"], -) -incoming_responses_counter = metrics.register_counter( - "responses", - labels=["method", "code"], -) +outgoing_requests_counter = Counter("synapse_http_matrixfederationclient_requests", + "", ["method"]) +incoming_responses_counter = Counter("synapse_http_matrixfederationclient_responses", + "", ["method", "code"]) MAX_LONG_RETRIES = 10 @@ -123,11 +122,22 @@ class MatrixFederationHttpClient(object): Fails with ``HTTPRequestException``: if we get an HTTP response code >= 300. + Fails with ``NotRetryingDestination`` if we are not yet ready to retry this server. + + Fails with ``FederationDeniedError`` if this destination + is not on our federation whitelist + (May also fail with plenty of other Exceptions for things like DNS failures, connection failures, SSL failures.) """ + if ( + self.hs.config.federation_domain_whitelist and + destination not in self.hs.config.federation_domain_whitelist + ): + raise FederationDeniedError(destination) + limiter = yield synapse.util.retryutils.get_retry_limiter( destination, self.clock, @@ -173,21 +183,21 @@ class MatrixFederationHttpClient(object): producer = body_callback(method, http_url_bytes, headers_dict) try: - def send_request(): - request_deferred = self.agent.request( - method, - url_bytes, - Headers(headers_dict), - producer - ) - - return self.clock.time_bound_deferred( - request_deferred, - time_out=timeout / 1000. if timeout else 60, - ) - - with logcontext.PreserveLoggingContext(): - response = yield send_request() + request_deferred = self.agent.request( + method, + url_bytes, + Headers(headers_dict), + producer + ) + add_timeout_to_deferred( + request_deferred, + timeout / 1000. if timeout else 60, + self.hs.get_reactor(), + cancelled_to_request_timed_out_error, + ) + response = yield make_deferred_yieldable( + request_deferred, + ) log_result = "%d %s" % (response.code, response.phrase,) break @@ -204,18 +214,15 @@ class MatrixFederationHttpClient(object): raise logger.warn( - "{%s} Sending request failed to %s: %s %s: %s - %s", + "{%s} Sending request failed to %s: %s %s: %s", txn_id, destination, method, url_bytes, - type(e).__name__, _flatten_response_never_received(e), ) - log_result = "%s - %s" % ( - type(e).__name__, _flatten_response_never_received(e), - ) + log_result = _flatten_response_never_received(e) if retries_left and not timeout: if long_retries: @@ -227,7 +234,7 @@ class MatrixFederationHttpClient(object): delay = min(delay, 2) delay *= random.uniform(0.8, 1.4) - yield sleep(delay) + yield self.clock.sleep(delay) retries_left -= 1 else: raise @@ -253,14 +260,35 @@ class MatrixFederationHttpClient(object): defer.returnValue(response) def sign_request(self, destination, method, url_bytes, headers_dict, - content=None): + content=None, destination_is=None): + """ + Signs a request by adding an Authorization header to headers_dict + Args: + destination (bytes|None): The desination home server of the request. + May be None if the destination is an identity server, in which case + destination_is must be non-None. + method (bytes): The HTTP method of the request + url_bytes (bytes): The URI path of the request + headers_dict (dict): Dictionary of request headers to append to + content (bytes): The body of the request + destination_is (bytes): As 'destination', but if the destination is an + identity server + + Returns: + None + """ request = { "method": method, "uri": url_bytes, "origin": self.server_name, - "destination": destination, } + if destination is not None: + request["destination"] = destination + + if destination_is is not None: + request["destination_is"] = destination_is + if content is not None: request["content"] = content @@ -278,7 +306,8 @@ class MatrixFederationHttpClient(object): headers_dict[b"Authorization"] = auth_headers @defer.inlineCallbacks - def put_json(self, destination, path, data={}, json_data_callback=None, + def put_json(self, destination, path, args={}, data={}, + json_data_callback=None, long_retries=False, timeout=None, ignore_backoff=False, backoff_on_404=False): @@ -288,6 +317,7 @@ class MatrixFederationHttpClient(object): destination (str): The remote server to send the HTTP request to. path (str): The HTTP path. + args (dict): query params data (dict): A dict containing the data that will be used as the request body. This will be encoded as JSON. json_data_callback (callable): A callable returning the dict to @@ -311,6 +341,9 @@ class MatrixFederationHttpClient(object): Fails with ``NotRetryingDestination`` if we are not yet ready to retry this server. + + Fails with ``FederationDeniedError`` if this destination + is not on our federation whitelist """ if not json_data_callback: @@ -331,6 +364,7 @@ class MatrixFederationHttpClient(object): path, body_callback=body_callback, headers_dict={"Content-Type": ["application/json"]}, + query_bytes=encode_query_args(args), long_retries=long_retries, timeout=timeout, ignore_backoff=ignore_backoff, @@ -347,7 +381,7 @@ class MatrixFederationHttpClient(object): @defer.inlineCallbacks def post_json(self, destination, path, data={}, long_retries=False, - timeout=None, ignore_backoff=False): + timeout=None, ignore_backoff=False, args={}): """ Sends the specifed json data using POST Args: @@ -362,6 +396,7 @@ class MatrixFederationHttpClient(object): giving up. None indicates no timeout. ignore_backoff (bool): true to ignore the historical backoff data and try the request anyway. + args (dict): query params Returns: Deferred: Succeeds when we get a 2xx HTTP response. The result will be the decoded JSON body. @@ -371,6 +406,9 @@ class MatrixFederationHttpClient(object): Fails with ``NotRetryingDestination`` if we are not yet ready to retry this server. + + Fails with ``FederationDeniedError`` if this destination + is not on our federation whitelist """ def body_callback(method, url_bytes, headers_dict): @@ -383,6 +421,7 @@ class MatrixFederationHttpClient(object): destination, "POST", path, + query_bytes=encode_query_args(args), body_callback=body_callback, headers_dict={"Content-Type": ["application/json"]}, long_retries=long_retries, @@ -424,16 +463,12 @@ class MatrixFederationHttpClient(object): Fails with ``NotRetryingDestination`` if we are not yet ready to retry this server. + + Fails with ``FederationDeniedError`` if this destination + is not on our federation whitelist """ logger.debug("get_json args: %s", args) - encoded_args = {} - for k, vs in args.items(): - if isinstance(vs, basestring): - vs = [vs] - encoded_args[k] = [v.encode("UTF-8") for v in vs] - - query_bytes = urllib.urlencode(encoded_args, True) logger.debug("Query bytes: %s Retry DNS: %s", args, retry_on_dns_fail) def body_callback(method, url_bytes, headers_dict): @@ -444,7 +479,7 @@ class MatrixFederationHttpClient(object): destination, "GET", path, - query_bytes=query_bytes, + query_bytes=encode_query_args(args), body_callback=body_callback, retry_on_dns_fail=retry_on_dns_fail, timeout=timeout, @@ -460,6 +495,55 @@ class MatrixFederationHttpClient(object): defer.returnValue(json.loads(body)) + @defer.inlineCallbacks + def delete_json(self, destination, path, long_retries=False, + timeout=None, ignore_backoff=False, args={}): + """Send a DELETE request to the remote expecting some json response + + Args: + destination (str): The remote server to send the HTTP request + to. + path (str): The HTTP path. + long_retries (bool): A boolean that indicates whether we should + retry for a short or long time. + timeout(int): How long to try (in ms) the destination for before + giving up. None indicates no timeout. + ignore_backoff (bool): true to ignore the historical backoff data and + try the request anyway. + Returns: + Deferred: Succeeds when we get a 2xx HTTP response. The result + will be the decoded JSON body. + + Fails with ``HTTPRequestException`` if we get an HTTP response + code >= 300. + + Fails with ``NotRetryingDestination`` if we are not yet ready + to retry this server. + + Fails with ``FederationDeniedError`` if this destination + is not on our federation whitelist + """ + + response = yield self._request( + destination, + "DELETE", + path, + query_bytes=encode_query_args(args), + headers_dict={"Content-Type": ["application/json"]}, + long_retries=long_retries, + timeout=timeout, + ignore_backoff=ignore_backoff, + ) + + if 200 <= response.code < 300: + # We need to update the transactions table to say it was sent? + check_content_type_is_json(response.headers) + + with logcontext.PreserveLoggingContext(): + body = yield readBody(response) + + defer.returnValue(json.loads(body)) + @defer.inlineCallbacks def get_file(self, destination, path, output_stream, args={}, retry_on_dns_fail=True, max_size=None, @@ -481,11 +565,14 @@ class MatrixFederationHttpClient(object): Fails with ``NotRetryingDestination`` if we are not yet ready to retry this server. + + Fails with ``FederationDeniedError`` if this destination + is not on our federation whitelist """ encoded_args = {} for k, vs in args.items(): - if isinstance(vs, basestring): + if isinstance(vs, string_types): vs = [vs] encoded_args[k] = [v.encode("UTF-8") for v in vs] @@ -513,7 +600,7 @@ class MatrixFederationHttpClient(object): length = yield _readBodyToFile( response, output_stream, max_size ) - except: + except Exception: logger.exception("Failed to download body") raise @@ -578,12 +665,14 @@ class _JsonProducer(object): def _flatten_response_never_received(e): if hasattr(e, "reasons"): - return ", ".join( + reasons = ", ".join( _flatten_response_never_received(f.value) for f in e.reasons ) + + return "%s:[%s]" % (type(e).__name__, reasons) else: - return "%s: %s" % (type(e).__name__, e.message,) + return repr(e) def check_content_type_is_json(headers): @@ -598,7 +687,7 @@ def check_content_type_is_json(headers): RuntimeError if the """ - c_type = headers.getRawHeaders("Content-Type") + c_type = headers.getRawHeaders(b"Content-Type") if c_type is None: raise RuntimeError( "No Content-Type header" @@ -610,3 +699,15 @@ def check_content_type_is_json(headers): raise RuntimeError( "Content-Type not application/json: was '%s'" % c_type ) + + +def encode_query_args(args): + encoded_args = {} + for k, vs in args.items(): + if isinstance(vs, string_types): + vs = [vs] + encoded_args[k] = [v.encode("UTF-8") for v in vs] + + query_bytes = urllib.urlencode(encoded_args, True) + + return query_bytes diff --git a/synapse/http/request_metrics.py b/synapse/http/request_metrics.py new file mode 100644 index 000000000..588e28057 --- /dev/null +++ b/synapse/http/request_metrics.py @@ -0,0 +1,231 @@ +# -*- coding: utf-8 -*- +# Copyright 2014-2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging + +from prometheus_client.core import Counter, Histogram + +from synapse.metrics import LaterGauge +from synapse.util.logcontext import LoggingContext + +logger = logging.getLogger(__name__) + + +# total number of responses served, split by method/servlet/tag +response_count = Counter( + "synapse_http_server_response_count", "", ["method", "servlet", "tag"] +) + +requests_counter = Counter( + "synapse_http_server_requests_received", "", ["method", "servlet"] +) + +outgoing_responses_counter = Counter( + "synapse_http_server_responses", "", ["method", "code"] +) + +response_timer = Histogram( + "synapse_http_server_response_time_seconds", "sec", + ["method", "servlet", "tag", "code"], +) + +response_ru_utime = Counter( + "synapse_http_server_response_ru_utime_seconds", "sec", ["method", "servlet", "tag"] +) + +response_ru_stime = Counter( + "synapse_http_server_response_ru_stime_seconds", "sec", ["method", "servlet", "tag"] +) + +response_db_txn_count = Counter( + "synapse_http_server_response_db_txn_count", "", ["method", "servlet", "tag"] +) + +# seconds spent waiting for db txns, excluding scheduling time, when processing +# this request +response_db_txn_duration = Counter( + "synapse_http_server_response_db_txn_duration_seconds", + "", + ["method", "servlet", "tag"], +) + +# seconds spent waiting for a db connection, when processing this request +response_db_sched_duration = Counter( + "synapse_http_server_response_db_sched_duration_seconds", + "", + ["method", "servlet", "tag"], +) + +# size in bytes of the response written +response_size = Counter( + "synapse_http_server_response_size", "", ["method", "servlet", "tag"] +) + +# In flight metrics are incremented while the requests are in flight, rather +# than when the response was written. + +in_flight_requests_ru_utime = Counter( + "synapse_http_server_in_flight_requests_ru_utime_seconds", + "", + ["method", "servlet"], +) + +in_flight_requests_ru_stime = Counter( + "synapse_http_server_in_flight_requests_ru_stime_seconds", + "", + ["method", "servlet"], +) + +in_flight_requests_db_txn_count = Counter( + "synapse_http_server_in_flight_requests_db_txn_count", "", ["method", "servlet"] +) + +# seconds spent waiting for db txns, excluding scheduling time, when processing +# this request +in_flight_requests_db_txn_duration = Counter( + "synapse_http_server_in_flight_requests_db_txn_duration_seconds", + "", + ["method", "servlet"], +) + +# seconds spent waiting for a db connection, when processing this request +in_flight_requests_db_sched_duration = Counter( + "synapse_http_server_in_flight_requests_db_sched_duration_seconds", + "", + ["method", "servlet"], +) + +# The set of all in flight requests, set[RequestMetrics] +_in_flight_requests = set() + + +def _get_in_flight_counts(): + """Returns a count of all in flight requests by (method, server_name) + + Returns: + dict[tuple[str, str], int] + """ + # Cast to a list to prevent it changing while the Prometheus + # thread is collecting metrics + reqs = list(_in_flight_requests) + + for rm in reqs: + rm.update_metrics() + + # Map from (method, name) -> int, the number of in flight requests of that + # type + counts = {} + for rm in reqs: + key = (rm.method, rm.name,) + counts[key] = counts.get(key, 0) + 1 + + return counts + + +LaterGauge( + "synapse_http_server_in_flight_requests_count", + "", + ["method", "servlet"], + _get_in_flight_counts, +) + + +class RequestMetrics(object): + def start(self, time_sec, name, method): + self.start = time_sec + self.start_context = LoggingContext.current_context() + self.name = name + self.method = method + + # _request_stats records resource usage that we have already added + # to the "in flight" metrics. + self._request_stats = self.start_context.get_resource_usage() + + _in_flight_requests.add(self) + + def stop(self, time_sec, request): + _in_flight_requests.discard(self) + + context = LoggingContext.current_context() + + tag = "" + if context: + tag = context.tag + + if context != self.start_context: + logger.warn( + "Context have unexpectedly changed %r, %r", + context, self.start_context + ) + return + + response_code = str(request.code) + + outgoing_responses_counter.labels(request.method, response_code).inc() + + response_count.labels(request.method, self.name, tag).inc() + + response_timer.labels(request.method, self.name, tag, response_code).observe( + time_sec - self.start + ) + + resource_usage = context.get_resource_usage() + + response_ru_utime.labels(request.method, self.name, tag).inc( + resource_usage.ru_utime, + ) + response_ru_stime.labels(request.method, self.name, tag).inc( + resource_usage.ru_stime, + ) + response_db_txn_count.labels(request.method, self.name, tag).inc( + resource_usage.db_txn_count + ) + response_db_txn_duration.labels(request.method, self.name, tag).inc( + resource_usage.db_txn_duration_sec + ) + response_db_sched_duration.labels(request.method, self.name, tag).inc( + resource_usage.db_sched_duration_sec + ) + + response_size.labels(request.method, self.name, tag).inc(request.sentLength) + + # We always call this at the end to ensure that we update the metrics + # regardless of whether a call to /metrics while the request was in + # flight. + self.update_metrics() + + def update_metrics(self): + """Updates the in flight metrics with values from this request. + """ + new_stats = self.start_context.get_resource_usage() + + diff = new_stats - self._request_stats + self._request_stats = new_stats + + in_flight_requests_ru_utime.labels(self.method, self.name).inc(diff.ru_utime) + in_flight_requests_ru_stime.labels(self.method, self.name).inc(diff.ru_stime) + + in_flight_requests_db_txn_count.labels(self.method, self.name).inc( + diff.db_txn_count + ) + + in_flight_requests_db_txn_duration.labels(self.method, self.name).inc( + diff.db_txn_duration_sec + ) + + in_flight_requests_db_sched_duration.labels(self.method, self.name).inc( + diff.db_sched_duration_sec + ) diff --git a/synapse/http/server.py b/synapse/http/server.py index 7ef3d526b..c70fdbdfd 100644 --- a/synapse/http/server.py +++ b/synapse/http/server.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2014-2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -12,148 +13,205 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. - - -from synapse.api.errors import ( - cs_exception, SynapseError, CodeMessageException, UnrecognizedRequestError, Codes -) -from synapse.util.logcontext import LoggingContext, PreserveLoggingContext -from synapse.util.caches import intern_dict -from synapse.util.metrics import Measure -import synapse.metrics -import synapse.events - -from canonicaljson import ( - encode_canonical_json, encode_pretty_printed_json -) - -from twisted.internet import defer -from twisted.web import server, resource -from twisted.web.server import NOT_DONE_YET -from twisted.web.util import redirectTo - +import cgi import collections import logging import urllib -import ujson + +from six.moves import http_client + +from canonicaljson import encode_canonical_json, encode_pretty_printed_json, json + +from twisted.internet import defer +from twisted.python import failure +from twisted.web import resource, server +from twisted.web.server import NOT_DONE_YET +from twisted.web.util import redirectTo + +import synapse.events +import synapse.metrics +from synapse.api.errors import ( + CodeMessageException, + Codes, + SynapseError, + UnrecognizedRequestError, + cs_exception, +) +from synapse.http.request_metrics import requests_counter +from synapse.util.caches import intern_dict +from synapse.util.logcontext import LoggingContext, PreserveLoggingContext +from synapse.util.metrics import Measure logger = logging.getLogger(__name__) -metrics = synapse.metrics.get_metrics_for(__name__) - -incoming_requests_counter = metrics.register_counter( - "requests", - labels=["method", "servlet", "tag"], -) -outgoing_responses_counter = metrics.register_counter( - "responses", - labels=["method", "code"], -) - -response_timer = metrics.register_distribution( - "response_time", - labels=["method", "servlet", "tag"] -) - -response_ru_utime = metrics.register_distribution( - "response_ru_utime", labels=["method", "servlet", "tag"] -) - -response_ru_stime = metrics.register_distribution( - "response_ru_stime", labels=["method", "servlet", "tag"] -) - -response_db_txn_count = metrics.register_distribution( - "response_db_txn_count", labels=["method", "servlet", "tag"] -) - -response_db_txn_duration = metrics.register_distribution( - "response_db_txn_duration", labels=["method", "servlet", "tag"] -) +HTML_ERROR_TEMPLATE = """ + + + + Error {code} + + +

{msg}

+ + +""" -_next_request_id = 0 +def wrap_json_request_handler(h): + """Wraps a request handler method with exception handling. + Also adds logging as per wrap_request_handler_with_logging. -def request_handler(include_metrics=False): - """Decorator for ``wrap_request_handler``""" - return lambda request_handler: wrap_request_handler(request_handler, include_metrics) + The handler method must have a signature of "handle_foo(self, request)", + where "self" must have a "clock" attribute (and "request" must be a + SynapseRequest). - -def wrap_request_handler(request_handler, include_metrics=False): - """Wraps a method that acts as a request handler with the necessary logging - and exception handling. - - The method must have a signature of "handle_foo(self, request)". The - argument "self" must have "version_string" and "clock" attributes. The - argument "request" must be a twisted HTTP request. - - The method must return a deferred. If the deferred succeeds we assume that + The handler must return a deferred. If the deferred succeeds we assume that a response has been sent. If the deferred fails with a SynapseError we use it to send a JSON response with the appropriate HTTP reponse code. If the deferred fails with any other type of error we send a 500 reponse. - - We insert a unique request-id into the logging context for this request and - log the response and duration for this request. """ @defer.inlineCallbacks def wrapped_request_handler(self, request): - global _next_request_id - request_id = "%s-%s" % (request.method, _next_request_id) - _next_request_id += 1 + try: + yield h(self, request) + except CodeMessageException as e: + code = e.code + if isinstance(e, SynapseError): + logger.info( + "%s SynapseError: %s - %s", request, code, e.msg + ) + else: + logger.exception(e) + respond_with_json( + request, code, cs_exception(e), send_cors=True, + pretty_print=_request_user_agent_is_curl(request), + ) + except Exception: + # failure.Failure() fishes the original Failure out + # of our stack, and thus gives us a sensible stack + # trace. + f = failure.Failure() + logger.error( + "Failed handle request via %r: %r: %s", + h, + request, + f.getTraceback().rstrip(), + ) + respond_with_json( + request, + 500, + { + "error": "Internal server error", + "errcode": Codes.UNKNOWN, + }, + send_cors=True, + pretty_print=_request_user_agent_is_curl(request), + ) + + return wrap_request_handler_with_logging(wrapped_request_handler) + + +def wrap_html_request_handler(h): + """Wraps a request handler method with exception handling. + + Also adds logging as per wrap_request_handler_with_logging. + + The handler method must have a signature of "handle_foo(self, request)", + where "self" must have a "clock" attribute (and "request" must be a + SynapseRequest). + """ + def wrapped_request_handler(self, request): + d = defer.maybeDeferred(h, self, request) + d.addErrback(_return_html_error, request) + return d + + return wrap_request_handler_with_logging(wrapped_request_handler) + + +def _return_html_error(f, request): + """Sends an HTML error page corresponding to the given failure + + Args: + f (twisted.python.failure.Failure): + request (twisted.web.iweb.IRequest): + """ + if f.check(CodeMessageException): + cme = f.value + code = cme.code + msg = cme.msg + + if isinstance(cme, SynapseError): + logger.info( + "%s SynapseError: %s - %s", request, code, msg + ) + else: + logger.error( + "Failed handle request %r: %s", + request, + f.getTraceback().rstrip(), + ) + else: + code = http_client.INTERNAL_SERVER_ERROR + msg = "Internal server error" + + logger.error( + "Failed handle request %r: %s", + request, + f.getTraceback().rstrip(), + ) + + body = HTML_ERROR_TEMPLATE.format( + code=code, msg=cgi.escape(msg), + ).encode("utf-8") + request.setResponseCode(code) + request.setHeader(b"Content-Type", b"text/html; charset=utf-8") + request.setHeader(b"Content-Length", b"%i" % (len(body),)) + request.write(body) + finish_request(request) + + +def wrap_request_handler_with_logging(h): + """Wraps a request handler to provide logging and metrics + + The handler method must have a signature of "handle_foo(self, request)", + where "self" must have a "clock" attribute (and "request" must be a + SynapseRequest). + + As well as calling `request.processing` (which will log the response and + duration for this request), the wrapped request handler will insert the + request id into the logging context. + """ + @defer.inlineCallbacks + def wrapped_request_handler(self, request): + """ + Args: + self: + request (synapse.http.site.SynapseRequest): + """ + + request_id = request.get_request_id() with LoggingContext(request_id) as request_context: + request_context.request = request_id with Measure(self.clock, "wrapped_request_handler"): - request_metrics = RequestMetrics() - request_metrics.start(self.clock, name=self.__class__.__name__) + # we start the request metrics timer here with an initial stab + # at the servlet name. For most requests that name will be + # JsonResource (or a subclass), and JsonResource._async_render + # will update it once it picks a servlet. + servlet_name = self.__class__.__name__ + with request.processing(servlet_name): + with PreserveLoggingContext(request_context): + d = defer.maybeDeferred(h, self, request) - request_context.request = request_id - with request.processing(): - try: - with PreserveLoggingContext(request_context): - if include_metrics: - yield request_handler(self, request, request_metrics) - else: - yield request_handler(self, request) - except CodeMessageException as e: - code = e.code - if isinstance(e, SynapseError): - logger.info( - "%s SynapseError: %s - %s", request, code, e.msg - ) - else: - logger.exception(e) - outgoing_responses_counter.inc(request.method, str(code)) - respond_with_json( - request, code, cs_exception(e), send_cors=True, - pretty_print=_request_user_agent_is_curl(request), - version_string=self.version_string, - ) - except: - logger.exception( - "Failed handle request %s.%s on %r: %r", - request_handler.__module__, - request_handler.__name__, - self, - request - ) - respond_with_json( - request, - 500, - { - "error": "Internal server error", - "errcode": Codes.UNKNOWN, - }, - send_cors=True - ) - finally: - try: - request_metrics.stop( - self.clock, request - ) - except Exception as e: - logger.warn("Failed to stop metrics: %r", e) + # record the arrival of the request *after* + # dispatching to the handler, so that the handler + # can update the servlet name in the request + # metrics + requests_counter.labels(request.method, + request.request_metrics.name).inc() + yield d return wrapped_request_handler @@ -183,7 +241,7 @@ class JsonResource(HttpServer, resource.Resource): """ This implements the HttpServer interface and provides JSON support for Resources. - Register callbacks via register_path() + Register callbacks via register_paths() Callbacks can return a tuple of status code and a dict in which case the the dict will automatically be sent to the client as a JSON object. @@ -203,7 +261,6 @@ class JsonResource(HttpServer, resource.Resource): self.canonical_json = canonical_json self.clock = hs.get_clock() self.path_regexs = {} - self.version_string = hs.version_string self.hs = hs def register_paths(self, method, path_patterns, callback): @@ -219,122 +276,103 @@ class JsonResource(HttpServer, resource.Resource): self._async_render(request) return server.NOT_DONE_YET - # Disable metric reporting because _async_render does its own metrics. - # It does its own metric reporting because _async_render dispatches to - # a callback and it's the class name of that callback we want to report - # against rather than the JsonResource itself. - @request_handler(include_metrics=True) + @wrap_json_request_handler @defer.inlineCallbacks - def _async_render(self, request, request_metrics): + def _async_render(self, request): """ This gets called from render() every time someone sends us a request. This checks if anyone has registered a callback for that method and path. """ - if request.method == "OPTIONS": - self._send_response(request, 200, {}) - return + callback, group_dict = self._get_handler_for_request(request) + + servlet_instance = getattr(callback, "__self__", None) + if servlet_instance is not None: + servlet_classname = servlet_instance.__class__.__name__ + else: + servlet_classname = "%r" % callback + request.request_metrics.name = servlet_classname + + # Now trigger the callback. If it returns a response, we send it + # here. If it throws an exception, that is handled by the wrapper + # installed by @request_handler. + + kwargs = intern_dict({ + name: urllib.unquote(value).decode("UTF-8") if value else value + for name, value in group_dict.items() + }) + + callback_return = yield callback(request, **kwargs) + if callback_return is not None: + code, response = callback_return + self._send_response(request, code, response) + + def _get_handler_for_request(self, request): + """Finds a callback method to handle the given request + + Args: + request (twisted.web.http.Request): + + Returns: + Tuple[Callable, dict[str, str]]: callback method, and the dict + mapping keys to path components as specified in the handler's + path match regexp. + + The callback will normally be a method registered via + register_paths, so will return (possibly via Deferred) either + None, or a tuple of (http code, response body). + """ + if request.method == b"OPTIONS": + return _options_handler, {} # Loop through all the registered callbacks to check if the method # and path regex match for path_entry in self.path_regexs.get(request.method, []): m = path_entry.pattern.match(request.path) - if not m: - continue - - # We found a match! Trigger callback and then return the - # returned response. We pass both the request and any - # matched groups from the regex to the callback. - - callback = path_entry.callback - - kwargs = intern_dict({ - name: urllib.unquote(value).decode("UTF-8") if value else value - for name, value in m.groupdict().items() - }) - - callback_return = yield callback(request, **kwargs) - if callback_return is not None: - code, response = callback_return - self._send_response(request, code, response) - - servlet_instance = getattr(callback, "__self__", None) - if servlet_instance is not None: - servlet_classname = servlet_instance.__class__.__name__ - else: - servlet_classname = "%r" % callback - - request_metrics.name = servlet_classname - - return + if m: + # We found a match! + return path_entry.callback, m.groupdict() # Huh. No one wanted to handle that? Fiiiiiine. Send 400. - raise UnrecognizedRequestError() + return _unrecognised_request_handler, {} def _send_response(self, request, code, response_json_object, response_code_message=None): - # could alternatively use request.notifyFinish() and flip a flag when - # the Deferred fires, but since the flag is RIGHT THERE it seems like - # a waste. - if request._disconnected: - logger.warn( - "Not sending response to request %s, already disconnected.", - request) - return - - outgoing_responses_counter.inc(request.method, str(code)) - # TODO: Only enable CORS for the requests that need it. respond_with_json( request, code, response_json_object, send_cors=True, response_code_message=response_code_message, pretty_print=_request_user_agent_is_curl(request), - version_string=self.version_string, canonical_json=self.canonical_json, ) -class RequestMetrics(object): - def start(self, clock, name): - self.start = clock.time_msec() - self.start_context = LoggingContext.current_context() - self.name = name +def _options_handler(request): + """Request handler for OPTIONS requests - def stop(self, clock, request): - context = LoggingContext.current_context() + This is a request handler suitable for return from + _get_handler_for_request. It returns a 200 and an empty body. - tag = "" - if context: - tag = context.tag + Args: + request (twisted.web.http.Request): - if context != self.start_context: - logger.warn( - "Context have unexpectedly changed %r, %r", - context, self.start_context - ) - return + Returns: + Tuple[int, dict]: http code, response body. + """ + return 200, {} - incoming_requests_counter.inc(request.method, self.name, tag) - response_timer.inc_by( - clock.time_msec() - self.start, request.method, - self.name, tag - ) +def _unrecognised_request_handler(request): + """Request handler for unrecognised requests - ru_utime, ru_stime = context.get_resource_usage() + This is a request handler suitable for return from + _get_handler_for_request. It actually just raises an + UnrecognizedRequestError. - response_ru_utime.inc_by( - ru_utime, request.method, self.name, tag - ) - response_ru_stime.inc_by( - ru_stime, request.method, self.name, tag - ) - response_db_txn_count.inc_by( - context.db_txn_count, request.method, self.name, tag - ) - response_db_txn_duration.inc_by( - context.db_txn_duration, request.method, self.name, tag - ) + Args: + request (twisted.web.http.Request): + """ + raise UnrecognizedRequestError() class RootRedirect(resource.Resource): @@ -355,26 +393,33 @@ class RootRedirect(resource.Resource): def respond_with_json(request, code, json_object, send_cors=False, response_code_message=None, pretty_print=False, - version_string="", canonical_json=True): + canonical_json=True): + # could alternatively use request.notifyFinish() and flip a flag when + # the Deferred fires, but since the flag is RIGHT THERE it seems like + # a waste. + if request._disconnected: + logger.warn( + "Not sending response to request %s, already disconnected.", + request) + return + if pretty_print: json_bytes = encode_pretty_printed_json(json_object) + "\n" else: if canonical_json or synapse.events.USE_FROZEN_DICTS: json_bytes = encode_canonical_json(json_object) else: - # ujson doesn't like frozen_dicts. - json_bytes = ujson.dumps(json_object, ensure_ascii=False) + json_bytes = json.dumps(json_object) return respond_with_json_bytes( request, code, json_bytes, send_cors=send_cors, response_code_message=response_code_message, - version_string=version_string ) def respond_with_json_bytes(request, code, json_bytes, send_cors=False, - version_string="", response_code_message=None): + response_code_message=None): """Sends encoded JSON in response to the given request. Args: @@ -388,8 +433,8 @@ def respond_with_json_bytes(request, code, json_bytes, send_cors=False, request.setResponseCode(code, message=response_code_message) request.setHeader(b"Content-Type", b"application/json") - request.setHeader(b"Server", version_string) request.setHeader(b"Content-Length", b"%d" % (len(json_bytes),)) + request.setHeader(b"Cache-Control", b"no-cache, no-store, must-revalidate") if send_cors: set_cors_headers(request) @@ -437,9 +482,9 @@ def finish_request(request): def _request_user_agent_is_curl(request): user_agents = request.requestHeaders.getRawHeaders( - "User-Agent", default=[] + b"User-Agent", default=[] ) for user_agent in user_agents: - if "curl" in user_agent: + if b"curl" in user_agent: return True return False diff --git a/synapse/http/servlet.py b/synapse/http/servlet.py index 9a4c36ad5..882816dc8 100644 --- a/synapse/http/servlet.py +++ b/synapse/http/servlet.py @@ -15,10 +15,11 @@ """ This module contains base REST classes for constructing REST servlets. """ -from synapse.api.errors import SynapseError, Codes - import logging -import simplejson + +from canonicaljson import json + +from synapse.api.errors import Codes, SynapseError logger = logging.getLogger(__name__) @@ -48,7 +49,7 @@ def parse_integer_from_args(args, name, default=None, required=False): if name in args: try: return int(args[name][0]) - except: + except Exception: message = "Query parameter %r must be an integer" % (name,) raise SynapseError(400, message) else: @@ -88,7 +89,7 @@ def parse_boolean_from_args(args, name, default=None, required=False): "true": True, "false": False, }[args[name][0]] - except: + except Exception: message = ( "Boolean query parameter %r must be one of" " ['true', 'false']" @@ -148,11 +149,13 @@ def parse_string_from_args(args, name, default=None, required=False, return default -def parse_json_value_from_request(request): +def parse_json_value_from_request(request, allow_empty_body=False): """Parse a JSON value from the body of a twisted HTTP request. Args: request: the twisted HTTP request. + allow_empty_body (bool): if True, an empty body will be accepted and + turned into None Returns: The JSON value. @@ -162,28 +165,39 @@ def parse_json_value_from_request(request): """ try: content_bytes = request.content.read() - except: + except Exception: raise SynapseError(400, "Error reading JSON content.") + if not content_bytes and allow_empty_body: + return None + try: - content = simplejson.loads(content_bytes) - except simplejson.JSONDecodeError: + content = json.loads(content_bytes) + except Exception as e: + logger.warn("Unable to parse JSON: %s", e) raise SynapseError(400, "Content not JSON.", errcode=Codes.NOT_JSON) return content -def parse_json_object_from_request(request): +def parse_json_object_from_request(request, allow_empty_body=False): """Parse a JSON object from the body of a twisted HTTP request. Args: request: the twisted HTTP request. + allow_empty_body (bool): if True, an empty body will be accepted and + turned into an empty dict. Raises: SynapseError if the request body couldn't be decoded as JSON or if it wasn't a JSON object. """ - content = parse_json_value_from_request(request) + content = parse_json_value_from_request( + request, allow_empty_body=allow_empty_body, + ) + + if allow_empty_body and content is None: + return {} if type(content) != dict: message = "Content must be a JSON object." @@ -192,7 +206,7 @@ def parse_json_object_from_request(request): return content -def assert_params_in_request(body, required): +def assert_params_in_dict(body, required): absent = [] for k in required: if k not in body: diff --git a/synapse/http/site.py b/synapse/http/site.py index 4b09d7ee6..5fd30a4c2 100644 --- a/synapse/http/site.py +++ b/synapse/http/site.py @@ -12,27 +12,50 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.util.logcontext import LoggingContext -from twisted.web.server import Site, Request - import contextlib import logging -import re import time -ACCESS_TOKEN_RE = re.compile(r'(\?.*access(_|%5[Ff])token=)[^&]*(.*)$') +from twisted.web.server import Request, Site + +from synapse.http import redact_uri +from synapse.http.request_metrics import RequestMetrics +from synapse.util.logcontext import ContextResourceUsage, LoggingContext + +logger = logging.getLogger(__name__) + +_next_request_seq = 0 class SynapseRequest(Request): - def __init__(self, site, *args, **kw): - Request.__init__(self, *args, **kw) + """Class which encapsulates an HTTP request to synapse. + + All of the requests processed in synapse are of this type. + + It extends twisted's twisted.web.server.Request, and adds: + * Unique request ID + * Redaction of access_token query-params in __repr__ + * Logging at start and end + * Metrics to record CPU, wallclock and DB time by endpoint. + + It provides a method `processing` which should be called by the Resource + which is handling the request, and returns a context manager. + + """ + def __init__(self, site, channel, *args, **kw): + Request.__init__(self, channel, *args, **kw) self.site = site + self._channel = channel self.authenticated_entity = None self.start_time = 0 + global _next_request_seq + self.request_seq = _next_request_seq + _next_request_seq += 1 + def __repr__(self): # We overwrite this so that we don't log ``access_token`` - return '<%s at 0x%x method=%s uri=%s clientproto=%s site=%s>' % ( + return '<%s at 0x%x method=%r uri=%r clientproto=%r site=%r>' % ( self.__class__.__name__, id(self), self.method, @@ -41,16 +64,27 @@ class SynapseRequest(Request): self.site.site_tag, ) + def get_request_id(self): + return "%s-%i" % (self.method, self.request_seq) + def get_redacted_uri(self): - return ACCESS_TOKEN_RE.sub( - r'\1\3', - self.uri - ) + return redact_uri(self.uri) def get_user_agent(self): - return self.requestHeaders.getRawHeaders("User-Agent", [None])[-1] + return self.requestHeaders.getRawHeaders(b"User-Agent", [None])[-1] + + def render(self, resrc): + # override the Server header which is set by twisted + self.setHeader("Server", self.site.server_version_string) + return Request.render(self, resrc) + + def _started_processing(self, servlet_name): + self.start_time = time.time() + self.request_metrics = RequestMetrics() + self.request_metrics.start( + self.start_time, name=servlet_name, method=self.method, + ) - def started_processing(self): self.site.access_logger.info( "%s - %s - Received request: %s %s", self.getClientIP(), @@ -58,44 +92,85 @@ class SynapseRequest(Request): self.method, self.get_redacted_uri() ) - self.start_time = int(time.time() * 1000) - - def finished_processing(self): + def _finished_processing(self): try: context = LoggingContext.current_context() - ru_utime, ru_stime = context.get_resource_usage() - db_txn_count = context.db_txn_count - db_txn_duration = context.db_txn_duration - except: - ru_utime, ru_stime = (0, 0) - db_txn_count, db_txn_duration = (0, 0) + usage = context.get_resource_usage() + except Exception: + usage = ContextResourceUsage() + + end_time = time.time() + + # need to decode as it could be raw utf-8 bytes + # from a IDN servname in an auth header + authenticated_entity = self.authenticated_entity + if authenticated_entity is not None: + authenticated_entity = authenticated_entity.decode("utf-8", "replace") + + # ...or could be raw utf-8 bytes in the User-Agent header. + # N.B. if you don't do this, the logger explodes cryptically + # with maximum recursion trying to log errors about + # the charset problem. + # c.f. https://github.com/matrix-org/synapse/issues/3471 + user_agent = self.get_user_agent() + if user_agent is not None: + user_agent = user_agent.decode("utf-8", "replace") self.site.access_logger.info( "%s - %s - {%s}" - " Processed request: %dms (%dms, %dms) (%dms/%d)" - " %sB %s \"%s %s %s\" \"%s\"", + " Processed request: %.3fsec (%.3fsec, %.3fsec) (%.3fsec/%.3fsec/%d)" + " %sB %s \"%s %s %s\" \"%s\" [%d dbevts]", self.getClientIP(), self.site.site_tag, - self.authenticated_entity, - int(time.time() * 1000) - self.start_time, - int(ru_utime * 1000), - int(ru_stime * 1000), - int(db_txn_duration * 1000), - int(db_txn_count), + authenticated_entity, + end_time - self.start_time, + usage.ru_utime, + usage.ru_stime, + usage.db_sched_duration_sec, + usage.db_txn_duration_sec, + int(usage.db_txn_count), self.sentLength, self.code, self.method, self.get_redacted_uri(), self.clientproto, - self.get_user_agent(), + user_agent, + usage.evt_db_fetch_count, ) + try: + self.request_metrics.stop(end_time, self) + except Exception as e: + logger.warn("Failed to stop metrics: %r", e) + @contextlib.contextmanager - def processing(self): - self.started_processing() + def processing(self, servlet_name): + """Record the fact that we are processing this request. + + Returns a context manager; the correct way to use this is: + + @defer.inlineCallbacks + def handle_request(request): + with request.processing("FooServlet"): + yield really_handle_the_request() + + This will log the request's arrival. Once the context manager is + closed, the completion of the request will be logged, and the various + metrics will be updated. + + Args: + servlet_name (str): the name of the servlet which will be + processing this request. This is used in the metrics. + + It is possible to update this afterwards by updating + self.request_metrics.servlet_name. + """ + # TODO: we should probably just move this into render() and finish(), + # to save having to call a separate method. + self._started_processing(servlet_name) yield - self.finished_processing() + self._finished_processing() class XForwardedForRequest(SynapseRequest): @@ -133,7 +208,8 @@ class SynapseSite(Site): Subclass of a twisted http Site that does access logging with python's standard logging """ - def __init__(self, logger_name, site_tag, config, resource, *args, **kwargs): + def __init__(self, logger_name, site_tag, config, resource, + server_version_string, *args, **kwargs): Site.__init__(self, resource, *args, **kwargs) self.site_tag = site_tag @@ -141,6 +217,7 @@ class SynapseSite(Site): proxied = config.get("x_forwarded", False) self.requestFactory = SynapseRequestFactory(self, proxied) self.access_logger = logging.getLogger(logger_name) + self.server_version_string = server_version_string def log(self, request): pass diff --git a/synapse/metrics/__init__.py b/synapse/metrics/__init__.py index 2265e6e8d..a9158fc06 100644 --- a/synapse/metrics/__init__.py +++ b/synapse/metrics/__init__.py @@ -13,118 +13,198 @@ # See the License for the specific language governing permissions and # limitations under the License. -import logging import functools -import time import gc +import logging +import os +import platform +import time + +import attr +from prometheus_client import Counter, Gauge, Histogram +from prometheus_client.core import REGISTRY, GaugeMetricFamily from twisted.internet import reactor -from .metric import ( - CounterMetric, CallbackMetric, DistributionMetric, CacheMetric, - MemoryUsageMetric, -) -from .process_collector import register_process_collector - - logger = logging.getLogger(__name__) - +running_on_pypy = platform.python_implementation() == "PyPy" all_metrics = [] all_collectors = [] +all_gauges = {} + +HAVE_PROC_SELF_STAT = os.path.exists("/proc/self/stat") -class Metrics(object): - """ A single Metrics object gives a (mutable) slice view of the all_metrics - dict, allowing callers to easily register new metrics that are namespaced - nicely.""" +class RegistryProxy(object): - def __init__(self, name): - self.name_prefix = name - - def make_subspace(self, name): - return Metrics("%s_%s" % (self.name_prefix, name)) - - def register_collector(self, func): - all_collectors.append(func) - - def _register(self, metric_class, name, *args, **kwargs): - full_name = "%s_%s" % (self.name_prefix, name) - - metric = metric_class(full_name, *args, **kwargs) - - all_metrics.append(metric) - return metric - - def register_counter(self, *args, **kwargs): - return self._register(CounterMetric, *args, **kwargs) - - def register_callback(self, *args, **kwargs): - return self._register(CallbackMetric, *args, **kwargs) - - def register_distribution(self, *args, **kwargs): - return self._register(DistributionMetric, *args, **kwargs) - - def register_cache(self, *args, **kwargs): - return self._register(CacheMetric, *args, **kwargs) + @staticmethod + def collect(): + for metric in REGISTRY.collect(): + if not metric.name.startswith("__"): + yield metric -def register_memory_metrics(hs): - try: - import psutil - process = psutil.Process() - process.memory_info().rss - except (ImportError, AttributeError): - logger.warn( - "psutil is not installed or incorrect version." - " Disabling memory metrics." - ) - return - metric = MemoryUsageMetric(hs, psutil) - all_metrics.append(metric) +@attr.s(hash=True) +class LaterGauge(object): + name = attr.ib() + desc = attr.ib() + labels = attr.ib(hash=False) + caller = attr.ib() -def get_metrics_for(pkg_name): - """ Returns a Metrics instance for conveniently creating metrics - namespaced with the given name prefix. """ + def collect(self): - # Convert a "package.name" to "package_name" because Prometheus doesn't - # let us use . in metric names - return Metrics(pkg_name.replace(".", "_")) + g = GaugeMetricFamily(self.name, self.desc, labels=self.labels) - -def render_all(): - strs = [] - - for collector in all_collectors: - collector() - - for metric in all_metrics: try: - strs += metric.render() + calls = self.caller() except Exception: - strs += ["# FAILED to render"] - logger.exception("Failed to render metric") + logger.exception( + "Exception running callback for LaterGauge(%s)", + self.name, + ) + yield g + return - strs.append("") # to generate a final CRLF + if isinstance(calls, dict): + for k, v in calls.items(): + g.add_metric(k, v) + else: + g.add_metric([], calls) - return "\n".join(strs) + yield g + + def __attrs_post_init__(self): + self._register() + + def _register(self): + if self.name in all_gauges.keys(): + logger.warning("%s already registered, reregistering" % (self.name,)) + REGISTRY.unregister(all_gauges.pop(self.name)) + + REGISTRY.register(self) + all_gauges[self.name] = self -register_process_collector(get_metrics_for("process")) +# +# Detailed CPU metrics +# + +class CPUMetrics(object): + + def __init__(self): + ticks_per_sec = 100 + try: + # Try and get the system config + ticks_per_sec = os.sysconf('SC_CLK_TCK') + except (ValueError, TypeError, AttributeError): + pass + + self.ticks_per_sec = ticks_per_sec + + def collect(self): + if not HAVE_PROC_SELF_STAT: + return + + with open("/proc/self/stat") as s: + line = s.read() + raw_stats = line.split(") ", 1)[1].split(" ") + + user = GaugeMetricFamily("process_cpu_user_seconds_total", "") + user.add_metric([], float(raw_stats[11]) / self.ticks_per_sec) + yield user + + sys = GaugeMetricFamily("process_cpu_system_seconds_total", "") + sys.add_metric([], float(raw_stats[12]) / self.ticks_per_sec) + yield sys -python_metrics = get_metrics_for("python") +REGISTRY.register(CPUMetrics()) -gc_time = python_metrics.register_distribution("gc_time", labels=["gen"]) -gc_unreachable = python_metrics.register_counter("gc_unreachable_total", labels=["gen"]) -python_metrics.register_callback( - "gc_counts", lambda: {(i,): v for i, v in enumerate(gc.get_count())}, labels=["gen"] +# +# Python GC metrics +# + +gc_unreachable = Gauge("python_gc_unreachable_total", "Unreachable GC objects", ["gen"]) +gc_time = Histogram( + "python_gc_time", + "Time taken to GC (sec)", + ["gen"], + buckets=[0.0025, 0.005, 0.01, 0.025, 0.05, 0.10, 0.25, 0.50, 1.00, 2.50, + 5.00, 7.50, 15.00, 30.00, 45.00, 60.00], ) -reactor_metrics = get_metrics_for("python.twisted.reactor") -tick_time = reactor_metrics.register_distribution("tick_time") -pending_calls_metric = reactor_metrics.register_distribution("pending_calls") + +class GCCounts(object): + + def collect(self): + cm = GaugeMetricFamily("python_gc_counts", "GC object counts", labels=["gen"]) + for n, m in enumerate(gc.get_count()): + cm.add_metric([str(n)], m) + + yield cm + + +if not running_on_pypy: + REGISTRY.register(GCCounts()) + +# +# Twisted reactor metrics +# + +tick_time = Histogram( + "python_twisted_reactor_tick_time", + "Tick time of the Twisted reactor (sec)", + buckets=[0.001, 0.002, 0.005, 0.01, 0.025, 0.05, 0.1, 0.2, 0.5, 1, 2, 5], +) +pending_calls_metric = Histogram( + "python_twisted_reactor_pending_calls", + "Pending calls", + buckets=[1, 2, 5, 10, 25, 50, 100, 250, 500, 1000], +) + +# +# Federation Metrics +# + +sent_edus_counter = Counter("synapse_federation_client_sent_edus", "") + +sent_transactions_counter = Counter("synapse_federation_client_sent_transactions", "") + +events_processed_counter = Counter("synapse_federation_client_events_processed", "") + +# Used to track where various components have processed in the event stream, +# e.g. federation sending, appservice sending, etc. +event_processing_positions = Gauge("synapse_event_processing_positions", "", ["name"]) + +# Used to track the current max events stream position +event_persisted_position = Gauge("synapse_event_persisted_position", "") + +# Used to track the received_ts of the last event processed by various +# components +event_processing_last_ts = Gauge("synapse_event_processing_last_ts", "", ["name"]) + +# Used to track the lag processing events. This is the time difference +# between the last processed event's received_ts and the time it was +# finished being processed. +event_processing_lag = Gauge("synapse_event_processing_lag", "", ["name"]) + +last_ticked = time.time() + + +class ReactorLastSeenMetric(object): + + def collect(self): + cm = GaugeMetricFamily( + "python_twisted_reactor_last_seen", + "Seconds since the Twisted reactor was last seen", + ) + cm.add_metric([], time.time() - last_ticked) + yield cm + + +REGISTRY.register(ReactorLastSeenMetric()) def runUntilCurrentTimer(func): @@ -146,12 +226,25 @@ def runUntilCurrentTimer(func): num_pending += 1 num_pending += len(reactor.threadCallQueue) - - start = time.time() * 1000 + start = time.time() ret = func(*args, **kwargs) - end = time.time() * 1000 - tick_time.inc_by(end - start) - pending_calls_metric.inc_by(num_pending) + end = time.time() + + # record the amount of wallclock time spent running pending calls. + # This is a proxy for the actual amount of time between reactor polls, + # since about 25% of time is actually spent running things triggered by + # I/O events, but that is harder to capture without rewriting half the + # reactor. + tick_time.observe(end - start) + pending_calls_metric.observe(num_pending) + + # Update the time we last ticked, for the metric to test whether + # Synapse's reactor has frozen + global last_ticked + last_ticked = end + + if running_on_pypy: + return ret # Check if we need to do a manual GC (since its been disabled), and do # one if necessary. @@ -161,12 +254,12 @@ def runUntilCurrentTimer(func): if threshold[i] < counts[i]: logger.info("Collecting gc %d", i) - start = time.time() * 1000 + start = time.time() unreachable = gc.collect(i) - end = time.time() * 1000 + end = time.time() - gc_time.inc_by(end - start, i) - gc_unreachable.inc_by(unreachable, i) + gc_time.labels(i).observe(end - start) + gc_unreachable.labels(i).set(unreachable) return ret @@ -185,6 +278,7 @@ try: # We manually run the GC each reactor tick so that we can get some metrics # about time spent doing GC, - gc.disable() + if not running_on_pypy: + gc.disable() except AttributeError: pass diff --git a/synapse/metrics/background_process_metrics.py b/synapse/metrics/background_process_metrics.py new file mode 100644 index 000000000..9d820e44a --- /dev/null +++ b/synapse/metrics/background_process_metrics.py @@ -0,0 +1,179 @@ +# -*- coding: utf-8 -*- +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import six + +from prometheus_client.core import REGISTRY, Counter, GaugeMetricFamily + +from twisted.internet import defer + +from synapse.util.logcontext import LoggingContext, PreserveLoggingContext + +_background_process_start_count = Counter( + "synapse_background_process_start_count", + "Number of background processes started", + ["name"], +) + +# we set registry=None in all of these to stop them getting registered with +# the default registry. Instead we collect them all via the CustomCollector, +# which ensures that we can update them before they are collected. +# +_background_process_ru_utime = Counter( + "synapse_background_process_ru_utime_seconds", + "User CPU time used by background processes, in seconds", + ["name"], + registry=None, +) + +_background_process_ru_stime = Counter( + "synapse_background_process_ru_stime_seconds", + "System CPU time used by background processes, in seconds", + ["name"], + registry=None, +) + +_background_process_db_txn_count = Counter( + "synapse_background_process_db_txn_count", + "Number of database transactions done by background processes", + ["name"], + registry=None, +) + +_background_process_db_txn_duration = Counter( + "synapse_background_process_db_txn_duration_seconds", + ("Seconds spent by background processes waiting for database " + "transactions, excluding scheduling time"), + ["name"], + registry=None, +) + +_background_process_db_sched_duration = Counter( + "synapse_background_process_db_sched_duration_seconds", + "Seconds spent by background processes waiting for database connections", + ["name"], + registry=None, +) + +# map from description to a counter, so that we can name our logcontexts +# incrementally. (It actually duplicates _background_process_start_count, but +# it's much simpler to do so than to try to combine them.) +_background_process_counts = dict() # type: dict[str, int] + +# map from description to the currently running background processes. +# +# it's kept as a dict of sets rather than a big set so that we can keep track +# of process descriptions that no longer have any active processes. +_background_processes = dict() # type: dict[str, set[_BackgroundProcess]] + + +class _Collector(object): + """A custom metrics collector for the background process metrics. + + Ensures that all of the metrics are up-to-date with any in-flight processes + before they are returned. + """ + def collect(self): + background_process_in_flight_count = GaugeMetricFamily( + "synapse_background_process_in_flight_count", + "Number of background processes in flight", + labels=["name"], + ) + + for desc, processes in six.iteritems(_background_processes): + background_process_in_flight_count.add_metric( + (desc,), len(processes), + ) + for process in processes: + process.update_metrics() + + yield background_process_in_flight_count + + # now we need to run collect() over each of the static Counters, and + # yield each metric they return. + for m in ( + _background_process_ru_utime, + _background_process_ru_stime, + _background_process_db_txn_count, + _background_process_db_txn_duration, + _background_process_db_sched_duration, + ): + for r in m.collect(): + yield r + + +REGISTRY.register(_Collector()) + + +class _BackgroundProcess(object): + def __init__(self, desc, ctx): + self.desc = desc + self._context = ctx + self._reported_stats = None + + def update_metrics(self): + """Updates the metrics with values from this process.""" + new_stats = self._context.get_resource_usage() + if self._reported_stats is None: + diff = new_stats + else: + diff = new_stats - self._reported_stats + self._reported_stats = new_stats + + _background_process_ru_utime.labels(self.desc).inc(diff.ru_utime) + _background_process_ru_stime.labels(self.desc).inc(diff.ru_stime) + _background_process_db_txn_count.labels(self.desc).inc( + diff.db_txn_count, + ) + _background_process_db_txn_duration.labels(self.desc).inc( + diff.db_txn_duration_sec, + ) + _background_process_db_sched_duration.labels(self.desc).inc( + diff.db_sched_duration_sec, + ) + + +def run_as_background_process(desc, func, *args, **kwargs): + """Run the given function in its own logcontext, with resource metrics + + This should be used to wrap processes which are fired off to run in the + background, instead of being associated with a particular request. + + Args: + desc (str): a description for this background process type + func: a function, which may return a Deferred + args: positional args for func + kwargs: keyword args for func + + Returns: None + """ + @defer.inlineCallbacks + def run(): + count = _background_process_counts.get(desc, 0) + _background_process_counts[desc] = count + 1 + _background_process_start_count.labels(desc).inc() + + with LoggingContext(desc) as context: + context.request = "%s-%i" % (desc, count) + proc = _BackgroundProcess(desc, context) + _background_processes.setdefault(desc, set()).add(proc) + try: + yield func(*args, **kwargs) + finally: + proc.update_metrics() + _background_processes[desc].remove(proc) + + with PreserveLoggingContext(): + run() diff --git a/synapse/metrics/metric.py b/synapse/metrics/metric.py deleted file mode 100644 index e87b2b80a..000000000 --- a/synapse/metrics/metric.py +++ /dev/null @@ -1,195 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2015, 2016 OpenMarket Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - - -from itertools import chain - - -# TODO(paul): I can't believe Python doesn't have one of these -def map_concat(func, items): - # flatten a list-of-lists - return list(chain.from_iterable(map(func, items))) - - -class BaseMetric(object): - - def __init__(self, name, labels=[]): - self.name = name - self.labels = labels # OK not to clone as we never write it - - def dimension(self): - return len(self.labels) - - def is_scalar(self): - return not len(self.labels) - - def _render_labelvalue(self, value): - # TODO: some kind of value escape - return '"%s"' % (value) - - def _render_key(self, values): - if self.is_scalar(): - return "" - return "{%s}" % ( - ",".join(["%s=%s" % (k, self._render_labelvalue(v)) - for k, v in zip(self.labels, values)]) - ) - - -class CounterMetric(BaseMetric): - """The simplest kind of metric; one that stores a monotonically-increasing - integer that counts events.""" - - def __init__(self, *args, **kwargs): - super(CounterMetric, self).__init__(*args, **kwargs) - - self.counts = {} - - # Scalar metrics are never empty - if self.is_scalar(): - self.counts[()] = 0 - - def inc_by(self, incr, *values): - if len(values) != self.dimension(): - raise ValueError( - "Expected as many values to inc() as labels (%d)" % (self.dimension()) - ) - - # TODO: should assert that the tag values are all strings - - if values not in self.counts: - self.counts[values] = incr - else: - self.counts[values] += incr - - def inc(self, *values): - self.inc_by(1, *values) - - def render_item(self, k): - return ["%s%s %d" % (self.name, self._render_key(k), self.counts[k])] - - def render(self): - return map_concat(self.render_item, sorted(self.counts.keys())) - - -class CallbackMetric(BaseMetric): - """A metric that returns the numeric value returned by a callback whenever - it is rendered. Typically this is used to implement gauges that yield the - size or other state of some in-memory object by actively querying it.""" - - def __init__(self, name, callback, labels=[]): - super(CallbackMetric, self).__init__(name, labels=labels) - - self.callback = callback - - def render(self): - value = self.callback() - - if self.is_scalar(): - return ["%s %.12g" % (self.name, value)] - - return ["%s%s %.12g" % (self.name, self._render_key(k), value[k]) - for k in sorted(value.keys())] - - -class DistributionMetric(object): - """A combination of an event counter and an accumulator, which counts - both the number of events and accumulates the total value. Typically this - could be used to keep track of method-running times, or other distributions - of values that occur in discrete occurances. - - TODO(paul): Try to export some heatmap-style stats? - """ - - def __init__(self, name, *args, **kwargs): - self.counts = CounterMetric(name + ":count", **kwargs) - self.totals = CounterMetric(name + ":total", **kwargs) - - def inc_by(self, inc, *values): - self.counts.inc(*values) - self.totals.inc_by(inc, *values) - - def render(self): - return self.counts.render() + self.totals.render() - - -class CacheMetric(object): - __slots__ = ("name", "cache_name", "hits", "misses", "size_callback") - - def __init__(self, name, size_callback, cache_name): - self.name = name - self.cache_name = cache_name - - self.hits = 0 - self.misses = 0 - - self.size_callback = size_callback - - def inc_hits(self): - self.hits += 1 - - def inc_misses(self): - self.misses += 1 - - def render(self): - size = self.size_callback() - hits = self.hits - total = self.misses + self.hits - - return [ - """%s:hits{name="%s"} %d""" % (self.name, self.cache_name, hits), - """%s:total{name="%s"} %d""" % (self.name, self.cache_name, total), - """%s:size{name="%s"} %d""" % (self.name, self.cache_name, size), - ] - - -class MemoryUsageMetric(object): - """Keeps track of the current memory usage, using psutil. - - The class will keep the current min/max/sum/counts of rss over the last - WINDOW_SIZE_SEC, by polling UPDATE_HZ times per second - """ - - UPDATE_HZ = 2 # number of times to get memory per second - WINDOW_SIZE_SEC = 30 # the size of the window in seconds - - def __init__(self, hs, psutil): - clock = hs.get_clock() - self.memory_snapshots = [] - - self.process = psutil.Process() - - clock.looping_call(self._update_curr_values, 1000 / self.UPDATE_HZ) - - def _update_curr_values(self): - max_size = self.UPDATE_HZ * self.WINDOW_SIZE_SEC - self.memory_snapshots.append(self.process.memory_info().rss) - self.memory_snapshots[:] = self.memory_snapshots[-max_size:] - - def render(self): - if not self.memory_snapshots: - return [] - - max_rss = max(self.memory_snapshots) - min_rss = min(self.memory_snapshots) - sum_rss = sum(self.memory_snapshots) - len_rss = len(self.memory_snapshots) - - return [ - "process_psutil_rss:max %d" % max_rss, - "process_psutil_rss:min %d" % min_rss, - "process_psutil_rss:total %d" % sum_rss, - "process_psutil_rss:count %d" % len_rss, - ] diff --git a/synapse/metrics/process_collector.py b/synapse/metrics/process_collector.py deleted file mode 100644 index 6fec3de39..000000000 --- a/synapse/metrics/process_collector.py +++ /dev/null @@ -1,122 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2015, 2016 OpenMarket Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -import os - - -TICKS_PER_SEC = 100 -BYTES_PER_PAGE = 4096 - -HAVE_PROC_STAT = os.path.exists("/proc/stat") -HAVE_PROC_SELF_STAT = os.path.exists("/proc/self/stat") -HAVE_PROC_SELF_LIMITS = os.path.exists("/proc/self/limits") -HAVE_PROC_SELF_FD = os.path.exists("/proc/self/fd") - -# Field indexes from /proc/self/stat, taken from the proc(5) manpage -STAT_FIELDS = { - "utime": 14, - "stime": 15, - "starttime": 22, - "vsize": 23, - "rss": 24, -} - - -stats = {} - -# In order to report process_start_time_seconds we need to know the -# machine's boot time, because the value in /proc/self/stat is relative to -# this -boot_time = None -if HAVE_PROC_STAT: - with open("/proc/stat") as _procstat: - for line in _procstat: - if line.startswith("btime "): - boot_time = int(line.split()[1]) - - -def update_resource_metrics(): - if HAVE_PROC_SELF_STAT: - global stats - with open("/proc/self/stat") as s: - line = s.read() - # line is PID (command) more stats go here ... - raw_stats = line.split(") ", 1)[1].split(" ") - - for (name, index) in STAT_FIELDS.iteritems(): - # subtract 3 from the index, because proc(5) is 1-based, and - # we've lost the first two fields in PID and COMMAND above - stats[name] = int(raw_stats[index - 3]) - - -def _count_fds(): - # Not every OS will have a /proc/self/fd directory - if not HAVE_PROC_SELF_FD: - return 0 - - return len(os.listdir("/proc/self/fd")) - - -def register_process_collector(process_metrics): - process_metrics.register_collector(update_resource_metrics) - - if HAVE_PROC_SELF_STAT: - process_metrics.register_callback( - "cpu_user_seconds_total", - lambda: float(stats["utime"]) / TICKS_PER_SEC - ) - process_metrics.register_callback( - "cpu_system_seconds_total", - lambda: float(stats["stime"]) / TICKS_PER_SEC - ) - process_metrics.register_callback( - "cpu_seconds_total", - lambda: (float(stats["utime"] + stats["stime"])) / TICKS_PER_SEC - ) - - process_metrics.register_callback( - "virtual_memory_bytes", - lambda: int(stats["vsize"]) - ) - process_metrics.register_callback( - "resident_memory_bytes", - lambda: int(stats["rss"]) * BYTES_PER_PAGE - ) - - process_metrics.register_callback( - "start_time_seconds", - lambda: boot_time + int(stats["starttime"]) / TICKS_PER_SEC - ) - - if HAVE_PROC_SELF_FD: - process_metrics.register_callback( - "open_fds", - lambda: _count_fds() - ) - - if HAVE_PROC_SELF_LIMITS: - def _get_max_fds(): - with open("/proc/self/limits") as limits: - for line in limits: - if not line.startswith("Max open files "): - continue - # Line is Max open files $SOFT $HARD - return int(line.split()[3]) - return None - - process_metrics.register_callback( - "max_fds", - lambda: _get_max_fds() - ) diff --git a/synapse/metrics/resource.py b/synapse/metrics/resource.py index 870f40060..978935907 100644 --- a/synapse/metrics/resource.py +++ b/synapse/metrics/resource.py @@ -13,27 +13,8 @@ # See the License for the specific language governing permissions and # limitations under the License. -from twisted.web.resource import Resource - -import synapse.metrics - +from prometheus_client.twisted import MetricsResource METRICS_PREFIX = "/_synapse/metrics" - -class MetricsResource(Resource): - isLeaf = True - - def __init__(self, hs): - Resource.__init__(self) # Resource is old-style, so no super() - - self.hs = hs - - def render_GET(self, request): - response = synapse.metrics.render_all() - - request.setHeader("Content-Type", "text/plain") - request.setHeader("Content-Length", str(len(response))) - - # Encode as UTF-8 (default) - return response.encode() +__all__ = ["MetricsResource", "METRICS_PREFIX"] diff --git a/synapse/module_api/__init__.py b/synapse/module_api/__init__.py new file mode 100644 index 000000000..097c844d3 --- /dev/null +++ b/synapse/module_api/__init__.py @@ -0,0 +1,123 @@ +# -*- coding: utf-8 -*- +# Copyright 2017 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +from twisted.internet import defer + +from synapse.types import UserID + + +class ModuleApi(object): + """A proxy object that gets passed to password auth providers so they + can register new users etc if necessary. + """ + def __init__(self, hs, auth_handler): + self.hs = hs + + self._store = hs.get_datastore() + self._auth = hs.get_auth() + self._auth_handler = auth_handler + + def get_user_by_req(self, req, allow_guest=False): + """Check the access_token provided for a request + + Args: + req (twisted.web.server.Request): Incoming HTTP request + allow_guest (bool): True if guest users should be allowed. If this + is False, and the access token is for a guest user, an + AuthError will be thrown + Returns: + twisted.internet.defer.Deferred[synapse.types.Requester]: + the requester for this request + Raises: + synapse.api.errors.AuthError: if no user by that token exists, + or the token is invalid. + """ + return self._auth.get_user_by_req(req, allow_guest) + + def get_qualified_user_id(self, username): + """Qualify a user id, if necessary + + Takes a user id provided by the user and adds the @ and :domain to + qualify it, if necessary + + Args: + username (str): provided user id + + Returns: + str: qualified @user:id + """ + if username.startswith('@'): + return username + return UserID(username, self.hs.hostname).to_string() + + def check_user_exists(self, user_id): + """Check if user exists. + + Args: + user_id (str): Complete @user:id + + Returns: + Deferred[str|None]: Canonical (case-corrected) user_id, or None + if the user is not registered. + """ + return self._auth_handler.check_user_exists(user_id) + + def register(self, localpart): + """Registers a new user with given localpart + + Returns: + Deferred: a 2-tuple of (user_id, access_token) + """ + reg = self.hs.get_handlers().registration_handler + return reg.register(localpart=localpart) + + @defer.inlineCallbacks + def invalidate_access_token(self, access_token): + """Invalidate an access token for a user + + Args: + access_token(str): access token + + Returns: + twisted.internet.defer.Deferred - resolves once the access token + has been removed. + + Raises: + synapse.api.errors.AuthError: the access token is invalid + """ + # see if the access token corresponds to a device + user_info = yield self._auth.get_user_by_access_token(access_token) + device_id = user_info.get("device_id") + user_id = user_info["user"].to_string() + if device_id: + # delete the device, which will also delete its access tokens + yield self.hs.get_device_handler().delete_device(user_id, device_id) + else: + # no associated device. Just delete the access token. + yield self._auth_handler.delete_access_token(access_token) + + def run_db_interaction(self, desc, func, *args, **kwargs): + """Run a function with a database connection + + Args: + desc (str): description for the transaction, for metrics etc + func (func): function to be run. Passed a database cursor object + as well as *args and **kwargs + *args: positional args to be passed to func + **kwargs: named args to be passed to func + + Returns: + Deferred[object]: result of func + """ + return self._store.runInteraction(desc, func, *args, **kwargs) diff --git a/synapse/notifier.py b/synapse/notifier.py index 385208b57..e650c3e49 100644 --- a/synapse/notifier.py +++ b/synapse/notifier.py @@ -13,34 +13,34 @@ # See the License for the specific language governing permissions and # limitations under the License. +import logging +from collections import namedtuple + +from prometheus_client import Counter + from twisted.internet import defer + from synapse.api.constants import EventTypes, Membership from synapse.api.errors import AuthError from synapse.handlers.presence import format_user_presence_state - -from synapse.util import DeferredTimedOutError -from synapse.util.logutils import log_function -from synapse.util.async import ObservableDeferred -from synapse.util.logcontext import PreserveLoggingContext, preserve_fn -from synapse.util.metrics import Measure +from synapse.metrics import LaterGauge from synapse.types import StreamToken +from synapse.util.async import ( + DeferredTimeoutError, + ObservableDeferred, + add_timeout_to_deferred, +) +from synapse.util.logcontext import PreserveLoggingContext, run_in_background +from synapse.util.logutils import log_function +from synapse.util.metrics import Measure from synapse.visibility import filter_events_for_client -import synapse.metrics - -from collections import namedtuple - -import logging - logger = logging.getLogger(__name__) -metrics = synapse.metrics.get_metrics_for(__name__) +notified_events_counter = Counter("synapse_notifier_notified_events", "") -notified_events_counter = metrics.register_counter("notified_events") - -users_woken_by_stream_counter = metrics.register_counter( - "users_woken_by_stream", labels=["stream"] -) +users_woken_by_stream_counter = Counter( + "synapse_notifier_users_woken_by_stream", "", ["stream"]) # TODO(paul): Should be shared somewhere @@ -105,7 +105,7 @@ class _NotifierUserStream(object): self.last_notified_ms = time_now_ms noify_deferred = self.notify_deferred - users_woken_by_stream_counter.inc(stream_key) + users_woken_by_stream_counter.labels(stream_key).inc() with PreserveLoggingContext(): self.notify_deferred = ObservableDeferred(defer.Deferred()) @@ -144,6 +144,7 @@ class _NotifierUserStream(object): class EventStreamResult(namedtuple("EventStreamResult", ("events", "tokens"))): def __nonzero__(self): return bool(self.events) + __bool__ = __nonzero__ # python3 class Notifier(object): @@ -159,6 +160,7 @@ class Notifier(object): self.user_to_user_stream = {} self.room_to_user_streams = {} + self.hs = hs self.event_sources = hs.get_event_sources() self.store = hs.get_datastore() self.pending_new_room_events = [] @@ -193,14 +195,14 @@ class Notifier(object): all_user_streams.add(x) return sum(stream.count_listeners() for stream in all_user_streams) - metrics.register_callback("listeners", count_listeners) + LaterGauge("synapse_notifier_listeners", "", [], count_listeners) - metrics.register_callback( - "rooms", + LaterGauge( + "synapse_notifier_rooms", "", [], lambda: count(bool, self.room_to_user_streams.values()), ) - metrics.register_callback( - "users", + LaterGauge( + "synapse_notifier_users", "", [], lambda: len(self.user_to_user_stream), ) @@ -250,14 +252,10 @@ class Notifier(object): def _on_new_room_event(self, event, room_stream_id, extra_users=[]): """Notify any user streams that are interested in this room event""" # poke any interested application service. - preserve_fn(self.appservice_handler.notify_interested_services)( - room_stream_id - ) + run_in_background(self._notify_app_services, room_stream_id) if self.federation_sender: - preserve_fn(self.federation_sender.notify_new_events)( - room_stream_id - ) + self.federation_sender.notify_new_events(room_stream_id) if event.type == EventTypes.Member and event.membership == Membership.JOIN: self._user_joined_room(event.state_key, event.room_id) @@ -268,8 +266,15 @@ class Notifier(object): rooms=[event.room_id], ) + @defer.inlineCallbacks + def _notify_app_services(self, room_stream_id): + try: + yield self.appservice_handler.notify_interested_services(room_stream_id) + except Exception: + logger.exception("Error notifying application services of event") + def on_new_event(self, stream_key, new_token, users=[], rooms=[]): - """ Used to inform listeners that something has happend event wise. + """ Used to inform listeners that something has happened event wise. Will wake up all listeners for the given users and rooms. """ @@ -289,7 +294,7 @@ class Notifier(object): for user_stream in user_streams: try: user_stream.notify(stream_key, new_token, time_now_ms) - except: + except Exception: logger.exception("Failed to notify listener") self.notify_replication() @@ -297,8 +302,7 @@ class Notifier(object): def on_new_replication_data(self): """Used to inform replication listeners that something has happend without waking up any of the normal user event streams""" - with PreserveLoggingContext(): - self.notify_replication() + self.notify_replication() @defer.inlineCallbacks def wait_for_events(self, user_id, timeout, callback, room_ids=None, @@ -333,11 +337,13 @@ class Notifier(object): # Now we wait for the _NotifierUserStream to be told there # is a new token. listener = user_stream.new_listener(prev_token) + add_timeout_to_deferred( + listener.deferred, + (end_time - now) / 1000., + self.hs.get_reactor(), + ) with PreserveLoggingContext(): - yield self.clock.time_bound_deferred( - listener.deferred, - time_out=(end_time - now) / 1000. - ) + yield listener.deferred current_token = user_stream.current_token @@ -348,7 +354,7 @@ class Notifier(object): # Update the prev_token to the current_token since nothing # has happened between the old prev_token and the current_token prev_token = current_token - except DeferredTimedOutError: + except DeferredTimeoutError: break except defer.CancelledError: break @@ -516,8 +522,14 @@ class Notifier(object): self.replication_deferred = ObservableDeferred(defer.Deferred()) deferred.callback(None) - for cb in self.replication_callbacks: - preserve_fn(cb)() + # the callbacks may well outlast the current request, so we run + # them in the sentinel logcontext. + # + # (ideally it would be up to the callbacks to know if they were + # starting off background processes and drop the logcontext + # accordingly, but that requires more changes) + for cb in self.replication_callbacks: + cb() @defer.inlineCallbacks def wait_for_replication(self, callback, timeout): @@ -547,13 +559,15 @@ class Notifier(object): if end_time <= now: break + add_timeout_to_deferred( + listener.deferred.addTimeout, + (end_time - now) / 1000., + self.hs.get_reactor(), + ) try: with PreserveLoggingContext(): - yield self.clock.time_bound_deferred( - listener.deferred, - time_out=(end_time - now) / 1000. - ) - except DeferredTimedOutError: + yield listener.deferred + except DeferredTimeoutError: break except defer.CancelledError: break diff --git a/synapse/push/action_generator.py b/synapse/push/action_generator.py index fe09d50d5..a5de75c48 100644 --- a/synapse/push/action_generator.py +++ b/synapse/push/action_generator.py @@ -13,13 +13,13 @@ # See the License for the specific language governing permissions and # limitations under the License. -from twisted.internet import defer +import logging -from .bulk_push_rule_evaluator import BulkPushRuleEvaluator +from twisted.internet import defer from synapse.util.metrics import Measure -import logging +from .bulk_push_rule_evaluator import BulkPushRuleEvaluator logger = logging.getLogger(__name__) @@ -40,10 +40,6 @@ class ActionGenerator(object): @defer.inlineCallbacks def handle_push_actions_for_event(self, event, context): with Measure(self.clock, "action_for_event_by_user"): - actions_by_user = yield self.bulk_evaluator.action_for_event_by_user( + yield self.bulk_evaluator.action_for_event_by_user( event, context ) - - context.push_actions = [ - (uid, actions) for uid, actions in actions_by_user.iteritems() - ] diff --git a/synapse/push/baserules.py b/synapse/push/baserules.py index 85effdfa4..8f0682c94 100644 --- a/synapse/push/baserules.py +++ b/synapse/push/baserules.py @@ -1,4 +1,5 @@ # Copyright 2015, 2016 OpenMarket Ltd +# Copyright 2017 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -12,9 +13,10 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.push.rulekinds import PRIORITY_CLASS_MAP, PRIORITY_CLASS_INVERSE_MAP import copy +from synapse.push.rulekinds import PRIORITY_CLASS_INVERSE_MAP, PRIORITY_CLASS_MAP + def list_with_base_rules(rawrules): """Combine the list of rules set by the user with the default push rules @@ -38,7 +40,7 @@ def list_with_base_rules(rawrules): rawrules = [r for r in rawrules if r['priority_class'] >= 0] # shove the server default rules for each kind onto the end of each - current_prio_class = PRIORITY_CLASS_INVERSE_MAP.keys()[-1] + current_prio_class = list(PRIORITY_CLASS_INVERSE_MAP)[-1] ruleslist.extend(make_base_prepend_rules( PRIORITY_CLASS_INVERSE_MAP[current_prio_class], modified_base_rules @@ -238,6 +240,28 @@ BASE_APPEND_OVERRIDE_RULES = [ } ] }, + { + 'rule_id': 'global/override/.m.rule.roomnotif', + 'conditions': [ + { + 'kind': 'event_match', + 'key': 'content.body', + 'pattern': '@room', + '_id': '_roomnotif_content', + }, + { + 'kind': 'sender_notification_permission', + 'key': 'room', + '_id': '_roomnotif_pl', + }, + ], + 'actions': [ + 'notify', { + 'set_tweak': 'highlight', + 'value': True, + } + ] + } ] diff --git a/synapse/push/bulk_push_rule_evaluator.py b/synapse/push/bulk_push_rule_evaluator.py index 9a96e6fe8..1d14d3639 100644 --- a/synapse/push/bulk_push_rule_evaluator.py +++ b/synapse/push/bulk_push_rule_evaluator.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2015 OpenMarket Ltd +# Copyright 2017 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -14,18 +15,22 @@ # limitations under the License. import logging +from collections import namedtuple + +from six import iteritems, itervalues + +from prometheus_client import Counter from twisted.internet import defer -from .push_rule_evaluator import PushRuleEvaluatorForEvent - -from synapse.visibility import filter_events_for_clients_context from synapse.api.constants import EventTypes, Membership -from synapse.util.caches.descriptors import cached +from synapse.event_auth import get_user_power_level +from synapse.state import POWER_KEY from synapse.util.async import Linearizer +from synapse.util.caches import register_cache +from synapse.util.caches.descriptors import cached -from collections import namedtuple - +from .push_rule_evaluator import PushRuleEvaluatorForEvent logger = logging.getLogger(__name__) @@ -33,6 +38,20 @@ logger = logging.getLogger(__name__) rules_by_room = {} +push_rules_invalidation_counter = Counter( + "synapse_push_bulk_push_rule_evaluator_push_rules_invalidation_counter", "") +push_rules_state_size_counter = Counter( + "synapse_push_bulk_push_rule_evaluator_push_rules_state_size_counter", "") + +# Measures whether we use the fast path of using state deltas, or if we have to +# recalculate from scratch +push_rules_delta_state_cache_metric = register_cache( + "cache", + "push_rules_delta_state_cache_metric", + cache=[], # Meaningless size, as this isn't a cache that stores values +) + + class BulkPushRuleEvaluator(object): """Calculates the outcome of push rules for an event for all users in the room at once. @@ -41,6 +60,13 @@ class BulkPushRuleEvaluator(object): def __init__(self, hs): self.hs = hs self.store = hs.get_datastore() + self.auth = hs.get_auth() + + self.room_push_rule_cache_metrics = register_cache( + "cache", + "room_push_rule_cache", + cache=[], # Meaningless size, as this isn't a cache that stores values + ) @defer.inlineCallbacks def _get_rules_for_event(self, event, context): @@ -79,37 +105,69 @@ class BulkPushRuleEvaluator(object): # It's important that RulesForRoom gets added to self._get_rules_for_room.cache # before any lookup methods get called on it as otherwise there may be # a race if invalidate_all gets called (which assumes its in the cache) - return RulesForRoom(self.hs, room_id, self._get_rules_for_room.cache) + return RulesForRoom( + self.hs, room_id, self._get_rules_for_room.cache, + self.room_push_rule_cache_metrics, + ) + + @defer.inlineCallbacks + def _get_power_levels_and_sender_level(self, event, context): + prev_state_ids = yield context.get_prev_state_ids(self.store) + pl_event_id = prev_state_ids.get(POWER_KEY) + if pl_event_id: + # fastpath: if there's a power level event, that's all we need, and + # not having a power level event is an extreme edge case + pl_event = yield self.store.get_event(pl_event_id) + auth_events = {POWER_KEY: pl_event} + else: + auth_events_ids = yield self.auth.compute_auth_events( + event, prev_state_ids, for_verification=False, + ) + auth_events = yield self.store.get_events(auth_events_ids) + auth_events = { + (e.type, e.state_key): e for e in itervalues(auth_events) + } + + sender_level = get_user_power_level(event.sender, auth_events) + + pl_event = auth_events.get(POWER_KEY) + + defer.returnValue((pl_event.content if pl_event else {}, sender_level)) @defer.inlineCallbacks def action_for_event_by_user(self, event, context): - """Given an event and context, evaluate the push rules and return - the results + """Given an event and context, evaluate the push rules and insert the + results into the event_push_actions_staging table. Returns: - dict of user_id -> action + Deferred """ rules_by_user = yield self._get_rules_for_event(event, context) actions_by_user = {} - # None of these users can be peeking since this list of users comes - # from the set of users in the room, so we know for sure they're all - # actually in the room. - user_tuples = [(u, False) for u in rules_by_user] - - filtered_by_user = yield filter_events_for_clients_context( - self.store, user_tuples, [event], {event.event_id: context} - ) - room_members = yield self.store.get_joined_users_from_context( event, context ) - evaluator = PushRuleEvaluatorForEvent(event, len(room_members)) + (power_levels, sender_power_level) = ( + yield self._get_power_levels_and_sender_level(event, context) + ) + + evaluator = PushRuleEvaluatorForEvent( + event, len(room_members), sender_power_level, power_levels, + ) condition_cache = {} - for uid, rules in rules_by_user.iteritems(): + for uid, rules in iteritems(rules_by_user): + if event.sender == uid: + continue + + if not event.is_state(): + is_ignored = yield self.store.is_ignored_by(event.sender, uid) + if is_ignored: + continue + display_name = None profile_info = room_members.get(uid) if profile_info: @@ -121,13 +179,6 @@ class BulkPushRuleEvaluator(object): if event.type == EventTypes.Member and event.state_key == uid: display_name = event.content.get("displayname", None) - filtered = filtered_by_user[uid] - if len(filtered) == 0: - continue - - if filtered[0].sender == uid: - continue - for rule in rules: if 'enabled' in rule and not rule['enabled']: continue @@ -138,9 +189,16 @@ class BulkPushRuleEvaluator(object): if matches: actions = [x for x in rule['actions'] if x != 'dont_notify'] if actions and 'notify' in actions: + # Push rules say we should notify the user of this event actions_by_user[uid] = actions break - defer.returnValue(actions_by_user) + + # Mark in the DB staging area the push actions for users who should be + # notified for this event. (This will then get handled when we persist + # the event) + yield self.store.add_push_actions_to_staging( + event.event_id, actions_by_user, + ) def _condition_checker(evaluator, conditions, uid, display_name, cache): @@ -170,17 +228,19 @@ class RulesForRoom(object): the entire cache for the room. """ - def __init__(self, hs, room_id, rules_for_room_cache): + def __init__(self, hs, room_id, rules_for_room_cache, room_push_rule_cache_metrics): """ Args: hs (HomeServer) room_id (str) rules_for_room_cache(Cache): The cache object that caches these RoomsForUser objects. + room_push_rule_cache_metrics (CacheMetric) """ self.room_id = room_id self.is_mine_id = hs.is_mine_id self.store = hs.get_datastore() + self.room_push_rule_cache_metrics = room_push_rule_cache_metrics self.linearizer = Linearizer(name="rules_for_room") @@ -222,11 +282,19 @@ class RulesForRoom(object): """ state_group = context.state_group + if state_group and self.state_group == state_group: + logger.debug("Using cached rules for %r", self.room_id) + self.room_push_rule_cache_metrics.inc_hits() + defer.returnValue(self.rules_by_user) + with (yield self.linearizer.queue(())): if state_group and self.state_group == state_group: logger.debug("Using cached rules for %r", self.room_id) + self.room_push_rule_cache_metrics.inc_hits() defer.returnValue(self.rules_by_user) + self.room_push_rule_cache_metrics.inc_misses() + ret_rules_by_user = {} missing_member_event_ids = {} if state_group and self.state_group == context.prev_group: @@ -234,8 +302,13 @@ class RulesForRoom(object): # results. ret_rules_by_user = self.rules_by_user current_state_ids = context.delta_ids + + push_rules_delta_state_cache_metric.inc_hits() else: - current_state_ids = context.current_state_ids + current_state_ids = yield context.get_current_state_ids(self.store) + push_rules_delta_state_cache_metric.inc_misses() + + push_rules_state_size_counter.inc(len(current_state_ids)) logger.debug( "Looking for member changes in %r %r", state_group, current_state_ids @@ -282,6 +355,14 @@ class RulesForRoom(object): yield self._update_rules_with_member_event_ids( ret_rules_by_user, missing_member_event_ids, state_group, event ) + else: + # The push rules didn't change but lets update the cache anyway + self.update_cache( + self.sequence, + members={}, # There were no membership changes + rules_by_user=ret_rules_by_user, + state_group=state_group + ) if logger.isEnabledFor(logging.DEBUG): logger.debug( @@ -324,7 +405,7 @@ class RulesForRoom(object): # If the event is a join event then it will be in current state evnts # map but not in the DB, so we have to explicitly insert it. if event.type == EventTypes.Member: - for event_id in member_event_ids.itervalues(): + for event_id in itervalues(member_event_ids): if event_id == event.event_id: members[event_id] = (event.state_key, event.membership) @@ -332,7 +413,7 @@ class RulesForRoom(object): logger.debug("Found members %r: %r", self.room_id, members.values()) interested_in_user_ids = set( - user_id for user_id, membership in members.itervalues() + user_id for user_id, membership in itervalues(members) if membership == Membership.JOIN ) @@ -344,7 +425,7 @@ class RulesForRoom(object): ) user_ids = set( - uid for uid, have_pusher in if_users_with_pushers.iteritems() if have_pusher + uid for uid, have_pusher in iteritems(if_users_with_pushers) if have_pusher ) logger.debug("With pushers: %r", user_ids) @@ -365,7 +446,7 @@ class RulesForRoom(object): ) ret_rules_by_user.update( - item for item in rules_by_user.iteritems() if item[0] is not None + item for item in iteritems(rules_by_user) if item[0] is not None ) self.update_cache(sequence, members, ret_rules_by_user, state_group) @@ -380,6 +461,7 @@ class RulesForRoom(object): self.state_group = object() self.member_map = {} self.rules_by_user = {} + push_rules_invalidation_counter.inc() def update_cache(self, sequence, members, rules_by_user, state_group): if sequence == self.sequence: diff --git a/synapse/push/clientformat.py b/synapse/push/clientformat.py index e0331b2d2..ecbf364a5 100644 --- a/synapse/push/clientformat.py +++ b/synapse/push/clientformat.py @@ -13,12 +13,10 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.push.rulekinds import ( - PRIORITY_CLASS_MAP, PRIORITY_CLASS_INVERSE_MAP -) - import copy +from synapse.push.rulekinds import PRIORITY_CLASS_INVERSE_MAP, PRIORITY_CLASS_MAP + def format_push_rules_for_user(user, ruleslist): """Converts a list of rawrules and a enabled map into nested dictionaries diff --git a/synapse/push/emailpusher.py b/synapse/push/emailpusher.py index a69dda7b0..d74637142 100644 --- a/synapse/push/emailpusher.py +++ b/synapse/push/emailpusher.py @@ -13,14 +13,13 @@ # See the License for the specific language governing permissions and # limitations under the License. -from twisted.internet import defer, reactor -from twisted.internet.error import AlreadyCalled, AlreadyCancelled - import logging -from synapse.util.metrics import Measure -from synapse.util.logcontext import LoggingContext +from twisted.internet import defer +from twisted.internet.error import AlreadyCalled, AlreadyCancelled +from synapse.util.logcontext import LoggingContext +from synapse.util.metrics import Measure logger = logging.getLogger(__name__) @@ -77,10 +76,13 @@ class EmailPusher(object): @defer.inlineCallbacks def on_started(self): if self.mailer is not None: - self.throttle_params = yield self.store.get_throttle_params_by_room( - self.pusher_id - ) - yield self._process() + try: + self.throttle_params = yield self.store.get_throttle_params_by_room( + self.pusher_id + ) + yield self._process() + except Exception: + logger.exception("Error starting email pusher") def on_stop(self): if self.timed_call: @@ -121,7 +123,7 @@ class EmailPusher(object): starting_max_ordering = self.max_stream_ordering try: yield self._unsafe_process() - except: + except Exception: logger.exception("Exception processing notifs") if self.max_stream_ordering == starting_max_ordering: break @@ -196,7 +198,7 @@ class EmailPusher(object): self.timed_call = None if soonest_due_at is not None: - self.timed_call = reactor.callLater( + self.timed_call = self.hs.get_reactor().callLater( self.seconds_until(soonest_due_at), self.on_timer ) diff --git a/synapse/push/httppusher.py b/synapse/push/httppusher.py index 8a5d47310..81e18bcf7 100644 --- a/synapse/push/httppusher.py +++ b/synapse/push/httppusher.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2015, 2016 OpenMarket Ltd +# Copyright 2017 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -12,21 +13,25 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. +import logging -from synapse.push import PusherConfigException +from prometheus_client import Counter -from twisted.internet import defer, reactor +from twisted.internet import defer from twisted.internet.error import AlreadyCalled, AlreadyCancelled -import logging -import push_rule_evaluator -import push_tools - +from synapse.push import PusherConfigException from synapse.util.logcontext import LoggingContext from synapse.util.metrics import Measure +from . import push_rule_evaluator, push_tools + logger = logging.getLogger(__name__) +http_push_processed_counter = Counter("synapse_http_httppusher_http_pushes_processed", "") + +http_push_failed_counter = Counter("synapse_http_httppusher_http_pushes_failed", "") + class HttpPusher(object): INITIAL_BACKOFF_SEC = 1 # in seconds because that's what Twisted takes @@ -84,7 +89,10 @@ class HttpPusher(object): @defer.inlineCallbacks def on_started(self): - yield self._process() + try: + yield self._process() + except Exception: + logger.exception("Error starting http pusher") @defer.inlineCallbacks def on_new_notifications(self, min_stream_ordering, max_stream_ordering): @@ -131,7 +139,7 @@ class HttpPusher(object): starting_max_ordering = self.max_stream_ordering try: yield self._unsafe_process() - except: + except Exception: logger.exception("Exception processing notifs") if self.max_stream_ordering == starting_max_ordering: break @@ -151,9 +159,16 @@ class HttpPusher(object): self.user_id, self.last_stream_ordering, self.max_stream_ordering ) + logger.info( + "Processing %i unprocessed push actions for %s starting at " + "stream_ordering %s", + len(unprocessed), self.name, self.last_stream_ordering, + ) + for push_action in unprocessed: processed = yield self._process_one(push_action) if processed: + http_push_processed_counter.inc() self.backoff_delay = HttpPusher.INITIAL_BACKOFF_SEC self.last_stream_ordering = push_action['stream_ordering'] yield self.store.update_pusher_last_stream_ordering_and_success( @@ -168,6 +183,7 @@ class HttpPusher(object): self.failing_since ) else: + http_push_failed_counter.inc() if not self.failing_since: self.failing_since = self.clock.time_msec() yield self.store.update_pusher_failing_since( @@ -204,7 +220,9 @@ class HttpPusher(object): ) else: logger.info("Push failed: delaying for %ds", self.backoff_delay) - self.timed_call = reactor.callLater(self.backoff_delay, self.on_timer) + self.timed_call = self.hs.get_reactor().callLater( + self.backoff_delay, self.on_timer + ) self.backoff_delay = min(self.backoff_delay * 2, self.MAX_BACKOFF_SEC) break @@ -244,6 +262,26 @@ class HttpPusher(object): @defer.inlineCallbacks def _build_notification_dict(self, event, tweaks, badge): + if self.data.get('format') == 'event_id_only': + d = { + 'notification': { + 'event_id': event.event_id, + 'room_id': event.room_id, + 'counts': { + 'unread': badge, + }, + 'devices': [ + { + 'app_id': self.app_id, + 'pushkey': self.pushkey, + 'pushkey_ts': long(self.pushkey_ts / 1000), + 'data': self.data_minus_url, + } + ] + } + } + defer.returnValue(d) + ctx = yield push_tools.get_context_for_event( self.store, self.state_handler, event, self.user_id ) @@ -275,7 +313,7 @@ class HttpPusher(object): if event.type == 'm.room.member': d['notification']['membership'] = event.content['membership'] d['notification']['user_is_target'] = event.state_key == self.user_id - if not self.hs.config.push_redact_content and 'content' in event: + if self.hs.config.push_include_content and 'content' in event: d['notification']['content'] = event.content # We no longer send aliases separately, instead, we send the human @@ -294,8 +332,11 @@ class HttpPusher(object): defer.returnValue([]) try: resp = yield self.http_client.post_json_get_json(self.url, notification_dict) - except: - logger.warn("Failed to push %s ", self.url) + except Exception: + logger.warn( + "Failed to push event %s to %s", + event.event_id, self.name, exc_info=True, + ) defer.returnValue(False) rejected = [] if 'rejected' in resp: @@ -304,7 +345,7 @@ class HttpPusher(object): @defer.inlineCallbacks def _send_badge(self, badge): - logger.info("Sending updated badge count %d to %r", badge, self.user_id) + logger.info("Sending updated badge count %d to %s", badge, self.name) d = { 'notification': { 'id': '', @@ -325,8 +366,11 @@ class HttpPusher(object): } try: resp = yield self.http_client.post_json_get_json(self.url, d) - except: - logger.exception("Failed to push %s ", self.url) + except Exception: + logger.warn( + "Failed to send badge count to %s", + self.name, exc_info=True, + ) defer.returnValue(False) rejected = [] if 'rejected' in resp: diff --git a/synapse/push/mailer.py b/synapse/push/mailer.py index b5cd9b426..9d601208f 100644 --- a/synapse/push/mailer.py +++ b/synapse/push/mailer.py @@ -13,30 +13,31 @@ # See the License for the specific language governing permissions and # limitations under the License. +import email.mime.multipart +import email.utils +import logging +import time +import urllib +from email.mime.multipart import MIMEMultipart +from email.mime.text import MIMEText + +import bleach +import jinja2 + from twisted.internet import defer from twisted.mail.smtp import sendmail -import email.utils -import email.mime.multipart -from email.mime.text import MIMEText -from email.mime.multipart import MIMEMultipart - -from synapse.util.async import concurrently_execute +from synapse.api.constants import EventTypes +from synapse.api.errors import StoreError from synapse.push.presentable_names import ( - calculate_room_name, name_from_member_event, descriptor_from_member_events + calculate_room_name, + descriptor_from_member_events, + name_from_member_event, ) from synapse.types import UserID -from synapse.api.errors import StoreError -from synapse.api.constants import EventTypes +from synapse.util.async import concurrently_execute from synapse.visibility import filter_events_for_client -import jinja2 -import bleach - -import time -import urllib - -import logging logger = logging.getLogger(__name__) @@ -229,7 +230,8 @@ class Mailer(object): if room_vars['notifs'] and 'messages' in room_vars['notifs'][-1]: prev_messages = room_vars['notifs'][-1]['messages'] for message in notifvars['messages']: - pm = filter(lambda pm: pm['id'] == message['id'], prev_messages) + pm = list(filter(lambda pm: pm['id'] == message['id'], + prev_messages)) if pm: if not message["is_historical"]: pm[0]["is_historical"] = False diff --git a/synapse/push/presentable_names.py b/synapse/push/presentable_names.py index 277da3cd3..eef6e18c2 100644 --- a/synapse/push/presentable_names.py +++ b/synapse/push/presentable_names.py @@ -13,10 +13,10 @@ # See the License for the specific language governing permissions and # limitations under the License. -from twisted.internet import defer - -import re import logging +import re + +from twisted.internet import defer logger = logging.getLogger(__name__) @@ -113,7 +113,7 @@ def calculate_room_name(store, room_state_ids, user_id, fallback_to_members=True # so find out who is in the room that isn't the user. if "m.room.member" in room_state_bytype_ids: member_events = yield store.get_events( - room_state_bytype_ids["m.room.member"].values() + list(room_state_bytype_ids["m.room.member"].values()) ) all_members = [ ev for ev in member_events.values() diff --git a/synapse/push/push_rule_evaluator.py b/synapse/push/push_rule_evaluator.py index 172c27c13..2bd321d53 100644 --- a/synapse/push/push_rule_evaluator.py +++ b/synapse/push/push_rule_evaluator.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2015, 2016 OpenMarket Ltd +# Copyright 2017 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -16,6 +17,8 @@ import logging import re +from six import string_types + from synapse.types import UserID from synapse.util.caches import CACHE_SIZE_FACTOR, register_cache from synapse.util.caches.lrucache import LruCache @@ -29,6 +32,21 @@ INEQUALITY_EXPR = re.compile("^([=<>]*)([0-9]*)$") def _room_member_count(ev, condition, room_member_count): + return _test_ineq_condition(condition, room_member_count) + + +def _sender_notification_permission(ev, condition, sender_power_level, power_levels): + notif_level_key = condition.get('key') + if notif_level_key is None: + return False + + notif_levels = power_levels.get('notifications', {}) + room_notif_level = notif_levels.get(notif_level_key, 50) + + return sender_power_level >= room_notif_level + + +def _test_ineq_condition(condition, number): if 'is' not in condition: return False m = INEQUALITY_EXPR.match(condition['is']) @@ -41,15 +59,15 @@ def _room_member_count(ev, condition, room_member_count): rhs = int(rhs) if ineq == '' or ineq == '==': - return room_member_count == rhs + return number == rhs elif ineq == '<': - return room_member_count < rhs + return number < rhs elif ineq == '>': - return room_member_count > rhs + return number > rhs elif ineq == '>=': - return room_member_count >= rhs + return number >= rhs elif ineq == '<=': - return room_member_count <= rhs + return number <= rhs else: return False @@ -65,9 +83,11 @@ def tweaks_for_actions(actions): class PushRuleEvaluatorForEvent(object): - def __init__(self, event, room_member_count): + def __init__(self, event, room_member_count, sender_power_level, power_levels): self._event = event self._room_member_count = room_member_count + self._sender_power_level = sender_power_level + self._power_levels = power_levels # Maps strings of e.g. 'content.body' -> event["content"]["body"] self._value_cache = _flatten_dict(event) @@ -81,6 +101,10 @@ class PushRuleEvaluatorForEvent(object): return _room_member_count( self._event, condition, self._room_member_count ) + elif condition['kind'] == 'sender_notification_permission': + return _sender_notification_permission( + self._event, condition, self._sender_power_level, self._power_levels, + ) else: return True @@ -128,7 +152,7 @@ class PushRuleEvaluatorForEvent(object): # Caches (glob, word_boundary) -> regex for push. See _glob_matches regex_cache = LruCache(50000 * CACHE_SIZE_FACTOR) -register_cache("regex_push_cache", regex_cache) +register_cache("cache", "regex_push_cache", regex_cache) def _glob_matches(glob, value, word_boundary=False): @@ -183,7 +207,7 @@ def _glob_to_re(glob, word_boundary): r, ) if word_boundary: - r = r"\b%s\b" % (r,) + r = _re_word_boundary(r) return re.compile(r, flags=re.IGNORECASE) else: @@ -192,7 +216,7 @@ def _glob_to_re(glob, word_boundary): return re.compile(r, flags=re.IGNORECASE) elif word_boundary: r = re.escape(glob) - r = r"\b%s\b" % (r,) + r = _re_word_boundary(r) return re.compile(r, flags=re.IGNORECASE) else: @@ -200,11 +224,23 @@ def _glob_to_re(glob, word_boundary): return re.compile(r, flags=re.IGNORECASE) +def _re_word_boundary(r): + """ + Adds word boundary characters to the start and end of an + expression to require that the match occur as a whole word, + but do so respecting the fact that strings starting or ending + with non-word characters will change word boundaries. + """ + # we can't use \b as it chokes on unicode. however \W seems to be okay + # as shorthand for [^0-9A-Za-z_]. + return r"(^|\W)%s(\W|$)" % (r,) + + def _flatten_dict(d, prefix=[], result=None): if result is None: result = {} for key, value in d.items(): - if isinstance(value, basestring): + if isinstance(value, string_types): result[".".join(prefix + [key])] = value.lower() elif hasattr(value, "items"): _flatten_dict(value, prefix=(prefix + [key]), result=result) diff --git a/synapse/push/push_tools.py b/synapse/push/push_tools.py index 6835f54e9..8049c298c 100644 --- a/synapse/push/push_tools.py +++ b/synapse/push/push_tools.py @@ -14,9 +14,8 @@ # limitations under the License. from twisted.internet import defer -from synapse.push.presentable_names import ( - calculate_room_name, name_from_member_event -) + +from synapse.push.presentable_names import calculate_room_name, name_from_member_event @defer.inlineCallbacks diff --git a/synapse/push/pusher.py b/synapse/push/pusher.py index 491f27bde..fcee6d9d7 100644 --- a/synapse/push/pusher.py +++ b/synapse/push/pusher.py @@ -13,9 +13,10 @@ # See the License for the specific language governing permissions and # limitations under the License. -from httppusher import HttpPusher - import logging + +from .httppusher import HttpPusher + logger = logging.getLogger(__name__) # We try importing this if we can (it will fail if we don't @@ -27,7 +28,7 @@ logger = logging.getLogger(__name__) try: from synapse.push.emailpusher import EmailPusher from synapse.push.mailer import Mailer, load_jinja2_templates -except: +except Exception: pass diff --git a/synapse/push/pusherpool.py b/synapse/push/pusherpool.py index 43cb6e9c0..36bb5bbc6 100644 --- a/synapse/push/pusherpool.py +++ b/synapse/push/pusherpool.py @@ -14,13 +14,12 @@ # See the License for the specific language governing permissions and # limitations under the License. +import logging + from twisted.internet import defer -from .pusher import PusherFactory -from synapse.util.logcontext import preserve_fn, preserve_context_over_deferred -from synapse.util.async import run_on_reactor - -import logging +from synapse.push.pusher import PusherFactory +from synapse.util.logcontext import make_deferred_yieldable, run_in_background logger = logging.getLogger(__name__) @@ -103,23 +102,28 @@ class PusherPool: yield self.remove_pusher(p['app_id'], p['pushkey'], p['user_name']) @defer.inlineCallbacks - def remove_pushers_by_user(self, user_id, except_access_token_id=None): - all = yield self.store.get_all_pushers() - logger.info( - "Removing all pushers for user %s except access tokens id %r", - user_id, except_access_token_id - ) - for p in all: - if p['user_name'] == user_id and p['access_token'] != except_access_token_id: + def remove_pushers_by_access_token(self, user_id, access_tokens): + """Remove the pushers for a given user corresponding to a set of + access_tokens. + + Args: + user_id (str): user to remove pushers for + access_tokens (Iterable[int]): access token *ids* to remove pushers + for + """ + tokens = set(access_tokens) + for p in (yield self.store.get_pushers_by_user_id(user_id)): + if p['access_token'] in tokens: logger.info( "Removing pusher for app id %s, pushkey %s, user %s", p['app_id'], p['pushkey'], p['user_name'] ) - yield self.remove_pusher(p['app_id'], p['pushkey'], p['user_name']) + yield self.remove_pusher( + p['app_id'], p['pushkey'], p['user_name'], + ) @defer.inlineCallbacks def on_new_notifications(self, min_stream_id, max_stream_id): - yield run_on_reactor() try: users_affected = yield self.store.get_push_action_users_in_range( min_stream_id, max_stream_id @@ -131,18 +135,20 @@ class PusherPool: if u in self.pushers: for p in self.pushers[u].values(): deferreds.append( - preserve_fn(p.on_new_notifications)( - min_stream_id, max_stream_id + run_in_background( + p.on_new_notifications, + min_stream_id, max_stream_id, ) ) - yield preserve_context_over_deferred(defer.gatherResults(deferreds)) - except: + yield make_deferred_yieldable( + defer.gatherResults(deferreds, consumeErrors=True), + ) + except Exception: logger.exception("Exception in pusher on_new_notifications") @defer.inlineCallbacks def on_new_receipts(self, min_stream_id, max_stream_id, affected_room_ids): - yield run_on_reactor() try: # Need to subtract 1 from the minimum because the lower bound here # is not inclusive @@ -158,11 +164,16 @@ class PusherPool: if u in self.pushers: for p in self.pushers[u].values(): deferreds.append( - preserve_fn(p.on_new_receipts)(min_stream_id, max_stream_id) + run_in_background( + p.on_new_receipts, + min_stream_id, max_stream_id, + ) ) - yield preserve_context_over_deferred(defer.gatherResults(deferreds)) - except: + yield make_deferred_yieldable( + defer.gatherResults(deferreds, consumeErrors=True), + ) + except Exception: logger.exception("Exception in pusher on_new_receipts") @defer.inlineCallbacks @@ -188,7 +199,7 @@ class PusherPool: for pusherdict in pushers: try: p = self.pusher_factory.create_pusher(pusherdict) - except: + except Exception: logger.exception("Couldn't start a pusher: caught Exception") continue if p: @@ -201,7 +212,7 @@ class PusherPool: if appid_pushkey in byuser: byuser[appid_pushkey].on_stop() byuser[appid_pushkey] = p - preserve_fn(p.on_started)() + run_in_background(p.on_started) logger.info("Started pushers") diff --git a/synapse/python_dependencies.py b/synapse/python_dependencies.py index ed7f1c89a..987eec3ef 100644 --- a/synapse/python_dependencies.py +++ b/synapse/python_dependencies.py @@ -1,5 +1,6 @@ # Copyright 2015, 2016 OpenMarket Ltd # Copyright 2017 Vector Creations Ltd +# Copyright 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -18,36 +19,52 @@ from distutils.version import LooseVersion logger = logging.getLogger(__name__) +# this dict maps from python package name to a list of modules we expect it to +# provide. +# +# the key is a "requirement specifier", as used as a parameter to `pip +# install`[1], or an `install_requires` argument to `setuptools.setup` [2]. +# +# the value is a sequence of strings; each entry should be the name of the +# python module, optionally followed by a version assertion which can be either +# ">=" or "==". +# +# [1] https://pip.pypa.io/en/stable/reference/pip_install/#requirement-specifiers. +# [2] https://setuptools.readthedocs.io/en/latest/setuptools.html#declaring-dependencies REQUIREMENTS = { "jsonschema>=2.5.1": ["jsonschema>=2.5.1"], "frozendict>=0.4": ["frozendict"], "unpaddedbase64>=1.1.0": ["unpaddedbase64>=1.1.0"], - "canonicaljson>=1.0.0": ["canonicaljson>=1.0.0"], + "canonicaljson>=1.1.3": ["canonicaljson>=1.1.3"], "signedjson>=1.0.0": ["signedjson>=1.0.0"], - "pynacl==0.3.0": ["nacl==0.3.0", "nacl.bindings"], + "pynacl>=1.2.1": ["nacl>=1.2.1", "nacl.bindings"], "service_identity>=1.0.0": ["service_identity>=1.0.0"], "Twisted>=16.0.0": ["twisted>=16.0.0"], - "pyopenssl>=0.14": ["OpenSSL>=0.14"], + + # We use crypto.get_elliptic_curve which is only supported in >=0.15 + "pyopenssl>=0.15": ["OpenSSL>=0.15"], + "pyyaml": ["yaml"], "pyasn1": ["pyasn1"], "daemonize": ["daemonize"], - "py-bcrypt": ["bcrypt"], + "bcrypt": ["bcrypt>=3.1.0"], "pillow": ["PIL"], "pydenticon": ["pydenticon"], - "ujson": ["ujson"], - "blist": ["blist"], - "pysaml2>=3.0.0,<4.0.0": ["saml2>=3.0.0,<4.0.0"], + "sortedcontainers": ["sortedcontainers"], + "pysaml2>=3.0.0": ["saml2>=3.0.0"], "pymacaroons-pynacl": ["pymacaroons"], "msgpack-python>=0.3.0": ["msgpack"], "phonenumbers>=8.2.0": ["phonenumbers"], + "six": ["six"], + "prometheus_client": ["prometheus_client"], + "attrs": ["attr"], + "netaddr>=0.7.18": ["netaddr"], } + CONDITIONAL_REQUIREMENTS = { "web_client": { "matrix_angular_sdk>=0.6.8": ["syweb>=0.6.8"], }, - "preview_url": { - "netaddr>=0.7.18": ["netaddr"], - }, "email.enable_notifs": { "Jinja2>=2.8": ["Jinja2>=2.8"], "bleach>=1.4.2": ["bleach>=1.4.2"], @@ -58,6 +75,9 @@ CONDITIONAL_REQUIREMENTS = { "psutil": { "psutil>=2.0.0": ["psutil>=2.0.0"], }, + "affinity": { + "affinity": ["affinity"], + }, } diff --git a/synapse/replication/http/__init__.py b/synapse/replication/http/__init__.py new file mode 100644 index 000000000..589ee94c6 --- /dev/null +++ b/synapse/replication/http/__init__.py @@ -0,0 +1,29 @@ +# -*- coding: utf-8 -*- +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from synapse.http.server import JsonResource +from synapse.replication.http import membership, send_event + +REPLICATION_PREFIX = "/_synapse/replication" + + +class ReplicationRestResource(JsonResource): + def __init__(self, hs): + JsonResource.__init__(self, hs, canonical_json=False) + self.register_servlets(hs) + + def register_servlets(self, hs): + send_event.register_servlets(hs, self) + membership.register_servlets(hs, self) diff --git a/synapse/replication/http/membership.py b/synapse/replication/http/membership.py new file mode 100644 index 000000000..6bfc8a5b8 --- /dev/null +++ b/synapse/replication/http/membership.py @@ -0,0 +1,334 @@ +# -*- coding: utf-8 -*- +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging +import re + +from twisted.internet import defer + +from synapse.api.errors import MatrixCodeMessageException, SynapseError +from synapse.http.servlet import RestServlet, parse_json_object_from_request +from synapse.types import Requester, UserID +from synapse.util.distributor import user_joined_room, user_left_room + +logger = logging.getLogger(__name__) + + +@defer.inlineCallbacks +def remote_join(client, host, port, requester, remote_room_hosts, + room_id, user_id, content): + """Ask the master to do a remote join for the given user to the given room + + Args: + client (SimpleHttpClient) + host (str): host of master + port (int): port on master listening for HTTP replication + requester (Requester) + remote_room_hosts (list[str]): Servers to try and join via + room_id (str) + user_id (str) + content (dict): The event content to use for the join event + + Returns: + Deferred + """ + uri = "http://%s:%s/_synapse/replication/remote_join" % (host, port) + + payload = { + "requester": requester.serialize(), + "remote_room_hosts": remote_room_hosts, + "room_id": room_id, + "user_id": user_id, + "content": content, + } + + try: + result = yield client.post_json_get_json(uri, payload) + except MatrixCodeMessageException as e: + # We convert to SynapseError as we know that it was a SynapseError + # on the master process that we should send to the client. (And + # importantly, not stack traces everywhere) + raise SynapseError(e.code, e.msg, e.errcode) + defer.returnValue(result) + + +@defer.inlineCallbacks +def remote_reject_invite(client, host, port, requester, remote_room_hosts, + room_id, user_id): + """Ask master to reject the invite for the user and room. + + Args: + client (SimpleHttpClient) + host (str): host of master + port (int): port on master listening for HTTP replication + requester (Requester) + remote_room_hosts (list[str]): Servers to try and reject via + room_id (str) + user_id (str) + + Returns: + Deferred + """ + uri = "http://%s:%s/_synapse/replication/remote_reject_invite" % (host, port) + + payload = { + "requester": requester.serialize(), + "remote_room_hosts": remote_room_hosts, + "room_id": room_id, + "user_id": user_id, + } + + try: + result = yield client.post_json_get_json(uri, payload) + except MatrixCodeMessageException as e: + # We convert to SynapseError as we know that it was a SynapseError + # on the master process that we should send to the client. (And + # importantly, not stack traces everywhere) + raise SynapseError(e.code, e.msg, e.errcode) + defer.returnValue(result) + + +@defer.inlineCallbacks +def get_or_register_3pid_guest(client, host, port, requester, + medium, address, inviter_user_id): + """Ask the master to get/create a guest account for given 3PID. + + Args: + client (SimpleHttpClient) + host (str): host of master + port (int): port on master listening for HTTP replication + requester (Requester) + medium (str) + address (str) + inviter_user_id (str): The user ID who is trying to invite the + 3PID + + Returns: + Deferred[(str, str)]: A 2-tuple of `(user_id, access_token)` of the + 3PID guest account. + """ + + uri = "http://%s:%s/_synapse/replication/get_or_register_3pid_guest" % (host, port) + + payload = { + "requester": requester.serialize(), + "medium": medium, + "address": address, + "inviter_user_id": inviter_user_id, + } + + try: + result = yield client.post_json_get_json(uri, payload) + except MatrixCodeMessageException as e: + # We convert to SynapseError as we know that it was a SynapseError + # on the master process that we should send to the client. (And + # importantly, not stack traces everywhere) + raise SynapseError(e.code, e.msg, e.errcode) + defer.returnValue(result) + + +@defer.inlineCallbacks +def notify_user_membership_change(client, host, port, user_id, room_id, change): + """Notify master that a user has joined or left the room + + Args: + client (SimpleHttpClient) + host (str): host of master + port (int): port on master listening for HTTP replication. + user_id (str) + room_id (str) + change (str): Either "join" or "left" + + Returns: + Deferred + """ + assert change in ("joined", "left") + + uri = "http://%s:%s/_synapse/replication/user_%s_room" % (host, port, change) + + payload = { + "user_id": user_id, + "room_id": room_id, + } + + try: + result = yield client.post_json_get_json(uri, payload) + except MatrixCodeMessageException as e: + # We convert to SynapseError as we know that it was a SynapseError + # on the master process that we should send to the client. (And + # importantly, not stack traces everywhere) + raise SynapseError(e.code, e.msg, e.errcode) + defer.returnValue(result) + + +class ReplicationRemoteJoinRestServlet(RestServlet): + PATTERNS = [re.compile("^/_synapse/replication/remote_join$")] + + def __init__(self, hs): + super(ReplicationRemoteJoinRestServlet, self).__init__() + + self.federation_handler = hs.get_handlers().federation_handler + self.store = hs.get_datastore() + self.clock = hs.get_clock() + + @defer.inlineCallbacks + def on_POST(self, request): + content = parse_json_object_from_request(request) + + remote_room_hosts = content["remote_room_hosts"] + room_id = content["room_id"] + user_id = content["user_id"] + event_content = content["content"] + + requester = Requester.deserialize(self.store, content["requester"]) + + if requester.user: + request.authenticated_entity = requester.user.to_string() + + logger.info( + "remote_join: %s into room: %s", + user_id, room_id, + ) + + yield self.federation_handler.do_invite_join( + remote_room_hosts, + room_id, + user_id, + event_content, + ) + + defer.returnValue((200, {})) + + +class ReplicationRemoteRejectInviteRestServlet(RestServlet): + PATTERNS = [re.compile("^/_synapse/replication/remote_reject_invite$")] + + def __init__(self, hs): + super(ReplicationRemoteRejectInviteRestServlet, self).__init__() + + self.federation_handler = hs.get_handlers().federation_handler + self.store = hs.get_datastore() + self.clock = hs.get_clock() + + @defer.inlineCallbacks + def on_POST(self, request): + content = parse_json_object_from_request(request) + + remote_room_hosts = content["remote_room_hosts"] + room_id = content["room_id"] + user_id = content["user_id"] + + requester = Requester.deserialize(self.store, content["requester"]) + + if requester.user: + request.authenticated_entity = requester.user.to_string() + + logger.info( + "remote_reject_invite: %s out of room: %s", + user_id, room_id, + ) + + try: + event = yield self.federation_handler.do_remotely_reject_invite( + remote_room_hosts, + room_id, + user_id, + ) + ret = event.get_pdu_json() + except Exception as e: + # if we were unable to reject the exception, just mark + # it as rejected on our end and plough ahead. + # + # The 'except' clause is very broad, but we need to + # capture everything from DNS failures upwards + # + logger.warn("Failed to reject invite: %s", e) + + yield self.store.locally_reject_invite( + user_id, room_id + ) + ret = {} + + defer.returnValue((200, ret)) + + +class ReplicationRegister3PIDGuestRestServlet(RestServlet): + PATTERNS = [re.compile("^/_synapse/replication/get_or_register_3pid_guest$")] + + def __init__(self, hs): + super(ReplicationRegister3PIDGuestRestServlet, self).__init__() + + self.registeration_handler = hs.get_handlers().registration_handler + self.store = hs.get_datastore() + self.clock = hs.get_clock() + + @defer.inlineCallbacks + def on_POST(self, request): + content = parse_json_object_from_request(request) + + medium = content["medium"] + address = content["address"] + inviter_user_id = content["inviter_user_id"] + + requester = Requester.deserialize(self.store, content["requester"]) + + if requester.user: + request.authenticated_entity = requester.user.to_string() + + logger.info("get_or_register_3pid_guest: %r", content) + + ret = yield self.registeration_handler.get_or_register_3pid_guest( + medium, address, inviter_user_id, + ) + + defer.returnValue((200, ret)) + + +class ReplicationUserJoinedLeftRoomRestServlet(RestServlet): + PATTERNS = [re.compile("^/_synapse/replication/user_(?Pjoined|left)_room$")] + + def __init__(self, hs): + super(ReplicationUserJoinedLeftRoomRestServlet, self).__init__() + + self.registeration_handler = hs.get_handlers().registration_handler + self.store = hs.get_datastore() + self.clock = hs.get_clock() + self.distributor = hs.get_distributor() + + def on_POST(self, request, change): + content = parse_json_object_from_request(request) + + user_id = content["user_id"] + room_id = content["room_id"] + + logger.info("user membership change: %s in %s", user_id, room_id) + + user = UserID.from_string(user_id) + + if change == "joined": + user_joined_room(self.distributor, user, room_id) + elif change == "left": + user_left_room(self.distributor, user, room_id) + else: + raise Exception("Unrecognized change: %r", change) + + return (200, {}) + + +def register_servlets(hs, http_server): + ReplicationRemoteJoinRestServlet(hs).register(http_server) + ReplicationRemoteRejectInviteRestServlet(hs).register(http_server) + ReplicationRegister3PIDGuestRestServlet(hs).register(http_server) + ReplicationUserJoinedLeftRoomRestServlet(hs).register(http_server) diff --git a/synapse/replication/http/send_event.py b/synapse/replication/http/send_event.py new file mode 100644 index 000000000..5227bc333 --- /dev/null +++ b/synapse/replication/http/send_event.py @@ -0,0 +1,165 @@ +# -*- coding: utf-8 -*- +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging +import re + +from twisted.internet import defer + +from synapse.api.errors import ( + CodeMessageException, + MatrixCodeMessageException, + SynapseError, +) +from synapse.events import FrozenEvent +from synapse.events.snapshot import EventContext +from synapse.http.servlet import RestServlet, parse_json_object_from_request +from synapse.types import Requester, UserID +from synapse.util.caches.response_cache import ResponseCache +from synapse.util.metrics import Measure + +logger = logging.getLogger(__name__) + + +@defer.inlineCallbacks +def send_event_to_master(clock, store, client, host, port, requester, event, context, + ratelimit, extra_users): + """Send event to be handled on the master + + Args: + clock (synapse.util.Clock) + store (DataStore) + client (SimpleHttpClient) + host (str): host of master + port (int): port on master listening for HTTP replication + requester (Requester) + event (FrozenEvent) + context (EventContext) + ratelimit (bool) + extra_users (list(UserID)): Any extra users to notify about event + """ + uri = "http://%s:%s/_synapse/replication/send_event/%s" % ( + host, port, event.event_id, + ) + + serialized_context = yield context.serialize(event, store) + + payload = { + "event": event.get_pdu_json(), + "internal_metadata": event.internal_metadata.get_dict(), + "rejected_reason": event.rejected_reason, + "context": serialized_context, + "requester": requester.serialize(), + "ratelimit": ratelimit, + "extra_users": [u.to_string() for u in extra_users], + } + + try: + # We keep retrying the same request for timeouts. This is so that we + # have a good idea that the request has either succeeded or failed on + # the master, and so whether we should clean up or not. + while True: + try: + result = yield client.put_json(uri, payload) + break + except CodeMessageException as e: + if e.code != 504: + raise + + logger.warn("send_event request timed out") + + # If we timed out we probably don't need to worry about backing + # off too much, but lets just wait a little anyway. + yield clock.sleep(1) + except MatrixCodeMessageException as e: + # We convert to SynapseError as we know that it was a SynapseError + # on the master process that we should send to the client. (And + # importantly, not stack traces everywhere) + raise SynapseError(e.code, e.msg, e.errcode) + defer.returnValue(result) + + +class ReplicationSendEventRestServlet(RestServlet): + """Handles events newly created on workers, including persisting and + notifying. + + The API looks like: + + POST /_synapse/replication/send_event/:event_id + + { + "event": { .. serialized event .. }, + "internal_metadata": { .. serialized internal_metadata .. }, + "rejected_reason": .., // The event.rejected_reason field + "context": { .. serialized event context .. }, + "requester": { .. serialized requester .. }, + "ratelimit": true, + "extra_users": [], + } + """ + PATTERNS = [re.compile("^/_synapse/replication/send_event/(?P[^/]+)$")] + + def __init__(self, hs): + super(ReplicationSendEventRestServlet, self).__init__() + + self.event_creation_handler = hs.get_event_creation_handler() + self.store = hs.get_datastore() + self.clock = hs.get_clock() + + # The responses are tiny, so we may as well cache them for a while + self.response_cache = ResponseCache(hs, "send_event", timeout_ms=30 * 60 * 1000) + + def on_PUT(self, request, event_id): + return self.response_cache.wrap( + event_id, + self._handle_request, + request + ) + + @defer.inlineCallbacks + def _handle_request(self, request): + with Measure(self.clock, "repl_send_event_parse"): + content = parse_json_object_from_request(request) + + event_dict = content["event"] + internal_metadata = content["internal_metadata"] + rejected_reason = content["rejected_reason"] + event = FrozenEvent(event_dict, internal_metadata, rejected_reason) + + requester = Requester.deserialize(self.store, content["requester"]) + context = yield EventContext.deserialize(self.store, content["context"]) + + ratelimit = content["ratelimit"] + extra_users = [UserID.from_string(u) for u in content["extra_users"]] + + if requester.user: + request.authenticated_entity = requester.user.to_string() + + logger.info( + "Got event to send with ID: %s into room: %s", + event.event_id, event.room_id, + ) + + yield self.event_creation_handler.persist_and_notify_client_event( + requester, event, context, + ratelimit=ratelimit, + extra_users=extra_users, + ) + + defer.returnValue((200, {})) + + +def register_servlets(hs, http_server): + ReplicationSendEventRestServlet(hs).register(http_server) diff --git a/synapse/replication/slave/storage/_base.py b/synapse/replication/slave/storage/_base.py index b96264116..3f7be74e0 100644 --- a/synapse/replication/slave/storage/_base.py +++ b/synapse/replication/slave/storage/_base.py @@ -13,19 +13,19 @@ # See the License for the specific language governing permissions and # limitations under the License. +import logging + from synapse.storage._base import SQLBaseStore from synapse.storage.engines import PostgresEngine from ._slaved_id_tracker import SlavedIdTracker -import logging - logger = logging.getLogger(__name__) class BaseSlavedStore(SQLBaseStore): def __init__(self, db_conn, hs): - super(BaseSlavedStore, self).__init__(hs) + super(BaseSlavedStore, self).__init__(db_conn, hs) if isinstance(self.database_engine, PostgresEngine): self._cache_id_gen = SlavedIdTracker( db_conn, "cache_invalidation_stream", "stream_id", diff --git a/synapse/replication/slave/storage/account_data.py b/synapse/replication/slave/storage/account_data.py index efbd87918..d9ba6d69b 100644 --- a/synapse/replication/slave/storage/account_data.py +++ b/synapse/replication/slave/storage/account_data.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -13,50 +14,20 @@ # See the License for the specific language governing permissions and # limitations under the License. -from ._base import BaseSlavedStore -from ._slaved_id_tracker import SlavedIdTracker -from synapse.storage import DataStore -from synapse.storage.account_data import AccountDataStore -from synapse.storage.tags import TagsStore -from synapse.util.caches.stream_change_cache import StreamChangeCache +from synapse.replication.slave.storage._base import BaseSlavedStore +from synapse.replication.slave.storage._slaved_id_tracker import SlavedIdTracker +from synapse.storage.account_data import AccountDataWorkerStore +from synapse.storage.tags import TagsWorkerStore -class SlavedAccountDataStore(BaseSlavedStore): +class SlavedAccountDataStore(TagsWorkerStore, AccountDataWorkerStore, BaseSlavedStore): def __init__(self, db_conn, hs): - super(SlavedAccountDataStore, self).__init__(db_conn, hs) self._account_data_id_gen = SlavedIdTracker( db_conn, "account_data_max_stream_id", "stream_id", ) - self._account_data_stream_cache = StreamChangeCache( - "AccountDataAndTagsChangeCache", - self._account_data_id_gen.get_current_token(), - ) - get_account_data_for_user = ( - AccountDataStore.__dict__["get_account_data_for_user"] - ) - - get_global_account_data_by_type_for_users = ( - AccountDataStore.__dict__["get_global_account_data_by_type_for_users"] - ) - - get_global_account_data_by_type_for_user = ( - AccountDataStore.__dict__["get_global_account_data_by_type_for_user"] - ) - - get_tags_for_user = TagsStore.__dict__["get_tags_for_user"] - get_tags_for_room = ( - DataStore.get_tags_for_room.__func__ - ) - get_account_data_for_room = ( - DataStore.get_account_data_for_room.__func__ - ) - - get_updated_tags = DataStore.get_updated_tags.__func__ - get_updated_account_data_for_user = ( - DataStore.get_updated_account_data_for_user.__func__ - ) + super(SlavedAccountDataStore, self).__init__(db_conn, hs) def get_max_account_data_stream_id(self): return self._account_data_id_gen.get_current_token() @@ -85,6 +56,10 @@ class SlavedAccountDataStore(BaseSlavedStore): (row.data_type, row.user_id,) ) self.get_account_data_for_user.invalidate((row.user_id,)) + self.get_account_data_for_room.invalidate((row.user_id, row.room_id,)) + self.get_account_data_for_room_and_type.invalidate( + (row.user_id, row.room_id, row.data_type,), + ) self._account_data_stream_cache.entity_has_changed( row.user_id, token ) diff --git a/synapse/replication/slave/storage/appservice.py b/synapse/replication/slave/storage/appservice.py index 0d3f31a50..b53a4c6bd 100644 --- a/synapse/replication/slave/storage/appservice.py +++ b/synapse/replication/slave/storage/appservice.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2015, 2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -13,33 +14,12 @@ # See the License for the specific language governing permissions and # limitations under the License. -from ._base import BaseSlavedStore -from synapse.storage import DataStore -from synapse.config.appservice import load_appservices -from synapse.storage.appservice import _make_exclusive_regex +from synapse.storage.appservice import ( + ApplicationServiceTransactionWorkerStore, + ApplicationServiceWorkerStore, +) -class SlavedApplicationServiceStore(BaseSlavedStore): - def __init__(self, db_conn, hs): - super(SlavedApplicationServiceStore, self).__init__(db_conn, hs) - self.services_cache = load_appservices( - hs.config.server_name, - hs.config.app_service_config_files - ) - self.exclusive_user_regex = _make_exclusive_regex(self.services_cache) - - get_app_service_by_token = DataStore.get_app_service_by_token.__func__ - get_app_service_by_user_id = DataStore.get_app_service_by_user_id.__func__ - get_app_services = DataStore.get_app_services.__func__ - get_new_events_for_appservice = DataStore.get_new_events_for_appservice.__func__ - create_appservice_txn = DataStore.create_appservice_txn.__func__ - get_appservices_by_state = DataStore.get_appservices_by_state.__func__ - get_oldest_unsent_txn = DataStore.get_oldest_unsent_txn.__func__ - _get_last_txn = DataStore._get_last_txn.__func__ - complete_appservice_txn = DataStore.complete_appservice_txn.__func__ - get_appservice_state = DataStore.get_appservice_state.__func__ - set_appservice_last_pos = DataStore.set_appservice_last_pos.__func__ - set_appservice_state = DataStore.set_appservice_state.__func__ - get_if_app_services_interested_in_user = ( - DataStore.get_if_app_services_interested_in_user.__func__ - ) +class SlavedApplicationServiceStore(ApplicationServiceTransactionWorkerStore, + ApplicationServiceWorkerStore): + pass diff --git a/synapse/replication/slave/storage/client_ips.py b/synapse/replication/slave/storage/client_ips.py index 65250285e..60641f1a4 100644 --- a/synapse/replication/slave/storage/client_ips.py +++ b/synapse/replication/slave/storage/client_ips.py @@ -13,11 +13,12 @@ # See the License for the specific language governing permissions and # limitations under the License. -from ._base import BaseSlavedStore from synapse.storage.client_ips import LAST_SEEN_GRANULARITY from synapse.util.caches import CACHE_SIZE_FACTOR from synapse.util.caches.descriptors import Cache +from ._base import BaseSlavedStore + class SlavedClientIpStore(BaseSlavedStore): def __init__(self, db_conn, hs): @@ -29,9 +30,8 @@ class SlavedClientIpStore(BaseSlavedStore): max_entries=50000 * CACHE_SIZE_FACTOR, ) - def insert_client_ip(self, user, access_token, ip, user_agent, device_id): + def insert_client_ip(self, user_id, access_token, ip, user_agent, device_id): now = int(self._clock.time_msec()) - user_id = user.to_string() key = (user_id, access_token, ip) try: diff --git a/synapse/replication/slave/storage/deviceinbox.py b/synapse/replication/slave/storage/deviceinbox.py index 6f3fb6477..87eaa5300 100644 --- a/synapse/replication/slave/storage/deviceinbox.py +++ b/synapse/replication/slave/storage/deviceinbox.py @@ -13,11 +13,12 @@ # See the License for the specific language governing permissions and # limitations under the License. +from synapse.storage import DataStore +from synapse.util.caches.expiringcache import ExpiringCache +from synapse.util.caches.stream_change_cache import StreamChangeCache + from ._base import BaseSlavedStore from ._slaved_id_tracker import SlavedIdTracker -from synapse.storage import DataStore -from synapse.util.caches.stream_change_cache import StreamChangeCache -from synapse.util.caches.expiringcache import ExpiringCache class SlavedDeviceInboxStore(BaseSlavedStore): diff --git a/synapse/replication/slave/storage/devices.py b/synapse/replication/slave/storage/devices.py index 7687867ae..8206a988f 100644 --- a/synapse/replication/slave/storage/devices.py +++ b/synapse/replication/slave/storage/devices.py @@ -13,12 +13,13 @@ # See the License for the specific language governing permissions and # limitations under the License. -from ._base import BaseSlavedStore -from ._slaved_id_tracker import SlavedIdTracker from synapse.storage import DataStore from synapse.storage.end_to_end_keys import EndToEndKeyStore from synapse.util.caches.stream_change_cache import StreamChangeCache +from ._base import BaseSlavedStore +from ._slaved_id_tracker import SlavedIdTracker + class SlavedDeviceStore(BaseSlavedStore): def __init__(self, db_conn, hs): diff --git a/synapse/replication/slave/storage/directory.py b/synapse/replication/slave/storage/directory.py index 7301d885f..1d1d48709 100644 --- a/synapse/replication/slave/storage/directory.py +++ b/synapse/replication/slave/storage/directory.py @@ -13,11 +13,10 @@ # See the License for the specific language governing permissions and # limitations under the License. +from synapse.storage.directory import DirectoryWorkerStore + from ._base import BaseSlavedStore -from synapse.storage.directory import DirectoryStore -class DirectoryStore(BaseSlavedStore): - get_aliases_for_room = DirectoryStore.__dict__[ - "get_aliases_for_room" - ] +class DirectoryStore(DirectoryWorkerStore, BaseSlavedStore): + pass diff --git a/synapse/replication/slave/storage/events.py b/synapse/replication/slave/storage/events.py index 94ebbffc1..bdb5eee4a 100644 --- a/synapse/replication/slave/storage/events.py +++ b/synapse/replication/slave/storage/events.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -12,20 +13,20 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. -from ._base import BaseSlavedStore -from ._slaved_id_tracker import SlavedIdTracker - -from synapse.api.constants import EventTypes -from synapse.storage import DataStore -from synapse.storage.roommember import RoomMemberStore -from synapse.storage.event_federation import EventFederationStore -from synapse.storage.event_push_actions import EventPushActionsStore -from synapse.storage.state import StateStore -from synapse.storage.stream import StreamStore -from synapse.util.caches.stream_change_cache import StreamChangeCache - import logging +from synapse.api.constants import EventTypes +from synapse.storage.event_federation import EventFederationWorkerStore +from synapse.storage.event_push_actions import EventPushActionsWorkerStore +from synapse.storage.events_worker import EventsWorkerStore +from synapse.storage.roommember import RoomMemberWorkerStore +from synapse.storage.signatures import SignatureWorkerStore +from synapse.storage.state import StateGroupWorkerStore +from synapse.storage.stream import StreamWorkerStore +from synapse.storage.user_erasure_store import UserErasureWorkerStore + +from ._base import BaseSlavedStore +from ._slaved_id_tracker import SlavedIdTracker logger = logging.getLogger(__name__) @@ -39,163 +40,34 @@ logger = logging.getLogger(__name__) # the method descriptor on the DataStore and chuck them into our class. -class SlavedEventStore(BaseSlavedStore): +class SlavedEventStore(EventFederationWorkerStore, + RoomMemberWorkerStore, + EventPushActionsWorkerStore, + StreamWorkerStore, + EventsWorkerStore, + StateGroupWorkerStore, + SignatureWorkerStore, + UserErasureWorkerStore, + BaseSlavedStore): def __init__(self, db_conn, hs): - super(SlavedEventStore, self).__init__(db_conn, hs) self._stream_id_gen = SlavedIdTracker( db_conn, "events", "stream_ordering", ) self._backfill_id_gen = SlavedIdTracker( db_conn, "events", "stream_ordering", step=-1 ) - events_max = self._stream_id_gen.get_current_token() - event_cache_prefill, min_event_val = self._get_cache_dict( - db_conn, "events", - entity_column="room_id", - stream_column="stream_ordering", - max_value=events_max, - ) - self._events_stream_cache = StreamChangeCache( - "EventsRoomStreamChangeCache", min_event_val, - prefilled_cache=event_cache_prefill, - ) - self._membership_stream_cache = StreamChangeCache( - "MembershipStreamChangeCache", events_max, - ) - self.stream_ordering_month_ago = 0 - self._stream_order_on_start = self.get_room_max_stream_ordering() + super(SlavedEventStore, self).__init__(db_conn, hs) # Cached functions can't be accessed through a class instance so we need # to reach inside the __dict__ to extract them. - get_rooms_for_user = RoomMemberStore.__dict__["get_rooms_for_user"] - get_users_in_room = RoomMemberStore.__dict__["get_users_in_room"] - get_hosts_in_room = RoomMemberStore.__dict__["get_hosts_in_room"] - get_users_who_share_room_with_user = ( - RoomMemberStore.__dict__["get_users_who_share_room_with_user"] - ) - get_latest_event_ids_in_room = EventFederationStore.__dict__[ - "get_latest_event_ids_in_room" - ] - get_invited_rooms_for_user = RoomMemberStore.__dict__[ - "get_invited_rooms_for_user" - ] - get_unread_event_push_actions_by_room_for_user = ( - EventPushActionsStore.__dict__["get_unread_event_push_actions_by_room_for_user"] - ) - _get_unread_counts_by_receipt_txn = ( - DataStore._get_unread_counts_by_receipt_txn.__func__ - ) - _get_unread_counts_by_pos_txn = ( - DataStore._get_unread_counts_by_pos_txn.__func__ - ) - _get_state_group_for_events = ( - StateStore.__dict__["_get_state_group_for_events"] - ) - _get_state_group_for_event = ( - StateStore.__dict__["_get_state_group_for_event"] - ) - _get_state_groups_from_groups = ( - StateStore.__dict__["_get_state_groups_from_groups"] - ) - _get_state_groups_from_groups_txn = ( - DataStore._get_state_groups_from_groups_txn.__func__ - ) - get_recent_event_ids_for_room = ( - StreamStore.__dict__["get_recent_event_ids_for_room"] - ) - get_current_state_ids = ( - StateStore.__dict__["get_current_state_ids"] - ) - get_state_group_delta = StateStore.__dict__["get_state_group_delta"] - _get_joined_hosts_cache = RoomMemberStore.__dict__["_get_joined_hosts_cache"] - has_room_changed_since = DataStore.has_room_changed_since.__func__ - get_unread_push_actions_for_user_in_range_for_http = ( - DataStore.get_unread_push_actions_for_user_in_range_for_http.__func__ - ) - get_unread_push_actions_for_user_in_range_for_email = ( - DataStore.get_unread_push_actions_for_user_in_range_for_email.__func__ - ) - get_push_action_users_in_range = ( - DataStore.get_push_action_users_in_range.__func__ - ) - get_event = DataStore.get_event.__func__ - get_events = DataStore.get_events.__func__ - get_rooms_for_user_where_membership_is = ( - DataStore.get_rooms_for_user_where_membership_is.__func__ - ) - get_membership_changes_for_user = ( - DataStore.get_membership_changes_for_user.__func__ - ) - get_room_events_max_id = DataStore.get_room_events_max_id.__func__ - get_room_events_stream_for_room = ( - DataStore.get_room_events_stream_for_room.__func__ - ) - get_events_around = DataStore.get_events_around.__func__ - get_state_for_event = DataStore.get_state_for_event.__func__ - get_state_for_events = DataStore.get_state_for_events.__func__ - get_state_groups = DataStore.get_state_groups.__func__ - get_state_groups_ids = DataStore.get_state_groups_ids.__func__ - get_state_ids_for_event = DataStore.get_state_ids_for_event.__func__ - get_state_ids_for_events = DataStore.get_state_ids_for_events.__func__ - get_joined_users_from_state = DataStore.get_joined_users_from_state.__func__ - get_joined_users_from_context = DataStore.get_joined_users_from_context.__func__ - _get_joined_users_from_context = ( - RoomMemberStore.__dict__["_get_joined_users_from_context"] - ) + def get_room_max_stream_ordering(self): + return self._stream_id_gen.get_current_token() - get_joined_hosts = DataStore.get_joined_hosts.__func__ - _get_joined_hosts = RoomMemberStore.__dict__["_get_joined_hosts"] - - get_recent_events_for_room = DataStore.get_recent_events_for_room.__func__ - get_room_events_stream_for_rooms = ( - DataStore.get_room_events_stream_for_rooms.__func__ - ) - is_host_joined = RoomMemberStore.__dict__["is_host_joined"] - get_stream_token_for_event = DataStore.get_stream_token_for_event.__func__ - - _set_before_and_after = staticmethod(DataStore._set_before_and_after) - - _get_events = DataStore._get_events.__func__ - _get_events_from_cache = DataStore._get_events_from_cache.__func__ - - _invalidate_get_event_cache = DataStore._invalidate_get_event_cache.__func__ - _enqueue_events = DataStore._enqueue_events.__func__ - _do_fetch = DataStore._do_fetch.__func__ - _fetch_event_rows = DataStore._fetch_event_rows.__func__ - _get_event_from_row = DataStore._get_event_from_row.__func__ - _get_rooms_for_user_where_membership_is_txn = ( - DataStore._get_rooms_for_user_where_membership_is_txn.__func__ - ) - _get_state_for_groups = DataStore._get_state_for_groups.__func__ - _get_all_state_from_cache = DataStore._get_all_state_from_cache.__func__ - _get_events_around_txn = DataStore._get_events_around_txn.__func__ - _get_some_state_from_cache = DataStore._get_some_state_from_cache.__func__ - - get_backfill_events = DataStore.get_backfill_events.__func__ - _get_backfill_events = DataStore._get_backfill_events.__func__ - get_missing_events = DataStore.get_missing_events.__func__ - _get_missing_events = DataStore._get_missing_events.__func__ - - get_auth_chain = DataStore.get_auth_chain.__func__ - get_auth_chain_ids = DataStore.get_auth_chain_ids.__func__ - _get_auth_chain_ids_txn = DataStore._get_auth_chain_ids_txn.__func__ - - get_room_max_stream_ordering = DataStore.get_room_max_stream_ordering.__func__ - - get_forward_extremeties_for_room = ( - DataStore.get_forward_extremeties_for_room.__func__ - ) - _get_forward_extremeties_for_room = ( - EventFederationStore.__dict__["_get_forward_extremeties_for_room"] - ) - - get_all_new_events_stream = DataStore.get_all_new_events_stream.__func__ - - get_federation_out_pos = DataStore.get_federation_out_pos.__func__ - update_federation_out_pos = DataStore.update_federation_out_pos.__func__ + def get_room_min_stream_ordering(self): + return self._backfill_id_gen.get_current_token() def stream_positions(self): result = super(SlavedEventStore, self).stream_positions() diff --git a/synapse/replication/slave/storage/filtering.py b/synapse/replication/slave/storage/filtering.py index 819ed6288..456a14cd5 100644 --- a/synapse/replication/slave/storage/filtering.py +++ b/synapse/replication/slave/storage/filtering.py @@ -13,9 +13,10 @@ # See the License for the specific language governing permissions and # limitations under the License. -from ._base import BaseSlavedStore from synapse.storage.filtering import FilteringStore +from ._base import BaseSlavedStore + class SlavedFilteringStore(BaseSlavedStore): def __init__(self, db_conn, hs): diff --git a/synapse/replication/slave/storage/groups.py b/synapse/replication/slave/storage/groups.py new file mode 100644 index 000000000..5777f07c8 --- /dev/null +++ b/synapse/replication/slave/storage/groups.py @@ -0,0 +1,55 @@ +# -*- coding: utf-8 -*- +# Copyright 2016 OpenMarket Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from synapse.storage import DataStore +from synapse.util.caches.stream_change_cache import StreamChangeCache + +from ._base import BaseSlavedStore +from ._slaved_id_tracker import SlavedIdTracker + + +class SlavedGroupServerStore(BaseSlavedStore): + def __init__(self, db_conn, hs): + super(SlavedGroupServerStore, self).__init__(db_conn, hs) + + self.hs = hs + + self._group_updates_id_gen = SlavedIdTracker( + db_conn, "local_group_updates", "stream_id", + ) + self._group_updates_stream_cache = StreamChangeCache( + "_group_updates_stream_cache", self._group_updates_id_gen.get_current_token(), + ) + + get_groups_changes_for_user = DataStore.get_groups_changes_for_user.__func__ + get_group_stream_token = DataStore.get_group_stream_token.__func__ + get_all_groups_for_user = DataStore.get_all_groups_for_user.__func__ + + def stream_positions(self): + result = super(SlavedGroupServerStore, self).stream_positions() + result["groups"] = self._group_updates_id_gen.get_current_token() + return result + + def process_replication_rows(self, stream_name, token, rows): + if stream_name == "groups": + self._group_updates_id_gen.advance(token) + for row in rows: + self._group_updates_stream_cache.entity_has_changed( + row.user_id, token + ) + + return super(SlavedGroupServerStore, self).process_replication_rows( + stream_name, token, rows + ) diff --git a/synapse/replication/slave/storage/keys.py b/synapse/replication/slave/storage/keys.py index dd2ae49e4..05ed16846 100644 --- a/synapse/replication/slave/storage/keys.py +++ b/synapse/replication/slave/storage/keys.py @@ -13,10 +13,11 @@ # See the License for the specific language governing permissions and # limitations under the License. -from ._base import BaseSlavedStore from synapse.storage import DataStore from synapse.storage.keys import KeyStore +from ._base import BaseSlavedStore + class SlavedKeyStore(BaseSlavedStore): _get_server_verify_key = KeyStore.__dict__[ diff --git a/synapse/replication/slave/storage/presence.py b/synapse/replication/slave/storage/presence.py index cfb928018..80b744082 100644 --- a/synapse/replication/slave/storage/presence.py +++ b/synapse/replication/slave/storage/presence.py @@ -13,12 +13,12 @@ # See the License for the specific language governing permissions and # limitations under the License. -from ._base import BaseSlavedStore -from ._slaved_id_tracker import SlavedIdTracker - -from synapse.util.caches.stream_change_cache import StreamChangeCache from synapse.storage import DataStore from synapse.storage.presence import PresenceStore +from synapse.util.caches.stream_change_cache import StreamChangeCache + +from ._base import BaseSlavedStore +from ._slaved_id_tracker import SlavedIdTracker class SlavedPresenceStore(BaseSlavedStore): diff --git a/synapse/replication/slave/storage/profile.py b/synapse/replication/slave/storage/profile.py new file mode 100644 index 000000000..46c28d417 --- /dev/null +++ b/synapse/replication/slave/storage/profile.py @@ -0,0 +1,21 @@ +# -*- coding: utf-8 -*- +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from synapse.replication.slave.storage._base import BaseSlavedStore +from synapse.storage.profile import ProfileWorkerStore + + +class SlavedProfileStore(ProfileWorkerStore, BaseSlavedStore): + pass diff --git a/synapse/replication/slave/storage/push_rule.py b/synapse/replication/slave/storage/push_rule.py index 83e880fdd..f0200c1e9 100644 --- a/synapse/replication/slave/storage/push_rule.py +++ b/synapse/replication/slave/storage/push_rule.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2015, 2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -13,31 +14,18 @@ # See the License for the specific language governing permissions and # limitations under the License. -from .events import SlavedEventStore +from synapse.storage.push_rule import PushRulesWorkerStore + from ._slaved_id_tracker import SlavedIdTracker -from synapse.storage import DataStore -from synapse.storage.push_rule import PushRuleStore -from synapse.util.caches.stream_change_cache import StreamChangeCache +from .events import SlavedEventStore -class SlavedPushRuleStore(SlavedEventStore): +class SlavedPushRuleStore(PushRulesWorkerStore, SlavedEventStore): def __init__(self, db_conn, hs): - super(SlavedPushRuleStore, self).__init__(db_conn, hs) self._push_rules_stream_id_gen = SlavedIdTracker( db_conn, "push_rules_stream", "stream_id", ) - self.push_rules_stream_cache = StreamChangeCache( - "PushRulesStreamChangeCache", - self._push_rules_stream_id_gen.get_current_token(), - ) - - get_push_rules_for_user = PushRuleStore.__dict__["get_push_rules_for_user"] - get_push_rules_enabled_for_user = ( - PushRuleStore.__dict__["get_push_rules_enabled_for_user"] - ) - have_push_rules_changed_for_user = ( - DataStore.have_push_rules_changed_for_user.__func__ - ) + super(SlavedPushRuleStore, self).__init__(db_conn, hs) def get_push_rules_stream_token(self): return ( @@ -45,6 +33,9 @@ class SlavedPushRuleStore(SlavedEventStore): self._stream_id_gen.get_current_token(), ) + def get_max_push_rules_stream_id(self): + return self._push_rules_stream_id_gen.get_current_token() + def stream_positions(self): result = super(SlavedPushRuleStore, self).stream_positions() result["push_rules"] = self._push_rules_stream_id_gen.get_current_token() diff --git a/synapse/replication/slave/storage/pushers.py b/synapse/replication/slave/storage/pushers.py index 4e8d68ece..3b2213c0d 100644 --- a/synapse/replication/slave/storage/pushers.py +++ b/synapse/replication/slave/storage/pushers.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -13,13 +14,13 @@ # See the License for the specific language governing permissions and # limitations under the License. +from synapse.storage.pusher import PusherWorkerStore + from ._base import BaseSlavedStore from ._slaved_id_tracker import SlavedIdTracker -from synapse.storage import DataStore - -class SlavedPusherStore(BaseSlavedStore): +class SlavedPusherStore(PusherWorkerStore, BaseSlavedStore): def __init__(self, db_conn, hs): super(SlavedPusherStore, self).__init__(db_conn, hs) @@ -28,13 +29,6 @@ class SlavedPusherStore(BaseSlavedStore): extra_tables=[("deleted_pushers", "stream_id")], ) - get_all_pushers = DataStore.get_all_pushers.__func__ - get_pushers_by = DataStore.get_pushers_by.__func__ - get_pushers_by_app_id_and_pushkey = ( - DataStore.get_pushers_by_app_id_and_pushkey.__func__ - ) - _decode_pushers_rows = DataStore._decode_pushers_rows.__func__ - def stream_positions(self): result = super(SlavedPusherStore, self).stream_positions() result["pushers"] = self._pushers_id_gen.get_current_token() diff --git a/synapse/replication/slave/storage/receipts.py b/synapse/replication/slave/storage/receipts.py index b371574ec..ed12342f4 100644 --- a/synapse/replication/slave/storage/receipts.py +++ b/synapse/replication/slave/storage/receipts.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -13,13 +14,11 @@ # See the License for the specific language governing permissions and # limitations under the License. +from synapse.storage.receipts import ReceiptsWorkerStore + from ._base import BaseSlavedStore from ._slaved_id_tracker import SlavedIdTracker -from synapse.storage import DataStore -from synapse.storage.receipts import ReceiptsStore -from synapse.util.caches.stream_change_cache import StreamChangeCache - # So, um, we want to borrow a load of functions intended for reading from # a DataStore, but we don't want to take functions that either write to the # DataStore or are cached and don't have cache invalidation logic. @@ -29,36 +28,19 @@ from synapse.util.caches.stream_change_cache import StreamChangeCache # the method descriptor on the DataStore and chuck them into our class. -class SlavedReceiptsStore(BaseSlavedStore): +class SlavedReceiptsStore(ReceiptsWorkerStore, BaseSlavedStore): def __init__(self, db_conn, hs): - super(SlavedReceiptsStore, self).__init__(db_conn, hs) - + # We instantiate this first as the ReceiptsWorkerStore constructor + # needs to be able to call get_max_receipt_stream_id self._receipts_id_gen = SlavedIdTracker( db_conn, "receipts_linearized", "stream_id" ) - self._receipts_stream_cache = StreamChangeCache( - "ReceiptsRoomChangeCache", self._receipts_id_gen.get_current_token() - ) + super(SlavedReceiptsStore, self).__init__(db_conn, hs) - get_receipts_for_user = ReceiptsStore.__dict__["get_receipts_for_user"] - get_linearized_receipts_for_room = ( - ReceiptsStore.__dict__["get_linearized_receipts_for_room"] - ) - _get_linearized_receipts_for_rooms = ( - ReceiptsStore.__dict__["_get_linearized_receipts_for_rooms"] - ) - get_last_receipt_event_id_for_user = ( - ReceiptsStore.__dict__["get_last_receipt_event_id_for_user"] - ) - - get_max_receipt_stream_id = DataStore.get_max_receipt_stream_id.__func__ - get_all_updated_receipts = DataStore.get_all_updated_receipts.__func__ - - get_linearized_receipts_for_rooms = ( - DataStore.get_linearized_receipts_for_rooms.__func__ - ) + def get_max_receipt_stream_id(self): + return self._receipts_id_gen.get_current_token() def stream_positions(self): result = super(SlavedReceiptsStore, self).stream_positions() @@ -67,10 +49,12 @@ class SlavedReceiptsStore(BaseSlavedStore): def invalidate_caches_for_receipt(self, room_id, receipt_type, user_id): self.get_receipts_for_user.invalidate((user_id, receipt_type)) - self.get_linearized_receipts_for_room.invalidate_many((room_id,)) + self._get_linearized_receipts_for_room.invalidate_many((room_id,)) self.get_last_receipt_event_id_for_user.invalidate( (user_id, room_id, receipt_type) ) + self._invalidate_get_users_with_receipts_in_room(room_id, receipt_type, user_id) + self.get_receipts_for_room.invalidate((room_id, receipt_type)) def process_replication_rows(self, stream_name, token, rows): if stream_name == "receipts": diff --git a/synapse/replication/slave/storage/registration.py b/synapse/replication/slave/storage/registration.py index e27c7332d..408d91df1 100644 --- a/synapse/replication/slave/storage/registration.py +++ b/synapse/replication/slave/storage/registration.py @@ -13,21 +13,10 @@ # See the License for the specific language governing permissions and # limitations under the License. +from synapse.storage.registration import RegistrationWorkerStore + from ._base import BaseSlavedStore -from synapse.storage import DataStore -from synapse.storage.registration import RegistrationStore -class SlavedRegistrationStore(BaseSlavedStore): - def __init__(self, db_conn, hs): - super(SlavedRegistrationStore, self).__init__(db_conn, hs) - - # TODO: use the cached version and invalidate deleted tokens - get_user_by_access_token = RegistrationStore.__dict__[ - "get_user_by_access_token" - ] - - _query_for_auth = DataStore._query_for_auth.__func__ - get_user_by_id = RegistrationStore.__dict__[ - "get_user_by_id" - ] +class SlavedRegistrationStore(RegistrationWorkerStore, BaseSlavedStore): + pass diff --git a/synapse/replication/slave/storage/room.py b/synapse/replication/slave/storage/room.py index f51038403..0cb474928 100644 --- a/synapse/replication/slave/storage/room.py +++ b/synapse/replication/slave/storage/room.py @@ -13,33 +13,21 @@ # See the License for the specific language governing permissions and # limitations under the License. +from synapse.storage.room import RoomWorkerStore + from ._base import BaseSlavedStore -from synapse.storage import DataStore -from synapse.storage.room import RoomStore from ._slaved_id_tracker import SlavedIdTracker -class RoomStore(BaseSlavedStore): +class RoomStore(RoomWorkerStore, BaseSlavedStore): def __init__(self, db_conn, hs): super(RoomStore, self).__init__(db_conn, hs) self._public_room_id_gen = SlavedIdTracker( db_conn, "public_room_list_stream", "stream_id" ) - get_public_room_ids = DataStore.get_public_room_ids.__func__ - get_current_public_room_stream_id = ( - DataStore.get_current_public_room_stream_id.__func__ - ) - get_public_room_ids_at_stream_id = ( - RoomStore.__dict__["get_public_room_ids_at_stream_id"] - ) - get_public_room_ids_at_stream_id_txn = ( - DataStore.get_public_room_ids_at_stream_id_txn.__func__ - ) - get_published_at_stream_id_txn = ( - DataStore.get_published_at_stream_id_txn.__func__ - ) - get_public_room_changes = DataStore.get_public_room_changes.__func__ + def get_current_public_room_stream_id(self): + return self._public_room_id_gen.get_current_token() def stream_positions(self): result = super(RoomStore, self).stream_positions() diff --git a/synapse/replication/slave/storage/transactions.py b/synapse/replication/slave/storage/transactions.py index fbb58f35d..9c9a5eadd 100644 --- a/synapse/replication/slave/storage/transactions.py +++ b/synapse/replication/slave/storage/transactions.py @@ -13,10 +13,11 @@ # See the License for the specific language governing permissions and # limitations under the License. -from ._base import BaseSlavedStore from synapse.storage import DataStore from synapse.storage.transactions import TransactionStore +from ._base import BaseSlavedStore + class TransactionStore(BaseSlavedStore): get_destination_retry_timings = TransactionStore.__dict__[ diff --git a/synapse/replication/tcp/client.py b/synapse/replication/tcp/client.py index 6d2513c4e..e592ab57b 100644 --- a/synapse/replication/tcp/client.py +++ b/synapse/replication/tcp/client.py @@ -15,17 +15,20 @@ """A replication client for use by synapse workers. """ -from twisted.internet import reactor, defer +import logging + +from twisted.internet import defer from twisted.internet.protocol import ReconnectingClientFactory from .commands import ( - FederationAckCommand, UserSyncCommand, RemovePusherCommand, InvalidateCacheCommand, + FederationAckCommand, + InvalidateCacheCommand, + RemovePusherCommand, UserIpCommand, + UserSyncCommand, ) from .protocol import ClientReplicationStreamProtocol -import logging - logger = logging.getLogger(__name__) @@ -44,7 +47,7 @@ class ReplicationClientFactory(ReconnectingClientFactory): self.server_name = hs.config.server_name self._clock = hs.get_clock() # As self.clock is defined in super class - reactor.addSystemEventTrigger("before", "shutdown", self.stopTrying) + hs.get_reactor().addSystemEventTrigger("before", "shutdown", self.stopTrying) def startedConnecting(self, connector): logger.info("Connecting to replication: %r", connector.getDestination()) @@ -95,7 +98,7 @@ class ReplicationClientHandler(object): factory = ReplicationClientFactory(hs, client_name, self) host = hs.config.worker_replication_host port = hs.config.worker_replication_port - reactor.connectTCP(host, port, factory) + hs.get_reactor().connectTCP(host, port, factory) def on_rdata(self, stream_name, token, rows): """Called when we get new replication data. By default this just pokes diff --git a/synapse/replication/tcp/commands.py b/synapse/replication/tcp/commands.py index a009214e4..f3908df64 100644 --- a/synapse/replication/tcp/commands.py +++ b/synapse/replication/tcp/commands.py @@ -19,8 +19,14 @@ allowed to be sent by which side. """ import logging -import ujson as json +import platform +if platform.python_implementation() == "PyPy": + import json + _json_encoder = json.JSONEncoder() +else: + import simplejson as json + _json_encoder = json.JSONEncoder(namedtuple_as_object=False) logger = logging.getLogger(__name__) @@ -107,7 +113,7 @@ class RdataCommand(Command): return " ".join(( self.stream_name, str(self.token) if self.token is not None else "batch", - json.dumps(self.row), + _json_encoder.encode(self.row), )) @@ -301,7 +307,9 @@ class InvalidateCacheCommand(Command): return cls(cache_func, json.loads(keys_json)) def to_line(self): - return " ".join((self.cache_func, json.dumps(self.keys))) + return " ".join(( + self.cache_func, _json_encoder.encode(self.keys), + )) class UserIpCommand(Command): @@ -323,14 +331,18 @@ class UserIpCommand(Command): @classmethod def from_line(cls, line): - user_id, access_token, ip, device_id, last_seen, user_agent = line.split(" ", 5) + user_id, jsn = line.split(" ", 1) - return cls(user_id, access_token, ip, user_agent, device_id, int(last_seen)) + access_token, ip, user_agent, device_id, last_seen = json.loads(jsn) + + return cls( + user_id, access_token, ip, user_agent, device_id, last_seen + ) def to_line(self): - return " ".join(( - self.user_id, self.access_token, self.ip, self.device_id, - str(self.last_seen), self.user_agent, + return self.user_id + " " + _json_encoder.encode(( + self.access_token, self.ip, self.user_agent, self.device_id, + self.last_seen, )) diff --git a/synapse/replication/tcp/protocol.py b/synapse/replication/tcp/protocol.py index 062272f8d..dec5ac091 100644 --- a/synapse/replication/tcp/protocol.py +++ b/synapse/replication/tcp/protocol.py @@ -49,32 +49,40 @@ indicate which side is sending, these are *not* included on the wire:: * connection closed by server * """ +import fcntl +import logging +import struct +from collections import defaultdict + +from six import iteritems, iterkeys + +from prometheus_client import Counter + from twisted.internet import defer from twisted.protocols.basic import LineOnlyReceiver from twisted.python.failure import Failure -from commands import ( - COMMAND_MAP, VALID_CLIENT_COMMANDS, VALID_SERVER_COMMANDS, - ErrorCommand, ServerCommand, RdataCommand, PositionCommand, PingCommand, - NameCommand, ReplicateCommand, UserSyncCommand, SyncCommand, -) -from streams import STREAMS_MAP - +from synapse.metrics import LaterGauge from synapse.util.stringutils import random_string -from synapse.metrics.metric import CounterMetric -import logging -import synapse.metrics -import struct -import fcntl - - -metrics = synapse.metrics.get_metrics_for(__name__) - -connection_close_counter = metrics.register_counter( - "close_reason", labels=["reason_type"], +from .commands import ( + COMMAND_MAP, + VALID_CLIENT_COMMANDS, + VALID_SERVER_COMMANDS, + ErrorCommand, + NameCommand, + PingCommand, + PositionCommand, + RdataCommand, + ReplicateCommand, + ServerCommand, + SyncCommand, + UserSyncCommand, ) +from .streams import STREAMS_MAP +connection_close_counter = Counter( + "synapse_replication_tcp_protocol_close_reason", "", ["reason_type"]) # A list of all connected protocols. This allows us to send metrics about the # connections. @@ -136,12 +144,8 @@ class BaseReplicationStreamProtocol(LineOnlyReceiver): # The LoopingCall for sending pings. self._send_ping_loop = None - self.inbound_commands_counter = CounterMetric( - "inbound_commands", labels=["command"], - ) - self.outbound_commands_counter = CounterMetric( - "outbound_commands", labels=["command"], - ) + self.inbound_commands_counter = defaultdict(int) + self.outbound_commands_counter = defaultdict(int) def connectionMade(self): logger.info("[%s] Connection established", self.id()) @@ -201,7 +205,8 @@ class BaseReplicationStreamProtocol(LineOnlyReceiver): self.last_received_command = self.clock.time_msec() - self.inbound_commands_counter.inc(cmd_name) + self.inbound_commands_counter[cmd_name] = ( + self.inbound_commands_counter[cmd_name] + 1) cmd_cls = COMMAND_MAP[cmd_name] try: @@ -244,15 +249,15 @@ class BaseReplicationStreamProtocol(LineOnlyReceiver): becoming full. """ if self.state == ConnectionStates.CLOSED: - logger.info("[%s] Not sending, connection closed", self.id()) + logger.debug("[%s] Not sending, connection closed", self.id()) return if do_buffer and self.state != ConnectionStates.ESTABLISHED: self._queue_command(cmd) return - self.outbound_commands_counter.inc(cmd.NAME) - + self.outbound_commands_counter[cmd.NAME] = ( + self.outbound_commands_counter[cmd.NAME] + 1) string = "%s %s" % (cmd.NAME, cmd.to_line(),) if "\n" in string: raise Exception("Unexpected newline in command: %r", string) @@ -264,7 +269,7 @@ class BaseReplicationStreamProtocol(LineOnlyReceiver): def _queue_command(self, cmd): """Queue the command until the connection is ready to write to again. """ - logger.info("[%s] Queing as conn %r, cmd: %r", self.id(), self.state, cmd) + logger.debug("[%s] Queing as conn %r, cmd: %r", self.id(), self.state, cmd) self.pending_commands.append(cmd) if len(self.pending_commands) > self.max_line_buffer: @@ -317,9 +322,9 @@ class BaseReplicationStreamProtocol(LineOnlyReceiver): def connectionLost(self, reason): logger.info("[%s] Replication connection closed: %r", self.id(), reason) if isinstance(reason, Failure): - connection_close_counter.inc(reason.type.__name__) + connection_close_counter.labels(reason.type.__name__).inc() else: - connection_close_counter.inc(reason.__class__.__name__) + connection_close_counter.labels(reason.__class__.__name__).inc() try: # Remove us from list of connections to be monitored @@ -392,7 +397,7 @@ class ServerReplicationStreamProtocol(BaseReplicationStreamProtocol): if stream_name == "ALL": # Subscribe to all streams we're publishing to. - for stream in self.streamer.streams_by_name.iterkeys(): + for stream in iterkeys(self.streamer.streams_by_name): self.subscribe_to_stream(stream, token) else: self.subscribe_to_stream(stream_name, token) @@ -498,7 +503,7 @@ class ClientReplicationStreamProtocol(BaseReplicationStreamProtocol): BaseReplicationStreamProtocol.connectionMade(self) # Once we've connected subscribe to the necessary streams - for stream_name, token in self.handler.get_streams_to_replicate().iteritems(): + for stream_name, token in iteritems(self.handler.get_streams_to_replicate()): self.replicate(stream_name, token) # Tell the server if we have any users currently syncing (should only @@ -517,25 +522,28 @@ class ClientReplicationStreamProtocol(BaseReplicationStreamProtocol): self.send_error("Wrong remote") def on_RDATA(self, cmd): + stream_name = cmd.stream_name + inbound_rdata_count.labels(stream_name).inc() + try: - row = STREAMS_MAP[cmd.stream_name].ROW_TYPE(*cmd.row) + row = STREAMS_MAP[stream_name].ROW_TYPE(*cmd.row) except Exception: logger.exception( "[%s] Failed to parse RDATA: %r %r", - self.id(), cmd.stream_name, cmd.row + self.id(), stream_name, cmd.row ) raise if cmd.token is None: # I.e. this is part of a batch of updates for this stream. Batch # until we get an update for the stream with a non None token - self.pending_batches.setdefault(cmd.stream_name, []).append(row) + self.pending_batches.setdefault(stream_name, []).append(row) else: # Check if this is the last of a batch of updates - rows = self.pending_batches.pop(cmd.stream_name, []) + rows = self.pending_batches.pop(stream_name, []) rows.append(row) - self.handler.on_rdata(cmd.stream_name, cmd.token, rows) + self.handler.on_rdata(stream_name, cmd.token, rows) def on_POSITION(self, cmd): self.handler.on_position(cmd.stream_name, cmd.token) @@ -563,13 +571,13 @@ class ClientReplicationStreamProtocol(BaseReplicationStreamProtocol): # The following simply registers metrics for the replication connections -metrics.register_callback( - "pending_commands", +pending_commands = LaterGauge( + "synapse_replication_tcp_protocol_pending_commands", + "", + ["name", "conn_id"], lambda: { - (p.name, p.conn_id): len(p.pending_commands) - for p in connected_connections + (p.name, p.conn_id): len(p.pending_commands) for p in connected_connections }, - labels=["name", "conn_id"], ) @@ -580,13 +588,13 @@ def transport_buffer_size(protocol): return 0 -metrics.register_callback( - "transport_send_buffer", +transport_send_buffer = LaterGauge( + "synapse_replication_tcp_protocol_transport_send_buffer", + "", + ["name", "conn_id"], lambda: { - (p.name, p.conn_id): transport_buffer_size(p) - for p in connected_connections + (p.name, p.conn_id): transport_buffer_size(p) for p in connected_connections }, - labels=["name", "conn_id"], ) @@ -605,42 +613,51 @@ def transport_kernel_read_buffer_size(protocol, read=True): return 0 -metrics.register_callback( - "transport_kernel_send_buffer", +tcp_transport_kernel_send_buffer = LaterGauge( + "synapse_replication_tcp_protocol_transport_kernel_send_buffer", + "", + ["name", "conn_id"], lambda: { (p.name, p.conn_id): transport_kernel_read_buffer_size(p, False) for p in connected_connections }, - labels=["name", "conn_id"], ) -metrics.register_callback( - "transport_kernel_read_buffer", +tcp_transport_kernel_read_buffer = LaterGauge( + "synapse_replication_tcp_protocol_transport_kernel_read_buffer", + "", + ["name", "conn_id"], lambda: { (p.name, p.conn_id): transport_kernel_read_buffer_size(p, True) for p in connected_connections }, - labels=["name", "conn_id"], ) -metrics.register_callback( - "inbound_commands", +tcp_inbound_commands = LaterGauge( + "synapse_replication_tcp_protocol_inbound_commands", + "", + ["command", "name", "conn_id"], lambda: { (k[0], p.name, p.conn_id): count for p in connected_connections - for k, count in p.inbound_commands_counter.counts.iteritems() + for k, count in iteritems(p.inbound_commands_counter) }, - labels=["command", "name", "conn_id"], ) -metrics.register_callback( - "outbound_commands", +tcp_outbound_commands = LaterGauge( + "synapse_replication_tcp_protocol_outbound_commands", + "", + ["command", "name", "conn_id"], lambda: { (k[0], p.name, p.conn_id): count for p in connected_connections - for k, count in p.outbound_commands_counter.counts.iteritems() + for k, count in iteritems(p.outbound_commands_counter) }, - labels=["command", "name", "conn_id"], +) + +# number of updates received for each RDATA stream +inbound_rdata_count = Counter( + "synapse_replication_tcp_protocol_inbound_rdata_count", "", ["stream_name"] ) diff --git a/synapse/replication/tcp/resource.py b/synapse/replication/tcp/resource.py index 3ea3ca5a6..611fb66e1 100644 --- a/synapse/replication/tcp/resource.py +++ b/synapse/replication/tcp/resource.py @@ -15,27 +15,29 @@ """The server side of the replication stream. """ -from twisted.internet import defer, reactor +import logging + +from six import itervalues + +from prometheus_client import Counter + +from twisted.internet import defer from twisted.internet.protocol import Factory -from streams import STREAMS_MAP, FederationStream -from protocol import ServerReplicationStreamProtocol - +from synapse.metrics import LaterGauge from synapse.util.metrics import Measure, measure_func -import logging -import synapse.metrics +from .protocol import ServerReplicationStreamProtocol +from .streams import STREAMS_MAP, FederationStream - -metrics = synapse.metrics.get_metrics_for(__name__) -stream_updates_counter = metrics.register_counter( - "stream_updates", labels=["stream_name"] -) -user_sync_counter = metrics.register_counter("user_sync") -federation_ack_counter = metrics.register_counter("federation_ack") -remove_pusher_counter = metrics.register_counter("remove_pusher") -invalidate_cache_counter = metrics.register_counter("invalidate_cache") -user_ip_cache_counter = metrics.register_counter("user_ip_cache") +stream_updates_counter = Counter("synapse_replication_tcp_resource_stream_updates", + "", ["stream_name"]) +user_sync_counter = Counter("synapse_replication_tcp_resource_user_sync", "") +federation_ack_counter = Counter("synapse_replication_tcp_resource_federation_ack", "") +remove_pusher_counter = Counter("synapse_replication_tcp_resource_remove_pusher", "") +invalidate_cache_counter = Counter("synapse_replication_tcp_resource_invalidate_cache", + "") +user_ip_cache_counter = Counter("synapse_replication_tcp_resource_user_ip_cache", "") logger = logging.getLogger(__name__) @@ -69,33 +71,34 @@ class ReplicationStreamer(object): self.presence_handler = hs.get_presence_handler() self.clock = hs.get_clock() self.notifier = hs.get_notifier() + self._server_notices_sender = hs.get_server_notices_sender() # Current connections. self.connections = [] - metrics.register_callback("total_connections", lambda: len(self.connections)) + LaterGauge("synapse_replication_tcp_resource_total_connections", "", [], + lambda: len(self.connections)) # List of streams that clients can subscribe to. # We only support federation stream if federation sending hase been # disabled on the master. self.streams = [ - stream(hs) for stream in STREAMS_MAP.itervalues() + stream(hs) for stream in itervalues(STREAMS_MAP) if stream != FederationStream or not hs.config.send_federation ] self.streams_by_name = {stream.NAME: stream for stream in self.streams} - metrics.register_callback( - "connections_per_stream", + LaterGauge( + "synapse_replication_tcp_resource_connections_per_stream", "", + ["stream_name"], lambda: { (stream_name,): len([ conn for conn in self.connections if stream_name in conn.replication_streams ]) for stream_name in self.streams_by_name - }, - labels=["stream_name"], - ) + }) self.federation_sender = None if not hs.config.send_federation: @@ -107,7 +110,7 @@ class ReplicationStreamer(object): self.is_looping = False self.pending_updates = False - reactor.addSystemEventTrigger("before", "shutdown", self.on_shutdown) + hs.get_reactor().addSystemEventTrigger("before", "shutdown", self.on_shutdown) def on_shutdown(self): # close all connections on shutdown @@ -160,7 +163,11 @@ class ReplicationStreamer(object): "Getting stream: %s: %s -> %s", stream.NAME, stream.last_token, stream.upto_token ) - updates, current_token = yield stream.get_updates() + try: + updates, current_token = yield stream.get_updates() + except Exception: + logger.info("Failed to handle stream %s", stream.NAME) + raise logger.debug( "Sending %d updates to %d connections", @@ -171,7 +178,7 @@ class ReplicationStreamer(object): logger.info( "Streaming: %s -> %s", stream.NAME, updates[-1][0] ) - stream_updates_counter.inc_by(len(updates), stream.NAME) + stream_updates_counter.labels(stream.NAME).inc(len(updates)) # Some streams return multiple rows with the same stream IDs, # we need to make sure they get sent out in batches. We do @@ -212,11 +219,12 @@ class ReplicationStreamer(object): self.federation_sender.federation_ack(token) @measure_func("repl.on_user_sync") + @defer.inlineCallbacks def on_user_sync(self, conn_id, user_id, is_syncing, last_sync_ms): """A client has started/stopped syncing on a worker. """ user_sync_counter.inc() - self.presence_handler.update_external_syncs_row( + yield self.presence_handler.update_external_syncs_row( conn_id, user_id, is_syncing, last_sync_ms, ) @@ -240,13 +248,15 @@ class ReplicationStreamer(object): getattr(self.store, cache_func).invalidate(tuple(keys)) @measure_func("repl.on_user_ip") + @defer.inlineCallbacks def on_user_ip(self, user_id, access_token, ip, user_agent, device_id, last_seen): """The client saw a user request """ user_ip_cache_counter.inc() - self.store.insert_client_ip( + yield self.store.insert_client_ip( user_id, access_token, ip, user_agent, device_id, last_seen, ) + yield self._server_notices_sender.on_user_ip(user_id) def send_sync_to_all_connections(self, data): """Sends a SYNC command to all clients. diff --git a/synapse/replication/tcp/streams.py b/synapse/replication/tcp/streams.py index fbafe12cc..55fe701c5 100644 --- a/synapse/replication/tcp/streams.py +++ b/synapse/replication/tcp/streams.py @@ -24,11 +24,10 @@ Each stream is defined by the following information: update_function: The function that returns a list of updates between two tokens """ -from twisted.internet import defer +import logging from collections import namedtuple -import logging - +from twisted.internet import defer logger = logging.getLogger(__name__) @@ -118,6 +117,12 @@ CurrentStateDeltaStreamRow = namedtuple("CurrentStateDeltaStream", ( "state_key", # str "event_id", # str, optional )) +GroupsStreamRow = namedtuple("GroupsStreamRow", ( + "group_id", # str + "user_id", # str + "type", # str + "content", # dict +)) class Stream(object): @@ -464,6 +469,19 @@ class CurrentStateDeltaStream(Stream): super(CurrentStateDeltaStream, self).__init__(hs) +class GroupServerStream(Stream): + NAME = "groups" + ROW_TYPE = GroupsStreamRow + + def __init__(self, hs): + store = hs.get_datastore() + + self.current_token = store.get_group_stream_token + self.update_function = store.get_all_groups_changes + + super(GroupServerStream, self).__init__(hs) + + STREAMS_MAP = { stream.NAME: stream for stream in ( @@ -482,5 +500,6 @@ STREAMS_MAP = { TagAccountDataStream, AccountDataStream, CurrentStateDeltaStream, + GroupServerStream, ) } diff --git a/synapse/rest/__init__.py b/synapse/rest/__init__.py index 3d809d181..3418f06fd 100644 --- a/synapse/rest/__init__.py +++ b/synapse/rest/__init__.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2014-2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -13,48 +14,50 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.rest.client import ( - versions, -) +from six import PY3 +from synapse.http.server import JsonResource +from synapse.rest.client import versions from synapse.rest.client.v1 import ( - room, - events, - profile, - presence, - initial_sync, - directory, - voip, admin, - pusher, - push_rule, - register as v1_register, + directory, + events, + initial_sync, login as v1_login, logout, + presence, + profile, + push_rule, + pusher, + room, + voip, ) - from synapse.rest.client.v2_alpha import ( - sync, - filter, account, - register, - auth, - receipts, - read_marker, - keys, - tokenrefresh, - tags, account_data, - report_event, - openid, - notifications, + auth, devices, - thirdparty, + filter, + groups, + keys, + notifications, + openid, + read_marker, + receipts, + register, + report_event, sendtodevice, + sync, + tags, + thirdparty, + tokenrefresh, user_directory, ) -from synapse.http.server import JsonResource +if not PY3: + from synapse.rest.client.v1_only import ( + register as v1_register, + ) class ClientRestResource(JsonResource): @@ -68,14 +71,22 @@ class ClientRestResource(JsonResource): def register_servlets(client_resource, hs): versions.register_servlets(client_resource) - # "v1" - room.register_servlets(hs, client_resource) + if not PY3: + # "v1" (Python 2 only) + v1_register.register_servlets(hs, client_resource) + + # Deprecated in r0 + initial_sync.register_servlets(hs, client_resource) + room.register_deprecated_servlets(hs, client_resource) + + # Partially deprecated in r0 events.register_servlets(hs, client_resource) - v1_register.register_servlets(hs, client_resource) + + # "v1" + "r0" + room.register_servlets(hs, client_resource) v1_login.register_servlets(hs, client_resource) profile.register_servlets(hs, client_resource) presence.register_servlets(hs, client_resource) - initial_sync.register_servlets(hs, client_resource) directory.register_servlets(hs, client_resource) voip.register_servlets(hs, client_resource) admin.register_servlets(hs, client_resource) @@ -102,3 +113,4 @@ class ClientRestResource(JsonResource): thirdparty.register_servlets(hs, client_resource) sendtodevice.register_servlets(hs, client_resource) user_directory.register_servlets(hs, client_resource) + groups.register_servlets(hs, client_resource) diff --git a/synapse/rest/client/transactions.py b/synapse/rest/client/transactions.py index fceca2ede..00b1b3066 100644 --- a/synapse/rest/client/transactions.py +++ b/synapse/rest/client/transactions.py @@ -17,37 +17,20 @@ to ensure idempotency when performing PUTs using the REST API.""" import logging -from synapse.api.auth import get_access_token_from_request from synapse.util.async import ObservableDeferred +from synapse.util.logcontext import make_deferred_yieldable, run_in_background logger = logging.getLogger(__name__) - -def get_transaction_key(request): - """A helper function which returns a transaction key that can be used - with TransactionCache for idempotent requests. - - Idempotency is based on the returned key being the same for separate - requests to the same endpoint. The key is formed from the HTTP request - path and the access_token for the requesting user. - - Args: - request (twisted.web.http.Request): The incoming request. Must - contain an access_token. - Returns: - str: A transaction key - """ - token = get_access_token_from_request(request) - return request.path + "/" + token - - CLEANUP_PERIOD_MS = 1000 * 60 * 30 # 30 mins class HttpTransactionCache(object): - def __init__(self, clock): - self.clock = clock + def __init__(self, hs): + self.hs = hs + self.auth = self.hs.get_auth() + self.clock = self.hs.get_clock() self.transactions = { # $txn_key: (ObservableDeferred<(res_code, res_json_body)>, timestamp) } @@ -55,6 +38,23 @@ class HttpTransactionCache(object): # for at *LEAST* 30 mins, and at *MOST* 60 mins. self.cleaner = self.clock.looping_call(self._cleanup, CLEANUP_PERIOD_MS) + def _get_transaction_key(self, request): + """A helper function which returns a transaction key that can be used + with TransactionCache for idempotent requests. + + Idempotency is based on the returned key being the same for separate + requests to the same endpoint. The key is formed from the HTTP request + path and the access_token for the requesting user. + + Args: + request (twisted.web.http.Request): The incoming request. Must + contain an access_token. + Returns: + str: A transaction key + """ + token = self.auth.get_access_token_from_request(request) + return request.path + "/" + token + def fetch_or_execute_request(self, request, fn, *args, **kwargs): """A helper function for fetch_or_execute which extracts a transaction key from the given request. @@ -63,7 +63,7 @@ class HttpTransactionCache(object): fetch_or_execute """ return self.fetch_or_execute( - get_transaction_key(request), fn, *args, **kwargs + self._get_transaction_key(request), fn, *args, **kwargs ) def fetch_or_execute(self, txn_key, fn, *args, **kwargs): @@ -80,31 +80,30 @@ class HttpTransactionCache(object): Returns: Deferred which resolves to a tuple of (response_code, response_dict). """ - try: - return self.transactions[txn_key][0].observe() - except (KeyError, IndexError): - pass # execute the function instead. + if txn_key in self.transactions: + observable = self.transactions[txn_key][0] + else: + # execute the function instead. + deferred = run_in_background(fn, *args, **kwargs) - deferred = fn(*args, **kwargs) + observable = ObservableDeferred(deferred) + self.transactions[txn_key] = (observable, self.clock.time_msec()) - # if the request fails with a Twisted failure, remove it - # from the transaction map. This is done to ensure that we don't - # cache transient errors like rate-limiting errors, etc. - def remove_from_map(err): - self.transactions.pop(txn_key, None) - return err - deferred.addErrback(remove_from_map) + # if the request fails with an exception, remove it + # from the transaction map. This is done to ensure that we don't + # cache transient errors like rate-limiting errors, etc. + def remove_from_map(err): + self.transactions.pop(txn_key, None) + # we deliberately do not propagate the error any further, as we + # expect the observers to have reported it. - # We don't add any other errbacks to the raw deferred, so we ask - # ObservableDeferred to swallow the error. This is fine as the error will - # still be reported to the observers. - observable = ObservableDeferred(deferred, consumeErrors=True) - self.transactions[txn_key] = (observable, self.clock.time_msec()) - return observable.observe() + deferred.addErrback(remove_from_map) + + return make_deferred_yieldable(observable.observe()) def _cleanup(self): now = self.clock.time_msec() - for key in self.transactions.keys(): + for key in list(self.transactions): ts = self.transactions[key][1] if now > (ts + CLEANUP_PERIOD_MS): # after cleanup period del self.transactions[key] diff --git a/synapse/rest/client/v1/admin.py b/synapse/rest/client/v1/admin.py index 7d786e8de..99f6c6e3c 100644 --- a/synapse/rest/client/v1/admin.py +++ b/synapse/rest/client/v1/admin.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2014-2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -13,17 +14,26 @@ # See the License for the specific language governing permissions and # limitations under the License. +import hashlib +import hmac +import logging + +from six.moves import http_client + from twisted.internet import defer from synapse.api.constants import Membership -from synapse.api.errors import AuthError, SynapseError +from synapse.api.errors import AuthError, Codes, NotFoundError, SynapseError +from synapse.http.servlet import ( + assert_params_in_dict, + parse_integer, + parse_json_object_from_request, + parse_string, +) from synapse.types import UserID, create_requester -from synapse.http.servlet import parse_json_object_from_request from .base import ClientV1RestServlet, client_path_patterns -import logging - logger = logging.getLogger(__name__) @@ -55,6 +65,125 @@ class UsersRestServlet(ClientV1RestServlet): defer.returnValue((200, ret)) +class UserRegisterServlet(ClientV1RestServlet): + """ + Attributes: + NONCE_TIMEOUT (int): Seconds until a generated nonce won't be accepted + nonces (dict[str, int]): The nonces that we will accept. A dict of + nonce to the time it was generated, in int seconds. + """ + PATTERNS = client_path_patterns("/admin/register") + NONCE_TIMEOUT = 60 + + def __init__(self, hs): + super(UserRegisterServlet, self).__init__(hs) + self.handlers = hs.get_handlers() + self.reactor = hs.get_reactor() + self.nonces = {} + self.hs = hs + + def _clear_old_nonces(self): + """ + Clear out old nonces that are older than NONCE_TIMEOUT. + """ + now = int(self.reactor.seconds()) + + for k, v in list(self.nonces.items()): + if now - v > self.NONCE_TIMEOUT: + del self.nonces[k] + + def on_GET(self, request): + """ + Generate a new nonce. + """ + self._clear_old_nonces() + + nonce = self.hs.get_secrets().token_hex(64) + self.nonces[nonce] = int(self.reactor.seconds()) + return (200, {"nonce": nonce.encode('ascii')}) + + @defer.inlineCallbacks + def on_POST(self, request): + self._clear_old_nonces() + + if not self.hs.config.registration_shared_secret: + raise SynapseError(400, "Shared secret registration is not enabled") + + body = parse_json_object_from_request(request) + + if "nonce" not in body: + raise SynapseError( + 400, "nonce must be specified", errcode=Codes.BAD_JSON, + ) + + nonce = body["nonce"] + + if nonce not in self.nonces: + raise SynapseError( + 400, "unrecognised nonce", + ) + + # Delete the nonce, so it can't be reused, even if it's invalid + del self.nonces[nonce] + + if "username" not in body: + raise SynapseError( + 400, "username must be specified", errcode=Codes.BAD_JSON, + ) + else: + if (not isinstance(body['username'], str) or len(body['username']) > 512): + raise SynapseError(400, "Invalid username") + + username = body["username"].encode("utf-8") + if b"\x00" in username: + raise SynapseError(400, "Invalid username") + + if "password" not in body: + raise SynapseError( + 400, "password must be specified", errcode=Codes.BAD_JSON, + ) + else: + if (not isinstance(body['password'], str) or len(body['password']) > 512): + raise SynapseError(400, "Invalid password") + + password = body["password"].encode("utf-8") + if b"\x00" in password: + raise SynapseError(400, "Invalid password") + + admin = body.get("admin", None) + got_mac = body["mac"] + + want_mac = hmac.new( + key=self.hs.config.registration_shared_secret.encode(), + digestmod=hashlib.sha1, + ) + want_mac.update(nonce) + want_mac.update(b"\x00") + want_mac.update(username) + want_mac.update(b"\x00") + want_mac.update(password) + want_mac.update(b"\x00") + want_mac.update(b"admin" if admin else b"notadmin") + want_mac = want_mac.hexdigest() + + if not hmac.compare_digest(want_mac, got_mac): + raise SynapseError( + 403, "HMAC incorrect", + ) + + # Reuse the parts of RegisterRestServlet to reduce code duplication + from synapse.rest.client.v2_alpha.register import RegisterRestServlet + register = RegisterRestServlet(self.hs) + + (user_id, _) = yield register.registration_handler.register( + localpart=username.lower(), password=password, admin=bool(admin), + generate_token=False, + ) + + result = yield register._create_registration_details(user_id, body) + defer.returnValue((200, result)) + + class WhoisRestServlet(ClientV1RestServlet): PATTERNS = client_path_patterns("/admin/whois/(?P[^/]*)") @@ -95,16 +224,8 @@ class PurgeMediaCacheRestServlet(ClientV1RestServlet): if not is_admin: raise AuthError(403, "You are not a server admin") - before_ts = request.args.get("before_ts", None) - if not before_ts: - raise SynapseError(400, "Missing 'before_ts' arg") - - logger.info("before_ts: %r", before_ts[0]) - - try: - before_ts = int(before_ts[0]) - except Exception: - raise SynapseError(400, "Invalid 'before_ts' arg") + before_ts = parse_integer(request, "before_ts", required=True) + logger.info("before_ts: %r", before_ts) ret = yield self.media_repository.delete_old_remote_media(before_ts) @@ -113,12 +234,18 @@ class PurgeMediaCacheRestServlet(ClientV1RestServlet): class PurgeHistoryRestServlet(ClientV1RestServlet): PATTERNS = client_path_patterns( - "/admin/purge_history/(?P[^/]*)/(?P[^/]*)" + "/admin/purge_history/(?P[^/]*)(/(?P[^/]+))?" ) def __init__(self, hs): + """ + + Args: + hs (synapse.server.HomeServer) + """ super(PurgeHistoryRestServlet, self).__init__(hs) - self.handlers = hs.get_handlers() + self.pagination_handler = hs.get_pagination_handler() + self.store = hs.get_datastore() @defer.inlineCallbacks def on_POST(self, request, room_id, event_id): @@ -128,20 +255,127 @@ class PurgeHistoryRestServlet(ClientV1RestServlet): if not is_admin: raise AuthError(403, "You are not a server admin") - yield self.handlers.message_handler.purge_history(room_id, event_id) + body = parse_json_object_from_request(request, allow_empty_body=True) - defer.returnValue((200, {})) + delete_local_events = bool(body.get("delete_local_events", False)) + + # establish the topological ordering we should keep events from. The + # user can provide an event_id in the URL or the request body, or can + # provide a timestamp in the request body. + if event_id is None: + event_id = body.get('purge_up_to_event_id') + + if event_id is not None: + event = yield self.store.get_event(event_id) + + if event.room_id != room_id: + raise SynapseError(400, "Event is for wrong room.") + + token = yield self.store.get_topological_token_for_event(event_id) + + logger.info( + "[purge] purging up to token %s (event_id %s)", + token, event_id, + ) + elif 'purge_up_to_ts' in body: + ts = body['purge_up_to_ts'] + if not isinstance(ts, int): + raise SynapseError( + 400, "purge_up_to_ts must be an int", + errcode=Codes.BAD_JSON, + ) + + stream_ordering = ( + yield self.store.find_first_stream_ordering_after_ts(ts) + ) + + r = ( + yield self.store.get_room_event_after_stream_ordering( + room_id, stream_ordering, + ) + ) + if not r: + logger.warn( + "[purge] purging events not possible: No event found " + "(received_ts %i => stream_ordering %i)", + ts, stream_ordering, + ) + raise SynapseError( + 404, + "there is no event to be purged", + errcode=Codes.NOT_FOUND, + ) + (stream, topo, _event_id) = r + token = "t%d-%d" % (topo, stream) + logger.info( + "[purge] purging up to token %s (received_ts %i => " + "stream_ordering %i)", + token, ts, stream_ordering, + ) + else: + raise SynapseError( + 400, + "must specify purge_up_to_event_id or purge_up_to_ts", + errcode=Codes.BAD_JSON, + ) + + purge_id = yield self.pagination_handler.start_purge_history( + room_id, token, + delete_local_events=delete_local_events, + ) + + defer.returnValue((200, { + "purge_id": purge_id, + })) + + +class PurgeHistoryStatusRestServlet(ClientV1RestServlet): + PATTERNS = client_path_patterns( + "/admin/purge_history_status/(?P[^/]+)" + ) + + def __init__(self, hs): + """ + + Args: + hs (synapse.server.HomeServer) + """ + super(PurgeHistoryStatusRestServlet, self).__init__(hs) + self.pagination_handler = hs.get_pagination_handler() + + @defer.inlineCallbacks + def on_GET(self, request, purge_id): + requester = yield self.auth.get_user_by_req(request) + is_admin = yield self.auth.is_server_admin(requester.user) + + if not is_admin: + raise AuthError(403, "You are not a server admin") + + purge_status = self.pagination_handler.get_purge_status(purge_id) + if purge_status is None: + raise NotFoundError("purge id '%s' not found" % purge_id) + + defer.returnValue((200, purge_status.asdict())) class DeactivateAccountRestServlet(ClientV1RestServlet): PATTERNS = client_path_patterns("/admin/deactivate/(?P[^/]*)") def __init__(self, hs): - self.store = hs.get_datastore() super(DeactivateAccountRestServlet, self).__init__(hs) + self._deactivate_account_handler = hs.get_deactivate_account_handler() @defer.inlineCallbacks def on_POST(self, request, target_user_id): + body = parse_json_object_from_request(request, allow_empty_body=True) + erase = body.get("erase", False) + if not isinstance(erase, bool): + raise SynapseError( + http_client.BAD_REQUEST, + "Param 'erase' must be a boolean, if given", + Codes.BAD_JSON, + ) + UserID.from_string(target_user_id) requester = yield self.auth.get_user_by_req(request) is_admin = yield self.auth.is_server_admin(requester.user) @@ -149,12 +383,9 @@ class DeactivateAccountRestServlet(ClientV1RestServlet): if not is_admin: raise AuthError(403, "You are not a server admin") - # FIXME: Theoretically there is a race here wherein user resets password - # using threepid. - yield self.store.user_delete_access_tokens(target_user_id) - yield self.store.user_delete_threepids(target_user_id) - yield self.store.user_set_password_hash(target_user_id, None) - + yield self._deactivate_account_handler.deactivate_account( + target_user_id, erase, + ) defer.returnValue((200, {})) @@ -168,14 +399,16 @@ class ShutdownRoomRestServlet(ClientV1RestServlet): DEFAULT_MESSAGE = ( "Sharing illegal content on this server is not permitted and rooms in" - " violatation will be blocked." + " violation will be blocked." ) def __init__(self, hs): super(ShutdownRoomRestServlet, self).__init__(hs) self.store = hs.get_datastore() - self.handlers = hs.get_handlers() self.state = hs.get_state_handler() + self._room_creation_handler = hs.get_room_creation_handler() + self.event_creation_handler = hs.get_event_creation_handler() + self.room_member_handler = hs.get_room_member_handler() @defer.inlineCallbacks def on_POST(self, request, room_id): @@ -185,17 +418,15 @@ class ShutdownRoomRestServlet(ClientV1RestServlet): raise AuthError(403, "You are not a server admin") content = parse_json_object_from_request(request) - - new_room_user_id = content.get("new_room_user_id") - if not new_room_user_id: - raise SynapseError(400, "Please provide field `new_room_user_id`") + assert_params_in_dict(content, ["new_room_user_id"]) + new_room_user_id = content["new_room_user_id"] room_creator_requester = create_requester(new_room_user_id) message = content.get("message", self.DEFAULT_MESSAGE) room_name = content.get("room_name", "Content Violation Notification") - info = yield self.handlers.room_creation_handler.create_room( + info = yield self._room_creation_handler.create_room( room_creator_requester, config={ "preset": "public_chat", @@ -208,8 +439,7 @@ class ShutdownRoomRestServlet(ClientV1RestServlet): ) new_room_id = info["room_id"] - msg_handler = self.handlers.message_handler - yield msg_handler.create_and_send_nonmember_event( + yield self.event_creation_handler.create_and_send_nonmember_event( room_creator_requester, { "type": "m.room.message", @@ -235,7 +465,7 @@ class ShutdownRoomRestServlet(ClientV1RestServlet): logger.info("Kicking %r from %r...", user_id, room_id) target_requester = create_requester(user_id) - yield self.handlers.room_member_handler.update_membership( + yield self.room_member_handler.update_membership( requester=target_requester, target=target_requester.user, room_id=room_id, @@ -244,9 +474,9 @@ class ShutdownRoomRestServlet(ClientV1RestServlet): ratelimit=False ) - yield self.handlers.room_member_handler.forget(target_requester.user, room_id) + yield self.room_member_handler.forget(target_requester.user, room_id) - yield self.handlers.room_member_handler.update_membership( + yield self.room_member_handler.update_membership( requester=target_requester, target=target_requester.user, room_id=new_room_id, @@ -294,9 +524,30 @@ class QuarantineMediaInRoom(ClientV1RestServlet): defer.returnValue((200, {"num_quarantined": num_quarantined})) +class ListMediaInRoom(ClientV1RestServlet): + """Lists all of the media in a given room. + """ + PATTERNS = client_path_patterns("/admin/room/(?P[^/]+)/media") + + def __init__(self, hs): + super(ListMediaInRoom, self).__init__(hs) + self.store = hs.get_datastore() + + @defer.inlineCallbacks + def on_GET(self, request, room_id): + requester = yield self.auth.get_user_by_req(request) + is_admin = yield self.auth.is_server_admin(requester.user) + if not is_admin: + raise AuthError(403, "You are not a server admin") + + local_mxcs, remote_mxcs = yield self.store.get_media_mxcs_in_room(room_id) + + defer.returnValue((200, {"local": local_mxcs, "remote": remote_mxcs})) + + class ResetPasswordRestServlet(ClientV1RestServlet): """Post request to allow an administrator reset password for a user. - This need a user have a administrator access in Synapse. + This needs user to have administrator access in Synapse. Example: http://localhost:8008/_matrix/client/api/v1/admin/reset_password/ @user:to_reset_password?access_token=admin_access_token @@ -314,12 +565,12 @@ class ResetPasswordRestServlet(ClientV1RestServlet): super(ResetPasswordRestServlet, self).__init__(hs) self.hs = hs self.auth = hs.get_auth() - self.auth_handler = hs.get_auth_handler() + self._set_password_handler = hs.get_set_password_handler() @defer.inlineCallbacks def on_POST(self, request, target_user_id): """Post request to allow an administrator reset password for a user. - This need a user have a administrator access in Synapse. + This needs user to have administrator access in Synapse. """ UserID.from_string(target_user_id) requester = yield self.auth.get_user_by_req(request) @@ -329,13 +580,12 @@ class ResetPasswordRestServlet(ClientV1RestServlet): raise AuthError(403, "You are not a server admin") params = parse_json_object_from_request(request) + assert_params_in_dict(params, ["new_password"]) new_password = params['new_password'] - if not new_password: - raise SynapseError(400, "Missing 'new_password' arg") logger.info("new_password: %r", new_password) - yield self.auth_handler.set_password( + yield self._set_password_handler.set_password( target_user_id, new_password, requester ) defer.returnValue((200, {})) @@ -343,7 +593,7 @@ class ResetPasswordRestServlet(ClientV1RestServlet): class GetUsersPaginatedRestServlet(ClientV1RestServlet): """Get request to get specific number of users from Synapse. - This need a user have a administrator access in Synapse. + This needs user to have administrator access in Synapse. Example: http://localhost:8008/_matrix/client/api/v1/admin/users_paginate/ @admin:user?access_token=admin_access_token&start=0&limit=10 @@ -362,7 +612,7 @@ class GetUsersPaginatedRestServlet(ClientV1RestServlet): @defer.inlineCallbacks def on_GET(self, request, target_user_id): """Get request to get specific number of users from Synapse. - This need a user have a administrator access in Synapse. + This needs user to have administrator access in Synapse. """ target_user = UserID.from_string(target_user_id) requester = yield self.auth.get_user_by_req(request) @@ -379,12 +629,9 @@ class GetUsersPaginatedRestServlet(ClientV1RestServlet): raise SynapseError(400, "Can only users a local user") order = "name" # order by name in user table - start = request.args.get("start")[0] - limit = request.args.get("limit")[0] - if not limit: - raise SynapseError(400, "Missing 'limit' arg") - if not start: - raise SynapseError(400, "Missing 'start' arg") + start = parse_integer(request, "start", required=True) + limit = parse_integer(request, "limit", required=True) + logger.info("limit: %s, start: %s", limit, start) ret = yield self.handlers.admin_handler.get_users_paginate( @@ -395,7 +642,7 @@ class GetUsersPaginatedRestServlet(ClientV1RestServlet): @defer.inlineCallbacks def on_POST(self, request, target_user_id): """Post request to get specific number of users from Synapse.. - This need a user have a administrator access in Synapse. + This needs user to have administrator access in Synapse. Example: http://localhost:8008/_matrix/client/api/v1/admin/users_paginate/ @admin:user?access_token=admin_access_token @@ -416,12 +663,9 @@ class GetUsersPaginatedRestServlet(ClientV1RestServlet): order = "name" # order by name in user table params = parse_json_object_from_request(request) + assert_params_in_dict(params, ["limit", "start"]) limit = params['limit'] start = params['start'] - if not limit: - raise SynapseError(400, "Missing 'limit' arg") - if not start: - raise SynapseError(400, "Missing 'start' arg") logger.info("limit: %s, start: %s", limit, start) ret = yield self.handlers.admin_handler.get_users_paginate( @@ -433,7 +677,7 @@ class GetUsersPaginatedRestServlet(ClientV1RestServlet): class SearchUsersRestServlet(ClientV1RestServlet): """Get request to search user table for specific users according to search term. - This need a user have a administrator access in Synapse. + This needs user to have administrator access in Synapse. Example: http://localhost:8008/_matrix/client/api/v1/admin/search_users/ @admin:user?access_token=admin_access_token&term=alice @@ -453,7 +697,7 @@ class SearchUsersRestServlet(ClientV1RestServlet): def on_GET(self, request, target_user_id): """Get request to search user table for specific users according to search term. - This need a user have a administrator access in Synapse. + This needs user to have a administrator access in Synapse. """ target_user = UserID.from_string(target_user_id) requester = yield self.auth.get_user_by_req(request) @@ -469,10 +713,7 @@ class SearchUsersRestServlet(ClientV1RestServlet): if not self.hs.is_mine(target_user): raise SynapseError(400, "Can only users a local user") - term = request.args.get("term")[0] - if not term: - raise SynapseError(400, "Missing 'term' arg") - + term = parse_string(request, "term", required=True) logger.info("term: %s ", term) ret = yield self.handlers.admin_handler.search_users( @@ -484,6 +725,7 @@ class SearchUsersRestServlet(ClientV1RestServlet): def register_servlets(hs, http_server): WhoisRestServlet(hs).register(http_server) PurgeMediaCacheRestServlet(hs).register(http_server) + PurgeHistoryStatusRestServlet(hs).register(http_server) DeactivateAccountRestServlet(hs).register(http_server) PurgeHistoryRestServlet(hs).register(http_server) UsersRestServlet(hs).register(http_server) @@ -492,3 +734,5 @@ def register_servlets(hs, http_server): SearchUsersRestServlet(hs).register(http_server) ShutdownRoomRestServlet(hs).register(http_server) QuarantineMediaInRoom(hs).register(http_server) + ListMediaInRoom(hs).register(http_server) + UserRegisterServlet(hs).register(http_server) diff --git a/synapse/rest/client/v1/base.py b/synapse/rest/client/v1/base.py index c7aa0bbf5..c77d7aba6 100644 --- a/synapse/rest/client/v1/base.py +++ b/synapse/rest/client/v1/base.py @@ -16,14 +16,12 @@ """This module contains base REST classes for constructing client v1 servlets. """ -from synapse.http.servlet import RestServlet -from synapse.api.urls import CLIENT_PREFIX -from synapse.rest.client.transactions import HttpTransactionCache - +import logging import re -import logging - +from synapse.api.urls import CLIENT_PREFIX +from synapse.http.servlet import RestServlet +from synapse.rest.client.transactions import HttpTransactionCache logger = logging.getLogger(__name__) @@ -52,6 +50,10 @@ class ClientV1RestServlet(RestServlet): """A base Synapse REST Servlet for the client version 1 API. """ + # This subclass was presumably created to allow the auth for the v1 + # protocol version to be different, however this behaviour was removed. + # it may no longer be necessary + def __init__(self, hs): """ Args: @@ -59,5 +61,5 @@ class ClientV1RestServlet(RestServlet): """ self.hs = hs self.builder_factory = hs.get_event_builder_factory() - self.auth = hs.get_v1auth() - self.txns = HttpTransactionCache(hs.get_clock()) + self.auth = hs.get_auth() + self.txns = HttpTransactionCache(hs) diff --git a/synapse/rest/client/v1/directory.py b/synapse/rest/client/v1/directory.py index f15aa5c13..69dcd618c 100644 --- a/synapse/rest/client/v1/directory.py +++ b/synapse/rest/client/v1/directory.py @@ -14,16 +14,15 @@ # limitations under the License. -from twisted.internet import defer - -from synapse.api.errors import AuthError, SynapseError, Codes -from synapse.types import RoomAlias -from synapse.http.servlet import parse_json_object_from_request - -from .base import ClientV1RestServlet, client_path_patterns - import logging +from twisted.internet import defer + +from synapse.api.errors import AuthError, Codes, SynapseError +from synapse.http.servlet import parse_json_object_from_request +from synapse.types import RoomAlias + +from .base import ClientV1RestServlet, client_path_patterns logger = logging.getLogger(__name__) @@ -53,15 +52,14 @@ class ClientDirectoryServer(ClientV1RestServlet): @defer.inlineCallbacks def on_PUT(self, request, room_alias): + room_alias = RoomAlias.from_string(room_alias) + content = parse_json_object_from_request(request) if "room_id" not in content: - raise SynapseError(400, "Missing room_id key", + raise SynapseError(400, 'Missing params: ["room_id"]', errcode=Codes.BAD_JSON) logger.debug("Got content: %s", content) - - room_alias = RoomAlias.from_string(room_alias) - logger.debug("Got room name: %s", room_alias.to_string()) room_id = content["room_id"] @@ -93,7 +91,7 @@ class ClientDirectoryServer(ClientV1RestServlet): ) except SynapseError as e: raise e - except: + except Exception: logger.exception("Failed to create association") raise except AuthError: diff --git a/synapse/rest/client/v1/events.py b/synapse/rest/client/v1/events.py index 701b6f549..b70c9c280 100644 --- a/synapse/rest/client/v1/events.py +++ b/synapse/rest/client/v1/events.py @@ -14,15 +14,15 @@ # limitations under the License. """This module contains REST servlets to do with event streaming, /events.""" +import logging + from twisted.internet import defer from synapse.api.errors import SynapseError -from synapse.streams.config import PaginationConfig -from .base import ClientV1RestServlet, client_path_patterns from synapse.events.utils import serialize_event +from synapse.streams.config import PaginationConfig -import logging - +from .base import ClientV1RestServlet, client_path_patterns logger = logging.getLogger(__name__) diff --git a/synapse/rest/client/v1/initial_sync.py b/synapse/rest/client/v1/initial_sync.py index 478e21eea..fd5f85b53 100644 --- a/synapse/rest/client/v1/initial_sync.py +++ b/synapse/rest/client/v1/initial_sync.py @@ -15,7 +15,9 @@ from twisted.internet import defer +from synapse.http.servlet import parse_boolean from synapse.streams.config import PaginationConfig + from .base import ClientV1RestServlet, client_path_patterns @@ -32,7 +34,7 @@ class InitialSyncRestServlet(ClientV1RestServlet): requester = yield self.auth.get_user_by_req(request) as_client_event = "raw" not in request.args pagination_config = PaginationConfig.from_request(request) - include_archived = request.args.get("archived", None) == ["true"] + include_archived = parse_boolean(request, "archived", default=False) content = yield self.initial_sync_handler.snapshot_all_rooms( user_id=requester.user.to_string(), pagin_config=pagination_config, diff --git a/synapse/rest/client/v1/login.py b/synapse/rest/client/v1/login.py index a43410fb3..cb85fa143 100644 --- a/synapse/rest/client/v1/login.py +++ b/synapse/rest/client/v1/login.py @@ -13,30 +13,27 @@ # See the License for the specific language governing permissions and # limitations under the License. -from twisted.internet import defer +import logging +import urllib +import xml.etree.ElementTree as ET -from synapse.api.errors import SynapseError, LoginError, Codes -from synapse.types import UserID +from six.moves.urllib import parse as urlparse + +from canonicaljson import json +from saml2 import BINDING_HTTP_POST, config +from saml2.client import Saml2Client + +from twisted.internet import defer +from twisted.web.client import PartialDownloadError + +from synapse.api.errors import Codes, LoginError, SynapseError from synapse.http.server import finish_request from synapse.http.servlet import parse_json_object_from_request +from synapse.types import UserID from synapse.util.msisdn import phone_number_to_msisdn from .base import ClientV1RestServlet, client_path_patterns -import simplejson as json -import urllib -import urlparse - -import logging -from saml2 import BINDING_HTTP_POST -from saml2 import config -from saml2.client import Saml2Client - -import xml.etree.ElementTree as ET - -from twisted.web.client import PartialDownloadError - - logger = logging.getLogger(__name__) @@ -85,7 +82,6 @@ def login_id_thirdparty_from_phone(identifier): class LoginRestServlet(ClientV1RestServlet): PATTERNS = client_path_patterns("/login$") - PASS_TYPE = "m.login.password" SAML2_TYPE = "m.login.saml2" CAS_TYPE = "m.login.cas" TOKEN_TYPE = "m.login.token" @@ -94,7 +90,6 @@ class LoginRestServlet(ClientV1RestServlet): def __init__(self, hs): super(LoginRestServlet, self).__init__(hs) self.idp_redirect_url = hs.config.saml2_idp_redirect_url - self.password_enabled = hs.config.password_enabled self.saml2_enabled = hs.config.saml2_enabled self.jwt_enabled = hs.config.jwt_enabled self.jwt_secret = hs.config.jwt_secret @@ -121,8 +116,10 @@ class LoginRestServlet(ClientV1RestServlet): # fall back to the fallback API if they don't understand one of the # login flow types returned. flows.append({"type": LoginRestServlet.TOKEN_TYPE}) - if self.password_enabled: - flows.append({"type": LoginRestServlet.PASS_TYPE}) + + flows.extend(( + {"type": t} for t in self.auth_handler.get_supported_login_types() + )) return (200, {"flows": flows}) @@ -133,14 +130,8 @@ class LoginRestServlet(ClientV1RestServlet): def on_POST(self, request): login_submission = parse_json_object_from_request(request) try: - if login_submission["type"] == LoginRestServlet.PASS_TYPE: - if not self.password_enabled: - raise SynapseError(400, "Password login has been disabled.") - - result = yield self.do_password_login(login_submission) - defer.returnValue(result) - elif self.saml2_enabled and (login_submission["type"] == - LoginRestServlet.SAML2_TYPE): + if self.saml2_enabled and (login_submission["type"] == + LoginRestServlet.SAML2_TYPE): relay_state = "" if "relay_state" in login_submission: relay_state = "&RelayState=" + urllib.quote( @@ -157,15 +148,31 @@ class LoginRestServlet(ClientV1RestServlet): result = yield self.do_token_login(login_submission) defer.returnValue(result) else: - raise SynapseError(400, "Bad login type.") + result = yield self._do_other_login(login_submission) + defer.returnValue(result) except KeyError: raise SynapseError(400, "Missing JSON keys.") @defer.inlineCallbacks - def do_password_login(self, login_submission): - if "password" not in login_submission: - raise SynapseError(400, "Missing parameter: password") + def _do_other_login(self, login_submission): + """Handle non-token/saml/jwt logins + Args: + login_submission: + + Returns: + (int, object): HTTP code/response + """ + # Log the request we got, but only certain fields to minimise the chance of + # logging someone's password (even if they accidentally put it in the wrong + # field) + logger.info( + "Got login request with identifier: %r, medium: %r, address: %r, user: %r", + login_submission.get('identifier'), + login_submission.get('medium'), + login_submission.get('address'), + login_submission.get('user'), + ) login_submission_legacy_convert(login_submission) if "identifier" not in login_submission: @@ -181,19 +188,25 @@ class LoginRestServlet(ClientV1RestServlet): # convert threepid identifiers to user IDs if identifier["type"] == "m.id.thirdparty": - if 'medium' not in identifier or 'address' not in identifier: + address = identifier.get('address') + medium = identifier.get('medium') + + if medium is None or address is None: raise SynapseError(400, "Invalid thirdparty identifier") - address = identifier['address'] - if identifier['medium'] == 'email': + if medium == 'email': # For emails, transform the address to lowercase. # We store all email addreses as lowercase in the DB. # (See add_threepid in synapse/handlers/auth.py) address = address.lower() user_id = yield self.hs.get_datastore().get_user_id_by_threepid( - identifier['medium'], address + medium, address, ) if not user_id: + logger.warn( + "unknown 3pid identifier medium %s, address %r", + medium, address, + ) raise LoginError(403, "", errcode=Codes.FORBIDDEN) identifier = { @@ -208,30 +221,29 @@ class LoginRestServlet(ClientV1RestServlet): if "user" not in identifier: raise SynapseError(400, "User identifier is missing 'user' key") - user_id = identifier["user"] - - if not user_id.startswith('@'): - user_id = UserID.create( - user_id, self.hs.hostname - ).to_string() - auth_handler = self.auth_handler - user_id = yield auth_handler.validate_password_login( - user_id=user_id, - password=login_submission["password"], + canonical_user_id, callback = yield auth_handler.validate_login( + identifier["user"], + login_submission, + ) + + device_id = yield self._register_device( + canonical_user_id, login_submission, ) - device_id = yield self._register_device(user_id, login_submission) access_token = yield auth_handler.get_access_token_for_user_id( - user_id, device_id, - login_submission.get("initial_device_display_name"), + canonical_user_id, device_id, ) + result = { - "user_id": user_id, # may have changed + "user_id": canonical_user_id, "access_token": access_token, "home_server": self.hs.hostname, "device_id": device_id, } + if callback is not None: + yield callback(result) + defer.returnValue((200, result)) @defer.inlineCallbacks @@ -244,7 +256,6 @@ class LoginRestServlet(ClientV1RestServlet): device_id = yield self._register_device(user_id, login_submission) access_token = yield auth_handler.get_access_token_for_user_id( user_id, device_id, - login_submission.get("initial_device_display_name"), ) result = { "user_id": user_id, # may have changed @@ -278,7 +289,7 @@ class LoginRestServlet(ClientV1RestServlet): if user is None: raise LoginError(401, "Invalid JWT", errcode=Codes.UNAUTHORIZED) - user_id = UserID.create(user, self.hs.hostname).to_string() + user_id = UserID(user, self.hs.hostname).to_string() auth_handler = self.auth_handler registered_user_id = yield auth_handler.check_user_exists(user_id) if registered_user_id: @@ -287,7 +298,6 @@ class LoginRestServlet(ClientV1RestServlet): ) access_token = yield auth_handler.get_access_token_for_user_id( registered_user_id, device_id, - login_submission.get("initial_device_display_name"), ) result = { @@ -444,7 +454,7 @@ class CasTicketServlet(ClientV1RestServlet): if required_value != actual_value: raise LoginError(401, "Unauthorized", errcode=Codes.UNAUTHORIZED) - user_id = UserID.create(user, self.hs.hostname).to_string() + user_id = UserID(user, self.hs.hostname).to_string() auth_handler = self.auth_handler registered_user_id = yield auth_handler.check_user_exists(user_id) if not registered_user_id: diff --git a/synapse/rest/client/v1/logout.py b/synapse/rest/client/v1/logout.py index 1358d0aca..430c69233 100644 --- a/synapse/rest/client/v1/logout.py +++ b/synapse/rest/client/v1/logout.py @@ -13,14 +13,13 @@ # See the License for the specific language governing permissions and # limitations under the License. -from twisted.internet import defer - -from synapse.api.auth import get_access_token_from_request - -from .base import ClientV1RestServlet, client_path_patterns - import logging +from twisted.internet import defer + +from synapse.api.errors import AuthError + +from .base import ClientV1RestServlet, client_path_patterns logger = logging.getLogger(__name__) @@ -30,15 +29,33 @@ class LogoutRestServlet(ClientV1RestServlet): def __init__(self, hs): super(LogoutRestServlet, self).__init__(hs) - self.store = hs.get_datastore() + self._auth = hs.get_auth() + self._auth_handler = hs.get_auth_handler() + self._device_handler = hs.get_device_handler() def on_OPTIONS(self, request): return (200, {}) @defer.inlineCallbacks def on_POST(self, request): - access_token = get_access_token_from_request(request) - yield self.store.delete_access_token(access_token) + try: + requester = yield self.auth.get_user_by_req(request) + except AuthError: + # this implies the access token has already been deleted. + defer.returnValue((401, { + "errcode": "M_UNKNOWN_TOKEN", + "error": "Access Token unknown or expired" + })) + else: + if requester.device_id is None: + # the acccess token wasn't associated with a device. + # Just delete the access token + access_token = self._auth.get_access_token_from_request(request) + yield self._auth_handler.delete_access_token(access_token) + else: + yield self._device_handler.delete_device( + requester.user.to_string(), requester.device_id) + defer.returnValue((200, {})) @@ -47,8 +64,9 @@ class LogoutAllRestServlet(ClientV1RestServlet): def __init__(self, hs): super(LogoutAllRestServlet, self).__init__(hs) - self.store = hs.get_datastore() self.auth = hs.get_auth() + self._auth_handler = hs.get_auth_handler() + self._device_handler = hs.get_device_handler() def on_OPTIONS(self, request): return (200, {}) @@ -57,7 +75,13 @@ class LogoutAllRestServlet(ClientV1RestServlet): def on_POST(self, request): requester = yield self.auth.get_user_by_req(request) user_id = requester.user.to_string() - yield self.store.user_delete_access_tokens(user_id) + + # first delete all of the user's devices + yield self._device_handler.delete_all_devices_for_user(user_id) + + # .. and then delete any access tokens which weren't associated with + # devices. + yield self._auth_handler.delete_access_tokens_for_user(user_id) defer.returnValue((200, {})) diff --git a/synapse/rest/client/v1/presence.py b/synapse/rest/client/v1/presence.py index 47b2dc45e..a14f0c807 100644 --- a/synapse/rest/client/v1/presence.py +++ b/synapse/rest/client/v1/presence.py @@ -15,15 +15,18 @@ """ This module contains REST servlets to do with presence: /presence/ """ +import logging + +from six import string_types + from twisted.internet import defer -from synapse.api.errors import SynapseError, AuthError -from synapse.types import UserID +from synapse.api.errors import AuthError, SynapseError from synapse.handlers.presence import format_user_presence_state from synapse.http.servlet import parse_json_object_from_request -from .base import ClientV1RestServlet, client_path_patterns +from synapse.types import UserID -import logging +from .base import ClientV1RestServlet, client_path_patterns logger = logging.getLogger(__name__) @@ -71,14 +74,14 @@ class PresenceStatusRestServlet(ClientV1RestServlet): if "status_msg" in content: state["status_msg"] = content.pop("status_msg") - if not isinstance(state["status_msg"], basestring): + if not isinstance(state["status_msg"], string_types): raise SynapseError(400, "status_msg must be a string.") if content: raise KeyError() except SynapseError as e: raise e - except: + except Exception: raise SynapseError(400, "Unable to parse state") yield self.presence_handler.set_state(user, state) @@ -129,7 +132,7 @@ class PresenceListRestServlet(ClientV1RestServlet): if "invite" in content: for u in content["invite"]: - if not isinstance(u, basestring): + if not isinstance(u, string_types): raise SynapseError(400, "Bad invite value.") if len(u) == 0: continue @@ -140,7 +143,7 @@ class PresenceListRestServlet(ClientV1RestServlet): if "drop" in content: for u in content["drop"]: - if not isinstance(u, basestring): + if not isinstance(u, string_types): raise SynapseError(400, "Bad drop value.") if len(u) == 0: continue diff --git a/synapse/rest/client/v1/profile.py b/synapse/rest/client/v1/profile.py index 1a5045c9e..a23edd8fe 100644 --- a/synapse/rest/client/v1/profile.py +++ b/synapse/rest/client/v1/profile.py @@ -16,9 +16,10 @@ """ This module contains REST servlets to do with profile: /profile/ """ from twisted.internet import defer -from .base import ClientV1RestServlet, client_path_patterns -from synapse.types import UserID from synapse.http.servlet import parse_json_object_from_request +from synapse.types import UserID + +from .base import ClientV1RestServlet, client_path_patterns class ProfileDisplaynameRestServlet(ClientV1RestServlet): @@ -26,13 +27,13 @@ class ProfileDisplaynameRestServlet(ClientV1RestServlet): def __init__(self, hs): super(ProfileDisplaynameRestServlet, self).__init__(hs) - self.handlers = hs.get_handlers() + self.profile_handler = hs.get_profile_handler() @defer.inlineCallbacks def on_GET(self, request, user_id): user = UserID.from_string(user_id) - displayname = yield self.handlers.profile_handler.get_displayname( + displayname = yield self.profile_handler.get_displayname( user, ) @@ -52,10 +53,10 @@ class ProfileDisplaynameRestServlet(ClientV1RestServlet): try: new_name = content["displayname"] - except: + except Exception: defer.returnValue((400, "Unable to parse name")) - yield self.handlers.profile_handler.set_displayname( + yield self.profile_handler.set_displayname( user, requester, new_name, is_admin) defer.returnValue((200, {})) @@ -69,13 +70,13 @@ class ProfileAvatarURLRestServlet(ClientV1RestServlet): def __init__(self, hs): super(ProfileAvatarURLRestServlet, self).__init__(hs) - self.handlers = hs.get_handlers() + self.profile_handler = hs.get_profile_handler() @defer.inlineCallbacks def on_GET(self, request, user_id): user = UserID.from_string(user_id) - avatar_url = yield self.handlers.profile_handler.get_avatar_url( + avatar_url = yield self.profile_handler.get_avatar_url( user, ) @@ -94,10 +95,10 @@ class ProfileAvatarURLRestServlet(ClientV1RestServlet): content = parse_json_object_from_request(request) try: new_name = content["avatar_url"] - except: + except Exception: defer.returnValue((400, "Unable to parse name")) - yield self.handlers.profile_handler.set_avatar_url( + yield self.profile_handler.set_avatar_url( user, requester, new_name, is_admin) defer.returnValue((200, {})) @@ -111,16 +112,16 @@ class ProfileRestServlet(ClientV1RestServlet): def __init__(self, hs): super(ProfileRestServlet, self).__init__(hs) - self.handlers = hs.get_handlers() + self.profile_handler = hs.get_profile_handler() @defer.inlineCallbacks def on_GET(self, request, user_id): user = UserID.from_string(user_id) - displayname = yield self.handlers.profile_handler.get_displayname( + displayname = yield self.profile_handler.get_displayname( user, ) - avatar_url = yield self.handlers.profile_handler.get_avatar_url( + avatar_url = yield self.profile_handler.get_avatar_url( user, ) diff --git a/synapse/rest/client/v1/push_rule.py b/synapse/rest/client/v1/push_rule.py index 6bb4821ec..6e95d9bec 100644 --- a/synapse/rest/client/v1/push_rule.py +++ b/synapse/rest/client/v1/push_rule.py @@ -16,16 +16,18 @@ from twisted.internet import defer from synapse.api.errors import ( - SynapseError, UnrecognizedRequestError, NotFoundError, StoreError + NotFoundError, + StoreError, + SynapseError, + UnrecognizedRequestError, ) -from .base import ClientV1RestServlet, client_path_patterns -from synapse.storage.push_rule import ( - InconsistentRuleException, RuleNotFoundException -) -from synapse.push.clientformat import format_push_rules_for_user +from synapse.http.servlet import parse_json_value_from_request, parse_string from synapse.push.baserules import BASE_RULE_IDS +from synapse.push.clientformat import format_push_rules_for_user from synapse.push.rulekinds import PRIORITY_CLASS_MAP -from synapse.http.servlet import parse_json_value_from_request +from synapse.storage.push_rule import InconsistentRuleException, RuleNotFoundException + +from .base import ClientV1RestServlet, client_path_patterns class PushRuleRestServlet(ClientV1RestServlet): @@ -73,13 +75,13 @@ class PushRuleRestServlet(ClientV1RestServlet): except InvalidRuleException as e: raise SynapseError(400, e.message) - before = request.args.get("before", None) + before = parse_string(request, "before") if before: - before = _namespaced_rule_id(spec, before[0]) + before = _namespaced_rule_id(spec, before) - after = request.args.get("after", None) + after = parse_string(request, "after") if after: - after = _namespaced_rule_id(spec, after[0]) + after = _namespaced_rule_id(spec, after) try: yield self.store.add_push_rule( diff --git a/synapse/rest/client/v1/pusher.py b/synapse/rest/client/v1/pusher.py index 9a2ed6ed8..182a68b1e 100644 --- a/synapse/rest/client/v1/pusher.py +++ b/synapse/rest/client/v1/pusher.py @@ -13,20 +13,22 @@ # See the License for the specific language governing permissions and # limitations under the License. +import logging + from twisted.internet import defer -from synapse.api.errors import SynapseError, Codes -from synapse.push import PusherConfigException -from synapse.http.servlet import ( - parse_json_object_from_request, parse_string, RestServlet -) +from synapse.api.errors import Codes, StoreError, SynapseError from synapse.http.server import finish_request -from synapse.api.errors import StoreError +from synapse.http.servlet import ( + RestServlet, + assert_params_in_dict, + parse_json_object_from_request, + parse_string, +) +from synapse.push import PusherConfigException from .base import ClientV1RestServlet, client_path_patterns -import logging - logger = logging.getLogger(__name__) @@ -73,6 +75,7 @@ class PushersSetRestServlet(ClientV1RestServlet): def __init__(self, hs): super(PushersSetRestServlet, self).__init__(hs) self.notifier = hs.get_notifier() + self.pusher_pool = self.hs.get_pusherpool() @defer.inlineCallbacks def on_POST(self, request): @@ -81,25 +84,19 @@ class PushersSetRestServlet(ClientV1RestServlet): content = parse_json_object_from_request(request) - pusher_pool = self.hs.get_pusherpool() - if ('pushkey' in content and 'app_id' in content and 'kind' in content and content['kind'] is None): - yield pusher_pool.remove_pusher( + yield self.pusher_pool.remove_pusher( content['app_id'], content['pushkey'], user_id=user.to_string() ) defer.returnValue((200, {})) - reqd = ['kind', 'app_id', 'app_display_name', - 'device_display_name', 'pushkey', 'lang', 'data'] - missing = [] - for i in reqd: - if i not in content: - missing.append(i) - if len(missing): - raise SynapseError(400, "Missing parameters: " + ','.join(missing), - errcode=Codes.MISSING_PARAM) + assert_params_in_dict( + content, + ['kind', 'app_id', 'app_display_name', + 'device_display_name', 'pushkey', 'lang', 'data'] + ) logger.debug("set pushkey %s to kind %s", content['pushkey'], content['kind']) logger.debug("Got pushers request with body: %r", content) @@ -109,14 +106,14 @@ class PushersSetRestServlet(ClientV1RestServlet): append = content['append'] if not append: - yield pusher_pool.remove_pushers_by_app_id_and_pushkey_not_user( + yield self.pusher_pool.remove_pushers_by_app_id_and_pushkey_not_user( app_id=content['app_id'], pushkey=content['pushkey'], not_user_id=user.to_string() ) try: - yield pusher_pool.add_pusher( + yield self.pusher_pool.add_pusher( user_id=user.to_string(), access_token=requester.access_token_id, kind=content['kind'], @@ -148,10 +145,11 @@ class PushersRemoveRestServlet(RestServlet): SUCCESS_HTML = "You have been unsubscribed" def __init__(self, hs): - super(RestServlet, self).__init__() + super(PushersRemoveRestServlet, self).__init__() self.hs = hs self.notifier = hs.get_notifier() - self.auth = hs.get_v1auth() + self.auth = hs.get_auth() + self.pusher_pool = self.hs.get_pusherpool() @defer.inlineCallbacks def on_GET(self, request): @@ -161,10 +159,8 @@ class PushersRemoveRestServlet(RestServlet): app_id = parse_string(request, "app_id", required=True) pushkey = parse_string(request, "pushkey", required=True) - pusher_pool = self.hs.get_pusherpool() - try: - yield pusher_pool.remove_pusher( + yield self.pusher_pool.remove_pusher( app_id=app_id, pushkey=pushkey, user_id=user.to_string(), @@ -178,7 +174,6 @@ class PushersRemoveRestServlet(RestServlet): request.setResponseCode(200) request.setHeader(b"Content-Type", b"text/html; charset=utf-8") - request.setHeader(b"Server", self.hs.version_string) request.setHeader(b"Content-Length", b"%d" % ( len(PushersRemoveRestServlet.SUCCESS_HTML), )) diff --git a/synapse/rest/client/v1/room.py b/synapse/rest/client/v1/room.py index cd388770c..b7bd878c9 100644 --- a/synapse/rest/client/v1/room.py +++ b/synapse/rest/client/v1/room.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2014-2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -14,22 +15,28 @@ # limitations under the License. """ This module contains REST servlets to do with rooms: /rooms/ """ +import logging + +from six.moves.urllib import parse as urlparse + +from canonicaljson import json + from twisted.internet import defer -from .base import ClientV1RestServlet, client_path_patterns -from synapse.api.errors import SynapseError, Codes, AuthError -from synapse.streams.config import PaginationConfig from synapse.api.constants import EventTypes, Membership +from synapse.api.errors import AuthError, Codes, SynapseError from synapse.api.filtering import Filter -from synapse.types import UserID, RoomID, RoomAlias, ThirdPartyInstanceID -from synapse.events.utils import serialize_event, format_event_for_client_v2 +from synapse.events.utils import format_event_for_client_v2, serialize_event from synapse.http.servlet import ( - parse_json_object_from_request, parse_string, parse_integer + assert_params_in_dict, + parse_integer, + parse_json_object_from_request, + parse_string, ) +from synapse.streams.config import PaginationConfig +from synapse.types import RoomAlias, RoomID, ThirdPartyInstanceID, UserID -import logging -import urllib -import ujson as json +from .base import ClientV1RestServlet, client_path_patterns logger = logging.getLogger(__name__) @@ -39,7 +46,7 @@ class RoomCreateRestServlet(ClientV1RestServlet): def __init__(self, hs): super(RoomCreateRestServlet, self).__init__(hs) - self.handlers = hs.get_handlers() + self._room_creation_handler = hs.get_room_creation_handler() def register(self, http_server): PATTERNS = "/createRoom" @@ -62,8 +69,7 @@ class RoomCreateRestServlet(ClientV1RestServlet): def on_POST(self, request): requester = yield self.auth.get_user_by_req(request) - handler = self.handlers.room_creation_handler - info = yield handler.create_room( + info = yield self._room_creation_handler.create_room( requester, self.get_room_config(request) ) @@ -82,6 +88,9 @@ class RoomStateEventRestServlet(ClientV1RestServlet): def __init__(self, hs): super(RoomStateEventRestServlet, self).__init__(hs) self.handlers = hs.get_handlers() + self.event_creation_hander = hs.get_event_creation_handler() + self.room_member_handler = hs.get_room_member_handler() + self.message_handler = hs.get_message_handler() def register(self, http_server): # /room/$roomid/state/$eventtype @@ -116,7 +125,7 @@ class RoomStateEventRestServlet(ClientV1RestServlet): format = parse_string(request, "format", default="content", allowed_values=["content", "event"]) - msg_handler = self.handlers.message_handler + msg_handler = self.message_handler data = yield msg_handler.get_room_data( user_id=requester.user.to_string(), room_id=room_id, @@ -154,7 +163,7 @@ class RoomStateEventRestServlet(ClientV1RestServlet): if event_type == EventTypes.Member: membership = content.get("membership", None) - event = yield self.handlers.room_member_handler.update_membership( + event = yield self.room_member_handler.update_membership( requester, target=UserID.from_string(state_key), room_id=room_id, @@ -162,16 +171,12 @@ class RoomStateEventRestServlet(ClientV1RestServlet): content=content, ) else: - msg_handler = self.handlers.message_handler - event, context = yield msg_handler.create_event( + event = yield self.event_creation_hander.create_and_send_nonmember_event( requester, event_dict, - token_id=requester.access_token_id, txn_id=txn_id, ) - yield msg_handler.send_nonmember_event(requester, event, context) - ret = {} if event: ret = {"event_id": event.event_id} @@ -183,7 +188,7 @@ class RoomSendEventRestServlet(ClientV1RestServlet): def __init__(self, hs): super(RoomSendEventRestServlet, self).__init__(hs) - self.handlers = hs.get_handlers() + self.event_creation_hander = hs.get_event_creation_handler() def register(self, http_server): # /rooms/$roomid/send/$event_type[/$txn_id] @@ -195,15 +200,19 @@ class RoomSendEventRestServlet(ClientV1RestServlet): requester = yield self.auth.get_user_by_req(request, allow_guest=True) content = parse_json_object_from_request(request) - msg_handler = self.handlers.message_handler - event = yield msg_handler.create_and_send_nonmember_event( + event_dict = { + "type": event_type, + "content": content, + "room_id": room_id, + "sender": requester.user.to_string(), + } + + if 'ts' in request.args and requester.app_service: + event_dict['origin_server_ts'] = parse_integer(request, "ts", 0) + + event = yield self.event_creation_hander.create_and_send_nonmember_event( requester, - { - "type": event_type, - "content": content, - "room_id": room_id, - "sender": requester.user.to_string(), - }, + event_dict, txn_id=txn_id, ) @@ -222,7 +231,7 @@ class RoomSendEventRestServlet(ClientV1RestServlet): class JoinRoomAliasServlet(ClientV1RestServlet): def __init__(self, hs): super(JoinRoomAliasServlet, self).__init__(hs) - self.handlers = hs.get_handlers() + self.room_member_handler = hs.get_room_member_handler() def register(self, http_server): # /join/$room_identifier[/$txn_id] @@ -238,7 +247,7 @@ class JoinRoomAliasServlet(ClientV1RestServlet): try: content = parse_json_object_from_request(request) - except: + except Exception: # Turns out we used to ignore the body entirely, and some clients # cheekily send invalid bodies. content = {} @@ -247,10 +256,10 @@ class JoinRoomAliasServlet(ClientV1RestServlet): room_id = room_identifier try: remote_room_hosts = request.args["server_name"] - except: + except Exception: remote_room_hosts = None elif RoomAlias.is_valid(room_identifier): - handler = self.handlers.room_member_handler + handler = self.room_member_handler room_alias = RoomAlias.from_string(room_identifier) room_id, remote_room_hosts = yield handler.lookup_room_alias(room_alias) room_id = room_id.to_string() @@ -259,7 +268,7 @@ class JoinRoomAliasServlet(ClientV1RestServlet): room_identifier, )) - yield self.handlers.room_member_handler.update_membership( + yield self.room_member_handler.update_membership( requester=requester, target=requester.user, room_id=room_id, @@ -369,14 +378,13 @@ class RoomMemberListRestServlet(ClientV1RestServlet): def __init__(self, hs): super(RoomMemberListRestServlet, self).__init__(hs) - self.handlers = hs.get_handlers() + self.message_handler = hs.get_message_handler() @defer.inlineCallbacks def on_GET(self, request, room_id): # TODO support Pagination stream API (limit/tokens) requester = yield self.auth.get_user_by_req(request) - handler = self.handlers.message_handler - events = yield handler.get_state_events( + events = yield self.message_handler.get_state_events( room_id=room_id, user_id=requester.user.to_string(), ) @@ -398,22 +406,18 @@ class JoinedRoomMemberListRestServlet(ClientV1RestServlet): def __init__(self, hs): super(JoinedRoomMemberListRestServlet, self).__init__(hs) - self.state = hs.get_state_handler() + self.message_handler = hs.get_message_handler() @defer.inlineCallbacks def on_GET(self, request, room_id): - yield self.auth.get_user_by_req(request) + requester = yield self.auth.get_user_by_req(request) - users_with_profile = yield self.state.get_current_user_in_room(room_id) + users_with_profile = yield self.message_handler.get_joined_members( + requester, room_id, + ) defer.returnValue((200, { - "joined": { - user_id: { - "avatar_url": profile.avatar_url, - "display_name": profile.display_name, - } - for user_id, profile in users_with_profile.iteritems() - } + "joined": users_with_profile, })) @@ -423,7 +427,7 @@ class RoomMessageListRestServlet(ClientV1RestServlet): def __init__(self, hs): super(RoomMessageListRestServlet, self).__init__(hs) - self.handlers = hs.get_handlers() + self.pagination_handler = hs.get_pagination_handler() @defer.inlineCallbacks def on_GET(self, request, room_id): @@ -432,14 +436,13 @@ class RoomMessageListRestServlet(ClientV1RestServlet): request, default_limit=10, ) as_client_event = "raw" not in request.args - filter_bytes = request.args.get("filter", None) + filter_bytes = parse_string(request, "filter") if filter_bytes: - filter_json = urllib.unquote(filter_bytes[-1]).decode("UTF-8") + filter_json = urlparse.unquote(filter_bytes).decode("UTF-8") event_filter = Filter(json.loads(filter_json)) else: event_filter = None - handler = self.handlers.message_handler - msgs = yield handler.get_messages( + msgs = yield self.pagination_handler.get_messages( room_id=room_id, requester=requester, pagin_config=pagination_config, @@ -456,14 +459,13 @@ class RoomStateRestServlet(ClientV1RestServlet): def __init__(self, hs): super(RoomStateRestServlet, self).__init__(hs) - self.handlers = hs.get_handlers() + self.message_handler = hs.get_message_handler() @defer.inlineCallbacks def on_GET(self, request, room_id): requester = yield self.auth.get_user_by_req(request, allow_guest=True) - handler = self.handlers.message_handler # Get all the current state for this room - events = yield handler.get_state_events( + events = yield self.message_handler.get_state_events( room_id=room_id, user_id=requester.user.to_string(), is_guest=requester.is_guest, @@ -491,23 +493,45 @@ class RoomInitialSyncRestServlet(ClientV1RestServlet): defer.returnValue((200, content)) -class RoomEventContext(ClientV1RestServlet): +class RoomEventServlet(ClientV1RestServlet): + PATTERNS = client_path_patterns( + "/rooms/(?P[^/]*)/event/(?P[^/]*)$" + ) + + def __init__(self, hs): + super(RoomEventServlet, self).__init__(hs) + self.clock = hs.get_clock() + self.event_handler = hs.get_event_handler() + + @defer.inlineCallbacks + def on_GET(self, request, room_id, event_id): + requester = yield self.auth.get_user_by_req(request) + event = yield self.event_handler.get_event(requester.user, event_id) + + time_now = self.clock.time_msec() + if event: + defer.returnValue((200, serialize_event(event, time_now))) + else: + defer.returnValue((404, "Event not found.")) + + +class RoomEventContextServlet(ClientV1RestServlet): PATTERNS = client_path_patterns( "/rooms/(?P[^/]*)/context/(?P[^/]*)$" ) def __init__(self, hs): - super(RoomEventContext, self).__init__(hs) + super(RoomEventContextServlet, self).__init__(hs) self.clock = hs.get_clock() - self.handlers = hs.get_handlers() + self.room_context_handler = hs.get_room_context_handler() @defer.inlineCallbacks def on_GET(self, request, room_id, event_id): requester = yield self.auth.get_user_by_req(request, allow_guest=True) - limit = int(request.args.get("limit", [10])[0]) + limit = parse_integer(request, "limit", default=10) - results = yield self.handlers.room_context_handler.get_event_context( + results = yield self.room_context_handler.get_event_context( requester.user, room_id, event_id, @@ -537,7 +561,7 @@ class RoomEventContext(ClientV1RestServlet): class RoomForgetRestServlet(ClientV1RestServlet): def __init__(self, hs): super(RoomForgetRestServlet, self).__init__(hs) - self.handlers = hs.get_handlers() + self.room_member_handler = hs.get_room_member_handler() def register(self, http_server): PATTERNS = ("/rooms/(?P[^/]*)/forget") @@ -550,7 +574,7 @@ class RoomForgetRestServlet(ClientV1RestServlet): allow_guest=False, ) - yield self.handlers.room_member_handler.forget( + yield self.room_member_handler.forget( user=requester.user, room_id=room_id, ) @@ -568,12 +592,12 @@ class RoomMembershipRestServlet(ClientV1RestServlet): def __init__(self, hs): super(RoomMembershipRestServlet, self).__init__(hs) - self.handlers = hs.get_handlers() + self.room_member_handler = hs.get_room_member_handler() def register(self, http_server): # /rooms/$roomid/[invite|join|leave] PATTERNS = ("/rooms/(?P[^/]*)/" - "(?Pjoin|invite|leave|ban|unban|kick|forget)") + "(?Pjoin|invite|leave|ban|unban|kick)") register_txn_path(self, PATTERNS, http_server) @defer.inlineCallbacks @@ -591,13 +615,13 @@ class RoomMembershipRestServlet(ClientV1RestServlet): try: content = parse_json_object_from_request(request) - except: + except Exception: # Turns out we used to ignore the body entirely, and some clients # cheekily send invalid bodies. content = {} if membership_action == "invite" and self._has_3pid_invite_keys(content): - yield self.handlers.room_member_handler.do_3pid_invite( + yield self.room_member_handler.do_3pid_invite( room_id, requester.user, content["medium"], @@ -611,15 +635,14 @@ class RoomMembershipRestServlet(ClientV1RestServlet): target = requester.user if membership_action in ["invite", "ban", "unban", "kick"]: - if "user_id" not in content: - raise SynapseError(400, "Missing user_id key.") + assert_params_in_dict(content, ["user_id"]) target = UserID.from_string(content["user_id"]) event_content = None if 'reason' in content and membership_action in ['kick', 'ban']: event_content = {'reason': content['reason']} - yield self.handlers.room_member_handler.update_membership( + yield self.room_member_handler.update_membership( requester=requester, target=target, room_id=room_id, @@ -629,7 +652,12 @@ class RoomMembershipRestServlet(ClientV1RestServlet): content=event_content, ) - defer.returnValue((200, {})) + return_value = {} + + if membership_action == "join": + return_value["room_id"] = room_id + + defer.returnValue((200, return_value)) def _has_3pid_invite_keys(self, content): for key in {"id_server", "medium", "address"}: @@ -647,6 +675,7 @@ class RoomRedactEventRestServlet(ClientV1RestServlet): def __init__(self, hs): super(RoomRedactEventRestServlet, self).__init__(hs) self.handlers = hs.get_handlers() + self.event_creation_handler = hs.get_event_creation_handler() def register(self, http_server): PATTERNS = ("/rooms/(?P[^/]*)/redact/(?P[^/]*)") @@ -657,8 +686,7 @@ class RoomRedactEventRestServlet(ClientV1RestServlet): requester = yield self.auth.get_user_by_req(request) content = parse_json_object_from_request(request) - msg_handler = self.handlers.message_handler - event = yield msg_handler.create_and_send_nonmember_event( + event = yield self.event_creation_handler.create_and_send_nonmember_event( requester, { "type": EventTypes.Redaction, @@ -692,8 +720,8 @@ class RoomTypingRestServlet(ClientV1RestServlet): def on_PUT(self, request, room_id, user_id): requester = yield self.auth.get_user_by_req(request) - room_id = urllib.unquote(room_id) - target_user = UserID.from_string(urllib.unquote(user_id)) + room_id = urlparse.unquote(room_id) + target_user = UserID.from_string(urlparse.unquote(user_id)) content = parse_json_object_from_request(request) @@ -734,7 +762,7 @@ class SearchRestServlet(ClientV1RestServlet): content = parse_json_object_from_request(request) - batch = request.args.get("next_batch", [None])[0] + batch = parse_string(request, "next_batch") results = yield self.handlers.search_handler.search( requester.user, content, @@ -802,9 +830,13 @@ def register_servlets(hs, http_server): RoomSendEventRestServlet(hs).register(http_server) PublicRoomListRestServlet(hs).register(http_server) RoomStateRestServlet(hs).register(http_server) - RoomInitialSyncRestServlet(hs).register(http_server) RoomRedactEventRestServlet(hs).register(http_server) RoomTypingRestServlet(hs).register(http_server) SearchRestServlet(hs).register(http_server) JoinedRoomsRestServlet(hs).register(http_server) - RoomEventContext(hs).register(http_server) + RoomEventServlet(hs).register(http_server) + RoomEventContextServlet(hs).register(http_server) + + +def register_deprecated_servlets(hs, http_server): + RoomInitialSyncRestServlet(hs).register(http_server) diff --git a/synapse/rest/client/v1/voip.py b/synapse/rest/client/v1/voip.py index c43b30b73..62f4c3d93 100644 --- a/synapse/rest/client/v1/voip.py +++ b/synapse/rest/client/v1/voip.py @@ -13,16 +13,15 @@ # See the License for the specific language governing permissions and # limitations under the License. +import base64 +import hashlib +import hmac + from twisted.internet import defer from .base import ClientV1RestServlet, client_path_patterns -import hmac -import hashlib -import base64 - - class VoipRestServlet(ClientV1RestServlet): PATTERNS = client_path_patterns("/voip/turnServer$") diff --git a/synapse/rest/client/v1_only/__init__.py b/synapse/rest/client/v1_only/__init__.py new file mode 100644 index 000000000..936f902ac --- /dev/null +++ b/synapse/rest/client/v1_only/__init__.py @@ -0,0 +1,3 @@ +""" +REST APIs that are only used in v1 (the legacy API). +""" diff --git a/synapse/rest/client/v1_only/base.py b/synapse/rest/client/v1_only/base.py new file mode 100644 index 000000000..9d4db7437 --- /dev/null +++ b/synapse/rest/client/v1_only/base.py @@ -0,0 +1,39 @@ +# -*- coding: utf-8 -*- +# Copyright 2014-2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +"""This module contains base REST classes for constructing client v1 servlets. +""" + +import re + +from synapse.api.urls import CLIENT_PREFIX + + +def v1_only_client_path_patterns(path_regex, include_in_unstable=True): + """Creates a regex compiled client path with the correct client path + prefix. + + Args: + path_regex (str): The regex string to match. This should NOT have a ^ + as this will be prefixed. + Returns: + list of SRE_Pattern + """ + patterns = [re.compile("^" + CLIENT_PREFIX + path_regex)] + if include_in_unstable: + unstable_prefix = CLIENT_PREFIX.replace("/api/v1", "/unstable") + patterns.append(re.compile("^" + unstable_prefix + path_regex)) + return patterns diff --git a/synapse/rest/client/v1/register.py b/synapse/rest/client/v1_only/register.py similarity index 86% rename from synapse/rest/client/v1/register.py rename to synapse/rest/client/v1_only/register.py index ecf7e311a..3439c3c6d 100644 --- a/synapse/rest/client/v1/register.py +++ b/synapse/rest/client/v1_only/register.py @@ -14,21 +14,20 @@ # limitations under the License. """This module contains REST servlets to do with registration: /register""" -from twisted.internet import defer - -from synapse.api.errors import SynapseError, Codes -from synapse.api.constants import LoginType -from synapse.api.auth import get_access_token_from_request -from .base import ClientV1RestServlet, client_path_patterns -import synapse.util.stringutils as stringutils -from synapse.http.servlet import parse_json_object_from_request -from synapse.types import create_requester - -from synapse.util.async import run_on_reactor - -from hashlib import sha1 import hmac import logging +from hashlib import sha1 + +from twisted.internet import defer + +import synapse.util.stringutils as stringutils +from synapse.api.constants import LoginType +from synapse.api.errors import Codes, SynapseError +from synapse.http.servlet import assert_params_in_dict, parse_json_object_from_request +from synapse.rest.client.v1.base import ClientV1RestServlet +from synapse.types import create_requester + +from .base import v1_only_client_path_patterns logger = logging.getLogger(__name__) @@ -51,7 +50,7 @@ class RegisterRestServlet(ClientV1RestServlet): handler doesn't have a concept of multi-stages or sessions. """ - PATTERNS = client_path_patterns("/register$", releases=(), include_in_unstable=False) + PATTERNS = v1_only_client_path_patterns("/register$", include_in_unstable=False) def __init__(self, hs): """ @@ -66,14 +65,20 @@ class RegisterRestServlet(ClientV1RestServlet): # TODO: persistent storage self.sessions = {} self.enable_registration = hs.config.enable_registration + self.auth = hs.get_auth() self.auth_handler = hs.get_auth_handler() self.handlers = hs.get_handlers() def on_GET(self, request): + + require_email = 'email' in self.hs.config.registrations_require_3pid + require_msisdn = 'msisdn' in self.hs.config.registrations_require_3pid + + flows = [] if self.hs.config.enable_registration_captcha: - return ( - 200, - {"flows": [ + # only support the email-only flow if we don't require MSISDN 3PIDs + if not require_msisdn: + flows.extend([ { "type": LoginType.RECAPTCHA, "stages": [ @@ -82,27 +87,34 @@ class RegisterRestServlet(ClientV1RestServlet): LoginType.PASSWORD ] }, + ]) + # only support 3PIDless registration if no 3PIDs are required + if not require_email and not require_msisdn: + flows.extend([ { "type": LoginType.RECAPTCHA, "stages": [LoginType.RECAPTCHA, LoginType.PASSWORD] } - ]} - ) + ]) else: - return ( - 200, - {"flows": [ + # only support the email-only flow if we don't require MSISDN 3PIDs + if require_email or not require_msisdn: + flows.extend([ { "type": LoginType.EMAIL_IDENTITY, "stages": [ LoginType.EMAIL_IDENTITY, LoginType.PASSWORD ] - }, + } + ]) + # only support 3PIDless registration if no 3PIDs are required + if not require_email and not require_msisdn: + flows.extend([ { "type": LoginType.PASSWORD } - ]} - ) + ]) + return (200, {"flows": flows}) @defer.inlineCallbacks def on_POST(self, request): @@ -111,8 +123,7 @@ class RegisterRestServlet(ClientV1RestServlet): session = (register_json["session"] if "session" in register_json else None) login_type = None - if "type" not in register_json: - raise SynapseError(400, "Missing 'type' key.") + assert_params_in_dict(register_json, ["type"]) try: login_type = register_json["type"] @@ -258,7 +269,6 @@ class RegisterRestServlet(ClientV1RestServlet): @defer.inlineCallbacks def _do_password(self, request, register_json, session): - yield run_on_reactor() if (self.hs.config.enable_registration_captcha and not session[LoginType.RECAPTCHA]): # captcha should've been done by this stage! @@ -298,11 +308,9 @@ class RegisterRestServlet(ClientV1RestServlet): @defer.inlineCallbacks def _do_app_service(self, request, register_json, session): - as_token = get_access_token_from_request(request) - - if "user" not in register_json: - raise SynapseError(400, "Expected 'user' key.") + as_token = self.auth.get_access_token_from_request(request) + assert_params_in_dict(register_json, ["user"]) user_localpart = register_json["user"].encode("utf-8") handler = self.handlers.registration_handler @@ -319,14 +327,7 @@ class RegisterRestServlet(ClientV1RestServlet): @defer.inlineCallbacks def _do_shared_secret(self, request, register_json, session): - yield run_on_reactor() - - if not isinstance(register_json.get("mac", None), basestring): - raise SynapseError(400, "Expected mac.") - if not isinstance(register_json.get("user", None), basestring): - raise SynapseError(400, "Expected 'user' key.") - if not isinstance(register_json.get("password", None), basestring): - raise SynapseError(400, "Expected 'password' key.") + assert_params_in_dict(register_json, ["mac", "user", "password"]) if not self.hs.config.registration_shared_secret: raise SynapseError(400, "Shared secret registration is not enabled") @@ -336,9 +337,9 @@ class RegisterRestServlet(ClientV1RestServlet): admin = register_json.get("admin", None) # Its important to check as we use null bytes as HMAC field separators - if "\x00" in user: + if b"\x00" in user: raise SynapseError(400, "Invalid user") - if "\x00" in password: + if b"\x00" in password: raise SynapseError(400, "Invalid password") # str() because otherwise hmac complains that 'unicode' does not @@ -346,20 +347,20 @@ class RegisterRestServlet(ClientV1RestServlet): got_mac = str(register_json["mac"]) want_mac = hmac.new( - key=self.hs.config.registration_shared_secret, + key=self.hs.config.registration_shared_secret.encode(), digestmod=sha1, ) want_mac.update(user) - want_mac.update("\x00") + want_mac.update(b"\x00") want_mac.update(password) - want_mac.update("\x00") - want_mac.update("admin" if admin else "notadmin") + want_mac.update(b"\x00") + want_mac.update(b"admin" if admin else b"notadmin") want_mac = want_mac.hexdigest() if compare_digest(want_mac, got_mac): handler = self.handlers.registration_handler user_id, token = yield handler.register( - localpart=user, + localpart=user.lower(), password=password, admin=bool(admin), ) @@ -379,7 +380,7 @@ class CreateUserRestServlet(ClientV1RestServlet): """Handles user creation via a server-to-server interface """ - PATTERNS = client_path_patterns("/createUser$", releases=()) + PATTERNS = v1_only_client_path_patterns("/createUser$") def __init__(self, hs): super(CreateUserRestServlet, self).__init__(hs) @@ -390,7 +391,7 @@ class CreateUserRestServlet(ClientV1RestServlet): def on_POST(self, request): user_json = parse_json_object_from_request(request) - access_token = get_access_token_from_request(request) + access_token = self.auth.get_access_token_from_request(request) app_service = self.store.get_app_service_by_token( access_token ) @@ -409,13 +410,7 @@ class CreateUserRestServlet(ClientV1RestServlet): @defer.inlineCallbacks def _do_create(self, requester, user_json): - yield run_on_reactor() - - if "localpart" not in user_json: - raise SynapseError(400, "Expected 'localpart' key.") - - if "displayname" not in user_json: - raise SynapseError(400, "Expected 'displayname' key.") + assert_params_in_dict(user_json, ["localpart", "displayname"]) localpart = user_json["localpart"].encode("utf-8") displayname = user_json["displayname"].encode("utf-8") diff --git a/synapse/rest/client/v2_alpha/_base.py b/synapse/rest/client/v2_alpha/_base.py index 1f5bc24cc..77434937f 100644 --- a/synapse/rest/client/v2_alpha/_base.py +++ b/synapse/rest/client/v2_alpha/_base.py @@ -15,12 +15,13 @@ """This module contains base REST classes for constructing client v1 servlets. """ - -from synapse.api.urls import CLIENT_V2_ALPHA_PREFIX +import logging import re -import logging +from twisted.internet import defer +from synapse.api.errors import InteractiveAuthIncompleteError +from synapse.api.urls import CLIENT_V2_ALPHA_PREFIX logger = logging.getLogger(__name__) @@ -57,3 +58,37 @@ def set_timeline_upper_limit(filter_json, filter_timeline_limit): filter_json['room']['timeline']["limit"] = min( filter_json['room']['timeline']['limit'], filter_timeline_limit) + + +def interactive_auth_handler(orig): + """Wraps an on_POST method to handle InteractiveAuthIncompleteErrors + + Takes a on_POST method which returns a deferred (errcode, body) response + and adds exception handling to turn a InteractiveAuthIncompleteError into + a 401 response. + + Normal usage is: + + @interactive_auth_handler + @defer.inlineCallbacks + def on_POST(self, request): + # ... + yield self.auth_handler.check_auth + """ + def wrapped(*args, **kwargs): + res = defer.maybeDeferred(orig, *args, **kwargs) + res.addErrback(_catch_incomplete_interactive_auth) + return res + return wrapped + + +def _catch_incomplete_interactive_auth(f): + """helper for interactive_auth_handler + + Catches InteractiveAuthIncompleteErrors and turns them into 401 responses + + Args: + f (failure.Failure): + """ + f.trap(InteractiveAuthIncompleteError) + return 401, f.value.result diff --git a/synapse/rest/client/v2_alpha/account.py b/synapse/rest/client/v2_alpha/account.py index 4990b22b9..eeae466d8 100644 --- a/synapse/rest/client/v2_alpha/account.py +++ b/synapse/rest/client/v2_alpha/account.py @@ -1,6 +1,7 @@ # -*- coding: utf-8 -*- # Copyright 2015, 2016 OpenMarket Ltd # Copyright 2017 Vector Creations Ltd +# Copyright 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -13,21 +14,23 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. +import logging + +from six.moves import http_client from twisted.internet import defer from synapse.api.constants import LoginType -from synapse.api.errors import LoginError, SynapseError, Codes +from synapse.api.errors import Codes, SynapseError from synapse.http.servlet import ( - RestServlet, parse_json_object_from_request, assert_params_in_request + RestServlet, + assert_params_in_dict, + parse_json_object_from_request, ) -from synapse.util.async import run_on_reactor from synapse.util.msisdn import phone_number_to_msisdn +from synapse.util.threepids import check_3pid_allowed -from ._base import client_v2_patterns - -import logging - +from ._base import client_v2_patterns, interactive_auth_handler logger = logging.getLogger(__name__) @@ -44,10 +47,15 @@ class EmailPasswordRequestTokenRestServlet(RestServlet): def on_POST(self, request): body = parse_json_object_from_request(request) - assert_params_in_request(body, [ + assert_params_in_dict(body, [ 'id_server', 'client_secret', 'email', 'send_attempt' ]) + if not check_3pid_allowed(self.hs, "email", body['email']): + raise SynapseError( + 403, "Third party identifier is not allowed", Codes.THREEPID_DENIED, + ) + existingUid = yield self.hs.get_datastore().get_user_id_by_threepid( 'email', body['email'] ) @@ -72,13 +80,18 @@ class MsisdnPasswordRequestTokenRestServlet(RestServlet): def on_POST(self, request): body = parse_json_object_from_request(request) - assert_params_in_request(body, [ + assert_params_in_dict(body, [ 'id_server', 'client_secret', 'country', 'phone_number', 'send_attempt', ]) msisdn = phone_number_to_msisdn(body['country'], body['phone_number']) + if not check_3pid_allowed(self.hs, "msisdn", msisdn): + raise SynapseError( + 403, "Third party identifier is not allowed", Codes.THREEPID_DENIED, + ) + existingUid = yield self.datastore.get_user_id_by_threepid( 'msisdn', msisdn ) @@ -99,56 +112,60 @@ class PasswordRestServlet(RestServlet): self.auth = hs.get_auth() self.auth_handler = hs.get_auth_handler() self.datastore = self.hs.get_datastore() + self._set_password_handler = hs.get_set_password_handler() + @interactive_auth_handler @defer.inlineCallbacks def on_POST(self, request): - yield run_on_reactor() - body = parse_json_object_from_request(request) - authed, result, params, _ = yield self.auth_handler.check_auth([ - [LoginType.PASSWORD], - [LoginType.EMAIL_IDENTITY], - [LoginType.MSISDN], - ], body, self.hs.get_ip_from_request(request)) + # there are two possibilities here. Either the user does not have an + # access token, and needs to do a password reset; or they have one and + # need to validate their identity. + # + # In the first case, we offer a couple of means of identifying + # themselves (email and msisdn, though it's unclear if msisdn actually + # works). + # + # In the second case, we require a password to confirm their identity. - if not authed: - defer.returnValue((401, result)) - - user_id = None - requester = None - - if LoginType.PASSWORD in result: - # if using password, they should also be logged in + if self.auth.has_access_token(request): requester = yield self.auth.get_user_by_req(request) - user_id = requester.user.to_string() - if user_id != result[LoginType.PASSWORD]: - raise LoginError(400, "", Codes.UNKNOWN) - elif LoginType.EMAIL_IDENTITY in result: - threepid = result[LoginType.EMAIL_IDENTITY] - if 'medium' not in threepid or 'address' not in threepid: - raise SynapseError(500, "Malformed threepid") - if threepid['medium'] == 'email': - # For emails, transform the address to lowercase. - # We store all email addreses as lowercase in the DB. - # (See add_threepid in synapse/handlers/auth.py) - threepid['address'] = threepid['address'].lower() - # if using email, we must know about the email they're authing with! - threepid_user_id = yield self.datastore.get_user_id_by_threepid( - threepid['medium'], threepid['address'] + params = yield self.auth_handler.validate_user_via_ui_auth( + requester, body, self.hs.get_ip_from_request(request), ) - if not threepid_user_id: - raise SynapseError(404, "Email address not found", Codes.NOT_FOUND) - user_id = threepid_user_id + user_id = requester.user.to_string() else: - logger.error("Auth succeeded but no known type!", result.keys()) - raise SynapseError(500, "", Codes.UNKNOWN) + requester = None + result, params, _ = yield self.auth_handler.check_auth( + [[LoginType.EMAIL_IDENTITY], [LoginType.MSISDN]], + body, self.hs.get_ip_from_request(request), + ) - if 'new_password' not in params: - raise SynapseError(400, "", Codes.MISSING_PARAM) + if LoginType.EMAIL_IDENTITY in result: + threepid = result[LoginType.EMAIL_IDENTITY] + if 'medium' not in threepid or 'address' not in threepid: + raise SynapseError(500, "Malformed threepid") + if threepid['medium'] == 'email': + # For emails, transform the address to lowercase. + # We store all email addreses as lowercase in the DB. + # (See add_threepid in synapse/handlers/auth.py) + threepid['address'] = threepid['address'].lower() + # if using email, we must know about the email they're authing with! + threepid_user_id = yield self.datastore.get_user_id_by_threepid( + threepid['medium'], threepid['address'] + ) + if not threepid_user_id: + raise SynapseError(404, "Email address not found", Codes.NOT_FOUND) + user_id = threepid_user_id + else: + logger.error("Auth succeeded but no known type! %r", result.keys()) + raise SynapseError(500, "", Codes.UNKNOWN) + + assert_params_in_dict(params, ["new_password"]) new_password = params['new_password'] - yield self.auth_handler.set_password( + yield self._set_password_handler.set_password( user_id, new_password, requester ) @@ -162,42 +179,39 @@ class DeactivateAccountRestServlet(RestServlet): PATTERNS = client_v2_patterns("/account/deactivate$") def __init__(self, hs): + super(DeactivateAccountRestServlet, self).__init__() self.hs = hs - self.store = hs.get_datastore() self.auth = hs.get_auth() self.auth_handler = hs.get_auth_handler() - super(DeactivateAccountRestServlet, self).__init__() + self._deactivate_account_handler = hs.get_deactivate_account_handler() + @interactive_auth_handler @defer.inlineCallbacks def on_POST(self, request): body = parse_json_object_from_request(request) + erase = body.get("erase", False) + if not isinstance(erase, bool): + raise SynapseError( + http_client.BAD_REQUEST, + "Param 'erase' must be a boolean, if given", + Codes.BAD_JSON, + ) - authed, result, params, _ = yield self.auth_handler.check_auth([ - [LoginType.PASSWORD], - ], body, self.hs.get_ip_from_request(request)) + requester = yield self.auth.get_user_by_req(request) - if not authed: - defer.returnValue((401, result)) - - user_id = None - requester = None - - if LoginType.PASSWORD in result: - # if using password, they should also be logged in - requester = yield self.auth.get_user_by_req(request) - user_id = requester.user.to_string() - if user_id != result[LoginType.PASSWORD]: - raise LoginError(400, "", Codes.UNKNOWN) - else: - logger.error("Auth succeeded but no known type!", result.keys()) - raise SynapseError(500, "", Codes.UNKNOWN) - - # FIXME: Theoretically there is a race here wherein user resets password - # using threepid. - yield self.store.user_delete_access_tokens(user_id) - yield self.store.user_delete_threepids(user_id) - yield self.store.user_set_password_hash(user_id, None) + # allow ASes to dectivate their own users + if requester.app_service: + yield self._deactivate_account_handler.deactivate_account( + requester.user.to_string(), erase, + ) + defer.returnValue((200, {})) + yield self.auth_handler.validate_user_via_ui_auth( + requester, body, self.hs.get_ip_from_request(request), + ) + yield self._deactivate_account_handler.deactivate_account( + requester.user.to_string(), erase, + ) defer.returnValue((200, {})) @@ -213,15 +227,15 @@ class EmailThreepidRequestTokenRestServlet(RestServlet): @defer.inlineCallbacks def on_POST(self, request): body = parse_json_object_from_request(request) + assert_params_in_dict( + body, + ['id_server', 'client_secret', 'email', 'send_attempt'], + ) - required = ['id_server', 'client_secret', 'email', 'send_attempt'] - absent = [] - for k in required: - if k not in body: - absent.append(k) - - if absent: - raise SynapseError(400, "Missing params: %r" % absent, Codes.MISSING_PARAM) + if not check_3pid_allowed(self.hs, "email", body['email']): + raise SynapseError( + 403, "Third party identifier is not allowed", Codes.THREEPID_DENIED, + ) existingUid = yield self.datastore.get_user_id_by_threepid( 'email', body['email'] @@ -246,21 +260,18 @@ class MsisdnThreepidRequestTokenRestServlet(RestServlet): @defer.inlineCallbacks def on_POST(self, request): body = parse_json_object_from_request(request) - - required = [ + assert_params_in_dict(body, [ 'id_server', 'client_secret', 'country', 'phone_number', 'send_attempt', - ] - absent = [] - for k in required: - if k not in body: - absent.append(k) - - if absent: - raise SynapseError(400, "Missing params: %r" % absent, Codes.MISSING_PARAM) + ]) msisdn = phone_number_to_msisdn(body['country'], body['phone_number']) + if not check_3pid_allowed(self.hs, "msisdn", msisdn): + raise SynapseError( + 403, "Third party identifier is not allowed", Codes.THREEPID_DENIED, + ) + existingUid = yield self.datastore.get_user_id_by_threepid( 'msisdn', msisdn ) @@ -285,8 +296,6 @@ class ThreepidRestServlet(RestServlet): @defer.inlineCallbacks def on_GET(self, request): - yield run_on_reactor() - requester = yield self.auth.get_user_by_req(request) threepids = yield self.datastore.user_get_threepids( @@ -297,8 +306,6 @@ class ThreepidRestServlet(RestServlet): @defer.inlineCallbacks def on_POST(self, request): - yield run_on_reactor() - body = parse_json_object_from_request(request) threePidCreds = body.get('threePidCreds') @@ -350,29 +357,40 @@ class ThreepidDeleteRestServlet(RestServlet): @defer.inlineCallbacks def on_POST(self, request): - yield run_on_reactor() - body = parse_json_object_from_request(request) - - required = ['medium', 'address'] - absent = [] - for k in required: - if k not in body: - absent.append(k) - - if absent: - raise SynapseError(400, "Missing params: %r" % absent, Codes.MISSING_PARAM) + assert_params_in_dict(body, ['medium', 'address']) requester = yield self.auth.get_user_by_req(request) user_id = requester.user.to_string() - yield self.auth_handler.delete_threepid( - user_id, body['medium'], body['address'] - ) + try: + yield self.auth_handler.delete_threepid( + user_id, body['medium'], body['address'] + ) + except Exception: + # NB. This endpoint should succeed if there is nothing to + # delete, so it should only throw if something is wrong + # that we ought to care about. + logger.exception("Failed to remove threepid") + raise SynapseError(500, "Failed to remove threepid") defer.returnValue((200, {})) +class WhoamiRestServlet(RestServlet): + PATTERNS = client_v2_patterns("/account/whoami$") + + def __init__(self, hs): + super(WhoamiRestServlet, self).__init__() + self.auth = hs.get_auth() + + @defer.inlineCallbacks + def on_GET(self, request): + requester = yield self.auth.get_user_by_req(request) + + defer.returnValue((200, {'user_id': requester.user.to_string()})) + + def register_servlets(hs, http_server): EmailPasswordRequestTokenRestServlet(hs).register(http_server) MsisdnPasswordRequestTokenRestServlet(hs).register(http_server) @@ -382,3 +400,4 @@ def register_servlets(hs, http_server): MsisdnThreepidRequestTokenRestServlet(hs).register(http_server) ThreepidRestServlet(hs).register(http_server) ThreepidDeleteRestServlet(hs).register(http_server) + WhoamiRestServlet(hs).register(http_server) diff --git a/synapse/rest/client/v2_alpha/account_data.py b/synapse/rest/client/v2_alpha/account_data.py index 0e0a187ef..371e9aa35 100644 --- a/synapse/rest/client/v2_alpha/account_data.py +++ b/synapse/rest/client/v2_alpha/account_data.py @@ -13,14 +13,14 @@ # See the License for the specific language governing permissions and # limitations under the License. -from ._base import client_v2_patterns - -from synapse.http.servlet import RestServlet, parse_json_object_from_request -from synapse.api.errors import AuthError, SynapseError +import logging from twisted.internet import defer -import logging +from synapse.api.errors import AuthError, SynapseError +from synapse.http.servlet import RestServlet, parse_json_object_from_request + +from ._base import client_v2_patterns logger = logging.getLogger(__name__) diff --git a/synapse/rest/client/v2_alpha/auth.py b/synapse/rest/client/v2_alpha/auth.py index 8e5577148..bd8b5f4af 100644 --- a/synapse/rest/client/v2_alpha/auth.py +++ b/synapse/rest/client/v2_alpha/auth.py @@ -13,6 +13,8 @@ # See the License for the specific language governing permissions and # limitations under the License. +import logging + from twisted.internet import defer from synapse.api.constants import LoginType @@ -23,9 +25,6 @@ from synapse.http.servlet import RestServlet from ._base import client_v2_patterns -import logging - - logger = logging.getLogger(__name__) RECAPTCHA_TEMPLATE = """ @@ -129,7 +128,6 @@ class AuthRestServlet(RestServlet): html_bytes = html.encode("utf8") request.setResponseCode(200) request.setHeader(b"Content-Type", b"text/html; charset=utf-8") - request.setHeader(b"Server", self.hs.version_string) request.setHeader(b"Content-Length", b"%d" % (len(html_bytes),)) request.write(html_bytes) @@ -175,7 +173,6 @@ class AuthRestServlet(RestServlet): html_bytes = html.encode("utf8") request.setResponseCode(200) request.setHeader(b"Content-Type", b"text/html; charset=utf-8") - request.setHeader(b"Server", self.hs.version_string) request.setHeader(b"Content-Length", b"%d" % (len(html_bytes),)) request.write(html_bytes) diff --git a/synapse/rest/client/v2_alpha/devices.py b/synapse/rest/client/v2_alpha/devices.py index b57ba95d2..9b75bb137 100644 --- a/synapse/rest/client/v2_alpha/devices.py +++ b/synapse/rest/client/v2_alpha/devices.py @@ -17,15 +17,20 @@ import logging from twisted.internet import defer -from synapse.api import constants, errors -from synapse.http import servlet -from ._base import client_v2_patterns +from synapse.api import errors +from synapse.http.servlet import ( + RestServlet, + assert_params_in_dict, + parse_json_object_from_request, +) + +from ._base import client_v2_patterns, interactive_auth_handler logger = logging.getLogger(__name__) -class DevicesRestServlet(servlet.RestServlet): - PATTERNS = client_v2_patterns("/devices$", releases=[], v2_alpha=False) +class DevicesRestServlet(RestServlet): + PATTERNS = client_v2_patterns("/devices$", v2_alpha=False) def __init__(self, hs): """ @@ -46,12 +51,12 @@ class DevicesRestServlet(servlet.RestServlet): defer.returnValue((200, {"devices": devices})) -class DeleteDevicesRestServlet(servlet.RestServlet): +class DeleteDevicesRestServlet(RestServlet): """ API for bulk deletion of devices. Accepts a JSON object with a devices key which lists the device_ids to delete. Requires user interactive auth. """ - PATTERNS = client_v2_patterns("/delete_devices", releases=[], v2_alpha=False) + PATTERNS = client_v2_patterns("/delete_devices", v2_alpha=False) def __init__(self, hs): super(DeleteDevicesRestServlet, self).__init__() @@ -60,31 +65,28 @@ class DeleteDevicesRestServlet(servlet.RestServlet): self.device_handler = hs.get_device_handler() self.auth_handler = hs.get_auth_handler() + @interactive_auth_handler @defer.inlineCallbacks def on_POST(self, request): + requester = yield self.auth.get_user_by_req(request) + try: - body = servlet.parse_json_object_from_request(request) + body = parse_json_object_from_request(request) except errors.SynapseError as e: if e.errcode == errors.Codes.NOT_JSON: - # deal with older clients which didn't pass a J*DELETESON dict + # DELETE + # deal with older clients which didn't pass a JSON dict # the same as those that pass an empty dict body = {} else: raise e - if 'devices' not in body: - raise errors.SynapseError( - 400, "No devices supplied", errcode=errors.Codes.MISSING_PARAM - ) + assert_params_in_dict(body, ["devices"]) - authed, result, params, _ = yield self.auth_handler.check_auth([ - [constants.LoginType.PASSWORD], - ], body, self.hs.get_ip_from_request(request)) + yield self.auth_handler.validate_user_via_ui_auth( + requester, body, self.hs.get_ip_from_request(request), + ) - if not authed: - defer.returnValue((401, result)) - - requester = yield self.auth.get_user_by_req(request) yield self.device_handler.delete_devices( requester.user.to_string(), body['devices'], @@ -92,9 +94,8 @@ class DeleteDevicesRestServlet(servlet.RestServlet): defer.returnValue((200, {})) -class DeviceRestServlet(servlet.RestServlet): - PATTERNS = client_v2_patterns("/devices/(?P[^/]*)$", - releases=[], v2_alpha=False) +class DeviceRestServlet(RestServlet): + PATTERNS = client_v2_patterns("/devices/(?P[^/]*)$", v2_alpha=False) def __init__(self, hs): """ @@ -116,10 +117,13 @@ class DeviceRestServlet(servlet.RestServlet): ) defer.returnValue((200, device)) + @interactive_auth_handler @defer.inlineCallbacks def on_DELETE(self, request, device_id): + requester = yield self.auth.get_user_by_req(request) + try: - body = servlet.parse_json_object_from_request(request) + body = parse_json_object_from_request(request) except errors.SynapseError as e: if e.errcode == errors.Codes.NOT_JSON: @@ -129,17 +133,12 @@ class DeviceRestServlet(servlet.RestServlet): else: raise - authed, result, params, _ = yield self.auth_handler.check_auth([ - [constants.LoginType.PASSWORD], - ], body, self.hs.get_ip_from_request(request)) + yield self.auth_handler.validate_user_via_ui_auth( + requester, body, self.hs.get_ip_from_request(request), + ) - if not authed: - defer.returnValue((401, result)) - - requester = yield self.auth.get_user_by_req(request) yield self.device_handler.delete_device( - requester.user.to_string(), - device_id, + requester.user.to_string(), device_id, ) defer.returnValue((200, {})) @@ -147,7 +146,7 @@ class DeviceRestServlet(servlet.RestServlet): def on_PUT(self, request, device_id): requester = yield self.auth.get_user_by_req(request, allow_guest=True) - body = servlet.parse_json_object_from_request(request) + body = parse_json_object_from_request(request) yield self.device_handler.update_device( requester.user.to_string(), device_id, diff --git a/synapse/rest/client/v2_alpha/filter.py b/synapse/rest/client/v2_alpha/filter.py index d2b2fd66e..ae8672887 100644 --- a/synapse/rest/client/v2_alpha/filter.py +++ b/synapse/rest/client/v2_alpha/filter.py @@ -13,17 +13,15 @@ # See the License for the specific language governing permissions and # limitations under the License. +import logging + from twisted.internet import defer -from synapse.api.errors import AuthError, SynapseError, StoreError, Codes +from synapse.api.errors import AuthError, Codes, StoreError, SynapseError from synapse.http.servlet import RestServlet, parse_json_object_from_request from synapse.types import UserID -from ._base import client_v2_patterns -from ._base import set_timeline_upper_limit - -import logging - +from ._base import client_v2_patterns, set_timeline_upper_limit logger = logging.getLogger(__name__) @@ -50,7 +48,7 @@ class GetFilterRestServlet(RestServlet): try: filter_id = int(filter_id) - except: + except Exception: raise SynapseError(400, "Invalid filter_id") try: diff --git a/synapse/rest/client/v2_alpha/groups.py b/synapse/rest/client/v2_alpha/groups.py new file mode 100644 index 000000000..21e02c07c --- /dev/null +++ b/synapse/rest/client/v2_alpha/groups.py @@ -0,0 +1,786 @@ +# -*- coding: utf-8 -*- +# Copyright 2017 Vector Creations Ltd +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging + +from twisted.internet import defer + +from synapse.http.servlet import RestServlet, parse_json_object_from_request +from synapse.types import GroupID + +from ._base import client_v2_patterns + +logger = logging.getLogger(__name__) + + +class GroupServlet(RestServlet): + """Get the group profile + """ + PATTERNS = client_v2_patterns("/groups/(?P[^/]*)/profile$") + + def __init__(self, hs): + super(GroupServlet, self).__init__() + self.auth = hs.get_auth() + self.clock = hs.get_clock() + self.groups_handler = hs.get_groups_local_handler() + + @defer.inlineCallbacks + def on_GET(self, request, group_id): + requester = yield self.auth.get_user_by_req(request, allow_guest=True) + requester_user_id = requester.user.to_string() + + group_description = yield self.groups_handler.get_group_profile( + group_id, + requester_user_id, + ) + + defer.returnValue((200, group_description)) + + @defer.inlineCallbacks + def on_POST(self, request, group_id): + requester = yield self.auth.get_user_by_req(request) + requester_user_id = requester.user.to_string() + + content = parse_json_object_from_request(request) + yield self.groups_handler.update_group_profile( + group_id, requester_user_id, content, + ) + + defer.returnValue((200, {})) + + +class GroupSummaryServlet(RestServlet): + """Get the full group summary + """ + PATTERNS = client_v2_patterns("/groups/(?P[^/]*)/summary$") + + def __init__(self, hs): + super(GroupSummaryServlet, self).__init__() + self.auth = hs.get_auth() + self.clock = hs.get_clock() + self.groups_handler = hs.get_groups_local_handler() + + @defer.inlineCallbacks + def on_GET(self, request, group_id): + requester = yield self.auth.get_user_by_req(request, allow_guest=True) + requester_user_id = requester.user.to_string() + + get_group_summary = yield self.groups_handler.get_group_summary( + group_id, + requester_user_id, + ) + + defer.returnValue((200, get_group_summary)) + + +class GroupSummaryRoomsCatServlet(RestServlet): + """Update/delete a rooms entry in the summary. + + Matches both: + - /groups/:group/summary/rooms/:room_id + - /groups/:group/summary/categories/:category/rooms/:room_id + """ + PATTERNS = client_v2_patterns( + "/groups/(?P[^/]*)/summary" + "(/categories/(?P[^/]+))?" + "/rooms/(?P[^/]*)$" + ) + + def __init__(self, hs): + super(GroupSummaryRoomsCatServlet, self).__init__() + self.auth = hs.get_auth() + self.clock = hs.get_clock() + self.groups_handler = hs.get_groups_local_handler() + + @defer.inlineCallbacks + def on_PUT(self, request, group_id, category_id, room_id): + requester = yield self.auth.get_user_by_req(request) + requester_user_id = requester.user.to_string() + + content = parse_json_object_from_request(request) + resp = yield self.groups_handler.update_group_summary_room( + group_id, requester_user_id, + room_id=room_id, + category_id=category_id, + content=content, + ) + + defer.returnValue((200, resp)) + + @defer.inlineCallbacks + def on_DELETE(self, request, group_id, category_id, room_id): + requester = yield self.auth.get_user_by_req(request) + requester_user_id = requester.user.to_string() + + resp = yield self.groups_handler.delete_group_summary_room( + group_id, requester_user_id, + room_id=room_id, + category_id=category_id, + ) + + defer.returnValue((200, resp)) + + +class GroupCategoryServlet(RestServlet): + """Get/add/update/delete a group category + """ + PATTERNS = client_v2_patterns( + "/groups/(?P[^/]*)/categories/(?P[^/]+)$" + ) + + def __init__(self, hs): + super(GroupCategoryServlet, self).__init__() + self.auth = hs.get_auth() + self.clock = hs.get_clock() + self.groups_handler = hs.get_groups_local_handler() + + @defer.inlineCallbacks + def on_GET(self, request, group_id, category_id): + requester = yield self.auth.get_user_by_req(request, allow_guest=True) + requester_user_id = requester.user.to_string() + + category = yield self.groups_handler.get_group_category( + group_id, requester_user_id, + category_id=category_id, + ) + + defer.returnValue((200, category)) + + @defer.inlineCallbacks + def on_PUT(self, request, group_id, category_id): + requester = yield self.auth.get_user_by_req(request) + requester_user_id = requester.user.to_string() + + content = parse_json_object_from_request(request) + resp = yield self.groups_handler.update_group_category( + group_id, requester_user_id, + category_id=category_id, + content=content, + ) + + defer.returnValue((200, resp)) + + @defer.inlineCallbacks + def on_DELETE(self, request, group_id, category_id): + requester = yield self.auth.get_user_by_req(request) + requester_user_id = requester.user.to_string() + + resp = yield self.groups_handler.delete_group_category( + group_id, requester_user_id, + category_id=category_id, + ) + + defer.returnValue((200, resp)) + + +class GroupCategoriesServlet(RestServlet): + """Get all group categories + """ + PATTERNS = client_v2_patterns( + "/groups/(?P[^/]*)/categories/$" + ) + + def __init__(self, hs): + super(GroupCategoriesServlet, self).__init__() + self.auth = hs.get_auth() + self.clock = hs.get_clock() + self.groups_handler = hs.get_groups_local_handler() + + @defer.inlineCallbacks + def on_GET(self, request, group_id): + requester = yield self.auth.get_user_by_req(request, allow_guest=True) + requester_user_id = requester.user.to_string() + + category = yield self.groups_handler.get_group_categories( + group_id, requester_user_id, + ) + + defer.returnValue((200, category)) + + +class GroupRoleServlet(RestServlet): + """Get/add/update/delete a group role + """ + PATTERNS = client_v2_patterns( + "/groups/(?P[^/]*)/roles/(?P[^/]+)$" + ) + + def __init__(self, hs): + super(GroupRoleServlet, self).__init__() + self.auth = hs.get_auth() + self.clock = hs.get_clock() + self.groups_handler = hs.get_groups_local_handler() + + @defer.inlineCallbacks + def on_GET(self, request, group_id, role_id): + requester = yield self.auth.get_user_by_req(request, allow_guest=True) + requester_user_id = requester.user.to_string() + + category = yield self.groups_handler.get_group_role( + group_id, requester_user_id, + role_id=role_id, + ) + + defer.returnValue((200, category)) + + @defer.inlineCallbacks + def on_PUT(self, request, group_id, role_id): + requester = yield self.auth.get_user_by_req(request) + requester_user_id = requester.user.to_string() + + content = parse_json_object_from_request(request) + resp = yield self.groups_handler.update_group_role( + group_id, requester_user_id, + role_id=role_id, + content=content, + ) + + defer.returnValue((200, resp)) + + @defer.inlineCallbacks + def on_DELETE(self, request, group_id, role_id): + requester = yield self.auth.get_user_by_req(request) + requester_user_id = requester.user.to_string() + + resp = yield self.groups_handler.delete_group_role( + group_id, requester_user_id, + role_id=role_id, + ) + + defer.returnValue((200, resp)) + + +class GroupRolesServlet(RestServlet): + """Get all group roles + """ + PATTERNS = client_v2_patterns( + "/groups/(?P[^/]*)/roles/$" + ) + + def __init__(self, hs): + super(GroupRolesServlet, self).__init__() + self.auth = hs.get_auth() + self.clock = hs.get_clock() + self.groups_handler = hs.get_groups_local_handler() + + @defer.inlineCallbacks + def on_GET(self, request, group_id): + requester = yield self.auth.get_user_by_req(request, allow_guest=True) + requester_user_id = requester.user.to_string() + + category = yield self.groups_handler.get_group_roles( + group_id, requester_user_id, + ) + + defer.returnValue((200, category)) + + +class GroupSummaryUsersRoleServlet(RestServlet): + """Update/delete a user's entry in the summary. + + Matches both: + - /groups/:group/summary/users/:room_id + - /groups/:group/summary/roles/:role/users/:user_id + """ + PATTERNS = client_v2_patterns( + "/groups/(?P[^/]*)/summary" + "(/roles/(?P[^/]+))?" + "/users/(?P[^/]*)$" + ) + + def __init__(self, hs): + super(GroupSummaryUsersRoleServlet, self).__init__() + self.auth = hs.get_auth() + self.clock = hs.get_clock() + self.groups_handler = hs.get_groups_local_handler() + + @defer.inlineCallbacks + def on_PUT(self, request, group_id, role_id, user_id): + requester = yield self.auth.get_user_by_req(request) + requester_user_id = requester.user.to_string() + + content = parse_json_object_from_request(request) + resp = yield self.groups_handler.update_group_summary_user( + group_id, requester_user_id, + user_id=user_id, + role_id=role_id, + content=content, + ) + + defer.returnValue((200, resp)) + + @defer.inlineCallbacks + def on_DELETE(self, request, group_id, role_id, user_id): + requester = yield self.auth.get_user_by_req(request) + requester_user_id = requester.user.to_string() + + resp = yield self.groups_handler.delete_group_summary_user( + group_id, requester_user_id, + user_id=user_id, + role_id=role_id, + ) + + defer.returnValue((200, resp)) + + +class GroupRoomServlet(RestServlet): + """Get all rooms in a group + """ + PATTERNS = client_v2_patterns("/groups/(?P[^/]*)/rooms$") + + def __init__(self, hs): + super(GroupRoomServlet, self).__init__() + self.auth = hs.get_auth() + self.clock = hs.get_clock() + self.groups_handler = hs.get_groups_local_handler() + + @defer.inlineCallbacks + def on_GET(self, request, group_id): + requester = yield self.auth.get_user_by_req(request, allow_guest=True) + requester_user_id = requester.user.to_string() + + result = yield self.groups_handler.get_rooms_in_group(group_id, requester_user_id) + + defer.returnValue((200, result)) + + +class GroupUsersServlet(RestServlet): + """Get all users in a group + """ + PATTERNS = client_v2_patterns("/groups/(?P[^/]*)/users$") + + def __init__(self, hs): + super(GroupUsersServlet, self).__init__() + self.auth = hs.get_auth() + self.clock = hs.get_clock() + self.groups_handler = hs.get_groups_local_handler() + + @defer.inlineCallbacks + def on_GET(self, request, group_id): + requester = yield self.auth.get_user_by_req(request, allow_guest=True) + requester_user_id = requester.user.to_string() + + result = yield self.groups_handler.get_users_in_group(group_id, requester_user_id) + + defer.returnValue((200, result)) + + +class GroupInvitedUsersServlet(RestServlet): + """Get users invited to a group + """ + PATTERNS = client_v2_patterns("/groups/(?P[^/]*)/invited_users$") + + def __init__(self, hs): + super(GroupInvitedUsersServlet, self).__init__() + self.auth = hs.get_auth() + self.clock = hs.get_clock() + self.groups_handler = hs.get_groups_local_handler() + + @defer.inlineCallbacks + def on_GET(self, request, group_id): + requester = yield self.auth.get_user_by_req(request) + requester_user_id = requester.user.to_string() + + result = yield self.groups_handler.get_invited_users_in_group( + group_id, + requester_user_id, + ) + + defer.returnValue((200, result)) + + +class GroupSettingJoinPolicyServlet(RestServlet): + """Set group join policy + """ + PATTERNS = client_v2_patterns("/groups/(?P[^/]*)/settings/m.join_policy$") + + def __init__(self, hs): + super(GroupSettingJoinPolicyServlet, self).__init__() + self.auth = hs.get_auth() + self.groups_handler = hs.get_groups_local_handler() + + @defer.inlineCallbacks + def on_PUT(self, request, group_id): + requester = yield self.auth.get_user_by_req(request) + requester_user_id = requester.user.to_string() + + content = parse_json_object_from_request(request) + + result = yield self.groups_handler.set_group_join_policy( + group_id, + requester_user_id, + content, + ) + + defer.returnValue((200, result)) + + +class GroupCreateServlet(RestServlet): + """Create a group + """ + PATTERNS = client_v2_patterns("/create_group$") + + def __init__(self, hs): + super(GroupCreateServlet, self).__init__() + self.auth = hs.get_auth() + self.clock = hs.get_clock() + self.groups_handler = hs.get_groups_local_handler() + self.server_name = hs.hostname + + @defer.inlineCallbacks + def on_POST(self, request): + requester = yield self.auth.get_user_by_req(request) + requester_user_id = requester.user.to_string() + + # TODO: Create group on remote server + content = parse_json_object_from_request(request) + localpart = content.pop("localpart") + group_id = GroupID(localpart, self.server_name).to_string() + + result = yield self.groups_handler.create_group( + group_id, + requester_user_id, + content, + ) + + defer.returnValue((200, result)) + + +class GroupAdminRoomsServlet(RestServlet): + """Add a room to the group + """ + PATTERNS = client_v2_patterns( + "/groups/(?P[^/]*)/admin/rooms/(?P[^/]*)$" + ) + + def __init__(self, hs): + super(GroupAdminRoomsServlet, self).__init__() + self.auth = hs.get_auth() + self.clock = hs.get_clock() + self.groups_handler = hs.get_groups_local_handler() + + @defer.inlineCallbacks + def on_PUT(self, request, group_id, room_id): + requester = yield self.auth.get_user_by_req(request) + requester_user_id = requester.user.to_string() + + content = parse_json_object_from_request(request) + result = yield self.groups_handler.add_room_to_group( + group_id, requester_user_id, room_id, content, + ) + + defer.returnValue((200, result)) + + @defer.inlineCallbacks + def on_DELETE(self, request, group_id, room_id): + requester = yield self.auth.get_user_by_req(request) + requester_user_id = requester.user.to_string() + + result = yield self.groups_handler.remove_room_from_group( + group_id, requester_user_id, room_id, + ) + + defer.returnValue((200, result)) + + +class GroupAdminRoomsConfigServlet(RestServlet): + """Update the config of a room in a group + """ + PATTERNS = client_v2_patterns( + "/groups/(?P[^/]*)/admin/rooms/(?P[^/]*)" + "/config/(?P[^/]*)$" + ) + + def __init__(self, hs): + super(GroupAdminRoomsConfigServlet, self).__init__() + self.auth = hs.get_auth() + self.clock = hs.get_clock() + self.groups_handler = hs.get_groups_local_handler() + + @defer.inlineCallbacks + def on_PUT(self, request, group_id, room_id, config_key): + requester = yield self.auth.get_user_by_req(request) + requester_user_id = requester.user.to_string() + + content = parse_json_object_from_request(request) + result = yield self.groups_handler.update_room_in_group( + group_id, requester_user_id, room_id, config_key, content, + ) + + defer.returnValue((200, result)) + + +class GroupAdminUsersInviteServlet(RestServlet): + """Invite a user to the group + """ + PATTERNS = client_v2_patterns( + "/groups/(?P[^/]*)/admin/users/invite/(?P[^/]*)$" + ) + + def __init__(self, hs): + super(GroupAdminUsersInviteServlet, self).__init__() + self.auth = hs.get_auth() + self.clock = hs.get_clock() + self.groups_handler = hs.get_groups_local_handler() + self.store = hs.get_datastore() + self.is_mine_id = hs.is_mine_id + + @defer.inlineCallbacks + def on_PUT(self, request, group_id, user_id): + requester = yield self.auth.get_user_by_req(request) + requester_user_id = requester.user.to_string() + + content = parse_json_object_from_request(request) + config = content.get("config", {}) + result = yield self.groups_handler.invite( + group_id, user_id, requester_user_id, config, + ) + + defer.returnValue((200, result)) + + +class GroupAdminUsersKickServlet(RestServlet): + """Kick a user from the group + """ + PATTERNS = client_v2_patterns( + "/groups/(?P[^/]*)/admin/users/remove/(?P[^/]*)$" + ) + + def __init__(self, hs): + super(GroupAdminUsersKickServlet, self).__init__() + self.auth = hs.get_auth() + self.clock = hs.get_clock() + self.groups_handler = hs.get_groups_local_handler() + + @defer.inlineCallbacks + def on_PUT(self, request, group_id, user_id): + requester = yield self.auth.get_user_by_req(request) + requester_user_id = requester.user.to_string() + + content = parse_json_object_from_request(request) + result = yield self.groups_handler.remove_user_from_group( + group_id, user_id, requester_user_id, content, + ) + + defer.returnValue((200, result)) + + +class GroupSelfLeaveServlet(RestServlet): + """Leave a joined group + """ + PATTERNS = client_v2_patterns( + "/groups/(?P[^/]*)/self/leave$" + ) + + def __init__(self, hs): + super(GroupSelfLeaveServlet, self).__init__() + self.auth = hs.get_auth() + self.clock = hs.get_clock() + self.groups_handler = hs.get_groups_local_handler() + + @defer.inlineCallbacks + def on_PUT(self, request, group_id): + requester = yield self.auth.get_user_by_req(request) + requester_user_id = requester.user.to_string() + + content = parse_json_object_from_request(request) + result = yield self.groups_handler.remove_user_from_group( + group_id, requester_user_id, requester_user_id, content, + ) + + defer.returnValue((200, result)) + + +class GroupSelfJoinServlet(RestServlet): + """Attempt to join a group, or knock + """ + PATTERNS = client_v2_patterns( + "/groups/(?P[^/]*)/self/join$" + ) + + def __init__(self, hs): + super(GroupSelfJoinServlet, self).__init__() + self.auth = hs.get_auth() + self.clock = hs.get_clock() + self.groups_handler = hs.get_groups_local_handler() + + @defer.inlineCallbacks + def on_PUT(self, request, group_id): + requester = yield self.auth.get_user_by_req(request) + requester_user_id = requester.user.to_string() + + content = parse_json_object_from_request(request) + result = yield self.groups_handler.join_group( + group_id, requester_user_id, content, + ) + + defer.returnValue((200, result)) + + +class GroupSelfAcceptInviteServlet(RestServlet): + """Accept a group invite + """ + PATTERNS = client_v2_patterns( + "/groups/(?P[^/]*)/self/accept_invite$" + ) + + def __init__(self, hs): + super(GroupSelfAcceptInviteServlet, self).__init__() + self.auth = hs.get_auth() + self.clock = hs.get_clock() + self.groups_handler = hs.get_groups_local_handler() + + @defer.inlineCallbacks + def on_PUT(self, request, group_id): + requester = yield self.auth.get_user_by_req(request) + requester_user_id = requester.user.to_string() + + content = parse_json_object_from_request(request) + result = yield self.groups_handler.accept_invite( + group_id, requester_user_id, content, + ) + + defer.returnValue((200, result)) + + +class GroupSelfUpdatePublicityServlet(RestServlet): + """Update whether we publicise a users membership of a group + """ + PATTERNS = client_v2_patterns( + "/groups/(?P[^/]*)/self/update_publicity$" + ) + + def __init__(self, hs): + super(GroupSelfUpdatePublicityServlet, self).__init__() + self.auth = hs.get_auth() + self.clock = hs.get_clock() + self.store = hs.get_datastore() + + @defer.inlineCallbacks + def on_PUT(self, request, group_id): + requester = yield self.auth.get_user_by_req(request) + requester_user_id = requester.user.to_string() + + content = parse_json_object_from_request(request) + publicise = content["publicise"] + yield self.store.update_group_publicity( + group_id, requester_user_id, publicise, + ) + + defer.returnValue((200, {})) + + +class PublicisedGroupsForUserServlet(RestServlet): + """Get the list of groups a user is advertising + """ + PATTERNS = client_v2_patterns( + "/publicised_groups/(?P[^/]*)$" + ) + + def __init__(self, hs): + super(PublicisedGroupsForUserServlet, self).__init__() + self.auth = hs.get_auth() + self.clock = hs.get_clock() + self.store = hs.get_datastore() + self.groups_handler = hs.get_groups_local_handler() + + @defer.inlineCallbacks + def on_GET(self, request, user_id): + yield self.auth.get_user_by_req(request, allow_guest=True) + + result = yield self.groups_handler.get_publicised_groups_for_user( + user_id + ) + + defer.returnValue((200, result)) + + +class PublicisedGroupsForUsersServlet(RestServlet): + """Get the list of groups a user is advertising + """ + PATTERNS = client_v2_patterns( + "/publicised_groups$" + ) + + def __init__(self, hs): + super(PublicisedGroupsForUsersServlet, self).__init__() + self.auth = hs.get_auth() + self.clock = hs.get_clock() + self.store = hs.get_datastore() + self.groups_handler = hs.get_groups_local_handler() + + @defer.inlineCallbacks + def on_POST(self, request): + yield self.auth.get_user_by_req(request, allow_guest=True) + + content = parse_json_object_from_request(request) + user_ids = content["user_ids"] + + result = yield self.groups_handler.bulk_get_publicised_groups( + user_ids + ) + + defer.returnValue((200, result)) + + +class GroupsForUserServlet(RestServlet): + """Get all groups the logged in user is joined to + """ + PATTERNS = client_v2_patterns( + "/joined_groups$" + ) + + def __init__(self, hs): + super(GroupsForUserServlet, self).__init__() + self.auth = hs.get_auth() + self.clock = hs.get_clock() + self.groups_handler = hs.get_groups_local_handler() + + @defer.inlineCallbacks + def on_GET(self, request): + requester = yield self.auth.get_user_by_req(request, allow_guest=True) + requester_user_id = requester.user.to_string() + + result = yield self.groups_handler.get_joined_groups(requester_user_id) + + defer.returnValue((200, result)) + + +def register_servlets(hs, http_server): + GroupServlet(hs).register(http_server) + GroupSummaryServlet(hs).register(http_server) + GroupInvitedUsersServlet(hs).register(http_server) + GroupUsersServlet(hs).register(http_server) + GroupRoomServlet(hs).register(http_server) + GroupSettingJoinPolicyServlet(hs).register(http_server) + GroupCreateServlet(hs).register(http_server) + GroupAdminRoomsServlet(hs).register(http_server) + GroupAdminRoomsConfigServlet(hs).register(http_server) + GroupAdminUsersInviteServlet(hs).register(http_server) + GroupAdminUsersKickServlet(hs).register(http_server) + GroupSelfLeaveServlet(hs).register(http_server) + GroupSelfJoinServlet(hs).register(http_server) + GroupSelfAcceptInviteServlet(hs).register(http_server) + GroupsForUserServlet(hs).register(http_server) + GroupCategoryServlet(hs).register(http_server) + GroupCategoriesServlet(hs).register(http_server) + GroupSummaryRoomsCatServlet(hs).register(http_server) + GroupRoleServlet(hs).register(http_server) + GroupRolesServlet(hs).register(http_server) + GroupSelfUpdatePublicityServlet(hs).register(http_server) + GroupSummaryUsersRoleServlet(hs).register(http_server) + PublicisedGroupsForUserServlet(hs).register(http_server) + PublicisedGroupsForUsersServlet(hs).register(http_server) diff --git a/synapse/rest/client/v2_alpha/keys.py b/synapse/rest/client/v2_alpha/keys.py index 6a3cfe84f..8486086b5 100644 --- a/synapse/rest/client/v2_alpha/keys.py +++ b/synapse/rest/client/v2_alpha/keys.py @@ -19,10 +19,13 @@ from twisted.internet import defer from synapse.api.errors import SynapseError from synapse.http.servlet import ( - RestServlet, parse_json_object_from_request, parse_integer + RestServlet, + parse_integer, + parse_json_object_from_request, + parse_string, ) -from synapse.http.servlet import parse_string from synapse.types import StreamToken + from ._base import client_v2_patterns logger = logging.getLogger(__name__) @@ -53,8 +56,7 @@ class KeyUploadServlet(RestServlet): }, } """ - PATTERNS = client_v2_patterns("/keys/upload(/(?P[^/]+))?$", - releases=()) + PATTERNS = client_v2_patterns("/keys/upload(/(?P[^/]+))?$") def __init__(self, hs): """ @@ -128,10 +130,7 @@ class KeyQueryServlet(RestServlet): } } } } } } """ - PATTERNS = client_v2_patterns( - "/keys/query$", - releases=() - ) + PATTERNS = client_v2_patterns("/keys/query$") def __init__(self, hs): """ @@ -160,10 +159,7 @@ class KeyChangesServlet(RestServlet): 200 OK { "changed": ["@foo:example.com"] } """ - PATTERNS = client_v2_patterns( - "/keys/changes$", - releases=() - ) + PATTERNS = client_v2_patterns("/keys/changes$") def __init__(self, hs): """ @@ -188,13 +184,11 @@ class KeyChangesServlet(RestServlet): user_id = requester.user.to_string() - changed = yield self.device_handler.get_user_ids_changed( + results = yield self.device_handler.get_user_ids_changed( user_id, from_token, ) - defer.returnValue((200, { - "changed": list(changed), - })) + defer.returnValue((200, results)) class OneTimeKeyServlet(RestServlet): @@ -215,10 +209,7 @@ class OneTimeKeyServlet(RestServlet): } } } } """ - PATTERNS = client_v2_patterns( - "/keys/claim$", - releases=() - ) + PATTERNS = client_v2_patterns("/keys/claim$") def __init__(self, hs): super(OneTimeKeyServlet, self).__init__() diff --git a/synapse/rest/client/v2_alpha/notifications.py b/synapse/rest/client/v2_alpha/notifications.py index fd2a3d69d..2a6ea3df5 100644 --- a/synapse/rest/client/v2_alpha/notifications.py +++ b/synapse/rest/client/v2_alpha/notifications.py @@ -13,24 +13,23 @@ # See the License for the specific language governing permissions and # limitations under the License. +import logging + from twisted.internet import defer -from synapse.http.servlet import ( - RestServlet, parse_string, parse_integer -) from synapse.events.utils import ( - serialize_event, format_event_for_client_v2_without_room_id, + format_event_for_client_v2_without_room_id, + serialize_event, ) +from synapse.http.servlet import RestServlet, parse_integer, parse_string from ._base import client_v2_patterns -import logging - logger = logging.getLogger(__name__) class NotificationsServlet(RestServlet): - PATTERNS = client_v2_patterns("/notifications$", releases=()) + PATTERNS = client_v2_patterns("/notifications$") def __init__(self, hs): super(NotificationsServlet, self).__init__() @@ -88,7 +87,7 @@ class NotificationsServlet(RestServlet): pa["topological_ordering"], pa["stream_ordering"] ) returned_push_actions.append(returned_pa) - next_token = pa["stream_ordering"] + next_token = str(pa["stream_ordering"]) defer.returnValue((200, { "notifications": returned_push_actions, diff --git a/synapse/rest/client/v2_alpha/openid.py b/synapse/rest/client/v2_alpha/openid.py index aa1cae8e1..01c90aa2a 100644 --- a/synapse/rest/client/v2_alpha/openid.py +++ b/synapse/rest/client/v2_alpha/openid.py @@ -14,15 +14,15 @@ # limitations under the License. -from ._base import client_v2_patterns - -from synapse.http.servlet import RestServlet, parse_json_object_from_request -from synapse.api.errors import AuthError -from synapse.util.stringutils import random_string +import logging from twisted.internet import defer -import logging +from synapse.api.errors import AuthError +from synapse.http.servlet import RestServlet, parse_json_object_from_request +from synapse.util.stringutils import random_string + +from ._base import client_v2_patterns logger = logging.getLogger(__name__) diff --git a/synapse/rest/client/v2_alpha/read_marker.py b/synapse/rest/client/v2_alpha/read_marker.py index 2f8784fe0..a6e582a5a 100644 --- a/synapse/rest/client/v2_alpha/read_marker.py +++ b/synapse/rest/client/v2_alpha/read_marker.py @@ -13,14 +13,14 @@ # See the License for the specific language governing permissions and # limitations under the License. +import logging + from twisted.internet import defer from synapse.http.servlet import RestServlet, parse_json_object_from_request + from ._base import client_v2_patterns -import logging - - logger = logging.getLogger(__name__) diff --git a/synapse/rest/client/v2_alpha/receipts.py b/synapse/rest/client/v2_alpha/receipts.py index 1fbff2edd..de370cac4 100644 --- a/synapse/rest/client/v2_alpha/receipts.py +++ b/synapse/rest/client/v2_alpha/receipts.py @@ -13,15 +13,15 @@ # See the License for the specific language governing permissions and # limitations under the License. +import logging + from twisted.internet import defer from synapse.api.errors import SynapseError from synapse.http.servlet import RestServlet + from ._base import client_v2_patterns -import logging - - logger = logging.getLogger(__name__) diff --git a/synapse/rest/client/v2_alpha/register.py b/synapse/rest/client/v2_alpha/register.py index 1421c1815..d6cf915d8 100644 --- a/synapse/rest/client/v2_alpha/register.py +++ b/synapse/rest/client/v2_alpha/register.py @@ -14,25 +14,29 @@ # See the License for the specific language governing permissions and # limitations under the License. +import hmac +import logging +from hashlib import sha1 + +from six import string_types + from twisted.internet import defer import synapse -from synapse.api.auth import get_access_token_from_request, has_access_token +import synapse.types from synapse.api.constants import LoginType -from synapse.api.errors import SynapseError, Codes, UnrecognizedRequestError +from synapse.api.errors import Codes, SynapseError, UnrecognizedRequestError from synapse.http.servlet import ( - RestServlet, parse_json_object_from_request, assert_params_in_request, parse_string + RestServlet, + assert_params_in_dict, + parse_json_object_from_request, + parse_string, ) from synapse.util.msisdn import phone_number_to_msisdn - -from ._base import client_v2_patterns - -import logging -import hmac -from hashlib import sha1 -from synapse.util.async import run_on_reactor from synapse.util.ratelimitutils import FederationRateLimiter +from synapse.util.threepids import check_3pid_allowed +from ._base import client_v2_patterns, interactive_auth_handler # We ought to be using hmac.compare_digest() but on older pythons it doesn't # exist. It's a _really minor_ security flaw to use plain string comparison @@ -64,10 +68,15 @@ class EmailRegisterRequestTokenRestServlet(RestServlet): def on_POST(self, request): body = parse_json_object_from_request(request) - assert_params_in_request(body, [ + assert_params_in_dict(body, [ 'id_server', 'client_secret', 'email', 'send_attempt' ]) + if not check_3pid_allowed(self.hs, "email", body['email']): + raise SynapseError( + 403, "Third party identifier is not allowed", Codes.THREEPID_DENIED, + ) + existingUid = yield self.hs.get_datastore().get_user_id_by_threepid( 'email', body['email'] ) @@ -95,7 +104,7 @@ class MsisdnRegisterRequestTokenRestServlet(RestServlet): def on_POST(self, request): body = parse_json_object_from_request(request) - assert_params_in_request(body, [ + assert_params_in_dict(body, [ 'id_server', 'client_secret', 'country', 'phone_number', 'send_attempt', @@ -103,6 +112,11 @@ class MsisdnRegisterRequestTokenRestServlet(RestServlet): msisdn = phone_number_to_msisdn(body['country'], body['phone_number']) + if not check_3pid_allowed(self.hs, "msisdn", msisdn): + raise SynapseError( + 403, "Third party identifier is not allowed", Codes.THREEPID_DENIED, + ) + existingUid = yield self.hs.get_datastore().get_user_id_by_threepid( 'msisdn', msisdn ) @@ -170,13 +184,13 @@ class RegisterRestServlet(RestServlet): self.auth_handler = hs.get_auth_handler() self.registration_handler = hs.get_handlers().registration_handler self.identity_handler = hs.get_handlers().identity_handler + self.room_member_handler = hs.get_room_member_handler() self.device_handler = hs.get_device_handler() self.macaroon_gen = hs.get_macaroon_generator() + @interactive_auth_handler @defer.inlineCallbacks def on_POST(self, request): - yield run_on_reactor() - body = parse_json_object_from_request(request) kind = "user" @@ -196,20 +210,20 @@ class RegisterRestServlet(RestServlet): # in sessions. Pull out the username/password provided to us. desired_password = None if 'password' in body: - if (not isinstance(body['password'], basestring) or + if (not isinstance(body['password'], string_types) or len(body['password']) > 512): raise SynapseError(400, "Invalid password") desired_password = body["password"] desired_username = None if 'username' in body: - if (not isinstance(body['username'], basestring) or + if (not isinstance(body['username'], string_types) or len(body['username']) > 512): raise SynapseError(400, "Invalid username") desired_username = body['username'] appservice = None - if has_access_token(request): + if self.auth.has_access_token(request): appservice = yield self.auth.get_appservice_by_req(request) # fork off as soon as possible for ASes and shared secret auth which @@ -221,15 +235,30 @@ class RegisterRestServlet(RestServlet): # 'user' key not 'username'). Since this is a new addition, we'll # fallback to 'username' if they gave one. desired_username = body.get("user", desired_username) - access_token = get_access_token_from_request(request) - if isinstance(desired_username, basestring): + # XXX we should check that desired_username is valid. Currently + # we give appservices carte blanche for any insanity in mxids, + # because the IRC bridges rely on being able to register stupid + # IDs. + + access_token = self.auth.get_access_token_from_request(request) + + if isinstance(desired_username, string_types): result = yield self._do_appservice_registration( desired_username, access_token, body ) defer.returnValue((200, result)) # we throw for non 200 responses return + # for either shared secret or regular registration, downcase the + # provided username before attempting to register it. This should mean + # that people who try to register with upper-case in their usernames + # don't get a nasty surprise. (Note that we treat username + # case-insenstively in login, so they are free to carry on imagining + # that their username is CrAzYh4cKeR if that keeps them happy) + if desired_username is not None: + desired_username = desired_username.lower() + # == Shared Secret Registration == (e.g. create new user scripts) if 'mac' in body: # FIXME: Should we really be determining if this is shared secret @@ -286,34 +315,66 @@ class RegisterRestServlet(RestServlet): if 'x_show_msisdn' in body and body['x_show_msisdn']: show_msisdn = True + # FIXME: need a better error than "no auth flow found" for scenarios + # where we required 3PID for registration but the user didn't give one + require_email = 'email' in self.hs.config.registrations_require_3pid + require_msisdn = 'msisdn' in self.hs.config.registrations_require_3pid + + flows = [] if self.hs.config.enable_registration_captcha: - flows = [ - [LoginType.RECAPTCHA], - [LoginType.EMAIL_IDENTITY, LoginType.RECAPTCHA], - ] + # only support 3PIDless registration if no 3PIDs are required + if not require_email and not require_msisdn: + flows.extend([[LoginType.RECAPTCHA]]) + # only support the email-only flow if we don't require MSISDN 3PIDs + if not require_msisdn: + flows.extend([[LoginType.EMAIL_IDENTITY, LoginType.RECAPTCHA]]) + if show_msisdn: + # only support the MSISDN-only flow if we don't require email 3PIDs + if not require_email: + flows.extend([[LoginType.MSISDN, LoginType.RECAPTCHA]]) + # always let users provide both MSISDN & email flows.extend([ - [LoginType.MSISDN, LoginType.RECAPTCHA], [LoginType.MSISDN, LoginType.EMAIL_IDENTITY, LoginType.RECAPTCHA], ]) else: - flows = [ - [LoginType.DUMMY], - [LoginType.EMAIL_IDENTITY], - ] + # only support 3PIDless registration if no 3PIDs are required + if not require_email and not require_msisdn: + flows.extend([[LoginType.DUMMY]]) + # only support the email-only flow if we don't require MSISDN 3PIDs + if not require_msisdn: + flows.extend([[LoginType.EMAIL_IDENTITY]]) + if show_msisdn: + # only support the MSISDN-only flow if we don't require email 3PIDs + if not require_email or require_msisdn: + flows.extend([[LoginType.MSISDN]]) + # always let users provide both MSISDN & email flows.extend([ - [LoginType.MSISDN], - [LoginType.MSISDN, LoginType.EMAIL_IDENTITY], + [LoginType.MSISDN, LoginType.EMAIL_IDENTITY] ]) - authed, auth_result, params, session_id = yield self.auth_handler.check_auth( + auth_result, params, session_id = yield self.auth_handler.check_auth( flows, body, self.hs.get_ip_from_request(request) ) - if not authed: - defer.returnValue((401, auth_result)) - return + # Check that we're not trying to register a denied 3pid. + # + # the user-facing checks will probably already have happened in + # /register/email/requestToken when we requested a 3pid, but that's not + # guaranteed. + + if auth_result: + for login_type in [LoginType.EMAIL_IDENTITY, LoginType.MSISDN]: + if login_type in auth_result: + medium = auth_result[login_type]['medium'] + address = auth_result[login_type]['address'] + + if not check_3pid_allowed(self.hs, medium, address): + raise SynapseError( + 403, "Third party identifier is not allowed", + Codes.THREEPID_DENIED, + ) if registered_user_id is not None: logger.info( @@ -325,14 +386,15 @@ class RegisterRestServlet(RestServlet): add_msisdn = False else: # NB: This may be from the auth handler and NOT from the POST - if 'password' not in params: - raise SynapseError(400, "Missing password.", - Codes.MISSING_PARAM) + assert_params_in_dict(params, ["password"]) desired_username = params.get("username", None) new_password = params.get("password", None) guest_access_token = params.get("guest_access_token", None) + if desired_username is not None: + desired_username = desired_username.lower() + (registered_user_id, _) = yield self.registration_handler.register( localpart=desired_username, password=new_password, @@ -383,15 +445,24 @@ class RegisterRestServlet(RestServlet): def _do_shared_secret_registration(self, username, password, body): if not self.hs.config.registration_shared_secret: raise SynapseError(400, "Shared secret registration is not enabled") + if not username: + raise SynapseError( + 400, "username must be specified", errcode=Codes.BAD_JSON, + ) - user = username.encode("utf-8") + # use the username from the original request rather than the + # downcased one in `username` for the mac calculation + user = body["username"].encode("utf-8") # str() because otherwise hmac complains that 'unicode' does not # have the buffer interface got_mac = str(body["mac"]) + # FIXME this is different to the /v1/register endpoint, which + # includes the password and admin flag in the hashed text. Why are + # these different? want_mac = hmac.new( - key=self.hs.config.registration_shared_secret, + key=self.hs.config.registration_shared_secret.encode(), msg=user, digestmod=sha1, ).hexdigest() @@ -492,11 +563,14 @@ class RegisterRestServlet(RestServlet): Returns: defer.Deferred: """ - reqd = ('medium', 'address', 'validated_at') - if any(x not in threepid for x in reqd): - # This will only happen if the ID server returns a malformed response - logger.info("Can't add incomplete 3pid") - defer.returnValue() + try: + assert_params_in_dict(threepid, ['medium', 'address', 'validated_at']) + except SynapseError as ex: + if ex.errcode == Codes.MISSING_PARAM: + # This will only happen if the ID server returns a malformed response + logger.info("Can't add incomplete 3pid") + defer.returnValue(None) + raise yield self.auth_handler.add_threepid( user_id, @@ -523,25 +597,28 @@ class RegisterRestServlet(RestServlet): Args: (str) user_id: full canonical @user:id (object) params: registration parameters, from which we pull - device_id and initial_device_name + device_id, initial_device_name and inhibit_login Returns: defer.Deferred: (object) dictionary for response from /register """ - device_id = yield self._register_device(user_id, params) - - access_token = ( - yield self.auth_handler.get_access_token_for_user_id( - user_id, device_id=device_id, - initial_display_name=params.get("initial_device_display_name") - ) - ) - - defer.returnValue({ + result = { "user_id": user_id, - "access_token": access_token, "home_server": self.hs.hostname, - "device_id": device_id, - }) + } + if not params.get("inhibit_login", False): + device_id = yield self._register_device(user_id, params) + + access_token = ( + yield self.auth_handler.get_access_token_for_user_id( + user_id, device_id=device_id, + ) + ) + + result.update({ + "access_token": access_token, + "device_id": device_id, + }) + defer.returnValue(result) def _register_device(self, user_id, params): """Register a device for a user. @@ -566,7 +643,7 @@ class RegisterRestServlet(RestServlet): @defer.inlineCallbacks def _do_guest_registration(self, params): if not self.hs.config.allow_guest_access: - defer.returnValue((403, "Guest access is disabled")) + raise SynapseError(403, "Guest access is disabled") user_id, _ = yield self.registration_handler.register( generate_token=False, make_guest=True diff --git a/synapse/rest/client/v2_alpha/report_event.py b/synapse/rest/client/v2_alpha/report_event.py index 8903e1240..95d2a71ec 100644 --- a/synapse/rest/client/v2_alpha/report_event.py +++ b/synapse/rest/client/v2_alpha/report_event.py @@ -13,13 +13,21 @@ # See the License for the specific language governing permissions and # limitations under the License. -from twisted.internet import defer - -from synapse.http.servlet import RestServlet, parse_json_object_from_request -from ._base import client_v2_patterns - import logging +from six import string_types +from six.moves import http_client + +from twisted.internet import defer + +from synapse.api.errors import Codes, SynapseError +from synapse.http.servlet import ( + RestServlet, + assert_params_in_dict, + parse_json_object_from_request, +) + +from ._base import client_v2_patterns logger = logging.getLogger(__name__) @@ -42,12 +50,26 @@ class ReportEventRestServlet(RestServlet): user_id = requester.user.to_string() body = parse_json_object_from_request(request) + assert_params_in_dict(body, ("reason", "score")) + + if not isinstance(body["reason"], string_types): + raise SynapseError( + http_client.BAD_REQUEST, + "Param 'reason' must be a string", + Codes.BAD_JSON, + ) + if not isinstance(body["score"], int): + raise SynapseError( + http_client.BAD_REQUEST, + "Param 'score' must be an integer", + Codes.BAD_JSON, + ) yield self.store.add_event_report( room_id=room_id, event_id=event_id, user_id=user_id, - reason=body.get("reason"), + reason=body["reason"], content=body, received_ts=self.clock.time_msec(), ) diff --git a/synapse/rest/client/v2_alpha/sendtodevice.py b/synapse/rest/client/v2_alpha/sendtodevice.py index d607bd297..a9e9a47a0 100644 --- a/synapse/rest/client/v2_alpha/sendtodevice.py +++ b/synapse/rest/client/v2_alpha/sendtodevice.py @@ -29,7 +29,7 @@ logger = logging.getLogger(__name__) class SendToDeviceRestServlet(servlet.RestServlet): PATTERNS = client_v2_patterns( "/sendToDevice/(?P[^/]*)/(?P[^/]*)$", - releases=[], v2_alpha=False + v2_alpha=False ) def __init__(self, hs): @@ -40,7 +40,7 @@ class SendToDeviceRestServlet(servlet.RestServlet): super(SendToDeviceRestServlet, self).__init__() self.hs = hs self.auth = hs.get_auth() - self.txns = HttpTransactionCache(hs.get_clock()) + self.txns = HttpTransactionCache(hs) self.device_message_handler = hs.get_device_message_handler() def on_PUT(self, request, message_type, txn_id): diff --git a/synapse/rest/client/v2_alpha/sync.py b/synapse/rest/client/v2_alpha/sync.py index 83e209d18..8aa06faf2 100644 --- a/synapse/rest/client/v2_alpha/sync.py +++ b/synapse/rest/client/v2_alpha/sync.py @@ -13,27 +13,26 @@ # See the License for the specific language governing permissions and # limitations under the License. -from twisted.internet import defer - -from synapse.http.servlet import ( - RestServlet, parse_string, parse_integer, parse_boolean -) -from synapse.handlers.presence import format_user_presence_state -from synapse.handlers.sync import SyncConfig -from synapse.types import StreamToken -from synapse.events.utils import ( - serialize_event, format_event_for_client_v2_without_room_id, -) -from synapse.api.filtering import FilterCollection, DEFAULT_FILTER_COLLECTION -from synapse.api.errors import SynapseError -from synapse.api.constants import PresenceState -from ._base import client_v2_patterns -from ._base import set_timeline_upper_limit - import itertools import logging -import ujson as json +from canonicaljson import json + +from twisted.internet import defer + +from synapse.api.constants import PresenceState +from synapse.api.errors import SynapseError +from synapse.api.filtering import DEFAULT_FILTER_COLLECTION, FilterCollection +from synapse.events.utils import ( + format_event_for_client_v2_without_room_id, + serialize_event, +) +from synapse.handlers.presence import format_user_presence_state +from synapse.handlers.sync import SyncConfig +from synapse.http.servlet import RestServlet, parse_boolean, parse_integer, parse_string +from synapse.types import StreamToken + +from ._base import client_v2_patterns, set_timeline_upper_limit logger = logging.getLogger(__name__) @@ -85,6 +84,7 @@ class SyncRestServlet(RestServlet): self.clock = hs.get_clock() self.filtering = hs.get_filtering() self.presence_handler = hs.get_presence_handler() + self._server_notices_sender = hs.get_server_notices_sender() @defer.inlineCallbacks def on_GET(self, request): @@ -110,7 +110,7 @@ class SyncRestServlet(RestServlet): filter_id = parse_string(request, "filter", default=None) full_state = parse_boolean(request, "full_state", default=False) - logger.info( + logger.debug( "/sync: user=%r, timeout=%r, since=%r," " set_presence=%r, filter_id=%r, device_id=%r" % ( user, timeout, since, set_presence, filter_id, device_id @@ -125,7 +125,7 @@ class SyncRestServlet(RestServlet): filter_object = json.loads(filter_id) set_timeline_upper_limit(filter_object, self.hs.config.filter_timeline_limit) - except: + except Exception: raise SynapseError(400, "Invalid filter JSON") self.filtering.check_valid_filter(filter_object) filter = FilterCollection(filter_object) @@ -149,6 +149,9 @@ class SyncRestServlet(RestServlet): else: since_token = None + # send any outstanding server notices to the user. + yield self._server_notices_sender.on_user_syncing(user.to_string()) + affect_presence = set_presence != PresenceState.OFFLINE if affect_presence: @@ -164,27 +167,35 @@ class SyncRestServlet(RestServlet): ) time_now = self.clock.time_msec() - - joined = self.encode_joined( - sync_result.joined, time_now, requester.access_token_id, filter.event_fields + response_content = self.encode_response( + time_now, sync_result, requester.access_token_id, filter ) - invited = self.encode_invited( - sync_result.invited, time_now, requester.access_token_id + defer.returnValue((200, response_content)) + + @staticmethod + def encode_response(time_now, sync_result, access_token_id, filter): + joined = SyncRestServlet.encode_joined( + sync_result.joined, time_now, access_token_id, filter.event_fields ) - archived = self.encode_archived( - sync_result.archived, time_now, requester.access_token_id, + invited = SyncRestServlet.encode_invited( + sync_result.invited, time_now, access_token_id, + ) + + archived = SyncRestServlet.encode_archived( + sync_result.archived, time_now, access_token_id, filter.event_fields, ) - response_content = { + return { "account_data": {"events": sync_result.account_data}, "to_device": {"events": sync_result.to_device}, "device_lists": { - "changed": list(sync_result.device_lists), + "changed": list(sync_result.device_lists.changed), + "left": list(sync_result.device_lists.left), }, - "presence": self.encode_presence( + "presence": SyncRestServlet.encode_presence( sync_result.presence, time_now ), "rooms": { @@ -192,13 +203,17 @@ class SyncRestServlet(RestServlet): "invite": invited, "leave": archived, }, + "groups": { + "join": sync_result.groups.join, + "invite": sync_result.groups.invite, + "leave": sync_result.groups.leave, + }, "device_one_time_keys_count": sync_result.device_one_time_keys_count, "next_batch": sync_result.next_batch.to_string(), } - defer.returnValue((200, response_content)) - - def encode_presence(self, events, time_now): + @staticmethod + def encode_presence(events, time_now): return { "events": [ { @@ -212,7 +227,8 @@ class SyncRestServlet(RestServlet): ] } - def encode_joined(self, rooms, time_now, token_id, event_fields): + @staticmethod + def encode_joined(rooms, time_now, token_id, event_fields): """ Encode the joined rooms in a sync result @@ -231,13 +247,14 @@ class SyncRestServlet(RestServlet): """ joined = {} for room in rooms: - joined[room.room_id] = self.encode_room( + joined[room.room_id] = SyncRestServlet.encode_room( room, time_now, token_id, only_fields=event_fields ) return joined - def encode_invited(self, rooms, time_now, token_id): + @staticmethod + def encode_invited(rooms, time_now, token_id): """ Encode the invited rooms in a sync result @@ -270,7 +287,8 @@ class SyncRestServlet(RestServlet): return invited - def encode_archived(self, rooms, time_now, token_id, event_fields): + @staticmethod + def encode_archived(rooms, time_now, token_id, event_fields): """ Encode the archived rooms in a sync result @@ -289,7 +307,7 @@ class SyncRestServlet(RestServlet): """ joined = {} for room in rooms: - joined[room.room_id] = self.encode_room( + joined[room.room_id] = SyncRestServlet.encode_room( room, time_now, token_id, joined=False, only_fields=event_fields ) diff --git a/synapse/rest/client/v2_alpha/tags.py b/synapse/rest/client/v2_alpha/tags.py index dac8603b0..4fea614e9 100644 --- a/synapse/rest/client/v2_alpha/tags.py +++ b/synapse/rest/client/v2_alpha/tags.py @@ -13,14 +13,14 @@ # See the License for the specific language governing permissions and # limitations under the License. -from ._base import client_v2_patterns - -from synapse.http.servlet import RestServlet, parse_json_object_from_request -from synapse.api.errors import AuthError +import logging from twisted.internet import defer -import logging +from synapse.api.errors import AuthError +from synapse.http.servlet import RestServlet, parse_json_object_from_request + +from ._base import client_v2_patterns logger = logging.getLogger(__name__) diff --git a/synapse/rest/client/v2_alpha/thirdparty.py b/synapse/rest/client/v2_alpha/thirdparty.py index 6fceb23e2..d9d379182 100644 --- a/synapse/rest/client/v2_alpha/thirdparty.py +++ b/synapse/rest/client/v2_alpha/thirdparty.py @@ -20,13 +20,14 @@ from twisted.internet import defer from synapse.api.constants import ThirdPartyEntityKind from synapse.http.servlet import RestServlet + from ._base import client_v2_patterns logger = logging.getLogger(__name__) class ThirdPartyProtocolsServlet(RestServlet): - PATTERNS = client_v2_patterns("/thirdparty/protocols", releases=()) + PATTERNS = client_v2_patterns("/thirdparty/protocols") def __init__(self, hs): super(ThirdPartyProtocolsServlet, self).__init__() @@ -43,8 +44,7 @@ class ThirdPartyProtocolsServlet(RestServlet): class ThirdPartyProtocolServlet(RestServlet): - PATTERNS = client_v2_patterns("/thirdparty/protocol/(?P[^/]+)$", - releases=()) + PATTERNS = client_v2_patterns("/thirdparty/protocol/(?P[^/]+)$") def __init__(self, hs): super(ThirdPartyProtocolServlet, self).__init__() @@ -66,8 +66,7 @@ class ThirdPartyProtocolServlet(RestServlet): class ThirdPartyUserServlet(RestServlet): - PATTERNS = client_v2_patterns("/thirdparty/user(/(?P[^/]+))?$", - releases=()) + PATTERNS = client_v2_patterns("/thirdparty/user(/(?P[^/]+))?$") def __init__(self, hs): super(ThirdPartyUserServlet, self).__init__() @@ -90,8 +89,7 @@ class ThirdPartyUserServlet(RestServlet): class ThirdPartyLocationServlet(RestServlet): - PATTERNS = client_v2_patterns("/thirdparty/location(/(?P[^/]+))?$", - releases=()) + PATTERNS = client_v2_patterns("/thirdparty/location(/(?P[^/]+))?$") def __init__(self, hs): super(ThirdPartyLocationServlet, self).__init__() diff --git a/synapse/rest/client/v2_alpha/user_directory.py b/synapse/rest/client/v2_alpha/user_directory.py index 6e012da4a..cac0624ba 100644 --- a/synapse/rest/client/v2_alpha/user_directory.py +++ b/synapse/rest/client/v2_alpha/user_directory.py @@ -19,6 +19,7 @@ from twisted.internet import defer from synapse.api.errors import SynapseError from synapse.http.servlet import RestServlet, parse_json_object_from_request + from ._base import client_v2_patterns logger = logging.getLogger(__name__) @@ -65,7 +66,7 @@ class UserDirectorySearchRestServlet(RestServlet): try: search_term = body["search_term"] - except: + except Exception: raise SynapseError(400, "`search_term` is required field") results = yield self.user_directory_handler.search_users( diff --git a/synapse/rest/client/versions.py b/synapse/rest/client/versions.py index e984ea47d..6ac2987b9 100644 --- a/synapse/rest/client/versions.py +++ b/synapse/rest/client/versions.py @@ -13,11 +13,11 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.http.servlet import RestServlet - import logging import re +from synapse.http.servlet import RestServlet + logger = logging.getLogger(__name__) @@ -30,6 +30,7 @@ class VersionsRestServlet(RestServlet): "r0.0.1", "r0.1.0", "r0.2.0", + "r0.3.0", ] }) diff --git a/synapse/rest/consent/__init__.py b/synapse/rest/consent/__init__.py new file mode 100644 index 000000000..e69de29bb diff --git a/synapse/rest/consent/consent_resource.py b/synapse/rest/consent/consent_resource.py new file mode 100644 index 000000000..147ff7d79 --- /dev/null +++ b/synapse/rest/consent/consent_resource.py @@ -0,0 +1,220 @@ +# -*- coding: utf-8 -*- +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import hmac +import logging +from hashlib import sha256 +from os import path + +from six.moves import http_client + +import jinja2 +from jinja2 import TemplateNotFound + +from twisted.internet import defer +from twisted.web.resource import Resource +from twisted.web.server import NOT_DONE_YET + +from synapse.api.errors import NotFoundError, StoreError, SynapseError +from synapse.config import ConfigError +from synapse.http.server import finish_request, wrap_html_request_handler +from synapse.http.servlet import parse_string +from synapse.types import UserID + +# language to use for the templates. TODO: figure this out from Accept-Language +TEMPLATE_LANGUAGE = "en" + +logger = logging.getLogger(__name__) + +# use hmac.compare_digest if we have it (python 2.7.7), else just use equality +if hasattr(hmac, "compare_digest"): + compare_digest = hmac.compare_digest +else: + def compare_digest(a, b): + return a == b + + +class ConsentResource(Resource): + """A twisted Resource to display a privacy policy and gather consent to it + + When accessed via GET, returns the privacy policy via a template. + + When accessed via POST, records the user's consent in the database and + displays a success page. + + The config should include a template_dir setting which contains templates + for the HTML. The directory should contain one subdirectory per language + (eg, 'en', 'fr'), and each language directory should contain the policy + document (named as '.html') and a success page (success.html). + + Both forms take a set of parameters from the browser. For the POST form, + these are normally sent as form parameters (but may be query-params); for + GET requests they must be query params. These are: + + u: the complete mxid, or the localpart of the user giving their + consent. Required for both GET (where it is used as an input to the + template) and for POST (where it is used to find the row in the db + to update). + + h: hmac_sha256(secret, u), where 'secret' is the privacy_secret in the + config file. If it doesn't match, the request is 403ed. + + v: the version of the privacy policy being agreed to. + + For GET: optional, and defaults to whatever was set in the config + file. Used to choose the version of the policy to pick from the + templates directory. + + For POST: required; gives the value to be recorded in the database + against the user. + """ + def __init__(self, hs): + """ + Args: + hs (synapse.server.HomeServer): homeserver + """ + Resource.__init__(self) + + self.hs = hs + self.store = hs.get_datastore() + + # this is required by the request_handler wrapper + self.clock = hs.get_clock() + + self._default_consent_version = hs.config.user_consent_version + if self._default_consent_version is None: + raise ConfigError( + "Consent resource is enabled but user_consent section is " + "missing in config file.", + ) + + # daemonize changes the cwd to /, so make the path absolute now. + consent_template_directory = path.abspath( + hs.config.user_consent_template_dir, + ) + if not path.isdir(consent_template_directory): + raise ConfigError( + "Could not find template directory '%s'" % ( + consent_template_directory, + ), + ) + + loader = jinja2.FileSystemLoader(consent_template_directory) + self._jinja_env = jinja2.Environment( + loader=loader, + autoescape=jinja2.select_autoescape(['html', 'htm', 'xml']), + ) + + if hs.config.form_secret is None: + raise ConfigError( + "Consent resource is enabled but form_secret is not set in " + "config file. It should be set to an arbitrary secret string.", + ) + + self._hmac_secret = hs.config.form_secret.encode("utf-8") + + def render_GET(self, request): + self._async_render_GET(request) + return NOT_DONE_YET + + @wrap_html_request_handler + @defer.inlineCallbacks + def _async_render_GET(self, request): + """ + Args: + request (twisted.web.http.Request): + """ + + version = parse_string(request, "v", + default=self._default_consent_version) + username = parse_string(request, "u", required=True) + userhmac = parse_string(request, "h", required=True) + + self._check_hash(username, userhmac) + + if username.startswith('@'): + qualified_user_id = username + else: + qualified_user_id = UserID(username, self.hs.hostname).to_string() + + u = yield self.store.get_user_by_id(qualified_user_id) + if u is None: + raise NotFoundError("Unknown user") + + try: + self._render_template( + request, "%s.html" % (version,), + user=username, userhmac=userhmac, version=version, + has_consented=(u["consent_version"] == version), + ) + except TemplateNotFound: + raise NotFoundError("Unknown policy version") + + def render_POST(self, request): + self._async_render_POST(request) + return NOT_DONE_YET + + @wrap_html_request_handler + @defer.inlineCallbacks + def _async_render_POST(self, request): + """ + Args: + request (twisted.web.http.Request): + """ + version = parse_string(request, "v", required=True) + username = parse_string(request, "u", required=True) + userhmac = parse_string(request, "h", required=True) + + self._check_hash(username, userhmac) + + if username.startswith('@'): + qualified_user_id = username + else: + qualified_user_id = UserID(username, self.hs.hostname).to_string() + + try: + yield self.store.user_set_consent_version(qualified_user_id, version) + except StoreError as e: + if e.code != 404: + raise + raise NotFoundError("Unknown user") + + try: + self._render_template(request, "success.html") + except TemplateNotFound: + raise NotFoundError("success.html not found") + + def _render_template(self, request, template_name, **template_args): + # get_template checks for ".." so we don't need to worry too much + # about path traversal here. + template_html = self._jinja_env.get_template( + path.join(TEMPLATE_LANGUAGE, template_name) + ) + html_bytes = template_html.render(**template_args).encode("utf8") + + request.setHeader(b"Content-Type", b"text/html; charset=utf-8") + request.setHeader(b"Content-Length", b"%i" % len(html_bytes)) + request.write(html_bytes) + finish_request(request) + + def _check_hash(self, userid, userhmac): + want_mac = hmac.new( + key=self._hmac_secret, + msg=userid, + digestmod=sha256, + ).hexdigest() + + if not compare_digest(want_mac, userhmac): + raise SynapseError(http_client.FORBIDDEN, "HMAC incorrect") diff --git a/synapse/rest/key/v1/server_key_resource.py b/synapse/rest/key/v1/server_key_resource.py index bd4fea577..b9ee6e1c1 100644 --- a/synapse/rest/key/v1/server_key_resource.py +++ b/synapse/rest/key/v1/server_key_resource.py @@ -14,14 +14,16 @@ # limitations under the License. -from twisted.web.resource import Resource -from synapse.http.server import respond_with_json_bytes -from signedjson.sign import sign_json -from unpaddedbase64 import encode_base64 -from canonicaljson import encode_canonical_json -from OpenSSL import crypto import logging +from canonicaljson import encode_canonical_json +from signedjson.sign import sign_json +from unpaddedbase64 import encode_base64 + +from OpenSSL import crypto +from twisted.web.resource import Resource + +from synapse.http.server import respond_with_json_bytes logger = logging.getLogger(__name__) @@ -49,7 +51,6 @@ class LocalKey(Resource): """ def __init__(self, hs): - self.version_string = hs.version_string self.response_body = encode_canonical_json( self.response_json_object(hs.config) ) @@ -84,7 +85,6 @@ class LocalKey(Resource): def render_GET(self, request): return respond_with_json_bytes( request, 200, self.response_body, - version_string=self.version_string ) def getChild(self, name, request): diff --git a/synapse/rest/key/v2/__init__.py b/synapse/rest/key/v2/__init__.py index a07224148..3491fd211 100644 --- a/synapse/rest/key/v2/__init__.py +++ b/synapse/rest/key/v2/__init__.py @@ -14,6 +14,7 @@ # limitations under the License. from twisted.web.resource import Resource + from .local_key_resource import LocalKey from .remote_key_resource import RemoteKey diff --git a/synapse/rest/key/v2/local_key_resource.py b/synapse/rest/key/v2/local_key_resource.py index be68d9a09..ec0ec7b43 100644 --- a/synapse/rest/key/v2/local_key_resource.py +++ b/synapse/rest/key/v2/local_key_resource.py @@ -14,13 +14,15 @@ # limitations under the License. -from twisted.web.resource import Resource -from synapse.http.server import respond_with_json_bytes -from signedjson.sign import sign_json -from unpaddedbase64 import encode_base64 -from canonicaljson import encode_canonical_json import logging +from canonicaljson import encode_canonical_json +from signedjson.sign import sign_json +from unpaddedbase64 import encode_base64 + +from twisted.web.resource import Resource + +from synapse.http.server import respond_with_json_bytes logger = logging.getLogger(__name__) @@ -63,7 +65,6 @@ class LocalKey(Resource): isLeaf = True def __init__(self, hs): - self.version_string = hs.version_string self.config = hs.config self.clock = hs.clock self.update_response_body(self.clock.time_msec()) @@ -115,5 +116,4 @@ class LocalKey(Resource): self.update_response_body(time_now) return respond_with_json_bytes( request, 200, self.response_body, - version_string=self.version_string ) diff --git a/synapse/rest/key/v2/remote_key_resource.py b/synapse/rest/key/v2/remote_key_resource.py index 9fe201365..7d67e4b06 100644 --- a/synapse/rest/key/v2/remote_key_resource.py +++ b/synapse/rest/key/v2/remote_key_resource.py @@ -12,18 +12,18 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.http.server import request_handler, respond_with_json_bytes -from synapse.http.servlet import parse_integer, parse_json_object_from_request -from synapse.api.errors import SynapseError, Codes -from synapse.crypto.keyring import KeyLookupError +import logging +from io import BytesIO +from twisted.internet import defer from twisted.web.resource import Resource from twisted.web.server import NOT_DONE_YET -from twisted.internet import defer +from synapse.api.errors import Codes, SynapseError +from synapse.crypto.keyring import KeyLookupError +from synapse.http.server import respond_with_json_bytes, wrap_json_request_handler +from synapse.http.servlet import parse_integer, parse_json_object_from_request -from io import BytesIO -import logging logger = logging.getLogger(__name__) @@ -91,14 +91,14 @@ class RemoteKey(Resource): def __init__(self, hs): self.keyring = hs.get_keyring() self.store = hs.get_datastore() - self.version_string = hs.version_string self.clock = hs.get_clock() + self.federation_domain_whitelist = hs.config.federation_domain_whitelist def render_GET(self, request): self.async_render_GET(request) return NOT_DONE_YET - @request_handler() + @wrap_json_request_handler @defer.inlineCallbacks def async_render_GET(self, request): if len(request.postpath) == 1: @@ -123,7 +123,7 @@ class RemoteKey(Resource): self.async_render_POST(request) return NOT_DONE_YET - @request_handler() + @wrap_json_request_handler @defer.inlineCallbacks def async_render_POST(self, request): content = parse_json_object_from_request(request) @@ -137,6 +137,13 @@ class RemoteKey(Resource): logger.info("Handling query for keys %r", query) store_queries = [] for server_name, key_ids in query.items(): + if ( + self.federation_domain_whitelist is not None and + server_name not in self.federation_domain_whitelist + ): + logger.debug("Federation denied with %s", server_name) + continue + if not key_ids: key_ids = (None,) for key_id in key_ids: @@ -213,7 +220,7 @@ class RemoteKey(Resource): ) except KeyLookupError as e: logger.info("Failed to fetch key: %s", e) - except: + except Exception: logger.exception("Failed to get key for %r", server_name) yield self.query_keys( request, query, query_remote_on_cache_miss=False @@ -232,5 +239,4 @@ class RemoteKey(Resource): respond_with_json_bytes( request, 200, result_io.getvalue(), - version_string=self.version_string ) diff --git a/synapse/rest/media/v0/content_repository.py b/synapse/rest/media/v0/content_repository.py index 956bd5da7..f255f2883 100644 --- a/synapse/rest/media/v0/content_repository.py +++ b/synapse/rest/media/v0/content_repository.py @@ -13,21 +13,19 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.http.server import respond_with_json_bytes, finish_request - -from synapse.api.errors import ( - Codes, cs_error -) - -from twisted.protocols.basic import FileSender -from twisted.web import server, resource - import base64 -import simplejson as json import logging import os import re +from canonicaljson import json + +from twisted.protocols.basic import FileSender +from twisted.web import resource, server + +from synapse.api.errors import Codes, cs_error +from synapse.http.server import finish_request, respond_with_json_bytes + logger = logging.getLogger(__name__) diff --git a/synapse/rest/media/v1/_base.py b/synapse/rest/media/v1/_base.py index b9600f216..65f4bd291 100644 --- a/synapse/rest/media/v1/_base.py +++ b/synapse/rest/media/v1/_base.py @@ -13,22 +13,20 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.http.server import respond_with_json, finish_request -from synapse.api.errors import ( - cs_error, Codes, SynapseError -) +import logging +import os +import urllib + +from six.moves.urllib import parse as urlparse from twisted.internet import defer from twisted.protocols.basic import FileSender +from synapse.api.errors import Codes, SynapseError, cs_error +from synapse.http.server import finish_request, respond_with_json +from synapse.util import logcontext from synapse.util.stringutils import is_ascii -import os - -import logging -import urllib -import urlparse - logger = logging.getLogger(__name__) @@ -44,7 +42,7 @@ def parse_media_id(request): except UnicodeDecodeError: pass return server_name, media_id, file_name - except: + except Exception: raise SynapseError( 404, "Invalid media id token %r" % (request.postpath,), @@ -69,42 +67,133 @@ def respond_with_file(request, media_type, file_path, logger.debug("Responding with %r", file_path) if os.path.isfile(file_path): - request.setHeader(b"Content-Type", media_type.encode("UTF-8")) - if upload_name: - if is_ascii(upload_name): - request.setHeader( - b"Content-Disposition", - b"inline; filename=%s" % ( - urllib.quote(upload_name.encode("utf-8")), - ), - ) - else: - request.setHeader( - b"Content-Disposition", - b"inline; filename*=utf-8''%s" % ( - urllib.quote(upload_name.encode("utf-8")), - ), - ) - - # cache for at least a day. - # XXX: we might want to turn this off for data we don't want to - # recommend caching as it's sensitive or private - or at least - # select private. don't bother setting Expires as all our - # clients are smart enough to be happy with Cache-Control - request.setHeader( - b"Cache-Control", b"public,max-age=86400,s-maxage=86400" - ) if file_size is None: stat = os.stat(file_path) file_size = stat.st_size - request.setHeader( - b"Content-Length", b"%d" % (file_size,) - ) + add_file_headers(request, media_type, file_size, upload_name) with open(file_path, "rb") as f: - yield FileSender().beginFileTransfer(f, request) + yield logcontext.make_deferred_yieldable( + FileSender().beginFileTransfer(f, request) + ) finish_request(request) else: respond_404(request) + + +def add_file_headers(request, media_type, file_size, upload_name): + """Adds the correct response headers in preparation for responding with the + media. + + Args: + request (twisted.web.http.Request) + media_type (str): The media/content type. + file_size (int): Size in bytes of the media, if known. + upload_name (str): The name of the requested file, if any. + """ + request.setHeader(b"Content-Type", media_type.encode("UTF-8")) + if upload_name: + if is_ascii(upload_name): + request.setHeader( + b"Content-Disposition", + b"inline; filename=%s" % ( + urllib.quote(upload_name.encode("utf-8")), + ), + ) + else: + request.setHeader( + b"Content-Disposition", + b"inline; filename*=utf-8''%s" % ( + urllib.quote(upload_name.encode("utf-8")), + ), + ) + + # cache for at least a day. + # XXX: we might want to turn this off for data we don't want to + # recommend caching as it's sensitive or private - or at least + # select private. don't bother setting Expires as all our + # clients are smart enough to be happy with Cache-Control + request.setHeader( + b"Cache-Control", b"public,max-age=86400,s-maxage=86400" + ) + + request.setHeader( + b"Content-Length", b"%d" % (file_size,) + ) + + +@defer.inlineCallbacks +def respond_with_responder(request, responder, media_type, file_size, upload_name=None): + """Responds to the request with given responder. If responder is None then + returns 404. + + Args: + request (twisted.web.http.Request) + responder (Responder|None) + media_type (str): The media/content type. + file_size (int|None): Size in bytes of the media. If not known it should be None + upload_name (str|None): The name of the requested file, if any. + """ + if not responder: + respond_404(request) + return + + logger.debug("Responding to media request with responder %s") + add_file_headers(request, media_type, file_size, upload_name) + with responder: + yield responder.write_to_consumer(request) + finish_request(request) + + +class Responder(object): + """Represents a response that can be streamed to the requester. + + Responder is a context manager which *must* be used, so that any resources + held can be cleaned up. + """ + def write_to_consumer(self, consumer): + """Stream response into consumer + + Args: + consumer (IConsumer) + + Returns: + Deferred: Resolves once the response has finished being written + """ + pass + + def __enter__(self): + pass + + def __exit__(self, exc_type, exc_val, exc_tb): + pass + + +class FileInfo(object): + """Details about a requested/uploaded file. + + Attributes: + server_name (str): The server name where the media originated from, + or None if local. + file_id (str): The local ID of the file. For local files this is the + same as the media_id + url_cache (bool): If the file is for the url preview cache + thumbnail (bool): Whether the file is a thumbnail or not. + thumbnail_width (int) + thumbnail_height (int) + thumbnail_method (str) + thumbnail_type (str): Content type of thumbnail, e.g. image/png + """ + def __init__(self, server_name, file_id, url_cache=False, + thumbnail=False, thumbnail_width=None, thumbnail_height=None, + thumbnail_method=None, thumbnail_type=None): + self.server_name = server_name + self.file_id = file_id + self.url_cache = url_cache + self.thumbnail = thumbnail + self.thumbnail_width = thumbnail_width + self.thumbnail_height = thumbnail_height + self.thumbnail_method = thumbnail_method + self.thumbnail_type = thumbnail_type diff --git a/synapse/rest/media/v1/download_resource.py b/synapse/rest/media/v1/download_resource.py index 6879249c8..fbfa85f74 100644 --- a/synapse/rest/media/v1/download_resource.py +++ b/synapse/rest/media/v1/download_resource.py @@ -12,17 +12,17 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. -import synapse.http.servlet - -from ._base import parse_media_id, respond_with_file, respond_404 -from twisted.web.resource import Resource -from synapse.http.server import request_handler, set_cors_headers - -from twisted.web.server import NOT_DONE_YET -from twisted.internet import defer - import logging +from twisted.internet import defer +from twisted.web.resource import Resource +from twisted.web.server import NOT_DONE_YET + +import synapse.http.servlet +from synapse.http.server import set_cors_headers, wrap_json_request_handler + +from ._base import parse_media_id, respond_404 + logger = logging.getLogger(__name__) @@ -32,18 +32,17 @@ class DownloadResource(Resource): def __init__(self, hs, media_repo): Resource.__init__(self) - self.filepaths = media_repo.filepaths self.media_repo = media_repo self.server_name = hs.hostname - self.store = hs.get_datastore() - self.version_string = hs.version_string + + # this is expected by @wrap_json_request_handler self.clock = hs.get_clock() def render_GET(self, request): self._async_render_GET(request) return NOT_DONE_YET - @request_handler() + @wrap_json_request_handler @defer.inlineCallbacks def _async_render_GET(self, request): set_cors_headers(request) @@ -57,59 +56,16 @@ class DownloadResource(Resource): ) server_name, media_id, name = parse_media_id(request) if server_name == self.server_name: - yield self._respond_local_file(request, media_id, name) + yield self.media_repo.get_local_media(request, media_id, name) else: - yield self._respond_remote_file( - request, server_name, media_id, name - ) + allow_remote = synapse.http.servlet.parse_boolean( + request, "allow_remote", default=True) + if not allow_remote: + logger.info( + "Rejecting request for remote media %s/%s due to allow_remote", + server_name, media_id, + ) + respond_404(request) + return - @defer.inlineCallbacks - def _respond_local_file(self, request, media_id, name): - media_info = yield self.store.get_local_media(media_id) - if not media_info or media_info["quarantined_by"]: - respond_404(request) - return - - media_type = media_info["media_type"] - media_length = media_info["media_length"] - upload_name = name if name else media_info["upload_name"] - if media_info["url_cache"]: - # TODO: Check the file still exists, if it doesn't we can redownload - # it from the url `media_info["url_cache"]` - file_path = self.filepaths.url_cache_filepath(media_id) - else: - file_path = self.filepaths.local_media_filepath(media_id) - - yield respond_with_file( - request, media_type, file_path, media_length, - upload_name=upload_name, - ) - - @defer.inlineCallbacks - def _respond_remote_file(self, request, server_name, media_id, name): - # don't forward requests for remote media if allow_remote is false - allow_remote = synapse.http.servlet.parse_boolean( - request, "allow_remote", default=True) - if not allow_remote: - logger.info( - "Rejecting request for remote media %s/%s due to allow_remote", - server_name, media_id, - ) - respond_404(request) - return - - media_info = yield self.media_repo.get_remote_media(server_name, media_id) - - media_type = media_info["media_type"] - media_length = media_info["media_length"] - filesystem_id = media_info["filesystem_id"] - upload_name = name if name else media_info["upload_name"] - - file_path = self.filepaths.remote_media_filepath( - server_name, filesystem_id - ) - - yield respond_with_file( - request, media_type, file_path, media_length, - upload_name=upload_name, - ) + yield self.media_repo.get_remote_media(request, server_name, media_id, name) diff --git a/synapse/rest/media/v1/filepath.py b/synapse/rest/media/v1/filepath.py index d92b7ff33..c8586fa28 100644 --- a/synapse/rest/media/v1/filepath.py +++ b/synapse/rest/media/v1/filepath.py @@ -13,79 +13,201 @@ # See the License for the specific language governing permissions and # limitations under the License. +import functools import os +import re + +NEW_FORMAT_ID_RE = re.compile(r"^\d\d\d\d-\d\d-\d\d") + + +def _wrap_in_base_path(func): + """Takes a function that returns a relative path and turns it into an + absolute path based on the location of the primary media store + """ + @functools.wraps(func) + def _wrapped(self, *args, **kwargs): + path = func(self, *args, **kwargs) + return os.path.join(self.base_path, path) + + return _wrapped class MediaFilePaths(object): + """Describes where files are stored on disk. - def __init__(self, base_path): - self.base_path = base_path + Most of the functions have a `*_rel` variant which returns a file path that + is relative to the base media store path. This is mainly used when we want + to write to the backup media store (when one is configured) + """ - def default_thumbnail(self, default_top_level, default_sub_type, width, - height, content_type, method): + def __init__(self, primary_base_path): + self.base_path = primary_base_path + + def default_thumbnail_rel(self, default_top_level, default_sub_type, width, + height, content_type, method): top_level_type, sub_type = content_type.split("/") file_name = "%i-%i-%s-%s-%s" % ( width, height, top_level_type, sub_type, method ) return os.path.join( - self.base_path, "default_thumbnails", default_top_level, + "default_thumbnails", default_top_level, default_sub_type, file_name ) - def local_media_filepath(self, media_id): + default_thumbnail = _wrap_in_base_path(default_thumbnail_rel) + + def local_media_filepath_rel(self, media_id): return os.path.join( - self.base_path, "local_content", + "local_content", media_id[0:2], media_id[2:4], media_id[4:] ) - def local_media_thumbnail(self, media_id, width, height, content_type, - method): + local_media_filepath = _wrap_in_base_path(local_media_filepath_rel) + + def local_media_thumbnail_rel(self, media_id, width, height, content_type, + method): top_level_type, sub_type = content_type.split("/") file_name = "%i-%i-%s-%s-%s" % ( width, height, top_level_type, sub_type, method ) return os.path.join( - self.base_path, "local_thumbnails", + "local_thumbnails", media_id[0:2], media_id[2:4], media_id[4:], file_name ) - def remote_media_filepath(self, server_name, file_id): + local_media_thumbnail = _wrap_in_base_path(local_media_thumbnail_rel) + + def remote_media_filepath_rel(self, server_name, file_id): return os.path.join( - self.base_path, "remote_content", server_name, + "remote_content", server_name, file_id[0:2], file_id[2:4], file_id[4:] ) - def remote_media_thumbnail(self, server_name, file_id, width, height, - content_type, method): + remote_media_filepath = _wrap_in_base_path(remote_media_filepath_rel) + + def remote_media_thumbnail_rel(self, server_name, file_id, width, height, + content_type, method): top_level_type, sub_type = content_type.split("/") file_name = "%i-%i-%s-%s" % (width, height, top_level_type, sub_type) return os.path.join( - self.base_path, "remote_thumbnail", server_name, + "remote_thumbnail", server_name, file_id[0:2], file_id[2:4], file_id[4:], file_name ) + remote_media_thumbnail = _wrap_in_base_path(remote_media_thumbnail_rel) + def remote_media_thumbnail_dir(self, server_name, file_id): return os.path.join( self.base_path, "remote_thumbnail", server_name, file_id[0:2], file_id[2:4], file_id[4:], ) - def url_cache_filepath(self, media_id): - return os.path.join( - self.base_path, "url_cache", - media_id[0:2], media_id[2:4], media_id[4:] - ) + def url_cache_filepath_rel(self, media_id): + if NEW_FORMAT_ID_RE.match(media_id): + # Media id is of the form + # E.g.: 2017-09-28-fsdRDt24DS234dsf + return os.path.join( + "url_cache", + media_id[:10], media_id[11:] + ) + else: + return os.path.join( + "url_cache", + media_id[0:2], media_id[2:4], media_id[4:], + ) + + url_cache_filepath = _wrap_in_base_path(url_cache_filepath_rel) + + def url_cache_filepath_dirs_to_delete(self, media_id): + "The dirs to try and remove if we delete the media_id file" + if NEW_FORMAT_ID_RE.match(media_id): + return [ + os.path.join( + self.base_path, "url_cache", + media_id[:10], + ), + ] + else: + return [ + os.path.join( + self.base_path, "url_cache", + media_id[0:2], media_id[2:4], + ), + os.path.join( + self.base_path, "url_cache", + media_id[0:2], + ), + ] + + def url_cache_thumbnail_rel(self, media_id, width, height, content_type, + method): + # Media id is of the form + # E.g.: 2017-09-28-fsdRDt24DS234dsf - def url_cache_thumbnail(self, media_id, width, height, content_type, - method): top_level_type, sub_type = content_type.split("/") file_name = "%i-%i-%s-%s-%s" % ( width, height, top_level_type, sub_type, method ) - return os.path.join( - self.base_path, "url_cache_thumbnails", - media_id[0:2], media_id[2:4], media_id[4:], - file_name - ) + + if NEW_FORMAT_ID_RE.match(media_id): + return os.path.join( + "url_cache_thumbnails", + media_id[:10], media_id[11:], + file_name + ) + else: + return os.path.join( + "url_cache_thumbnails", + media_id[0:2], media_id[2:4], media_id[4:], + file_name + ) + + url_cache_thumbnail = _wrap_in_base_path(url_cache_thumbnail_rel) + + def url_cache_thumbnail_directory(self, media_id): + # Media id is of the form + # E.g.: 2017-09-28-fsdRDt24DS234dsf + + if NEW_FORMAT_ID_RE.match(media_id): + return os.path.join( + self.base_path, "url_cache_thumbnails", + media_id[:10], media_id[11:], + ) + else: + return os.path.join( + self.base_path, "url_cache_thumbnails", + media_id[0:2], media_id[2:4], media_id[4:], + ) + + def url_cache_thumbnail_dirs_to_delete(self, media_id): + "The dirs to try and remove if we delete the media_id thumbnails" + # Media id is of the form + # E.g.: 2017-09-28-fsdRDt24DS234dsf + if NEW_FORMAT_ID_RE.match(media_id): + return [ + os.path.join( + self.base_path, "url_cache_thumbnails", + media_id[:10], media_id[11:], + ), + os.path.join( + self.base_path, "url_cache_thumbnails", + media_id[:10], + ), + ] + else: + return [ + os.path.join( + self.base_path, "url_cache_thumbnails", + media_id[0:2], media_id[2:4], media_id[4:], + ), + os.path.join( + self.base_path, "url_cache_thumbnails", + media_id[0:2], media_id[2:4], + ), + os.path.join( + self.base_path, "url_cache_thumbnails", + media_id[0:2], + ), + ] diff --git a/synapse/rest/media/v1/identicon_resource.py b/synapse/rest/media/v1/identicon_resource.py index 66f2b6bd3..bdbd8d50d 100644 --- a/synapse/rest/media/v1/identicon_resource.py +++ b/synapse/rest/media/v1/identicon_resource.py @@ -13,8 +13,11 @@ # limitations under the License. from pydenticon import Generator + from twisted.web.resource import Resource +from synapse.http.servlet import parse_integer + FOREGROUND = [ "rgb(45,79,255)", "rgb(254,180,44)", @@ -55,8 +58,8 @@ class IdenticonResource(Resource): def render_GET(self, request): name = "/".join(request.postpath) - width = int(request.args.get("width", [96])[0]) - height = int(request.args.get("height", [96])[0]) + width = parse_integer(request, "width", default=96) + height = parse_integer(request, "height", default=96) identicon_bytes = self.generate_identicon(name, width, height) request.setHeader(b"Content-Type", b"image/png") request.setHeader( diff --git a/synapse/rest/media/v1/media_repository.py b/synapse/rest/media/v1/media_repository.py index 0ea1248ce..30242c525 100644 --- a/synapse/rest/media/v1/media_repository.py +++ b/synapse/rest/media/v1/media_repository.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2014-2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -13,45 +14,52 @@ # See the License for the specific language governing permissions and # limitations under the License. -from twisted.internet import defer, threads -import twisted.internet.error -import twisted.web.http -from twisted.web.resource import Resource - -from .upload_resource import UploadResource -from .download_resource import DownloadResource -from .thumbnail_resource import ThumbnailResource -from .identicon_resource import IdenticonResource -from .preview_url_resource import PreviewUrlResource -from .filepath import MediaFilePaths -from .thumbnailer import Thumbnailer - -from synapse.http.matrixfederationclient import MatrixFederationHttpClient -from synapse.util.stringutils import random_string -from synapse.api.errors import SynapseError, HttpResponseException, \ - NotFoundError - -from synapse.util.async import Linearizer -from synapse.util.stringutils import is_ascii -from synapse.util.logcontext import preserve_context_over_fn -from synapse.util.retryutils import NotRetryingDestination - -import os +import cgi import errno +import logging +import os import shutil -import cgi -import logging -import urlparse +from six import iteritems +from six.moves.urllib import parse as urlparse + +import twisted.internet.error +import twisted.web.http +from twisted.internet import defer, threads +from twisted.web.resource import Resource + +from synapse.api.errors import ( + FederationDeniedError, + HttpResponseException, + NotFoundError, + SynapseError, +) +from synapse.http.matrixfederationclient import MatrixFederationHttpClient +from synapse.util.async import Linearizer +from synapse.util.logcontext import make_deferred_yieldable +from synapse.util.retryutils import NotRetryingDestination +from synapse.util.stringutils import is_ascii, random_string + +from ._base import FileInfo, respond_404, respond_with_responder +from .download_resource import DownloadResource +from .filepath import MediaFilePaths +from .identicon_resource import IdenticonResource +from .media_storage import MediaStorage +from .preview_url_resource import PreviewUrlResource +from .storage_provider import StorageProviderWrapper +from .thumbnail_resource import ThumbnailResource +from .thumbnailer import Thumbnailer +from .upload_resource import UploadResource logger = logging.getLogger(__name__) -UPDATE_RECENTLY_ACCESSED_REMOTES_TS = 60 * 1000 +UPDATE_RECENTLY_ACCESSED_TS = 60 * 1000 class MediaRepository(object): def __init__(self, hs): + self.hs = hs self.auth = hs.get_auth() self.client = MatrixFederationHttpClient(hs) self.clock = hs.get_clock() @@ -59,46 +67,90 @@ class MediaRepository(object): self.store = hs.get_datastore() self.max_upload_size = hs.config.max_upload_size self.max_image_pixels = hs.config.max_image_pixels - self.filepaths = MediaFilePaths(hs.config.media_store_path) + + self.primary_base_path = hs.config.media_store_path + self.filepaths = MediaFilePaths(self.primary_base_path) + self.dynamic_thumbnails = hs.config.dynamic_thumbnails self.thumbnail_requirements = hs.config.thumbnail_requirements self.remote_media_linearizer = Linearizer(name="media_remote") self.recently_accessed_remotes = set() + self.recently_accessed_locals = set() + + self.federation_domain_whitelist = hs.config.federation_domain_whitelist + + # List of StorageProviders where we should search for media and + # potentially upload to. + storage_providers = [] + + for clz, provider_config, wrapper_config in hs.config.media_storage_providers: + backend = clz(hs, provider_config) + provider = StorageProviderWrapper( + backend, + store_local=wrapper_config.store_local, + store_remote=wrapper_config.store_remote, + store_synchronous=wrapper_config.store_synchronous, + ) + storage_providers.append(provider) + + self.media_storage = MediaStorage( + self.hs, self.primary_base_path, self.filepaths, storage_providers, + ) self.clock.looping_call( - self._update_recently_accessed_remotes, - UPDATE_RECENTLY_ACCESSED_REMOTES_TS + self._update_recently_accessed, + UPDATE_RECENTLY_ACCESSED_TS, ) @defer.inlineCallbacks - def _update_recently_accessed_remotes(self): - media = self.recently_accessed_remotes + def _update_recently_accessed(self): + remote_media = self.recently_accessed_remotes self.recently_accessed_remotes = set() + local_media = self.recently_accessed_locals + self.recently_accessed_locals = set() + yield self.store.update_cached_last_access_time( - media, self.clock.time_msec() + local_media, remote_media, self.clock.time_msec() ) - @staticmethod - def _makedirs(filepath): - dirname = os.path.dirname(filepath) - if not os.path.exists(dirname): - os.makedirs(dirname) + def mark_recently_accessed(self, server_name, media_id): + """Mark the given media as recently accessed. + + Args: + server_name (str|None): Origin server of media, or None if local + media_id (str): The media ID of the content + """ + if server_name: + self.recently_accessed_remotes.add((server_name, media_id)) + else: + self.recently_accessed_locals.add(media_id) @defer.inlineCallbacks def create_content(self, media_type, upload_name, content, content_length, auth_user): + """Store uploaded content for a local user and return the mxc URL + + Args: + media_type(str): The content type of the file + upload_name(str): The name of the file + content: A file like object that is the content to store + content_length(int): The length of the content + auth_user(str): The user_id of the uploader + + Returns: + Deferred[str]: The mxc url of the stored content + """ media_id = random_string(24) - fname = self.filepaths.local_media_filepath(media_id) - self._makedirs(fname) + file_info = FileInfo( + server_name=None, + file_id=media_id, + ) - # This shouldn't block for very long because the content will have - # already been uploaded at this point. - with open(fname, "wb") as f: - f.write(content) + fname = yield self.media_storage.store_file(content, file_info) logger.info("Stored local media in file %r", fname) @@ -110,131 +162,275 @@ class MediaRepository(object): media_length=content_length, user_id=auth_user, ) - media_info = { - "media_type": media_type, - "media_length": content_length, - } - yield self._generate_local_thumbnails(media_id, media_info) + yield self._generate_thumbnails( + None, media_id, media_id, media_type, + ) defer.returnValue("mxc://%s/%s" % (self.server_name, media_id)) @defer.inlineCallbacks - def get_remote_media(self, server_name, media_id): + def get_local_media(self, request, media_id, name): + """Responds to reqests for local media, if exists, or returns 404. + + Args: + request(twisted.web.http.Request) + media_id (str): The media ID of the content. (This is the same as + the file_id for local content.) + name (str|None): Optional name that, if specified, will be used as + the filename in the Content-Disposition header of the response. + + Returns: + Deferred: Resolves once a response has successfully been written + to request + """ + media_info = yield self.store.get_local_media(media_id) + if not media_info or media_info["quarantined_by"]: + respond_404(request) + return + + self.mark_recently_accessed(None, media_id) + + media_type = media_info["media_type"] + media_length = media_info["media_length"] + upload_name = name if name else media_info["upload_name"] + url_cache = media_info["url_cache"] + + file_info = FileInfo( + None, media_id, + url_cache=url_cache, + ) + + responder = yield self.media_storage.fetch_media(file_info) + yield respond_with_responder( + request, responder, media_type, media_length, upload_name, + ) + + @defer.inlineCallbacks + def get_remote_media(self, request, server_name, media_id, name): + """Respond to requests for remote media. + + Args: + request(twisted.web.http.Request) + server_name (str): Remote server_name where the media originated. + media_id (str): The media ID of the content (as defined by the + remote server). + name (str|None): Optional name that, if specified, will be used as + the filename in the Content-Disposition header of the response. + + Returns: + Deferred: Resolves once a response has successfully been written + to request + """ + if ( + self.federation_domain_whitelist is not None and + server_name not in self.federation_domain_whitelist + ): + raise FederationDeniedError(server_name) + + self.mark_recently_accessed(server_name, media_id) + + # We linearize here to ensure that we don't try and download remote + # media multiple times concurrently key = (server_name, media_id) with (yield self.remote_media_linearizer.queue(key)): - media_info = yield self._get_remote_media_impl(server_name, media_id) + responder, media_info = yield self._get_remote_media_impl( + server_name, media_id, + ) + + # We deliberately stream the file outside the lock + if responder: + media_type = media_info["media_type"] + media_length = media_info["media_length"] + upload_name = name if name else media_info["upload_name"] + yield respond_with_responder( + request, responder, media_type, media_length, upload_name, + ) + else: + respond_404(request) + + @defer.inlineCallbacks + def get_remote_media_info(self, server_name, media_id): + """Gets the media info associated with the remote file, downloading + if necessary. + + Args: + server_name (str): Remote server_name where the media originated. + media_id (str): The media ID of the content (as defined by the + remote server). + + Returns: + Deferred[dict]: The media_info of the file + """ + if ( + self.federation_domain_whitelist is not None and + server_name not in self.federation_domain_whitelist + ): + raise FederationDeniedError(server_name) + + # We linearize here to ensure that we don't try and download remote + # media multiple times concurrently + key = (server_name, media_id) + with (yield self.remote_media_linearizer.queue(key)): + responder, media_info = yield self._get_remote_media_impl( + server_name, media_id, + ) + + # Ensure we actually use the responder so that it releases resources + if responder: + with responder: + pass + defer.returnValue(media_info) @defer.inlineCallbacks def _get_remote_media_impl(self, server_name, media_id): + """Looks for media in local cache, if not there then attempt to + download from remote server. + + Args: + server_name (str): Remote server_name where the media originated. + media_id (str): The media ID of the content (as defined by the + remote server). + + Returns: + Deferred[(Responder, media_info)] + """ media_info = yield self.store.get_cached_remote_media( server_name, media_id ) - if not media_info: - media_info = yield self._download_remote_file( - server_name, media_id - ) - elif media_info["quarantined_by"]: - raise NotFoundError() + + # file_id is the ID we use to track the file locally. If we've already + # seen the file then reuse the existing ID, otherwise genereate a new + # one. + if media_info: + file_id = media_info["filesystem_id"] else: - self.recently_accessed_remotes.add((server_name, media_id)) - yield self.store.update_cached_last_access_time( - [(server_name, media_id)], self.clock.time_msec() - ) - defer.returnValue(media_info) + file_id = random_string(24) + + file_info = FileInfo(server_name, file_id) + + # If we have an entry in the DB, try and look for it + if media_info: + if media_info["quarantined_by"]: + logger.info("Media is quarantined") + raise NotFoundError() + + responder = yield self.media_storage.fetch_media(file_info) + if responder: + defer.returnValue((responder, media_info)) + + # Failed to find the file anywhere, lets download it. + + media_info = yield self._download_remote_file( + server_name, media_id, file_id + ) + + responder = yield self.media_storage.fetch_media(file_info) + defer.returnValue((responder, media_info)) @defer.inlineCallbacks - def _download_remote_file(self, server_name, media_id): - file_id = random_string(24) + def _download_remote_file(self, server_name, media_id, file_id): + """Attempt to download the remote file from the given server name, + using the given file_id as the local id. - fname = self.filepaths.remote_media_filepath( - server_name, file_id + Args: + server_name (str): Originating server + media_id (str): The media ID of the content (as defined by the + remote server). This is different than the file_id, which is + locally generated. + file_id (str): Local file ID + + Returns: + Deferred[MediaInfo] + """ + + file_info = FileInfo( + server_name=server_name, + file_id=file_id, ) - self._makedirs(fname) - try: - with open(fname, "wb") as f: - request_path = "/".join(( - "/_matrix/media/v1/download", server_name, media_id, - )) + with self.media_storage.store_into_file(file_info) as (f, fname, finish): + request_path = "/".join(( + "/_matrix/media/v1/download", server_name, media_id, + )) + try: + length, headers = yield self.client.get_file( + server_name, request_path, output_stream=f, + max_size=self.max_upload_size, args={ + # tell the remote server to 404 if it doesn't + # recognise the server_name, to make sure we don't + # end up with a routing loop. + "allow_remote": "false", + } + ) + except twisted.internet.error.DNSLookupError as e: + logger.warn("HTTP error fetching remote media %s/%s: %r", + server_name, media_id, e) + raise NotFoundError() + + except HttpResponseException as e: + logger.warn("HTTP error fetching remote media %s/%s: %s", + server_name, media_id, e.response) + if e.code == twisted.web.http.NOT_FOUND: + raise SynapseError.from_http_response_exception(e) + raise SynapseError(502, "Failed to fetch remote media") + + except SynapseError: + logger.exception("Failed to fetch remote media %s/%s", + server_name, media_id) + raise + except NotRetryingDestination: + logger.warn("Not retrying destination %r", server_name) + raise SynapseError(502, "Failed to fetch remote media") + except Exception: + logger.exception("Failed to fetch remote media %s/%s", + server_name, media_id) + raise SynapseError(502, "Failed to fetch remote media") + + yield finish() + + media_type = headers["Content-Type"][0] + + time_now_ms = self.clock.time_msec() + + content_disposition = headers.get("Content-Disposition", None) + if content_disposition: + _, params = cgi.parse_header(content_disposition[0],) + upload_name = None + + # First check if there is a valid UTF-8 filename + upload_name_utf8 = params.get("filename*", None) + if upload_name_utf8: + if upload_name_utf8.lower().startswith("utf-8''"): + upload_name = upload_name_utf8[7:] + + # If there isn't check for an ascii name. + if not upload_name: + upload_name_ascii = params.get("filename", None) + if upload_name_ascii and is_ascii(upload_name_ascii): + upload_name = upload_name_ascii + + if upload_name: + upload_name = urlparse.unquote(upload_name) try: - length, headers = yield self.client.get_file( - server_name, request_path, output_stream=f, - max_size=self.max_upload_size, args={ - # tell the remote server to 404 if it doesn't - # recognise the server_name, to make sure we don't - # end up with a routing loop. - "allow_remote": "false", - } - ) - except twisted.internet.error.DNSLookupError as e: - logger.warn("HTTP error fetching remote media %s/%s: %r", - server_name, media_id, e) - raise NotFoundError() + upload_name = upload_name.decode("utf-8") + except UnicodeDecodeError: + upload_name = None + else: + upload_name = None - except HttpResponseException as e: - logger.warn("HTTP error fetching remote media %s/%s: %s", - server_name, media_id, e.response) - if e.code == twisted.web.http.NOT_FOUND: - raise SynapseError.from_http_response_exception(e) - raise SynapseError(502, "Failed to fetch remote media") + logger.info("Stored remote media in file %r", fname) - except SynapseError: - logger.exception("Failed to fetch remote media %s/%s", - server_name, media_id) - raise - except NotRetryingDestination: - logger.warn("Not retrying destination %r", server_name) - raise SynapseError(502, "Failed to fetch remote media") - except Exception: - logger.exception("Failed to fetch remote media %s/%s", - server_name, media_id) - raise SynapseError(502, "Failed to fetch remote media") - - media_type = headers["Content-Type"][0] - time_now_ms = self.clock.time_msec() - - content_disposition = headers.get("Content-Disposition", None) - if content_disposition: - _, params = cgi.parse_header(content_disposition[0],) - upload_name = None - - # First check if there is a valid UTF-8 filename - upload_name_utf8 = params.get("filename*", None) - if upload_name_utf8: - if upload_name_utf8.lower().startswith("utf-8''"): - upload_name = upload_name_utf8[7:] - - # If there isn't check for an ascii name. - if not upload_name: - upload_name_ascii = params.get("filename", None) - if upload_name_ascii and is_ascii(upload_name_ascii): - upload_name = upload_name_ascii - - if upload_name: - upload_name = urlparse.unquote(upload_name) - try: - upload_name = upload_name.decode("utf-8") - except UnicodeDecodeError: - upload_name = None - else: - upload_name = None - - logger.info("Stored remote media in file %r", fname) - - yield self.store.store_cached_remote_media( - origin=server_name, - media_id=media_id, - media_type=media_type, - time_now_ms=self.clock.time_msec(), - upload_name=upload_name, - media_length=length, - filesystem_id=file_id, - ) - except: - os.remove(fname) - raise + yield self.store.store_cached_remote_media( + origin=server_name, + media_id=media_id, + media_type=media_type, + time_now_ms=self.clock.time_msec(), + upload_name=upload_name, + media_length=length, + filesystem_id=file_id, + ) media_info = { "media_type": media_type, @@ -244,8 +440,8 @@ class MediaRepository(object): "filesystem_id": file_id, } - yield self._generate_remote_thumbnails( - server_name, media_id, media_info + yield self._generate_thumbnails( + server_name, media_id, file_id, media_type, ) defer.returnValue(media_info) @@ -253,9 +449,8 @@ class MediaRepository(object): def _get_thumbnail_requirements(self, media_type): return self.thumbnail_requirements.get(media_type, ()) - def _generate_thumbnail(self, input_path, t_path, t_width, t_height, + def _generate_thumbnail(self, thumbnailer, t_width, t_height, t_method, t_type): - thumbnailer = Thumbnailer(input_path) m_width = thumbnailer.width m_height = thumbnailer.height @@ -267,75 +462,125 @@ class MediaRepository(object): return if t_method == "crop": - t_len = thumbnailer.crop(t_path, t_width, t_height, t_type) + t_byte_source = thumbnailer.crop(t_width, t_height, t_type) elif t_method == "scale": t_width, t_height = thumbnailer.aspect(t_width, t_height) t_width = min(m_width, t_width) t_height = min(m_height, t_height) - t_len = thumbnailer.scale(t_path, t_width, t_height, t_type) + t_byte_source = thumbnailer.scale(t_width, t_height, t_type) else: - t_len = None + t_byte_source = None - return t_len + return t_byte_source @defer.inlineCallbacks def generate_local_exact_thumbnail(self, media_id, t_width, t_height, - t_method, t_type): - input_path = self.filepaths.local_media_filepath(media_id) + t_method, t_type, url_cache): + input_path = yield self.media_storage.ensure_media_is_in_local_cache(FileInfo( + None, media_id, url_cache=url_cache, + )) - t_path = self.filepaths.local_media_thumbnail( - media_id, t_width, t_height, t_type, t_method - ) - self._makedirs(t_path) - - t_len = yield preserve_context_over_fn( - threads.deferToThread, + thumbnailer = Thumbnailer(input_path) + t_byte_source = yield make_deferred_yieldable(threads.deferToThread( self._generate_thumbnail, - input_path, t_path, t_width, t_height, t_method, t_type - ) + thumbnailer, t_width, t_height, t_method, t_type + )) + + if t_byte_source: + try: + file_info = FileInfo( + server_name=None, + file_id=media_id, + url_cache=url_cache, + thumbnail=True, + thumbnail_width=t_width, + thumbnail_height=t_height, + thumbnail_method=t_method, + thumbnail_type=t_type, + ) + + output_path = yield self.media_storage.store_file( + t_byte_source, file_info, + ) + finally: + t_byte_source.close() + + logger.info("Stored thumbnail in file %r", output_path) + + t_len = os.path.getsize(output_path) - if t_len: yield self.store.store_local_thumbnail( media_id, t_width, t_height, t_type, t_method, t_len ) - defer.returnValue(t_path) + defer.returnValue(output_path) @defer.inlineCallbacks def generate_remote_exact_thumbnail(self, server_name, file_id, media_id, t_width, t_height, t_method, t_type): - input_path = self.filepaths.remote_media_filepath(server_name, file_id) + input_path = yield self.media_storage.ensure_media_is_in_local_cache(FileInfo( + server_name, file_id, url_cache=False, + )) - t_path = self.filepaths.remote_media_thumbnail( - server_name, file_id, t_width, t_height, t_type, t_method - ) - self._makedirs(t_path) - - t_len = yield preserve_context_over_fn( - threads.deferToThread, + thumbnailer = Thumbnailer(input_path) + t_byte_source = yield make_deferred_yieldable(threads.deferToThread( self._generate_thumbnail, - input_path, t_path, t_width, t_height, t_method, t_type - ) + thumbnailer, t_width, t_height, t_method, t_type + )) + + if t_byte_source: + try: + file_info = FileInfo( + server_name=server_name, + file_id=media_id, + thumbnail=True, + thumbnail_width=t_width, + thumbnail_height=t_height, + thumbnail_method=t_method, + thumbnail_type=t_type, + ) + + output_path = yield self.media_storage.store_file( + t_byte_source, file_info, + ) + finally: + t_byte_source.close() + + logger.info("Stored thumbnail in file %r", output_path) + + t_len = os.path.getsize(output_path) - if t_len: yield self.store.store_remote_media_thumbnail( server_name, media_id, file_id, t_width, t_height, t_type, t_method, t_len ) - defer.returnValue(t_path) + defer.returnValue(output_path) @defer.inlineCallbacks - def _generate_local_thumbnails(self, media_id, media_info, url_cache=False): - media_type = media_info["media_type"] + def _generate_thumbnails(self, server_name, media_id, file_id, media_type, + url_cache=False): + """Generate and store thumbnails for an image. + + Args: + server_name (str|None): The server name if remote media, else None if local + media_id (str): The media ID of the content. (This is the same as + the file_id for local content) + file_id (str): Local file ID + media_type (str): The content type of the file + url_cache (bool): If we are thumbnailing images downloaded for the URL cache, + used exclusively by the url previewer + + Returns: + Deferred[dict]: Dict with "width" and "height" keys of original image + """ requirements = self._get_thumbnail_requirements(media_type) if not requirements: return - if url_cache: - input_path = self.filepaths.url_cache_filepath(media_id) - else: - input_path = self.filepaths.local_media_filepath(media_id) + input_path = yield self.media_storage.ensure_media_is_in_local_cache(FileInfo( + server_name, file_id, url_cache=url_cache, + )) thumbnailer = Thumbnailer(input_path) m_width = thumbnailer.width @@ -348,135 +593,68 @@ class MediaRepository(object): ) return - local_thumbnails = [] + # We deduplicate the thumbnail sizes by ignoring the cropped versions if + # they have the same dimensions of a scaled one. + thumbnails = {} + for r_width, r_height, r_method, r_type in requirements: + if r_method == "crop": + thumbnails.setdefault((r_width, r_height, r_type), r_method) + elif r_method == "scale": + t_width, t_height = thumbnailer.aspect(r_width, r_height) + t_width = min(m_width, t_width) + t_height = min(m_height, t_height) + thumbnails[(t_width, t_height, r_type)] = r_method - def generate_thumbnails(): - scales = set() - crops = set() - for r_width, r_height, r_method, r_type in requirements: - if r_method == "scale": - t_width, t_height = thumbnailer.aspect(r_width, r_height) - scales.add(( - min(m_width, t_width), min(m_height, t_height), r_type, - )) - elif r_method == "crop": - crops.add((r_width, r_height, r_type)) - - for t_width, t_height, t_type in scales: - t_method = "scale" - if url_cache: - t_path = self.filepaths.url_cache_thumbnail( - media_id, t_width, t_height, t_type, t_method - ) - else: - t_path = self.filepaths.local_media_thumbnail( - media_id, t_width, t_height, t_type, t_method - ) - self._makedirs(t_path) - t_len = thumbnailer.scale(t_path, t_width, t_height, t_type) - - local_thumbnails.append(( - media_id, t_width, t_height, t_type, t_method, t_len + # Now we generate the thumbnails for each dimension, store it + for (t_width, t_height, t_type), t_method in iteritems(thumbnails): + # Generate the thumbnail + if t_method == "crop": + t_byte_source = yield make_deferred_yieldable(threads.deferToThread( + thumbnailer.crop, + t_width, t_height, t_type, )) - - for t_width, t_height, t_type in crops: - if (t_width, t_height, t_type) in scales: - # If the aspect ratio of the cropped thumbnail matches a purely - # scaled one then there is no point in calculating a separate - # thumbnail. - continue - t_method = "crop" - if url_cache: - t_path = self.filepaths.url_cache_thumbnail( - media_id, t_width, t_height, t_type, t_method - ) - else: - t_path = self.filepaths.local_media_thumbnail( - media_id, t_width, t_height, t_type, t_method - ) - self._makedirs(t_path) - t_len = thumbnailer.crop(t_path, t_width, t_height, t_type) - local_thumbnails.append(( - media_id, t_width, t_height, t_type, t_method, t_len + elif t_method == "scale": + t_byte_source = yield make_deferred_yieldable(threads.deferToThread( + thumbnailer.scale, + t_width, t_height, t_type, )) + else: + logger.error("Unrecognized method: %r", t_method) + continue - yield preserve_context_over_fn(threads.deferToThread, generate_thumbnails) + if not t_byte_source: + continue - for l in local_thumbnails: - yield self.store.store_local_thumbnail(*l) - - defer.returnValue({ - "width": m_width, - "height": m_height, - }) - - @defer.inlineCallbacks - def _generate_remote_thumbnails(self, server_name, media_id, media_info): - media_type = media_info["media_type"] - file_id = media_info["filesystem_id"] - requirements = self._get_thumbnail_requirements(media_type) - if not requirements: - return - - remote_thumbnails = [] - - input_path = self.filepaths.remote_media_filepath(server_name, file_id) - thumbnailer = Thumbnailer(input_path) - m_width = thumbnailer.width - m_height = thumbnailer.height - - def generate_thumbnails(): - if m_width * m_height >= self.max_image_pixels: - logger.info( - "Image too large to thumbnail %r x %r > %r", - m_width, m_height, self.max_image_pixels + try: + file_info = FileInfo( + server_name=server_name, + file_id=file_id, + thumbnail=True, + thumbnail_width=t_width, + thumbnail_height=t_height, + thumbnail_method=t_method, + thumbnail_type=t_type, + url_cache=url_cache, ) - return - scales = set() - crops = set() - for r_width, r_height, r_method, r_type in requirements: - if r_method == "scale": - t_width, t_height = thumbnailer.aspect(r_width, r_height) - scales.add(( - min(m_width, t_width), min(m_height, t_height), r_type, - )) - elif r_method == "crop": - crops.add((r_width, r_height, r_type)) - - for t_width, t_height, t_type in scales: - t_method = "scale" - t_path = self.filepaths.remote_media_thumbnail( - server_name, file_id, t_width, t_height, t_type, t_method + output_path = yield self.media_storage.store_file( + t_byte_source, file_info, ) - self._makedirs(t_path) - t_len = thumbnailer.scale(t_path, t_width, t_height, t_type) - remote_thumbnails.append([ + finally: + t_byte_source.close() + + t_len = os.path.getsize(output_path) + + # Write to database + if server_name: + yield self.store.store_remote_media_thumbnail( server_name, media_id, file_id, t_width, t_height, t_type, t_method, t_len - ]) - - for t_width, t_height, t_type in crops: - if (t_width, t_height, t_type) in scales: - # If the aspect ratio of the cropped thumbnail matches a purely - # scaled one then there is no point in calculating a separate - # thumbnail. - continue - t_method = "crop" - t_path = self.filepaths.remote_media_thumbnail( - server_name, file_id, t_width, t_height, t_type, t_method ) - self._makedirs(t_path) - t_len = thumbnailer.crop(t_path, t_width, t_height, t_type) - remote_thumbnails.append([ - server_name, media_id, file_id, - t_width, t_height, t_type, t_method, t_len - ]) - - yield preserve_context_over_fn(threads.deferToThread, generate_thumbnails) - - for r in remote_thumbnails: - yield self.store.store_remote_media_thumbnail(*r) + else: + yield self.store.store_local_thumbnail( + media_id, t_width, t_height, t_type, t_method, t_len + ) defer.returnValue({ "width": m_width, @@ -497,6 +675,8 @@ class MediaRepository(object): logger.info("Deleting: %r", key) + # TODO: Should we delete from the backup store + with (yield self.remote_media_linearizer.queue(key)): full_path = self.filepaths.remote_media_filepath(origin, file_id) try: @@ -571,7 +751,11 @@ class MediaRepositoryResource(Resource): self.putChild("upload", UploadResource(hs, media_repo)) self.putChild("download", DownloadResource(hs, media_repo)) - self.putChild("thumbnail", ThumbnailResource(hs, media_repo)) + self.putChild("thumbnail", ThumbnailResource( + hs, media_repo, media_repo.media_storage, + )) self.putChild("identicon", IdenticonResource()) if hs.config.url_preview_enabled: - self.putChild("preview_url", PreviewUrlResource(hs, media_repo)) + self.putChild("preview_url", PreviewUrlResource( + hs, media_repo, media_repo.media_storage, + )) diff --git a/synapse/rest/media/v1/media_storage.py b/synapse/rest/media/v1/media_storage.py new file mode 100644 index 000000000..b25993fcb --- /dev/null +++ b/synapse/rest/media/v1/media_storage.py @@ -0,0 +1,265 @@ +# -*- coding: utf-8 -*- +# Copyright 2018 New Vecotr Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import contextlib +import logging +import os +import shutil +import sys + +import six + +from twisted.internet import defer, threads +from twisted.protocols.basic import FileSender + +from synapse.util.file_consumer import BackgroundFileConsumer +from synapse.util.logcontext import make_deferred_yieldable + +from ._base import Responder + +logger = logging.getLogger(__name__) + + +class MediaStorage(object): + """Responsible for storing/fetching files from local sources. + + Args: + hs (synapse.server.Homeserver) + local_media_directory (str): Base path where we store media on disk + filepaths (MediaFilePaths) + storage_providers ([StorageProvider]): List of StorageProvider that are + used to fetch and store files. + """ + + def __init__(self, hs, local_media_directory, filepaths, storage_providers): + self.hs = hs + self.local_media_directory = local_media_directory + self.filepaths = filepaths + self.storage_providers = storage_providers + + @defer.inlineCallbacks + def store_file(self, source, file_info): + """Write `source` to the on disk media store, and also any other + configured storage providers + + Args: + source: A file like object that should be written + file_info (FileInfo): Info about the file to store + + Returns: + Deferred[str]: the file path written to in the primary media store + """ + + with self.store_into_file(file_info) as (f, fname, finish_cb): + # Write to the main repository + yield make_deferred_yieldable(threads.deferToThread( + _write_file_synchronously, source, f, + )) + yield finish_cb() + + defer.returnValue(fname) + + @contextlib.contextmanager + def store_into_file(self, file_info): + """Context manager used to get a file like object to write into, as + described by file_info. + + Actually yields a 3-tuple (file, fname, finish_cb), where file is a file + like object that can be written to, fname is the absolute path of file + on disk, and finish_cb is a function that returns a Deferred. + + fname can be used to read the contents from after upload, e.g. to + generate thumbnails. + + finish_cb must be called and waited on after the file has been + successfully been written to. Should not be called if there was an + error. + + Args: + file_info (FileInfo): Info about the file to store + + Example: + + with media_storage.store_into_file(info) as (f, fname, finish_cb): + # .. write into f ... + yield finish_cb() + """ + + path = self._file_info_to_path(file_info) + fname = os.path.join(self.local_media_directory, path) + + dirname = os.path.dirname(fname) + if not os.path.exists(dirname): + os.makedirs(dirname) + + finished_called = [False] + + @defer.inlineCallbacks + def finish(): + for provider in self.storage_providers: + yield provider.store_file(path, file_info) + + finished_called[0] = True + + try: + with open(fname, "wb") as f: + yield f, fname, finish + except Exception: + t, v, tb = sys.exc_info() + try: + os.remove(fname) + except Exception: + pass + six.reraise(t, v, tb) + + if not finished_called: + raise Exception("Finished callback not called") + + @defer.inlineCallbacks + def fetch_media(self, file_info): + """Attempts to fetch media described by file_info from the local cache + and configured storage providers. + + Args: + file_info (FileInfo) + + Returns: + Deferred[Responder|None]: Returns a Responder if the file was found, + otherwise None. + """ + + path = self._file_info_to_path(file_info) + local_path = os.path.join(self.local_media_directory, path) + if os.path.exists(local_path): + defer.returnValue(FileResponder(open(local_path, "rb"))) + + for provider in self.storage_providers: + res = yield provider.fetch(path, file_info) + if res: + defer.returnValue(res) + + defer.returnValue(None) + + @defer.inlineCallbacks + def ensure_media_is_in_local_cache(self, file_info): + """Ensures that the given file is in the local cache. Attempts to + download it from storage providers if it isn't. + + Args: + file_info (FileInfo) + + Returns: + Deferred[str]: Full path to local file + """ + path = self._file_info_to_path(file_info) + local_path = os.path.join(self.local_media_directory, path) + if os.path.exists(local_path): + defer.returnValue(local_path) + + dirname = os.path.dirname(local_path) + if not os.path.exists(dirname): + os.makedirs(dirname) + + for provider in self.storage_providers: + res = yield provider.fetch(path, file_info) + if res: + with res: + consumer = BackgroundFileConsumer( + open(local_path, "w"), self.hs.get_reactor()) + yield res.write_to_consumer(consumer) + yield consumer.wait() + defer.returnValue(local_path) + + raise Exception("file could not be found") + + def _file_info_to_path(self, file_info): + """Converts file_info into a relative path. + + The path is suitable for storing files under a directory, e.g. used to + store files on local FS under the base media repository directory. + + Args: + file_info (FileInfo) + + Returns: + str + """ + if file_info.url_cache: + if file_info.thumbnail: + return self.filepaths.url_cache_thumbnail_rel( + media_id=file_info.file_id, + width=file_info.thumbnail_width, + height=file_info.thumbnail_height, + content_type=file_info.thumbnail_type, + method=file_info.thumbnail_method, + ) + return self.filepaths.url_cache_filepath_rel(file_info.file_id) + + if file_info.server_name: + if file_info.thumbnail: + return self.filepaths.remote_media_thumbnail_rel( + server_name=file_info.server_name, + file_id=file_info.file_id, + width=file_info.thumbnail_width, + height=file_info.thumbnail_height, + content_type=file_info.thumbnail_type, + method=file_info.thumbnail_method + ) + return self.filepaths.remote_media_filepath_rel( + file_info.server_name, file_info.file_id, + ) + + if file_info.thumbnail: + return self.filepaths.local_media_thumbnail_rel( + media_id=file_info.file_id, + width=file_info.thumbnail_width, + height=file_info.thumbnail_height, + content_type=file_info.thumbnail_type, + method=file_info.thumbnail_method + ) + return self.filepaths.local_media_filepath_rel( + file_info.file_id, + ) + + +def _write_file_synchronously(source, dest): + """Write `source` to the file like `dest` synchronously. Should be called + from a thread. + + Args: + source: A file like object that's to be written + dest: A file like object to be written to + """ + source.seek(0) # Ensure we read from the start of the file + shutil.copyfileobj(source, dest) + + +class FileResponder(Responder): + """Wraps an open file that can be sent to a request. + + Args: + open_file (file): A file like object to be streamed ot the client, + is closed when finished streaming. + """ + def __init__(self, open_file): + self.open_file = open_file + + def write_to_consumer(self, consumer): + return make_deferred_yieldable( + FileSender().beginFileTransfer(self.open_file, consumer) + ) + + def __exit__(self, exc_type, exc_val, exc_tb): + self.open_file.close() diff --git a/synapse/rest/media/v1/preview_url_resource.py b/synapse/rest/media/v1/preview_url_resource.py index b81a336c5..b70b15c4c 100644 --- a/synapse/rest/media/v1/preview_url_resource.py +++ b/synapse/rest/media/v1/preview_url_resource.py @@ -12,80 +12,98 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. - -from twisted.web.server import NOT_DONE_YET -from twisted.internet import defer -from twisted.web.resource import Resource - -from synapse.api.errors import ( - SynapseError, Codes, -) -from synapse.util.stringutils import random_string -from synapse.util.caches.expiringcache import ExpiringCache -from synapse.http.client import SpiderHttpClient -from synapse.http.server import ( - request_handler, respond_with_json_bytes -) -from synapse.util.async import ObservableDeferred -from synapse.util.stringutils import is_ascii - +import cgi +import datetime +import errno +import fnmatch +import itertools +import logging import os import re -import fnmatch -import cgi -import ujson as json -import urlparse -import itertools +import shutil +import sys +import traceback + +from six import string_types +from six.moves import urllib_parse as urlparse + +from canonicaljson import json + +from twisted.internet import defer +from twisted.web.resource import Resource +from twisted.web.server import NOT_DONE_YET + +from synapse.api.errors import Codes, SynapseError +from synapse.http.client import SpiderHttpClient +from synapse.http.server import ( + respond_with_json, + respond_with_json_bytes, + wrap_json_request_handler, +) +from synapse.http.servlet import parse_integer, parse_string +from synapse.util.async import ObservableDeferred +from synapse.util.caches.expiringcache import ExpiringCache +from synapse.util.logcontext import make_deferred_yieldable, run_in_background +from synapse.util.stringutils import is_ascii, random_string + +from ._base import FileInfo -import logging logger = logging.getLogger(__name__) class PreviewUrlResource(Resource): isLeaf = True - def __init__(self, hs, media_repo): + def __init__(self, hs, media_repo, media_storage): Resource.__init__(self) self.auth = hs.get_auth() self.clock = hs.get_clock() - self.version_string = hs.version_string self.filepaths = media_repo.filepaths self.max_spider_size = hs.config.max_spider_size self.server_name = hs.hostname self.store = hs.get_datastore() self.client = SpiderHttpClient(hs) self.media_repo = media_repo + self.primary_base_path = media_repo.primary_base_path + self.media_storage = media_storage self.url_preview_url_blacklist = hs.config.url_preview_url_blacklist - # simple memory cache mapping urls to OG metadata - self.cache = ExpiringCache( + # memory cache mapping urls to an ObservableDeferred returning + # JSON-encoded OG metadata + self._cache = ExpiringCache( cache_name="url_previews", clock=self.clock, # don't spider URLs more often than once an hour expiry_ms=60 * 60 * 1000, ) - self.cache.start() + self._cache.start() - self.downloads = {} + self._cleaner_loop = self.clock.looping_call( + self._expire_url_cache_data, 10 * 1000 + ) + + def render_OPTIONS(self, request): + return respond_with_json(request, 200, {}, send_cors=True) def render_GET(self, request): self._async_render_GET(request) return NOT_DONE_YET - @request_handler() + @wrap_json_request_handler @defer.inlineCallbacks def _async_render_GET(self, request): # XXX: if get_user_by_req fails, what should we do in an async render? requester = yield self.auth.get_user_by_req(request) - url = request.args.get("url")[0] + url = parse_string(request, "url") if "ts" in request.args: - ts = int(request.args.get("ts")[0]) + ts = parse_integer(request, "ts") else: ts = self.clock.time_msec() + # XXX: we could move this into _do_preview if we wanted. url_tuple = urlparse.urlsplit(url) for entry in self.url_preview_url_blacklist: match = True @@ -118,53 +136,62 @@ class PreviewUrlResource(Resource): Codes.UNKNOWN ) - # first check the memory cache - good to handle all the clients on this - # HS thundering away to preview the same URL at the same time. - og = self.cache.get(url) - if og: - respond_with_json_bytes(request, 200, json.dumps(og), send_cors=True) - return + # the in-memory cache: + # * ensures that only one request is active at a time + # * takes load off the DB for the thundering herds + # * also caches any failures (unlike the DB) so we don't keep + # requesting the same endpoint - # then check the URL cache in the DB (which will also provide us with + observable = self._cache.get(url) + + if not observable: + download = run_in_background( + self._do_preview, + url, requester.user, ts, + ) + observable = ObservableDeferred( + download, + consumeErrors=True + ) + self._cache[url] = observable + else: + logger.info("Returning cached response") + + og = yield make_deferred_yieldable(observable.observe()) + respond_with_json_bytes(request, 200, og, send_cors=True) + + @defer.inlineCallbacks + def _do_preview(self, url, user, ts): + """Check the db, and download the URL and build a preview + + Args: + url (str): + user (str): + ts (int): + + Returns: + Deferred[str]: json-encoded og data + """ + # check the URL cache in the DB (which will also provide us with # historical previews, if we have any) cache_result = yield self.store.get_url_cache(url, ts) if ( cache_result and - cache_result["download_ts"] + cache_result["expires"] > ts and + cache_result["expires_ts"] > ts and cache_result["response_code"] / 100 == 2 ): - respond_with_json_bytes( - request, 200, cache_result["og"].encode('utf-8'), - send_cors=True - ) + defer.returnValue(cache_result["og"]) return - # Ensure only one download for a given URL is active at a time - download = self.downloads.get(url) - if download is None: - download = self._download_url(url, requester.user) - download = ObservableDeferred( - download, - consumeErrors=True - ) - self.downloads[url] = download - - @download.addBoth - def callback(media_info): - del self.downloads[url] - return media_info - media_info = yield download.observe() - - # FIXME: we should probably update our cache now anyway, so that - # even if the OG calculation raises, we don't keep hammering on the - # remote server. For now, leave it uncached to aid debugging OG - # calculation problems + media_info = yield self._download_url(url, user) logger.debug("got media_info of '%s'" % media_info) if _is_media(media_info['media_type']): - dims = yield self.media_repo._generate_local_thumbnails( - media_info['filesystem_id'], media_info, url_cache=True, + file_id = media_info['filesystem_id'] + dims = yield self.media_repo._generate_thumbnails( + None, file_id, file_id, media_info["media_type"], + url_cache=True, ) og = { @@ -204,13 +231,15 @@ class PreviewUrlResource(Resource): # just rely on the caching on the master request to speed things up. if 'og:image' in og and og['og:image']: image_info = yield self._download_url( - _rebase_url(og['og:image'], media_info['uri']), requester.user + _rebase_url(og['og:image'], media_info['uri']), user ) if _is_media(image_info['media_type']): # TODO: make sure we don't choke on white-on-transparent images - dims = yield self.media_repo._generate_local_thumbnails( - image_info['filesystem_id'], image_info, url_cache=True, + file_id = image_info['filesystem_id'] + dims = yield self.media_repo._generate_thumbnails( + None, file_id, file_id, image_info["media_type"], + url_cache=True, ) if dims: og["og:image:width"] = dims['width'] @@ -231,21 +260,20 @@ class PreviewUrlResource(Resource): logger.debug("Calculated OG for %s as %s" % (url, og)) - # store OG in ephemeral in-memory cache - self.cache[url] = og + jsonog = json.dumps(og) # store OG in history-aware DB cache yield self.store.store_url_cache( url, media_info["response_code"], media_info["etag"], - media_info["expires"], - json.dumps(og), + media_info["expires"] + media_info["created_ts"], + jsonog, media_info["filesystem_id"], media_info["created_ts"], ) - respond_with_json_bytes(request, 200, json.dumps(og), send_cors=True) + defer.returnValue(jsonog) @defer.inlineCallbacks def _download_url(self, url, user): @@ -253,21 +281,36 @@ class PreviewUrlResource(Resource): # we're most likely being explicitly triggered by a human rather than a # bot, so are we really a robot? - # XXX: horrible duplication with base_resource's _download_remote_file() - file_id = random_string(24) + file_id = datetime.date.today().isoformat() + '_' + random_string(16) - fname = self.filepaths.url_cache_filepath(file_id) - self.media_repo._makedirs(fname) + file_info = FileInfo( + server_name=None, + file_id=file_id, + url_cache=True, + ) - try: - with open(fname, "wb") as f: + with self.media_storage.store_into_file(file_info) as (f, fname, finish): + try: logger.debug("Trying to get url '%s'" % url) length, headers, uri, code = yield self.client.get_file( url, output_stream=f, max_size=self.max_spider_size, ) + except Exception as e: # FIXME: pass through 404s and other error messages nicely + logger.warn("Error downloading %s: %r", url, e) + raise SynapseError( + 500, "Failed to download content: %s" % ( + traceback.format_exception_only(sys.exc_type, e), + ), + Codes.UNKNOWN, + ) + yield finish() - media_type = headers["Content-Type"][0] + try: + if "Content-Type" in headers: + media_type = headers["Content-Type"][0] + else: + media_type = "application/octet-stream" time_now_ms = self.clock.time_msec() content_disposition = headers.get("Content-Disposition", None) @@ -307,11 +350,11 @@ class PreviewUrlResource(Resource): ) except Exception as e: - os.remove(fname) - raise SynapseError( - 500, ("Failed to download content: %s" % e), - Codes.UNKNOWN - ) + logger.error("Error handling downloaded %s: %r", url, e) + # TODO: we really ought to delete the downloaded file in this + # case, since we won't have recorded it in the db, and will + # therefore not expire it. + raise defer.returnValue({ "media_type": media_type, @@ -328,6 +371,95 @@ class PreviewUrlResource(Resource): "etag": headers["ETag"][0] if "ETag" in headers else None, }) + @defer.inlineCallbacks + def _expire_url_cache_data(self): + """Clean up expired url cache content, media and thumbnails. + """ + # TODO: Delete from backup media store + + now = self.clock.time_msec() + + logger.info("Running url preview cache expiry") + + if not (yield self.store.has_completed_background_updates()): + logger.info("Still running DB updates; skipping expiry") + return + + # First we delete expired url cache entries + media_ids = yield self.store.get_expired_url_cache(now) + + removed_media = [] + for media_id in media_ids: + fname = self.filepaths.url_cache_filepath(media_id) + try: + os.remove(fname) + except OSError as e: + # If the path doesn't exist, meh + if e.errno != errno.ENOENT: + logger.warn("Failed to remove media: %r: %s", media_id, e) + continue + + removed_media.append(media_id) + + try: + dirs = self.filepaths.url_cache_filepath_dirs_to_delete(media_id) + for dir in dirs: + os.rmdir(dir) + except Exception: + pass + + yield self.store.delete_url_cache(removed_media) + + if removed_media: + logger.info("Deleted %d entries from url cache", len(removed_media)) + + # Now we delete old images associated with the url cache. + # These may be cached for a bit on the client (i.e., they + # may have a room open with a preview url thing open). + # So we wait a couple of days before deleting, just in case. + expire_before = now - 2 * 24 * 60 * 60 * 1000 + media_ids = yield self.store.get_url_cache_media_before(expire_before) + + removed_media = [] + for media_id in media_ids: + fname = self.filepaths.url_cache_filepath(media_id) + try: + os.remove(fname) + except OSError as e: + # If the path doesn't exist, meh + if e.errno != errno.ENOENT: + logger.warn("Failed to remove media: %r: %s", media_id, e) + continue + + try: + dirs = self.filepaths.url_cache_filepath_dirs_to_delete(media_id) + for dir in dirs: + os.rmdir(dir) + except Exception: + pass + + thumbnail_dir = self.filepaths.url_cache_thumbnail_directory(media_id) + try: + shutil.rmtree(thumbnail_dir) + except OSError as e: + # If the path doesn't exist, meh + if e.errno != errno.ENOENT: + logger.warn("Failed to remove media: %r: %s", media_id, e) + continue + + removed_media.append(media_id) + + try: + dirs = self.filepaths.url_cache_thumbnail_dirs_to_delete(media_id) + for dir in dirs: + os.rmdir(dir) + except Exception: + pass + + yield self.store.delete_url_cache_media(removed_media) + + logger.info("Deleted %d media from url cache", len(removed_media)) + def decode_and_calc_og(body, media_uri, request_encoding=None): from lxml import etree @@ -425,7 +557,14 @@ def _calc_og(tree, media_uri): from lxml import etree TAGS_TO_REMOVE = ( - "header", "nav", "aside", "footer", "script", "style", etree.Comment + "header", + "nav", + "aside", + "footer", + "script", + "noscript", + "style", + etree.Comment ) # Split all the text nodes into paragraphs (by splitting on new @@ -452,8 +591,8 @@ def _iterate_over_text(tree, *tags_to_ignore): # to be returned. elements = iter([tree]) while True: - el = elements.next() - if isinstance(el, basestring): + el = next(elements) + if isinstance(el, string_types): yield el elif el is not None and el.tag not in tags_to_ignore: # el.text is the text before the first child, so we can immediately diff --git a/synapse/rest/media/v1/storage_provider.py b/synapse/rest/media/v1/storage_provider.py new file mode 100644 index 000000000..7b9f8b4d7 --- /dev/null +++ b/synapse/rest/media/v1/storage_provider.py @@ -0,0 +1,144 @@ +# -*- coding: utf-8 -*- +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging +import os +import shutil + +from twisted.internet import defer, threads + +from synapse.config._base import Config +from synapse.util.logcontext import run_in_background + +from .media_storage import FileResponder + +logger = logging.getLogger(__name__) + + +class StorageProvider(object): + """A storage provider is a service that can store uploaded media and + retrieve them. + """ + def store_file(self, path, file_info): + """Store the file described by file_info. The actual contents can be + retrieved by reading the file in file_info.upload_path. + + Args: + path (str): Relative path of file in local cache + file_info (FileInfo) + + Returns: + Deferred + """ + pass + + def fetch(self, path, file_info): + """Attempt to fetch the file described by file_info and stream it + into writer. + + Args: + path (str): Relative path of file in local cache + file_info (FileInfo) + + Returns: + Deferred(Responder): Returns a Responder if the provider has the file, + otherwise returns None. + """ + pass + + +class StorageProviderWrapper(StorageProvider): + """Wraps a storage provider and provides various config options + + Args: + backend (StorageProvider) + store_local (bool): Whether to store new local files or not. + store_synchronous (bool): Whether to wait for file to be successfully + uploaded, or todo the upload in the backgroud. + store_remote (bool): Whether remote media should be uploaded + """ + def __init__(self, backend, store_local, store_synchronous, store_remote): + self.backend = backend + self.store_local = store_local + self.store_synchronous = store_synchronous + self.store_remote = store_remote + + def store_file(self, path, file_info): + if not file_info.server_name and not self.store_local: + return defer.succeed(None) + + if file_info.server_name and not self.store_remote: + return defer.succeed(None) + + if self.store_synchronous: + return self.backend.store_file(path, file_info) + else: + # TODO: Handle errors. + def store(): + try: + return self.backend.store_file(path, file_info) + except Exception: + logger.exception("Error storing file") + run_in_background(store) + return defer.succeed(None) + + def fetch(self, path, file_info): + return self.backend.fetch(path, file_info) + + +class FileStorageProviderBackend(StorageProvider): + """A storage provider that stores files in a directory on a filesystem. + + Args: + hs (HomeServer) + config: The config returned by `parse_config`. + """ + + def __init__(self, hs, config): + self.cache_directory = hs.config.media_store_path + self.base_directory = config + + def store_file(self, path, file_info): + """See StorageProvider.store_file""" + + primary_fname = os.path.join(self.cache_directory, path) + backup_fname = os.path.join(self.base_directory, path) + + dirname = os.path.dirname(backup_fname) + if not os.path.exists(dirname): + os.makedirs(dirname) + + return threads.deferToThread( + shutil.copyfile, primary_fname, backup_fname, + ) + + def fetch(self, path, file_info): + """See StorageProvider.fetch""" + + backup_fname = os.path.join(self.base_directory, path) + if os.path.isfile(backup_fname): + return FileResponder(open(backup_fname, "rb")) + + @staticmethod + def parse_config(config): + """Called on startup to parse config supplied. This should parse + the config and raise if there is a problem. + + The returned value is passed into the constructor. + + In this case we only care about a single param, the directory, so let's + just pull that out. + """ + return Config.ensure_directory(config["directory"]) diff --git a/synapse/rest/media/v1/thumbnail_resource.py b/synapse/rest/media/v1/thumbnail_resource.py index 68d56b2b1..5305e9175 100644 --- a/synapse/rest/media/v1/thumbnail_resource.py +++ b/synapse/rest/media/v1/thumbnail_resource.py @@ -14,38 +14,44 @@ # limitations under the License. -from ._base import parse_media_id, respond_404, respond_with_file -from twisted.web.resource import Resource -from synapse.http.servlet import parse_string, parse_integer -from synapse.http.server import request_handler, set_cors_headers - -from twisted.web.server import NOT_DONE_YET -from twisted.internet import defer - import logging +from twisted.internet import defer +from twisted.web.resource import Resource +from twisted.web.server import NOT_DONE_YET + +from synapse.http.server import set_cors_headers, wrap_json_request_handler +from synapse.http.servlet import parse_integer, parse_string + +from ._base import ( + FileInfo, + parse_media_id, + respond_404, + respond_with_file, + respond_with_responder, +) + logger = logging.getLogger(__name__) class ThumbnailResource(Resource): isLeaf = True - def __init__(self, hs, media_repo): + def __init__(self, hs, media_repo, media_storage): Resource.__init__(self) self.store = hs.get_datastore() - self.filepaths = media_repo.filepaths self.media_repo = media_repo + self.media_storage = media_storage self.dynamic_thumbnails = hs.config.dynamic_thumbnails self.server_name = hs.hostname - self.version_string = hs.version_string self.clock = hs.get_clock() def render_GET(self, request): self._async_render_GET(request) return NOT_DONE_YET - @request_handler() + @wrap_json_request_handler @defer.inlineCallbacks def _async_render_GET(self, request): set_cors_headers(request) @@ -64,6 +70,7 @@ class ThumbnailResource(Resource): yield self._respond_local_thumbnail( request, media_id, width, height, method, m_type ) + self.media_repo.mark_recently_accessed(None, media_id) else: if self.dynamic_thumbnails: yield self._select_or_generate_remote_thumbnail( @@ -75,20 +82,20 @@ class ThumbnailResource(Resource): request, server_name, media_id, width, height, method, m_type ) + self.media_repo.mark_recently_accessed(server_name, media_id) @defer.inlineCallbacks def _respond_local_thumbnail(self, request, media_id, width, height, method, m_type): media_info = yield self.store.get_local_media(media_id) - if not media_info or media_info["quarantined_by"]: + if not media_info: + respond_404(request) + return + if media_info["quarantined_by"]: + logger.info("Media is quarantined") respond_404(request) return - - # if media_info["media_type"] == "image/svg+xml": - # file_path = self.filepaths.local_media_filepath(media_id) - # yield respond_with_file(request, media_info["media_type"], file_path) - # return thumbnail_infos = yield self.store.get_local_media_thumbnails(media_id) @@ -96,42 +103,39 @@ class ThumbnailResource(Resource): thumbnail_info = self._select_thumbnail( width, height, method, m_type, thumbnail_infos ) - t_width = thumbnail_info["thumbnail_width"] - t_height = thumbnail_info["thumbnail_height"] - t_type = thumbnail_info["thumbnail_type"] - t_method = thumbnail_info["thumbnail_method"] - if media_info["url_cache"]: - # TODO: Check the file still exists, if it doesn't we can redownload - # it from the url `media_info["url_cache"]` - file_path = self.filepaths.url_cache_thumbnail( - media_id, t_width, t_height, t_type, t_method, - ) - else: - file_path = self.filepaths.local_media_thumbnail( - media_id, t_width, t_height, t_type, t_method, - ) - yield respond_with_file(request, t_type, file_path) - - else: - yield self._respond_default_thumbnail( - request, media_info, width, height, method, m_type, + file_info = FileInfo( + server_name=None, file_id=media_id, + url_cache=media_info["url_cache"], + thumbnail=True, + thumbnail_width=thumbnail_info["thumbnail_width"], + thumbnail_height=thumbnail_info["thumbnail_height"], + thumbnail_type=thumbnail_info["thumbnail_type"], + thumbnail_method=thumbnail_info["thumbnail_method"], ) + t_type = file_info.thumbnail_type + t_length = thumbnail_info["thumbnail_length"] + + responder = yield self.media_storage.fetch_media(file_info) + yield respond_with_responder(request, responder, t_type, t_length) + else: + logger.info("Couldn't find any generated thumbnails") + respond_404(request) + @defer.inlineCallbacks def _select_or_generate_local_thumbnail(self, request, media_id, desired_width, desired_height, desired_method, desired_type): media_info = yield self.store.get_local_media(media_id) - if not media_info or media_info["quarantined_by"]: + if not media_info: + respond_404(request) + return + if media_info["quarantined_by"]: + logger.info("Media is quarantined") respond_404(request) return - - # if media_info["media_type"] == "image/svg+xml": - # file_path = self.filepaths.local_media_filepath(media_id) - # yield respond_with_file(request, media_info["media_type"], file_path) - # return thumbnail_infos = yield self.store.get_local_media_thumbnails(media_id) for info in thumbnail_infos: @@ -141,46 +145,43 @@ class ThumbnailResource(Resource): t_type = info["thumbnail_type"] == desired_type if t_w and t_h and t_method and t_type: - if media_info["url_cache"]: - # TODO: Check the file still exists, if it doesn't we can redownload - # it from the url `media_info["url_cache"]` - file_path = self.filepaths.url_cache_thumbnail( - media_id, desired_width, desired_height, desired_type, - desired_method, - ) - else: - file_path = self.filepaths.local_media_thumbnail( - media_id, desired_width, desired_height, desired_type, - desired_method, - ) - yield respond_with_file(request, desired_type, file_path) - return + file_info = FileInfo( + server_name=None, file_id=media_id, + url_cache=media_info["url_cache"], + thumbnail=True, + thumbnail_width=info["thumbnail_width"], + thumbnail_height=info["thumbnail_height"], + thumbnail_type=info["thumbnail_type"], + thumbnail_method=info["thumbnail_method"], + ) - logger.debug("We don't have a local thumbnail of that size. Generating") + t_type = file_info.thumbnail_type + t_length = info["thumbnail_length"] + + responder = yield self.media_storage.fetch_media(file_info) + if responder: + yield respond_with_responder(request, responder, t_type, t_length) + return + + logger.debug("We don't have a thumbnail of that size. Generating") # Okay, so we generate one. file_path = yield self.media_repo.generate_local_exact_thumbnail( - media_id, desired_width, desired_height, desired_method, desired_type + media_id, desired_width, desired_height, desired_method, desired_type, + url_cache=media_info["url_cache"], ) if file_path: yield respond_with_file(request, desired_type, file_path) else: - yield self._respond_default_thumbnail( - request, media_info, desired_width, desired_height, - desired_method, desired_type, - ) + logger.warn("Failed to generate thumbnail") + respond_404(request) @defer.inlineCallbacks def _select_or_generate_remote_thumbnail(self, request, server_name, media_id, desired_width, desired_height, desired_method, desired_type): - media_info = yield self.media_repo.get_remote_media(server_name, media_id) - - # if media_info["media_type"] == "image/svg+xml": - # file_path = self.filepaths.remote_media_filepath(server_name, media_id) - # yield respond_with_file(request, media_info["media_type"], file_path) - # return + media_info = yield self.media_repo.get_remote_media_info(server_name, media_id) thumbnail_infos = yield self.store.get_remote_media_thumbnails( server_name, media_id, @@ -195,14 +196,24 @@ class ThumbnailResource(Resource): t_type = info["thumbnail_type"] == desired_type if t_w and t_h and t_method and t_type: - file_path = self.filepaths.remote_media_thumbnail( - server_name, file_id, desired_width, desired_height, - desired_type, desired_method, + file_info = FileInfo( + server_name=server_name, file_id=media_info["filesystem_id"], + thumbnail=True, + thumbnail_width=info["thumbnail_width"], + thumbnail_height=info["thumbnail_height"], + thumbnail_type=info["thumbnail_type"], + thumbnail_method=info["thumbnail_method"], ) - yield respond_with_file(request, desired_type, file_path) - return - logger.debug("We don't have a local thumbnail of that size. Generating") + t_type = file_info.thumbnail_type + t_length = info["thumbnail_length"] + + responder = yield self.media_storage.fetch_media(file_info) + if responder: + yield respond_with_responder(request, responder, t_type, t_length) + return + + logger.debug("We don't have a thumbnail of that size. Generating") # Okay, so we generate one. file_path = yield self.media_repo.generate_remote_exact_thumbnail( @@ -213,22 +224,16 @@ class ThumbnailResource(Resource): if file_path: yield respond_with_file(request, desired_type, file_path) else: - yield self._respond_default_thumbnail( - request, media_info, desired_width, desired_height, - desired_method, desired_type, - ) + logger.warn("Failed to generate thumbnail") + respond_404(request) @defer.inlineCallbacks def _respond_remote_thumbnail(self, request, server_name, media_id, width, height, method, m_type): # TODO: Don't download the whole remote file - # We should proxy the thumbnail from the remote server instead. - media_info = yield self.media_repo.get_remote_media(server_name, media_id) - - # if media_info["media_type"] == "image/svg+xml": - # file_path = self.filepaths.remote_media_filepath(server_name, media_id) - # yield respond_with_file(request, media_info["media_type"], file_path) - # return + # We should proxy the thumbnail from the remote server instead of + # downloading the remote file and generating our own thumbnails. + media_info = yield self.media_repo.get_remote_media_info(server_name, media_id) thumbnail_infos = yield self.store.get_remote_media_thumbnails( server_name, media_id, @@ -238,59 +243,23 @@ class ThumbnailResource(Resource): thumbnail_info = self._select_thumbnail( width, height, method, m_type, thumbnail_infos ) - t_width = thumbnail_info["thumbnail_width"] - t_height = thumbnail_info["thumbnail_height"] - t_type = thumbnail_info["thumbnail_type"] - t_method = thumbnail_info["thumbnail_method"] - file_id = thumbnail_info["filesystem_id"] + file_info = FileInfo( + server_name=server_name, file_id=media_info["filesystem_id"], + thumbnail=True, + thumbnail_width=thumbnail_info["thumbnail_width"], + thumbnail_height=thumbnail_info["thumbnail_height"], + thumbnail_type=thumbnail_info["thumbnail_type"], + thumbnail_method=thumbnail_info["thumbnail_method"], + ) + + t_type = file_info.thumbnail_type t_length = thumbnail_info["thumbnail_length"] - file_path = self.filepaths.remote_media_thumbnail( - server_name, file_id, t_width, t_height, t_type, t_method, - ) - yield respond_with_file(request, t_type, file_path, t_length) + responder = yield self.media_storage.fetch_media(file_info) + yield respond_with_responder(request, responder, t_type, t_length) else: - yield self._respond_default_thumbnail( - request, media_info, width, height, method, m_type, - ) - - @defer.inlineCallbacks - def _respond_default_thumbnail(self, request, media_info, width, height, - method, m_type): - # XXX: how is this meant to work? store.get_default_thumbnails - # appears to always return [] so won't this always 404? - media_type = media_info["media_type"] - top_level_type = media_type.split("/")[0] - sub_type = media_type.split("/")[-1].split(";")[0] - thumbnail_infos = yield self.store.get_default_thumbnails( - top_level_type, sub_type, - ) - if not thumbnail_infos: - thumbnail_infos = yield self.store.get_default_thumbnails( - top_level_type, "_default", - ) - if not thumbnail_infos: - thumbnail_infos = yield self.store.get_default_thumbnails( - "_default", "_default", - ) - if not thumbnail_infos: + logger.info("Failed to find any generated thumbnails") respond_404(request) - return - - thumbnail_info = self._select_thumbnail( - width, height, "crop", m_type, thumbnail_infos - ) - - t_width = thumbnail_info["thumbnail_width"] - t_height = thumbnail_info["thumbnail_height"] - t_type = thumbnail_info["thumbnail_type"] - t_method = thumbnail_info["thumbnail_method"] - t_length = thumbnail_info["thumbnail_length"] - - file_path = self.filepaths.default_thumbnail( - top_level_type, sub_type, t_width, t_height, t_type, t_method, - ) - yield respond_with_file(request, t_type, file_path, t_length) def _select_thumbnail(self, desired_width, desired_height, desired_method, desired_type, thumbnail_infos): diff --git a/synapse/rest/media/v1/thumbnailer.py b/synapse/rest/media/v1/thumbnailer.py index 3868d4f65..a4b26c258 100644 --- a/synapse/rest/media/v1/thumbnailer.py +++ b/synapse/rest/media/v1/thumbnailer.py @@ -13,10 +13,10 @@ # See the License for the specific language governing permissions and # limitations under the License. -import PIL.Image as Image +import logging from io import BytesIO -import logging +import PIL.Image as Image logger = logging.getLogger(__name__) @@ -50,12 +50,16 @@ class Thumbnailer(object): else: return ((max_height * self.width) // self.height, max_height) - def scale(self, output_path, width, height, output_type): - """Rescales the image to the given dimensions""" - scaled = self.image.resize((width, height), Image.ANTIALIAS) - return self.save_image(scaled, output_type, output_path) + def scale(self, width, height, output_type): + """Rescales the image to the given dimensions. - def crop(self, output_path, width, height, output_type): + Returns: + BytesIO: the bytes of the encoded image ready to be written to disk + """ + scaled = self.image.resize((width, height), Image.ANTIALIAS) + return self._encode_image(scaled, output_type) + + def crop(self, width, height, output_type): """Rescales and crops the image to the given dimensions preserving aspect:: (w_in / h_in) = (w_scaled / h_scaled) @@ -65,6 +69,9 @@ class Thumbnailer(object): Args: max_width: The largest possible width. max_height: The larget possible height. + + Returns: + BytesIO: the bytes of the encoded image ready to be written to disk """ if width * self.height > height * self.width: scaled_height = (width * self.height) // self.width @@ -82,13 +89,9 @@ class Thumbnailer(object): crop_left = (scaled_width - width) // 2 crop_right = width + crop_left cropped = scaled_image.crop((crop_left, 0, crop_right, height)) - return self.save_image(cropped, output_type, output_path) + return self._encode_image(cropped, output_type) - def save_image(self, output_image, output_type, output_path): + def _encode_image(self, output_image, output_type): output_bytes_io = BytesIO() output_image.save(output_bytes_io, self.FORMATS[output_type], quality=80) - output_bytes = output_bytes_io.getvalue() - with open(output_path, "wb") as output_file: - output_file.write(output_bytes) - logger.info("Stored thumbnail in file %r", output_path) - return len(output_bytes) + return output_bytes_io diff --git a/synapse/rest/media/v1/upload_resource.py b/synapse/rest/media/v1/upload_resource.py index 4ab33f73b..9b22d204a 100644 --- a/synapse/rest/media/v1/upload_resource.py +++ b/synapse/rest/media/v1/upload_resource.py @@ -13,16 +13,15 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.http.server import respond_with_json, request_handler +import logging + +from twisted.internet import defer +from twisted.web.resource import Resource +from twisted.web.server import NOT_DONE_YET from synapse.api.errors import SynapseError - -from twisted.web.server import NOT_DONE_YET -from twisted.internet import defer - -from twisted.web.resource import Resource - -import logging +from synapse.http.server import respond_with_json, wrap_json_request_handler +from synapse.http.servlet import parse_string logger = logging.getLogger(__name__) @@ -40,7 +39,6 @@ class UploadResource(Resource): self.server_name = hs.hostname self.auth = hs.get_auth() self.max_upload_size = hs.config.max_upload_size - self.version_string = hs.version_string self.clock = hs.get_clock() def render_POST(self, request): @@ -51,7 +49,7 @@ class UploadResource(Resource): respond_with_json(request, 200, {}, send_cors=True) return NOT_DONE_YET - @request_handler() + @wrap_json_request_handler @defer.inlineCallbacks def _async_render_POST(self, request): requester = yield self.auth.get_user_by_req(request) @@ -68,10 +66,10 @@ class UploadResource(Resource): code=413, ) - upload_name = request.args.get("filename", None) + upload_name = parse_string(request, "filename") if upload_name: try: - upload_name = upload_name[0].decode('UTF-8') + upload_name = upload_name.decode('UTF-8') except UnicodeDecodeError: raise SynapseError( msg="Invalid UTF-8 filename parameter: %r" % (upload_name), @@ -81,19 +79,19 @@ class UploadResource(Resource): headers = request.requestHeaders if headers.hasHeader("Content-Type"): - media_type = headers.getRawHeaders("Content-Type")[0] + media_type = headers.getRawHeaders(b"Content-Type")[0] else: raise SynapseError( msg="Upload request missing 'Content-Type'", code=400, ) - # if headers.hasHeader("Content-Disposition"): - # disposition = headers.getRawHeaders("Content-Disposition")[0] + # if headers.hasHeader(b"Content-Disposition"): + # disposition = headers.getRawHeaders(b"Content-Disposition")[0] # TODO(markjh): parse content-dispostion content_uri = yield self.media_repo.create_content( - media_type, upload_name, request.content.read(), + media_type, upload_name, request.content, content_length, requester.user ) diff --git a/synapse/secrets.py b/synapse/secrets.py new file mode 100644 index 000000000..f397daaa5 --- /dev/null +++ b/synapse/secrets.py @@ -0,0 +1,42 @@ +# -*- coding: utf-8 -*- +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +""" +Injectable secrets module for Synapse. + +See https://docs.python.org/3/library/secrets.html#module-secrets for the API +used in Python 3.6, and the API emulated in Python 2.7. +""" + +import six + +if six.PY3: + import secrets + + def Secrets(): + return secrets + + +else: + + import os + import binascii + + class Secrets(object): + def token_bytes(self, nbytes=32): + return os.urandom(nbytes) + + def token_hex(self, nbytes=32): + return binascii.hexlify(self.token_bytes(nbytes)) diff --git a/synapse/server.py b/synapse/server.py index a38e5179e..140be9ebe 100644 --- a/synapse/server.py +++ b/synapse/server.py @@ -31,32 +31,55 @@ from synapse.appservice.api import ApplicationServiceApi from synapse.appservice.scheduler import ApplicationServiceScheduler from synapse.crypto.keyring import Keyring from synapse.events.builder import EventBuilderFactory -from synapse.federation import initialize_http_replication +from synapse.events.spamcheck import SpamChecker +from synapse.federation.federation_client import FederationClient +from synapse.federation.federation_server import ( + FederationHandlerRegistry, + FederationServer, +) from synapse.federation.send_queue import FederationRemoteSendQueue -from synapse.federation.transport.client import TransportLayerClient from synapse.federation.transaction_queue import TransactionQueue +from synapse.federation.transport.client import TransportLayerClient +from synapse.groups.attestations import GroupAttestationSigning, GroupAttestionRenewer +from synapse.groups.groups_server import GroupsServerHandler from synapse.handlers import Handlers from synapse.handlers.appservice import ApplicationServicesHandler -from synapse.handlers.auth import AuthHandler, MacaroonGeneartor -from synapse.handlers.devicemessage import DeviceMessageHandler +from synapse.handlers.auth import AuthHandler, MacaroonGenerator +from synapse.handlers.deactivate_account import DeactivateAccountHandler from synapse.handlers.device import DeviceHandler +from synapse.handlers.devicemessage import DeviceMessageHandler from synapse.handlers.e2e_keys import E2eKeysHandler +from synapse.handlers.events import EventHandler, EventStreamHandler +from synapse.handlers.groups_local import GroupsLocalHandler +from synapse.handlers.initial_sync import InitialSyncHandler +from synapse.handlers.message import EventCreationHandler, MessageHandler +from synapse.handlers.pagination import PaginationHandler from synapse.handlers.presence import PresenceHandler +from synapse.handlers.profile import ProfileHandler +from synapse.handlers.read_marker import ReadMarkerHandler +from synapse.handlers.receipts import ReceiptsHandler +from synapse.handlers.room import RoomContextHandler, RoomCreationHandler from synapse.handlers.room_list import RoomListHandler +from synapse.handlers.room_member import RoomMemberMasterHandler +from synapse.handlers.room_member_worker import RoomMemberWorkerHandler +from synapse.handlers.set_password import SetPasswordHandler from synapse.handlers.sync import SyncHandler from synapse.handlers.typing import TypingHandler -from synapse.handlers.events import EventHandler, EventStreamHandler -from synapse.handlers.initial_sync import InitialSyncHandler -from synapse.handlers.receipts import ReceiptsHandler -from synapse.handlers.read_marker import ReadMarkerHandler -from synapse.handlers.user_directory import UserDirectoyHandler -from synapse.http.client import SimpleHttpClient, InsecureInterceptableContextFactory +from synapse.handlers.user_directory import UserDirectoryHandler +from synapse.http.client import InsecureInterceptableContextFactory, SimpleHttpClient from synapse.http.matrixfederationclient import MatrixFederationHttpClient from synapse.notifier import Notifier from synapse.push.action_generator import ActionGenerator from synapse.push.pusherpool import PusherPool -from synapse.rest.media.v1.media_repository import MediaRepository -from synapse.state import StateHandler +from synapse.rest.media.v1.media_repository import ( + MediaRepository, + MediaRepositoryResource, +) +from synapse.secrets import Secrets +from synapse.server_notices.server_notices_manager import ServerNoticesManager +from synapse.server_notices.server_notices_sender import ServerNoticesSender +from synapse.server_notices.worker_server_notices_sender import WorkerServerNoticesSender +from synapse.state import StateHandler, StateResolutionHandler from synapse.storage import DataStore from synapse.streams.events import EventSources from synapse.util import Clock @@ -82,21 +105,21 @@ class HomeServer(object): which must be implemented by the subclass. This code may call any of the required "get" methods on the instance to obtain the sub-dependencies that one requires. + + Attributes: + config (synapse.config.homeserver.HomeserverConfig): """ DEPENDENCIES = [ - 'config', - 'clock', 'http_client', 'db_pool', - 'persistence_service', - 'replication_layer', - 'datastore', + 'federation_client', + 'federation_server', 'handlers', - 'v1auth', 'auth', - 'rest_servlet_factory', + 'room_creation_handler', 'state_handler', + 'state_resolution_handler', 'presence_handler', 'sync_handler', 'typing_handler', @@ -111,19 +134,12 @@ class HomeServer(object): 'application_service_scheduler', 'application_service_handler', 'device_message_handler', + 'profile_handler', + 'event_creation_handler', + 'deactivate_account_handler', + 'set_password_handler', 'notifier', - 'distributor', - 'client_resource', - 'resource_for_federation', - 'resource_for_static_content', - 'resource_for_web_client', - 'resource_for_content_repo', - 'resource_for_server_key', - 'resource_for_server_key_v2', - 'resource_for_media_repository', - 'resource_for_metrics', 'event_sources', - 'ratelimiter', 'keyring', 'pusherpool', 'event_builder_factory', @@ -131,6 +147,7 @@ class HomeServer(object): 'http_client_context_factory', 'simple_http_client', 'media_repository', + 'media_repository_resource', 'federation_transport_client', 'federation_sender', 'receipts_handler', @@ -139,17 +156,34 @@ class HomeServer(object): 'read_marker_handler', 'action_generator', 'user_directory_handler', + 'groups_local_handler', + 'groups_server_handler', + 'groups_attestation_signing', + 'groups_attestation_renewer', + 'secrets', + 'spam_checker', + 'room_member_handler', + 'federation_registry', + 'server_notices_manager', + 'server_notices_sender', + 'message_handler', + 'pagination_handler', + 'room_context_handler', ] - def __init__(self, hostname, **kwargs): + def __init__(self, hostname, reactor=None, **kwargs): """ Args: hostname : The hostname for the server. """ + if not reactor: + from twisted.internet import reactor + + self._reactor = reactor self.hostname = hostname self._building = {} - self.clock = Clock() + self.clock = Clock(reactor) self.distributor = Distributor() self.ratelimiter = Ratelimiter() @@ -162,6 +196,12 @@ class HomeServer(object): self.datastore = DataStore(self.get_db_conn(), self) logger.info("Finished setting up.") + def get_reactor(self): + """ + Fetch the Twisted reactor in use by this HomeServer. + """ + return self._reactor + def get_ip_from_request(self, request): # X-Forwarded-For is handled by our custom request type. return request.getClientIP() @@ -172,8 +212,26 @@ class HomeServer(object): def is_mine_id(self, string): return string.split(":", 1)[1] == self.hostname - def build_replication_layer(self): - return initialize_http_replication(self) + def get_clock(self): + return self.clock + + def get_datastore(self): + return self.datastore + + def get_config(self): + return self.config + + def get_distributor(self): + return self.distributor + + def get_ratelimiter(self): + return self.ratelimiter + + def build_federation_client(self): + return FederationClient(self) + + def build_federation_server(self): + return FederationServer(self) def build_handlers(self): return Handlers(self) @@ -194,18 +252,15 @@ class HomeServer(object): def build_simple_http_client(self): return SimpleHttpClient(self) - def build_v1auth(self): - orf = Auth(self) - # Matrix spec makes no reference to what HTTP status code is returned, - # but the V1 API uses 403 where it means 401, and the webclient - # relies on this behaviour, so V1 gets its own copy of the auth - # with backwards compat behaviour. - orf.TOKEN_NOT_FOUND_HTTP_STATUS = 403 - return orf + def build_room_creation_handler(self): + return RoomCreationHandler(self) def build_state_handler(self): return StateHandler(self) + def build_state_resolution_handler(self): + return StateResolutionHandler(self) + def build_presence_handler(self): return PresenceHandler(self) @@ -222,7 +277,7 @@ class HomeServer(object): return AuthHandler(self) def build_macaroon_generator(self): - return MacaroonGeneartor(self) + return MacaroonGenerator(self) def build_device_handler(self): return DeviceHandler(self) @@ -251,6 +306,18 @@ class HomeServer(object): def build_initial_sync_handler(self): return InitialSyncHandler(self) + def build_profile_handler(self): + return ProfileHandler(self) + + def build_event_creation_handler(self): + return EventCreationHandler(self) + + def build_deactivate_account_handler(self): + return DeactivateAccountHandler(self) + + def build_set_password_handler(self): + return SetPasswordHandler(self) + def build_event_sources(self): return EventSources(self) @@ -277,9 +344,32 @@ class HomeServer(object): return adbapi.ConnectionPool( name, + cp_reactor=self.get_reactor(), **self.db_config.get("args", {}) ) + def get_db_conn(self, run_new_connection=True): + """Makes a new connection to the database, skipping the db pool + + Returns: + Connection: a connection object implementing the PEP-249 spec + """ + # Any param beginning with cp_ is a parameter for adbapi, and should + # not be passed to the database engine. + db_params = { + k: v for k, v in self.db_config.get("args", {}).items() + if not k.startswith("cp_") + } + db_conn = self.database_engine.module.connect(**db_params) + if run_new_connection: + self.database_engine.on_new_connection(db_conn) + return db_conn + + def build_media_repository_resource(self): + # build the media repo resource. This indirects through the HomeServer + # to ensure that we only have a single instance of + return MediaRepositoryResource(self) + def build_media_repository(self): return MediaRepository(self) @@ -307,7 +397,52 @@ class HomeServer(object): return ActionGenerator(self) def build_user_directory_handler(self): - return UserDirectoyHandler(self) + return UserDirectoryHandler(self) + + def build_groups_local_handler(self): + return GroupsLocalHandler(self) + + def build_groups_server_handler(self): + return GroupsServerHandler(self) + + def build_groups_attestation_signing(self): + return GroupAttestationSigning(self) + + def build_groups_attestation_renewer(self): + return GroupAttestionRenewer(self) + + def build_secrets(self): + return Secrets() + + def build_spam_checker(self): + return SpamChecker(self) + + def build_room_member_handler(self): + if self.config.worker_app: + return RoomMemberWorkerHandler(self) + return RoomMemberMasterHandler(self) + + def build_federation_registry(self): + return FederationHandlerRegistry() + + def build_server_notices_manager(self): + if self.config.worker_app: + raise Exception("Workers cannot send server notices") + return ServerNoticesManager(self) + + def build_server_notices_sender(self): + if self.config.worker_app: + return WorkerServerNoticesSender(self) + return ServerNoticesSender(self) + + def build_message_handler(self): + return MessageHandler(self) + + def build_pagination_handler(self): + return PaginationHandler(self) + + def build_room_context_handler(self): + return RoomContextHandler(self) def remove_pusher(self, app_id, push_key, user_id): return self.get_pusherpool().remove_pusher(app_id, push_key, user_id) diff --git a/synapse/server.pyi b/synapse/server.pyi index 9570df553..ce2848623 100644 --- a/synapse/server.pyi +++ b/synapse/server.pyi @@ -1,12 +1,25 @@ import synapse.api.auth +import synapse.config.homeserver +import synapse.federation.transaction_queue +import synapse.federation.transport.client import synapse.handlers import synapse.handlers.auth +import synapse.handlers.deactivate_account import synapse.handlers.device import synapse.handlers.e2e_keys -import synapse.storage +import synapse.handlers.set_password +import synapse.rest.media.v1.media_repository +import synapse.server_notices.server_notices_manager +import synapse.server_notices.server_notices_sender import synapse.state +import synapse.storage + class HomeServer(object): + @property + def config(self) -> synapse.config.homeserver.HomeServerConfig: + pass + def get_auth(self) -> synapse.api.auth.Auth: pass @@ -27,3 +40,36 @@ class HomeServer(object): def get_state_handler(self) -> synapse.state.StateHandler: pass + + def get_state_resolution_handler(self) -> synapse.state.StateResolutionHandler: + pass + + def get_deactivate_account_handler(self) -> synapse.handlers.deactivate_account.DeactivateAccountHandler: + pass + + def get_room_creation_handler(self) -> synapse.handlers.room.RoomCreationHandler: + pass + + def get_event_creation_handler(self) -> synapse.handlers.message.EventCreationHandler: + pass + + def get_set_password_handler(self) -> synapse.handlers.set_password.SetPasswordHandler: + pass + + def get_federation_sender(self) -> synapse.federation.transaction_queue.TransactionQueue: + pass + + def get_federation_transport_client(self) -> synapse.federation.transport.client.TransportLayerClient: + pass + + def get_media_repository_resource(self) -> synapse.rest.media.v1.media_repository.MediaRepositoryResource: + pass + + def get_media_repository(self) -> synapse.rest.media.v1.media_repository.MediaRepository: + pass + + def get_server_notices_manager(self) -> synapse.server_notices.server_notices_manager.ServerNoticesManager: + pass + + def get_server_notices_sender(self) -> synapse.server_notices.server_notices_sender.ServerNoticesSender: + pass diff --git a/synapse/server_notices/__init__.py b/synapse/server_notices/__init__.py new file mode 100644 index 000000000..e69de29bb diff --git a/synapse/server_notices/consent_server_notices.py b/synapse/server_notices/consent_server_notices.py new file mode 100644 index 000000000..5e3044d16 --- /dev/null +++ b/synapse/server_notices/consent_server_notices.py @@ -0,0 +1,138 @@ +# -*- coding: utf-8 -*- +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import logging + +from six import iteritems, string_types + +from twisted.internet import defer + +from synapse.api.errors import SynapseError +from synapse.api.urls import ConsentURIBuilder +from synapse.config import ConfigError +from synapse.types import get_localpart_from_id + +logger = logging.getLogger(__name__) + + +class ConsentServerNotices(object): + """Keeps track of whether we need to send users server_notices about + privacy policy consent, and sends one if we do. + """ + def __init__(self, hs): + """ + + Args: + hs (synapse.server.HomeServer): + """ + self._server_notices_manager = hs.get_server_notices_manager() + self._store = hs.get_datastore() + + self._users_in_progress = set() + + self._current_consent_version = hs.config.user_consent_version + self._server_notice_content = hs.config.user_consent_server_notice_content + self._send_to_guests = hs.config.user_consent_server_notice_to_guests + + if self._server_notice_content is not None: + if not self._server_notices_manager.is_enabled(): + raise ConfigError( + "user_consent configuration requires server notices, but " + "server notices are not enabled.", + ) + if 'body' not in self._server_notice_content: + raise ConfigError( + "user_consent server_notice_consent must contain a 'body' " + "key.", + ) + + self._consent_uri_builder = ConsentURIBuilder(hs.config) + + @defer.inlineCallbacks + def maybe_send_server_notice_to_user(self, user_id): + """Check if we need to send a notice to this user, and does so if so + + Args: + user_id (str): user to check + + Returns: + Deferred + """ + if self._server_notice_content is None: + # not enabled + return + + # make sure we don't send two messages to the same user at once + if user_id in self._users_in_progress: + return + self._users_in_progress.add(user_id) + try: + u = yield self._store.get_user_by_id(user_id) + + if u["is_guest"] and not self._send_to_guests: + # don't send to guests + return + + if u["consent_version"] == self._current_consent_version: + # user has already consented + return + + if u["consent_server_notice_sent"] == self._current_consent_version: + # we've already sent a notice to the user + return + + # need to send a message. + try: + consent_uri = self._consent_uri_builder.build_user_consent_uri( + get_localpart_from_id(user_id), + ) + content = copy_with_str_subst( + self._server_notice_content, { + 'consent_uri': consent_uri, + }, + ) + yield self._server_notices_manager.send_notice( + user_id, content, + ) + yield self._store.user_set_consent_server_notice_sent( + user_id, self._current_consent_version, + ) + except SynapseError as e: + logger.error("Error sending server notice about user consent: %s", e) + finally: + self._users_in_progress.remove(user_id) + + +def copy_with_str_subst(x, substitutions): + """Deep-copy a structure, carrying out string substitions on any strings + + Args: + x (object): structure to be copied + substitutions (object): substitutions to be made - passed into the + string '%' operator + + Returns: + copy of x + """ + if isinstance(x, string_types): + return x % substitutions + if isinstance(x, dict): + return { + k: copy_with_str_subst(v, substitutions) for (k, v) in iteritems(x) + } + if isinstance(x, (list, tuple)): + return [copy_with_str_subst(y) for y in x] + + # assume it's uninterested and can be shallow-copied. + return x diff --git a/synapse/server_notices/server_notices_manager.py b/synapse/server_notices/server_notices_manager.py new file mode 100644 index 000000000..a26deace5 --- /dev/null +++ b/synapse/server_notices/server_notices_manager.py @@ -0,0 +1,146 @@ +# -*- coding: utf-8 -*- +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import logging + +from twisted.internet import defer + +from synapse.api.constants import EventTypes, Membership, RoomCreationPreset +from synapse.types import create_requester +from synapse.util.caches.descriptors import cachedInlineCallbacks + +logger = logging.getLogger(__name__) + + +class ServerNoticesManager(object): + def __init__(self, hs): + """ + + Args: + hs (synapse.server.HomeServer): + """ + + self._store = hs.get_datastore() + self._config = hs.config + self._room_creation_handler = hs.get_room_creation_handler() + self._event_creation_handler = hs.get_event_creation_handler() + self._is_mine_id = hs.is_mine_id + + def is_enabled(self): + """Checks if server notices are enabled on this server. + + Returns: + bool + """ + return self._config.server_notices_mxid is not None + + @defer.inlineCallbacks + def send_notice(self, user_id, event_content): + """Send a notice to the given user + + Creates the server notices room, if none exists. + + Args: + user_id (str): mxid of user to send event to. + event_content (dict): content of event to send + + Returns: + Deferred[None] + """ + room_id = yield self.get_notice_room_for_user(user_id) + + system_mxid = self._config.server_notices_mxid + requester = create_requester(system_mxid) + + logger.info("Sending server notice to %s", user_id) + + yield self._event_creation_handler.create_and_send_nonmember_event( + requester, { + "type": EventTypes.Message, + "room_id": room_id, + "sender": system_mxid, + "content": event_content, + }, + ratelimit=False, + ) + + @cachedInlineCallbacks() + def get_notice_room_for_user(self, user_id): + """Get the room for notices for a given user + + If we have not yet created a notice room for this user, create it + + Args: + user_id (str): complete user id for the user we want a room for + + Returns: + str: room id of notice room. + """ + if not self.is_enabled(): + raise Exception("Server notices not enabled") + + assert self._is_mine_id(user_id), \ + "Cannot send server notices to remote users" + + rooms = yield self._store.get_rooms_for_user_where_membership_is( + user_id, [Membership.INVITE, Membership.JOIN], + ) + system_mxid = self._config.server_notices_mxid + for room in rooms: + # it's worth noting that there is an asymmetry here in that we + # expect the user to be invited or joined, but the system user must + # be joined. This is kinda deliberate, in that if somebody somehow + # manages to invite the system user to a room, that doesn't make it + # the server notices room. + user_ids = yield self._store.get_users_in_room(room.room_id) + if system_mxid in user_ids: + # we found a room which our user shares with the system notice + # user + logger.info("Using room %s", room.room_id) + defer.returnValue(room.room_id) + + # apparently no existing notice room: create a new one + logger.info("Creating server notices room for %s", user_id) + + # see if we want to override the profile info for the server user. + # note that if we want to override either the display name or the + # avatar, we have to use both. + join_profile = None + if ( + self._config.server_notices_mxid_display_name is not None or + self._config.server_notices_mxid_avatar_url is not None + ): + join_profile = { + "displayname": self._config.server_notices_mxid_display_name, + "avatar_url": self._config.server_notices_mxid_avatar_url, + } + + requester = create_requester(system_mxid) + info = yield self._room_creation_handler.create_room( + requester, + config={ + "preset": RoomCreationPreset.PRIVATE_CHAT, + "name": self._config.server_notices_room_name, + "power_level_content_override": { + "users_default": -10, + }, + "invite": (user_id,) + }, + ratelimit=False, + creator_join_profile=join_profile, + ) + room_id = info['room_id'] + + logger.info("Created server notices room %s for %s", room_id, user_id) + defer.returnValue(room_id) diff --git a/synapse/server_notices/server_notices_sender.py b/synapse/server_notices/server_notices_sender.py new file mode 100644 index 000000000..5d23965f3 --- /dev/null +++ b/synapse/server_notices/server_notices_sender.py @@ -0,0 +1,58 @@ +# -*- coding: utf-8 -*- +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +from synapse.server_notices.consent_server_notices import ConsentServerNotices + + +class ServerNoticesSender(object): + """A centralised place which sends server notices automatically when + Certain Events take place + """ + def __init__(self, hs): + """ + + Args: + hs (synapse.server.HomeServer): + """ + # todo: it would be nice to make this more dynamic + self._consent_server_notices = ConsentServerNotices(hs) + + def on_user_syncing(self, user_id): + """Called when the user performs a sync operation. + + Args: + user_id (str): mxid of user who synced + + Returns: + Deferred + """ + return self._consent_server_notices.maybe_send_server_notice_to_user( + user_id, + ) + + def on_user_ip(self, user_id): + """Called on the master when a worker process saw a client request. + + Args: + user_id (str): mxid + + Returns: + Deferred + """ + # The synchrotrons use a stubbed version of ServerNoticesSender, so + # we check for notices to send to the user in on_user_ip as well as + # in on_user_syncing + return self._consent_server_notices.maybe_send_server_notice_to_user( + user_id, + ) diff --git a/synapse/server_notices/worker_server_notices_sender.py b/synapse/server_notices/worker_server_notices_sender.py new file mode 100644 index 000000000..4a133026c --- /dev/null +++ b/synapse/server_notices/worker_server_notices_sender.py @@ -0,0 +1,46 @@ +# -*- coding: utf-8 -*- +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +from twisted.internet import defer + + +class WorkerServerNoticesSender(object): + """Stub impl of ServerNoticesSender which does nothing""" + def __init__(self, hs): + """ + Args: + hs (synapse.server.HomeServer): + """ + + def on_user_syncing(self, user_id): + """Called when the user performs a sync operation. + + Args: + user_id (str): mxid of user who synced + + Returns: + Deferred + """ + return defer.succeed(None) + + def on_user_ip(self, user_id): + """Called on the master when a worker process saw a client request. + + Args: + user_id (str): mxid + + Returns: + Deferred + """ + raise AssertionError("on_user_ip unexpectedly called on worker") diff --git a/synapse/state.py b/synapse/state.py index 390799fbd..033f55d96 100644 --- a/synapse/state.py +++ b/synapse/state.py @@ -14,23 +14,25 @@ # limitations under the License. +import hashlib +import logging +from collections import namedtuple + +from six import iteritems, iterkeys, itervalues + +from frozendict import frozendict + from twisted.internet import defer from synapse import event_auth -from synapse.util.logutils import log_function -from synapse.util.caches.expiringcache import ExpiringCache -from synapse.util.metrics import Measure from synapse.api.constants import EventTypes from synapse.api.errors import AuthError from synapse.events.snapshot import EventContext from synapse.util.async import Linearizer from synapse.util.caches import CACHE_SIZE_FACTOR - -from collections import namedtuple -from frozendict import frozendict - -import logging -import hashlib +from synapse.util.caches.expiringcache import ExpiringCache +from synapse.util.logutils import log_function +from synapse.util.metrics import Measure logger = logging.getLogger(__name__) @@ -58,7 +60,11 @@ class _StateCacheEntry(object): __slots__ = ["state", "state_group", "state_id", "prev_group", "delta_ids"] def __init__(self, state, state_group, prev_group=None, delta_ids=None): + # dict[(str, str), str] map from (type, state_key) to event_id self.state = frozendict(state) + + # the ID of a state group if one and only one is involved. + # otherwise, None otherwise? self.state_group = state_group self.prev_group = prev_group @@ -81,31 +87,19 @@ class _StateCacheEntry(object): class StateHandler(object): - """ Responsible for doing state conflict resolution. + """Fetches bits of state from the stores, and does state resolution + where necessary """ def __init__(self, hs): self.clock = hs.get_clock() self.store = hs.get_datastore() self.hs = hs - - # dict of set of event_ids -> _StateCacheEntry. - self._state_cache = None - self.resolve_linearizer = Linearizer(name="state_resolve_lock") + self._state_resolution_handler = hs.get_state_resolution_handler() def start_caching(self): - logger.debug("start_caching") - - self._state_cache = ExpiringCache( - cache_name="state_cache", - clock=self.clock, - max_len=SIZE_OF_CACHE, - expiry_ms=EVICTION_TIMEOUT_SECONDS * 1000, - iterable=True, - reset_expiry_on_get=True, - ) - - self._state_cache.start() + # TODO: remove this shim + self._state_resolution_handler.start_caching() @defer.inlineCallbacks def get_current_state(self, room_id, event_type=None, state_key="", @@ -127,7 +121,7 @@ class StateHandler(object): latest_event_ids = yield self.store.get_latest_event_ids_in_room(room_id) logger.debug("calling resolve_state_groups from get_current_state") - ret = yield self.resolve_state_groups(room_id, latest_event_ids) + ret = yield self.resolve_state_groups_for_events(room_id, latest_event_ids) state = ret.state if event_type: @@ -138,27 +132,36 @@ class StateHandler(object): defer.returnValue(event) return - state_map = yield self.store.get_events(state.values(), get_prev_content=False) + state_map = yield self.store.get_events(list(state.values()), + get_prev_content=False) state = { - key: state_map[e_id] for key, e_id in state.items() if e_id in state_map + key: state_map[e_id] for key, e_id in iteritems(state) if e_id in state_map } defer.returnValue(state) @defer.inlineCallbacks - def get_current_state_ids(self, room_id, event_type=None, state_key="", - latest_event_ids=None): + def get_current_state_ids(self, room_id, latest_event_ids=None): + """Get the current state, or the state at a set of events, for a room + + Args: + room_id (str): + + latest_event_ids (iterable[str]|None): if given, the forward + extremities to resolve. If None, we look them up from the + database (via a cache) + + Returns: + Deferred[dict[(str, str), str)]]: the state dict, mapping from + (event_type, state_key) -> event_id + """ if not latest_event_ids: latest_event_ids = yield self.store.get_latest_event_ids_in_room(room_id) logger.debug("calling resolve_state_groups from get_current_state_ids") - ret = yield self.resolve_state_groups(room_id, latest_event_ids) + ret = yield self.resolve_state_groups_for_events(room_id, latest_event_ids) state = ret.state - if event_type: - defer.returnValue(state.get((event_type, state_key))) - return - defer.returnValue(state) @defer.inlineCallbacks @@ -166,7 +169,7 @@ class StateHandler(object): if not latest_event_ids: latest_event_ids = yield self.store.get_latest_event_ids_in_room(room_id) logger.debug("calling resolve_state_groups from get_current_user_in_room") - entry = yield self.resolve_state_groups(room_id, latest_event_ids) + entry = yield self.resolve_state_groups_for_events(room_id, latest_event_ids) joined_users = yield self.store.get_joined_users_from_state(room_id, entry) defer.returnValue(joined_users) @@ -175,7 +178,7 @@ class StateHandler(object): if not latest_event_ids: latest_event_ids = yield self.store.get_latest_event_ids_in_room(room_id) logger.debug("calling resolve_state_groups from get_current_hosts_in_room") - entry = yield self.resolve_state_groups(room_id, latest_event_ids) + entry = yield self.resolve_state_groups_for_events(room_id, latest_event_ids) joined_hosts = yield self.store.get_joined_hosts(room_id, entry) defer.returnValue(joined_hosts) @@ -183,8 +186,15 @@ class StateHandler(object): def compute_event_context(self, event, old_state=None): """Build an EventContext structure for the event. + This works out what the current state should be for the event, and + generates a new state group if necessary. + Args: event (synapse.events.EventBase): + old_state (dict|None): The state at the event if it can't be + calculated from existing events. This is normally only specified + when receiving an event from federation where we don't have the + prev events for, e.g. when backfilling. Returns: synapse.events.snapshot.EventContext: """ @@ -193,113 +203,158 @@ class StateHandler(object): # If this is an outlier, then we know it shouldn't have any current # state. Certainly store.get_current_state won't return any, and # persisting the event won't store the state group. - context = EventContext() if old_state: - context.prev_state_ids = { + prev_state_ids = { (s.type, s.state_key): s.event_id for s in old_state } if event.is_state(): - context.current_state_ids = dict(context.prev_state_ids) + current_state_ids = dict(prev_state_ids) key = (event.type, event.state_key) - context.current_state_ids[key] = event.event_id + current_state_ids[key] = event.event_id else: - context.current_state_ids = context.prev_state_ids + current_state_ids = prev_state_ids else: - context.current_state_ids = {} - context.prev_state_ids = {} - context.prev_state_events = [] - context.state_group = self.store.get_next_state_group() + current_state_ids = {} + prev_state_ids = {} + + # We don't store state for outliers, so we don't generate a state + # group for it. + context = EventContext.with_state( + state_group=None, + current_state_ids=current_state_ids, + prev_state_ids=prev_state_ids, + ) + defer.returnValue(context) if old_state: - context = EventContext() - context.prev_state_ids = { + # We already have the state, so we don't need to calculate it. + # Let's just correctly fill out the context and create a + # new state group for it. + + prev_state_ids = { (s.type, s.state_key): s.event_id for s in old_state } - context.state_group = self.store.get_next_state_group() if event.is_state(): key = (event.type, event.state_key) - if key in context.prev_state_ids: - replaces = context.prev_state_ids[key] + if key in prev_state_ids: + replaces = prev_state_ids[key] if replaces != event.event_id: # Paranoia check event.unsigned["replaces_state"] = replaces - context.current_state_ids = dict(context.prev_state_ids) - context.current_state_ids[key] = event.event_id + current_state_ids = dict(prev_state_ids) + current_state_ids[key] = event.event_id else: - context.current_state_ids = context.prev_state_ids + current_state_ids = prev_state_ids + + state_group = yield self.store.store_state_group( + event.event_id, + event.room_id, + prev_group=None, + delta_ids=None, + current_state_ids=current_state_ids, + ) + + context = EventContext.with_state( + state_group=state_group, + current_state_ids=current_state_ids, + prev_state_ids=prev_state_ids, + ) - context.prev_state_events = [] defer.returnValue(context) logger.debug("calling resolve_state_groups from compute_event_context") - entry = yield self.resolve_state_groups( + entry = yield self.resolve_state_groups_for_events( event.room_id, [e for e, _ in event.prev_events], ) - curr_state = entry.state + prev_state_ids = entry.state + prev_group = None + delta_ids = None - context = EventContext() - context.prev_state_ids = curr_state if event.is_state(): - context.state_group = self.store.get_next_state_group() + # If this is a state event then we need to create a new state + # group for the state after this event. key = (event.type, event.state_key) - if key in context.prev_state_ids: - replaces = context.prev_state_ids[key] + if key in prev_state_ids: + replaces = prev_state_ids[key] event.unsigned["replaces_state"] = replaces - context.current_state_ids = dict(context.prev_state_ids) - context.current_state_ids[key] = event.event_id + current_state_ids = dict(prev_state_ids) + current_state_ids[key] = event.event_id if entry.state_group: - context.prev_group = entry.state_group - context.delta_ids = { + # If the state at the event has a state group assigned then + # we can use that as the prev group + prev_group = entry.state_group + delta_ids = { key: event.event_id } elif entry.prev_group: - context.prev_group = entry.prev_group - context.delta_ids = dict(entry.delta_ids) - context.delta_ids[key] = event.event_id + # If the state at the event only has a prev group, then we can + # use that as a prev group too. + prev_group = entry.prev_group + delta_ids = dict(entry.delta_ids) + delta_ids[key] = event.event_id + + state_group = yield self.store.store_state_group( + event.event_id, + event.room_id, + prev_group=prev_group, + delta_ids=delta_ids, + current_state_ids=current_state_ids, + ) else: + current_state_ids = prev_state_ids + prev_group = entry.prev_group + delta_ids = entry.delta_ids + if entry.state_group is None: - entry.state_group = self.store.get_next_state_group() + entry.state_group = yield self.store.store_state_group( + event.event_id, + event.room_id, + prev_group=entry.prev_group, + delta_ids=entry.delta_ids, + current_state_ids=current_state_ids, + ) entry.state_id = entry.state_group - context.state_group = entry.state_group - context.current_state_ids = context.prev_state_ids - context.prev_group = entry.prev_group - context.delta_ids = entry.delta_ids + state_group = entry.state_group + + context = EventContext.with_state( + state_group=state_group, + current_state_ids=current_state_ids, + prev_state_ids=prev_state_ids, + prev_group=prev_group, + delta_ids=delta_ids, + ) - context.prev_state_events = [] defer.returnValue(context) @defer.inlineCallbacks - @log_function - def resolve_state_groups(self, room_id, event_ids): + def resolve_state_groups_for_events(self, room_id, event_ids): """ Given a list of event_ids this method fetches the state at each event, resolves conflicts between them and returns them. + Args: + room_id (str): + event_ids (list[str]): + Returns: - a Deferred tuple of (`state_group`, `state`, `prev_state`). - `state_group` is the name of a state group if one and only one is - involved. `state` is a map from (type, state_key) to event, and - `prev_state` is a list of event ids. + Deferred[_StateCacheEntry]: resolved state """ logger.debug("resolve_state_groups event_ids %s", event_ids) + # map from state group id to the state in that state group (where + # 'state' is a map from state key to event id) + # dict[int, dict[(str, str), str]] state_groups_ids = yield self.store.get_state_groups_ids( room_id, event_ids ) - logger.debug( - "resolve_state_groups state_groups %s", - state_groups_ids.keys() - ) - - group_names = frozenset(state_groups_ids.keys()) - if len(group_names) == 1: - name, state_list = state_groups_ids.items().pop() + if len(state_groups_ids) == 1: + name, state_list = list(state_groups_ids.items()).pop() prev_group, delta_ids = yield self.store.get_state_group_delta(name) @@ -310,76 +365,15 @@ class StateHandler(object): delta_ids=delta_ids, )) - with (yield self.resolve_linearizer.queue(group_names)): - if self._state_cache is not None: - cache = self._state_cache.get(group_names, None) - if cache: - defer.returnValue(cache) + result = yield self._state_resolution_handler.resolve_state_groups( + room_id, state_groups_ids, None, self._state_map_factory, + ) + defer.returnValue(result) - logger.info( - "Resolving state for %s with %d groups", room_id, len(state_groups_ids) - ) - - state = {} - for st in state_groups_ids.values(): - for key, e_id in st.items(): - state.setdefault(key, set()).add(e_id) - - conflicted_state = { - k: list(v) - for k, v in state.items() - if len(v) > 1 - } - - if conflicted_state: - logger.info("Resolving conflicted state for %r", room_id) - with Measure(self.clock, "state._resolve_events"): - new_state = yield resolve_events( - state_groups_ids.values(), - state_map_factory=lambda ev_ids: self.store.get_events( - ev_ids, get_prev_content=False, check_redacted=False, - ), - ) - else: - new_state = { - key: e_ids.pop() for key, e_ids in state.items() - } - - state_group = None - new_state_event_ids = frozenset(new_state.values()) - for sg, events in state_groups_ids.items(): - if new_state_event_ids == frozenset(e_id for e_id in events): - state_group = sg - break - - # TODO: We want to create a state group for this set of events, to - # increase cache hits, but we need to make sure that it doesn't - # end up as a prev_group without being added to the database - - prev_group = None - delta_ids = None - for old_group, old_ids in state_groups_ids.iteritems(): - if not set(new_state) - set(old_ids): - n_delta_ids = { - k: v - for k, v in new_state.iteritems() - if old_ids.get(k) != v - } - if not delta_ids or len(n_delta_ids) < len(delta_ids): - prev_group = old_group - delta_ids = n_delta_ids - - cache = _StateCacheEntry( - state=new_state, - state_group=state_group, - prev_group=prev_group, - delta_ids=delta_ids, - ) - - if self._state_cache is not None: - self._state_cache[group_names] = cache - - defer.returnValue(cache) + def _state_map_factory(self, ev_ids): + return self.store.get_events( + ev_ids, get_prev_content=False, check_redacted=False, + ) def resolve_events(self, state_sets, event): logger.info( @@ -397,35 +391,208 @@ class StateHandler(object): } with Measure(self.clock, "state._resolve_events"): - new_state = resolve_events(state_set_ids, state_map) + new_state = resolve_events_with_state_map(state_set_ids, state_map) new_state = { - key: state_map[ev_id] for key, ev_id in new_state.items() + key: state_map[ev_id] for key, ev_id in iteritems(new_state) } return new_state +class StateResolutionHandler(object): + """Responsible for doing state conflict resolution. + + Note that the storage layer depends on this handler, so all functions must + be storage-independent. + """ + def __init__(self, hs): + self.clock = hs.get_clock() + + # dict of set of event_ids -> _StateCacheEntry. + self._state_cache = None + self.resolve_linearizer = Linearizer(name="state_resolve_lock") + + def start_caching(self): + logger.debug("start_caching") + + self._state_cache = ExpiringCache( + cache_name="state_cache", + clock=self.clock, + max_len=SIZE_OF_CACHE, + expiry_ms=EVICTION_TIMEOUT_SECONDS * 1000, + iterable=True, + reset_expiry_on_get=True, + ) + + self._state_cache.start() + + @defer.inlineCallbacks + @log_function + def resolve_state_groups( + self, room_id, state_groups_ids, event_map, state_map_factory, + ): + """Resolves conflicts between a set of state groups + + Always generates a new state group (unless we hit the cache), so should + not be called for a single state group + + Args: + room_id (str): room we are resolving for (used for logging) + state_groups_ids (dict[int, dict[(str, str), str]]): + map from state group id to the state in that state group + (where 'state' is a map from state key to event id) + + event_map(dict[str,FrozenEvent]|None): + a dict from event_id to event, for any events that we happen to + have in flight (eg, those currently being persisted). This will be + used as a starting point fof finding the state we need; any missing + events will be requested via state_map_factory. + + If None, all events will be fetched via state_map_factory. + + Returns: + Deferred[_StateCacheEntry]: resolved state + """ + logger.debug( + "resolve_state_groups state_groups %s", + state_groups_ids.keys() + ) + + group_names = frozenset(state_groups_ids.keys()) + + with (yield self.resolve_linearizer.queue(group_names)): + if self._state_cache is not None: + cache = self._state_cache.get(group_names, None) + if cache: + defer.returnValue(cache) + + logger.info( + "Resolving state for %s with %d groups", room_id, len(state_groups_ids) + ) + + # start by assuming we won't have any conflicted state, and build up the new + # state map by iterating through the state groups. If we discover a conflict, + # we give up and instead use `resolve_events_with_factory`. + # + # XXX: is this actually worthwhile, or should we just let + # resolve_events_with_factory do it? + new_state = {} + conflicted_state = False + for st in itervalues(state_groups_ids): + for key, e_id in iteritems(st): + if key in new_state: + conflicted_state = True + break + new_state[key] = e_id + if conflicted_state: + break + + if conflicted_state: + logger.info("Resolving conflicted state for %r", room_id) + with Measure(self.clock, "state._resolve_events"): + new_state = yield resolve_events_with_factory( + list(itervalues(state_groups_ids)), + event_map=event_map, + state_map_factory=state_map_factory, + ) + + # if the new state matches any of the input state groups, we can + # use that state group again. Otherwise we will generate a state_id + # which will be used as a cache key for future resolutions, but + # not get persisted. + + with Measure(self.clock, "state.create_group_ids"): + cache = _make_state_cache_entry(new_state, state_groups_ids) + + if self._state_cache is not None: + self._state_cache[group_names] = cache + + defer.returnValue(cache) + + +def _make_state_cache_entry( + new_state, + state_groups_ids, +): + """Given a resolved state, and a set of input state groups, pick one to base + a new state group on (if any), and return an appropriately-constructed + _StateCacheEntry. + + Args: + new_state (dict[(str, str), str]): resolved state map (mapping from + (type, state_key) to event_id) + + state_groups_ids (dict[int, dict[(str, str), str]]): + map from state group id to the state in that state group + (where 'state' is a map from state key to event id) + + Returns: + _StateCacheEntry + """ + # if the new state matches any of the input state groups, we can + # use that state group again. Otherwise we will generate a state_id + # which will be used as a cache key for future resolutions, but + # not get persisted. + + # first look for exact matches + new_state_event_ids = set(itervalues(new_state)) + for sg, state in iteritems(state_groups_ids): + if len(new_state_event_ids) != len(state): + continue + + old_state_event_ids = set(itervalues(state)) + if new_state_event_ids == old_state_event_ids: + # got an exact match. + return _StateCacheEntry( + state=new_state, + state_group=sg, + ) + + # TODO: We want to create a state group for this set of events, to + # increase cache hits, but we need to make sure that it doesn't + # end up as a prev_group without being added to the database + + # failing that, look for the closest match. + prev_group = None + delta_ids = None + + for old_group, old_state in iteritems(state_groups_ids): + n_delta_ids = { + k: v + for k, v in iteritems(new_state) + if old_state.get(k) != v + } + if not delta_ids or len(n_delta_ids) < len(delta_ids): + prev_group = old_group + delta_ids = n_delta_ids + + return _StateCacheEntry( + state=new_state, + state_group=None, + prev_group=prev_group, + delta_ids=delta_ids, + ) + + def _ordered_events(events): def key_func(e): - return -int(e.depth), hashlib.sha1(e.event_id).hexdigest() + return -int(e.depth), hashlib.sha1(e.event_id.encode()).hexdigest() return sorted(events, key=key_func) -def resolve_events(state_sets, state_map_factory): +def resolve_events_with_state_map(state_sets, state_map): """ Args: state_sets(list): List of dicts of (type, state_key) -> event_id, which are the different state groups to resolve. - state_map_factory(dict|callable): If callable, then will be called - with a list of event_ids that are needed, and should return with - a Deferred of dict of event_id to event. Otherwise, should be - a dict from event_id to event of all events in state_sets. + state_map(dict): a dict from event_id to event, for all events in + state_sets. Returns - dict[(str, str), synapse.events.FrozenEvent] is a map from - (type, state_key) to event. + dict[(str, str), str]: + a map from (type, state_key) to event_id. """ if len(state_sets) == 1: return state_sets[0] @@ -434,13 +601,6 @@ def resolve_events(state_sets, state_map_factory): state_sets, ) - if callable(state_map_factory): - return _resolve_with_state_fac( - unconflicted_state, conflicted_state, state_map_factory - ) - - state_map = state_map_factory - auth_events = _create_auth_events_from_maps( unconflicted_state, conflicted_state, state_map ) @@ -454,12 +614,28 @@ def _seperate(state_sets): """Takes the state_sets and figures out which keys are conflicted and which aren't. i.e., which have multiple different event_ids associated with them in different state sets. + + Args: + state_sets(iterable[dict[(str, str), str]]): + List of dicts of (type, state_key) -> event_id, which are the + different state groups to resolve. + + Returns: + (dict[(str, str), str], dict[(str, str), set[str]]): + A tuple of (unconflicted_state, conflicted_state), where: + + unconflicted_state is a dict mapping (type, state_key)->event_id + for unconflicted state keys. + + conflicted_state is a dict mapping (type, state_key) to a set of + event ids for conflicted state keys. """ - unconflicted_state = dict(state_sets[0]) + state_set_iterator = iter(state_sets) + unconflicted_state = dict(next(state_set_iterator)) conflicted_state = {} - for state_set in state_sets[1:]: - for key, value in state_set.iteritems(): + for state_set in state_set_iterator: + for key, value in iteritems(state_set): # Check if there is an unconflicted entry for the state key. unconflicted_value = unconflicted_state.get(key) if unconflicted_value is None: @@ -484,24 +660,63 @@ def _seperate(state_sets): @defer.inlineCallbacks -def _resolve_with_state_fac(unconflicted_state, conflicted_state, - state_map_factory): +def resolve_events_with_factory(state_sets, event_map, state_map_factory): + """ + Args: + state_sets(list): List of dicts of (type, state_key) -> event_id, + which are the different state groups to resolve. + + event_map(dict[str,FrozenEvent]|None): + a dict from event_id to event, for any events that we happen to + have in flight (eg, those currently being persisted). This will be + used as a starting point fof finding the state we need; any missing + events will be requested via state_map_factory. + + If None, all events will be fetched via state_map_factory. + + state_map_factory(func): will be called + with a list of event_ids that are needed, and should return with + a Deferred of dict of event_id to event. + + Returns + Deferred[dict[(str, str), str]]: + a map from (type, state_key) to event_id. + """ + if len(state_sets) == 1: + defer.returnValue(state_sets[0]) + + unconflicted_state, conflicted_state = _seperate( + state_sets, + ) + needed_events = set( event_id - for event_ids in conflicted_state.itervalues() + for event_ids in itervalues(conflicted_state) for event_id in event_ids ) + if event_map is not None: + needed_events -= set(iterkeys(event_map)) logger.info("Asking for %d conflicted events", len(needed_events)) + # dict[str, FrozenEvent]: a map from state event id to event. Only includes + # the state events which are in conflict (and those in event_map) state_map = yield state_map_factory(needed_events) + if event_map is not None: + state_map.update(event_map) + # get the ids of the auth events which allow us to authenticate the + # conflicted state, picking only from the unconflicting state. + # + # dict[(str, str), str]: a map from state key to event id auth_events = _create_auth_events_from_maps( unconflicted_state, conflicted_state, state_map ) - new_needed_events = set(auth_events.itervalues()) + new_needed_events = set(itervalues(auth_events)) new_needed_events -= needed_events + if event_map is not None: + new_needed_events -= set(iterkeys(event_map)) logger.info("Asking for %d auth events", len(new_needed_events)) @@ -515,7 +730,7 @@ def _resolve_with_state_fac(unconflicted_state, conflicted_state, def _create_auth_events_from_maps(unconflicted_state, conflicted_state, state_map): auth_events = {} - for event_ids in conflicted_state.itervalues(): + for event_ids in itervalues(conflicted_state): for event_id in event_ids: if event_id in state_map: keys = event_auth.auth_types_for_event(state_map[event_id]) @@ -527,10 +742,10 @@ def _create_auth_events_from_maps(unconflicted_state, conflicted_state, state_ma return auth_events -def _resolve_with_state(unconflicted_state_ids, conflicted_state_ds, auth_event_ids, +def _resolve_with_state(unconflicted_state_ids, conflicted_state_ids, auth_event_ids, state_map): conflicted_state = {} - for key, event_ids in conflicted_state_ds.iteritems(): + for key, event_ids in iteritems(conflicted_state_ids): events = [state_map[ev_id] for ev_id in event_ids if ev_id in state_map] if len(events) > 1: conflicted_state[key] = events @@ -539,7 +754,7 @@ def _resolve_with_state(unconflicted_state_ids, conflicted_state_ds, auth_event_ auth_events = { key: state_map[ev_id] - for key, ev_id in auth_event_ids.items() + for key, ev_id in iteritems(auth_event_ids) if ev_id in state_map } @@ -547,12 +762,12 @@ def _resolve_with_state(unconflicted_state_ids, conflicted_state_ds, auth_event_ resolved_state = _resolve_state_events( conflicted_state, auth_events ) - except: + except Exception: logger.exception("Failed to resolve state") raise new_state = unconflicted_state_ids - for key, event in resolved_state.iteritems(): + for key, event in iteritems(resolved_state): new_state[key] = event.event_id return new_state @@ -577,7 +792,7 @@ def _resolve_state_events(conflicted_state, auth_events): auth_events.update(resolved_state) - for key, events in conflicted_state.items(): + for key, events in iteritems(conflicted_state): if key[0] == EventTypes.JoinRules: logger.debug("Resolving conflicted join rules %r", events) resolved_state[key] = _resolve_auth_events( @@ -587,7 +802,7 @@ def _resolve_state_events(conflicted_state, auth_events): auth_events.update(resolved_state) - for key, events in conflicted_state.items(): + for key, events in iteritems(conflicted_state): if key[0] == EventTypes.Member: logger.debug("Resolving conflicted member lists %r", events) resolved_state[key] = _resolve_auth_events( @@ -597,7 +812,7 @@ def _resolve_state_events(conflicted_state, auth_events): auth_events.update(resolved_state) - for key, events in conflicted_state.items(): + for key, events in iteritems(conflicted_state): if key not in resolved_state: logger.debug("Resolving conflicted state %r:%r", key, events) resolved_state[key] = _resolve_normal_events( diff --git a/synapse/storage/__init__.py b/synapse/storage/__init__.py index b92472df3..ba88a5497 100644 --- a/synapse/storage/__init__.py +++ b/synapse/storage/__init__.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2014-2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -13,53 +14,49 @@ # See the License for the specific language governing permissions and # limitations under the License. -from twisted.internet import defer +import datetime +import logging +import time -from synapse.storage.devices import DeviceStore -from .appservice import ( - ApplicationServiceStore, ApplicationServiceTransactionStore -) -from ._base import LoggingTransaction -from .directory import DirectoryStore -from .events import EventsStore -from .presence import PresenceStore, UserPresenceState -from .profile import ProfileStore -from .registration import RegistrationStore -from .room import RoomStore -from .roommember import RoomMemberStore -from .stream import StreamStore -from .transactions import TransactionStore -from .keys import KeyStore -from .event_federation import EventFederationStore -from .pusher import PusherStore -from .push_rule import PushRuleStore -from .media_repository import MediaRepositoryStore -from .rejections import RejectionsStore -from .event_push_actions import EventPushActionsStore -from .deviceinbox import DeviceInboxStore - -from .state import StateStore -from .signatures import SignatureStore -from .filtering import FilteringStore -from .end_to_end_keys import EndToEndKeyStore - -from .receipts import ReceiptsStore -from .search import SearchStore -from .tags import TagsStore -from .account_data import AccountDataStore -from .openid import OpenIdStore -from .client_ips import ClientIpStore -from .user_directory import UserDirectoryStore - -from .util.id_generators import IdGenerator, StreamIdGenerator, ChainedIdGenerator -from .engines import PostgresEngine +from dateutil import tz from synapse.api.constants import PresenceState +from synapse.storage.devices import DeviceStore +from synapse.storage.user_erasure_store import UserErasureStore from synapse.util.caches.stream_change_cache import StreamChangeCache - -import logging - +from .account_data import AccountDataStore +from .appservice import ApplicationServiceStore, ApplicationServiceTransactionStore +from .client_ips import ClientIpStore +from .deviceinbox import DeviceInboxStore +from .directory import DirectoryStore +from .end_to_end_keys import EndToEndKeyStore +from .engines import PostgresEngine +from .event_federation import EventFederationStore +from .event_push_actions import EventPushActionsStore +from .events import EventsStore +from .filtering import FilteringStore +from .group_server import GroupServerStore +from .keys import KeyStore +from .media_repository import MediaRepositoryStore +from .openid import OpenIdStore +from .presence import PresenceStore, UserPresenceState +from .profile import ProfileStore +from .push_rule import PushRuleStore +from .pusher import PusherStore +from .receipts import ReceiptsStore +from .registration import RegistrationStore +from .rejections import RejectionsStore +from .room import RoomStore +from .roommember import RoomMemberStore +from .search import SearchStore +from .signatures import SignatureStore +from .state import StateStore +from .stream import StreamStore +from .tags import TagsStore +from .transactions import TransactionStore +from .user_directory import UserDirectoryStore +from .util.id_generators import ChainedIdGenerator, IdGenerator, StreamIdGenerator logger = logging.getLogger(__name__) @@ -88,6 +85,8 @@ class DataStore(RoomMemberStore, RoomStore, DeviceStore, DeviceInboxStore, UserDirectoryStore, + GroupServerStore, + UserErasureStore, ): def __init__(self, db_conn, hs): @@ -103,12 +102,6 @@ class DataStore(RoomMemberStore, RoomStore, db_conn, "events", "stream_ordering", step=-1, extra_tables=[("ex_outlier_stream", "event_stream_ordering")] ) - self._receipts_id_gen = StreamIdGenerator( - db_conn, "receipts_linearized", "stream_id" - ) - self._account_data_id_gen = StreamIdGenerator( - db_conn, "account_data_max_stream_id", "stream_id" - ) self._presence_id_gen = StreamIdGenerator( db_conn, "presence_stream", "stream_id" ) @@ -123,7 +116,6 @@ class DataStore(RoomMemberStore, RoomStore, ) self._transaction_id_gen = IdGenerator(db_conn, "sent_transactions", "id") - self._state_groups_id_gen = IdGenerator(db_conn, "state_groups", "id") self._access_tokens_id_gen = IdGenerator(db_conn, "access_tokens", "id") self._event_reports_id_gen = IdGenerator(db_conn, "event_reports", "id") self._push_rule_id_gen = IdGenerator(db_conn, "push_rules", "id") @@ -135,6 +127,9 @@ class DataStore(RoomMemberStore, RoomStore, db_conn, "pushers", "id", extra_tables=[("deleted_pushers", "stream_id")], ) + self._group_updates_id_gen = StreamIdGenerator( + db_conn, "local_group_updates", "stream_id", + ) if isinstance(self.database_engine, PostgresEngine): self._cache_id_gen = StreamIdGenerator( @@ -143,27 +138,6 @@ class DataStore(RoomMemberStore, RoomStore, else: self._cache_id_gen = None - events_max = self._stream_id_gen.get_current_token() - event_cache_prefill, min_event_val = self._get_cache_dict( - db_conn, "events", - entity_column="room_id", - stream_column="stream_ordering", - max_value=events_max, - ) - self._events_stream_cache = StreamChangeCache( - "EventsRoomStreamChangeCache", min_event_val, - prefilled_cache=event_cache_prefill, - ) - - self._membership_stream_cache = StreamChangeCache( - "MembershipStreamChangeCache", events_max, - ) - - account_max = self._account_data_id_gen.get_current_token() - self._account_data_stream_cache = StreamChangeCache( - "AccountDataAndTagsChangeCache", account_max, - ) - self._presence_on_startup = self._get_active_presence(db_conn) presence_cache_prefill, min_presence_val = self._get_cache_dict( @@ -177,18 +151,6 @@ class DataStore(RoomMemberStore, RoomStore, prefilled_cache=presence_cache_prefill ) - push_rules_prefill, push_rules_id = self._get_cache_dict( - db_conn, "push_rules_stream", - entity_column="user_id", - stream_column="stream_id", - max_value=self._push_rules_stream_id_gen.get_current_token()[0], - ) - - self.push_rules_stream_cache = StreamChangeCache( - "PushRulesStreamChangeCache", push_rules_id, - prefilled_cache=push_rules_prefill, - ) - max_device_inbox_id = self._device_inbox_id_gen.get_current_token() device_inbox_prefill, min_device_inbox_id = self._get_cache_dict( db_conn, "device_inbox", @@ -223,6 +185,7 @@ class DataStore(RoomMemberStore, RoomStore, "DeviceListFederationStreamChangeCache", device_list_max, ) + events_max = self._stream_id_gen.get_current_token() curr_state_delta_prefill, min_curr_state_delta_id = self._get_cache_dict( db_conn, "current_state_delta_stream", entity_column="room_id", @@ -235,24 +198,25 @@ class DataStore(RoomMemberStore, RoomStore, prefilled_cache=curr_state_delta_prefill, ) - cur = LoggingTransaction( - db_conn.cursor(), - name="_find_stream_orderings_for_times_txn", - database_engine=self.database_engine, - after_callbacks=[], - final_callbacks=[], + _group_updates_prefill, min_group_updates_id = self._get_cache_dict( + db_conn, "local_group_updates", + entity_column="user_id", + stream_column="stream_id", + max_value=self._group_updates_id_gen.get_current_token(), + limit=1000, ) - self._find_stream_orderings_for_times_txn(cur) - cur.close() - - self.find_stream_orderings_looping_call = self._clock.looping_call( - self._find_stream_orderings_for_times, 10 * 60 * 1000 + self._group_updates_stream_cache = StreamChangeCache( + "_group_updates_stream_cache", min_group_updates_id, + prefilled_cache=_group_updates_prefill, ) self._stream_order_on_start = self.get_room_max_stream_ordering() self._min_stream_order_on_start = self.get_room_min_stream_ordering() - super(DataStore, self).__init__(hs) + # Used in _generate_user_daily_visits to keep track of progress + self._last_user_visit_update = self._get_start_of_day() + + super(DataStore, self).__init__(db_conn, hs) def take_presence_startup_info(self): active_on_startup = self._presence_on_startup @@ -281,13 +245,12 @@ class DataStore(RoomMemberStore, RoomStore, return [UserPresenceState(**row) for row in rows] - @defer.inlineCallbacks def count_daily_users(self): """ Counts the number of users who used this homeserver in the last 24 hours. """ def _count_users(txn): - yesterday = int(self._clock.time_msec()) - (1000 * 60 * 60 * 24), + yesterday = int(self._clock.time_msec()) - (1000 * 60 * 60 * 24) sql = """ SELECT COALESCE(count(*), 0) FROM ( @@ -301,8 +264,154 @@ class DataStore(RoomMemberStore, RoomStore, count, = txn.fetchone() return count - ret = yield self.runInteraction("count_users", _count_users) - defer.returnValue(ret) + return self.runInteraction("count_users", _count_users) + + def count_r30_users(self): + """ + Counts the number of 30 day retained users, defined as:- + * Users who have created their accounts more than 30 days ago + * Where last seen at most 30 days ago + * Where account creation and last_seen are > 30 days apart + + Returns counts globaly for a given user as well as breaking + by platform + """ + def _count_r30_users(txn): + thirty_days_in_secs = 86400 * 30 + now = int(self._clock.time()) + thirty_days_ago_in_secs = now - thirty_days_in_secs + + sql = """ + SELECT platform, COALESCE(count(*), 0) FROM ( + SELECT + users.name, platform, users.creation_ts * 1000, + MAX(uip.last_seen) + FROM users + INNER JOIN ( + SELECT + user_id, + last_seen, + CASE + WHEN user_agent LIKE '%%Android%%' THEN 'android' + WHEN user_agent LIKE '%%iOS%%' THEN 'ios' + WHEN user_agent LIKE '%%Electron%%' THEN 'electron' + WHEN user_agent LIKE '%%Mozilla%%' THEN 'web' + WHEN user_agent LIKE '%%Gecko%%' THEN 'web' + ELSE 'unknown' + END + AS platform + FROM user_ips + ) uip + ON users.name = uip.user_id + AND users.appservice_id is NULL + AND users.creation_ts < ? + AND uip.last_seen/1000 > ? + AND (uip.last_seen/1000) - users.creation_ts > 86400 * 30 + GROUP BY users.name, platform, users.creation_ts + ) u GROUP BY platform + """ + + results = {} + txn.execute(sql, (thirty_days_ago_in_secs, + thirty_days_ago_in_secs)) + + for row in txn: + if row[0] is 'unknown': + pass + results[row[0]] = row[1] + + sql = """ + SELECT COALESCE(count(*), 0) FROM ( + SELECT users.name, users.creation_ts * 1000, + MAX(uip.last_seen) + FROM users + INNER JOIN ( + SELECT + user_id, + last_seen + FROM user_ips + ) uip + ON users.name = uip.user_id + AND appservice_id is NULL + AND users.creation_ts < ? + AND uip.last_seen/1000 > ? + AND (uip.last_seen/1000) - users.creation_ts > 86400 * 30 + GROUP BY users.name, users.creation_ts + ) u + """ + + txn.execute(sql, (thirty_days_ago_in_secs, + thirty_days_ago_in_secs)) + + count, = txn.fetchone() + results['all'] = count + + return results + + return self.runInteraction("count_r30_users", _count_r30_users) + + def _get_start_of_day(self): + """ + Returns millisecond unixtime for start of UTC day. + """ + now = datetime.datetime.utcnow() + today_start = datetime.datetime(now.year, now.month, + now.day, tzinfo=tz.tzutc()) + return int(time.mktime(today_start.timetuple())) * 1000 + + def generate_user_daily_visits(self): + """ + Generates daily visit data for use in cohort/ retention analysis + """ + def _generate_user_daily_visits(txn): + logger.info("Calling _generate_user_daily_visits") + today_start = self._get_start_of_day() + a_day_in_milliseconds = 24 * 60 * 60 * 1000 + now = self.clock.time_msec() + + sql = """ + INSERT INTO user_daily_visits (user_id, device_id, timestamp) + SELECT u.user_id, u.device_id, ? + FROM user_ips AS u + LEFT JOIN ( + SELECT user_id, device_id, timestamp FROM user_daily_visits + WHERE timestamp = ? + ) udv + ON u.user_id = udv.user_id AND u.device_id=udv.device_id + INNER JOIN users ON users.name=u.user_id + WHERE last_seen > ? AND last_seen <= ? + AND udv.timestamp IS NULL AND users.is_guest=0 + AND users.appservice_id IS NULL + GROUP BY u.user_id, u.device_id + """ + + # This means that the day has rolled over but there could still + # be entries from the previous day. There is an edge case + # where if the user logs in at 23:59 and overwrites their + # last_seen at 00:01 then they will not be counted in the + # previous day's stats - it is important that the query is run + # often to minimise this case. + if today_start > self._last_user_visit_update: + yesterday_start = today_start - a_day_in_milliseconds + txn.execute(sql, ( + yesterday_start, yesterday_start, + self._last_user_visit_update, today_start + )) + self._last_user_visit_update = today_start + + txn.execute(sql, ( + today_start, today_start, + self._last_user_visit_update, + now + )) + # Update _last_user_visit_update to now. The reason to do this + # rather just clamping to the beginning of the day is to limit + # the size of the join - meaning that the query can be run more + # frequently + self._last_user_visit_update = now + + return self.runInteraction("generate_user_daily_visits", + _generate_user_daily_visits) def get_users(self): """Function to reterive a list of users in users table. diff --git a/synapse/storage/_base.py b/synapse/storage/_base.py index 6f54036d6..1d41d8d44 100644 --- a/synapse/storage/_base.py +++ b/synapse/storage/_base.py @@ -13,36 +13,38 @@ # See the License for the specific language governing permissions and # limitations under the License. import logging +import sys +import threading +import time -from synapse.api.errors import StoreError -from synapse.util.logcontext import LoggingContext, PreserveLoggingContext -from synapse.util.caches import CACHE_SIZE_FACTOR -from synapse.util.caches.dictionary_cache import DictionaryCache -from synapse.util.caches.descriptors import Cache -from synapse.storage.engines import PostgresEngine -import synapse.metrics +from six import iteritems, iterkeys, itervalues +from six.moves import intern, range +from prometheus_client import Histogram from twisted.internet import defer -import sys -import time -import threading - +from synapse.api.errors import StoreError +from synapse.storage.engines import PostgresEngine +from synapse.util.caches.descriptors import Cache +from synapse.util.logcontext import LoggingContext, PreserveLoggingContext logger = logging.getLogger(__name__) +try: + MAX_TXN_ID = sys.maxint - 1 +except AttributeError: + # python 3 does not have a maximum int value + MAX_TXN_ID = 2**63 - 1 + sql_logger = logging.getLogger("synapse.storage.SQL") transaction_logger = logging.getLogger("synapse.storage.txn") perf_logger = logging.getLogger("synapse.storage.TIME") +sql_scheduling_timer = Histogram("synapse_storage_schedule_time", "sec") -metrics = synapse.metrics.get_metrics_for("synapse.storage") - -sql_scheduling_timer = metrics.register_distribution("schedule_time") - -sql_query_timer = metrics.register_distribution("query_time", labels=["verb"]) -sql_txn_timer = metrics.register_distribution("transaction_time", labels=["desc"]) +sql_query_timer = Histogram("synapse_storage_query_time", "sec", ["verb"]) +sql_txn_timer = Histogram("synapse_storage_transaction_time", "sec", ["desc"]) class LoggingTransaction(object): @@ -50,16 +52,16 @@ class LoggingTransaction(object): passed to the constructor. Adds logging and metrics to the .execute() method.""" __slots__ = [ - "txn", "name", "database_engine", "after_callbacks", "final_callbacks", + "txn", "name", "database_engine", "after_callbacks", "exception_callbacks", ] def __init__(self, txn, name, database_engine, after_callbacks, - final_callbacks): + exception_callbacks): object.__setattr__(self, "txn", txn) object.__setattr__(self, "name", name) object.__setattr__(self, "database_engine", database_engine) object.__setattr__(self, "after_callbacks", after_callbacks) - object.__setattr__(self, "final_callbacks", final_callbacks) + object.__setattr__(self, "exception_callbacks", exception_callbacks) def call_after(self, callback, *args, **kwargs): """Call the given callback on the main twisted thread after the @@ -68,8 +70,8 @@ class LoggingTransaction(object): """ self.after_callbacks.append((callback, args, kwargs)) - def call_finally(self, callback, *args, **kwargs): - self.final_callbacks.append((callback, args, kwargs)) + def call_on_exception(self, callback, *args, **kwargs): + self.exception_callbacks.append((callback, args, kwargs)) def __getattr__(self, name): return getattr(self.txn, name) @@ -103,11 +105,11 @@ class LoggingTransaction(object): "[SQL values] {%s} %r", self.name, args[0] ) - except: + except Exception: # Don't let logging failures stop SQL from working pass - start = time.time() * 1000 + start = time.time() try: return func( @@ -117,9 +119,9 @@ class LoggingTransaction(object): logger.debug("[SQL FAIL] {%s} %s", self.name, e) raise finally: - msecs = (time.time() * 1000) - start - sql_logger.debug("[SQL time] {%s} %f", self.name, msecs) - sql_query_timer.inc_by(msecs, sql.split()[0]) + secs = time.time() - start + sql_logger.debug("[SQL time] {%s} %f sec", self.name, secs) + sql_query_timer.labels(sql.split()[0]).observe(secs) class PerformanceCounters(object): @@ -129,7 +131,7 @@ class PerformanceCounters(object): def update(self, key, start_time, end_time=None): if end_time is None: - end_time = time.time() * 1000 + end_time = time.time() duration = end_time - start_time count, cum_time = self.current_counters.get(key, (0, 0)) count += 1 @@ -139,7 +141,7 @@ class PerformanceCounters(object): def interval(self, interval_duration, limit=3): counters = [] - for name, (count, cum_time) in self.current_counters.iteritems(): + for name, (count, cum_time) in iteritems(self.current_counters): prev_count, prev_time = self.previous_counters.get(name, (0, 0)) counters.append(( (cum_time - prev_time) / interval_duration, @@ -162,7 +164,7 @@ class PerformanceCounters(object): class SQLBaseStore(object): _TXN_ID = 0 - def __init__(self, hs): + def __init__(self, db_conn, hs): self.hs = hs self._clock = hs.get_clock() self._db_pool = hs.get_db_pool() @@ -180,10 +182,6 @@ class SQLBaseStore(object): self._get_event_cache = Cache("*getEvent*", keylen=3, max_entries=hs.config.event_cache_size) - self._state_group_cache = DictionaryCache( - "*stateGroupCache*", 100000 * CACHE_SIZE_FACTOR - ) - self._event_fetch_lock = threading.Condition() self._event_fetch_list = [] self._event_fetch_ongoing = 0 @@ -221,14 +219,14 @@ class SQLBaseStore(object): self._clock.looping_call(loop, 10000) - def _new_transaction(self, conn, desc, after_callbacks, final_callbacks, - logging_context, func, *args, **kwargs): - start = time.time() * 1000 + def _new_transaction(self, conn, desc, after_callbacks, exception_callbacks, + func, *args, **kwargs): + start = time.time() txn_id = self._TXN_ID # We don't really need these to be unique, so lets stop it from # growing really large. - self._TXN_ID = (self._TXN_ID + 1) % (sys.maxint - 1) + self._TXN_ID = (self._TXN_ID + 1) % (MAX_TXN_ID) name = "%s-%x" % (desc, txn_id, ) @@ -242,7 +240,7 @@ class SQLBaseStore(object): txn = conn.cursor() txn = LoggingTransaction( txn, name, self.database_engine, after_callbacks, - final_callbacks, + exception_callbacks, ) r = func(txn, *args, **kwargs) conn.commit() @@ -283,73 +281,85 @@ class SQLBaseStore(object): logger.debug("[TXN FAIL] {%s} %s", name, e) raise finally: - end = time.time() * 1000 + end = time.time() duration = end - start - if logging_context is not None: - logging_context.add_database_transaction(duration) + LoggingContext.current_context().add_database_transaction(duration) - transaction_logger.debug("[TXN END] {%s} %f", name, duration) + transaction_logger.debug("[TXN END] {%s} %f sec", name, duration) self._current_txn_total_time += duration self._txn_perf_counters.update(desc, start, end) - sql_txn_timer.inc_by(duration, desc) + sql_txn_timer.labels(desc).observe(duration) @defer.inlineCallbacks def runInteraction(self, desc, func, *args, **kwargs): - """Wraps the .runInteraction() method on the underlying db_pool.""" - current_context = LoggingContext.current_context() + """Starts a transaction on the database and runs a given function - start_time = time.time() * 1000 + Arguments: + desc (str): description of the transaction, for logging and metrics + func (func): callback function, which will be called with a + database transaction (twisted.enterprise.adbapi.Transaction) as + its first argument, followed by `args` and `kwargs`. + args (list): positional args to pass to `func` + kwargs (dict): named args to pass to `func` + + Returns: + Deferred: The result of func + """ after_callbacks = [] - final_callbacks = [] - - def inner_func(conn, *args, **kwargs): - with LoggingContext("runInteraction") as context: - sql_scheduling_timer.inc_by(time.time() * 1000 - start_time) - - if self.database_engine.is_connection_closed(conn): - logger.debug("Reconnecting closed database connection") - conn.reconnect() - - current_context.copy_to(context) - return self._new_transaction( - conn, desc, after_callbacks, final_callbacks, current_context, - func, *args, **kwargs - ) + exception_callbacks = [] try: - with PreserveLoggingContext(): - result = yield self._db_pool.runWithConnection( - inner_func, *args, **kwargs - ) + result = yield self.runWithConnection( + self._new_transaction, + desc, after_callbacks, exception_callbacks, func, + *args, **kwargs + ) for after_callback, after_args, after_kwargs in after_callbacks: after_callback(*after_args, **after_kwargs) - finally: - for after_callback, after_args, after_kwargs in final_callbacks: + except: # noqa: E722, as we reraise the exception this is fine. + for after_callback, after_args, after_kwargs in exception_callbacks: after_callback(*after_args, **after_kwargs) + raise defer.returnValue(result) @defer.inlineCallbacks def runWithConnection(self, func, *args, **kwargs): - """Wraps the .runInteraction() method on the underlying db_pool.""" - current_context = LoggingContext.current_context() + """Wraps the .runWithConnection() method on the underlying db_pool. - start_time = time.time() * 1000 + Arguments: + func (func): callback function, which will be called with a + database connection (twisted.enterprise.adbapi.Connection) as + its first argument, followed by `args` and `kwargs`. + args (list): positional args to pass to `func` + kwargs (dict): named args to pass to `func` + + Returns: + Deferred: The result of func + """ + parent_context = LoggingContext.current_context() + if parent_context == LoggingContext.sentinel: + logger.warn( + "Starting db connection from sentinel context: metrics will be lost", + ) + parent_context = None + + start_time = time.time() def inner_func(conn, *args, **kwargs): - with LoggingContext("runWithConnection") as context: - sql_scheduling_timer.inc_by(time.time() * 1000 - start_time) + with LoggingContext("runWithConnection", parent_context) as context: + sched_duration_sec = time.time() - start_time + sql_scheduling_timer.observe(sched_duration_sec) + context.add_database_scheduled(sched_duration_sec) if self.database_engine.is_connection_closed(conn): logger.debug("Reconnecting closed database connection") conn.reconnect() - current_context.copy_to(context) - return func(conn, *args, **kwargs) with PreserveLoggingContext(): @@ -368,7 +378,7 @@ class SQLBaseStore(object): Returns: A list of dicts where the key is the column header. """ - col_headers = list(intern(column[0]) for column in cursor.description) + col_headers = list(intern(str(column[0])) for column in cursor.description) results = list( dict(zip(col_headers, row)) for row in cursor ) @@ -475,23 +485,53 @@ class SQLBaseStore(object): txn.executemany(sql, vals) + @defer.inlineCallbacks def _simple_upsert(self, table, keyvalues, values, insertion_values={}, desc="_simple_upsert", lock=True): """ + + `lock` should generally be set to True (the default), but can be set + to False if either of the following are true: + + * there is a UNIQUE INDEX on the key columns. In this case a conflict + will cause an IntegrityError in which case this function will retry + the update. + + * we somehow know that we are the only thread which will be updating + this table. + Args: table (str): The table to upsert into keyvalues (dict): The unique key tables and their new values values (dict): The nonunique columns and their new values - insertion_values (dict): key/values to use when inserting + insertion_values (dict): additional key/values to use only when + inserting + lock (bool): True to lock the table when doing the upsert. Returns: Deferred(bool): True if a new entry was created, False if an existing one was updated. """ - return self.runInteraction( - desc, - self._simple_upsert_txn, table, keyvalues, values, insertion_values, - lock - ) + attempts = 0 + while True: + try: + result = yield self.runInteraction( + desc, + self._simple_upsert_txn, table, keyvalues, values, insertion_values, + lock=lock + ) + defer.returnValue(result) + except self.database_engine.module.IntegrityError as e: + attempts += 1 + if attempts >= 5: + # don't retry forever, because things other than races + # can cause IntegrityErrors + raise + + # presumably we raced with another transaction: let's retry. + logger.warn( + "IntegrityError when upserting into %s; retrying: %s", + table, e + ) def _simple_upsert_txn(self, txn, table, keyvalues, values, insertion_values={}, lock=True): @@ -499,37 +539,38 @@ class SQLBaseStore(object): if lock: self.database_engine.lock_table(txn, table) - # Try to update + # First try to update. sql = "UPDATE %s SET %s WHERE %s" % ( table, ", ".join("%s = ?" % (k,) for k in values), " AND ".join("%s = ?" % (k,) for k in keyvalues) ) - sqlargs = values.values() + keyvalues.values() + sqlargs = list(values.values()) + list(keyvalues.values()) txn.execute(sql, sqlargs) - if txn.rowcount == 0: - # We didn't update and rows so insert a new one - allvalues = {} - allvalues.update(keyvalues) - allvalues.update(values) - allvalues.update(insertion_values) - - sql = "INSERT INTO %s (%s) VALUES (%s)" % ( - table, - ", ".join(k for k in allvalues), - ", ".join("?" for _ in allvalues) - ) - txn.execute(sql, allvalues.values()) - - return True - else: + if txn.rowcount > 0: + # successfully updated at least one row. return False + # We didn't update any rows so insert a new one + allvalues = {} + allvalues.update(keyvalues) + allvalues.update(values) + allvalues.update(insertion_values) + + sql = "INSERT INTO %s (%s) VALUES (%s)" % ( + table, + ", ".join(k for k in allvalues), + ", ".join("?" for _ in allvalues) + ) + txn.execute(sql, list(allvalues.values())) + # successfully inserted + return True + def _simple_select_one(self, table, keyvalues, retcols, allow_none=False, desc="_simple_select_one"): """Executes a SELECT query on the named table, which is expected to - return a single row, returning a single column from it. + return a single row, returning multiple columns from it. Args: table : string giving the table name @@ -582,20 +623,18 @@ class SQLBaseStore(object): @staticmethod def _simple_select_onecol_txn(txn, table, keyvalues, retcol): - if keyvalues: - where = "WHERE %s" % " AND ".join("%s = ?" % k for k in keyvalues.iterkeys()) - else: - where = "" - sql = ( - "SELECT %(retcol)s FROM %(table)s %(where)s" + "SELECT %(retcol)s FROM %(table)s" ) % { "retcol": retcol, "table": table, - "where": where, } - txn.execute(sql, keyvalues.values()) + if keyvalues: + sql += " WHERE %s" % " AND ".join("%s = ?" % k for k in iterkeys(keyvalues)) + txn.execute(sql, list(keyvalues.values())) + else: + txn.execute(sql) return [r[0] for r in txn] @@ -606,7 +645,7 @@ class SQLBaseStore(object): Args: table (str): table name - keyvalues (dict): column names and values to select the rows with + keyvalues (dict|None): column names and values to select the rows with retcol (str): column whos value we wish to retrieve. Returns: @@ -657,7 +696,7 @@ class SQLBaseStore(object): table, " AND ".join("%s = ?" % (k, ) for k in keyvalues) ) - txn.execute(sql, keyvalues.values()) + txn.execute(sql, list(keyvalues.values())) else: sql = "SELECT %s FROM %s" % ( ", ".join(retcols), @@ -688,9 +727,12 @@ class SQLBaseStore(object): if not iterable: defer.returnValue(results) + # iterables can not be sliced, so convert it to a list first + it_list = list(iterable) + chunks = [ - iterable[i:i + batch_size] - for i in xrange(0, len(iterable), batch_size) + it_list[i:i + batch_size] + for i in range(0, len(it_list), batch_size) ] for chunk in chunks: rows = yield self.runInteraction( @@ -730,7 +772,7 @@ class SQLBaseStore(object): ) values.extend(iterable) - for key, value in keyvalues.iteritems(): + for key, value in iteritems(keyvalues): clauses.append("%s = ?" % (key,)) values.append(value) @@ -743,6 +785,33 @@ class SQLBaseStore(object): txn.execute(sql, values) return cls.cursor_to_dict(txn) + def _simple_update(self, table, keyvalues, updatevalues, desc): + return self.runInteraction( + desc, + self._simple_update_txn, + table, keyvalues, updatevalues, + ) + + @staticmethod + def _simple_update_txn(txn, table, keyvalues, updatevalues): + if keyvalues: + where = "WHERE %s" % " AND ".join("%s = ?" % k for k in iterkeys(keyvalues)) + else: + where = "" + + update_sql = "UPDATE %s SET %s %s" % ( + table, + ", ".join("%s = ?" % (k,) for k in updatevalues), + where, + ) + + txn.execute( + update_sql, + list(updatevalues.values()) + list(keyvalues.values()) + ) + + return txn.rowcount + def _simple_update_one(self, table, keyvalues, updatevalues, desc="_simple_update_one"): """Executes an UPDATE query on the named table, setting new values for @@ -768,27 +837,13 @@ class SQLBaseStore(object): table, keyvalues, updatevalues, ) - @staticmethod - def _simple_update_one_txn(txn, table, keyvalues, updatevalues): - if keyvalues: - where = "WHERE %s" % " AND ".join("%s = ?" % k for k in keyvalues.iterkeys()) - else: - where = "" + @classmethod + def _simple_update_one_txn(cls, txn, table, keyvalues, updatevalues): + rowcount = cls._simple_update_txn(txn, table, keyvalues, updatevalues) - update_sql = "UPDATE %s SET %s %s" % ( - table, - ", ".join("%s = ?" % (k,) for k in updatevalues), - where, - ) - - txn.execute( - update_sql, - updatevalues.values() + keyvalues.values() - ) - - if txn.rowcount == 0: + if rowcount == 0: raise StoreError(404, "No row found") - if txn.rowcount > 1: + if rowcount > 1: raise StoreError(500, "More than one row matched") @staticmethod @@ -800,7 +855,7 @@ class SQLBaseStore(object): " AND ".join("%s = ?" % (k,) for k in keyvalues) ) - txn.execute(select_sql, keyvalues.values()) + txn.execute(select_sql, list(keyvalues.values())) row = txn.fetchone() if not row: @@ -838,7 +893,7 @@ class SQLBaseStore(object): " AND ".join("%s = ?" % (k, ) for k in keyvalues) ) - txn.execute(sql, keyvalues.values()) + txn.execute(sql, list(keyvalues.values())) if txn.rowcount == 0: raise StoreError(404, "No row found") if txn.rowcount > 1: @@ -856,7 +911,7 @@ class SQLBaseStore(object): " AND ".join("%s = ?" % (k, ) for k in keyvalues) ) - return txn.execute(sql, keyvalues.values()) + return txn.execute(sql, list(keyvalues.values())) def _simple_delete_many(self, table, column, iterable, keyvalues, desc): return self.runInteraction( @@ -888,7 +943,7 @@ class SQLBaseStore(object): ) values.extend(iterable) - for key, value in keyvalues.iteritems(): + for key, value in iteritems(keyvalues): clauses.append("%s = ?" % (key,)) values.append(value) @@ -928,7 +983,7 @@ class SQLBaseStore(object): txn.close() if cache: - min_val = min(cache.itervalues()) + min_val = min(itervalues(cache)) else: min_val = max_value @@ -951,7 +1006,8 @@ class SQLBaseStore(object): # __exit__ called after the transaction finishes. ctx = self._cache_id_gen.get_next() stream_id = ctx.__enter__() - txn.call_finally(ctx.__exit__, None, None, None) + txn.call_on_exception(ctx.__exit__, None, None, None) + txn.call_after(ctx.__exit__, None, None, None) txn.call_after(self.hs.get_notifier().on_new_replication_data) self._simple_insert_txn( @@ -1042,7 +1098,7 @@ class SQLBaseStore(object): " AND ".join("%s = ?" % (k,) for k in keyvalues), " ? ASC LIMIT ? OFFSET ?" ) - txn.execute(sql, keyvalues.values() + pagevalues) + txn.execute(sql, list(keyvalues.values()) + list(pagevalues)) else: sql = "SELECT %s FROM %s ORDER BY %s" % ( ", ".join(retcols), diff --git a/synapse/storage/account_data.py b/synapse/storage/account_data.py index aa84ffc2b..bbc3355c7 100644 --- a/synapse/storage/account_data.py +++ b/synapse/storage/account_data.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2014-2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -13,18 +14,46 @@ # See the License for the specific language governing permissions and # limitations under the License. -from ._base import SQLBaseStore +import abc +import logging + +from canonicaljson import json + from twisted.internet import defer -from synapse.util.caches.descriptors import cached, cachedList, cachedInlineCallbacks - -import ujson as json -import logging +from synapse.storage._base import SQLBaseStore +from synapse.storage.util.id_generators import StreamIdGenerator +from synapse.util.caches.descriptors import cached, cachedInlineCallbacks +from synapse.util.caches.stream_change_cache import StreamChangeCache logger = logging.getLogger(__name__) -class AccountDataStore(SQLBaseStore): +class AccountDataWorkerStore(SQLBaseStore): + """This is an abstract base class where subclasses must implement + `get_max_account_data_stream_id` which can be called in the initializer. + """ + + # This ABCMeta metaclass ensures that we cannot be instantiated without + # the abstract methods being implemented. + __metaclass__ = abc.ABCMeta + + def __init__(self, db_conn, hs): + account_max = self.get_max_account_data_stream_id() + self._account_data_stream_cache = StreamChangeCache( + "AccountDataAndTagsChangeCache", account_max, + ) + + super(AccountDataWorkerStore, self).__init__(db_conn, hs) + + @abc.abstractmethod + def get_max_account_data_stream_id(self): + """Get the current max stream ID for account data stream + + Returns: + int + """ + raise NotImplementedError() @cached() def get_account_data_for_user(self, user_id): @@ -63,7 +92,7 @@ class AccountDataStore(SQLBaseStore): "get_account_data_for_user", get_account_data_for_user_txn ) - @cachedInlineCallbacks(num_args=2) + @cachedInlineCallbacks(num_args=2, max_entries=5000) def get_global_account_data_by_type_for_user(self, data_type, user_id): """ Returns: @@ -85,25 +114,7 @@ class AccountDataStore(SQLBaseStore): else: defer.returnValue(None) - @cachedList(cached_method_name="get_global_account_data_by_type_for_user", - num_args=2, list_name="user_ids", inlineCallbacks=True) - def get_global_account_data_by_type_for_users(self, data_type, user_ids): - rows = yield self._simple_select_many_batch( - table="account_data", - column="user_id", - iterable=user_ids, - keyvalues={ - "account_data_type": data_type, - }, - retcols=("user_id", "content",), - desc="get_global_account_data_by_type_for_users", - ) - - defer.returnValue({ - row["user_id"]: json.loads(row["content"]) if row["content"] else None - for row in rows - }) - + @cached(num_args=2) def get_account_data_for_room(self, user_id, room_id): """Get all the client account_data for a user for a room. @@ -127,6 +138,38 @@ class AccountDataStore(SQLBaseStore): "get_account_data_for_room", get_account_data_for_room_txn ) + @cached(num_args=3, max_entries=5000) + def get_account_data_for_room_and_type(self, user_id, room_id, account_data_type): + """Get the client account_data of given type for a user for a room. + + Args: + user_id(str): The user to get the account_data for. + room_id(str): The room to get the account_data for. + account_data_type (str): The account data type to get. + Returns: + A deferred of the room account_data for that type, or None if + there isn't any set. + """ + def get_account_data_for_room_and_type_txn(txn): + content_json = self._simple_select_one_onecol_txn( + txn, + table="room_account_data", + keyvalues={ + "user_id": user_id, + "room_id": room_id, + "account_data_type": account_data_type, + }, + retcol="content", + allow_none=True + ) + + return json.loads(content_json) if content_json else None + + return self.runInteraction( + "get_account_data_for_room_and_type", + get_account_data_for_room_and_type_txn, + ) + def get_all_updated_account_data(self, last_global_id, last_room_id, current_id, limit): """Get all the client account_data that has changed on the server @@ -209,6 +252,36 @@ class AccountDataStore(SQLBaseStore): "get_updated_account_data_for_user", get_updated_account_data_for_user_txn ) + @cachedInlineCallbacks(num_args=2, cache_context=True, max_entries=5000) + def is_ignored_by(self, ignored_user_id, ignorer_user_id, cache_context): + ignored_account_data = yield self.get_global_account_data_by_type_for_user( + "m.ignored_user_list", ignorer_user_id, + on_invalidate=cache_context.invalidate, + ) + if not ignored_account_data: + defer.returnValue(False) + + defer.returnValue( + ignored_user_id in ignored_account_data.get("ignored_users", {}) + ) + + +class AccountDataStore(AccountDataWorkerStore): + def __init__(self, db_conn, hs): + self._account_data_id_gen = StreamIdGenerator( + db_conn, "account_data_max_stream_id", "stream_id" + ) + + super(AccountDataStore, self).__init__(db_conn, hs) + + def get_max_account_data_stream_id(self): + """Get the current max stream id for the private user data stream + + Returns: + A deferred int. + """ + return self._account_data_id_gen.get_current_token() + @defer.inlineCallbacks def add_account_data_to_room(self, user_id, room_id, account_data_type, content): """Add some account_data to a room for a user. @@ -222,9 +295,12 @@ class AccountDataStore(SQLBaseStore): """ content_json = json.dumps(content) - def add_account_data_txn(txn, next_id): - self._simple_upsert_txn( - txn, + with self._account_data_id_gen.get_next() as next_id: + # no need to lock here as room_account_data has a unique constraint + # on (user_id, room_id, account_data_type) so _simple_upsert will + # retry if there is a conflict. + yield self._simple_upsert( + desc="add_room_account_data", table="room_account_data", keyvalues={ "user_id": user_id, @@ -234,18 +310,23 @@ class AccountDataStore(SQLBaseStore): values={ "stream_id": next_id, "content": content_json, - } + }, + lock=False, ) - txn.call_after( - self._account_data_stream_cache.entity_has_changed, - user_id, next_id, - ) - txn.call_after(self.get_account_data_for_user.invalidate, (user_id,)) - self._update_max_stream_id(txn, next_id) - with self._account_data_id_gen.get_next() as next_id: - yield self.runInteraction( - "add_room_account_data", add_account_data_txn, next_id + # it's theoretically possible for the above to succeed and the + # below to fail - in which case we might reuse a stream id on + # restart, and the above update might not get propagated. That + # doesn't sound any worse than the whole update getting lost, + # which is what would happen if we combined the two into one + # transaction. + yield self._update_max_stream_id(next_id) + + self._account_data_stream_cache.entity_has_changed(user_id, next_id) + self.get_account_data_for_user.invalidate((user_id,)) + self.get_account_data_for_room.invalidate((user_id, room_id,)) + self.get_account_data_for_room_and_type.prefill( + (user_id, room_id, account_data_type,), content, ) result = self._account_data_id_gen.get_current_token() @@ -263,9 +344,12 @@ class AccountDataStore(SQLBaseStore): """ content_json = json.dumps(content) - def add_account_data_txn(txn, next_id): - self._simple_upsert_txn( - txn, + with self._account_data_id_gen.get_next() as next_id: + # no need to lock here as account_data has a unique constraint on + # (user_id, account_data_type) so _simple_upsert will retry if + # there is a conflict. + yield self._simple_upsert( + desc="add_user_account_data", table="account_data", keyvalues={ "user_id": user_id, @@ -274,37 +358,43 @@ class AccountDataStore(SQLBaseStore): values={ "stream_id": next_id, "content": content_json, - } + }, + lock=False, ) - txn.call_after( - self._account_data_stream_cache.entity_has_changed, + + # it's theoretically possible for the above to succeed and the + # below to fail - in which case we might reuse a stream id on + # restart, and the above update might not get propagated. That + # doesn't sound any worse than the whole update getting lost, + # which is what would happen if we combined the two into one + # transaction. + yield self._update_max_stream_id(next_id) + + self._account_data_stream_cache.entity_has_changed( user_id, next_id, ) - txn.call_after(self.get_account_data_for_user.invalidate, (user_id,)) - txn.call_after( - self.get_global_account_data_by_type_for_user.invalidate, + self.get_account_data_for_user.invalidate((user_id,)) + self.get_global_account_data_by_type_for_user.invalidate( (account_data_type, user_id,) ) - self._update_max_stream_id(txn, next_id) - - with self._account_data_id_gen.get_next() as next_id: - yield self.runInteraction( - "add_user_account_data", add_account_data_txn, next_id - ) result = self._account_data_id_gen.get_current_token() defer.returnValue(result) - def _update_max_stream_id(self, txn, next_id): + def _update_max_stream_id(self, next_id): """Update the max stream_id Args: - txn: The database cursor next_id(int): The the revision to advance to. """ - update_max_id_sql = ( - "UPDATE account_data_max_stream_id" - " SET stream_id = ?" - " WHERE stream_id < ?" + def _update(txn): + update_max_id_sql = ( + "UPDATE account_data_max_stream_id" + " SET stream_id = ?" + " WHERE stream_id < ?" + ) + txn.execute(update_max_id_sql, (next_id, next_id)) + return self.runInteraction( + "update_account_data_max_stream_id", + _update, ) - txn.execute(update_max_id_sql, (next_id, next_id)) diff --git a/synapse/storage/appservice.py b/synapse/storage/appservice.py index c63935cb0..9f12b360b 100644 --- a/synapse/storage/appservice.py +++ b/synapse/storage/appservice.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2015, 2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -14,15 +15,16 @@ # limitations under the License. import logging import re -import simplejson as json + +from canonicaljson import json + from twisted.internet import defer -from synapse.api.constants import Membership from synapse.appservice import AppServiceTransaction from synapse.config.appservice import load_appservices -from synapse.storage.roommember import RoomsForUser -from ._base import SQLBaseStore +from synapse.storage.events import EventsWorkerStore +from ._base import SQLBaseStore logger = logging.getLogger(__name__) @@ -46,17 +48,16 @@ def _make_exclusive_regex(services_cache): return exclusive_user_regex -class ApplicationServiceStore(SQLBaseStore): - - def __init__(self, hs): - super(ApplicationServiceStore, self).__init__(hs) - self.hostname = hs.hostname +class ApplicationServiceWorkerStore(SQLBaseStore): + def __init__(self, db_conn, hs): self.services_cache = load_appservices( hs.hostname, hs.config.app_service_config_files ) self.exclusive_user_regex = _make_exclusive_regex(self.services_cache) + super(ApplicationServiceWorkerStore, self).__init__(db_conn, hs) + def get_app_services(self): return self.services_cache @@ -99,83 +100,30 @@ class ApplicationServiceStore(SQLBaseStore): return service return None - def get_app_service_rooms(self, service): - """Get a list of RoomsForUser for this application service. - - Application services may be "interested" in lots of rooms depending on - the room ID, the room aliases, or the members in the room. This function - takes all of these into account and returns a list of RoomsForUser which - represent the entire list of room IDs that this application service - wants to know about. + def get_app_service_by_id(self, as_id): + """Get the application service with the given appservice ID. Args: - service: The application service to get a room list for. + as_id (str): The application service ID. Returns: - A list of RoomsForUser. + synapse.appservice.ApplicationService or None. """ - return self.runInteraction( - "get_app_service_rooms", - self._get_app_service_rooms_txn, - service, - ) - - def _get_app_service_rooms_txn(self, txn, service): - # get all rooms matching the room ID regex. - room_entries = self._simple_select_list_txn( - txn=txn, table="rooms", keyvalues=None, retcols=["room_id"] - ) - matching_room_list = set([ - r["room_id"] for r in room_entries if - service.is_interested_in_room(r["room_id"]) - ]) - - # resolve room IDs for matching room alias regex. - room_alias_mappings = self._simple_select_list_txn( - txn=txn, table="room_aliases", keyvalues=None, - retcols=["room_id", "room_alias"] - ) - matching_room_list |= set([ - r["room_id"] for r in room_alias_mappings if - service.is_interested_in_alias(r["room_alias"]) - ]) - - # get all rooms for every user for this AS. This is scoped to users on - # this HS only. - user_list = self._simple_select_list_txn( - txn=txn, table="users", keyvalues=None, retcols=["name"] - ) - user_list = [ - u["name"] for u in user_list if - service.is_interested_in_user(u["name"]) - ] - rooms_for_user_matching_user_id = set() # RoomsForUser list - for user_id in user_list: - # FIXME: This assumes this store is linked with RoomMemberStore :( - rooms_for_user = self._get_rooms_for_user_where_membership_is_txn( - txn=txn, - user_id=user_id, - membership_list=[Membership.JOIN] - ) - rooms_for_user_matching_user_id |= set(rooms_for_user) - - # make RoomsForUser tuples for room ids and aliases which are not in the - # main rooms_for_user_list - e.g. they are rooms which do not have AS - # registered users in it. - known_room_ids = [r.room_id for r in rooms_for_user_matching_user_id] - missing_rooms_for_user = [ - RoomsForUser(r, service.sender, "join") for r in - matching_room_list if r not in known_room_ids - ] - rooms_for_user_matching_user_id |= set(missing_rooms_for_user) - - return rooms_for_user_matching_user_id + for service in self.services_cache: + if service.id == as_id: + return service + return None -class ApplicationServiceTransactionStore(SQLBaseStore): +class ApplicationServiceStore(ApplicationServiceWorkerStore): + # This is currently empty due to there not being any AS storage functions + # that can't be run on the workers. Since this may change in future, and + # to keep consistency with the other stores, we keep this empty class for + # now. + pass - def __init__(self, hs): - super(ApplicationServiceTransactionStore, self).__init__(hs) +class ApplicationServiceTransactionWorkerStore(ApplicationServiceWorkerStore, + EventsWorkerStore): @defer.inlineCallbacks def get_appservices_by_state(self, state): """Get a list of application services based on their state. @@ -420,3 +368,11 @@ class ApplicationServiceTransactionStore(SQLBaseStore): events = yield self._get_events(event_ids) defer.returnValue((upper_bound, events)) + + +class ApplicationServiceTransactionStore(ApplicationServiceTransactionWorkerStore): + # This is currently empty due to there not being any AS storage functions + # that can't be run on the workers. Since this may change in future, and + # to keep consistency with the other stores, we keep this empty class for + # now. + pass diff --git a/synapse/storage/background_updates.py b/synapse/storage/background_updates.py index 7157fb1df..5fe1ca2de 100644 --- a/synapse/storage/background_updates.py +++ b/synapse/storage/background_updates.py @@ -12,15 +12,17 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. -import synapse.util.async -from ._base import SQLBaseStore -from . import engines +import logging + +from canonicaljson import json from twisted.internet import defer -import ujson as json -import logging +from synapse.metrics.background_process_metrics import run_as_background_process + +from . import engines +from ._base import SQLBaseStore logger = logging.getLogger(__name__) @@ -80,25 +82,30 @@ class BackgroundUpdateStore(SQLBaseStore): BACKGROUND_UPDATE_INTERVAL_MS = 1000 BACKGROUND_UPDATE_DURATION_MS = 100 - def __init__(self, hs): - super(BackgroundUpdateStore, self).__init__(hs) + def __init__(self, db_conn, hs): + super(BackgroundUpdateStore, self).__init__(db_conn, hs) self._background_update_performance = {} self._background_update_queue = [] self._background_update_handlers = {} + self._all_done = False + + def start_doing_background_updates(self): + run_as_background_process( + "background_updates", self._run_background_updates, + ) @defer.inlineCallbacks - def start_doing_background_updates(self): + def _run_background_updates(self): logger.info("Starting background schema updates") - while True: - yield synapse.util.async.sleep( + yield self.hs.get_clock().sleep( self.BACKGROUND_UPDATE_INTERVAL_MS / 1000.) try: result = yield self.do_next_background_update( self.BACKGROUND_UPDATE_DURATION_MS ) - except: + except Exception: logger.exception("Error doing update") else: if result is None: @@ -106,8 +113,40 @@ class BackgroundUpdateStore(SQLBaseStore): "No more background updates to do." " Unscheduling background update task." ) + self._all_done = True defer.returnValue(None) + @defer.inlineCallbacks + def has_completed_background_updates(self): + """Check if all the background updates have completed + + Returns: + Deferred[bool]: True if all background updates have completed + """ + # if we've previously determined that there is nothing left to do, that + # is easy + if self._all_done: + defer.returnValue(True) + + # obviously, if we have things in our queue, we're not done. + if self._background_update_queue: + defer.returnValue(False) + + # otherwise, check if there are updates to be run. This is important, + # as we may be running on a worker which doesn't perform the bg updates + # itself, but still wants to wait for them to happen. + updates = yield self._simple_select_onecol( + "background_updates", + keyvalues=None, + retcol="1", + desc="check_background_updates", + ) + if not updates: + self._all_done = True + defer.returnValue(True) + + defer.returnValue(False) + @defer.inlineCallbacks def do_next_background_update(self, desired_duration_ms): """Does some amount of work on the next queued background update @@ -209,6 +248,25 @@ class BackgroundUpdateStore(SQLBaseStore): """ self._background_update_handlers[update_name] = update_handler + def register_noop_background_update(self, update_name): + """Register a noop handler for a background update. + + This is useful when we previously did a background update, but no + longer wish to do the update. In this case the background update should + be removed from the schema delta files, but there may still be some + users who have the background update queued, so this method should + also be called to clear the update. + + Args: + update_name (str): Name of update + """ + @defer.inlineCallbacks + def noop_update(progress, batch_size): + yield self._end_background_update(update_name) + defer.returnValue(1) + + self.register_background_update_handler(update_name, noop_update) + def register_background_index_update(self, update_name, index_name, table, columns, where_clause=None, unique=False, @@ -269,7 +327,7 @@ class BackgroundUpdateStore(SQLBaseStore): # Sqlite doesn't support concurrent creation of indexes. # # We don't use partial indices on SQLite as it wasn't introduced - # until 3.8, and wheezy has 3.7 + # until 3.8, and wheezy and CentOS 7 have 3.7 # # We assume that sqlite doesn't give us invalid indices; however # we may still end up with the index existing but the diff --git a/synapse/storage/client_ips.py b/synapse/storage/client_ips.py index fc468ea18..77ae10da3 100644 --- a/synapse/storage/client_ips.py +++ b/synapse/storage/client_ips.py @@ -15,13 +15,15 @@ import logging -from twisted.internet import defer, reactor +from six import iteritems -from ._base import Cache -from . import background_updates +from twisted.internet import defer +from synapse.metrics.background_process_metrics import run_as_background_process from synapse.util.caches import CACHE_SIZE_FACTOR +from . import background_updates +from ._base import Cache logger = logging.getLogger(__name__) @@ -32,14 +34,14 @@ LAST_SEEN_GRANULARITY = 120 * 1000 class ClientIpStore(background_updates.BackgroundUpdateStore): - def __init__(self, hs): + def __init__(self, db_conn, hs): self.client_ip_last_seen = Cache( name="client_ip_last_seen", keylen=4, max_entries=50000 * CACHE_SIZE_FACTOR, ) - super(ClientIpStore, self).__init__(hs) + super(ClientIpStore, self).__init__(db_conn, hs) self.register_background_index_update( "user_ips_device_index", @@ -48,17 +50,35 @@ class ClientIpStore(background_updates.BackgroundUpdateStore): columns=["user_id", "device_id", "last_seen"], ) + self.register_background_index_update( + "user_ips_last_seen_index", + index_name="user_ips_last_seen", + table="user_ips", + columns=["user_id", "last_seen"], + ) + + self.register_background_index_update( + "user_ips_last_seen_only_index", + index_name="user_ips_last_seen_only", + table="user_ips", + columns=["last_seen"], + ) + # (user_id, access_token, ip) -> (user_agent, device_id, last_seen) self._batch_row_update = {} self._client_ip_looper = self._clock.looping_call( self._update_client_ips_batch, 5 * 1000 ) - reactor.addSystemEventTrigger("before", "shutdown", self._update_client_ips_batch) + self.hs.get_reactor().addSystemEventTrigger( + "before", "shutdown", self._update_client_ips_batch + ) - def insert_client_ip(self, user, access_token, ip, user_agent, device_id): - now = int(self._clock.time_msec()) - key = (user.to_string(), access_token, ip) + def insert_client_ip(self, user_id, access_token, ip, user_agent, device_id, + now=None): + if not now: + now = int(self._clock.time_msec()) + key = (user_id, access_token, ip) try: last_seen = self.client_ip_last_seen.get(key) @@ -74,16 +94,22 @@ class ClientIpStore(background_updates.BackgroundUpdateStore): self._batch_row_update[key] = (user_agent, device_id, now) def _update_client_ips_batch(self): - to_update = self._batch_row_update - self._batch_row_update = {} - return self.runInteraction( - "_update_client_ips_batch", self._update_client_ips_batch_txn, to_update + def update(): + to_update = self._batch_row_update + self._batch_row_update = {} + return self.runInteraction( + "_update_client_ips_batch", self._update_client_ips_batch_txn, + to_update, + ) + + run_as_background_process( + "update_client_ips", update, ) def _update_client_ips_batch_txn(self, txn, to_update): self.database_engine.lock_table(txn, "user_ips") - for entry in to_update.iteritems(): + for entry in iteritems(to_update): (user_id, access_token, ip), (user_agent, device_id, last_seen) = entry self._simple_upsert_txn( @@ -215,5 +241,5 @@ class ClientIpStore(background_updates.BackgroundUpdateStore): "user_agent": user_agent, "last_seen": last_seen, } - for (access_token, ip), (user_agent, last_seen) in results.iteritems() + for (access_token, ip), (user_agent, last_seen) in iteritems(results) )) diff --git a/synapse/storage/deviceinbox.py b/synapse/storage/deviceinbox.py index 0b62b493d..73646da02 100644 --- a/synapse/storage/deviceinbox.py +++ b/synapse/storage/deviceinbox.py @@ -14,14 +14,14 @@ # limitations under the License. import logging -import ujson + +from canonicaljson import json from twisted.internet import defer -from .background_updates import BackgroundUpdateStore - from synapse.util.caches.expiringcache import ExpiringCache +from .background_updates import BackgroundUpdateStore logger = logging.getLogger(__name__) @@ -29,8 +29,8 @@ logger = logging.getLogger(__name__) class DeviceInboxStore(BackgroundUpdateStore): DEVICE_INBOX_STREAM_ID = "device_inbox_stream_drop" - def __init__(self, hs): - super(DeviceInboxStore, self).__init__(hs) + def __init__(self, db_conn, hs): + super(DeviceInboxStore, self).__init__(db_conn, hs) self.register_background_index_update( "device_inbox_stream_index", @@ -85,7 +85,7 @@ class DeviceInboxStore(BackgroundUpdateStore): ) rows = [] for destination, edu in remote_messages_by_destination.items(): - edu_json = ujson.dumps(edu) + edu_json = json.dumps(edu) rows.append((destination, stream_id, now_ms, edu_json)) txn.executemany(sql, rows) @@ -177,7 +177,7 @@ class DeviceInboxStore(BackgroundUpdateStore): " WHERE user_id = ?" ) txn.execute(sql, (user_id,)) - message_json = ujson.dumps(messages_by_device["*"]) + message_json = json.dumps(messages_by_device["*"]) for row in txn: # Add the message for all devices for this user on this # server. @@ -199,7 +199,7 @@ class DeviceInboxStore(BackgroundUpdateStore): # Only insert into the local inbox if the device exists on # this server device = row[0] - message_json = ujson.dumps(messages_by_device[device]) + message_json = json.dumps(messages_by_device[device]) messages_json_for_user[device] = message_json if messages_json_for_user: @@ -253,7 +253,7 @@ class DeviceInboxStore(BackgroundUpdateStore): messages = [] for row in txn: stream_pos = row[0] - messages.append(ujson.loads(row[1])) + messages.append(json.loads(row[1])) if len(messages) < limit: stream_pos = current_stream_id return (messages, stream_pos) @@ -389,7 +389,7 @@ class DeviceInboxStore(BackgroundUpdateStore): messages = [] for row in txn: stream_pos = row[0] - messages.append(ujson.loads(row[1])) + messages.append(json.loads(row[1])) if len(messages) < limit: stream_pos = current_stream_id return (messages, stream_pos) diff --git a/synapse/storage/devices.py b/synapse/storage/devices.py index bb27fd1f7..cc3cdf2eb 100644 --- a/synapse/storage/devices.py +++ b/synapse/storage/devices.py @@ -13,21 +13,24 @@ # See the License for the specific language governing permissions and # limitations under the License. import logging -import ujson as json + +from six import iteritems, itervalues + +from canonicaljson import json from twisted.internet import defer from synapse.api.errors import StoreError -from ._base import SQLBaseStore, Cache -from synapse.util.caches.descriptors import cached, cachedList, cachedInlineCallbacks +from synapse.util.caches.descriptors import cached, cachedInlineCallbacks, cachedList +from ._base import Cache, SQLBaseStore logger = logging.getLogger(__name__) class DeviceStore(SQLBaseStore): - def __init__(self, hs): - super(DeviceStore, self).__init__(hs) + def __init__(self, db_conn, hs): + super(DeviceStore, self).__init__(db_conn, hs) # Map of (user_id, device_id) -> bool. If there is an entry that implies # the device exists. @@ -245,17 +248,31 @@ class DeviceStore(SQLBaseStore): def _update_remote_device_list_cache_entry_txn(self, txn, user_id, device_id, content, stream_id): - self._simple_upsert_txn( - txn, - table="device_lists_remote_cache", - keyvalues={ - "user_id": user_id, - "device_id": device_id, - }, - values={ - "content": json.dumps(content), - } - ) + if content.get("deleted"): + self._simple_delete_txn( + txn, + table="device_lists_remote_cache", + keyvalues={ + "user_id": user_id, + "device_id": device_id, + }, + ) + + txn.call_after( + self.device_id_exists_cache.invalidate, (user_id, device_id,) + ) + else: + self._simple_upsert_txn( + txn, + table="device_lists_remote_cache", + keyvalues={ + "user_id": user_id, + "device_id": device_id, + }, + values={ + "content": json.dumps(content), + } + ) txn.call_after(self._get_cached_user_device.invalidate, (user_id, device_id,)) txn.call_after(self._get_cached_devices_for_user.invalidate, (user_id,)) @@ -360,10 +377,10 @@ class DeviceStore(SQLBaseStore): return (now_stream_id, []) if len(query_map) >= 20: - now_stream_id = max(stream_id for stream_id in query_map.itervalues()) + now_stream_id = max(stream_id for stream_id in itervalues(query_map)) devices = self._get_e2e_device_keys_txn( - txn, query_map.keys(), include_all_devices=True + txn, query_map.keys(), include_all_devices=True, include_deleted_devices=True ) prev_sent_id_sql = """ @@ -373,13 +390,13 @@ class DeviceStore(SQLBaseStore): """ results = [] - for user_id, user_devices in devices.iteritems(): + for user_id, user_devices in iteritems(devices): # The prev_id for the first row is always the last row before # `from_stream_id` txn.execute(prev_sent_id_sql, (destination, user_id, from_stream_id)) rows = txn.fetchall() prev_id = rows[0][0] - for device_id, device in user_devices.iteritems(): + for device_id, device in iteritems(user_devices): stream_id = query_map[(user_id, device_id)] result = { "user_id": user_id, @@ -390,12 +407,15 @@ class DeviceStore(SQLBaseStore): prev_id = stream_id - key_json = device.get("key_json", None) - if key_json: - result["keys"] = json.loads(key_json) - device_display_name = device.get("device_display_name", None) - if device_display_name: - result["device_display_name"] = device_display_name + if device is not None: + key_json = device.get("key_json", None) + if key_json: + result["keys"] = json.loads(key_json) + device_display_name = device.get("device_display_name", None) + if device_display_name: + result["device_display_name"] = device_display_name + else: + result["deleted"] = True results.append(result) @@ -483,7 +503,7 @@ class DeviceStore(SQLBaseStore): if devices: user_devices = devices[user_id] results = [] - for device_id, device in user_devices.iteritems(): + for device_id, device in iteritems(user_devices): result = { "device_id": device_id, } diff --git a/synapse/storage/directory.py b/synapse/storage/directory.py index 79e7c540a..808194236 100644 --- a/synapse/storage/directory.py +++ b/synapse/storage/directory.py @@ -13,15 +13,14 @@ # See the License for the specific language governing permissions and # limitations under the License. -from ._base import SQLBaseStore -from synapse.util.caches.descriptors import cached - -from synapse.api.errors import SynapseError +from collections import namedtuple from twisted.internet import defer -from collections import namedtuple +from synapse.api.errors import SynapseError +from synapse.util.caches.descriptors import cached +from ._base import SQLBaseStore RoomAliasMapping = namedtuple( "RoomAliasMapping", @@ -29,8 +28,7 @@ RoomAliasMapping = namedtuple( ) -class DirectoryStore(SQLBaseStore): - +class DirectoryWorkerStore(SQLBaseStore): @defer.inlineCallbacks def get_association_from_room_alias(self, room_alias): """ Get's the room_id and server list for a given room_alias @@ -69,6 +67,28 @@ class DirectoryStore(SQLBaseStore): RoomAliasMapping(room_id, room_alias.to_string(), servers) ) + def get_room_alias_creator(self, room_alias): + return self._simple_select_one_onecol( + table="room_aliases", + keyvalues={ + "room_alias": room_alias, + }, + retcol="creator", + desc="get_room_alias_creator", + allow_none=True + ) + + @cached(max_entries=5000) + def get_aliases_for_room(self, room_id): + return self._simple_select_onecol( + "room_aliases", + {"room_id": room_id}, + "room_alias", + desc="get_aliases_for_room", + ) + + +class DirectoryStore(DirectoryWorkerStore): @defer.inlineCallbacks def create_room_alias_association(self, room_alias, room_id, servers, creator=None): """ Creates an associatin between a room alias and room_id/servers @@ -116,17 +136,6 @@ class DirectoryStore(SQLBaseStore): ) defer.returnValue(ret) - def get_room_alias_creator(self, room_alias): - return self._simple_select_one_onecol( - table="room_aliases", - keyvalues={ - "room_alias": room_alias, - }, - retcol="creator", - desc="get_room_alias_creator", - allow_none=True - ) - @defer.inlineCallbacks def delete_room_alias(self, room_alias): room_id = yield self.runInteraction( @@ -135,7 +144,6 @@ class DirectoryStore(SQLBaseStore): room_alias, ) - self.get_aliases_for_room.invalidate((room_id,)) defer.returnValue(room_id) def _delete_room_alias_txn(self, txn, room_alias): @@ -160,17 +168,12 @@ class DirectoryStore(SQLBaseStore): (room_alias.to_string(),) ) - return room_id - - @cached(max_entries=5000) - def get_aliases_for_room(self, room_id): - return self._simple_select_onecol( - "room_aliases", - {"room_id": room_id}, - "room_alias", - desc="get_aliases_for_room", + self._invalidate_cache_and_stream( + txn, self.get_aliases_for_room, (room_id,) ) + return room_id + def update_aliases_for_room(self, old_room_id, new_room_id, creator): def _update_aliases_for_room_txn(txn): sql = "UPDATE room_aliases SET room_id = ?, creator = ? WHERE room_id = ?" diff --git a/synapse/storage/end_to_end_keys.py b/synapse/storage/end_to_end_keys.py index 2cebb203c..523b4360c 100644 --- a/synapse/storage/end_to_end_keys.py +++ b/synapse/storage/end_to_end_keys.py @@ -12,13 +12,14 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. +from six import iteritems + +from canonicaljson import encode_canonical_json, json + from twisted.internet import defer from synapse.util.caches.descriptors import cached -from canonicaljson import encode_canonical_json -import ujson as json - from ._base import SQLBaseStore @@ -63,12 +64,18 @@ class EndToEndKeyStore(SQLBaseStore): ) @defer.inlineCallbacks - def get_e2e_device_keys(self, query_list, include_all_devices=False): + def get_e2e_device_keys( + self, query_list, include_all_devices=False, + include_deleted_devices=False, + ): """Fetch a list of device keys. Args: query_list(list): List of pairs of user_ids and device_ids. include_all_devices (bool): whether to include entries for devices that don't have device keys + include_deleted_devices (bool): whether to include null entries for + devices which no longer exist (but were in the query_list). + This option only takes effect if include_all_devices is true. Returns: Dict mapping from user-id to dict mapping from device_id to dict containing "key_json", "device_display_name". @@ -78,19 +85,28 @@ class EndToEndKeyStore(SQLBaseStore): results = yield self.runInteraction( "get_e2e_device_keys", self._get_e2e_device_keys_txn, - query_list, include_all_devices, + query_list, include_all_devices, include_deleted_devices, ) - for user_id, device_keys in results.iteritems(): - for device_id, device_info in device_keys.iteritems(): + for user_id, device_keys in iteritems(results): + for device_id, device_info in iteritems(device_keys): device_info["keys"] = json.loads(device_info.pop("key_json")) defer.returnValue(results) - def _get_e2e_device_keys_txn(self, txn, query_list, include_all_devices): + def _get_e2e_device_keys_txn( + self, txn, query_list, include_all_devices=False, + include_deleted_devices=False, + ): query_clauses = [] query_params = [] + if include_all_devices is False: + include_deleted_devices = False + + if include_deleted_devices: + deleted_devices = set(query_list) + for (user_id, device_id) in query_list: query_clause = "user_id = ?" query_params.append(user_id) @@ -118,8 +134,14 @@ class EndToEndKeyStore(SQLBaseStore): result = {} for row in rows: + if include_deleted_devices: + deleted_devices.remove((row["user_id"], row["device_id"])) result.setdefault(row["user_id"], {})[row["device_id"]] = row + if include_deleted_devices: + for user_id, device_id in deleted_devices: + result.setdefault(user_id, {})[device_id] = None + return result @defer.inlineCallbacks diff --git a/synapse/storage/engines/__init__.py b/synapse/storage/engines/__init__.py index 338b49561..e2f9de845 100644 --- a/synapse/storage/engines/__init__.py +++ b/synapse/storage/engines/__init__.py @@ -13,13 +13,13 @@ # See the License for the specific language governing permissions and # limitations under the License. +import importlib +import platform + from ._base import IncorrectDatabaseSetup from .postgres import PostgresEngine from .sqlite3 import Sqlite3Engine -import importlib - - SUPPORTED_MODULE = { "sqlite3": Sqlite3Engine, "psycopg2": PostgresEngine, @@ -31,6 +31,10 @@ def create_engine(database_config): engine_class = SUPPORTED_MODULE.get(name, None) if engine_class: + # pypy requires psycopg2cffi rather than psycopg2 + if (name == "psycopg2" and + platform.python_implementation() == "PyPy"): + name = "psycopg2cffi" module = importlib.import_module(name) return engine_class(module, database_config) diff --git a/synapse/storage/engines/postgres.py b/synapse/storage/engines/postgres.py index a6ae79dfa..8a0386c1a 100644 --- a/synapse/storage/engines/postgres.py +++ b/synapse/storage/engines/postgres.py @@ -62,3 +62,9 @@ class PostgresEngine(object): def lock_table(self, txn, table): txn.execute("LOCK TABLE %s in EXCLUSIVE MODE" % (table,)) + + def get_next_state_group_id(self, txn): + """Returns an int that can be used as a new state_group ID + """ + txn.execute("SELECT nextval('state_group_id_seq')") + return txn.fetchone()[0] diff --git a/synapse/storage/engines/sqlite3.py b/synapse/storage/engines/sqlite3.py index 755c9a1f0..19949fc47 100644 --- a/synapse/storage/engines/sqlite3.py +++ b/synapse/storage/engines/sqlite3.py @@ -13,9 +13,10 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.storage.prepare_database import prepare_database - import struct +import threading + +from synapse.storage.prepare_database import prepare_database class Sqlite3Engine(object): @@ -24,6 +25,11 @@ class Sqlite3Engine(object): def __init__(self, database_module, database_config): self.module = database_module + # The current max state_group, or None if we haven't looked + # in the DB yet. + self._current_state_group_id = None + self._current_state_group_id_lock = threading.Lock() + def check_database(self, txn): pass @@ -43,6 +49,19 @@ class Sqlite3Engine(object): def lock_table(self, txn, table): return + def get_next_state_group_id(self, txn): + """Returns an int that can be used as a new state_group ID + """ + # We do application locking here since if we're using sqlite then + # we are a single process synapse. + with self._current_state_group_id_lock: + if self._current_state_group_id is None: + txn.execute("SELECT COALESCE(max(id), 0) FROM state_groups") + self._current_state_group_id = txn.fetchone()[0] + + self._current_state_group_id += 1 + return self._current_state_group_id + # Following functions taken from: https://github.com/coleifer/peewee diff --git a/synapse/storage/event_federation.py b/synapse/storage/event_federation.py index e8133de2f..8d366d1b9 100644 --- a/synapse/storage/event_federation.py +++ b/synapse/storage/event_federation.py @@ -12,45 +12,27 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. +import logging +import random + +from six.moves import range +from six.moves.queue import Empty, PriorityQueue + +from unpaddedbase64 import encode_base64 from twisted.internet import defer -from ._base import SQLBaseStore from synapse.api.errors import StoreError +from synapse.storage._base import SQLBaseStore +from synapse.storage.events import EventsWorkerStore +from synapse.storage.signatures import SignatureWorkerStore from synapse.util.caches.descriptors import cached -from unpaddedbase64 import encode_base64 - -import logging -from Queue import PriorityQueue, Empty - logger = logging.getLogger(__name__) -class EventFederationStore(SQLBaseStore): - """ Responsible for storing and serving up the various graphs associated - with an event. Including the main event graph and the auth chains for an - event. - - Also has methods for getting the front (latest) and back (oldest) edges - of the event graphs. These are used to generate the parents for new events - and backfilling from another server respectively. - """ - - EVENT_AUTH_STATE_ONLY = "event_auth_state_only" - - def __init__(self, hs): - super(EventFederationStore, self).__init__(hs) - - self.register_background_update_handler( - self.EVENT_AUTH_STATE_ONLY, - self._background_delete_non_state_event_auth, - ) - - hs.get_clock().looping_call( - self._delete_old_forward_extrem_cache, 60 * 60 * 1000 - ) - +class EventFederationWorkerStore(EventsWorkerStore, SignatureWorkerStore, + SQLBaseStore): def get_auth_chain(self, event_ids, include_given=False): """Get auth events for given event_ids. The events *must* be state events. @@ -97,7 +79,7 @@ class EventFederationStore(SQLBaseStore): front_list = list(front) chunks = [ front_list[x:x + 100] - for x in xrange(0, len(front), 100) + for x in range(0, len(front), 100) ] for chunk in chunks: txn.execute( @@ -152,7 +134,47 @@ class EventFederationStore(SQLBaseStore): retcol="event_id", ) + @defer.inlineCallbacks + def get_prev_events_for_room(self, room_id): + """ + Gets a subset of the current forward extremities in the given room. + + Limits the result to 10 extremities, so that we can avoid creating + events which refer to hundreds of prev_events. + + Args: + room_id (str): room_id + + Returns: + Deferred[list[(str, dict[str, str], int)]] + for each event, a tuple of (event_id, hashes, depth) + where *hashes* is a map from algorithm to hash. + """ + res = yield self.get_latest_event_ids_and_hashes_in_room(room_id) + if len(res) > 10: + # Sort by reverse depth, so we point to the most recent. + res.sort(key=lambda a: -a[2]) + + # we use half of the limit for the actual most recent events, and + # the other half to randomly point to some of the older events, to + # make sure that we don't completely ignore the older events. + res = res[0:5] + random.sample(res[5:], 5) + + defer.returnValue(res) + def get_latest_event_ids_and_hashes_in_room(self, room_id): + """ + Gets the current forward extremities in the given room + + Args: + room_id (str): room_id + + Returns: + Deferred[list[(str, dict[str, str], int)]] + for each event, a tuple of (event_id, hashes, depth) + where *hashes* is a map from algorithm to hash. + """ + return self.runInteraction( "get_latest_event_ids_and_hashes_in_room", self._get_latest_event_ids_and_hashes_in_room, @@ -201,22 +223,6 @@ class EventFederationStore(SQLBaseStore): room_id, ) - @defer.inlineCallbacks - def get_max_depth_of_events(self, event_ids): - sql = ( - "SELECT MAX(depth) FROM events WHERE event_id IN (%s)" - ) % (",".join(["?"] * len(event_ids)),) - - rows = yield self._execute( - "get_max_depth_of_events", None, - sql, *event_ids - ) - - if rows: - defer.returnValue(rows[0][0]) - else: - defer.returnValue(1) - def _get_min_depth_interaction(self, txn, room_id): min_depth = self._simple_select_one_onecol_txn( txn, @@ -228,88 +234,6 @@ class EventFederationStore(SQLBaseStore): return int(min_depth) if min_depth is not None else None - def _update_min_depth_for_room_txn(self, txn, room_id, depth): - min_depth = self._get_min_depth_interaction(txn, room_id) - - if min_depth and depth >= min_depth: - return - - self._simple_upsert_txn( - txn, - table="room_depth", - keyvalues={ - "room_id": room_id, - }, - values={ - "min_depth": depth, - }, - ) - - def _handle_mult_prev_events(self, txn, events): - """ - For the given event, update the event edges table and forward and - backward extremities tables. - """ - self._simple_insert_many_txn( - txn, - table="event_edges", - values=[ - { - "event_id": ev.event_id, - "prev_event_id": e_id, - "room_id": ev.room_id, - "is_state": False, - } - for ev in events - for e_id, _ in ev.prev_events - ], - ) - - self._update_backward_extremeties(txn, events) - - def _update_backward_extremeties(self, txn, events): - """Updates the event_backward_extremities tables based on the new/updated - events being persisted. - - This is called for new events *and* for events that were outliers, but - are now being persisted as non-outliers. - - Forward extremities are handled when we first start persisting the events. - """ - events_by_room = {} - for ev in events: - events_by_room.setdefault(ev.room_id, []).append(ev) - - query = ( - "INSERT INTO event_backward_extremities (event_id, room_id)" - " SELECT ?, ? WHERE NOT EXISTS (" - " SELECT 1 FROM event_backward_extremities" - " WHERE event_id = ? AND room_id = ?" - " )" - " AND NOT EXISTS (" - " SELECT 1 FROM events WHERE event_id = ? AND room_id = ? " - " AND outlier = ?" - " )" - ) - - txn.executemany(query, [ - (e_id, ev.room_id, e_id, ev.room_id, e_id, ev.room_id, False) - for ev in events for e_id, _ in ev.prev_events - if not ev.internal_metadata.is_outlier() - ]) - - query = ( - "DELETE FROM event_backward_extremities" - " WHERE event_id = ? AND room_id = ?" - ) - txn.executemany( - query, - [ - (ev.event_id, ev.room_id) for ev in events - if not ev.internal_metadata.is_outlier() - ] - ) - def get_forward_extremeties_for_room(self, room_id, stream_ordering): """For a given room_id and stream_ordering, return the forward extremeties of the room at that point in "time". @@ -371,28 +295,6 @@ class EventFederationStore(SQLBaseStore): get_forward_extremeties_for_room_txn ) - def _delete_old_forward_extrem_cache(self): - def _delete_old_forward_extrem_cache_txn(txn): - # Delete entries older than a month, while making sure we don't delete - # the only entries for a room. - sql = (""" - DELETE FROM stream_ordering_to_exterm - WHERE - room_id IN ( - SELECT room_id - FROM stream_ordering_to_exterm - WHERE stream_ordering > ? - ) AND stream_ordering < ? - """) - txn.execute( - sql, - (self.stream_ordering_month_ago, self.stream_ordering_month_ago,) - ) - return self.runInteraction( - "_delete_old_forward_extrem_cache", - _delete_old_forward_extrem_cache_txn - ) - def get_backfill_events(self, room_id, event_list, limit): """Get a list of Events for a given topic that occurred before (and including) the events in event_list. Return a list of max size `limit` @@ -522,6 +424,135 @@ class EventFederationStore(SQLBaseStore): return event_results + +class EventFederationStore(EventFederationWorkerStore): + """ Responsible for storing and serving up the various graphs associated + with an event. Including the main event graph and the auth chains for an + event. + + Also has methods for getting the front (latest) and back (oldest) edges + of the event graphs. These are used to generate the parents for new events + and backfilling from another server respectively. + """ + + EVENT_AUTH_STATE_ONLY = "event_auth_state_only" + + def __init__(self, db_conn, hs): + super(EventFederationStore, self).__init__(db_conn, hs) + + self.register_background_update_handler( + self.EVENT_AUTH_STATE_ONLY, + self._background_delete_non_state_event_auth, + ) + + hs.get_clock().looping_call( + self._delete_old_forward_extrem_cache, 60 * 60 * 1000 + ) + + def _update_min_depth_for_room_txn(self, txn, room_id, depth): + min_depth = self._get_min_depth_interaction(txn, room_id) + + if min_depth and depth >= min_depth: + return + + self._simple_upsert_txn( + txn, + table="room_depth", + keyvalues={ + "room_id": room_id, + }, + values={ + "min_depth": depth, + }, + ) + + def _handle_mult_prev_events(self, txn, events): + """ + For the given event, update the event edges table and forward and + backward extremities tables. + """ + self._simple_insert_many_txn( + txn, + table="event_edges", + values=[ + { + "event_id": ev.event_id, + "prev_event_id": e_id, + "room_id": ev.room_id, + "is_state": False, + } + for ev in events + for e_id, _ in ev.prev_events + ], + ) + + self._update_backward_extremeties(txn, events) + + def _update_backward_extremeties(self, txn, events): + """Updates the event_backward_extremities tables based on the new/updated + events being persisted. + + This is called for new events *and* for events that were outliers, but + are now being persisted as non-outliers. + + Forward extremities are handled when we first start persisting the events. + """ + events_by_room = {} + for ev in events: + events_by_room.setdefault(ev.room_id, []).append(ev) + + query = ( + "INSERT INTO event_backward_extremities (event_id, room_id)" + " SELECT ?, ? WHERE NOT EXISTS (" + " SELECT 1 FROM event_backward_extremities" + " WHERE event_id = ? AND room_id = ?" + " )" + " AND NOT EXISTS (" + " SELECT 1 FROM events WHERE event_id = ? AND room_id = ? " + " AND outlier = ?" + " )" + ) + + txn.executemany(query, [ + (e_id, ev.room_id, e_id, ev.room_id, e_id, ev.room_id, False) + for ev in events for e_id, _ in ev.prev_events + if not ev.internal_metadata.is_outlier() + ]) + + query = ( + "DELETE FROM event_backward_extremities" + " WHERE event_id = ? AND room_id = ?" + ) + txn.executemany( + query, + [ + (ev.event_id, ev.room_id) for ev in events + if not ev.internal_metadata.is_outlier() + ] + ) + + def _delete_old_forward_extrem_cache(self): + def _delete_old_forward_extrem_cache_txn(txn): + # Delete entries older than a month, while making sure we don't delete + # the only entries for a room. + sql = (""" + DELETE FROM stream_ordering_to_exterm + WHERE + room_id IN ( + SELECT room_id + FROM stream_ordering_to_exterm + WHERE stream_ordering > ? + ) AND stream_ordering < ? + """) + txn.execute( + sql, + (self.stream_ordering_month_ago, self.stream_ordering_month_ago,) + ) + return self.runInteraction( + "_delete_old_forward_extrem_cache", + _delete_old_forward_extrem_cache_txn + ) + def clean_room_for_join(self, room_id): return self.runInteraction( "clean_room_for_join", diff --git a/synapse/storage/event_push_actions.py b/synapse/storage/event_push_actions.py index d6d8723b4..29b511ae5 100644 --- a/synapse/storage/event_push_actions.py +++ b/synapse/storage/event_push_actions.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2015 OpenMarket Ltd +# Copyright 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -13,15 +14,16 @@ # See the License for the specific language governing permissions and # limitations under the License. -from ._base import SQLBaseStore -from twisted.internet import defer -from synapse.util.async import sleep -from synapse.util.caches.descriptors import cachedInlineCallbacks -from synapse.types import RoomStreamToken -from .stream import lower_bound - import logging -import ujson as json + +from six import iteritems + +from canonicaljson import json + +from twisted.internet import defer + +from synapse.storage._base import LoggingTransaction, SQLBaseStore +from synapse.util.caches.descriptors import cachedInlineCallbacks logger = logging.getLogger(__name__) @@ -62,59 +64,29 @@ def _deserialize_action(actions, is_highlight): return DEFAULT_NOTIF_ACTION -class EventPushActionsStore(SQLBaseStore): - EPA_HIGHLIGHT_INDEX = "epa_highlight_index" +class EventPushActionsWorkerStore(SQLBaseStore): + def __init__(self, db_conn, hs): + super(EventPushActionsWorkerStore, self).__init__(db_conn, hs) - def __init__(self, hs): - super(EventPushActionsStore, self).__init__(hs) + # These get correctly set by _find_stream_orderings_for_times_txn + self.stream_ordering_month_ago = None + self.stream_ordering_day_ago = None - self.register_background_index_update( - self.EPA_HIGHLIGHT_INDEX, - index_name="event_push_actions_u_highlight", - table="event_push_actions", - columns=["user_id", "stream_ordering"], + cur = LoggingTransaction( + db_conn.cursor(), + name="_find_stream_orderings_for_times_txn", + database_engine=self.database_engine, + after_callbacks=[], + exception_callbacks=[], ) + self._find_stream_orderings_for_times_txn(cur) + cur.close() - self.register_background_index_update( - "event_push_actions_highlights_index", - index_name="event_push_actions_highlights_index", - table="event_push_actions", - columns=["user_id", "room_id", "topological_ordering", "stream_ordering"], - where_clause="highlight=1" + self.find_stream_orderings_looping_call = self._clock.looping_call( + self._find_stream_orderings_for_times, 10 * 60 * 1000 ) - - self._doing_notif_rotation = False - self._rotate_notif_loop = self._clock.looping_call( - self._rotate_notifs, 30 * 60 * 1000 - ) - - def _set_push_actions_for_event_and_users_txn(self, txn, event, tuples): - """ - Args: - event: the event set actions for - tuples: list of tuples of (user_id, actions) - """ - values = [] - for uid, actions in tuples: - is_highlight = 1 if _action_has_highlight(actions) else 0 - - values.append({ - 'room_id': event.room_id, - 'event_id': event.event_id, - 'user_id': uid, - 'actions': _serialize_action(actions, is_highlight), - 'stream_ordering': event.internal_metadata.stream_ordering, - 'topological_ordering': event.depth, - 'notif': 1, - 'highlight': is_highlight, - }) - - for uid, __ in tuples: - txn.call_after( - self.get_unread_event_push_actions_by_room_for_user.invalidate_many, - (event.room_id, uid) - ) - self._simple_insert_many_txn(txn, "event_push_actions", values) + self._rotate_delay = 3 + self._rotate_count = 10000 @cachedInlineCallbacks(num_args=3, tree=True, max_entries=5000) def get_unread_event_push_actions_by_room_for_user( @@ -130,7 +102,7 @@ class EventPushActionsStore(SQLBaseStore): def _get_unread_counts_by_receipt_txn(self, txn, room_id, user_id, last_read_event_id): sql = ( - "SELECT stream_ordering, topological_ordering" + "SELECT stream_ordering" " FROM events" " WHERE room_id = ? AND event_id = ?" ) @@ -142,17 +114,12 @@ class EventPushActionsStore(SQLBaseStore): return {"notify_count": 0, "highlight_count": 0} stream_ordering = results[0][0] - topological_ordering = results[0][1] return self._get_unread_counts_by_pos_txn( - txn, room_id, user_id, topological_ordering, stream_ordering + txn, room_id, user_id, stream_ordering ) - def _get_unread_counts_by_pos_txn(self, txn, room_id, user_id, topological_ordering, - stream_ordering): - token = RoomStreamToken( - topological_ordering, stream_ordering - ) + def _get_unread_counts_by_pos_txn(self, txn, room_id, user_id, stream_ordering): # First get number of notifications. # We don't need to put a notif=1 clause as all rows always have @@ -163,10 +130,10 @@ class EventPushActionsStore(SQLBaseStore): " WHERE" " user_id = ?" " AND room_id = ?" - " AND %s" - ) % (lower_bound(token, self.database_engine, inclusive=False),) + " AND stream_ordering > ?" + ) - txn.execute(sql, (user_id, room_id)) + txn.execute(sql, (user_id, room_id, stream_ordering)) row = txn.fetchone() notify_count = row[0] if row else 0 @@ -186,10 +153,10 @@ class EventPushActionsStore(SQLBaseStore): " highlight = 1" " AND user_id = ?" " AND room_id = ?" - " AND %s" - ) % (lower_bound(token, self.database_engine, inclusive=False),) + " AND stream_ordering > ?" + ) - txn.execute(sql, (user_id, room_id)) + txn.execute(sql, (user_id, room_id, stream_ordering)) row = txn.fetchone() highlight_count = row[0] if row else 0 @@ -240,7 +207,6 @@ class EventPushActionsStore(SQLBaseStore): " ep.highlight " " FROM (" " SELECT room_id," - " MAX(topological_ordering) as topological_ordering," " MAX(stream_ordering) as stream_ordering" " FROM events" " INNER JOIN receipts_linearized USING (room_id, event_id)" @@ -250,13 +216,7 @@ class EventPushActionsStore(SQLBaseStore): " event_push_actions AS ep" " WHERE" " ep.room_id = rl.room_id" - " AND (" - " ep.topological_ordering > rl.topological_ordering" - " OR (" - " ep.topological_ordering = rl.topological_ordering" - " AND ep.stream_ordering > rl.stream_ordering" - " )" - " )" + " AND ep.stream_ordering > rl.stream_ordering" " AND ep.user_id = ?" " AND ep.stream_ordering > ?" " AND ep.stream_ordering <= ?" @@ -349,7 +309,6 @@ class EventPushActionsStore(SQLBaseStore): " ep.highlight, e.received_ts" " FROM (" " SELECT room_id," - " MAX(topological_ordering) as topological_ordering," " MAX(stream_ordering) as stream_ordering" " FROM events" " INNER JOIN receipts_linearized USING (room_id, event_id)" @@ -360,13 +319,7 @@ class EventPushActionsStore(SQLBaseStore): " INNER JOIN events AS e USING (room_id, event_id)" " WHERE" " ep.room_id = rl.room_id" - " AND (" - " ep.topological_ordering > rl.topological_ordering" - " OR (" - " ep.topological_ordering = rl.topological_ordering" - " AND ep.stream_ordering > rl.stream_ordering" - " )" - " )" + " AND ep.stream_ordering > rl.stream_ordering" " AND ep.user_id = ?" " AND ep.stream_ordering > ?" " AND ep.stream_ordering <= ?" @@ -432,6 +385,290 @@ class EventPushActionsStore(SQLBaseStore): # Now return the first `limit` defer.returnValue(notifs[:limit]) + def add_push_actions_to_staging(self, event_id, user_id_actions): + """Add the push actions for the event to the push action staging area. + + Args: + event_id (str) + user_id_actions (dict[str, list[dict|str])]): A dictionary mapping + user_id to list of push actions, where an action can either be + a string or dict. + + Returns: + Deferred + """ + + if not user_id_actions: + return + + # This is a helper function for generating the necessary tuple that + # can be used to inert into the `event_push_actions_staging` table. + def _gen_entry(user_id, actions): + is_highlight = 1 if _action_has_highlight(actions) else 0 + return ( + event_id, # event_id column + user_id, # user_id column + _serialize_action(actions, is_highlight), # actions column + 1, # notif column + is_highlight, # highlight column + ) + + def _add_push_actions_to_staging_txn(txn): + # We don't use _simple_insert_many here to avoid the overhead + # of generating lists of dicts. + + sql = """ + INSERT INTO event_push_actions_staging + (event_id, user_id, actions, notif, highlight) + VALUES (?, ?, ?, ?, ?) + """ + + txn.executemany(sql, ( + _gen_entry(user_id, actions) + for user_id, actions in iteritems(user_id_actions) + )) + + return self.runInteraction( + "add_push_actions_to_staging", _add_push_actions_to_staging_txn + ) + + @defer.inlineCallbacks + def remove_push_actions_from_staging(self, event_id): + """Called if we failed to persist the event to ensure that stale push + actions don't build up in the DB + + Args: + event_id (str) + """ + + try: + res = yield self._simple_delete( + table="event_push_actions_staging", + keyvalues={ + "event_id": event_id, + }, + desc="remove_push_actions_from_staging", + ) + defer.returnValue(res) + except Exception: + # this method is called from an exception handler, so propagating + # another exception here really isn't helpful - there's nothing + # the caller can do about it. Just log the exception and move on. + logger.exception( + "Error removing push actions after event persistence failure", + ) + + @defer.inlineCallbacks + def _find_stream_orderings_for_times(self): + yield self.runInteraction( + "_find_stream_orderings_for_times", + self._find_stream_orderings_for_times_txn + ) + + def _find_stream_orderings_for_times_txn(self, txn): + logger.info("Searching for stream ordering 1 month ago") + self.stream_ordering_month_ago = self._find_first_stream_ordering_after_ts_txn( + txn, self._clock.time_msec() - 30 * 24 * 60 * 60 * 1000 + ) + logger.info( + "Found stream ordering 1 month ago: it's %d", + self.stream_ordering_month_ago + ) + logger.info("Searching for stream ordering 1 day ago") + self.stream_ordering_day_ago = self._find_first_stream_ordering_after_ts_txn( + txn, self._clock.time_msec() - 24 * 60 * 60 * 1000 + ) + logger.info( + "Found stream ordering 1 day ago: it's %d", + self.stream_ordering_day_ago + ) + + def find_first_stream_ordering_after_ts(self, ts): + """Gets the stream ordering corresponding to a given timestamp. + + Specifically, finds the stream_ordering of the first event that was + received on or after the timestamp. This is done by a binary search on + the events table, since there is no index on received_ts, so is + relatively slow. + + Args: + ts (int): timestamp in millis + + Returns: + Deferred[int]: stream ordering of the first event received on/after + the timestamp + """ + return self.runInteraction( + "_find_first_stream_ordering_after_ts_txn", + self._find_first_stream_ordering_after_ts_txn, + ts, + ) + + @staticmethod + def _find_first_stream_ordering_after_ts_txn(txn, ts): + """ + Find the stream_ordering of the first event that was received on or + after a given timestamp. This is relatively slow as there is no index + on received_ts but we can then use this to delete push actions before + this. + + received_ts must necessarily be in the same order as stream_ordering + and stream_ordering is indexed, so we manually binary search using + stream_ordering + + Args: + txn (twisted.enterprise.adbapi.Transaction): + ts (int): timestamp to search for + + Returns: + int: stream ordering + """ + txn.execute("SELECT MAX(stream_ordering) FROM events") + max_stream_ordering = txn.fetchone()[0] + + if max_stream_ordering is None: + return 0 + + # We want the first stream_ordering in which received_ts is greater + # than or equal to ts. Call this point X. + # + # We maintain the invariants: + # + # range_start <= X <= range_end + # + range_start = 0 + range_end = max_stream_ordering + 1 + + # Given a stream_ordering, look up the timestamp at that + # stream_ordering. + # + # The array may be sparse (we may be missing some stream_orderings). + # We treat the gaps as the same as having the same value as the + # preceding entry, because we will pick the lowest stream_ordering + # which satisfies our requirement of received_ts >= ts. + # + # For example, if our array of events indexed by stream_ordering is + # [10, , 20], we should treat this as being equivalent to + # [10, 10, 20]. + # + sql = ( + "SELECT received_ts FROM events" + " WHERE stream_ordering <= ?" + " ORDER BY stream_ordering DESC" + " LIMIT 1" + ) + + while range_end - range_start > 0: + middle = (range_end + range_start) // 2 + txn.execute(sql, (middle,)) + row = txn.fetchone() + if row is None: + # no rows with stream_ordering<=middle + range_start = middle + 1 + continue + + middle_ts = row[0] + if ts > middle_ts: + # we got a timestamp lower than the one we were looking for. + # definitely need to look higher: X > middle. + range_start = middle + 1 + else: + # we got a timestamp higher than (or the same as) the one we + # were looking for. We aren't yet sure about the point we + # looked up, but we can be sure that X <= middle. + range_end = middle + + return range_end + + +class EventPushActionsStore(EventPushActionsWorkerStore): + EPA_HIGHLIGHT_INDEX = "epa_highlight_index" + + def __init__(self, db_conn, hs): + super(EventPushActionsStore, self).__init__(db_conn, hs) + + self.register_background_index_update( + self.EPA_HIGHLIGHT_INDEX, + index_name="event_push_actions_u_highlight", + table="event_push_actions", + columns=["user_id", "stream_ordering"], + ) + + self.register_background_index_update( + "event_push_actions_highlights_index", + index_name="event_push_actions_highlights_index", + table="event_push_actions", + columns=["user_id", "room_id", "topological_ordering", "stream_ordering"], + where_clause="highlight=1" + ) + + self._doing_notif_rotation = False + self._rotate_notif_loop = self._clock.looping_call( + self._rotate_notifs, 30 * 60 * 1000 + ) + + def _set_push_actions_for_event_and_users_txn(self, txn, events_and_contexts, + all_events_and_contexts): + """Handles moving push actions from staging table to main + event_push_actions table for all events in `events_and_contexts`. + + Also ensures that all events in `all_events_and_contexts` are removed + from the push action staging area. + + Args: + events_and_contexts (list[(EventBase, EventContext)]): events + we are persisting + all_events_and_contexts (list[(EventBase, EventContext)]): all + events that we were going to persist. This includes events + we've already persisted, etc, that wouldn't appear in + events_and_context. + """ + + sql = """ + INSERT INTO event_push_actions ( + room_id, event_id, user_id, actions, stream_ordering, + topological_ordering, notif, highlight + ) + SELECT ?, event_id, user_id, actions, ?, ?, notif, highlight + FROM event_push_actions_staging + WHERE event_id = ? + """ + + if events_and_contexts: + txn.executemany(sql, ( + ( + event.room_id, event.internal_metadata.stream_ordering, + event.depth, event.event_id, + ) + for event, _ in events_and_contexts + )) + + for event, _ in events_and_contexts: + user_ids = self._simple_select_onecol_txn( + txn, + table="event_push_actions_staging", + keyvalues={ + "event_id": event.event_id, + }, + retcol="user_id", + ) + + for uid in user_ids: + txn.call_after( + self.get_unread_event_push_actions_by_room_for_user.invalidate_many, + (event.room_id, uid,) + ) + + # Now we delete the staging area for *all* events that were being + # persisted. + txn.executemany( + "DELETE FROM event_push_actions_staging WHERE event_id = ?", + ( + (event.event_id,) + for event, _ in all_events_and_contexts + ) + ) + @defer.inlineCallbacks def get_push_actions_for_user(self, user_id, before=None, limit=50, only_highlight=False): @@ -509,10 +746,10 @@ class EventPushActionsStore(SQLBaseStore): ) def _remove_old_push_actions_before_txn(self, txn, room_id, user_id, - topological_ordering, stream_ordering): + stream_ordering): """ Purges old push actions for a user and room before a given - topological_ordering. + stream_ordering. We however keep a months worth of highlighted notifications, so that users can still get a list of recent highlights. @@ -521,7 +758,7 @@ class EventPushActionsStore(SQLBaseStore): txn: The transcation room_id: Room ID to delete from user_id: user ID to delete for - topological_ordering: The lowest topological ordering which will + stream_ordering: The lowest stream ordering which will not be deleted. """ txn.call_after( @@ -540,9 +777,9 @@ class EventPushActionsStore(SQLBaseStore): txn.execute( "DELETE FROM event_push_actions " " WHERE user_id = ? AND room_id = ? AND " - " topological_ordering <= ?" + " stream_ordering <= ?" " AND ((stream_ordering < ? AND highlight = 1) or highlight = 0)", - (user_id, room_id, topological_ordering, self.stream_ordering_month_ago) + (user_id, room_id, stream_ordering, self.stream_ordering_month_ago) ) txn.execute(""" @@ -550,69 +787,6 @@ class EventPushActionsStore(SQLBaseStore): WHERE room_id = ? AND user_id = ? AND stream_ordering <= ? """, (room_id, user_id, stream_ordering)) - @defer.inlineCallbacks - def _find_stream_orderings_for_times(self): - yield self.runInteraction( - "_find_stream_orderings_for_times", - self._find_stream_orderings_for_times_txn - ) - - def _find_stream_orderings_for_times_txn(self, txn): - logger.info("Searching for stream ordering 1 month ago") - self.stream_ordering_month_ago = self._find_first_stream_ordering_after_ts_txn( - txn, self._clock.time_msec() - 30 * 24 * 60 * 60 * 1000 - ) - logger.info( - "Found stream ordering 1 month ago: it's %d", - self.stream_ordering_month_ago - ) - logger.info("Searching for stream ordering 1 day ago") - self.stream_ordering_day_ago = self._find_first_stream_ordering_after_ts_txn( - txn, self._clock.time_msec() - 24 * 60 * 60 * 1000 - ) - logger.info( - "Found stream ordering 1 day ago: it's %d", - self.stream_ordering_day_ago - ) - - def _find_first_stream_ordering_after_ts_txn(self, txn, ts): - """ - Find the stream_ordering of the first event that was received after - a given timestamp. This is relatively slow as there is no index on - received_ts but we can then use this to delete push actions before - this. - - received_ts must necessarily be in the same order as stream_ordering - and stream_ordering is indexed, so we manually binary search using - stream_ordering - """ - txn.execute("SELECT MAX(stream_ordering) FROM events") - max_stream_ordering = txn.fetchone()[0] - - if max_stream_ordering is None: - return 0 - - range_start = 0 - range_end = max_stream_ordering - - sql = ( - "SELECT received_ts FROM events" - " WHERE stream_ordering > ?" - " ORDER BY stream_ordering" - " LIMIT 1" - ) - - while range_end - range_start > 1: - middle = int((range_end + range_start) / 2) - txn.execute(sql, (middle,)) - middle_ts = txn.fetchone()[0] - if ts > middle_ts: - range_start = middle - else: - range_end = middle - - return range_end - @defer.inlineCallbacks def _rotate_notifs(self): if self._doing_notif_rotation or self.stream_ordering_day_ago is None: @@ -629,7 +803,7 @@ class EventPushActionsStore(SQLBaseStore): ) if caught_up: break - yield sleep(5) + yield self.hs.get_clock().sleep(self._rotate_delay) finally: self._doing_notif_rotation = False @@ -650,8 +824,8 @@ class EventPushActionsStore(SQLBaseStore): txn.execute(""" SELECT stream_ordering FROM event_push_actions WHERE stream_ordering > ? - ORDER BY stream_ordering ASC LIMIT 1 OFFSET 50000 - """, (old_rotate_stream_ordering,)) + ORDER BY stream_ordering ASC LIMIT 1 OFFSET ? + """, (old_rotate_stream_ordering, self._rotate_count)) stream_row = txn.fetchone() if stream_row: offset_stream_ordering, = stream_row diff --git a/synapse/storage/events.py b/synapse/storage/events.py index 7002b3752..906a40503 100644 --- a/synapse/storage/events.py +++ b/synapse/storage/events.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2014-2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -12,64 +13,59 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. -from ._base import SQLBaseStore -from twisted.internet import defer, reactor - -from synapse.events import FrozenEvent, USE_FROZEN_DICTS -from synapse.events.utils import prune_event - -from synapse.util.async import ObservableDeferred -from synapse.util.logcontext import ( - preserve_fn, PreserveLoggingContext, preserve_context_over_deferred -) -from synapse.util.logutils import log_function -from synapse.util.metrics import Measure -from synapse.api.constants import EventTypes -from synapse.api.errors import SynapseError -from synapse.state import resolve_events -from synapse.util.caches.descriptors import cached -from synapse.types import get_domain_from_id - -from canonicaljson import encode_canonical_json -from collections import deque, namedtuple, OrderedDict +import itertools +import logging +from collections import OrderedDict, deque, namedtuple from functools import wraps +from six import iteritems +from six.moves import range + +from canonicaljson import json +from prometheus_client import Counter + +from twisted.internet import defer + import synapse.metrics - -import logging -import ujson as json - +from synapse.api.constants import EventTypes +from synapse.api.errors import SynapseError # these are only included to make the type annotations work -from synapse.events import EventBase # noqa: F401 -from synapse.events.snapshot import EventContext # noqa: F401 +from synapse.events import EventBase # noqa: F401 +from synapse.events.snapshot import EventContext # noqa: F401 +from synapse.metrics.background_process_metrics import run_as_background_process +from synapse.storage.events_worker import EventsWorkerStore +from synapse.types import RoomStreamToken, get_domain_from_id +from synapse.util.async import ObservableDeferred +from synapse.util.caches.descriptors import cached, cachedInlineCallbacks +from synapse.util.frozenutils import frozendict_json_encoder +from synapse.util.logcontext import PreserveLoggingContext, make_deferred_yieldable +from synapse.util.logutils import log_function +from synapse.util.metrics import Measure logger = logging.getLogger(__name__) +persist_event_counter = Counter("synapse_storage_events_persisted_events", "") +event_counter = Counter("synapse_storage_events_persisted_events_sep", "", + ["type", "origin_type", "origin_entity"]) -metrics = synapse.metrics.get_metrics_for(__name__) -persist_event_counter = metrics.register_counter("persisted_events") -event_counter = metrics.register_counter( - "persisted_events_sep", labels=["type", "origin_type", "origin_entity"] -) +# The number of times we are recalculating the current state +state_delta_counter = Counter("synapse_storage_events_state_delta", "") + +# The number of times we are recalculating state when there is only a +# single forward extremity +state_delta_single_event_counter = Counter( + "synapse_storage_events_state_delta_single_event", "") + +# The number of times we are reculating state when we could have resonably +# calculated the delta when we calculated the state for an event we were +# persisting. +state_delta_reuse_delta_counter = Counter( + "synapse_storage_events_state_delta_reuse_delta", "") def encode_json(json_object): - if USE_FROZEN_DICTS: - # ujson doesn't like frozen_dicts - return encode_canonical_json(json_object) - else: - return json.dumps(json_object, ensure_ascii=False) - - -# These values are used in the `enqueus_event` and `_do_fetch` methods to -# control how we batch/bulk fetch events from the database. -# The values are plucked out of thing air to make initial sync run faster -# on jki.re -# TODO: Make these configurable. -EVENT_QUEUE_THREADS = 3 # Max number of threads that will fetch events -EVENT_QUEUE_ITERATIONS = 3 # No. times we block waiting for requests for events -EVENT_QUEUE_TIMEOUT_S = 0.1 # Timeout when waiting for requests for events + return frozendict_json_encoder.encode(json_object) class _EventPeristenceQueue(object): @@ -88,19 +84,29 @@ class _EventPeristenceQueue(object): def add_to_queue(self, room_id, events_and_contexts, backfilled): """Add events to the queue, with the given persist_event options. + NB: due to the normal usage pattern of this method, it does *not* + follow the synapse logcontext rules, and leaves the logcontext in + place whether or not the returned deferred is ready. + Args: room_id (str): events_and_contexts (list[(EventBase, EventContext)]): backfilled (bool): + + Returns: + defer.Deferred: a deferred which will resolve once the events are + persisted. Runs its callbacks *without* a logcontext. """ queue = self._event_persist_queues.setdefault(room_id, deque()) if queue: + # if the last item in the queue has the same `backfilled` setting, + # we can just add these new events to that item. end_item = queue[-1] if end_item.backfilled == backfilled: end_item.events_and_contexts.extend(events_and_contexts) return end_item.deferred.observe() - deferred = ObservableDeferred(defer.Deferred()) + deferred = ObservableDeferred(defer.Deferred(), consumeErrors=True) queue.append(self._EventPersistQueueItem( events_and_contexts=events_and_contexts, @@ -113,11 +119,11 @@ class _EventPeristenceQueue(object): def handle_queue(self, room_id, per_item_callback): """Attempts to handle the queue for a room if not already being handled. - The given callback will be invoked with for each item in the queue,1 + The given callback will be invoked with for each item in the queue, of type _EventPersistQueueItem. The per_item_callback will continuously be called with new items, unless the queue becomnes empty. The return value of the function will be given to the deferreds waiting on the item, - exceptions will be passed to the deferres as well. + exceptions will be passed to the deferreds as well. This function should therefore be called whenever anything is added to the queue. @@ -136,18 +142,23 @@ class _EventPeristenceQueue(object): try: queue = self._get_drainining_queue(room_id) for item in queue: + # handle_queue_loop runs in the sentinel logcontext, so + # there is no need to preserve_fn when running the + # callbacks on the deferred. try: ret = yield per_item_callback(item) - item.deferred.callback(ret) - except Exception as e: - item.deferred.errback(e) + with PreserveLoggingContext(): + item.deferred.callback(ret) + except Exception: + item.deferred.errback() finally: queue = self._event_persist_queues.pop(room_id, None) if queue: self._event_persist_queues[room_id] = queue self._currently_persisting_rooms.discard(room_id) - preserve_fn(handle_queue_loop)() + # set handle_queue_loop off in the background + run_as_background_process("persist_events", handle_queue_loop) def _get_drainining_queue(self, room_id): queue = self._event_persist_queues.setdefault(room_id, deque()) @@ -183,13 +194,12 @@ def _retry_on_integrity_error(func): return f -class EventsStore(SQLBaseStore): +class EventsStore(EventsWorkerStore): EVENT_ORIGIN_SERVER_TS_NAME = "event_origin_server_ts" EVENT_FIELDS_SENDER_URL_UPDATE_NAME = "event_fields_sender_url" - def __init__(self, hs): - super(EventsStore, self).__init__(hs) - self._clock = hs.get_clock() + def __init__(self, db_conn, hs): + super(EventsStore, self).__init__(db_conn, hs) self.register_background_update_handler( self.EVENT_ORIGIN_SERVER_TS_NAME, self._background_reindex_origin_server_ts ) @@ -220,6 +230,8 @@ class EventsStore(SQLBaseStore): self._event_persist_queue = _EventPeristenceQueue() + self._state_resolution_handler = hs.get_state_resolution_handler() + def persist_events(self, events_and_contexts, backfilled=False): """ Write events to the database @@ -232,8 +244,8 @@ class EventsStore(SQLBaseStore): partitioned.setdefault(event.room_id, []).append((event, ctx)) deferreds = [] - for room_id, evs_ctxs in partitioned.iteritems(): - d = preserve_fn(self._event_persist_queue.add_to_queue)( + for room_id, evs_ctxs in iteritems(partitioned): + d = self._event_persist_queue.add_to_queue( room_id, evs_ctxs, backfilled=backfilled, ) @@ -242,7 +254,7 @@ class EventsStore(SQLBaseStore): for room_id in partitioned: self._maybe_start_persisting(room_id) - return preserve_context_over_deferred( + return make_deferred_yieldable( defer.gatherResults(deferreds, consumeErrors=True) ) @@ -267,7 +279,7 @@ class EventsStore(SQLBaseStore): self._maybe_start_persisting(event.room_id) - yield preserve_context_over_deferred(deferred) + yield make_deferred_yieldable(deferred) max_persisted_id = yield self._stream_id_gen.get_current_token() defer.returnValue((event.internal_metadata.stream_ordering, max_persisted_id)) @@ -275,10 +287,11 @@ class EventsStore(SQLBaseStore): def _maybe_start_persisting(self, room_id): @defer.inlineCallbacks def persisting_queue(item): - yield self._persist_events( - item.events_and_contexts, - backfilled=item.backfilled, - ) + with Measure(self._clock, "persist_events"): + yield self._persist_events( + item.events_and_contexts, + backfilled=item.backfilled, + ) self._event_persist_queue.handle_queue(room_id, persisting_queue) @@ -316,7 +329,7 @@ class EventsStore(SQLBaseStore): chunks = [ events_and_contexts[x:x + 100] - for x in xrange(0, len(events_and_contexts), 100) + for x in range(0, len(events_and_contexts), 100) ] for chunk in chunks: @@ -325,8 +338,23 @@ class EventsStore(SQLBaseStore): # NB: Assumes that we are only persisting events for one room # at a time. + + # map room_id->list[event_ids] giving the new forward + # extremities in each room new_forward_extremeties = {} + + # map room_id->(type,state_key)->event_id tracking the full + # state in each room after adding these events. + # This is simply used to prefill the get_current_state_ids + # cache current_state_for_room = {} + + # map room_id->(to_delete, to_insert) where to_delete is a list + # of type/state keys to remove from current state, and to_insert + # is a map (type,key)->event_id giving the state delta in each + # room + state_delta_for_room = {} + if not backfilled: with Measure(self._clock, "_calculate_state_and_extrem"): # Work out the new "current state" for each room. @@ -338,7 +366,7 @@ class EventsStore(SQLBaseStore): (event, context) ) - for room_id, ev_ctx_rm in events_by_room.iteritems(): + for room_id, ev_ctx_rm in iteritems(events_by_room): # Work out new extremities by recursively adding and removing # the new events. latest_event_ids = yield self.get_latest_event_ids_in_room( @@ -348,7 +376,8 @@ class EventsStore(SQLBaseStore): room_id, ev_ctx_rm, latest_event_ids ) - if new_latest_event_ids == set(latest_event_ids): + latest_event_ids = set(latest_event_ids) + if new_latest_event_ids == latest_event_ids: # No change in extremities, so no change in state continue @@ -369,11 +398,63 @@ class EventsStore(SQLBaseStore): if all_single_prev_not_state: continue - state = yield self._calculate_state_delta( - room_id, ev_ctx_rm, new_latest_event_ids + state_delta_counter.inc() + if len(new_latest_event_ids) == 1: + state_delta_single_event_counter.inc() + + # This is a fairly handwavey check to see if we could + # have guessed what the delta would have been when + # processing one of these events. + # What we're interested in is if the latest extremities + # were the same when we created the event as they are + # now. When this server creates a new event (as opposed + # to receiving it over federation) it will use the + # forward extremities as the prev_events, so we can + # guess this by looking at the prev_events and checking + # if they match the current forward extremities. + for ev, _ in ev_ctx_rm: + prev_event_ids = set(e for e, _ in ev.prev_events) + if latest_event_ids == prev_event_ids: + state_delta_reuse_delta_counter.inc() + break + + logger.info( + "Calculating state delta for room %s", room_id, ) - if state: - current_state_for_room[room_id] = state + with Measure( + self._clock, + "persist_events.get_new_state_after_events", + ): + res = yield self._get_new_state_after_events( + room_id, + ev_ctx_rm, + latest_event_ids, + new_latest_event_ids, + ) + current_state, delta_ids = res + + # If either are not None then there has been a change, + # and we need to work out the delta (or use that + # given) + if delta_ids is not None: + # If there is a delta we know that we've + # only added or replaced state, never + # removed keys entirely. + state_delta_for_room[room_id] = ([], delta_ids) + elif current_state is not None: + with Measure( + self._clock, + "persist_events.calculate_state_delta", + ): + delta = yield self._calculate_state_delta( + room_id, current_state, + ) + state_delta_for_room[room_id] = delta + + # If we have the current_state then lets prefill + # the cache with it. + if current_state is not None: + current_state_for_room[room_id] = current_state yield self.runInteraction( "persist_events", @@ -381,10 +462,13 @@ class EventsStore(SQLBaseStore): events_and_contexts=chunk, backfilled=backfilled, delete_existing=delete_existing, - current_state_for_room=current_state_for_room, + state_delta_for_room=state_delta_for_room, new_forward_extremeties=new_forward_extremeties, ) - persist_event_counter.inc_by(len(chunk)) + persist_event_counter.inc(len(chunk)) + synapse.metrics.event_persisted_position.set( + chunk[-1][0].internal_metadata.stream_ordering, + ) for event, context in chunk: if context.app_service: origin_type = "local" @@ -396,14 +480,14 @@ class EventsStore(SQLBaseStore): origin_type = "remote" origin_entity = get_domain_from_id(event.sender) - event_counter.inc(event.type, origin_type, origin_entity) + event_counter.labels(event.type, origin_type, origin_entity).inc() - for room_id, (_, _, new_state) in current_state_for_room.iteritems(): + for room_id, new_state in iteritems(current_state_for_room): self.get_current_state_ids.prefill( (room_id, ), new_state ) - for room_id, latest_event_ids in new_forward_extremeties.iteritems(): + for room_id, latest_event_ids in iteritems(new_forward_extremeties): self.get_latest_event_ids_in_room.prefill( (room_id,), list(latest_event_ids) ) @@ -450,183 +534,187 @@ class EventsStore(SQLBaseStore): defer.returnValue(new_latest_event_ids) @defer.inlineCallbacks - def _calculate_state_delta(self, room_id, events_context, new_latest_event_ids): + def _get_new_state_after_events(self, room_id, events_context, old_latest_event_ids, + new_latest_event_ids): + """Calculate the current state dict after adding some new events to + a room + + Args: + room_id (str): + room to which the events are being added. Used for logging etc + + events_context (list[(EventBase, EventContext)]): + events and contexts which are being added to the room + + old_latest_event_ids (iterable[str]): + the old forward extremities for the room. + + new_latest_event_ids (iterable[str]): + the new forward extremities for the room. + + Returns: + Deferred[tuple[dict[(str,str), str]|None, dict[(str,str), str]|None]]: + Returns a tuple of two state maps, the first being the full new current + state and the second being the delta to the existing current state. + If both are None then there has been no change. + + If there has been a change then we only return the delta if its + already been calculated. Conversely if we do know the delta then + the new current state is only returned if we've already calculated + it. + """ + + if not new_latest_event_ids: + return + + # map from state_group to ((type, key) -> event_id) state map + state_groups_map = {} + + # Map from (prev state group, new state group) -> delta state dict + state_group_deltas = {} + + for ev, ctx in events_context: + if ctx.state_group is None: + # I don't think this can happen, but let's double-check + raise Exception( + "Context for new extremity event %s has no state " + "group" % (ev.event_id, ), + ) + + if ctx.state_group in state_groups_map: + continue + + # We're only interested in pulling out state that has already + # been cached in the context. We'll pull stuff out of the DB later + # if necessary. + current_state_ids = ctx.get_cached_current_state_ids() + if current_state_ids is not None: + state_groups_map[ctx.state_group] = current_state_ids + + if ctx.prev_group: + state_group_deltas[(ctx.prev_group, ctx.state_group)] = ctx.delta_ids + + # We need to map the event_ids to their state groups. First, let's + # check if the event is one we're persisting, in which case we can + # pull the state group from its context. + # Otherwise we need to pull the state group from the database. + + # Set of events we need to fetch groups for. (We know none of the old + # extremities are going to be in events_context). + missing_event_ids = set(old_latest_event_ids) + + event_id_to_state_group = {} + for event_id in new_latest_event_ids: + # First search in the list of new events we're adding. + for ev, ctx in events_context: + if event_id == ev.event_id: + event_id_to_state_group[event_id] = ctx.state_group + break + else: + # If we couldn't find it, then we'll need to pull + # the state from the database + missing_event_ids.add(event_id) + + if missing_event_ids: + # Now pull out the state groups for any missing events from DB + event_to_groups = yield self._get_state_group_for_events( + missing_event_ids, + ) + event_id_to_state_group.update(event_to_groups) + + # State groups of old_latest_event_ids + old_state_groups = set( + event_id_to_state_group[evid] for evid in old_latest_event_ids + ) + + # State groups of new_latest_event_ids + new_state_groups = set( + event_id_to_state_group[evid] for evid in new_latest_event_ids + ) + + # If they old and new groups are the same then we don't need to do + # anything. + if old_state_groups == new_state_groups: + defer.returnValue((None, None)) + + if len(new_state_groups) == 1 and len(old_state_groups) == 1: + # If we're going from one state group to another, lets check if + # we have a delta for that transition. If we do then we can just + # return that. + + new_state_group = next(iter(new_state_groups)) + old_state_group = next(iter(old_state_groups)) + + delta_ids = state_group_deltas.get( + (old_state_group, new_state_group,), None + ) + if delta_ids is not None: + # We have a delta from the existing to new current state, + # so lets just return that. If we happen to already have + # the current state in memory then lets also return that, + # but it doesn't matter if we don't. + new_state = state_groups_map.get(new_state_group) + defer.returnValue((new_state, delta_ids)) + + # Now that we have calculated new_state_groups we need to get + # their state IDs so we can resolve to a single state set. + missing_state = new_state_groups - set(state_groups_map) + if missing_state: + group_to_state = yield self._get_state_for_groups(missing_state) + state_groups_map.update(group_to_state) + + if len(new_state_groups) == 1: + # If there is only one state group, then we know what the current + # state is. + defer.returnValue((state_groups_map[new_state_groups.pop()], None)) + + # Ok, we need to defer to the state handler to resolve our state sets. + + def get_events(ev_ids): + return self.get_events( + ev_ids, get_prev_content=False, check_redacted=False, + ) + + state_groups = { + sg: state_groups_map[sg] for sg in new_state_groups + } + + events_map = {ev.event_id: ev for ev, _ in events_context} + logger.debug("calling resolve_state_groups from preserve_events") + res = yield self._state_resolution_handler.resolve_state_groups( + room_id, state_groups, events_map, get_events + ) + + defer.returnValue((res.state, None)) + + @defer.inlineCallbacks + def _calculate_state_delta(self, room_id, current_state): """Calculate the new state deltas for a room. Assumes that we are only persisting events for one room at a time. Returns: - 3-tuple (to_delete, to_insert, new_state) where both are state dicts, - i.e. (type, state_key) -> event_id. `to_delete` are the entries to - first be deleted from current_state_events, `to_insert` are entries - to insert. `new_state` is the full set of state. - May return None if there are no changes to be applied. + tuple[list, dict] (to_delete, to_insert): where to_delete are the + type/state_keys to remove from current_state_events and `to_insert` + are the updates to current_state_events. """ - # Now we need to work out the different state sets for - # each state extremities - state_sets = [] - state_groups = set() - missing_event_ids = [] - was_updated = False - for event_id in new_latest_event_ids: - # First search in the list of new events we're adding, - # and then use the current state from that - for ev, ctx in events_context: - if event_id == ev.event_id: - if ctx.current_state_ids is None: - raise Exception("Unknown current state") - - # If we've already seen the state group don't bother adding - # it to the state sets again - if ctx.state_group not in state_groups: - state_sets.append(ctx.current_state_ids) - if ctx.delta_ids or hasattr(ev, "state_key"): - was_updated = True - if ctx.state_group: - # Add this as a seen state group (if it has a state - # group) - state_groups.add(ctx.state_group) - break - else: - # If we couldn't find it, then we'll need to pull - # the state from the database - was_updated = True - missing_event_ids.append(event_id) - - if missing_event_ids: - # Now pull out the state for any missing events from DB - event_to_groups = yield self._get_state_group_for_events( - missing_event_ids, - ) - - groups = set(event_to_groups.itervalues()) - state_groups - - if groups: - group_to_state = yield self._get_state_for_groups(groups) - state_sets.extend(group_to_state.itervalues()) - - if not new_latest_event_ids: - current_state = {} - elif was_updated: - if len(state_sets) == 1: - # If there is only one state set, then we know what the current - # state is. - current_state = state_sets[0] - else: - # We work out the current state by passing the state sets to the - # state resolution algorithm. It may ask for some events, including - # the events we have yet to persist, so we need a slightly more - # complicated event lookup function than simply looking the events - # up in the db. - events_map = {ev.event_id: ev for ev, _ in events_context} - - @defer.inlineCallbacks - def get_events(ev_ids): - # We get the events by first looking at the list of events we - # are trying to persist, and then fetching the rest from the DB. - db = [] - to_return = {} - for ev_id in ev_ids: - ev = events_map.get(ev_id, None) - if ev: - to_return[ev_id] = ev - else: - db.append(ev_id) - - if db: - evs = yield self.get_events( - ev_ids, get_prev_content=False, check_redacted=False, - ) - to_return.update(evs) - defer.returnValue(to_return) - - current_state = yield resolve_events( - state_sets, - state_map_factory=get_events, - ) - else: - return - existing_state = yield self.get_current_state_ids(room_id) - existing_events = set(existing_state.itervalues()) - new_events = set(ev_id for ev_id in current_state.itervalues()) - changed_events = existing_events ^ new_events + to_delete = [ + key for key in existing_state + if key not in current_state + ] - if not changed_events: - return - - to_delete = { - key: ev_id for key, ev_id in existing_state.iteritems() - if ev_id in changed_events - } - events_to_insert = (new_events - existing_events) to_insert = { - key: ev_id for key, ev_id in current_state.iteritems() - if ev_id in events_to_insert + key: ev_id for key, ev_id in iteritems(current_state) + if ev_id != existing_state.get(key) } - defer.returnValue((to_delete, to_insert, current_state)) - - @defer.inlineCallbacks - def get_event(self, event_id, check_redacted=True, - get_prev_content=False, allow_rejected=False, - allow_none=False): - """Get an event from the database by event_id. - - Args: - event_id (str): The event_id of the event to fetch - check_redacted (bool): If True, check if event has been redacted - and redact it. - get_prev_content (bool): If True and event is a state event, - include the previous states content in the unsigned field. - allow_rejected (bool): If True return rejected events. - allow_none (bool): If True, return None if no event found, if - False throw an exception. - - Returns: - Deferred : A FrozenEvent. - """ - events = yield self._get_events( - [event_id], - check_redacted=check_redacted, - get_prev_content=get_prev_content, - allow_rejected=allow_rejected, - ) - - if not events and not allow_none: - raise SynapseError(404, "Could not find event %s" % (event_id,)) - - defer.returnValue(events[0] if events else None) - - @defer.inlineCallbacks - def get_events(self, event_ids, check_redacted=True, - get_prev_content=False, allow_rejected=False): - """Get events from the database - - Args: - event_ids (list): The event_ids of the events to fetch - check_redacted (bool): If True, check if event has been redacted - and redact it. - get_prev_content (bool): If True and event is a state event, - include the previous states content in the unsigned field. - allow_rejected (bool): If True return rejected events. - - Returns: - Deferred : Dict from event_id to event. - """ - events = yield self._get_events( - event_ids, - check_redacted=check_redacted, - get_prev_content=get_prev_content, - allow_rejected=allow_rejected, - ) - - defer.returnValue({e.event_id: e for e in events}) + defer.returnValue((to_delete, to_insert)) @log_function def _persist_events_txn(self, txn, events_and_contexts, backfilled, - delete_existing=False, current_state_for_room={}, + delete_existing=False, state_delta_for_room={}, new_forward_extremeties={}): """Insert some number of room events into the necessary database tables. @@ -642,19 +730,21 @@ class EventsStore(SQLBaseStore): delete_existing (bool): True to purge existing table rows for the events from the database. This is useful when retrying due to IntegrityError. - current_state_for_room (dict[str, (list[str], list[str])]): + state_delta_for_room (dict[str, (list, dict)]): The current-state delta for each room. For each room, a tuple - (to_delete, to_insert), being a list of event ids to be removed - from the current state, and a list of event ids to be added to + (to_delete, to_insert), being a list of type/state keys to be + removed from the current state, and a state set to be added to the current state. new_forward_extremeties (dict[str, list[str]]): The new forward extremities for each room. For each room, a list of the event ids which are the forward extremities. """ + all_events_and_contexts = events_and_contexts + max_stream_order = events_and_contexts[-1][0].internal_metadata.stream_ordering - self._update_current_state_txn(txn, current_state_for_room, max_stream_order) + self._update_current_state_txn(txn, state_delta_for_room, max_stream_order) self._update_forward_extremities_txn( txn, @@ -698,9 +788,8 @@ class EventsStore(SQLBaseStore): events_and_contexts=events_and_contexts, ) - # Insert into the state_groups, state_groups_state, and - # event_to_state_groups tables. - self._store_mult_state_groups_txn(txn, events_and_contexts) + # Insert into event_to_state_groups. + self._store_event_state_mappings_txn(txn, events_and_contexts) # _store_rejected_events_txn filters out any events which were # rejected, and returns the filtered list. @@ -715,15 +804,53 @@ class EventsStore(SQLBaseStore): self._update_metadata_tables_txn( txn, events_and_contexts=events_and_contexts, + all_events_and_contexts=all_events_and_contexts, backfilled=backfilled, ) def _update_current_state_txn(self, txn, state_delta_by_room, max_stream_order): - for room_id, current_state_tuple in state_delta_by_room.iteritems(): - to_delete, to_insert, _ = current_state_tuple + for room_id, current_state_tuple in iteritems(state_delta_by_room): + to_delete, to_insert = current_state_tuple + + # First we add entries to the current_state_delta_stream. We + # do this before updating the current_state_events table so + # that we can use it to calculate the `prev_event_id`. (This + # allows us to not have to pull out the existing state + # unnecessarily). + sql = """ + INSERT INTO current_state_delta_stream + (stream_id, room_id, type, state_key, event_id, prev_event_id) + SELECT ?, ?, ?, ?, ?, ( + SELECT event_id FROM current_state_events + WHERE room_id = ? AND type = ? AND state_key = ? + ) + """ + txn.executemany(sql, ( + ( + max_stream_order, room_id, etype, state_key, None, + room_id, etype, state_key, + ) + for etype, state_key in to_delete + # We sanity check that we're deleting rather than updating + if (etype, state_key) not in to_insert + )) + txn.executemany(sql, ( + ( + max_stream_order, room_id, etype, state_key, ev_id, + room_id, etype, state_key, + ) + for (etype, state_key), ev_id in iteritems(to_insert) + )) + + # Now we actually update the current_state_events table + txn.executemany( - "DELETE FROM current_state_events WHERE event_id = ?", - [(ev_id,) for ev_id in to_delete.itervalues()], + "DELETE FROM current_state_events" + " WHERE room_id = ? AND type = ? AND state_key = ?", + ( + (room_id, etype, state_key) + for etype, state_key in itertools.chain(to_delete, to_insert) + ), ) self._simple_insert_many_txn( @@ -736,30 +863,12 @@ class EventsStore(SQLBaseStore): "type": key[0], "state_key": key[1], } - for key, ev_id in to_insert.iteritems() + for key, ev_id in iteritems(to_insert) ], ) - state_deltas = {key: None for key in to_delete} - state_deltas.update(to_insert) - - self._simple_insert_many_txn( - txn, - table="current_state_delta_stream", - values=[ - { - "stream_id": max_stream_order, - "room_id": room_id, - "type": key[0], - "state_key": key[1], - "event_id": ev_id, - "prev_event_id": to_delete.get(key, None), - } - for key, ev_id in state_deltas.iteritems() - ] - ) - - self._curr_state_delta_stream_cache.entity_has_changed( + txn.call_after( + self._curr_state_delta_stream_cache.entity_has_changed, room_id, max_stream_order, ) @@ -771,19 +880,23 @@ class EventsStore(SQLBaseStore): # and which we have added, then we invlidate the caches for all # those users. members_changed = set( - state_key for ev_type, state_key in state_deltas + state_key + for ev_type, state_key in itertools.chain(to_delete, to_insert) if ev_type == EventTypes.Member ) for member in members_changed: self._invalidate_cache_and_stream( - txn, self.get_rooms_for_user, (member,) + txn, self.get_rooms_for_user_with_stream_ordering, (member,) ) for host in set(get_domain_from_id(u) for u in members_changed): self._invalidate_cache_and_stream( txn, self.is_host_joined, (room_id, host) ) + self._invalidate_cache_and_stream( + txn, self.was_host_joined, (room_id, host) + ) self._invalidate_cache_and_stream( txn, self.get_users_in_room, (room_id,) @@ -795,7 +908,7 @@ class EventsStore(SQLBaseStore): def _update_forward_extremities_txn(self, txn, new_forward_extremities, max_stream_order): - for room_id, new_extrem in new_forward_extremities.iteritems(): + for room_id, new_extrem in iteritems(new_forward_extremities): self._simple_delete_txn( txn, table="event_forward_extremities", @@ -813,7 +926,7 @@ class EventsStore(SQLBaseStore): "event_id": ev_id, "room_id": room_id, } - for room_id, new_extrem in new_forward_extremities.iteritems() + for room_id, new_extrem in iteritems(new_forward_extremities) for ev_id in new_extrem ], ) @@ -830,7 +943,7 @@ class EventsStore(SQLBaseStore): "event_id": event_id, "stream_ordering": max_stream_order, } - for room_id, new_extrem in new_forward_extremities.iteritems() + for room_id, new_extrem in iteritems(new_forward_extremities) for event_id in new_extrem ] ) @@ -858,7 +971,7 @@ class EventsStore(SQLBaseStore): new_events_and_contexts[event.event_id] = (event, context) else: new_events_and_contexts[event.event_id] = (event, context) - return new_events_and_contexts.values() + return list(new_events_and_contexts.values()) def _update_room_depths_txn(self, txn, events_and_contexts, backfilled): """Update min_depth for each room @@ -884,7 +997,7 @@ class EventsStore(SQLBaseStore): event.depth, depth_updates.get(event.room_id, event.depth) ) - for room_id, depth in depth_updates.iteritems(): + for room_id, depth in iteritems(depth_updates): self._update_min_depth_for_room_txn(txn, room_id, depth) def _update_outliers_txn(self, txn, events_and_contexts): @@ -932,10 +1045,9 @@ class EventsStore(SQLBaseStore): # an outlier in the database. We now have some state at that # so we need to update the state_groups table with that state. - # insert into the state_group, state_groups_state and - # event_to_state_groups tables. + # insert into event_to_state_groups. try: - self._store_mult_state_groups_txn(txn, ((event, context),)) + self._store_event_state_mappings_txn(txn, ((event, context),)) except Exception: logger.exception("") raise @@ -1001,7 +1113,6 @@ class EventsStore(SQLBaseStore): "event_edge_hashes", "event_edges", "event_forward_extremities", - "event_push_actions", "event_reference_hashes", "event_search", "event_signatures", @@ -1021,6 +1132,14 @@ class EventsStore(SQLBaseStore): [(ev.event_id,) for ev, _ in events_and_contexts] ) + for table in ( + "event_push_actions", + ): + txn.executemany( + "DELETE FROM %s WHERE room_id = ? AND event_id = ?" % (table,), + [(ev.event_id,) for ev, _ in events_and_contexts] + ) + def _store_event_txn(self, txn, events_and_contexts): """Insert new events into the event and event_json tables @@ -1110,27 +1229,33 @@ class EventsStore(SQLBaseStore): ec for ec in events_and_contexts if ec[0] not in to_remove ] - def _update_metadata_tables_txn(self, txn, events_and_contexts, backfilled): + def _update_metadata_tables_txn(self, txn, events_and_contexts, + all_events_and_contexts, backfilled): """Update all the miscellaneous tables for new events Args: txn (twisted.enterprise.adbapi.Connection): db connection events_and_contexts (list[(EventBase, EventContext)]): events we are persisting + all_events_and_contexts (list[(EventBase, EventContext)]): all + events that we were going to persist. This includes events + we've already persisted, etc, that wouldn't appear in + events_and_context. backfilled (bool): True if the events were backfilled """ + # Insert all the push actions into the event_push_actions table. + self._set_push_actions_for_event_and_users_txn( + txn, + events_and_contexts=events_and_contexts, + all_events_and_contexts=all_events_and_contexts, + ) + if not events_and_contexts: # nothing to do here return for event, context in events_and_contexts: - # Insert all the push actions into the event_push_actions table. - if context.push_actions: - self._set_push_actions_for_event_and_users_txn( - txn, event, context.push_actions - ) - if event.type == EventTypes.Redaction and event.redacts is not None: # Remove the entries in the event_push_actions table for the # redacted event. @@ -1263,7 +1388,7 @@ class EventsStore(SQLBaseStore): " WHERE e.event_id IN (%s)" ) % (",".join(["?"] * len(ev_map)),) - txn.execute(sql, ev_map.keys()) + txn.execute(sql, list(ev_map)) rows = self.cursor_to_dict(txn) for row in rows: event = ev_map[row["event_id"]] @@ -1302,13 +1427,49 @@ class EventsStore(SQLBaseStore): defer.returnValue(set(r["event_id"] for r in rows)) - def have_events(self, event_ids): + @defer.inlineCallbacks + def have_seen_events(self, event_ids): """Given a list of event ids, check if we have already processed them. + Args: + event_ids (iterable[str]): + Returns: - dict: Has an entry for each event id we already have seen. Maps to - the rejected reason string if we rejected the event, else maps to - None. + Deferred[set[str]]: The events we have already seen. + """ + results = set() + + def have_seen_events_txn(txn, chunk): + sql = ( + "SELECT event_id FROM events as e WHERE e.event_id IN (%s)" + % (",".join("?" * len(chunk)), ) + ) + txn.execute(sql, chunk) + for (event_id, ) in txn: + results.add(event_id) + + # break the input up into chunks of 100 + input_iterator = iter(event_ids) + for chunk in iter(lambda: list(itertools.islice(input_iterator, 100)), + []): + yield self.runInteraction( + "have_seen_events", + have_seen_events_txn, + chunk, + ) + defer.returnValue(results) + + def get_seen_events_with_rejections(self, event_ids): + """Given a list of event ids, check if we rejected them. + + Args: + event_ids (list[str]) + + Returns: + Deferred[dict[str, str|None): + Has an entry for each event id we already have seen. Maps to + the rejected reason string if we rejected the event, else maps + to None. """ if not event_ids: return defer.succeed({}) @@ -1330,295 +1491,7 @@ class EventsStore(SQLBaseStore): return res - return self.runInteraction( - "have_events", f, - ) - - @defer.inlineCallbacks - def _get_events(self, event_ids, check_redacted=True, - get_prev_content=False, allow_rejected=False): - if not event_ids: - defer.returnValue([]) - - event_id_list = event_ids - event_ids = set(event_ids) - - event_entry_map = self._get_events_from_cache( - event_ids, - allow_rejected=allow_rejected, - ) - - missing_events_ids = [e for e in event_ids if e not in event_entry_map] - - if missing_events_ids: - missing_events = yield self._enqueue_events( - missing_events_ids, - check_redacted=check_redacted, - allow_rejected=allow_rejected, - ) - - event_entry_map.update(missing_events) - - events = [] - for event_id in event_id_list: - entry = event_entry_map.get(event_id, None) - if not entry: - continue - - if allow_rejected or not entry.event.rejected_reason: - if check_redacted and entry.redacted_event: - event = entry.redacted_event - else: - event = entry.event - - events.append(event) - - if get_prev_content: - if "replaces_state" in event.unsigned: - prev = yield self.get_event( - event.unsigned["replaces_state"], - get_prev_content=False, - allow_none=True, - ) - if prev: - event.unsigned = dict(event.unsigned) - event.unsigned["prev_content"] = prev.content - event.unsigned["prev_sender"] = prev.sender - - defer.returnValue(events) - - def _invalidate_get_event_cache(self, event_id): - self._get_event_cache.invalidate((event_id,)) - - def _get_events_from_cache(self, events, allow_rejected, update_metrics=True): - """Fetch events from the caches - - Args: - events (list(str)): list of event_ids to fetch - allow_rejected (bool): Whether to teturn events that were rejected - update_metrics (bool): Whether to update the cache hit ratio metrics - - Returns: - dict of event_id -> _EventCacheEntry for each event_id in cache. If - allow_rejected is `False` then there will still be an entry but it - will be `None` - """ - event_map = {} - - for event_id in events: - ret = self._get_event_cache.get( - (event_id,), None, - update_metrics=update_metrics, - ) - if not ret: - continue - - if allow_rejected or not ret.event.rejected_reason: - event_map[event_id] = ret - else: - event_map[event_id] = None - - return event_map - - def _do_fetch(self, conn): - """Takes a database connection and waits for requests for events from - the _event_fetch_list queue. - """ - event_list = [] - i = 0 - while True: - try: - with self._event_fetch_lock: - event_list = self._event_fetch_list - self._event_fetch_list = [] - - if not event_list: - single_threaded = self.database_engine.single_threaded - if single_threaded or i > EVENT_QUEUE_ITERATIONS: - self._event_fetch_ongoing -= 1 - return - else: - self._event_fetch_lock.wait(EVENT_QUEUE_TIMEOUT_S) - i += 1 - continue - i = 0 - - event_id_lists = zip(*event_list)[0] - event_ids = [ - item for sublist in event_id_lists for item in sublist - ] - - rows = self._new_transaction( - conn, "do_fetch", [], [], None, self._fetch_event_rows, event_ids - ) - - row_dict = { - r["event_id"]: r - for r in rows - } - - # We only want to resolve deferreds from the main thread - def fire(lst, res): - for ids, d in lst: - if not d.called: - try: - with PreserveLoggingContext(): - d.callback([ - res[i] - for i in ids - if i in res - ]) - except: - logger.exception("Failed to callback") - with PreserveLoggingContext(): - reactor.callFromThread(fire, event_list, row_dict) - except Exception as e: - logger.exception("do_fetch") - - # We only want to resolve deferreds from the main thread - def fire(evs): - for _, d in evs: - if not d.called: - with PreserveLoggingContext(): - d.errback(e) - - if event_list: - with PreserveLoggingContext(): - reactor.callFromThread(fire, event_list) - - @defer.inlineCallbacks - def _enqueue_events(self, events, check_redacted=True, allow_rejected=False): - """Fetches events from the database using the _event_fetch_list. This - allows batch and bulk fetching of events - it allows us to fetch events - without having to create a new transaction for each request for events. - """ - if not events: - defer.returnValue({}) - - events_d = defer.Deferred() - with self._event_fetch_lock: - self._event_fetch_list.append( - (events, events_d) - ) - - self._event_fetch_lock.notify() - - if self._event_fetch_ongoing < EVENT_QUEUE_THREADS: - self._event_fetch_ongoing += 1 - should_start = True - else: - should_start = False - - if should_start: - with PreserveLoggingContext(): - self.runWithConnection( - self._do_fetch - ) - - logger.debug("Loading %d events", len(events)) - with PreserveLoggingContext(): - rows = yield events_d - logger.debug("Loaded %d events (%d rows)", len(events), len(rows)) - - if not allow_rejected: - rows[:] = [r for r in rows if not r["rejects"]] - - res = yield preserve_context_over_deferred(defer.gatherResults( - [ - preserve_fn(self._get_event_from_row)( - row["internal_metadata"], row["json"], row["redacts"], - rejected_reason=row["rejects"], - ) - for row in rows - ], - consumeErrors=True - )) - - defer.returnValue({ - e.event.event_id: e - for e in res if e - }) - - def _fetch_event_rows(self, txn, events): - rows = [] - N = 200 - for i in range(1 + len(events) / N): - evs = events[i * N:(i + 1) * N] - if not evs: - break - - sql = ( - "SELECT " - " e.event_id as event_id, " - " e.internal_metadata," - " e.json," - " r.redacts as redacts," - " rej.event_id as rejects " - " FROM event_json as e" - " LEFT JOIN rejections as rej USING (event_id)" - " LEFT JOIN redactions as r ON e.event_id = r.redacts" - " WHERE e.event_id IN (%s)" - ) % (",".join(["?"] * len(evs)),) - - txn.execute(sql, evs) - rows.extend(self.cursor_to_dict(txn)) - - return rows - - @defer.inlineCallbacks - def _get_event_from_row(self, internal_metadata, js, redacted, - rejected_reason=None): - with Measure(self._clock, "_get_event_from_row"): - d = json.loads(js) - internal_metadata = json.loads(internal_metadata) - - if rejected_reason: - rejected_reason = yield self._simple_select_one_onecol( - table="rejections", - keyvalues={"event_id": rejected_reason}, - retcol="reason", - desc="_get_event_from_row_rejected_reason", - ) - - original_ev = FrozenEvent( - d, - internal_metadata_dict=internal_metadata, - rejected_reason=rejected_reason, - ) - - redacted_event = None - if redacted: - redacted_event = prune_event(original_ev) - - redaction_id = yield self._simple_select_one_onecol( - table="redactions", - keyvalues={"redacts": redacted_event.event_id}, - retcol="event_id", - desc="_get_event_from_row_redactions", - ) - - redacted_event.unsigned["redacted_by"] = redaction_id - # Get the redaction event. - - because = yield self.get_event( - redaction_id, - check_redacted=False, - allow_none=True, - ) - - if because: - # It's fine to do add the event directly, since get_pdu_json - # will serialise this field correctly - redacted_event.unsigned["redacted_because"] = because - - cache_entry = _EventCacheEntry( - event=original_ev, - redacted_event=redacted_event, - ) - - self._get_event_cache.prefill((original_ev.event_id,), cache_entry) - - defer.returnValue(cache_entry) + return self.runInteraction("get_rejection_reasons", f) @defer.inlineCallbacks def count_daily_messages(self): @@ -1778,7 +1651,7 @@ class EventsStore(SQLBaseStore): chunks = [ event_ids[i:i + 100] - for i in xrange(0, len(event_ids), 100) + for i in range(0, len(event_ids), 100) ] for chunk in chunks: ev_rows = self._simple_select_many_txn( @@ -2005,15 +1878,32 @@ class EventsStore(SQLBaseStore): ) return self.runInteraction("get_all_new_events", get_all_new_events_txn) - def delete_old_state(self, room_id, topological_ordering): + def purge_history( + self, room_id, token, delete_local_events, + ): + """Deletes room history before a certain point + + Args: + room_id (str): + + token (str): A topological token to delete events before + + delete_local_events (bool): + if True, we will delete local events as well as remote ones + (instead of just marking them as outliers and deleting their + state groups). + """ + return self.runInteraction( - "delete_old_state", - self._delete_old_state_txn, room_id, topological_ordering + "purge_history", + self._purge_history_txn, room_id, token, + delete_local_events, ) - def _delete_old_state_txn(self, txn, room_id, topological_ordering): - """Deletes old room state - """ + def _purge_history_txn( + self, txn, room_id, token_str, delete_local_events, + ): + token = RoomStreamToken.parse(token_str) # Tables that should be pruned: # event_auth @@ -2035,6 +1925,37 @@ class EventsStore(SQLBaseStore): # state_groups # state_groups_state + # we will build a temporary table listing the events so that we don't + # have to keep shovelling the list back and forth across the + # connection. Annoyingly the python sqlite driver commits the + # transaction on CREATE, so let's do this first. + # + # furthermore, we might already have the table from a previous (failed) + # purge attempt, so let's drop the table first. + + txn.execute("DROP TABLE IF EXISTS events_to_purge") + + txn.execute( + "CREATE TEMPORARY TABLE events_to_purge (" + " event_id TEXT NOT NULL," + " should_delete BOOLEAN NOT NULL" + ")" + ) + + # create an index on should_delete because later we'll be looking for + # the should_delete / shouldn't_delete subsets + txn.execute( + "CREATE INDEX events_to_purge_should_delete" + " ON events_to_purge(should_delete)", + ) + + # We do joins against events_to_purge for e.g. calculating state + # groups to purge, etc., so lets make an index. + txn.execute( + "CREATE INDEX events_to_purge_id" + " ON events_to_purge(event_id)", + ) + # First ensure that we're not about to delete all the forward extremeties txn.execute( "SELECT e.event_id, e.depth FROM events as e " @@ -2047,7 +1968,7 @@ class EventsStore(SQLBaseStore): rows = txn.fetchall() max_depth = max(row[0] for row in rows) - if max_depth <= topological_ordering: + if max_depth <= token.topological: # We need to ensure we don't delete all the events from the datanase # otherwise we wouldn't be able to send any events (due to not # having any backwards extremeties) @@ -2055,42 +1976,48 @@ class EventsStore(SQLBaseStore): 400, "topological_ordering is greater than forward extremeties" ) - logger.debug("[purge] looking for events to delete") + logger.info("[purge] looking for events to delete") + + should_delete_expr = "state_key IS NULL" + should_delete_params = () + if not delete_local_events: + should_delete_expr += " AND event_id NOT LIKE ?" + should_delete_params += ("%:" + self.hs.hostname, ) + + should_delete_params += (room_id, token.topological) txn.execute( - "SELECT event_id, state_key FROM events" - " LEFT JOIN state_events USING (room_id, event_id)" - " WHERE room_id = ? AND topological_ordering < ?", - (room_id, topological_ordering,) + "INSERT INTO events_to_purge" + " SELECT event_id, %s" + " FROM events AS e LEFT JOIN state_events USING (event_id)" + " WHERE e.room_id = ? AND topological_ordering < ?" % ( + should_delete_expr, + ), + should_delete_params, + ) + txn.execute( + "SELECT event_id, should_delete FROM events_to_purge" ) event_rows = txn.fetchall() - - to_delete = [ - (event_id,) for event_id, state_key in event_rows - if state_key is None and not self.hs.is_mine_id(event_id) - ] logger.info( - "[purge] found %i events before cutoff, of which %i are remote" - " non-state events to delete", len(event_rows), len(to_delete)) + "[purge] found %i events before cutoff, of which %i can be deleted", + len(event_rows), sum(1 for e in event_rows if e[1]), + ) - for event_id, state_key in event_rows: - txn.call_after(self._get_state_group_for_event.invalidate, (event_id,)) - - logger.debug("[purge] Finding new backward extremities") + logger.info("[purge] Finding new backward extremities") # We calculate the new entries for the backward extremeties by finding - # all events that point to events that are to be purged + # events to be purged that are pointed to by events we're not going to + # purge. txn.execute( - "SELECT DISTINCT e.event_id FROM events as e" - " INNER JOIN event_edges as ed ON e.event_id = ed.prev_event_id" - " INNER JOIN events as e2 ON e2.event_id = ed.event_id" - " WHERE e.room_id = ? AND e.topological_ordering < ?" - " AND e2.topological_ordering >= ?", - (room_id, topological_ordering, topological_ordering) + "SELECT DISTINCT e.event_id FROM events_to_purge AS e" + " INNER JOIN event_edges AS ed ON e.event_id = ed.prev_event_id" + " LEFT JOIN events_to_purge AS ep2 ON ed.event_id = ep2.event_id" + " WHERE ep2.event_id IS NULL", ) new_backwards_extrems = txn.fetchall() - logger.debug("[purge] replacing backward extremities: %r", new_backwards_extrems) + logger.info("[purge] replacing backward extremities: %r", new_backwards_extrems) txn.execute( "DELETE FROM event_backward_extremities WHERE room_id = ?", @@ -2106,34 +2033,39 @@ class EventsStore(SQLBaseStore): ] ) - logger.debug("[purge] finding redundant state groups") + logger.info("[purge] finding redundant state groups") # Get all state groups that are only referenced by events that are # to be deleted. - txn.execute( - "SELECT state_group FROM event_to_state_groups" - " INNER JOIN events USING (event_id)" - " WHERE state_group IN (" - " SELECT DISTINCT state_group FROM events" - " INNER JOIN event_to_state_groups USING (event_id)" - " WHERE room_id = ? AND topological_ordering < ?" - " )" - " GROUP BY state_group HAVING MAX(topological_ordering) < ?", - (room_id, topological_ordering, topological_ordering) - ) + # This works by first getting state groups that we may want to delete, + # joining against event_to_state_groups to get events that use that + # state group, then left joining against events_to_purge again. Any + # state group where the left join produce *no nulls* are referenced + # only by events that are going to be purged. + txn.execute(""" + SELECT state_group FROM + ( + SELECT DISTINCT state_group FROM events_to_purge + INNER JOIN event_to_state_groups USING (event_id) + ) AS sp + INNER JOIN event_to_state_groups USING (state_group) + LEFT JOIN events_to_purge AS ep USING (event_id) + GROUP BY state_group + HAVING SUM(CASE WHEN ep.event_id IS NULL THEN 1 ELSE 0 END) = 0 + """) state_rows = txn.fetchall() - logger.debug("[purge] found %i redundant state groups", len(state_rows)) + logger.info("[purge] found %i redundant state groups", len(state_rows)) # make a set of the redundant state groups, so that we can look them up # efficiently state_groups_to_delete = set([sg for sg, in state_rows]) # Now we get all the state groups that rely on these state groups - logger.debug("[purge] finding state groups which depend on redundant" - " state groups") + logger.info("[purge] finding state groups which depend on redundant" + " state groups") remaining_state_groups = [] - for i in xrange(0, len(state_rows), 100): + for i in range(0, len(state_rows), 100): chunk = [sg for sg, in state_rows[i:i + 100]] # look for state groups whose prev_state_group is one we are about # to delete @@ -2156,7 +2088,7 @@ class EventsStore(SQLBaseStore): # Now we turn the state groups that reference to-be-deleted state # groups to non delta versions. for sg in remaining_state_groups: - logger.debug("[purge] de-delta-ing remaining state group %s", sg) + logger.info("[purge] de-delta-ing remaining state group %s", sg) curr_state = self._get_state_groups_from_groups_txn( txn, [sg], types=None ) @@ -2189,11 +2121,11 @@ class EventsStore(SQLBaseStore): "state_key": key[1], "event_id": state_id, } - for key, state_id in curr_state.iteritems() + for key, state_id in iteritems(curr_state) ], ) - logger.debug("[purge] removing redundant state groups") + logger.info("[purge] removing redundant state groups") txn.executemany( "DELETE FROM state_groups_state WHERE state_group = ?", state_rows @@ -2203,18 +2135,15 @@ class EventsStore(SQLBaseStore): state_rows ) - # Delete all non-state - logger.debug("[purge] removing events from event_to_state_groups") - txn.executemany( - "DELETE FROM event_to_state_groups WHERE event_id = ?", - [(event_id,) for event_id, _ in event_rows] - ) - - logger.debug("[purge] updating room_depth") + logger.info("[purge] removing events from event_to_state_groups") txn.execute( - "UPDATE room_depth SET min_depth = ? WHERE room_id = ?", - (topological_ordering, room_id,) + "DELETE FROM event_to_state_groups " + "WHERE event_id IN (SELECT event_id from events_to_purge)" ) + for event_id, _ in event_rows: + txn.call_after(self._get_state_group_for_event.invalidate, ( + event_id, + )) # Delete all remote non-state events for table in ( @@ -2226,28 +2155,75 @@ class EventsStore(SQLBaseStore): "event_edge_hashes", "event_edges", "event_forward_extremities", - "event_push_actions", "event_reference_hashes", "event_search", "event_signatures", "rejections", ): - logger.debug("[purge] removing remote non-state events from %s", table) + logger.info("[purge] removing events from %s", table) - txn.executemany( - "DELETE FROM %s WHERE event_id = ?" % (table,), - to_delete + txn.execute( + "DELETE FROM %s WHERE event_id IN (" + " SELECT event_id FROM events_to_purge WHERE should_delete" + ")" % (table,), + ) + + # event_push_actions lacks an index on event_id, and has one on + # (room_id, event_id) instead. + for table in ( + "event_push_actions", + ): + logger.info("[purge] removing events from %s", table) + + txn.execute( + "DELETE FROM %s WHERE room_id = ? AND event_id IN (" + " SELECT event_id FROM events_to_purge WHERE should_delete" + ")" % (table,), + (room_id, ) ) # Mark all state and own events as outliers - logger.debug("[purge] marking remaining events as outliers") - txn.executemany( + logger.info("[purge] marking remaining events as outliers") + txn.execute( "UPDATE events SET outlier = ?" - " WHERE event_id = ?", - [ - (True, event_id,) for event_id, state_key in event_rows - if state_key is not None or self.hs.is_mine_id(event_id) - ] + " WHERE event_id IN (" + " SELECT event_id FROM events_to_purge " + " WHERE NOT should_delete" + ")", + (True,), + ) + + # synapse tries to take out an exclusive lock on room_depth whenever it + # persists events (because upsert), and once we run this update, we + # will block that for the rest of our transaction. + # + # So, let's stick it at the end so that we don't block event + # persistence. + # + # We do this by calculating the minimum depth of the backwards + # extremities. However, the events in event_backward_extremities + # are ones we don't have yet so we need to look at the events that + # point to it via event_edges table. + txn.execute(""" + SELECT COALESCE(MIN(depth), 0) + FROM event_backward_extremities AS eb + INNER JOIN event_edges AS eg ON eg.prev_event_id = eb.event_id + INNER JOIN events AS e ON e.event_id = eg.event_id + WHERE eb.room_id = ? + """, (room_id,)) + min_depth, = txn.fetchone() + + logger.info("[purge] updating room_depth to %d", min_depth) + + txn.execute( + "UPDATE room_depth SET min_depth = ? WHERE room_id = ?", + (min_depth, room_id,) + ) + + # finally, drop the temp table. this will commit the txn in sqlite, + # so make sure to keep this actually last. + txn.execute( + "DROP TABLE events_to_purge" ) logger.info("[purge] done") @@ -2260,7 +2236,7 @@ class EventsStore(SQLBaseStore): to_2, so_2 = yield self._get_event_ordering(event_id2) defer.returnValue((to_1, so_1) > (to_2, so_2)) - @defer.inlineCallbacks + @cachedInlineCallbacks(max_entries=5000) def _get_event_ordering(self, event_id): res = yield self._simple_select_one( table="events", diff --git a/synapse/storage/events_worker.py b/synapse/storage/events_worker.py new file mode 100644 index 000000000..f28239a80 --- /dev/null +++ b/synapse/storage/events_worker.py @@ -0,0 +1,436 @@ +# -*- coding: utf-8 -*- +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import logging +from collections import namedtuple + +from canonicaljson import json + +from twisted.internet import defer + +from synapse.api.errors import SynapseError +# these are only included to make the type annotations work +from synapse.events import EventBase # noqa: F401 +from synapse.events import FrozenEvent +from synapse.events.snapshot import EventContext # noqa: F401 +from synapse.events.utils import prune_event +from synapse.metrics.background_process_metrics import run_as_background_process +from synapse.util.logcontext import ( + LoggingContext, + PreserveLoggingContext, + make_deferred_yieldable, + run_in_background, +) +from synapse.util.metrics import Measure + +from ._base import SQLBaseStore + +logger = logging.getLogger(__name__) + + +# These values are used in the `enqueus_event` and `_do_fetch` methods to +# control how we batch/bulk fetch events from the database. +# The values are plucked out of thing air to make initial sync run faster +# on jki.re +# TODO: Make these configurable. +EVENT_QUEUE_THREADS = 3 # Max number of threads that will fetch events +EVENT_QUEUE_ITERATIONS = 3 # No. times we block waiting for requests for events +EVENT_QUEUE_TIMEOUT_S = 0.1 # Timeout when waiting for requests for events + + +_EventCacheEntry = namedtuple("_EventCacheEntry", ("event", "redacted_event")) + + +class EventsWorkerStore(SQLBaseStore): + def get_received_ts(self, event_id): + """Get received_ts (when it was persisted) for the event. + + Raises an exception for unknown events. + + Args: + event_id (str) + + Returns: + Deferred[int|None]: Timestamp in milliseconds, or None for events + that were persisted before received_ts was implemented. + """ + return self._simple_select_one_onecol( + table="events", + keyvalues={ + "event_id": event_id, + }, + retcol="received_ts", + desc="get_received_ts", + ) + + @defer.inlineCallbacks + def get_event(self, event_id, check_redacted=True, + get_prev_content=False, allow_rejected=False, + allow_none=False): + """Get an event from the database by event_id. + + Args: + event_id (str): The event_id of the event to fetch + check_redacted (bool): If True, check if event has been redacted + and redact it. + get_prev_content (bool): If True and event is a state event, + include the previous states content in the unsigned field. + allow_rejected (bool): If True return rejected events. + allow_none (bool): If True, return None if no event found, if + False throw an exception. + + Returns: + Deferred : A FrozenEvent. + """ + events = yield self._get_events( + [event_id], + check_redacted=check_redacted, + get_prev_content=get_prev_content, + allow_rejected=allow_rejected, + ) + + if not events and not allow_none: + raise SynapseError(404, "Could not find event %s" % (event_id,)) + + defer.returnValue(events[0] if events else None) + + @defer.inlineCallbacks + def get_events(self, event_ids, check_redacted=True, + get_prev_content=False, allow_rejected=False): + """Get events from the database + + Args: + event_ids (list): The event_ids of the events to fetch + check_redacted (bool): If True, check if event has been redacted + and redact it. + get_prev_content (bool): If True and event is a state event, + include the previous states content in the unsigned field. + allow_rejected (bool): If True return rejected events. + + Returns: + Deferred : Dict from event_id to event. + """ + events = yield self._get_events( + event_ids, + check_redacted=check_redacted, + get_prev_content=get_prev_content, + allow_rejected=allow_rejected, + ) + + defer.returnValue({e.event_id: e for e in events}) + + @defer.inlineCallbacks + def _get_events(self, event_ids, check_redacted=True, + get_prev_content=False, allow_rejected=False): + if not event_ids: + defer.returnValue([]) + + event_id_list = event_ids + event_ids = set(event_ids) + + event_entry_map = self._get_events_from_cache( + event_ids, + allow_rejected=allow_rejected, + ) + + missing_events_ids = [e for e in event_ids if e not in event_entry_map] + + if missing_events_ids: + log_ctx = LoggingContext.current_context() + log_ctx.record_event_fetch(len(missing_events_ids)) + + missing_events = yield self._enqueue_events( + missing_events_ids, + check_redacted=check_redacted, + allow_rejected=allow_rejected, + ) + + event_entry_map.update(missing_events) + + events = [] + for event_id in event_id_list: + entry = event_entry_map.get(event_id, None) + if not entry: + continue + + if allow_rejected or not entry.event.rejected_reason: + if check_redacted and entry.redacted_event: + event = entry.redacted_event + else: + event = entry.event + + events.append(event) + + if get_prev_content: + if "replaces_state" in event.unsigned: + prev = yield self.get_event( + event.unsigned["replaces_state"], + get_prev_content=False, + allow_none=True, + ) + if prev: + event.unsigned = dict(event.unsigned) + event.unsigned["prev_content"] = prev.content + event.unsigned["prev_sender"] = prev.sender + + defer.returnValue(events) + + def _invalidate_get_event_cache(self, event_id): + self._get_event_cache.invalidate((event_id,)) + + def _get_events_from_cache(self, events, allow_rejected, update_metrics=True): + """Fetch events from the caches + + Args: + events (list(str)): list of event_ids to fetch + allow_rejected (bool): Whether to teturn events that were rejected + update_metrics (bool): Whether to update the cache hit ratio metrics + + Returns: + dict of event_id -> _EventCacheEntry for each event_id in cache. If + allow_rejected is `False` then there will still be an entry but it + will be `None` + """ + event_map = {} + + for event_id in events: + ret = self._get_event_cache.get( + (event_id,), None, + update_metrics=update_metrics, + ) + if not ret: + continue + + if allow_rejected or not ret.event.rejected_reason: + event_map[event_id] = ret + else: + event_map[event_id] = None + + return event_map + + def _do_fetch(self, conn): + """Takes a database connection and waits for requests for events from + the _event_fetch_list queue. + """ + i = 0 + while True: + with self._event_fetch_lock: + event_list = self._event_fetch_list + self._event_fetch_list = [] + + if not event_list: + single_threaded = self.database_engine.single_threaded + if single_threaded or i > EVENT_QUEUE_ITERATIONS: + self._event_fetch_ongoing -= 1 + return + else: + self._event_fetch_lock.wait(EVENT_QUEUE_TIMEOUT_S) + i += 1 + continue + i = 0 + + self._fetch_event_list(conn, event_list) + + def _fetch_event_list(self, conn, event_list): + """Handle a load of requests from the _event_fetch_list queue + + Args: + conn (twisted.enterprise.adbapi.Connection): database connection + + event_list (list[Tuple[list[str], Deferred]]): + The fetch requests. Each entry consists of a list of event + ids to be fetched, and a deferred to be completed once the + events have been fetched. + + """ + with Measure(self._clock, "_fetch_event_list"): + try: + event_id_lists = zip(*event_list)[0] + event_ids = [ + item for sublist in event_id_lists for item in sublist + ] + + rows = self._new_transaction( + conn, "do_fetch", [], [], + self._fetch_event_rows, event_ids, + ) + + row_dict = { + r["event_id"]: r + for r in rows + } + + # We only want to resolve deferreds from the main thread + def fire(lst, res): + for ids, d in lst: + if not d.called: + try: + with PreserveLoggingContext(): + d.callback([ + res[i] + for i in ids + if i in res + ]) + except Exception: + logger.exception("Failed to callback") + with PreserveLoggingContext(): + self.hs.get_reactor().callFromThread(fire, event_list, row_dict) + except Exception as e: + logger.exception("do_fetch") + + # We only want to resolve deferreds from the main thread + def fire(evs): + for _, d in evs: + if not d.called: + with PreserveLoggingContext(): + d.errback(e) + + with PreserveLoggingContext(): + self.hs.get_reactor().callFromThread(fire, event_list) + + @defer.inlineCallbacks + def _enqueue_events(self, events, check_redacted=True, allow_rejected=False): + """Fetches events from the database using the _event_fetch_list. This + allows batch and bulk fetching of events - it allows us to fetch events + without having to create a new transaction for each request for events. + """ + if not events: + defer.returnValue({}) + + events_d = defer.Deferred() + with self._event_fetch_lock: + self._event_fetch_list.append( + (events, events_d) + ) + + self._event_fetch_lock.notify() + + if self._event_fetch_ongoing < EVENT_QUEUE_THREADS: + self._event_fetch_ongoing += 1 + should_start = True + else: + should_start = False + + if should_start: + run_as_background_process( + "fetch_events", + self.runWithConnection, + self._do_fetch, + ) + + logger.debug("Loading %d events", len(events)) + with PreserveLoggingContext(): + rows = yield events_d + logger.debug("Loaded %d events (%d rows)", len(events), len(rows)) + + if not allow_rejected: + rows[:] = [r for r in rows if not r["rejects"]] + + res = yield make_deferred_yieldable(defer.gatherResults( + [ + run_in_background( + self._get_event_from_row, + row["internal_metadata"], row["json"], row["redacts"], + rejected_reason=row["rejects"], + ) + for row in rows + ], + consumeErrors=True + )) + + defer.returnValue({ + e.event.event_id: e + for e in res if e + }) + + def _fetch_event_rows(self, txn, events): + rows = [] + N = 200 + for i in range(1 + len(events) // N): + evs = events[i * N:(i + 1) * N] + if not evs: + break + + sql = ( + "SELECT " + " e.event_id as event_id, " + " e.internal_metadata," + " e.json," + " r.redacts as redacts," + " rej.event_id as rejects " + " FROM event_json as e" + " LEFT JOIN rejections as rej USING (event_id)" + " LEFT JOIN redactions as r ON e.event_id = r.redacts" + " WHERE e.event_id IN (%s)" + ) % (",".join(["?"] * len(evs)),) + + txn.execute(sql, evs) + rows.extend(self.cursor_to_dict(txn)) + + return rows + + @defer.inlineCallbacks + def _get_event_from_row(self, internal_metadata, js, redacted, + rejected_reason=None): + with Measure(self._clock, "_get_event_from_row"): + d = json.loads(js) + internal_metadata = json.loads(internal_metadata) + + if rejected_reason: + rejected_reason = yield self._simple_select_one_onecol( + table="rejections", + keyvalues={"event_id": rejected_reason}, + retcol="reason", + desc="_get_event_from_row_rejected_reason", + ) + + original_ev = FrozenEvent( + d, + internal_metadata_dict=internal_metadata, + rejected_reason=rejected_reason, + ) + + redacted_event = None + if redacted: + redacted_event = prune_event(original_ev) + + redaction_id = yield self._simple_select_one_onecol( + table="redactions", + keyvalues={"redacts": redacted_event.event_id}, + retcol="event_id", + desc="_get_event_from_row_redactions", + ) + + redacted_event.unsigned["redacted_by"] = redaction_id + # Get the redaction event. + + because = yield self.get_event( + redaction_id, + check_redacted=False, + allow_none=True, + ) + + if because: + # It's fine to do add the event directly, since get_pdu_json + # will serialise this field correctly + redacted_event.unsigned["redacted_because"] = because + + cache_entry = _EventCacheEntry( + event=original_ev, + redacted_event=redacted_event, + ) + + self._get_event_cache.prefill((original_ev.event_id,), cache_entry) + + defer.returnValue(cache_entry) diff --git a/synapse/storage/filtering.py b/synapse/storage/filtering.py index 78b1e3094..2d5896c5b 100644 --- a/synapse/storage/filtering.py +++ b/synapse/storage/filtering.py @@ -13,14 +13,14 @@ # See the License for the specific language governing permissions and # limitations under the License. +from canonicaljson import encode_canonical_json, json + from twisted.internet import defer -from ._base import SQLBaseStore -from synapse.api.errors import SynapseError, Codes +from synapse.api.errors import Codes, SynapseError from synapse.util.caches.descriptors import cachedInlineCallbacks -from canonicaljson import encode_canonical_json -import simplejson as json +from ._base import SQLBaseStore class FilteringStore(SQLBaseStore): @@ -44,7 +44,7 @@ class FilteringStore(SQLBaseStore): desc="get_user_filter", ) - defer.returnValue(json.loads(str(def_json).decode("utf-8"))) + defer.returnValue(json.loads(bytes(def_json).decode("utf-8"))) def add_user_filter(self, user_localpart, user_filter): def_json = encode_canonical_json(user_filter) diff --git a/synapse/storage/group_server.py b/synapse/storage/group_server.py new file mode 100644 index 000000000..592d1b4c2 --- /dev/null +++ b/synapse/storage/group_server.py @@ -0,0 +1,1252 @@ +# -*- coding: utf-8 -*- +# Copyright 2017 Vector Creations Ltd +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from canonicaljson import json + +from twisted.internet import defer + +from synapse.api.errors import SynapseError + +from ._base import SQLBaseStore + +# The category ID for the "default" category. We don't store as null in the +# database to avoid the fun of null != null +_DEFAULT_CATEGORY_ID = "" +_DEFAULT_ROLE_ID = "" + + +class GroupServerStore(SQLBaseStore): + def set_group_join_policy(self, group_id, join_policy): + """Set the join policy of a group. + + join_policy can be one of: + * "invite" + * "open" + """ + return self._simple_update_one( + table="groups", + keyvalues={ + "group_id": group_id, + }, + updatevalues={ + "join_policy": join_policy, + }, + desc="set_group_join_policy", + ) + + def get_group(self, group_id): + return self._simple_select_one( + table="groups", + keyvalues={ + "group_id": group_id, + }, + retcols=( + "name", "short_description", "long_description", + "avatar_url", "is_public", "join_policy", + ), + allow_none=True, + desc="get_group", + ) + + def get_users_in_group(self, group_id, include_private=False): + # TODO: Pagination + + keyvalues = { + "group_id": group_id, + } + if not include_private: + keyvalues["is_public"] = True + + return self._simple_select_list( + table="group_users", + keyvalues=keyvalues, + retcols=("user_id", "is_public", "is_admin",), + desc="get_users_in_group", + ) + + def get_invited_users_in_group(self, group_id): + # TODO: Pagination + + return self._simple_select_onecol( + table="group_invites", + keyvalues={ + "group_id": group_id, + }, + retcol="user_id", + desc="get_invited_users_in_group", + ) + + def get_rooms_in_group(self, group_id, include_private=False): + # TODO: Pagination + + keyvalues = { + "group_id": group_id, + } + if not include_private: + keyvalues["is_public"] = True + + return self._simple_select_list( + table="group_rooms", + keyvalues=keyvalues, + retcols=("room_id", "is_public",), + desc="get_rooms_in_group", + ) + + def get_rooms_for_summary_by_category(self, group_id, include_private=False): + """Get the rooms and categories that should be included in a summary request + + Returns ([rooms], [categories]) + """ + def _get_rooms_for_summary_txn(txn): + keyvalues = { + "group_id": group_id, + } + if not include_private: + keyvalues["is_public"] = True + + sql = """ + SELECT room_id, is_public, category_id, room_order + FROM group_summary_rooms + WHERE group_id = ? + """ + + if not include_private: + sql += " AND is_public = ?" + txn.execute(sql, (group_id, True)) + else: + txn.execute(sql, (group_id,)) + + rooms = [ + { + "room_id": row[0], + "is_public": row[1], + "category_id": row[2] if row[2] != _DEFAULT_CATEGORY_ID else None, + "order": row[3], + } + for row in txn + ] + + sql = """ + SELECT category_id, is_public, profile, cat_order + FROM group_summary_room_categories + INNER JOIN group_room_categories USING (group_id, category_id) + WHERE group_id = ? + """ + + if not include_private: + sql += " AND is_public = ?" + txn.execute(sql, (group_id, True)) + else: + txn.execute(sql, (group_id,)) + + categories = { + row[0]: { + "is_public": row[1], + "profile": json.loads(row[2]), + "order": row[3], + } + for row in txn + } + + return rooms, categories + return self.runInteraction( + "get_rooms_for_summary", _get_rooms_for_summary_txn + ) + + def add_room_to_summary(self, group_id, room_id, category_id, order, is_public): + return self.runInteraction( + "add_room_to_summary", self._add_room_to_summary_txn, + group_id, room_id, category_id, order, is_public, + ) + + def _add_room_to_summary_txn(self, txn, group_id, room_id, category_id, order, + is_public): + """Add (or update) room's entry in summary. + + Args: + group_id (str) + room_id (str) + category_id (str): If not None then adds the category to the end of + the summary if its not already there. [Optional] + order (int): If not None inserts the room at that position, e.g. + an order of 1 will put the room first. Otherwise, the room gets + added to the end. + """ + room_in_group = self._simple_select_one_onecol_txn( + txn, + table="group_rooms", + keyvalues={ + "group_id": group_id, + "room_id": room_id, + }, + retcol="room_id", + allow_none=True, + ) + if not room_in_group: + raise SynapseError(400, "room not in group") + + if category_id is None: + category_id = _DEFAULT_CATEGORY_ID + else: + cat_exists = self._simple_select_one_onecol_txn( + txn, + table="group_room_categories", + keyvalues={ + "group_id": group_id, + "category_id": category_id, + }, + retcol="group_id", + allow_none=True, + ) + if not cat_exists: + raise SynapseError(400, "Category doesn't exist") + + # TODO: Check category is part of summary already + cat_exists = self._simple_select_one_onecol_txn( + txn, + table="group_summary_room_categories", + keyvalues={ + "group_id": group_id, + "category_id": category_id, + }, + retcol="group_id", + allow_none=True, + ) + if not cat_exists: + # If not, add it with an order larger than all others + txn.execute(""" + INSERT INTO group_summary_room_categories + (group_id, category_id, cat_order) + SELECT ?, ?, COALESCE(MAX(cat_order), 0) + 1 + FROM group_summary_room_categories + WHERE group_id = ? AND category_id = ? + """, (group_id, category_id, group_id, category_id)) + + existing = self._simple_select_one_txn( + txn, + table="group_summary_rooms", + keyvalues={ + "group_id": group_id, + "room_id": room_id, + "category_id": category_id, + }, + retcols=("room_order", "is_public",), + allow_none=True, + ) + + if order is not None: + # Shuffle other room orders that come after the given order + sql = """ + UPDATE group_summary_rooms SET room_order = room_order + 1 + WHERE group_id = ? AND category_id = ? AND room_order >= ? + """ + txn.execute(sql, (group_id, category_id, order,)) + elif not existing: + sql = """ + SELECT COALESCE(MAX(room_order), 0) + 1 FROM group_summary_rooms + WHERE group_id = ? AND category_id = ? + """ + txn.execute(sql, (group_id, category_id,)) + order, = txn.fetchone() + + if existing: + to_update = {} + if order is not None: + to_update["room_order"] = order + if is_public is not None: + to_update["is_public"] = is_public + self._simple_update_txn( + txn, + table="group_summary_rooms", + keyvalues={ + "group_id": group_id, + "category_id": category_id, + "room_id": room_id, + }, + values=to_update, + ) + else: + if is_public is None: + is_public = True + + self._simple_insert_txn( + txn, + table="group_summary_rooms", + values={ + "group_id": group_id, + "category_id": category_id, + "room_id": room_id, + "room_order": order, + "is_public": is_public, + }, + ) + + def remove_room_from_summary(self, group_id, room_id, category_id): + if category_id is None: + category_id = _DEFAULT_CATEGORY_ID + + return self._simple_delete( + table="group_summary_rooms", + keyvalues={ + "group_id": group_id, + "category_id": category_id, + "room_id": room_id, + }, + desc="remove_room_from_summary", + ) + + @defer.inlineCallbacks + def get_group_categories(self, group_id): + rows = yield self._simple_select_list( + table="group_room_categories", + keyvalues={ + "group_id": group_id, + }, + retcols=("category_id", "is_public", "profile"), + desc="get_group_categories", + ) + + defer.returnValue({ + row["category_id"]: { + "is_public": row["is_public"], + "profile": json.loads(row["profile"]), + } + for row in rows + }) + + @defer.inlineCallbacks + def get_group_category(self, group_id, category_id): + category = yield self._simple_select_one( + table="group_room_categories", + keyvalues={ + "group_id": group_id, + "category_id": category_id, + }, + retcols=("is_public", "profile"), + desc="get_group_category", + ) + + category["profile"] = json.loads(category["profile"]) + + defer.returnValue(category) + + def upsert_group_category(self, group_id, category_id, profile, is_public): + """Add/update room category for group + """ + insertion_values = {} + update_values = {"category_id": category_id} # This cannot be empty + + if profile is None: + insertion_values["profile"] = "{}" + else: + update_values["profile"] = json.dumps(profile) + + if is_public is None: + insertion_values["is_public"] = True + else: + update_values["is_public"] = is_public + + return self._simple_upsert( + table="group_room_categories", + keyvalues={ + "group_id": group_id, + "category_id": category_id, + }, + values=update_values, + insertion_values=insertion_values, + desc="upsert_group_category", + ) + + def remove_group_category(self, group_id, category_id): + return self._simple_delete( + table="group_room_categories", + keyvalues={ + "group_id": group_id, + "category_id": category_id, + }, + desc="remove_group_category", + ) + + @defer.inlineCallbacks + def get_group_roles(self, group_id): + rows = yield self._simple_select_list( + table="group_roles", + keyvalues={ + "group_id": group_id, + }, + retcols=("role_id", "is_public", "profile"), + desc="get_group_roles", + ) + + defer.returnValue({ + row["role_id"]: { + "is_public": row["is_public"], + "profile": json.loads(row["profile"]), + } + for row in rows + }) + + @defer.inlineCallbacks + def get_group_role(self, group_id, role_id): + role = yield self._simple_select_one( + table="group_roles", + keyvalues={ + "group_id": group_id, + "role_id": role_id, + }, + retcols=("is_public", "profile"), + desc="get_group_role", + ) + + role["profile"] = json.loads(role["profile"]) + + defer.returnValue(role) + + def upsert_group_role(self, group_id, role_id, profile, is_public): + """Add/remove user role + """ + insertion_values = {} + update_values = {"role_id": role_id} # This cannot be empty + + if profile is None: + insertion_values["profile"] = "{}" + else: + update_values["profile"] = json.dumps(profile) + + if is_public is None: + insertion_values["is_public"] = True + else: + update_values["is_public"] = is_public + + return self._simple_upsert( + table="group_roles", + keyvalues={ + "group_id": group_id, + "role_id": role_id, + }, + values=update_values, + insertion_values=insertion_values, + desc="upsert_group_role", + ) + + def remove_group_role(self, group_id, role_id): + return self._simple_delete( + table="group_roles", + keyvalues={ + "group_id": group_id, + "role_id": role_id, + }, + desc="remove_group_role", + ) + + def add_user_to_summary(self, group_id, user_id, role_id, order, is_public): + return self.runInteraction( + "add_user_to_summary", self._add_user_to_summary_txn, + group_id, user_id, role_id, order, is_public, + ) + + def _add_user_to_summary_txn(self, txn, group_id, user_id, role_id, order, + is_public): + """Add (or update) user's entry in summary. + + Args: + group_id (str) + user_id (str) + role_id (str): If not None then adds the role to the end of + the summary if its not already there. [Optional] + order (int): If not None inserts the user at that position, e.g. + an order of 1 will put the user first. Otherwise, the user gets + added to the end. + """ + user_in_group = self._simple_select_one_onecol_txn( + txn, + table="group_users", + keyvalues={ + "group_id": group_id, + "user_id": user_id, + }, + retcol="user_id", + allow_none=True, + ) + if not user_in_group: + raise SynapseError(400, "user not in group") + + if role_id is None: + role_id = _DEFAULT_ROLE_ID + else: + role_exists = self._simple_select_one_onecol_txn( + txn, + table="group_roles", + keyvalues={ + "group_id": group_id, + "role_id": role_id, + }, + retcol="group_id", + allow_none=True, + ) + if not role_exists: + raise SynapseError(400, "Role doesn't exist") + + # TODO: Check role is part of the summary already + role_exists = self._simple_select_one_onecol_txn( + txn, + table="group_summary_roles", + keyvalues={ + "group_id": group_id, + "role_id": role_id, + }, + retcol="group_id", + allow_none=True, + ) + if not role_exists: + # If not, add it with an order larger than all others + txn.execute(""" + INSERT INTO group_summary_roles + (group_id, role_id, role_order) + SELECT ?, ?, COALESCE(MAX(role_order), 0) + 1 + FROM group_summary_roles + WHERE group_id = ? AND role_id = ? + """, (group_id, role_id, group_id, role_id)) + + existing = self._simple_select_one_txn( + txn, + table="group_summary_users", + keyvalues={ + "group_id": group_id, + "user_id": user_id, + "role_id": role_id, + }, + retcols=("user_order", "is_public",), + allow_none=True, + ) + + if order is not None: + # Shuffle other users orders that come after the given order + sql = """ + UPDATE group_summary_users SET user_order = user_order + 1 + WHERE group_id = ? AND role_id = ? AND user_order >= ? + """ + txn.execute(sql, (group_id, role_id, order,)) + elif not existing: + sql = """ + SELECT COALESCE(MAX(user_order), 0) + 1 FROM group_summary_users + WHERE group_id = ? AND role_id = ? + """ + txn.execute(sql, (group_id, role_id,)) + order, = txn.fetchone() + + if existing: + to_update = {} + if order is not None: + to_update["user_order"] = order + if is_public is not None: + to_update["is_public"] = is_public + self._simple_update_txn( + txn, + table="group_summary_users", + keyvalues={ + "group_id": group_id, + "role_id": role_id, + "user_id": user_id, + }, + values=to_update, + ) + else: + if is_public is None: + is_public = True + + self._simple_insert_txn( + txn, + table="group_summary_users", + values={ + "group_id": group_id, + "role_id": role_id, + "user_id": user_id, + "user_order": order, + "is_public": is_public, + }, + ) + + def remove_user_from_summary(self, group_id, user_id, role_id): + if role_id is None: + role_id = _DEFAULT_ROLE_ID + + return self._simple_delete( + table="group_summary_users", + keyvalues={ + "group_id": group_id, + "role_id": role_id, + "user_id": user_id, + }, + desc="remove_user_from_summary", + ) + + def get_users_for_summary_by_role(self, group_id, include_private=False): + """Get the users and roles that should be included in a summary request + + Returns ([users], [roles]) + """ + def _get_users_for_summary_txn(txn): + keyvalues = { + "group_id": group_id, + } + if not include_private: + keyvalues["is_public"] = True + + sql = """ + SELECT user_id, is_public, role_id, user_order + FROM group_summary_users + WHERE group_id = ? + """ + + if not include_private: + sql += " AND is_public = ?" + txn.execute(sql, (group_id, True)) + else: + txn.execute(sql, (group_id,)) + + users = [ + { + "user_id": row[0], + "is_public": row[1], + "role_id": row[2] if row[2] != _DEFAULT_ROLE_ID else None, + "order": row[3], + } + for row in txn + ] + + sql = """ + SELECT role_id, is_public, profile, role_order + FROM group_summary_roles + INNER JOIN group_roles USING (group_id, role_id) + WHERE group_id = ? + """ + + if not include_private: + sql += " AND is_public = ?" + txn.execute(sql, (group_id, True)) + else: + txn.execute(sql, (group_id,)) + + roles = { + row[0]: { + "is_public": row[1], + "profile": json.loads(row[2]), + "order": row[3], + } + for row in txn + } + + return users, roles + return self.runInteraction( + "get_users_for_summary_by_role", _get_users_for_summary_txn + ) + + def is_user_in_group(self, user_id, group_id): + return self._simple_select_one_onecol( + table="group_users", + keyvalues={ + "group_id": group_id, + "user_id": user_id, + }, + retcol="user_id", + allow_none=True, + desc="is_user_in_group", + ).addCallback(lambda r: bool(r)) + + def is_user_admin_in_group(self, group_id, user_id): + return self._simple_select_one_onecol( + table="group_users", + keyvalues={ + "group_id": group_id, + "user_id": user_id, + }, + retcol="is_admin", + allow_none=True, + desc="is_user_admin_in_group", + ) + + def add_group_invite(self, group_id, user_id): + """Record that the group server has invited a user + """ + return self._simple_insert( + table="group_invites", + values={ + "group_id": group_id, + "user_id": user_id, + }, + desc="add_group_invite", + ) + + def is_user_invited_to_local_group(self, group_id, user_id): + """Has the group server invited a user? + """ + return self._simple_select_one_onecol( + table="group_invites", + keyvalues={ + "group_id": group_id, + "user_id": user_id, + }, + retcol="user_id", + desc="is_user_invited_to_local_group", + allow_none=True, + ) + + def get_users_membership_info_in_group(self, group_id, user_id): + """Get a dict describing the membership of a user in a group. + + Example if joined: + + { + "membership": "join", + "is_public": True, + "is_privileged": False, + } + + Returns an empty dict if the user is not join/invite/etc + """ + def _get_users_membership_in_group_txn(txn): + row = self._simple_select_one_txn( + txn, + table="group_users", + keyvalues={ + "group_id": group_id, + "user_id": user_id, + }, + retcols=("is_admin", "is_public"), + allow_none=True, + ) + + if row: + return { + "membership": "join", + "is_public": row["is_public"], + "is_privileged": row["is_admin"], + } + + row = self._simple_select_one_onecol_txn( + txn, + table="group_invites", + keyvalues={ + "group_id": group_id, + "user_id": user_id, + }, + retcol="user_id", + allow_none=True, + ) + + if row: + return { + "membership": "invite", + } + + return {} + + return self.runInteraction( + "get_users_membership_info_in_group", _get_users_membership_in_group_txn, + ) + + def add_user_to_group(self, group_id, user_id, is_admin=False, is_public=True, + local_attestation=None, remote_attestation=None): + """Add a user to the group server. + + Args: + group_id (str) + user_id (str) + is_admin (bool) + is_public (bool) + local_attestation (dict): The attestation the GS created to give + to the remote server. Optional if the user and group are on the + same server + remote_attestation (dict): The attestation given to GS by remote + server. Optional if the user and group are on the same server + """ + def _add_user_to_group_txn(txn): + self._simple_insert_txn( + txn, + table="group_users", + values={ + "group_id": group_id, + "user_id": user_id, + "is_admin": is_admin, + "is_public": is_public, + }, + ) + + self._simple_delete_txn( + txn, + table="group_invites", + keyvalues={ + "group_id": group_id, + "user_id": user_id, + }, + ) + + if local_attestation: + self._simple_insert_txn( + txn, + table="group_attestations_renewals", + values={ + "group_id": group_id, + "user_id": user_id, + "valid_until_ms": local_attestation["valid_until_ms"], + }, + ) + if remote_attestation: + self._simple_insert_txn( + txn, + table="group_attestations_remote", + values={ + "group_id": group_id, + "user_id": user_id, + "valid_until_ms": remote_attestation["valid_until_ms"], + "attestation_json": json.dumps(remote_attestation), + }, + ) + + return self.runInteraction( + "add_user_to_group", _add_user_to_group_txn + ) + + def remove_user_from_group(self, group_id, user_id): + def _remove_user_from_group_txn(txn): + self._simple_delete_txn( + txn, + table="group_users", + keyvalues={ + "group_id": group_id, + "user_id": user_id, + }, + ) + self._simple_delete_txn( + txn, + table="group_invites", + keyvalues={ + "group_id": group_id, + "user_id": user_id, + }, + ) + self._simple_delete_txn( + txn, + table="group_attestations_renewals", + keyvalues={ + "group_id": group_id, + "user_id": user_id, + }, + ) + self._simple_delete_txn( + txn, + table="group_attestations_remote", + keyvalues={ + "group_id": group_id, + "user_id": user_id, + }, + ) + self._simple_delete_txn( + txn, + table="group_summary_users", + keyvalues={ + "group_id": group_id, + "user_id": user_id, + }, + ) + return self.runInteraction("remove_user_from_group", _remove_user_from_group_txn) + + def add_room_to_group(self, group_id, room_id, is_public): + return self._simple_insert( + table="group_rooms", + values={ + "group_id": group_id, + "room_id": room_id, + "is_public": is_public, + }, + desc="add_room_to_group", + ) + + def update_room_in_group_visibility(self, group_id, room_id, is_public): + return self._simple_update( + table="group_rooms", + keyvalues={ + "group_id": group_id, + "room_id": room_id, + }, + updatevalues={ + "is_public": is_public, + }, + desc="update_room_in_group_visibility", + ) + + def remove_room_from_group(self, group_id, room_id): + def _remove_room_from_group_txn(txn): + self._simple_delete_txn( + txn, + table="group_rooms", + keyvalues={ + "group_id": group_id, + "room_id": room_id, + }, + ) + + self._simple_delete_txn( + txn, + table="group_summary_rooms", + keyvalues={ + "group_id": group_id, + "room_id": room_id, + }, + ) + return self.runInteraction( + "remove_room_from_group", _remove_room_from_group_txn, + ) + + def get_publicised_groups_for_user(self, user_id): + """Get all groups a user is publicising + """ + return self._simple_select_onecol( + table="local_group_membership", + keyvalues={ + "user_id": user_id, + "membership": "join", + "is_publicised": True, + }, + retcol="group_id", + desc="get_publicised_groups_for_user", + ) + + def update_group_publicity(self, group_id, user_id, publicise): + """Update whether the user is publicising their membership of the group + """ + return self._simple_update_one( + table="local_group_membership", + keyvalues={ + "group_id": group_id, + "user_id": user_id, + }, + updatevalues={ + "is_publicised": publicise, + }, + desc="update_group_publicity" + ) + + @defer.inlineCallbacks + def register_user_group_membership(self, group_id, user_id, membership, + is_admin=False, content={}, + local_attestation=None, + remote_attestation=None, + is_publicised=False, + ): + """Registers that a local user is a member of a (local or remote) group. + + Args: + group_id (str) + user_id (str) + membership (str) + is_admin (bool) + content (dict): Content of the membership, e.g. includes the inviter + if the user has been invited. + local_attestation (dict): If remote group then store the fact that we + have given out an attestation, else None. + remote_attestation (dict): If remote group then store the remote + attestation from the group, else None. + """ + def _register_user_group_membership_txn(txn, next_id): + # TODO: Upsert? + self._simple_delete_txn( + txn, + table="local_group_membership", + keyvalues={ + "group_id": group_id, + "user_id": user_id, + }, + ) + self._simple_insert_txn( + txn, + table="local_group_membership", + values={ + "group_id": group_id, + "user_id": user_id, + "is_admin": is_admin, + "membership": membership, + "is_publicised": is_publicised, + "content": json.dumps(content), + }, + ) + + self._simple_insert_txn( + txn, + table="local_group_updates", + values={ + "stream_id": next_id, + "group_id": group_id, + "user_id": user_id, + "type": "membership", + "content": json.dumps({"membership": membership, "content": content}), + } + ) + self._group_updates_stream_cache.entity_has_changed(user_id, next_id) + + # TODO: Insert profile to ensure it comes down stream if its a join. + + if membership == "join": + if local_attestation: + self._simple_insert_txn( + txn, + table="group_attestations_renewals", + values={ + "group_id": group_id, + "user_id": user_id, + "valid_until_ms": local_attestation["valid_until_ms"], + } + ) + if remote_attestation: + self._simple_insert_txn( + txn, + table="group_attestations_remote", + values={ + "group_id": group_id, + "user_id": user_id, + "valid_until_ms": remote_attestation["valid_until_ms"], + "attestation_json": json.dumps(remote_attestation), + } + ) + else: + self._simple_delete_txn( + txn, + table="group_attestations_renewals", + keyvalues={ + "group_id": group_id, + "user_id": user_id, + }, + ) + self._simple_delete_txn( + txn, + table="group_attestations_remote", + keyvalues={ + "group_id": group_id, + "user_id": user_id, + }, + ) + + return next_id + + with self._group_updates_id_gen.get_next() as next_id: + res = yield self.runInteraction( + "register_user_group_membership", + _register_user_group_membership_txn, next_id, + ) + defer.returnValue(res) + + @defer.inlineCallbacks + def create_group(self, group_id, user_id, name, avatar_url, short_description, + long_description,): + yield self._simple_insert( + table="groups", + values={ + "group_id": group_id, + "name": name, + "avatar_url": avatar_url, + "short_description": short_description, + "long_description": long_description, + "is_public": True, + }, + desc="create_group", + ) + + @defer.inlineCallbacks + def update_group_profile(self, group_id, profile,): + yield self._simple_update_one( + table="groups", + keyvalues={ + "group_id": group_id, + }, + updatevalues=profile, + desc="update_group_profile", + ) + + def get_attestations_need_renewals(self, valid_until_ms): + """Get all attestations that need to be renewed until givent time + """ + def _get_attestations_need_renewals_txn(txn): + sql = """ + SELECT group_id, user_id FROM group_attestations_renewals + WHERE valid_until_ms <= ? + """ + txn.execute(sql, (valid_until_ms,)) + return self.cursor_to_dict(txn) + return self.runInteraction( + "get_attestations_need_renewals", _get_attestations_need_renewals_txn + ) + + def update_attestation_renewal(self, group_id, user_id, attestation): + """Update an attestation that we have renewed + """ + return self._simple_update_one( + table="group_attestations_renewals", + keyvalues={ + "group_id": group_id, + "user_id": user_id, + }, + updatevalues={ + "valid_until_ms": attestation["valid_until_ms"], + }, + desc="update_attestation_renewal", + ) + + def update_remote_attestion(self, group_id, user_id, attestation): + """Update an attestation that a remote has renewed + """ + return self._simple_update_one( + table="group_attestations_remote", + keyvalues={ + "group_id": group_id, + "user_id": user_id, + }, + updatevalues={ + "valid_until_ms": attestation["valid_until_ms"], + "attestation_json": json.dumps(attestation) + }, + desc="update_remote_attestion", + ) + + def remove_attestation_renewal(self, group_id, user_id): + """Remove an attestation that we thought we should renew, but actually + shouldn't. Ideally this would never get called as we would never + incorrectly try and do attestations for local users on local groups. + + Args: + group_id (str) + user_id (str) + """ + return self._simple_delete( + table="group_attestations_renewals", + keyvalues={ + "group_id": group_id, + "user_id": user_id, + }, + desc="remove_attestation_renewal", + ) + + @defer.inlineCallbacks + def get_remote_attestation(self, group_id, user_id): + """Get the attestation that proves the remote agrees that the user is + in the group. + """ + row = yield self._simple_select_one( + table="group_attestations_remote", + keyvalues={ + "group_id": group_id, + "user_id": user_id, + }, + retcols=("valid_until_ms", "attestation_json"), + desc="get_remote_attestation", + allow_none=True, + ) + + now = int(self._clock.time_msec()) + if row and now < row["valid_until_ms"]: + defer.returnValue(json.loads(row["attestation_json"])) + + defer.returnValue(None) + + def get_joined_groups(self, user_id): + return self._simple_select_onecol( + table="local_group_membership", + keyvalues={ + "user_id": user_id, + "membership": "join", + }, + retcol="group_id", + desc="get_joined_groups", + ) + + def get_all_groups_for_user(self, user_id, now_token): + def _get_all_groups_for_user_txn(txn): + sql = """ + SELECT group_id, type, membership, u.content + FROM local_group_updates AS u + INNER JOIN local_group_membership USING (group_id, user_id) + WHERE user_id = ? AND membership != 'leave' + AND stream_id <= ? + """ + txn.execute(sql, (user_id, now_token,)) + return [ + { + "group_id": row[0], + "type": row[1], + "membership": row[2], + "content": json.loads(row[3]), + } + for row in txn + ] + return self.runInteraction( + "get_all_groups_for_user", _get_all_groups_for_user_txn, + ) + + def get_groups_changes_for_user(self, user_id, from_token, to_token): + from_token = int(from_token) + has_changed = self._group_updates_stream_cache.has_entity_changed( + user_id, from_token, + ) + if not has_changed: + return [] + + def _get_groups_changes_for_user_txn(txn): + sql = """ + SELECT group_id, membership, type, u.content + FROM local_group_updates AS u + INNER JOIN local_group_membership USING (group_id, user_id) + WHERE user_id = ? AND ? < stream_id AND stream_id <= ? + """ + txn.execute(sql, (user_id, from_token, to_token,)) + return [{ + "group_id": group_id, + "membership": membership, + "type": gtype, + "content": json.loads(content_json), + } for group_id, membership, gtype, content_json in txn] + return self.runInteraction( + "get_groups_changes_for_user", _get_groups_changes_for_user_txn, + ) + + def get_all_groups_changes(self, from_token, to_token, limit): + from_token = int(from_token) + has_changed = self._group_updates_stream_cache.has_any_entity_changed( + from_token, + ) + if not has_changed: + return [] + + def _get_all_groups_changes_txn(txn): + sql = """ + SELECT stream_id, group_id, user_id, type, content + FROM local_group_updates + WHERE ? < stream_id AND stream_id <= ? + LIMIT ? + """ + txn.execute(sql, (from_token, to_token, limit,)) + return [( + stream_id, + group_id, + user_id, + gtype, + json.loads(content_json), + ) for stream_id, group_id, user_id, gtype, content_json in txn] + return self.runInteraction( + "get_all_groups_changes", _get_all_groups_changes_txn, + ) + + def get_group_stream_token(self): + return self._group_updates_id_gen.get_current_token() diff --git a/synapse/storage/keys.py b/synapse/storage/keys.py index 3b5e0a4fb..f54797760 100644 --- a/synapse/storage/keys.py +++ b/synapse/storage/keys.py @@ -13,19 +13,29 @@ # See the License for the specific language governing permissions and # limitations under the License. -from ._base import SQLBaseStore -from synapse.util.caches.descriptors import cachedInlineCallbacks - -from twisted.internet import defer - -import OpenSSL -from signedjson.key import decode_verify_key_bytes import hashlib - import logging +import six + +from signedjson.key import decode_verify_key_bytes + +import OpenSSL +from twisted.internet import defer + +from synapse.util.caches.descriptors import cachedInlineCallbacks + +from ._base import SQLBaseStore + logger = logging.getLogger(__name__) +# py2 sqlite has buffer hardcoded as only binary type, so we must use it, +# despite being deprecated and removed in favor of memoryview +if six.PY2: + db_binary_type = buffer +else: + db_binary_type = memoryview + class KeyStore(SQLBaseStore): """Persistence for signature verification keys and tls X.509 certificates @@ -72,7 +82,7 @@ class KeyStore(SQLBaseStore): values={ "from_server": from_server, "ts_added_ms": time_now_ms, - "tls_certificate": buffer(tls_certificate_bytes), + "tls_certificate": db_binary_type(tls_certificate_bytes), }, desc="store_server_certificate", ) @@ -92,7 +102,7 @@ class KeyStore(SQLBaseStore): if verify_key_bytes: defer.returnValue(decode_verify_key_bytes( - key_id, str(verify_key_bytes) + key_id, bytes(verify_key_bytes) )) @defer.inlineCallbacks @@ -113,30 +123,37 @@ class KeyStore(SQLBaseStore): keys[key_id] = key defer.returnValue(keys) - @defer.inlineCallbacks def store_server_verify_key(self, server_name, from_server, time_now_ms, verify_key): """Stores a NACL verification key for the given server. Args: server_name (str): The name of the server. - key_id (str): The version of the key for the server. from_server (str): Where the verification key was looked up - ts_now_ms (int): The time now in milliseconds - verification_key (VerifyKey): The NACL verify key. + time_now_ms (int): The time now in milliseconds + verify_key (nacl.signing.VerifyKey): The NACL verify key. """ - yield self._simple_upsert( - table="server_signature_keys", - keyvalues={ - "server_name": server_name, - "key_id": "%s:%s" % (verify_key.alg, verify_key.version), - }, - values={ - "from_server": from_server, - "ts_added_ms": time_now_ms, - "verify_key": buffer(verify_key.encode()), - }, - desc="store_server_verify_key", - ) + key_id = "%s:%s" % (verify_key.alg, verify_key.version) + + def _txn(txn): + self._simple_upsert_txn( + txn, + table="server_signature_keys", + keyvalues={ + "server_name": server_name, + "key_id": key_id, + }, + values={ + "from_server": from_server, + "ts_added_ms": time_now_ms, + "verify_key": db_binary_type(verify_key.encode()), + }, + ) + txn.call_after( + self._get_server_verify_key.invalidate, + (server_name, key_id) + ) + + return self.runInteraction("store_server_verify_key", _txn) def store_server_keys_json(self, server_name, key_id, from_server, ts_now_ms, ts_expires_ms, key_json_bytes): @@ -165,7 +182,7 @@ class KeyStore(SQLBaseStore): "from_server": from_server, "ts_added_ms": ts_now_ms, "ts_valid_until_ms": ts_expires_ms, - "key_json": buffer(key_json_bytes), + "key_json": db_binary_type(key_json_bytes), }, desc="store_server_keys_json", ) diff --git a/synapse/storage/media_repository.py b/synapse/storage/media_repository.py index 82bb61b81..e6cdbb054 100644 --- a/synapse/storage/media_repository.py +++ b/synapse/storage/media_repository.py @@ -12,15 +12,22 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. - -from ._base import SQLBaseStore +from synapse.storage.background_updates import BackgroundUpdateStore -class MediaRepositoryStore(SQLBaseStore): +class MediaRepositoryStore(BackgroundUpdateStore): """Persistence for attachments and avatars""" - def get_default_thumbnails(self, top_level_type, sub_type): - return [] + def __init__(self, db_conn, hs): + super(MediaRepositoryStore, self).__init__(db_conn, hs) + + self.register_background_index_update( + update_name='local_media_repository_url_idx', + index_name='local_media_repository_url_idx', + table='local_media_repository', + columns=['created_ts'], + where_clause='url_cache IS NOT NULL', + ) def get_local_media(self, media_id): """Get the metadata for a local piece of media @@ -62,7 +69,7 @@ class MediaRepositoryStore(SQLBaseStore): def get_url_cache_txn(txn): # get the most recently cached result (relative to the given ts) sql = ( - "SELECT response_code, etag, expires, og, media_id, download_ts" + "SELECT response_code, etag, expires_ts, og, media_id, download_ts" " FROM local_media_repository_url_cache" " WHERE url = ? AND download_ts <= ?" " ORDER BY download_ts DESC LIMIT 1" @@ -74,7 +81,7 @@ class MediaRepositoryStore(SQLBaseStore): # ...or if we've requested a timestamp older than the oldest # copy in the cache, return the oldest copy (if any) sql = ( - "SELECT response_code, etag, expires, og, media_id, download_ts" + "SELECT response_code, etag, expires_ts, og, media_id, download_ts" " FROM local_media_repository_url_cache" " WHERE url = ? AND download_ts > ?" " ORDER BY download_ts ASC LIMIT 1" @@ -86,14 +93,14 @@ class MediaRepositoryStore(SQLBaseStore): return None return dict(zip(( - 'response_code', 'etag', 'expires', 'og', 'media_id', 'download_ts' + 'response_code', 'etag', 'expires_ts', 'og', 'media_id', 'download_ts' ), row)) return self.runInteraction( "get_url_cache", get_url_cache_txn ) - def store_url_cache(self, url, response_code, etag, expires, og, media_id, + def store_url_cache(self, url, response_code, etag, expires_ts, og, media_id, download_ts): return self._simple_insert( "local_media_repository_url_cache", @@ -101,7 +108,7 @@ class MediaRepositoryStore(SQLBaseStore): "url": url, "response_code": response_code, "etag": etag, - "expires": expires, + "expires_ts": expires_ts, "og": og, "media_id": media_id, "download_ts": download_ts, @@ -166,7 +173,14 @@ class MediaRepositoryStore(SQLBaseStore): desc="store_cached_remote_media", ) - def update_cached_last_access_time(self, origin_id_tuples, time_ts): + def update_cached_last_access_time(self, local_media, remote_media, time_ms): + """Updates the last access time of the given media + + Args: + local_media (iterable[str]): Set of media_ids + remote_media (iterable[(str, str)]): Set of (server_name, media_id) + time_ms: Current time in milliseconds + """ def update_cache_txn(txn): sql = ( "UPDATE remote_media_cache SET last_access_ts = ?" @@ -174,8 +188,18 @@ class MediaRepositoryStore(SQLBaseStore): ) txn.executemany(sql, ( - (time_ts, media_origin, media_id) - for media_origin, media_id in origin_id_tuples + (time_ms, media_origin, media_id) + for media_origin, media_id in remote_media + )) + + sql = ( + "UPDATE local_media_repository SET last_access_ts = ?" + " WHERE media_id = ?" + ) + + txn.executemany(sql, ( + (time_ms, media_id) + for media_id in local_media )) return self.runInteraction("update_cached_last_access_time", update_cache_txn) @@ -238,3 +262,70 @@ class MediaRepositoryStore(SQLBaseStore): }, ) return self.runInteraction("delete_remote_media", delete_remote_media_txn) + + def get_expired_url_cache(self, now_ts): + sql = ( + "SELECT media_id FROM local_media_repository_url_cache" + " WHERE expires_ts < ?" + " ORDER BY expires_ts ASC" + " LIMIT 500" + ) + + def _get_expired_url_cache_txn(txn): + txn.execute(sql, (now_ts,)) + return [row[0] for row in txn] + + return self.runInteraction("get_expired_url_cache", _get_expired_url_cache_txn) + + def delete_url_cache(self, media_ids): + if len(media_ids) == 0: + return + + sql = ( + "DELETE FROM local_media_repository_url_cache" + " WHERE media_id = ?" + ) + + def _delete_url_cache_txn(txn): + txn.executemany(sql, [(media_id,) for media_id in media_ids]) + + return self.runInteraction("delete_url_cache", _delete_url_cache_txn) + + def get_url_cache_media_before(self, before_ts): + sql = ( + "SELECT media_id FROM local_media_repository" + " WHERE created_ts < ? AND url_cache IS NOT NULL" + " ORDER BY created_ts ASC" + " LIMIT 500" + ) + + def _get_url_cache_media_before_txn(txn): + txn.execute(sql, (before_ts,)) + return [row[0] for row in txn] + + return self.runInteraction( + "get_url_cache_media_before", _get_url_cache_media_before_txn, + ) + + def delete_url_cache_media(self, media_ids): + if len(media_ids) == 0: + return + + def _delete_url_cache_media_txn(txn): + sql = ( + "DELETE FROM local_media_repository" + " WHERE media_id = ?" + ) + + txn.executemany(sql, [(media_id,) for media_id in media_ids]) + + sql = ( + "DELETE FROM local_media_repository_thumbnails" + " WHERE media_id = ?" + ) + + txn.executemany(sql, [(media_id,) for media_id in media_ids]) + + return self.runInteraction( + "delete_url_cache_media", _delete_url_cache_media_txn, + ) diff --git a/synapse/storage/prepare_database.py b/synapse/storage/prepare_database.py index 72b670b83..b290f834b 100644 --- a/synapse/storage/prepare_database.py +++ b/synapse/storage/prepare_database.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2014 - 2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -19,13 +20,12 @@ import logging import os import re - logger = logging.getLogger(__name__) # Remember to update this number every time a change is made to database # schema files, so the users will be informed on server restarts. -SCHEMA_VERSION = 43 +SCHEMA_VERSION = 50 dir_path = os.path.abspath(os.path.dirname(__file__)) @@ -44,6 +44,13 @@ def prepare_database(db_conn, database_engine, config): If `config` is None then prepare_database will assert that no upgrade is necessary, *or* will create a fresh database if the database is empty. + + Args: + db_conn: + database_engine: + config (synapse.config.homeserver.HomeServerConfig|None): + application config, or None if we are connecting to an existing + database which we expect to be configured already """ try: cur = db_conn.cursor() @@ -64,9 +71,13 @@ def prepare_database(db_conn, database_engine, config): else: _setup_new_database(cur, database_engine) + # check if any of our configured dynamic modules want a database + if config is not None: + _apply_module_schemas(cur, database_engine, config) + cur.close() db_conn.commit() - except: + except Exception: db_conn.rollback() raise @@ -283,6 +294,65 @@ def _upgrade_existing_database(cur, current_version, applied_delta_files, ) +def _apply_module_schemas(txn, database_engine, config): + """Apply the module schemas for the dynamic modules, if any + + Args: + cur: database cursor + database_engine: synapse database engine class + config (synapse.config.homeserver.HomeServerConfig): + application config + """ + for (mod, _config) in config.password_providers: + if not hasattr(mod, 'get_db_schema_files'): + continue + modname = ".".join((mod.__module__, mod.__name__)) + _apply_module_schema_files( + txn, database_engine, modname, mod.get_db_schema_files(), + ) + + +def _apply_module_schema_files(cur, database_engine, modname, names_and_streams): + """Apply the module schemas for a single module + + Args: + cur: database cursor + database_engine: synapse database engine class + modname (str): fully qualified name of the module + names_and_streams (Iterable[(str, file)]): the names and streams of + schemas to be applied + """ + cur.execute( + database_engine.convert_param_style( + "SELECT file FROM applied_module_schemas WHERE module_name = ?" + ), + (modname,) + ) + applied_deltas = set(d for d, in cur) + for (name, stream) in names_and_streams: + if name in applied_deltas: + continue + + root_name, ext = os.path.splitext(name) + if ext != '.sql': + raise PrepareDatabaseException( + "only .sql files are currently supported for module schemas", + ) + + logger.info("applying schema %s for %s", name, modname) + for statement in get_statements(stream): + cur.execute(statement) + + # Mark as done. + cur.execute( + database_engine.convert_param_style( + "INSERT INTO applied_module_schemas (module_name, file)" + " VALUES (?,?)", + ), + (modname, name) + ) + + def get_statements(f): statement_buffer = "" in_comment = False # If we're in a /* ... */ style comment diff --git a/synapse/storage/presence.py b/synapse/storage/presence.py index 9e9d3c259..a0c7a0dc8 100644 --- a/synapse/storage/presence.py +++ b/synapse/storage/presence.py @@ -13,12 +13,15 @@ # See the License for the specific language governing permissions and # limitations under the License. -from ._base import SQLBaseStore +from collections import namedtuple + +from twisted.internet import defer + from synapse.api.constants import PresenceState +from synapse.util import batch_iter from synapse.util.caches.descriptors import cached, cachedInlineCallbacks, cachedList -from collections import namedtuple -from twisted.internet import defer +from ._base import SQLBaseStore class UserPresenceState(namedtuple("UserPresenceState", @@ -115,11 +118,7 @@ class PresenceStore(SQLBaseStore): " AND user_id IN (%s)" ) - batches = ( - presence_states[i:i + 50] - for i in xrange(0, len(presence_states), 50) - ) - for states in batches: + for states in batch_iter(presence_states, 50): args = [stream_id] args.extend(s.user_id for s in states) txn.execute( diff --git a/synapse/storage/profile.py b/synapse/storage/profile.py index 26a40905a..60295da25 100644 --- a/synapse/storage/profile.py +++ b/synapse/storage/profile.py @@ -13,15 +13,37 @@ # See the License for the specific language governing permissions and # limitations under the License. +from twisted.internet import defer + +from synapse.api.errors import StoreError +from synapse.storage.roommember import ProfileInfo + from ._base import SQLBaseStore -class ProfileStore(SQLBaseStore): - def create_profile(self, user_localpart): - return self._simple_insert( - table="profiles", - values={"user_id": user_localpart}, - desc="create_profile", +class ProfileWorkerStore(SQLBaseStore): + @defer.inlineCallbacks + def get_profileinfo(self, user_localpart): + try: + profile = yield self._simple_select_one( + table="profiles", + keyvalues={"user_id": user_localpart}, + retcols=("displayname", "avatar_url"), + desc="get_profileinfo", + ) + except StoreError as e: + if e.code == 404: + # no match + defer.returnValue(ProfileInfo(None, None)) + return + else: + raise + + defer.returnValue( + ProfileInfo( + avatar_url=profile['avatar_url'], + display_name=profile['displayname'], + ) ) def get_profile_displayname(self, user_localpart): @@ -32,14 +54,6 @@ class ProfileStore(SQLBaseStore): desc="get_profile_displayname", ) - def set_profile_displayname(self, user_localpart, new_displayname): - return self._simple_update_one( - table="profiles", - keyvalues={"user_id": user_localpart}, - updatevalues={"displayname": new_displayname}, - desc="set_profile_displayname", - ) - def get_profile_avatar_url(self, user_localpart): return self._simple_select_one_onecol( table="profiles", @@ -48,6 +62,32 @@ class ProfileStore(SQLBaseStore): desc="get_profile_avatar_url", ) + def get_from_remote_profile_cache(self, user_id): + return self._simple_select_one( + table="remote_profile_cache", + keyvalues={"user_id": user_id}, + retcols=("displayname", "avatar_url",), + allow_none=True, + desc="get_from_remote_profile_cache", + ) + + +class ProfileStore(ProfileWorkerStore): + def create_profile(self, user_localpart): + return self._simple_insert( + table="profiles", + values={"user_id": user_localpart}, + desc="create_profile", + ) + + def set_profile_displayname(self, user_localpart, new_displayname): + return self._simple_update_one( + table="profiles", + keyvalues={"user_id": user_localpart}, + updatevalues={"displayname": new_displayname}, + desc="set_profile_displayname", + ) + def set_profile_avatar_url(self, user_localpart, new_avatar_url): return self._simple_update_one( table="profiles", @@ -55,3 +95,90 @@ class ProfileStore(SQLBaseStore): updatevalues={"avatar_url": new_avatar_url}, desc="set_profile_avatar_url", ) + + def add_remote_profile_cache(self, user_id, displayname, avatar_url): + """Ensure we are caching the remote user's profiles. + + This should only be called when `is_subscribed_remote_profile_for_user` + would return true for the user. + """ + return self._simple_upsert( + table="remote_profile_cache", + keyvalues={"user_id": user_id}, + values={ + "displayname": displayname, + "avatar_url": avatar_url, + "last_check": self._clock.time_msec(), + }, + desc="add_remote_profile_cache", + ) + + def update_remote_profile_cache(self, user_id, displayname, avatar_url): + return self._simple_update( + table="remote_profile_cache", + keyvalues={"user_id": user_id}, + values={ + "displayname": displayname, + "avatar_url": avatar_url, + "last_check": self._clock.time_msec(), + }, + desc="update_remote_profile_cache", + ) + + @defer.inlineCallbacks + def maybe_delete_remote_profile_cache(self, user_id): + """Check if we still care about the remote user's profile, and if we + don't then remove their profile from the cache + """ + subscribed = yield self.is_subscribed_remote_profile_for_user(user_id) + if not subscribed: + yield self._simple_delete( + table="remote_profile_cache", + keyvalues={"user_id": user_id}, + desc="delete_remote_profile_cache", + ) + + def get_remote_profile_cache_entries_that_expire(self, last_checked): + """Get all users who haven't been checked since `last_checked` + """ + def _get_remote_profile_cache_entries_that_expire_txn(txn): + sql = """ + SELECT user_id, displayname, avatar_url + FROM remote_profile_cache + WHERE last_check < ? + """ + + txn.execute(sql, (last_checked,)) + + return self.cursor_to_dict(txn) + + return self.runInteraction( + "get_remote_profile_cache_entries_that_expire", + _get_remote_profile_cache_entries_that_expire_txn, + ) + + @defer.inlineCallbacks + def is_subscribed_remote_profile_for_user(self, user_id): + """Check whether we are interested in a remote user's profile. + """ + res = yield self._simple_select_one_onecol( + table="group_users", + keyvalues={"user_id": user_id}, + retcol="user_id", + allow_none=True, + desc="should_update_remote_profile_cache_for_user", + ) + + if res: + defer.returnValue(True) + + res = yield self._simple_select_one_onecol( + table="group_invites", + keyvalues={"user_id": user_id}, + retcol="user_id", + allow_none=True, + desc="should_update_remote_profile_cache_for_user", + ) + + if res: + defer.returnValue(True) diff --git a/synapse/storage/push_rule.py b/synapse/storage/push_rule.py index 8758b1c0c..6a5028961 100644 --- a/synapse/storage/push_rule.py +++ b/synapse/storage/push_rule.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2014-2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -13,14 +14,22 @@ # See the License for the specific language governing permissions and # limitations under the License. -from ._base import SQLBaseStore -from synapse.util.caches.descriptors import cachedInlineCallbacks, cachedList -from synapse.push.baserules import list_with_base_rules -from synapse.api.constants import EventTypes +import abc +import logging + +from canonicaljson import json + from twisted.internet import defer -import logging -import simplejson as json +from synapse.push.baserules import list_with_base_rules +from synapse.storage.appservice import ApplicationServiceWorkerStore +from synapse.storage.pusher import PusherWorkerStore +from synapse.storage.receipts import ReceiptsWorkerStore +from synapse.storage.roommember import RoomMemberWorkerStore +from synapse.util.caches.descriptors import cachedInlineCallbacks, cachedList +from synapse.util.caches.stream_change_cache import StreamChangeCache + +from ._base import SQLBaseStore logger = logging.getLogger(__name__) @@ -48,7 +57,43 @@ def _load_rules(rawrules, enabled_map): return rules -class PushRuleStore(SQLBaseStore): +class PushRulesWorkerStore(ApplicationServiceWorkerStore, + ReceiptsWorkerStore, + PusherWorkerStore, + RoomMemberWorkerStore, + SQLBaseStore): + """This is an abstract base class where subclasses must implement + `get_max_push_rules_stream_id` which can be called in the initializer. + """ + + # This ABCMeta metaclass ensures that we cannot be instantiated without + # the abstract methods being implemented. + __metaclass__ = abc.ABCMeta + + def __init__(self, db_conn, hs): + super(PushRulesWorkerStore, self).__init__(db_conn, hs) + + push_rules_prefill, push_rules_id = self._get_cache_dict( + db_conn, "push_rules_stream", + entity_column="user_id", + stream_column="stream_id", + max_value=self.get_max_push_rules_stream_id(), + ) + + self.push_rules_stream_cache = StreamChangeCache( + "PushRulesStreamChangeCache", push_rules_id, + prefilled_cache=push_rules_prefill, + ) + + @abc.abstractmethod + def get_max_push_rules_stream_id(self): + """Get the position of the push rules stream. + + Returns: + int + """ + raise NotImplementedError() + @cachedInlineCallbacks(max_entries=5000) def get_push_rules_for_user(self, user_id): rows = yield self._simple_select_list( @@ -89,6 +134,22 @@ class PushRuleStore(SQLBaseStore): r['rule_id']: False if r['enabled'] == 0 else True for r in results }) + def have_push_rules_changed_for_user(self, user_id, last_id): + if not self.push_rules_stream_cache.has_entity_changed(user_id, last_id): + return defer.succeed(False) + else: + def have_push_rules_changed_txn(txn): + sql = ( + "SELECT COUNT(stream_id) FROM push_rules_stream" + " WHERE user_id = ? AND ? < stream_id" + ) + txn.execute(sql, (user_id, last_id)) + count, = txn.fetchone() + return bool(count) + return self.runInteraction( + "have_push_rules_changed", have_push_rules_changed_txn + ) + @cachedList(cached_method_name="get_push_rules_for_user", list_name="user_ids", num_args=1, inlineCallbacks=True) def bulk_get_push_rules(self, user_ids): @@ -124,6 +185,7 @@ class PushRuleStore(SQLBaseStore): defer.returnValue(results) + @defer.inlineCallbacks def bulk_get_push_rules_for_room(self, event, context): state_group = context.state_group if not state_group: @@ -133,9 +195,11 @@ class PushRuleStore(SQLBaseStore): # To do this we set the state_group to a new object as object() != object() state_group = object() - return self._bulk_get_push_rules_for_room( - event.room_id, state_group, context.current_state_ids, event=event + current_state_ids = yield context.get_current_state_ids(self) + result = yield self._bulk_get_push_rules_for_room( + event.room_id, state_group, current_state_ids, event=event ) + defer.returnValue(result) @cachedInlineCallbacks(num_args=2, cache_context=True) def _bulk_get_push_rules_for_room(self, room_id, state_group, current_state_ids, @@ -185,18 +249,6 @@ class PushRuleStore(SQLBaseStore): if uid in local_users_in_room: user_ids.add(uid) - forgotten = yield self.who_forgot_in_room( - event.room_id, on_invalidate=cache_context.invalidate, - ) - - for row in forgotten: - user_id = row["user_id"] - event_id = row["event_id"] - - mem_id = current_state_ids.get((EventTypes.Member, user_id), None) - if event_id == mem_id: - user_ids.discard(user_id) - rules_by_user = yield self.bulk_get_push_rules( user_ids, on_invalidate=cache_context.invalidate, ) @@ -228,6 +280,8 @@ class PushRuleStore(SQLBaseStore): results.setdefault(row['user_name'], {})[row['rule_id']] = enabled defer.returnValue(results) + +class PushRuleStore(PushRulesWorkerStore): @defer.inlineCallbacks def add_push_rule( self, user_id, rule_id, priority_class, conditions, actions, @@ -526,21 +580,8 @@ class PushRuleStore(SQLBaseStore): room stream ordering it corresponds to.""" return self._push_rules_stream_id_gen.get_current_token() - def have_push_rules_changed_for_user(self, user_id, last_id): - if not self.push_rules_stream_cache.has_entity_changed(user_id, last_id): - return defer.succeed(False) - else: - def have_push_rules_changed_txn(txn): - sql = ( - "SELECT COUNT(stream_id) FROM push_rules_stream" - " WHERE user_id = ? AND ? < stream_id" - ) - txn.execute(sql, (user_id, last_id)) - count, = txn.fetchone() - return bool(count) - return self.runInteraction( - "have_push_rules_changed", have_push_rules_changed_txn - ) + def get_max_push_rules_stream_id(self): + return self.get_push_rules_stream_token()[0] class RuleNotFoundException(Exception): diff --git a/synapse/storage/pusher.py b/synapse/storage/pusher.py index 34d2f82b7..8443bd4c1 100644 --- a/synapse/storage/pusher.py +++ b/synapse/storage/pusher.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2014-2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -13,21 +14,21 @@ # See the License for the specific language governing permissions and # limitations under the License. -from ._base import SQLBaseStore -from twisted.internet import defer +import logging +import types -from canonicaljson import encode_canonical_json +from canonicaljson import encode_canonical_json, json + +from twisted.internet import defer from synapse.util.caches.descriptors import cachedInlineCallbacks, cachedList -import logging -import simplejson as json -import types +from ._base import SQLBaseStore logger = logging.getLogger(__name__) -class PusherStore(SQLBaseStore): +class PusherWorkerStore(SQLBaseStore): def _decode_pushers_rows(self, rows): for r in rows: dataJson = r['data'] @@ -102,9 +103,6 @@ class PusherStore(SQLBaseStore): rows = yield self.runInteraction("get_all_pushers", get_pushers) defer.returnValue(rows) - def get_pushers_stream_token(self): - return self._pushers_id_gen.get_current_token() - def get_all_updated_pushers(self, last_id, current_id, limit): if last_id == current_id: return defer.succeed(([], [])) @@ -198,56 +196,74 @@ class PusherStore(SQLBaseStore): defer.returnValue(result) + +class PusherStore(PusherWorkerStore): + def get_pushers_stream_token(self): + return self._pushers_id_gen.get_current_token() + @defer.inlineCallbacks def add_pusher(self, user_id, access_token, kind, app_id, app_display_name, device_display_name, pushkey, pushkey_ts, lang, data, last_stream_ordering, profile_tag=""): with self._pushers_id_gen.get_next() as stream_id: - def f(txn): - newly_inserted = self._simple_upsert_txn( - txn, - "pushers", - { - "app_id": app_id, - "pushkey": pushkey, - "user_name": user_id, - }, - { - "access_token": access_token, - "kind": kind, - "app_display_name": app_display_name, - "device_display_name": device_display_name, - "ts": pushkey_ts, - "lang": lang, - "data": encode_canonical_json(data), - "last_stream_ordering": last_stream_ordering, - "profile_tag": profile_tag, - "id": stream_id, - }, - ) - if newly_inserted: - # get_if_user_has_pusher only cares if the user has - # at least *one* pusher. - txn.call_after(self.get_if_user_has_pusher.invalidate, (user_id,)) + # no need to lock because `pushers` has a unique key on + # (app_id, pushkey, user_name) so _simple_upsert will retry + newly_inserted = yield self._simple_upsert( + table="pushers", + keyvalues={ + "app_id": app_id, + "pushkey": pushkey, + "user_name": user_id, + }, + values={ + "access_token": access_token, + "kind": kind, + "app_display_name": app_display_name, + "device_display_name": device_display_name, + "ts": pushkey_ts, + "lang": lang, + "data": encode_canonical_json(data), + "last_stream_ordering": last_stream_ordering, + "profile_tag": profile_tag, + "id": stream_id, + }, + desc="add_pusher", + lock=False, + ) - yield self.runInteraction("add_pusher", f) + if newly_inserted: + yield self.runInteraction( + "add_pusher", + self._invalidate_cache_and_stream, + self.get_if_user_has_pusher, (user_id,) + ) @defer.inlineCallbacks def delete_pusher_by_app_id_pushkey_user_id(self, app_id, pushkey, user_id): def delete_pusher_txn(txn, stream_id): - txn.call_after(self.get_if_user_has_pusher.invalidate, (user_id,)) + self._invalidate_cache_and_stream( + txn, self.get_if_user_has_pusher, (user_id,) + ) self._simple_delete_one_txn( txn, "pushers", {"app_id": app_id, "pushkey": pushkey, "user_name": user_id} ) - self._simple_upsert_txn( + + # it's possible for us to end up with duplicate rows for + # (app_id, pushkey, user_id) at different stream_ids, but that + # doesn't really matter. + self._simple_insert_txn( txn, - "deleted_pushers", - {"app_id": app_id, "pushkey": pushkey, "user_id": user_id}, - {"stream_id": stream_id}, + table="deleted_pushers", + values={ + "stream_id": stream_id, + "app_id": app_id, + "pushkey": pushkey, + "user_id": user_id, + }, ) with self._pushers_id_gen.get_next() as stream_id: @@ -310,9 +326,12 @@ class PusherStore(SQLBaseStore): @defer.inlineCallbacks def set_throttle_params(self, pusher_id, room_id, params): + # no need to lock because `pusher_throttle` has a primary key on + # (pusher, room_id) so _simple_upsert will retry yield self._simple_upsert( "pusher_throttle", {"pusher": pusher_id, "room_id": room_id}, params, - desc="set_throttle_params" + desc="set_throttle_params", + lock=False, ) diff --git a/synapse/storage/receipts.py b/synapse/storage/receipts.py index f42b8014c..0ac665e96 100644 --- a/synapse/storage/receipts.py +++ b/synapse/storage/receipts.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2014-2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -13,52 +14,52 @@ # See the License for the specific language governing permissions and # limitations under the License. -from ._base import SQLBaseStore -from synapse.util.caches.descriptors import cachedInlineCallbacks, cachedList, cached -from synapse.util.caches.stream_change_cache import StreamChangeCache +import abc +import logging + +from canonicaljson import json from twisted.internet import defer -import logging -import ujson as json +from synapse.util.caches.descriptors import cached, cachedInlineCallbacks, cachedList +from synapse.util.caches.stream_change_cache import StreamChangeCache +from ._base import SQLBaseStore +from .util.id_generators import StreamIdGenerator logger = logging.getLogger(__name__) -class ReceiptsStore(SQLBaseStore): - def __init__(self, hs): - super(ReceiptsStore, self).__init__(hs) +class ReceiptsWorkerStore(SQLBaseStore): + """This is an abstract base class where subclasses must implement + `get_max_receipt_stream_id` which can be called in the initializer. + """ + + # This ABCMeta metaclass ensures that we cannot be instantiated without + # the abstract methods being implemented. + __metaclass__ = abc.ABCMeta + + def __init__(self, db_conn, hs): + super(ReceiptsWorkerStore, self).__init__(db_conn, hs) self._receipts_stream_cache = StreamChangeCache( - "ReceiptsRoomChangeCache", self._receipts_id_gen.get_current_token() + "ReceiptsRoomChangeCache", self.get_max_receipt_stream_id() ) + @abc.abstractmethod + def get_max_receipt_stream_id(self): + """Get the current max stream ID for receipts stream + + Returns: + int + """ + raise NotImplementedError() + @cachedInlineCallbacks() def get_users_with_read_receipts_in_room(self, room_id): receipts = yield self.get_receipts_for_room(room_id, "m.read") defer.returnValue(set(r['user_id'] for r in receipts)) - def _invalidate_get_users_with_receipts_in_room(self, room_id, receipt_type, - user_id): - if receipt_type != "m.read": - return - - # Returns an ObservableDeferred - res = self.get_users_with_read_receipts_in_room.cache.get( - room_id, None, update_metrics=False, - ) - - if res: - if isinstance(res, defer.Deferred) and res.called: - res = res.result - if user_id in res: - # We'd only be adding to the set, so no point invalidating if the - # user is already there - return - - self.get_users_with_read_receipts_in_room.invalidate((room_id,)) - @cached(num_args=2) def get_receipts_for_room(self, room_id, receipt_type): return self._simple_select_list( @@ -139,7 +140,9 @@ class ReceiptsStore(SQLBaseStore): """ room_ids = set(room_ids) - if from_key: + if from_key is not None: + # Only ask the database about rooms where there have been new + # receipts added since `from_key` room_ids = yield self._receipts_stream_cache.get_entities_changed( room_ids, from_key ) @@ -150,7 +153,6 @@ class ReceiptsStore(SQLBaseStore): defer.returnValue([ev for res in results.values() for ev in res]) - @cachedInlineCallbacks(num_args=3, tree=True) def get_linearized_receipts_for_room(self, room_id, to_key, from_key=None): """Get receipts for a single room for sending to clients. @@ -161,7 +163,19 @@ class ReceiptsStore(SQLBaseStore): from the start. Returns: - list: A list of receipts. + Deferred[list]: A list of receipts. + """ + if from_key is not None: + # Check the cache first to see if any new receipts have been added + # since`from_key`. If not we can no-op. + if not self._receipts_stream_cache.has_entity_changed(room_id, from_key): + defer.succeed([]) + + return self._get_linearized_receipts_for_room(room_id, to_key, from_key) + + @cachedInlineCallbacks(num_args=3, tree=True) + def _get_linearized_receipts_for_room(self, room_id, to_key, from_key=None): + """See get_linearized_receipts_for_room """ def f(txn): if from_key: @@ -210,7 +224,7 @@ class ReceiptsStore(SQLBaseStore): "content": content, }]) - @cachedList(cached_method_name="get_linearized_receipts_for_room", + @cachedList(cached_method_name="_get_linearized_receipts_for_room", list_name="room_ids", num_args=3, inlineCallbacks=True) def _get_linearized_receipts_for_rooms(self, room_ids, to_key, from_key=None): if not room_ids: @@ -270,11 +284,97 @@ class ReceiptsStore(SQLBaseStore): } defer.returnValue(results) + def get_all_updated_receipts(self, last_id, current_id, limit=None): + if last_id == current_id: + return defer.succeed([]) + + def get_all_updated_receipts_txn(txn): + sql = ( + "SELECT stream_id, room_id, receipt_type, user_id, event_id, data" + " FROM receipts_linearized" + " WHERE ? < stream_id AND stream_id <= ?" + " ORDER BY stream_id ASC" + ) + args = [last_id, current_id] + if limit is not None: + sql += " LIMIT ?" + args.append(limit) + txn.execute(sql, args) + + return txn.fetchall() + return self.runInteraction( + "get_all_updated_receipts", get_all_updated_receipts_txn + ) + + def _invalidate_get_users_with_receipts_in_room(self, room_id, receipt_type, + user_id): + if receipt_type != "m.read": + return + + # Returns either an ObservableDeferred or the raw result + res = self.get_users_with_read_receipts_in_room.cache.get( + room_id, None, update_metrics=False, + ) + + # first handle the Deferred case + if isinstance(res, defer.Deferred): + if res.called: + res = res.result + else: + res = None + + if res and user_id in res: + # We'd only be adding to the set, so no point invalidating if the + # user is already there + return + + self.get_users_with_read_receipts_in_room.invalidate((room_id,)) + + +class ReceiptsStore(ReceiptsWorkerStore): + def __init__(self, db_conn, hs): + # We instantiate this first as the ReceiptsWorkerStore constructor + # needs to be able to call get_max_receipt_stream_id + self._receipts_id_gen = StreamIdGenerator( + db_conn, "receipts_linearized", "stream_id" + ) + + super(ReceiptsStore, self).__init__(db_conn, hs) + def get_max_receipt_stream_id(self): return self._receipts_id_gen.get_current_token() def insert_linearized_receipt_txn(self, txn, room_id, receipt_type, user_id, event_id, data, stream_id): + res = self._simple_select_one_txn( + txn, + table="events", + retcols=["topological_ordering", "stream_ordering"], + keyvalues={"event_id": event_id}, + allow_none=True + ) + + stream_ordering = int(res["stream_ordering"]) if res else None + + # We don't want to clobber receipts for more recent events, so we + # have to compare orderings of existing receipts + if stream_ordering is not None: + sql = ( + "SELECT stream_ordering, event_id FROM events" + " INNER JOIN receipts_linearized as r USING (event_id, room_id)" + " WHERE r.room_id = ? AND r.receipt_type = ? AND r.user_id = ?" + ) + txn.execute(sql, (room_id, receipt_type, user_id)) + + for so, eid in txn: + if int(so) >= stream_ordering: + logger.debug( + "Ignoring new receipt for %s in favour of existing " + "one for later event %s", + event_id, eid, + ) + return False + txn.call_after( self.get_receipts_for_room.invalidate, (room_id, receipt_type) ) @@ -286,7 +386,7 @@ class ReceiptsStore(SQLBaseStore): self.get_receipts_for_user.invalidate, (user_id, receipt_type) ) # FIXME: This shouldn't invalidate the whole cache - txn.call_after(self.get_linearized_receipts_for_room.invalidate_many, (room_id,)) + txn.call_after(self._get_linearized_receipts_for_room.invalidate_many, (room_id,)) txn.call_after( self._receipts_stream_cache.entity_has_changed, @@ -298,34 +398,6 @@ class ReceiptsStore(SQLBaseStore): (user_id, room_id, receipt_type) ) - res = self._simple_select_one_txn( - txn, - table="events", - retcols=["topological_ordering", "stream_ordering"], - keyvalues={"event_id": event_id}, - allow_none=True - ) - - topological_ordering = int(res["topological_ordering"]) if res else None - stream_ordering = int(res["stream_ordering"]) if res else None - - # We don't want to clobber receipts for more recent events, so we - # have to compare orderings of existing receipts - sql = ( - "SELECT topological_ordering, stream_ordering, event_id FROM events" - " INNER JOIN receipts_linearized as r USING (event_id, room_id)" - " WHERE r.room_id = ? AND r.receipt_type = ? AND r.user_id = ?" - ) - - txn.execute(sql, (room_id, receipt_type, user_id)) - - if topological_ordering: - for to, so, _ in txn: - if int(to) > topological_ordering: - return False - elif int(to) == topological_ordering and int(so) >= stream_ordering: - return False - self._simple_delete_txn( txn, table="receipts_linearized", @@ -349,12 +421,11 @@ class ReceiptsStore(SQLBaseStore): } ) - if receipt_type == "m.read" and topological_ordering: + if receipt_type == "m.read" and stream_ordering is not None: self._remove_old_push_actions_before_txn( txn, room_id=room_id, user_id=user_id, - topological_ordering=topological_ordering, stream_ordering=stream_ordering, ) @@ -435,7 +506,7 @@ class ReceiptsStore(SQLBaseStore): self.get_receipts_for_user.invalidate, (user_id, receipt_type) ) # FIXME: This shouldn't invalidate the whole cache - txn.call_after(self.get_linearized_receipts_for_room.invalidate_many, (room_id,)) + txn.call_after(self._get_linearized_receipts_for_room.invalidate_many, (room_id,)) self._simple_delete_txn( txn, @@ -457,25 +528,3 @@ class ReceiptsStore(SQLBaseStore): "data": json.dumps(data), } ) - - def get_all_updated_receipts(self, last_id, current_id, limit=None): - if last_id == current_id: - return defer.succeed([]) - - def get_all_updated_receipts_txn(txn): - sql = ( - "SELECT stream_id, room_id, receipt_type, user_id, event_id, data" - " FROM receipts_linearized" - " WHERE ? < stream_id AND stream_id <= ?" - " ORDER BY stream_id ASC" - ) - args = [last_id, current_id] - if limit is not None: - sql += " LIMIT ?" - args.append(limit) - txn.execute(sql, args) - - return txn.fetchall() - return self.runInteraction( - "get_all_updated_receipts", get_all_updated_receipts_txn - ) diff --git a/synapse/storage/registration.py b/synapse/storage/registration.py index 20acd58fc..07333f777 100644 --- a/synapse/storage/registration.py +++ b/synapse/storage/registration.py @@ -15,17 +15,83 @@ import re +from six.moves import range + from twisted.internet import defer -from synapse.api.errors import StoreError, Codes +from synapse.api.errors import Codes, StoreError from synapse.storage import background_updates +from synapse.storage._base import SQLBaseStore from synapse.util.caches.descriptors import cached, cachedInlineCallbacks -class RegistrationStore(background_updates.BackgroundUpdateStore): +class RegistrationWorkerStore(SQLBaseStore): + @cached() + def get_user_by_id(self, user_id): + return self._simple_select_one( + table="users", + keyvalues={ + "name": user_id, + }, + retcols=[ + "name", "password_hash", "is_guest", + "consent_version", "consent_server_notice_sent", + "appservice_id", + ], + allow_none=True, + desc="get_user_by_id", + ) - def __init__(self, hs): - super(RegistrationStore, self).__init__(hs) + @cached() + def get_user_by_access_token(self, token): + """Get a user from the given access token. + + Args: + token (str): The access token of a user. + Returns: + defer.Deferred: None, if the token did not match, otherwise dict + including the keys `name`, `is_guest`, `device_id`, `token_id`. + """ + return self.runInteraction( + "get_user_by_access_token", + self._query_for_auth, + token + ) + + @defer.inlineCallbacks + def is_server_admin(self, user): + res = yield self._simple_select_one_onecol( + table="users", + keyvalues={"name": user.to_string()}, + retcol="admin", + allow_none=True, + desc="is_server_admin", + ) + + defer.returnValue(res if res else False) + + def _query_for_auth(self, txn, token): + sql = ( + "SELECT users.name, users.is_guest, access_tokens.id as token_id," + " access_tokens.device_id" + " FROM users" + " INNER JOIN access_tokens on users.name = access_tokens.user_id" + " WHERE token = ?" + ) + + txn.execute(sql, (token,)) + rows = self.cursor_to_dict(txn) + if rows: + return rows[0] + + return None + + +class RegistrationStore(RegistrationWorkerStore, + background_updates.BackgroundUpdateStore): + + def __init__(self, db_conn, hs): + super(RegistrationStore, self).__init__(db_conn, hs) self.clock = hs.get_clock() @@ -37,12 +103,17 @@ class RegistrationStore(background_updates.BackgroundUpdateStore): ) self.register_background_index_update( - "refresh_tokens_device_index", - index_name="refresh_tokens_device_id", - table="refresh_tokens", - columns=["user_id", "device_id"], + "users_creation_ts", + index_name="users_creation_ts", + table="users", + columns=["creation_ts"], ) + # we no longer use refresh tokens, but it's possible that some people + # might have a background update queued to build this index. Just + # clear the background update. + self.register_noop_background_update("refresh_tokens_device_index") + @defer.inlineCallbacks def add_access_token_to_user(self, user_id, token, device_id=None): """Adds an access token for the given user. @@ -177,9 +248,11 @@ class RegistrationStore(background_updates.BackgroundUpdateStore): ) if create_profile_with_localpart: + # set a default displayname serverside to avoid ugly race + # between auto-joins and clients trying to set displaynames txn.execute( - "INSERT INTO profiles(user_id) VALUES (?)", - (create_profile_with_localpart,) + "INSERT INTO profiles(user_id, displayname) VALUES (?,?)", + (create_profile_with_localpart, create_profile_with_localpart) ) self._invalidate_cache_and_stream( @@ -187,18 +260,6 @@ class RegistrationStore(background_updates.BackgroundUpdateStore): ) txn.call_after(self.is_guest.invalidate, (user_id,)) - @cached() - def get_user_by_id(self, user_id): - return self._simple_select_one( - table="users", - keyvalues={ - "name": user_id, - }, - retcols=["name", "password_hash", "is_guest"], - allow_none=True, - desc="get_user_by_id", - ) - def get_users_by_id_case_insensitive(self, user_id): """Gets users that match user_id case insensitively. Returns a mapping of user_id -> password_hash. @@ -236,12 +297,57 @@ class RegistrationStore(background_updates.BackgroundUpdateStore): "user_set_password_hash", user_set_password_hash_txn ) - @defer.inlineCallbacks - def user_delete_access_tokens(self, user_id, except_token_id=None, - device_id=None, - delete_refresh_tokens=False): + def user_set_consent_version(self, user_id, consent_version): + """Updates the user table to record privacy policy consent + + Args: + user_id (str): full mxid of the user to update + consent_version (str): version of the policy the user has consented + to + + Raises: + StoreError(404) if user not found """ - Invalidate access/refresh tokens belonging to a user + def f(txn): + self._simple_update_one_txn( + txn, + table='users', + keyvalues={'name': user_id, }, + updatevalues={'consent_version': consent_version, }, + ) + self._invalidate_cache_and_stream( + txn, self.get_user_by_id, (user_id,) + ) + return self.runInteraction("user_set_consent_version", f) + + def user_set_consent_server_notice_sent(self, user_id, consent_version): + """Updates the user table to record that we have sent the user a server + notice about privacy policy consent + + Args: + user_id (str): full mxid of the user to update + consent_version (str): version of the policy we have notified the + user about + + Raises: + StoreError(404) if user not found + """ + def f(txn): + self._simple_update_one_txn( + txn, + table='users', + keyvalues={'name': user_id, }, + updatevalues={'consent_server_notice_sent': consent_version, }, + ) + self._invalidate_cache_and_stream( + txn, self.get_user_by_id, (user_id,) + ) + return self.runInteraction("user_set_consent_server_notice_sent", f) + + def user_delete_access_tokens(self, user_id, except_token_id=None, + device_id=None): + """ + Invalidate access tokens belonging to a user Args: user_id (str): ID of user the tokens belong to @@ -250,10 +356,9 @@ class RegistrationStore(background_updates.BackgroundUpdateStore): device_id (str|None): ID of device the tokens are associated with. If None, tokens associated with any device (or no device) will be deleted - delete_refresh_tokens (bool): True to delete refresh tokens as - well as access tokens. Returns: - defer.Deferred: + defer.Deferred[list[str, int, str|None, int]]: a list of + (token, token id, device id) for each of the deleted tokens """ def f(txn): keyvalues = { @@ -262,13 +367,6 @@ class RegistrationStore(background_updates.BackgroundUpdateStore): if device_id is not None: keyvalues["device_id"] = device_id - if delete_refresh_tokens: - self._simple_delete_txn( - txn, - table="refresh_tokens", - keyvalues=keyvalues, - ) - items = keyvalues.items() where_clause = " AND ".join(k + " = ?" for k, _ in items) values = [v for _, v in items] @@ -277,14 +375,14 @@ class RegistrationStore(background_updates.BackgroundUpdateStore): values.append(except_token_id) txn.execute( - "SELECT token FROM access_tokens WHERE %s" % where_clause, + "SELECT token, id, device_id FROM access_tokens WHERE %s" % where_clause, values ) - rows = self.cursor_to_dict(txn) + tokens_and_devices = [(r[0], r[1], r[2]) for r in txn] - for row in rows: + for token, _, _ in tokens_and_devices: self._invalidate_cache_and_stream( - txn, self.get_user_by_access_token, (row["token"],) + txn, self.get_user_by_access_token, (token,) ) txn.execute( @@ -292,7 +390,9 @@ class RegistrationStore(background_updates.BackgroundUpdateStore): values ) - yield self.runInteraction( + return tokens_and_devices + + return self.runInteraction( "user_delete_access_tokens", f, ) @@ -312,34 +412,6 @@ class RegistrationStore(background_updates.BackgroundUpdateStore): return self.runInteraction("delete_access_token", f) - @cached() - def get_user_by_access_token(self, token): - """Get a user from the given access token. - - Args: - token (str): The access token of a user. - Returns: - defer.Deferred: None, if the token did not match, otherwise dict - including the keys `name`, `is_guest`, `device_id`, `token_id`. - """ - return self.runInteraction( - "get_user_by_access_token", - self._query_for_auth, - token - ) - - @defer.inlineCallbacks - def is_server_admin(self, user): - res = yield self._simple_select_one_onecol( - table="users", - keyvalues={"name": user.to_string()}, - retcol="admin", - allow_none=True, - desc="is_server_admin", - ) - - defer.returnValue(res if res else False) - @cachedInlineCallbacks() def is_guest(self, user_id): res = yield self._simple_select_one_onecol( @@ -352,22 +424,6 @@ class RegistrationStore(background_updates.BackgroundUpdateStore): defer.returnValue(res if res else False) - def _query_for_auth(self, txn, token): - sql = ( - "SELECT users.name, users.is_guest, access_tokens.id as token_id," - " access_tokens.device_id" - " FROM users" - " INNER JOIN access_tokens on users.name = access_tokens.user_id" - " WHERE token = ?" - ) - - txn.execute(sql, (token,)) - rows = self.cursor_to_dict(txn) - if rows: - return rows[0] - - return None - @defer.inlineCallbacks def user_add_threepid(self, user_id, medium, address, validated_at, added_at): yield self._simple_upsert("user_threepids", { @@ -404,15 +460,6 @@ class RegistrationStore(background_updates.BackgroundUpdateStore): defer.returnValue(ret['user_id']) defer.returnValue(None) - def user_delete_threepids(self, user_id): - return self._simple_delete( - "user_threepids", - keyvalues={ - "user_id": user_id, - }, - desc="user_delete_threepids", - ) - def user_delete_threepid(self, user_id, medium, address): return self._simple_delete( "user_threepids", @@ -437,6 +484,35 @@ class RegistrationStore(background_updates.BackgroundUpdateStore): ret = yield self.runInteraction("count_users", _count_users) defer.returnValue(ret) + def count_daily_user_type(self): + """ + Counts 1) native non guest users + 2) native guests users + 3) bridged users + who registered on the homeserver in the past 24 hours + """ + def _count_daily_user_type(txn): + yesterday = int(self._clock.time()) - (60 * 60 * 24) + + sql = """ + SELECT user_type, COALESCE(count(*), 0) AS count FROM ( + SELECT + CASE + WHEN is_guest=0 AND appservice_id IS NULL THEN 'native' + WHEN is_guest=1 AND appservice_id IS NULL THEN 'guest' + WHEN is_guest=0 AND appservice_id IS NOT NULL THEN 'bridged' + END AS user_type + FROM users + WHERE creation_ts > ? + ) AS t GROUP BY user_type + """ + results = {'native': 0, 'guest': 0, 'bridged': 0} + txn.execute(sql, (yesterday,)) + for row in txn: + results[row[0]] = row[1] + return results + return self.runInteraction("count_daily_user_type", _count_daily_user_type) + @defer.inlineCallbacks def count_nonbridged_users(self): def _count_users(txn): @@ -464,18 +540,16 @@ class RegistrationStore(background_updates.BackgroundUpdateStore): """ def _find_next_generated_user_id(txn): txn.execute("SELECT name FROM users") - rows = self.cursor_to_dict(txn) regex = re.compile("^@(\d+):") found = set() - for r in rows: - user_id = r["name"] + for user_id, in txn: match = regex.search(user_id) if match: found.add(int(match.group(1))) - for i in xrange(len(found) + 1): + for i in range(len(found) + 1): if i not in found: return i @@ -530,3 +604,44 @@ class RegistrationStore(background_updates.BackgroundUpdateStore): except self.database_engine.module.IntegrityError: ret = yield self.get_3pid_guest_access_token(medium, address) defer.returnValue(ret) + + def add_user_pending_deactivation(self, user_id): + """ + Adds a user to the table of users who need to be parted from all the rooms they're + in + """ + return self._simple_insert( + "users_pending_deactivation", + values={ + "user_id": user_id, + }, + desc="add_user_pending_deactivation", + ) + + def del_user_pending_deactivation(self, user_id): + """ + Removes the given user to the table of users who need to be parted from all the + rooms they're in, effectively marking that user as fully deactivated. + """ + # XXX: This should be simple_delete_one but we failed to put a unique index on + # the table, so somehow duplicate entries have ended up in it. + return self._simple_delete( + "users_pending_deactivation", + keyvalues={ + "user_id": user_id, + }, + desc="del_user_pending_deactivation", + ) + + def get_user_pending_deactivation(self): + """ + Gets one user from the table of users waiting to be parted from all the rooms + they're in. + """ + return self._simple_select_one_onecol( + "users_pending_deactivation", + keyvalues={}, + retcol="user_id", + allow_none=True, + desc="get_users_pending_deactivation", + ) diff --git a/synapse/storage/rejections.py b/synapse/storage/rejections.py index 40acb5c4e..880f047ad 100644 --- a/synapse/storage/rejections.py +++ b/synapse/storage/rejections.py @@ -13,10 +13,10 @@ # See the License for the specific language governing permissions and # limitations under the License. -from ._base import SQLBaseStore - import logging +from ._base import SQLBaseStore + logger = logging.getLogger(__name__) diff --git a/synapse/storage/room.py b/synapse/storage/room.py index 23688430b..3147fb682 100644 --- a/synapse/storage/room.py +++ b/synapse/storage/room.py @@ -13,19 +13,19 @@ # See the License for the specific language governing permissions and # limitations under the License. +import collections +import logging +import re + +from canonicaljson import json + from twisted.internet import defer from synapse.api.errors import StoreError +from synapse.storage._base import SQLBaseStore +from synapse.storage.search import SearchStore from synapse.util.caches.descriptors import cached, cachedInlineCallbacks -from ._base import SQLBaseStore -from .engines import PostgresEngine, Sqlite3Engine - -import collections -import logging -import ujson as json -import re - logger = logging.getLogger(__name__) @@ -40,7 +40,138 @@ RatelimitOverride = collections.namedtuple( ) -class RoomStore(SQLBaseStore): +class RoomWorkerStore(SQLBaseStore): + def get_public_room_ids(self): + return self._simple_select_onecol( + table="rooms", + keyvalues={ + "is_public": True, + }, + retcol="room_id", + desc="get_public_room_ids", + ) + + @cached(num_args=2, max_entries=100) + def get_public_room_ids_at_stream_id(self, stream_id, network_tuple): + """Get pulbic rooms for a particular list, or across all lists. + + Args: + stream_id (int) + network_tuple (ThirdPartyInstanceID): The list to use (None, None) + means the main list, None means all lsits. + """ + return self.runInteraction( + "get_public_room_ids_at_stream_id", + self.get_public_room_ids_at_stream_id_txn, + stream_id, network_tuple=network_tuple + ) + + def get_public_room_ids_at_stream_id_txn(self, txn, stream_id, + network_tuple): + return { + rm + for rm, vis in self.get_published_at_stream_id_txn( + txn, stream_id, network_tuple=network_tuple + ).items() + if vis + } + + def get_published_at_stream_id_txn(self, txn, stream_id, network_tuple): + if network_tuple: + # We want to get from a particular list. No aggregation required. + + sql = (""" + SELECT room_id, visibility FROM public_room_list_stream + INNER JOIN ( + SELECT room_id, max(stream_id) AS stream_id + FROM public_room_list_stream + WHERE stream_id <= ? %s + GROUP BY room_id + ) grouped USING (room_id, stream_id) + """) + + if network_tuple.appservice_id is not None: + txn.execute( + sql % ("AND appservice_id = ? AND network_id = ?",), + (stream_id, network_tuple.appservice_id, network_tuple.network_id,) + ) + else: + txn.execute( + sql % ("AND appservice_id IS NULL",), + (stream_id,) + ) + return dict(txn) + else: + # We want to get from all lists, so we need to aggregate the results + + logger.info("Executing full list") + + sql = (""" + SELECT room_id, visibility + FROM public_room_list_stream + INNER JOIN ( + SELECT + room_id, max(stream_id) AS stream_id, appservice_id, + network_id + FROM public_room_list_stream + WHERE stream_id <= ? + GROUP BY room_id, appservice_id, network_id + ) grouped USING (room_id, stream_id) + """) + + txn.execute( + sql, + (stream_id,) + ) + + results = {} + # A room is visible if its visible on any list. + for room_id, visibility in txn: + results[room_id] = bool(visibility) or results.get(room_id, False) + + return results + + def get_public_room_changes(self, prev_stream_id, new_stream_id, + network_tuple): + def get_public_room_changes_txn(txn): + then_rooms = self.get_public_room_ids_at_stream_id_txn( + txn, prev_stream_id, network_tuple + ) + + now_rooms_dict = self.get_published_at_stream_id_txn( + txn, new_stream_id, network_tuple + ) + + now_rooms_visible = set( + rm for rm, vis in now_rooms_dict.items() if vis + ) + now_rooms_not_visible = set( + rm for rm, vis in now_rooms_dict.items() if not vis + ) + + newly_visible = now_rooms_visible - then_rooms + newly_unpublished = now_rooms_not_visible & then_rooms + + return newly_visible, newly_unpublished + + return self.runInteraction( + "get_public_room_changes", get_public_room_changes_txn + ) + + @cached(max_entries=10000) + def is_room_blocked(self, room_id): + return self._simple_select_one_onecol( + table="blocked_rooms", + keyvalues={ + "room_id": room_id, + }, + retcol="1", + allow_none=True, + desc="is_room_blocked", + ) + + +class RoomStore(RoomWorkerStore, SearchStore): @defer.inlineCallbacks def store_room(self, room_id, room_creator_user_id, is_public): @@ -227,16 +358,6 @@ class RoomStore(SQLBaseStore): ) self.hs.get_notifier().on_new_replication_data() - def get_public_room_ids(self): - return self._simple_select_onecol( - table="rooms", - keyvalues={ - "is_public": True, - }, - retcol="room_id", - desc="get_public_room_ids", - ) - def get_room_count(self): """Retrieve a list of all rooms """ @@ -263,8 +384,8 @@ class RoomStore(SQLBaseStore): }, ) - self._store_event_search_txn( - txn, event, "content.topic", event.content["topic"] + self.store_event_search_txn( + txn, event, "content.topic", event.content["topic"], ) def _store_room_name_txn(self, txn, event): @@ -279,14 +400,14 @@ class RoomStore(SQLBaseStore): } ) - self._store_event_search_txn( - txn, event, "content.name", event.content["name"] + self.store_event_search_txn( + txn, event, "content.name", event.content["name"], ) def _store_room_message_txn(self, txn, event): if hasattr(event, "content") and "body" in event.content: - self._store_event_search_txn( - txn, event, "content.body", event.content["body"] + self.store_event_search_txn( + txn, event, "content.body", event.content["body"], ) def _store_history_visibility_txn(self, txn, event): @@ -308,31 +429,6 @@ class RoomStore(SQLBaseStore): event.content[key] )) - def _store_event_search_txn(self, txn, event, key, value): - if isinstance(self.database_engine, PostgresEngine): - sql = ( - "INSERT INTO event_search" - " (event_id, room_id, key, vector, stream_ordering, origin_server_ts)" - " VALUES (?,?,?,to_tsvector('english', ?),?,?)" - ) - txn.execute( - sql, - ( - event.event_id, event.room_id, key, value, - event.internal_metadata.stream_ordering, - event.origin_server_ts, - ) - ) - elif isinstance(self.database_engine, Sqlite3Engine): - sql = ( - "INSERT INTO event_search (event_id, room_id, key, value)" - " VALUES (?,?,?,?)" - ) - txn.execute(sql, (event.event_id, event.room_id, key, value,)) - else: - # This should be unreachable. - raise Exception("Unrecognized database engine") - def add_event_report(self, room_id, event_id, user_id, reason, content, received_ts): next_id = self._event_reports_id_gen.get_next() @@ -353,113 +449,6 @@ class RoomStore(SQLBaseStore): def get_current_public_room_stream_id(self): return self._public_room_id_gen.get_current_token() - @cached(num_args=2, max_entries=100) - def get_public_room_ids_at_stream_id(self, stream_id, network_tuple): - """Get pulbic rooms for a particular list, or across all lists. - - Args: - stream_id (int) - network_tuple (ThirdPartyInstanceID): The list to use (None, None) - means the main list, None means all lsits. - """ - return self.runInteraction( - "get_public_room_ids_at_stream_id", - self.get_public_room_ids_at_stream_id_txn, - stream_id, network_tuple=network_tuple - ) - - def get_public_room_ids_at_stream_id_txn(self, txn, stream_id, - network_tuple): - return { - rm - for rm, vis in self.get_published_at_stream_id_txn( - txn, stream_id, network_tuple=network_tuple - ).items() - if vis - } - - def get_published_at_stream_id_txn(self, txn, stream_id, network_tuple): - if network_tuple: - # We want to get from a particular list. No aggregation required. - - sql = (""" - SELECT room_id, visibility FROM public_room_list_stream - INNER JOIN ( - SELECT room_id, max(stream_id) AS stream_id - FROM public_room_list_stream - WHERE stream_id <= ? %s - GROUP BY room_id - ) grouped USING (room_id, stream_id) - """) - - if network_tuple.appservice_id is not None: - txn.execute( - sql % ("AND appservice_id = ? AND network_id = ?",), - (stream_id, network_tuple.appservice_id, network_tuple.network_id,) - ) - else: - txn.execute( - sql % ("AND appservice_id IS NULL",), - (stream_id,) - ) - return dict(txn) - else: - # We want to get from all lists, so we need to aggregate the results - - logger.info("Executing full list") - - sql = (""" - SELECT room_id, visibility - FROM public_room_list_stream - INNER JOIN ( - SELECT - room_id, max(stream_id) AS stream_id, appservice_id, - network_id - FROM public_room_list_stream - WHERE stream_id <= ? - GROUP BY room_id, appservice_id, network_id - ) grouped USING (room_id, stream_id) - """) - - txn.execute( - sql, - (stream_id,) - ) - - results = {} - # A room is visible if its visible on any list. - for room_id, visibility in txn: - results[room_id] = bool(visibility) or results.get(room_id, False) - - return results - - def get_public_room_changes(self, prev_stream_id, new_stream_id, - network_tuple): - def get_public_room_changes_txn(txn): - then_rooms = self.get_public_room_ids_at_stream_id_txn( - txn, prev_stream_id, network_tuple - ) - - now_rooms_dict = self.get_published_at_stream_id_txn( - txn, new_stream_id, network_tuple - ) - - now_rooms_visible = set( - rm for rm, vis in now_rooms_dict.items() if vis - ) - now_rooms_not_visible = set( - rm for rm, vis in now_rooms_dict.items() if not vis - ) - - newly_visible = now_rooms_visible - then_rooms - newly_unpublished = now_rooms_not_visible & then_rooms - - return newly_visible, newly_unpublished - - return self.runInteraction( - "get_public_room_changes", get_public_room_changes_txn - ) - def get_all_new_public_rooms(self, prev_id, current_id, limit): def get_all_new_public_rooms(txn): sql = (""" @@ -509,18 +498,6 @@ class RoomStore(SQLBaseStore): else: defer.returnValue(None) - @cached(max_entries=10000) - def is_room_blocked(self, room_id): - return self._simple_select_one_onecol( - table="blocked_rooms", - keyvalues={ - "room_id": room_id, - }, - retcol="1", - allow_none=True, - desc="is_room_blocked", - ) - @defer.inlineCallbacks def block_room(self, room_id, user_id): yield self._simple_insert( @@ -531,75 +508,121 @@ class RoomStore(SQLBaseStore): }, desc="block_room", ) - self.is_room_blocked.invalidate((room_id,)) + yield self.runInteraction( + "block_room_invalidation", + self._invalidate_cache_and_stream, + self.is_room_blocked, (room_id,), + ) + + def get_media_mxcs_in_room(self, room_id): + """Retrieves all the local and remote media MXC URIs in a given room + + Args: + room_id (str) + + Returns: + The local and remote media as a lists of tuples where the key is + the hostname and the value is the media ID. + """ + def _get_media_mxcs_in_room_txn(txn): + local_mxcs, remote_mxcs = self._get_media_mxcs_in_room_txn(txn, room_id) + local_media_mxcs = [] + remote_media_mxcs = [] + + # Convert the IDs to MXC URIs + for media_id in local_mxcs: + local_media_mxcs.append("mxc://%s/%s" % (self.hs.hostname, media_id)) + for hostname, media_id in remote_mxcs: + remote_media_mxcs.append("mxc://%s/%s" % (hostname, media_id)) + + return local_media_mxcs, remote_media_mxcs + return self.runInteraction("get_media_ids_in_room", _get_media_mxcs_in_room_txn) def quarantine_media_ids_in_room(self, room_id, quarantined_by): """For a room loops through all events with media and quarantines the associated media """ - def _get_media_ids_in_room(txn): - mxc_re = re.compile("^mxc://([^/]+)/([^/#?]+)") - - next_token = self.get_current_events_token() + 1 - + def _quarantine_media_in_room_txn(txn): + local_mxcs, remote_mxcs = self._get_media_mxcs_in_room_txn(txn, room_id) total_media_quarantined = 0 - while next_token: - sql = """ - SELECT stream_ordering, content FROM events - WHERE room_id = ? - AND stream_ordering < ? - AND contains_url = ? AND outlier = ? - ORDER BY stream_ordering DESC - LIMIT ? + # Now update all the tables to set the quarantined_by flag + + txn.executemany(""" + UPDATE local_media_repository + SET quarantined_by = ? + WHERE media_id = ? + """, ((quarantined_by, media_id) for media_id in local_mxcs)) + + txn.executemany( """ - txn.execute(sql, (room_id, next_token, True, False, 100)) - - next_token = None - local_media_mxcs = [] - remote_media_mxcs = [] - for stream_ordering, content_json in txn: - next_token = stream_ordering - content = json.loads(content_json) - - content_url = content.get("url") - thumbnail_url = content.get("info", {}).get("thumbnail_url") - - for url in (content_url, thumbnail_url): - if not url: - continue - matches = mxc_re.match(url) - if matches: - hostname = matches.group(1) - media_id = matches.group(2) - if hostname == self.hostname: - local_media_mxcs.append(media_id) - else: - remote_media_mxcs.append((hostname, media_id)) - - # Now update all the tables to set the quarantined_by flag - - txn.executemany(""" - UPDATE local_media_repository + UPDATE remote_media_cache SET quarantined_by = ? - WHERE media_id = ? - """, ((quarantined_by, media_id) for media_id in local_media_mxcs)) - - txn.executemany( - """ - UPDATE remote_media_cache - SET quarantined_by = ? - WHERE media_origin AND media_id = ? - """, - ( - (quarantined_by, origin, media_id) - for origin, media_id in remote_media_mxcs - ) + WHERE media_origin = ? AND media_id = ? + """, + ( + (quarantined_by, origin, media_id) + for origin, media_id in remote_mxcs ) + ) - total_media_quarantined += len(local_media_mxcs) - total_media_quarantined += len(remote_media_mxcs) + total_media_quarantined += len(local_mxcs) + total_media_quarantined += len(remote_mxcs) return total_media_quarantined - return self.runInteraction("get_media_ids_in_room", _get_media_ids_in_room) + return self.runInteraction( + "quarantine_media_in_room", + _quarantine_media_in_room_txn, + ) + + def _get_media_mxcs_in_room_txn(self, txn, room_id): + """Retrieves all the local and remote media MXC URIs in a given room + + Args: + txn (cursor) + room_id (str) + + Returns: + The local and remote media as a lists of tuples where the key is + the hostname and the value is the media ID. + """ + mxc_re = re.compile("^mxc://([^/]+)/([^/#?]+)") + + next_token = self.get_current_events_token() + 1 + local_media_mxcs = [] + remote_media_mxcs = [] + + while next_token: + sql = """ + SELECT stream_ordering, json FROM events + JOIN event_json USING (room_id, event_id) + WHERE room_id = ? + AND stream_ordering < ? + AND contains_url = ? AND outlier = ? + ORDER BY stream_ordering DESC + LIMIT ? + """ + txn.execute(sql, (room_id, next_token, True, False, 100)) + + next_token = None + for stream_ordering, content_json in txn: + next_token = stream_ordering + event_json = json.loads(content_json) + content = event_json["content"] + content_url = content.get("url") + thumbnail_url = content.get("info", {}).get("thumbnail_url") + + for url in (content_url, thumbnail_url): + if not url: + continue + matches = mxc_re.match(url) + if matches: + hostname = matches.group(1) + media_id = matches.group(2) + if hostname == self.hs.hostname: + local_media_mxcs.append(media_id) + else: + remote_media_mxcs.append((hostname, media_id)) + + return local_media_mxcs, remote_media_mxcs diff --git a/synapse/storage/roommember.py b/synapse/storage/roommember.py index 457ca288d..01697ab2c 100644 --- a/synapse/storage/roommember.py +++ b/synapse/storage/roommember.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2014-2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -13,22 +14,23 @@ # See the License for the specific language governing permissions and # limitations under the License. -from twisted.internet import defer - +import logging from collections import namedtuple -from ._base import SQLBaseStore +from six import iteritems, itervalues + +from canonicaljson import json + +from twisted.internet import defer + +from synapse.api.constants import EventTypes, Membership +from synapse.storage.events import EventsWorkerStore +from synapse.types import get_domain_from_id from synapse.util.async import Linearizer from synapse.util.caches import intern_string from synapse.util.caches.descriptors import cached, cachedInlineCallbacks from synapse.util.stringutils import to_ascii -from synapse.api.constants import Membership, EventTypes -from synapse.types import get_domain_from_id - -import logging -import ujson as json - logger = logging.getLogger(__name__) @@ -37,6 +39,11 @@ RoomsForUser = namedtuple( ("room_id", "sender", "membership", "event_id", "stream_ordering") ) +GetRoomsForUserWithStreamOrdering = namedtuple( + "_GetRoomsForUserWithStreamOrdering", + ("room_id", "stream_ordering",) +) + # We store this using a namedtuple so that we save about 3x space over using a # dict. @@ -48,97 +55,7 @@ ProfileInfo = namedtuple( _MEMBERSHIP_PROFILE_UPDATE_NAME = "room_membership_profile_update" -class RoomMemberStore(SQLBaseStore): - def __init__(self, hs): - super(RoomMemberStore, self).__init__(hs) - self.register_background_update_handler( - _MEMBERSHIP_PROFILE_UPDATE_NAME, self._background_add_membership_profile - ) - - def _store_room_members_txn(self, txn, events, backfilled): - """Store a room member in the database. - """ - self._simple_insert_many_txn( - txn, - table="room_memberships", - values=[ - { - "event_id": event.event_id, - "user_id": event.state_key, - "sender": event.user_id, - "room_id": event.room_id, - "membership": event.membership, - "display_name": event.content.get("displayname", None), - "avatar_url": event.content.get("avatar_url", None), - } - for event in events - ] - ) - - for event in events: - txn.call_after( - self._membership_stream_cache.entity_has_changed, - event.state_key, event.internal_metadata.stream_ordering - ) - txn.call_after( - self.get_invited_rooms_for_user.invalidate, (event.state_key,) - ) - - # We update the local_invites table only if the event is "current", - # i.e., its something that has just happened. - # The only current event that can also be an outlier is if its an - # invite that has come in across federation. - is_new_state = not backfilled and ( - not event.internal_metadata.is_outlier() - or event.internal_metadata.is_invite_from_remote() - ) - is_mine = self.hs.is_mine_id(event.state_key) - if is_new_state and is_mine: - if event.membership == Membership.INVITE: - self._simple_insert_txn( - txn, - table="local_invites", - values={ - "event_id": event.event_id, - "invitee": event.state_key, - "inviter": event.sender, - "room_id": event.room_id, - "stream_id": event.internal_metadata.stream_ordering, - } - ) - else: - sql = ( - "UPDATE local_invites SET stream_id = ?, replaced_by = ? WHERE" - " room_id = ? AND invitee = ? AND locally_rejected is NULL" - " AND replaced_by is NULL" - ) - - txn.execute(sql, ( - event.internal_metadata.stream_ordering, - event.event_id, - event.room_id, - event.state_key, - )) - - @defer.inlineCallbacks - def locally_reject_invite(self, user_id, room_id): - sql = ( - "UPDATE local_invites SET stream_id = ?, locally_rejected = ? WHERE" - " room_id = ? AND invitee = ? AND locally_rejected is NULL" - " AND replaced_by is NULL" - ) - - def f(txn, stream_ordering): - txn.execute(sql, ( - stream_ordering, - True, - room_id, - user_id, - )) - - with self._stream_id_gen.get_next() as stream_ordering: - yield self.runInteraction("locally_reject_invite", f, stream_ordering) - +class RoomMemberWorkerStore(EventsWorkerStore): @cachedInlineCallbacks(max_entries=100000, iterable=True, cache_context=True) def get_hosts_in_room(self, room_id, cache_context): """Returns the set of all hosts currently in the room @@ -270,12 +187,32 @@ class RoomMemberStore(SQLBaseStore): return results @cachedInlineCallbacks(max_entries=500000, iterable=True) - def get_rooms_for_user(self, user_id): + def get_rooms_for_user_with_stream_ordering(self, user_id): """Returns a set of room_ids the user is currently joined to + + Args: + user_id (str) + + Returns: + Deferred[frozenset[GetRoomsForUserWithStreamOrdering]]: Returns + the rooms the user is in currently, along with the stream ordering + of the most recent join for that user and room. """ rooms = yield self.get_rooms_for_user_where_membership_is( user_id, membership_list=[Membership.JOIN], ) + defer.returnValue(frozenset( + GetRoomsForUserWithStreamOrdering(r.room_id, r.stream_ordering) + for r in rooms + )) + + @defer.inlineCallbacks + def get_rooms_for_user(self, user_id, on_invalidate=None): + """Returns a set of room_ids the user is currently joined to + """ + rooms = yield self.get_rooms_for_user_with_stream_ordering( + user_id, on_invalidate=on_invalidate, + ) defer.returnValue(frozenset(r.room_id for r in rooms)) @cachedInlineCallbacks(max_entries=500000, cache_context=True, iterable=True) @@ -295,89 +232,7 @@ class RoomMemberStore(SQLBaseStore): defer.returnValue(user_who_share_room) - def forget(self, user_id, room_id): - """Indicate that user_id wishes to discard history for room_id.""" - def f(txn): - sql = ( - "UPDATE" - " room_memberships" - " SET" - " forgotten = 1" - " WHERE" - " user_id = ?" - " AND" - " room_id = ?" - ) - txn.execute(sql, (user_id, room_id)) - - txn.call_after(self.was_forgotten_at.invalidate_all) - txn.call_after(self.did_forget.invalidate, (user_id, room_id)) - self._invalidate_cache_and_stream( - txn, self.who_forgot_in_room, (room_id,) - ) - return self.runInteraction("forget_membership", f) - - @cachedInlineCallbacks(num_args=2) - def did_forget(self, user_id, room_id): - """Returns whether user_id has elected to discard history for room_id. - - Returns False if they have since re-joined.""" - def f(txn): - sql = ( - "SELECT" - " COUNT(*)" - " FROM" - " room_memberships" - " WHERE" - " user_id = ?" - " AND" - " room_id = ?" - " AND" - " forgotten = 0" - ) - txn.execute(sql, (user_id, room_id)) - rows = txn.fetchall() - return rows[0][0] - count = yield self.runInteraction("did_forget_membership", f) - defer.returnValue(count == 0) - - @cachedInlineCallbacks(num_args=3) - def was_forgotten_at(self, user_id, room_id, event_id): - """Returns whether user_id has elected to discard history for room_id at - event_id. - - event_id must be a membership event.""" - def f(txn): - sql = ( - "SELECT" - " forgotten" - " FROM" - " room_memberships" - " WHERE" - " user_id = ?" - " AND" - " room_id = ?" - " AND" - " event_id = ?" - ) - txn.execute(sql, (user_id, room_id, event_id)) - rows = txn.fetchall() - return rows[0][0] - forgot = yield self.runInteraction("did_forget_membership_at", f) - defer.returnValue(forgot == 1) - - @cached() - def who_forgot_in_room(self, room_id): - return self._simple_select_list( - table="room_memberships", - retcols=("user_id", "event_id"), - keyvalues={ - "room_id": room_id, - "forgotten": 1, - }, - desc="who_forgot" - ) - + @defer.inlineCallbacks def get_joined_users_from_context(self, event, context): state_group = context.state_group if not state_group: @@ -387,11 +242,13 @@ class RoomMemberStore(SQLBaseStore): # To do this we set the state_group to a new object as object() != object() state_group = object() - return self._get_joined_users_from_context( - event.room_id, state_group, context.current_state_ids, + current_state_ids = yield context.get_current_state_ids(self) + result = yield self._get_joined_users_from_context( + event.room_id, state_group, current_state_ids, event=event, context=context, ) + defer.returnValue(result) def get_joined_users_from_state(self, room_id, state_entry): state_group = state_entry.state_group @@ -419,7 +276,7 @@ class RoomMemberStore(SQLBaseStore): users_in_room = {} member_event_ids = [ e_id - for key, e_id in current_state_ids.iteritems() + for key, e_id in iteritems(current_state_ids) if key[0] == EventTypes.Member ] @@ -436,7 +293,7 @@ class RoomMemberStore(SQLBaseStore): users_in_room = dict(prev_res) member_event_ids = [ e_id - for key, e_id in context.delta_ids.iteritems() + for key, e_id in iteritems(context.delta_ids) if key[0] == EventTypes.Member ] for etype, state_key in context.delta_ids: @@ -533,6 +390,46 @@ class RoomMemberStore(SQLBaseStore): defer.returnValue(True) + @cachedInlineCallbacks() + def was_host_joined(self, room_id, host): + """Check whether the server is or ever was in the room. + + Args: + room_id (str) + host (str) + + Returns: + Deferred: Resolves to True if the host is/was in the room, otherwise + False. + """ + if '%' in host or '_' in host: + raise Exception("Invalid host name") + + sql = """ + SELECT user_id FROM room_memberships + WHERE room_id = ? + AND user_id LIKE ? + AND membership = 'join' + LIMIT 1 + """ + + # We do need to be careful to ensure that host doesn't have any wild cards + # in it, but we checked above for known ones and we'll check below that + # the returned user actually has the correct domain. + like_clause = "%:" + host + + rows = yield self._execute("was_host_joined", None, sql, room_id, like_clause) + + if not rows: + defer.returnValue(False) + + user_id = rows[0][0] + if get_domain_from_id(user_id) != host: + # This can only happen if the host name has something funky in it + raise Exception("Invalid host name") + + defer.returnValue(True) + def get_joined_hosts(self, room_id, state_entry): state_group = state_entry.state_group if not state_group: @@ -560,6 +457,144 @@ class RoomMemberStore(SQLBaseStore): defer.returnValue(joined_hosts) + @cached(max_entries=10000) + def _get_joined_hosts_cache(self, room_id): + return _JoinedHostsCache(self, room_id) + + +class RoomMemberStore(RoomMemberWorkerStore): + def __init__(self, db_conn, hs): + super(RoomMemberStore, self).__init__(db_conn, hs) + self.register_background_update_handler( + _MEMBERSHIP_PROFILE_UPDATE_NAME, self._background_add_membership_profile + ) + + def _store_room_members_txn(self, txn, events, backfilled): + """Store a room member in the database. + """ + self._simple_insert_many_txn( + txn, + table="room_memberships", + values=[ + { + "event_id": event.event_id, + "user_id": event.state_key, + "sender": event.user_id, + "room_id": event.room_id, + "membership": event.membership, + "display_name": event.content.get("displayname", None), + "avatar_url": event.content.get("avatar_url", None), + } + for event in events + ] + ) + + for event in events: + txn.call_after( + self._membership_stream_cache.entity_has_changed, + event.state_key, event.internal_metadata.stream_ordering + ) + txn.call_after( + self.get_invited_rooms_for_user.invalidate, (event.state_key,) + ) + + # We update the local_invites table only if the event is "current", + # i.e., its something that has just happened. + # The only current event that can also be an outlier is if its an + # invite that has come in across federation. + is_new_state = not backfilled and ( + not event.internal_metadata.is_outlier() + or event.internal_metadata.is_invite_from_remote() + ) + is_mine = self.hs.is_mine_id(event.state_key) + if is_new_state and is_mine: + if event.membership == Membership.INVITE: + self._simple_insert_txn( + txn, + table="local_invites", + values={ + "event_id": event.event_id, + "invitee": event.state_key, + "inviter": event.sender, + "room_id": event.room_id, + "stream_id": event.internal_metadata.stream_ordering, + } + ) + else: + sql = ( + "UPDATE local_invites SET stream_id = ?, replaced_by = ? WHERE" + " room_id = ? AND invitee = ? AND locally_rejected is NULL" + " AND replaced_by is NULL" + ) + + txn.execute(sql, ( + event.internal_metadata.stream_ordering, + event.event_id, + event.room_id, + event.state_key, + )) + + @defer.inlineCallbacks + def locally_reject_invite(self, user_id, room_id): + sql = ( + "UPDATE local_invites SET stream_id = ?, locally_rejected = ? WHERE" + " room_id = ? AND invitee = ? AND locally_rejected is NULL" + " AND replaced_by is NULL" + ) + + def f(txn, stream_ordering): + txn.execute(sql, ( + stream_ordering, + True, + room_id, + user_id, + )) + + with self._stream_id_gen.get_next() as stream_ordering: + yield self.runInteraction("locally_reject_invite", f, stream_ordering) + + def forget(self, user_id, room_id): + """Indicate that user_id wishes to discard history for room_id.""" + def f(txn): + sql = ( + "UPDATE" + " room_memberships" + " SET" + " forgotten = 1" + " WHERE" + " user_id = ?" + " AND" + " room_id = ?" + ) + txn.execute(sql, (user_id, room_id)) + + txn.call_after(self.did_forget.invalidate, (user_id, room_id)) + return self.runInteraction("forget_membership", f) + + @cachedInlineCallbacks(num_args=2) + def did_forget(self, user_id, room_id): + """Returns whether user_id has elected to discard history for room_id. + + Returns False if they have since re-joined.""" + def f(txn): + sql = ( + "SELECT" + " COUNT(*)" + " FROM" + " room_memberships" + " WHERE" + " user_id = ?" + " AND" + " room_id = ?" + " AND" + " forgotten = 0" + ) + txn.execute(sql, (user_id, room_id)) + rows = txn.fetchall() + return rows[0][0] + count = yield self.runInteraction("did_forget_membership", f) + defer.returnValue(count == 0) + @defer.inlineCallbacks def _background_add_membership_profile(self, progress, batch_size): target_min_stream_id = progress.get( @@ -573,8 +608,9 @@ class RoomMemberStore(SQLBaseStore): def add_membership_profile_txn(txn): sql = (""" - SELECT stream_ordering, event_id, events.room_id, content + SELECT stream_ordering, event_id, events.room_id, event_json.json FROM events + INNER JOIN event_json USING (event_id) INNER JOIN room_memberships USING (event_id) WHERE ? <= stream_ordering AND stream_ordering < ? AND type = 'm.room.member' @@ -595,8 +631,9 @@ class RoomMemberStore(SQLBaseStore): event_id = row["event_id"] room_id = row["room_id"] try: - content = json.loads(row["content"]) - except: + event_json = json.loads(row["json"]) + content = event_json['content'] + except Exception: continue display_name = content.get("displayname", None) @@ -635,10 +672,6 @@ class RoomMemberStore(SQLBaseStore): defer.returnValue(result) - @cached(max_entries=10000, iterable=True) - def _get_joined_hosts_cache(self, room_id): - return _JoinedHostsCache(self, room_id) - class _JoinedHostsCache(object): """Cache for joined hosts in a room that is optimised to handle updates @@ -671,7 +704,7 @@ class _JoinedHostsCache(object): if state_entry.state_group == self.state_group: pass elif state_entry.prev_group == self.state_group: - for (typ, state_key), event_id in state_entry.delta_ids.iteritems(): + for (typ, state_key), event_id in iteritems(state_entry.delta_ids): if typ != EventTypes.Member: continue @@ -701,7 +734,7 @@ class _JoinedHostsCache(object): self.state_group = state_entry.state_group else: self.state_group = object() - self._len = sum(len(v) for v in self.hosts_to_joined_users.itervalues()) + self._len = sum(len(v) for v in itervalues(self.hosts_to_joined_users)) defer.returnValue(frozenset(self.hosts_to_joined_users)) def __len__(self): diff --git a/synapse/storage/schema/delta/14/upgrade_appservice_db.py b/synapse/storage/schema/delta/14/upgrade_appservice_db.py index 8755bb2e4..4d725b92f 100644 --- a/synapse/storage/schema/delta/14/upgrade_appservice_db.py +++ b/synapse/storage/schema/delta/14/upgrade_appservice_db.py @@ -12,9 +12,10 @@ # See the License for the specific language governing permissions and # limitations under the License. -import json import logging +import simplejson as json + logger = logging.getLogger(__name__) diff --git a/synapse/storage/schema/delta/25/fts.py b/synapse/storage/schema/delta/25/fts.py index 4269ac69a..4b2ffd35f 100644 --- a/synapse/storage/schema/delta/25/fts.py +++ b/synapse/storage/schema/delta/25/fts.py @@ -14,10 +14,10 @@ import logging -from synapse.storage.prepare_database import get_statements -from synapse.storage.engines import PostgresEngine, Sqlite3Engine +import simplejson -import ujson +from synapse.storage.engines import PostgresEngine, Sqlite3Engine +from synapse.storage.prepare_database import get_statements logger = logging.getLogger(__name__) @@ -66,7 +66,7 @@ def run_create(cur, database_engine, *args, **kwargs): "max_stream_id_exclusive": max_stream_id + 1, "rows_inserted": 0, } - progress_json = ujson.dumps(progress) + progress_json = simplejson.dumps(progress) sql = ( "INSERT into background_updates (update_name, progress_json)" diff --git a/synapse/storage/schema/delta/27/ts.py b/synapse/storage/schema/delta/27/ts.py index 71b12a273..414f9f5aa 100644 --- a/synapse/storage/schema/delta/27/ts.py +++ b/synapse/storage/schema/delta/27/ts.py @@ -14,9 +14,9 @@ import logging -from synapse.storage.prepare_database import get_statements +import simplejson -import ujson +from synapse.storage.prepare_database import get_statements logger = logging.getLogger(__name__) @@ -45,7 +45,7 @@ def run_create(cur, database_engine, *args, **kwargs): "max_stream_id_exclusive": max_stream_id + 1, "rows_inserted": 0, } - progress_json = ujson.dumps(progress) + progress_json = simplejson.dumps(progress) sql = ( "INSERT into background_updates (update_name, progress_json)" diff --git a/synapse/storage/schema/delta/30/as_users.py b/synapse/storage/schema/delta/30/as_users.py index 5b7d8d1ab..ef7ec3434 100644 --- a/synapse/storage/schema/delta/30/as_users.py +++ b/synapse/storage/schema/delta/30/as_users.py @@ -12,8 +12,10 @@ # See the License for the specific language governing permissions and # limitations under the License. import logging -from synapse.config.appservice import load_appservices +from six.moves import range + +from synapse.config.appservice import load_appservices logger = logging.getLogger(__name__) @@ -22,7 +24,7 @@ def run_create(cur, database_engine, *args, **kwargs): # NULL indicates user was not registered by an appservice. try: cur.execute("ALTER TABLE users ADD COLUMN appservice_id TEXT") - except: + except Exception: # Maybe we already added the column? Hope so... pass @@ -58,7 +60,7 @@ def run_upgrade(cur, database_engine, config, *args, **kwargs): for as_id, user_ids in owned.items(): n = 100 - user_chunks = (user_ids[i:i + 100] for i in xrange(0, len(user_ids), n)) + user_chunks = (user_ids[i:i + 100] for i in range(0, len(user_ids), n)) for chunk in user_chunks: cur.execute( database_engine.convert_param_style( diff --git a/synapse/storage/schema/delta/31/search_update.py b/synapse/storage/schema/delta/31/search_update.py index 470ae0c00..7d8ca5f93 100644 --- a/synapse/storage/schema/delta/31/search_update.py +++ b/synapse/storage/schema/delta/31/search_update.py @@ -12,12 +12,13 @@ # See the License for the specific language governing permissions and # limitations under the License. +import logging + +import simplejson + from synapse.storage.engines import PostgresEngine from synapse.storage.prepare_database import get_statements -import logging -import ujson - logger = logging.getLogger(__name__) @@ -49,7 +50,7 @@ def run_create(cur, database_engine, *args, **kwargs): "rows_inserted": 0, "have_added_indexes": False, } - progress_json = ujson.dumps(progress) + progress_json = simplejson.dumps(progress) sql = ( "INSERT into background_updates (update_name, progress_json)" diff --git a/synapse/storage/schema/delta/33/event_fields.py b/synapse/storage/schema/delta/33/event_fields.py index 83066cccc..bff1256a7 100644 --- a/synapse/storage/schema/delta/33/event_fields.py +++ b/synapse/storage/schema/delta/33/event_fields.py @@ -12,10 +12,11 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.storage.prepare_database import get_statements - import logging -import ujson + +import simplejson + +from synapse.storage.prepare_database import get_statements logger = logging.getLogger(__name__) @@ -44,7 +45,7 @@ def run_create(cur, database_engine, *args, **kwargs): "max_stream_id_exclusive": max_stream_id + 1, "rows_inserted": 0, } - progress_json = ujson.dumps(progress) + progress_json = simplejson.dumps(progress) sql = ( "INSERT into background_updates (update_name, progress_json)" diff --git a/synapse/storage/schema/delta/33/remote_media_ts.py b/synapse/storage/schema/delta/33/remote_media_ts.py index 55ae43f39..9754d3ccf 100644 --- a/synapse/storage/schema/delta/33/remote_media_ts.py +++ b/synapse/storage/schema/delta/33/remote_media_ts.py @@ -14,7 +14,6 @@ import time - ALTER_TABLE = "ALTER TABLE remote_media_cache ADD COLUMN last_access_ts BIGINT" diff --git a/synapse/storage/schema/delta/34/cache_stream.py b/synapse/storage/schema/delta/34/cache_stream.py index 3b63a1562..cf09e43e2 100644 --- a/synapse/storage/schema/delta/34/cache_stream.py +++ b/synapse/storage/schema/delta/34/cache_stream.py @@ -12,11 +12,11 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.storage.prepare_database import get_statements -from synapse.storage.engines import PostgresEngine - import logging +from synapse.storage.engines import PostgresEngine +from synapse.storage.prepare_database import get_statements + logger = logging.getLogger(__name__) diff --git a/synapse/storage/schema/delta/34/received_txn_purge.py b/synapse/storage/schema/delta/34/received_txn_purge.py index 033144341..67d505e68 100644 --- a/synapse/storage/schema/delta/34/received_txn_purge.py +++ b/synapse/storage/schema/delta/34/received_txn_purge.py @@ -12,10 +12,10 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.storage.engines import PostgresEngine - import logging +from synapse.storage.engines import PostgresEngine + logger = logging.getLogger(__name__) diff --git a/synapse/storage/schema/delta/34/sent_txn_purge.py b/synapse/storage/schema/delta/34/sent_txn_purge.py index 81948e343..0ffab10b6 100644 --- a/synapse/storage/schema/delta/34/sent_txn_purge.py +++ b/synapse/storage/schema/delta/34/sent_txn_purge.py @@ -12,10 +12,10 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.storage.engines import PostgresEngine - import logging +from synapse.storage.engines import PostgresEngine + logger = logging.getLogger(__name__) diff --git a/synapse/storage/schema/delta/37/remove_auth_idx.py b/synapse/storage/schema/delta/37/remove_auth_idx.py index 20ad8bd5a..a37788416 100644 --- a/synapse/storage/schema/delta/37/remove_auth_idx.py +++ b/synapse/storage/schema/delta/37/remove_auth_idx.py @@ -12,11 +12,11 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.storage.prepare_database import get_statements -from synapse.storage.engines import PostgresEngine - import logging +from synapse.storage.engines import PostgresEngine +from synapse.storage.prepare_database import get_statements + logger = logging.getLogger(__name__) DROP_INDICES = """ diff --git a/synapse/storage/schema/delta/38/postgres_fts_gist.sql b/synapse/storage/schema/delta/38/postgres_fts_gist.sql index f090a7b75..515e6b8e8 100644 --- a/synapse/storage/schema/delta/38/postgres_fts_gist.sql +++ b/synapse/storage/schema/delta/38/postgres_fts_gist.sql @@ -13,5 +13,7 @@ * limitations under the License. */ - INSERT into background_updates (update_name, progress_json) - VALUES ('event_search_postgres_gist', '{}'); +-- We no longer do this given we back it out again in schema 47 + +-- INSERT into background_updates (update_name, progress_json) +-- VALUES ('event_search_postgres_gist', '{}'); diff --git a/synapse/storage/schema/delta/42/user_dir.py b/synapse/storage/schema/delta/42/user_dir.py index ea6a18196..506f326f4 100644 --- a/synapse/storage/schema/delta/42/user_dir.py +++ b/synapse/storage/schema/delta/42/user_dir.py @@ -14,8 +14,8 @@ import logging -from synapse.storage.prepare_database import get_statements from synapse.storage.engines import PostgresEngine, Sqlite3Engine +from synapse.storage.prepare_database import get_statements logger = logging.getLogger(__name__) diff --git a/synapse/storage/schema/delta/43/user_share.sql b/synapse/storage/schema/delta/43/user_share.sql index 4501d90cb..ee7062abe 100644 --- a/synapse/storage/schema/delta/43/user_share.sql +++ b/synapse/storage/schema/delta/43/user_share.sql @@ -29,5 +29,5 @@ CREATE INDEX users_who_share_rooms_r_idx ON users_who_share_rooms(room_id); CREATE INDEX users_who_share_rooms_o_idx ON users_who_share_rooms(other_user_id); --- Make sure that we popualte the table initially +-- Make sure that we populate the table initially UPDATE user_directory_stream_pos SET stream_id = NULL; diff --git a/synapse/storage/schema/delta/44/expire_url_cache.sql b/synapse/storage/schema/delta/44/expire_url_cache.sql new file mode 100644 index 000000000..b12f9b2eb --- /dev/null +++ b/synapse/storage/schema/delta/44/expire_url_cache.sql @@ -0,0 +1,41 @@ +/* Copyright 2017 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- this didn't work on SQLite 3.7 (because of lack of partial indexes), so was +-- removed and replaced with 46/local_media_repository_url_idx.sql. +-- +-- CREATE INDEX local_media_repository_url_idx ON local_media_repository(created_ts) WHERE url_cache IS NOT NULL; + +-- we need to change `expires` to `expires_ts` so that we can index on it. SQLite doesn't support +-- indices on expressions until 3.9. +CREATE TABLE local_media_repository_url_cache_new( + url TEXT, + response_code INTEGER, + etag TEXT, + expires_ts BIGINT, + og TEXT, + media_id TEXT, + download_ts BIGINT +); + +INSERT INTO local_media_repository_url_cache_new + SELECT url, response_code, etag, expires + download_ts, og, media_id, download_ts FROM local_media_repository_url_cache; + +DROP TABLE local_media_repository_url_cache; +ALTER TABLE local_media_repository_url_cache_new RENAME TO local_media_repository_url_cache; + +CREATE INDEX local_media_repository_url_cache_expires_idx ON local_media_repository_url_cache(expires_ts); +CREATE INDEX local_media_repository_url_cache_by_url_download_ts ON local_media_repository_url_cache(url, download_ts); +CREATE INDEX local_media_repository_url_cache_media_idx ON local_media_repository_url_cache(media_id); diff --git a/synapse/storage/schema/delta/45/group_server.sql b/synapse/storage/schema/delta/45/group_server.sql new file mode 100644 index 000000000..b2333848a --- /dev/null +++ b/synapse/storage/schema/delta/45/group_server.sql @@ -0,0 +1,167 @@ +/* Copyright 2017 Vector Creations Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CREATE TABLE groups ( + group_id TEXT NOT NULL, + name TEXT, -- the display name of the room + avatar_url TEXT, + short_description TEXT, + long_description TEXT +); + +CREATE UNIQUE INDEX groups_idx ON groups(group_id); + + +-- list of users the group server thinks are joined +CREATE TABLE group_users ( + group_id TEXT NOT NULL, + user_id TEXT NOT NULL, + is_admin BOOLEAN NOT NULL, + is_public BOOLEAN NOT NULL -- whether the users membership can be seen by everyone +); + + +CREATE INDEX groups_users_g_idx ON group_users(group_id, user_id); +CREATE INDEX groups_users_u_idx ON group_users(user_id); + +-- list of users the group server thinks are invited +CREATE TABLE group_invites ( + group_id TEXT NOT NULL, + user_id TEXT NOT NULL +); + +CREATE INDEX groups_invites_g_idx ON group_invites(group_id, user_id); +CREATE INDEX groups_invites_u_idx ON group_invites(user_id); + + +CREATE TABLE group_rooms ( + group_id TEXT NOT NULL, + room_id TEXT NOT NULL, + is_public BOOLEAN NOT NULL -- whether the room can be seen by everyone +); + +CREATE UNIQUE INDEX groups_rooms_g_idx ON group_rooms(group_id, room_id); +CREATE INDEX groups_rooms_r_idx ON group_rooms(room_id); + + +-- Rooms to include in the summary +CREATE TABLE group_summary_rooms ( + group_id TEXT NOT NULL, + room_id TEXT NOT NULL, + category_id TEXT NOT NULL, + room_order BIGINT NOT NULL, + is_public BOOLEAN NOT NULL, -- whether the room should be show to everyone + UNIQUE (group_id, category_id, room_id, room_order), + CHECK (room_order > 0) +); + +CREATE UNIQUE INDEX group_summary_rooms_g_idx ON group_summary_rooms(group_id, room_id, category_id); + + +-- Categories to include in the summary +CREATE TABLE group_summary_room_categories ( + group_id TEXT NOT NULL, + category_id TEXT NOT NULL, + cat_order BIGINT NOT NULL, + UNIQUE (group_id, category_id, cat_order), + CHECK (cat_order > 0) +); + +-- The categories in the group +CREATE TABLE group_room_categories ( + group_id TEXT NOT NULL, + category_id TEXT NOT NULL, + profile TEXT NOT NULL, + is_public BOOLEAN NOT NULL, -- whether the category should be show to everyone + UNIQUE (group_id, category_id) +); + +-- The users to include in the group summary +CREATE TABLE group_summary_users ( + group_id TEXT NOT NULL, + user_id TEXT NOT NULL, + role_id TEXT NOT NULL, + user_order BIGINT NOT NULL, + is_public BOOLEAN NOT NULL -- whether the user should be show to everyone +); + +CREATE INDEX group_summary_users_g_idx ON group_summary_users(group_id); + +-- The roles to include in the group summary +CREATE TABLE group_summary_roles ( + group_id TEXT NOT NULL, + role_id TEXT NOT NULL, + role_order BIGINT NOT NULL, + UNIQUE (group_id, role_id, role_order), + CHECK (role_order > 0) +); + + +-- The roles in a groups +CREATE TABLE group_roles ( + group_id TEXT NOT NULL, + role_id TEXT NOT NULL, + profile TEXT NOT NULL, + is_public BOOLEAN NOT NULL, -- whether the role should be show to everyone + UNIQUE (group_id, role_id) +); + + +-- List of attestations we've given out and need to renew +CREATE TABLE group_attestations_renewals ( + group_id TEXT NOT NULL, + user_id TEXT NOT NULL, + valid_until_ms BIGINT NOT NULL +); + +CREATE INDEX group_attestations_renewals_g_idx ON group_attestations_renewals(group_id, user_id); +CREATE INDEX group_attestations_renewals_u_idx ON group_attestations_renewals(user_id); +CREATE INDEX group_attestations_renewals_v_idx ON group_attestations_renewals(valid_until_ms); + + +-- List of attestations we've received from remotes and are interested in. +CREATE TABLE group_attestations_remote ( + group_id TEXT NOT NULL, + user_id TEXT NOT NULL, + valid_until_ms BIGINT NOT NULL, + attestation_json TEXT NOT NULL +); + +CREATE INDEX group_attestations_remote_g_idx ON group_attestations_remote(group_id, user_id); +CREATE INDEX group_attestations_remote_u_idx ON group_attestations_remote(user_id); +CREATE INDEX group_attestations_remote_v_idx ON group_attestations_remote(valid_until_ms); + + +-- The group membership for the HS's users +CREATE TABLE local_group_membership ( + group_id TEXT NOT NULL, + user_id TEXT NOT NULL, + is_admin BOOLEAN NOT NULL, + membership TEXT NOT NULL, + is_publicised BOOLEAN NOT NULL, -- if the user is publicising their membership + content TEXT NOT NULL +); + +CREATE INDEX local_group_membership_u_idx ON local_group_membership(user_id, group_id); +CREATE INDEX local_group_membership_g_idx ON local_group_membership(group_id); + + +CREATE TABLE local_group_updates ( + stream_id BIGINT NOT NULL, + group_id TEXT NOT NULL, + user_id TEXT NOT NULL, + type TEXT NOT NULL, + content TEXT NOT NULL +); diff --git a/synapse/storage/schema/delta/45/profile_cache.sql b/synapse/storage/schema/delta/45/profile_cache.sql new file mode 100644 index 000000000..e5ddc84df --- /dev/null +++ b/synapse/storage/schema/delta/45/profile_cache.sql @@ -0,0 +1,28 @@ +/* Copyright 2017 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + + +-- A subset of remote users whose profiles we have cached. +-- Whether a user is in this table or not is defined by the storage function +-- `is_subscribed_remote_profile_for_user` +CREATE TABLE remote_profile_cache ( + user_id TEXT NOT NULL, + displayname TEXT, + avatar_url TEXT, + last_check BIGINT NOT NULL +); + +CREATE UNIQUE INDEX remote_profile_cache_user_id ON remote_profile_cache(user_id); +CREATE INDEX remote_profile_cache_time ON remote_profile_cache(last_check); diff --git a/synapse/storage/schema/delta/33/refreshtoken_device_index.sql b/synapse/storage/schema/delta/46/drop_refresh_tokens.sql similarity index 79% rename from synapse/storage/schema/delta/33/refreshtoken_device_index.sql rename to synapse/storage/schema/delta/46/drop_refresh_tokens.sql index bb225dafb..68c48a89a 100644 --- a/synapse/storage/schema/delta/33/refreshtoken_device_index.sql +++ b/synapse/storage/schema/delta/46/drop_refresh_tokens.sql @@ -1,4 +1,4 @@ -/* Copyright 2016 OpenMarket Ltd +/* Copyright 2017 New Vector Ltd * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. @@ -13,5 +13,5 @@ * limitations under the License. */ -INSERT INTO background_updates (update_name, progress_json) VALUES - ('refresh_tokens_device_index', '{}'); +/* we no longer use (or create) the refresh_tokens table */ +DROP TABLE IF EXISTS refresh_tokens; diff --git a/synapse/storage/schema/delta/46/drop_unique_deleted_pushers.sql b/synapse/storage/schema/delta/46/drop_unique_deleted_pushers.sql new file mode 100644 index 000000000..bb307889c --- /dev/null +++ b/synapse/storage/schema/delta/46/drop_unique_deleted_pushers.sql @@ -0,0 +1,35 @@ +/* Copyright 2017 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- drop the unique constraint on deleted_pushers so that we can just insert +-- into it rather than upserting. + +CREATE TABLE deleted_pushers2 ( + stream_id BIGINT NOT NULL, + app_id TEXT NOT NULL, + pushkey TEXT NOT NULL, + user_id TEXT NOT NULL +); + +INSERT INTO deleted_pushers2 (stream_id, app_id, pushkey, user_id) + SELECT stream_id, app_id, pushkey, user_id from deleted_pushers; + +DROP TABLE deleted_pushers; +ALTER TABLE deleted_pushers2 RENAME TO deleted_pushers; + +-- create the index after doing the inserts because that's more efficient. +-- it also means we can give it the same name as the old one without renaming. +CREATE INDEX deleted_pushers_stream_id ON deleted_pushers (stream_id); + diff --git a/synapse/storage/schema/delta/46/group_server.sql b/synapse/storage/schema/delta/46/group_server.sql new file mode 100644 index 000000000..097679bc9 --- /dev/null +++ b/synapse/storage/schema/delta/46/group_server.sql @@ -0,0 +1,32 @@ +/* Copyright 2017 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CREATE TABLE groups_new ( + group_id TEXT NOT NULL, + name TEXT, -- the display name of the room + avatar_url TEXT, + short_description TEXT, + long_description TEXT, + is_public BOOL NOT NULL -- whether non-members can access group APIs +); + +-- NB: awful hack to get the default to be true on postgres and 1 on sqlite +INSERT INTO groups_new + SELECT group_id, name, avatar_url, short_description, long_description, (1=1) FROM groups; + +DROP TABLE groups; +ALTER TABLE groups_new RENAME TO groups; + +CREATE UNIQUE INDEX groups_idx ON groups(group_id); diff --git a/synapse/storage/schema/delta/46/local_media_repository_url_idx.sql b/synapse/storage/schema/delta/46/local_media_repository_url_idx.sql new file mode 100644 index 000000000..bbfc7f5d1 --- /dev/null +++ b/synapse/storage/schema/delta/46/local_media_repository_url_idx.sql @@ -0,0 +1,24 @@ +/* Copyright 2017 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- register a background update which will recreate the +-- local_media_repository_url_idx index. +-- +-- We do this as a bg update not because it is a particularly onerous +-- operation, but because we'd like it to be a partial index if possible, and +-- the background_index_update code will understand whether we are on +-- postgres or sqlite and behave accordingly. +INSERT INTO background_updates (update_name, progress_json) VALUES + ('local_media_repository_url_idx', '{}'); diff --git a/synapse/storage/schema/delta/46/user_dir_null_room_ids.sql b/synapse/storage/schema/delta/46/user_dir_null_room_ids.sql new file mode 100644 index 000000000..cb0d5a257 --- /dev/null +++ b/synapse/storage/schema/delta/46/user_dir_null_room_ids.sql @@ -0,0 +1,35 @@ +/* Copyright 2017 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- change the user_directory table to also cover global local user profiles +-- rather than just profiles within specific rooms. + +CREATE TABLE user_directory2 ( + user_id TEXT NOT NULL, + room_id TEXT, + display_name TEXT, + avatar_url TEXT +); + +INSERT INTO user_directory2(user_id, room_id, display_name, avatar_url) + SELECT user_id, room_id, display_name, avatar_url from user_directory; + +DROP TABLE user_directory; +ALTER TABLE user_directory2 RENAME TO user_directory; + +-- create indexes after doing the inserts because that's more efficient. +-- it also means we can give it the same name as the old one without renaming. +CREATE INDEX user_directory_room_idx ON user_directory(room_id); +CREATE UNIQUE INDEX user_directory_user_idx ON user_directory(user_id); diff --git a/synapse/storage/schema/delta/46/user_dir_typos.sql b/synapse/storage/schema/delta/46/user_dir_typos.sql new file mode 100644 index 000000000..d9505f8da --- /dev/null +++ b/synapse/storage/schema/delta/46/user_dir_typos.sql @@ -0,0 +1,24 @@ +/* Copyright 2017 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- this is just embarassing :| +ALTER TABLE users_in_pubic_room RENAME TO users_in_public_rooms; + +-- this is only 300K rows on matrix.org and takes ~3s to generate the index, +-- so is hopefully not going to block anyone else for that long... +CREATE INDEX users_in_public_rooms_room_idx ON users_in_public_rooms(room_id); +CREATE UNIQUE INDEX users_in_public_rooms_user_idx ON users_in_public_rooms(user_id); +DROP INDEX users_in_pubic_room_room_idx; +DROP INDEX users_in_pubic_room_user_idx; diff --git a/synapse/storage/schema/delta/33/refreshtoken_device.sql b/synapse/storage/schema/delta/47/last_access_media.sql similarity index 84% rename from synapse/storage/schema/delta/33/refreshtoken_device.sql rename to synapse/storage/schema/delta/47/last_access_media.sql index 290bd6da8..f505fb22b 100644 --- a/synapse/storage/schema/delta/33/refreshtoken_device.sql +++ b/synapse/storage/schema/delta/47/last_access_media.sql @@ -1,4 +1,4 @@ -/* Copyright 2016 OpenMarket Ltd +/* Copyright 2018 New Vector Ltd * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. @@ -13,4 +13,4 @@ * limitations under the License. */ -ALTER TABLE refresh_tokens ADD COLUMN device_id TEXT; +ALTER TABLE local_media_repository ADD COLUMN last_access_ts BIGINT; diff --git a/synapse/storage/schema/delta/23/refresh_tokens.sql b/synapse/storage/schema/delta/47/postgres_fts_gin.sql similarity index 75% rename from synapse/storage/schema/delta/23/refresh_tokens.sql rename to synapse/storage/schema/delta/47/postgres_fts_gin.sql index 34db0cf12..31d7a817e 100644 --- a/synapse/storage/schema/delta/23/refresh_tokens.sql +++ b/synapse/storage/schema/delta/47/postgres_fts_gin.sql @@ -1,4 +1,4 @@ -/* Copyright 2015, 2016 OpenMarket Ltd +/* Copyright 2018 New Vector Ltd * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. @@ -13,9 +13,5 @@ * limitations under the License. */ -CREATE TABLE IF NOT EXISTS refresh_tokens( - id INTEGER PRIMARY KEY, - token TEXT NOT NULL, - user_id TEXT NOT NULL, - UNIQUE (token) -); +INSERT into background_updates (update_name, progress_json) + VALUES ('event_search_postgres_gin', '{}'); diff --git a/synapse/storage/schema/delta/47/push_actions_staging.sql b/synapse/storage/schema/delta/47/push_actions_staging.sql new file mode 100644 index 000000000..edccf4a96 --- /dev/null +++ b/synapse/storage/schema/delta/47/push_actions_staging.sql @@ -0,0 +1,28 @@ +/* Copyright 2018 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- Temporary staging area for push actions that have been calculated for an +-- event, but the event hasn't yet been persisted. +-- When the event is persisted the rows are moved over to the +-- event_push_actions table. +CREATE TABLE event_push_actions_staging ( + event_id TEXT NOT NULL, + user_id TEXT NOT NULL, + actions TEXT NOT NULL, + notif SMALLINT NOT NULL, + highlight SMALLINT NOT NULL +); + +CREATE INDEX event_push_actions_staging_id ON event_push_actions_staging(event_id); diff --git a/synapse/storage/schema/delta/47/state_group_seq.py b/synapse/storage/schema/delta/47/state_group_seq.py new file mode 100644 index 000000000..f6766501d --- /dev/null +++ b/synapse/storage/schema/delta/47/state_group_seq.py @@ -0,0 +1,37 @@ +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from synapse.storage.engines import PostgresEngine + + +def run_create(cur, database_engine, *args, **kwargs): + if isinstance(database_engine, PostgresEngine): + # if we already have some state groups, we want to start making new + # ones with a higher id. + cur.execute("SELECT max(id) FROM state_groups") + row = cur.fetchone() + + if row[0] is None: + start_val = 1 + else: + start_val = row[0] + 1 + + cur.execute( + "CREATE SEQUENCE state_group_id_seq START WITH %s", + (start_val, ), + ) + + +def run_upgrade(*args, **kwargs): + pass diff --git a/synapse/storage/schema/delta/48/add_user_consent.sql b/synapse/storage/schema/delta/48/add_user_consent.sql new file mode 100644 index 000000000..523749150 --- /dev/null +++ b/synapse/storage/schema/delta/48/add_user_consent.sql @@ -0,0 +1,18 @@ +/* Copyright 2018 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +/* record the version of the privacy policy the user has consented to + */ +ALTER TABLE users ADD COLUMN consent_version TEXT; diff --git a/synapse/storage/schema/delta/48/add_user_ips_last_seen_index.sql b/synapse/storage/schema/delta/48/add_user_ips_last_seen_index.sql new file mode 100644 index 000000000..9248b0b24 --- /dev/null +++ b/synapse/storage/schema/delta/48/add_user_ips_last_seen_index.sql @@ -0,0 +1,17 @@ +/* Copyright 2018 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +INSERT into background_updates (update_name, progress_json) + VALUES ('user_ips_last_seen_index', '{}'); diff --git a/synapse/storage/schema/delta/48/deactivated_users.sql b/synapse/storage/schema/delta/48/deactivated_users.sql new file mode 100644 index 000000000..e9013a696 --- /dev/null +++ b/synapse/storage/schema/delta/48/deactivated_users.sql @@ -0,0 +1,25 @@ +/* Copyright 2018 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +/* + * Store any accounts that have been requested to be deactivated. + * We part the account from all the rooms its in when its + * deactivated. This can take some time and synapse may be restarted + * before it completes, so store the user IDs here until the process + * is complete. + */ +CREATE TABLE users_pending_deactivation ( + user_id TEXT NOT NULL +); diff --git a/synapse/storage/schema/delta/48/group_unique_indexes.py b/synapse/storage/schema/delta/48/group_unique_indexes.py new file mode 100644 index 000000000..2233af87d --- /dev/null +++ b/synapse/storage/schema/delta/48/group_unique_indexes.py @@ -0,0 +1,57 @@ +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from synapse.storage.engines import PostgresEngine +from synapse.storage.prepare_database import get_statements + +FIX_INDEXES = """ +-- rebuild indexes as uniques +DROP INDEX groups_invites_g_idx; +CREATE UNIQUE INDEX group_invites_g_idx ON group_invites(group_id, user_id); +DROP INDEX groups_users_g_idx; +CREATE UNIQUE INDEX group_users_g_idx ON group_users(group_id, user_id); + +-- rename other indexes to actually match their table names.. +DROP INDEX groups_users_u_idx; +CREATE INDEX group_users_u_idx ON group_users(user_id); +DROP INDEX groups_invites_u_idx; +CREATE INDEX group_invites_u_idx ON group_invites(user_id); +DROP INDEX groups_rooms_g_idx; +CREATE UNIQUE INDEX group_rooms_g_idx ON group_rooms(group_id, room_id); +DROP INDEX groups_rooms_r_idx; +CREATE INDEX group_rooms_r_idx ON group_rooms(room_id); +""" + + +def run_create(cur, database_engine, *args, **kwargs): + rowid = "ctid" if isinstance(database_engine, PostgresEngine) else "rowid" + + # remove duplicates from group_users & group_invites tables + cur.execute(""" + DELETE FROM group_users WHERE %s NOT IN ( + SELECT min(%s) FROM group_users GROUP BY group_id, user_id + ); + """ % (rowid, rowid)) + cur.execute(""" + DELETE FROM group_invites WHERE %s NOT IN ( + SELECT min(%s) FROM group_invites GROUP BY group_id, user_id + ); + """ % (rowid, rowid)) + + for statement in get_statements(FIX_INDEXES.splitlines()): + cur.execute(statement) + + +def run_upgrade(*args, **kwargs): + pass diff --git a/synapse/storage/schema/delta/48/groups_joinable.sql b/synapse/storage/schema/delta/48/groups_joinable.sql new file mode 100644 index 000000000..ce26eaf0c --- /dev/null +++ b/synapse/storage/schema/delta/48/groups_joinable.sql @@ -0,0 +1,22 @@ +/* Copyright 2018 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +/* + * This isn't a real ENUM because sqlite doesn't support it + * and we use a default of NULL for inserted rows and interpret + * NULL at the python store level as necessary so that existing + * rows are given the correct default policy. + */ +ALTER TABLE groups ADD COLUMN join_policy TEXT NOT NULL DEFAULT 'invite'; diff --git a/synapse/storage/schema/delta/49/add_user_consent_server_notice_sent.sql b/synapse/storage/schema/delta/49/add_user_consent_server_notice_sent.sql new file mode 100644 index 000000000..14dcf18d7 --- /dev/null +++ b/synapse/storage/schema/delta/49/add_user_consent_server_notice_sent.sql @@ -0,0 +1,20 @@ +/* Copyright 2018 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +/* record whether we have sent a server notice about consenting to the + * privacy policy. Specifically records the version of the policy we sent + * a message about. + */ +ALTER TABLE users ADD COLUMN consent_server_notice_sent TEXT; diff --git a/synapse/storage/schema/delta/49/add_user_daily_visits.sql b/synapse/storage/schema/delta/49/add_user_daily_visits.sql new file mode 100644 index 000000000..3dd478196 --- /dev/null +++ b/synapse/storage/schema/delta/49/add_user_daily_visits.sql @@ -0,0 +1,21 @@ +/* Copyright 2018 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + + +CREATE TABLE user_daily_visits ( user_id TEXT NOT NULL, + device_id TEXT, + timestamp BIGINT NOT NULL ); +CREATE INDEX user_daily_visits_uts_idx ON user_daily_visits(user_id, timestamp); +CREATE INDEX user_daily_visits_ts_idx ON user_daily_visits(timestamp); diff --git a/synapse/storage/schema/delta/49/add_user_ips_last_seen_only_index.sql b/synapse/storage/schema/delta/49/add_user_ips_last_seen_only_index.sql new file mode 100644 index 000000000..3a4ed59b5 --- /dev/null +++ b/synapse/storage/schema/delta/49/add_user_ips_last_seen_only_index.sql @@ -0,0 +1,17 @@ +/* Copyright 2018 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +INSERT into background_updates (update_name, progress_json) + VALUES ('user_ips_last_seen_only_index', '{}'); diff --git a/synapse/storage/schema/delta/50/add_creation_ts_users_index.sql b/synapse/storage/schema/delta/50/add_creation_ts_users_index.sql new file mode 100644 index 000000000..c93ae4753 --- /dev/null +++ b/synapse/storage/schema/delta/50/add_creation_ts_users_index.sql @@ -0,0 +1,19 @@ +/* Copyright 2018 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + + + +INSERT into background_updates (update_name, progress_json) + VALUES ('users_creation_ts', '{}'); diff --git a/synapse/storage/schema/delta/50/erasure_store.sql b/synapse/storage/schema/delta/50/erasure_store.sql new file mode 100644 index 000000000..5d8641a9a --- /dev/null +++ b/synapse/storage/schema/delta/50/erasure_store.sql @@ -0,0 +1,21 @@ +/* Copyright 2018 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- a table of users who have requested that their details be erased +CREATE TABLE erased_users ( + user_id TEXT NOT NULL +); + +CREATE UNIQUE INDEX erased_users_user ON erased_users(user_id); diff --git a/synapse/storage/schema/schema_version.sql b/synapse/storage/schema/schema_version.sql index a7ade6998..42e5cb6df 100644 --- a/synapse/storage/schema/schema_version.sql +++ b/synapse/storage/schema/schema_version.sql @@ -25,3 +25,10 @@ CREATE TABLE IF NOT EXISTS applied_schema_deltas( file TEXT NOT NULL, UNIQUE(version, file) ); + +-- a list of schema files we have loaded on behalf of dynamic modules +CREATE TABLE IF NOT EXISTS applied_module_schemas( + module_name TEXT NOT NULL, + file TEXT NOT NULL, + UNIQUE(module_name, file) +); diff --git a/synapse/storage/search.py b/synapse/storage/search.py index 8f2b3c443..d5b5df93e 100644 --- a/synapse/storage/search.py +++ b/synapse/storage/search.py @@ -13,28 +13,38 @@ # See the License for the specific language governing permissions and # limitations under the License. +import logging +import re +from collections import namedtuple + +from six import string_types + +from canonicaljson import json + from twisted.internet import defer -from .background_updates import BackgroundUpdateStore from synapse.api.errors import SynapseError from synapse.storage.engines import PostgresEngine, Sqlite3Engine -import logging -import re -import ujson as json - +from .background_updates import BackgroundUpdateStore logger = logging.getLogger(__name__) +SearchEntry = namedtuple('SearchEntry', [ + 'key', 'value', 'event_id', 'room_id', 'stream_ordering', + 'origin_server_ts', +]) + class SearchStore(BackgroundUpdateStore): EVENT_SEARCH_UPDATE_NAME = "event_search" EVENT_SEARCH_ORDER_UPDATE_NAME = "event_search_order" EVENT_SEARCH_USE_GIST_POSTGRES_NAME = "event_search_postgres_gist" + EVENT_SEARCH_USE_GIN_POSTGRES_NAME = "event_search_postgres_gin" - def __init__(self, hs): - super(SearchStore, self).__init__(hs) + def __init__(self, db_conn, hs): + super(SearchStore, self).__init__(db_conn, hs) self.register_background_update_handler( self.EVENT_SEARCH_UPDATE_NAME, self._background_reindex_search ) @@ -42,23 +52,35 @@ class SearchStore(BackgroundUpdateStore): self.EVENT_SEARCH_ORDER_UPDATE_NAME, self._background_reindex_search_order ) - self.register_background_update_handler( + + # we used to have a background update to turn the GIN index into a + # GIST one; we no longer do that (obviously) because we actually want + # a GIN index. However, it's possible that some people might still have + # the background update queued, so we register a handler to clear the + # background update. + self.register_noop_background_update( self.EVENT_SEARCH_USE_GIST_POSTGRES_NAME, - self._background_reindex_gist_search + ) + + self.register_background_update_handler( + self.EVENT_SEARCH_USE_GIN_POSTGRES_NAME, + self._background_reindex_gin_search ) @defer.inlineCallbacks def _background_reindex_search(self, progress, batch_size): + # we work through the events table from highest stream id to lowest target_min_stream_id = progress["target_min_stream_id_inclusive"] max_stream_id = progress["max_stream_id_exclusive"] rows_inserted = progress.get("rows_inserted", 0) - INSERT_CLUMP_SIZE = 1000 TYPES = ["m.room.name", "m.room.message", "m.room.topic"] def reindex_search_txn(txn): sql = ( - "SELECT stream_ordering, event_id, room_id, type, content FROM events" + "SELECT stream_ordering, event_id, room_id, type, json, " + " origin_server_ts FROM events" + " JOIN event_json USING (room_id, event_id)" " WHERE ? <= stream_ordering AND stream_ordering < ?" " AND (%s)" " ORDER BY stream_ordering DESC" @@ -67,6 +89,10 @@ class SearchStore(BackgroundUpdateStore): txn.execute(sql, (target_min_stream_id, max_stream_id, batch_size)) + # we could stream straight from the results into + # store_search_entries_txn with a generator function, but that + # would mean having two cursors open on the database at once. + # Instead we just build a list of results. rows = self.cursor_to_dict(txn) if not rows: return 0 @@ -79,9 +105,12 @@ class SearchStore(BackgroundUpdateStore): event_id = row["event_id"] room_id = row["room_id"] etype = row["type"] + stream_ordering = row["stream_ordering"] + origin_server_ts = row["origin_server_ts"] try: - content = json.loads(row["content"]) - except: + event_json = json.loads(row["json"]) + content = event_json["content"] + except Exception: continue if etype == "m.room.message": @@ -93,35 +122,28 @@ class SearchStore(BackgroundUpdateStore): elif etype == "m.room.name": key = "content.name" value = content["name"] + else: + raise Exception("unexpected event type %s" % etype) except (KeyError, AttributeError): # If the event is missing a necessary field then # skip over it. continue - if not isinstance(value, basestring): + if not isinstance(value, string_types): # If the event body, name or topic isn't a string # then skip over it continue - event_search_rows.append((event_id, room_id, key, value)) + event_search_rows.append(SearchEntry( + key=key, + value=value, + event_id=event_id, + room_id=room_id, + stream_ordering=stream_ordering, + origin_server_ts=origin_server_ts, + )) - if isinstance(self.database_engine, PostgresEngine): - sql = ( - "INSERT INTO event_search (event_id, room_id, key, vector)" - " VALUES (?,?,?,to_tsvector('english', ?))" - ) - elif isinstance(self.database_engine, Sqlite3Engine): - sql = ( - "INSERT INTO event_search (event_id, room_id, key, value)" - " VALUES (?,?,?,?)" - ) - else: - # This should be unreachable. - raise Exception("Unrecognized database engine") - - for index in range(0, len(event_search_rows), INSERT_CLUMP_SIZE): - clump = event_search_rows[index:index + INSERT_CLUMP_SIZE] - txn.executemany(sql, clump) + self.store_search_entries_txn(txn, event_search_rows) progress = { "target_min_stream_id_inclusive": target_min_stream_id, @@ -145,25 +167,48 @@ class SearchStore(BackgroundUpdateStore): defer.returnValue(result) @defer.inlineCallbacks - def _background_reindex_gist_search(self, progress, batch_size): + def _background_reindex_gin_search(self, progress, batch_size): + """This handles old synapses which used GIST indexes, if any; + converting them back to be GIN as per the actual schema. + """ + def create_index(conn): conn.rollback() + + # we have to set autocommit, because postgres refuses to + # CREATE INDEX CONCURRENTLY without it. conn.set_session(autocommit=True) - c = conn.cursor() - c.execute( - "CREATE INDEX CONCURRENTLY event_search_fts_idx_gist" - " ON event_search USING GIST (vector)" - ) + try: + c = conn.cursor() - c.execute("DROP INDEX event_search_fts_idx") + # if we skipped the conversion to GIST, we may already/still + # have an event_search_fts_idx; unfortunately postgres 9.4 + # doesn't support CREATE INDEX IF EXISTS so we just catch the + # exception and ignore it. + import psycopg2 + try: + c.execute( + "CREATE INDEX CONCURRENTLY event_search_fts_idx" + " ON event_search USING GIN (vector)" + ) + except psycopg2.ProgrammingError as e: + logger.warn( + "Ignoring error %r when trying to switch from GIST to GIN", + e + ) - conn.set_session(autocommit=False) + # we should now be able to delete the GIST index. + c.execute( + "DROP INDEX IF EXISTS event_search_fts_idx_gist" + ) + finally: + conn.set_session(autocommit=False) if isinstance(self.database_engine, PostgresEngine): yield self.runWithConnection(create_index) - yield self._end_background_update(self.EVENT_SEARCH_USE_GIST_POSTGRES_NAME) + yield self._end_background_update(self.EVENT_SEARCH_USE_GIN_POSTGRES_NAME) defer.returnValue(1) @defer.inlineCallbacks @@ -242,6 +287,85 @@ class SearchStore(BackgroundUpdateStore): defer.returnValue(num_rows) + def store_event_search_txn(self, txn, event, key, value): + """Add event to the search table + + Args: + txn (cursor): + event (EventBase): + key (str): + value (str): + """ + self.store_search_entries_txn( + txn, + (SearchEntry( + key=key, + value=value, + event_id=event.event_id, + room_id=event.room_id, + stream_ordering=event.internal_metadata.stream_ordering, + origin_server_ts=event.origin_server_ts, + ),), + ) + + def store_search_entries_txn(self, txn, entries): + """Add entries to the search table + + Args: + txn (cursor): + entries (iterable[SearchEntry]): + entries to be added to the table + """ + if isinstance(self.database_engine, PostgresEngine): + sql = ( + "INSERT INTO event_search" + " (event_id, room_id, key, vector, stream_ordering, origin_server_ts)" + " VALUES (?,?,?,to_tsvector('english', ?),?,?)" + ) + + args = (( + entry.event_id, entry.room_id, entry.key, entry.value, + entry.stream_ordering, entry.origin_server_ts, + ) for entry in entries) + + # inserts to a GIN index are normally batched up into a pending + # list, and then all committed together once the list gets to a + # certain size. The trouble with that is that postgres (pre-9.5) + # uses work_mem to determine the length of the list, and work_mem + # is typically very large. + # + # We therefore reduce work_mem while we do the insert. + # + # (postgres 9.5 uses the separate gin_pending_list_limit setting, + # so doesn't suffer the same problem, but changing work_mem will + # be harmless) + # + # Note that we don't need to worry about restoring it on + # exception, because exceptions will cause the transaction to be + # rolled back, including the effects of the SET command. + # + # Also: we use SET rather than SET LOCAL because there's lots of + # other stuff going on in this transaction, which want to have the + # normal work_mem setting. + + txn.execute("SET work_mem='256kB'") + txn.executemany(sql, args) + txn.execute("RESET work_mem") + + elif isinstance(self.database_engine, Sqlite3Engine): + sql = ( + "INSERT INTO event_search (event_id, room_id, key, value)" + " VALUES (?,?,?,?)" + ) + args = (( + entry.event_id, entry.room_id, entry.key, entry.value, + ) for entry in entries) + + txn.executemany(sql, args) + else: + # This should be unreachable. + raise Exception("Unrecognized database engine") + @defer.inlineCallbacks def search_msgs(self, room_ids, search_term, keys): """Performs a full text search over events with given keys. @@ -326,7 +450,7 @@ class SearchStore(BackgroundUpdateStore): "search_msgs", self.cursor_to_dict, sql, *args ) - results = filter(lambda row: row["room_id"] in room_ids, results) + results = list(filter(lambda row: row["room_id"] in room_ids, results)) events = yield self._get_events([r["event_id"] for r in results]) @@ -407,7 +531,7 @@ class SearchStore(BackgroundUpdateStore): origin_server_ts, stream = pagination_token.split(",") origin_server_ts = int(origin_server_ts) stream = int(stream) - except: + except Exception: raise SynapseError(400, "Invalid pagination token") clauses.append( @@ -481,7 +605,7 @@ class SearchStore(BackgroundUpdateStore): "search_rooms", self.cursor_to_dict, sql, *args ) - results = filter(lambda row: row["room_id"] in room_ids, results) + results = list(filter(lambda row: row["room_id"] in room_ids, results)) events = yield self._get_events([r["event_id"] for r in results]) diff --git a/synapse/storage/signatures.py b/synapse/storage/signatures.py index 67d5d9969..470212aa2 100644 --- a/synapse/storage/signatures.py +++ b/synapse/storage/signatures.py @@ -13,21 +13,31 @@ # See the License for the specific language governing permissions and # limitations under the License. -from twisted.internet import defer - -from ._base import SQLBaseStore +import six from unpaddedbase64 import encode_base64 + +from twisted.internet import defer + from synapse.crypto.event_signing import compute_event_reference_hash from synapse.util.caches.descriptors import cached, cachedList +from ._base import SQLBaseStore -class SignatureStore(SQLBaseStore): - """Persistence for event signatures and hashes""" +# py2 sqlite has buffer hardcoded as only binary type, so we must use it, +# despite being deprecated and removed in favor of memoryview +if six.PY2: + db_binary_type = buffer +else: + db_binary_type = memoryview + +class SignatureWorkerStore(SQLBaseStore): @cached() def get_event_reference_hash(self, event_id): - return self._get_event_reference_hashes_txn(event_id) + # This is a dummy function to allow get_event_reference_hashes + # to use its cache + raise NotImplementedError() @cachedList(cached_method_name="get_event_reference_hash", list_name="event_ids", num_args=1) @@ -56,7 +66,7 @@ class SignatureStore(SQLBaseStore): for e_id, h in hashes.items() } - defer.returnValue(hashes.items()) + defer.returnValue(list(hashes.items())) def _get_event_reference_hashes_txn(self, txn, event_id): """Get all the hashes for a given PDU. @@ -74,6 +84,10 @@ class SignatureStore(SQLBaseStore): txn.execute(query, (event_id, )) return {k: v for k, v in txn} + +class SignatureStore(SignatureWorkerStore): + """Persistence for event signatures and hashes""" + def _store_event_reference_hashes_txn(self, txn, events): """Store a hash for a PDU Args: @@ -87,7 +101,7 @@ class SignatureStore(SQLBaseStore): vals.append({ "event_id": event.event_id, "algorithm": ref_alg, - "hash": buffer(ref_hash_bytes), + "hash": db_binary_type(ref_hash_bytes), }) self._simple_insert_many_txn( diff --git a/synapse/storage/state.py b/synapse/storage/state.py index 5673e4aa9..89a05c461 100644 --- a/synapse/storage/state.py +++ b/synapse/storage/state.py @@ -13,16 +13,22 @@ # See the License for the specific language governing permissions and # limitations under the License. -from ._base import SQLBaseStore -from synapse.util.caches.descriptors import cached, cachedList -from synapse.util.caches import intern_string -from synapse.util.stringutils import to_ascii -from synapse.storage.engines import PostgresEngine - -from twisted.internet import defer +import logging from collections import namedtuple -import logging +from six import iteritems, itervalues +from six.moves import range + +from twisted.internet import defer + +from synapse.storage.background_updates import BackgroundUpdateStore +from synapse.storage.engines import PostgresEngine +from synapse.util.caches import get_cache_factor_for, intern_string +from synapse.util.caches.descriptors import cached, cachedList +from synapse.util.caches.dictionary_cache import DictionaryCache +from synapse.util.stringutils import to_ascii + +from ._base import SQLBaseStore logger = logging.getLogger(__name__) @@ -40,45 +46,19 @@ class _GetStateGroupDelta(namedtuple("_GetStateGroupDelta", ("prev_group", "delt return len(self.delta_ids) if self.delta_ids else 0 -class StateStore(SQLBaseStore): - """ Keeps track of the state at a given event. - - This is done by the concept of `state groups`. Every event is a assigned - a state group (identified by an arbitrary string), which references a - collection of state events. The current state of an event is then the - collection of state events referenced by the event's state group. - - Hence, every change in the current state causes a new state group to be - generated. However, if no change happens (e.g., if we get a message event - with only one parent it inherits the state group from its parent.) - - There are three tables: - * `state_groups`: Stores group name, first event with in the group and - room id. - * `event_to_state_groups`: Maps events to state groups. - * `state_groups_state`: Maps state group to state events. +class StateGroupWorkerStore(SQLBaseStore): + """The parts of StateGroupStore that can be called from workers. """ STATE_GROUP_DEDUPLICATION_UPDATE_NAME = "state_group_state_deduplication" STATE_GROUP_INDEX_UPDATE_NAME = "state_group_state_type_index" CURRENT_STATE_INDEX_UPDATE_NAME = "current_state_members_idx" - def __init__(self, hs): - super(StateStore, self).__init__(hs) - self.register_background_update_handler( - self.STATE_GROUP_DEDUPLICATION_UPDATE_NAME, - self._background_deduplicate_state, - ) - self.register_background_update_handler( - self.STATE_GROUP_INDEX_UPDATE_NAME, - self._background_index_state, - ) - self.register_background_index_update( - self.CURRENT_STATE_INDEX_UPDATE_NAME, - index_name="current_state_events_member_index", - table="current_state_events", - columns=["state_key"], - where_clause="type='m.room.member'", + def __init__(self, db_conn, hs): + super(StateGroupWorkerStore, self).__init__(db_conn, hs) + + self._state_group_cache = DictionaryCache( + "*stateGroupCache*", 500000 * get_cache_factor_for("stateGroupCache") ) @cached(max_entries=100000, iterable=True) @@ -158,11 +138,25 @@ class StateStore(SQLBaseStore): event_ids, ) - groups = set(event_to_groups.itervalues()) + groups = set(itervalues(event_to_groups)) group_to_state = yield self._get_state_for_groups(groups) defer.returnValue(group_to_state) + @defer.inlineCallbacks + def get_state_ids_for_group(self, state_group): + """Get the state IDs for the given state group + + Args: + state_group (int) + + Returns: + Deferred[dict]: Resolves to a map of (type, state_key) -> event_id + """ + group_to_state = yield self._get_state_for_groups((state_group,)) + + defer.returnValue(group_to_state[state_group]) + @defer.inlineCallbacks def get_state_groups(self, room_id, event_ids): """ Get the state groups for the given list of event_ids @@ -176,199 +170,27 @@ class StateStore(SQLBaseStore): state_event_map = yield self.get_events( [ - ev_id for group_ids in group_to_ids.itervalues() - for ev_id in group_ids.itervalues() + ev_id for group_ids in itervalues(group_to_ids) + for ev_id in itervalues(group_ids) ], get_prev_content=False ) defer.returnValue({ group: [ - state_event_map[v] for v in event_id_map.itervalues() + state_event_map[v] for v in itervalues(event_id_map) if v in state_event_map ] - for group, event_id_map in group_to_ids.iteritems() + for group, event_id_map in iteritems(group_to_ids) }) - def _have_persisted_state_group_txn(self, txn, state_group): - txn.execute( - "SELECT count(*) FROM state_groups WHERE id = ?", - (state_group,) - ) - row = txn.fetchone() - return row and row[0] - - def _store_mult_state_groups_txn(self, txn, events_and_contexts): - state_groups = {} - for event, context in events_and_contexts: - if event.internal_metadata.is_outlier(): - continue - - if context.current_state_ids is None: - # AFAIK, this can never happen - logger.error( - "Non-outlier event %s had current_state_ids==None", - event.event_id) - continue - - # if the event was rejected, just give it the same state as its - # predecessor. - if context.rejected: - state_groups[event.event_id] = context.prev_group - continue - - state_groups[event.event_id] = context.state_group - - if self._have_persisted_state_group_txn(txn, context.state_group): - continue - - self._simple_insert_txn( - txn, - table="state_groups", - values={ - "id": context.state_group, - "room_id": event.room_id, - "event_id": event.event_id, - }, - ) - - # We persist as a delta if we can, while also ensuring the chain - # of deltas isn't tooo long, as otherwise read performance degrades. - if context.prev_group: - is_in_db = self._simple_select_one_onecol_txn( - txn, - table="state_groups", - keyvalues={"id": context.prev_group}, - retcol="id", - allow_none=True, - ) - if not is_in_db: - raise Exception( - "Trying to persist state with unpersisted prev_group: %r" - % (context.prev_group,) - ) - - potential_hops = self._count_state_group_hops_txn( - txn, context.prev_group - ) - if context.prev_group and potential_hops < MAX_STATE_DELTA_HOPS: - self._simple_insert_txn( - txn, - table="state_group_edges", - values={ - "state_group": context.state_group, - "prev_state_group": context.prev_group, - }, - ) - - self._simple_insert_many_txn( - txn, - table="state_groups_state", - values=[ - { - "state_group": context.state_group, - "room_id": event.room_id, - "type": key[0], - "state_key": key[1], - "event_id": state_id, - } - for key, state_id in context.delta_ids.iteritems() - ], - ) - else: - self._simple_insert_many_txn( - txn, - table="state_groups_state", - values=[ - { - "state_group": context.state_group, - "room_id": event.room_id, - "type": key[0], - "state_key": key[1], - "event_id": state_id, - } - for key, state_id in context.current_state_ids.iteritems() - ], - ) - - # Prefill the state group cache with this group. - # It's fine to use the sequence like this as the state group map - # is immutable. (If the map wasn't immutable then this prefill could - # race with another update) - txn.call_after( - self._state_group_cache.update, - self._state_group_cache.sequence, - key=context.state_group, - value=dict(context.current_state_ids), - full=True, - ) - - self._simple_insert_many_txn( - txn, - table="event_to_state_groups", - values=[ - { - "state_group": state_group_id, - "event_id": event_id, - } - for event_id, state_group_id in state_groups.iteritems() - ], - ) - - for event_id, state_group_id in state_groups.iteritems(): - txn.call_after( - self._get_state_group_for_event.prefill, - (event_id,), state_group_id - ) - - def _count_state_group_hops_txn(self, txn, state_group): - """Given a state group, count how many hops there are in the tree. - - This is used to ensure the delta chains don't get too long. - """ - if isinstance(self.database_engine, PostgresEngine): - sql = (""" - WITH RECURSIVE state(state_group) AS ( - VALUES(?::bigint) - UNION ALL - SELECT prev_state_group FROM state_group_edges e, state s - WHERE s.state_group = e.state_group - ) - SELECT count(*) FROM state; - """) - - txn.execute(sql, (state_group,)) - row = txn.fetchone() - if row and row[0]: - return row[0] - else: - return 0 - else: - # We don't use WITH RECURSIVE on sqlite3 as there are distributions - # that ship with an sqlite3 version that doesn't support it (e.g. wheezy) - next_group = state_group - count = 0 - - while next_group: - next_group = self._simple_select_one_onecol_txn( - txn, - table="state_group_edges", - keyvalues={"state_group": next_group}, - retcol="prev_state_group", - allow_none=True, - ) - if next_group: - count += 1 - - return count - @defer.inlineCallbacks def _get_state_groups_from_groups(self, groups, types): """Returns dictionary state_group -> (dict of (type, state_key) -> event id) """ results = {} - chunks = [groups[i:i + 100] for i in xrange(0, len(groups), 100)] + chunks = [groups[i:i + 100] for i in range(0, len(groups), 100)] for chunk in chunks: res = yield self.runInteraction( "_get_state_groups_from_groups", @@ -422,6 +244,9 @@ class StateStore(SQLBaseStore): ( "AND type = ? AND state_key = ?", (etype, state_key) + ) if state_key is not None else ( + "AND type = ?", + (etype,) ) for etype, state_key in types ] @@ -441,10 +266,19 @@ class StateStore(SQLBaseStore): key = (typ, state_key) results[group][key] = event_id else: + where_args = [] + where_clauses = [] + wildcard_types = False if types is not None: - where_clause = "AND (%s)" % ( - " OR ".join(["(type = ? AND state_key = ?)"] * len(types)), - ) + for typ in types: + if typ[1] is None: + where_clauses.append("(type = ?)") + where_args.append(typ[0]) + wildcard_types = True + else: + where_clauses.append("(type = ? AND state_key = ?)") + where_args.extend([typ[0], typ[1]]) + where_clause = "AND (%s)" % (" OR ".join(where_clauses)) else: where_clause = "" @@ -461,7 +295,7 @@ class StateStore(SQLBaseStore): # after we finish deduping state, which requires this func) args = [next_group] if types: - args.extend(i for typ in types for i in typ) + args.extend(where_args) txn.execute( "SELECT type, state_key, event_id FROM state_groups_state" @@ -474,9 +308,17 @@ class StateStore(SQLBaseStore): if (typ, state_key) not in results[group] ) - # If the lengths match then we must have all the types, - # so no need to go walk further down the tree. - if types is not None and len(results[group]) == len(types): + # If the number of entries in the (type,state_key)->event_id dict + # matches the number of (type,state_keys) types we were searching + # for, then we must have found them all, so no need to go walk + # further down the tree... UNLESS our types filter contained + # wildcards (i.e. Nones) in which case we have to do an exhaustive + # search + if ( + types is not None and + not wildcard_types and + len(results[group]) == len(types) + ): break next_group = self._simple_select_one_onecol_txn( @@ -509,21 +351,21 @@ class StateStore(SQLBaseStore): event_ids, ) - groups = set(event_to_groups.itervalues()) + groups = set(itervalues(event_to_groups)) group_to_state = yield self._get_state_for_groups(groups, types) state_event_map = yield self.get_events( - [ev_id for sd in group_to_state.itervalues() for ev_id in sd.itervalues()], + [ev_id for sd in itervalues(group_to_state) for ev_id in itervalues(sd)], get_prev_content=False ) event_to_state = { event_id: { k: state_event_map[v] - for k, v in group_to_state[group].iteritems() + for k, v in iteritems(group_to_state[group]) if v in state_event_map } - for event_id, group in event_to_groups.iteritems() + for event_id, group in iteritems(event_to_groups) } defer.returnValue({event: event_to_state[event] for event in event_ids}) @@ -546,12 +388,12 @@ class StateStore(SQLBaseStore): event_ids, ) - groups = set(event_to_groups.itervalues()) + groups = set(itervalues(event_to_groups)) group_to_state = yield self._get_state_for_groups(groups, types) event_to_state = { event_id: group_to_state[group] - for event_id, group in event_to_groups.iteritems() + for event_id, group in iteritems(event_to_groups) } defer.returnValue({event: event_to_state[event] for event in event_ids}) @@ -665,7 +507,7 @@ class StateStore(SQLBaseStore): got_all = is_all or not missing_types return { - k: v for k, v in state_dict_ids.iteritems() + k: v for k, v in iteritems(state_dict_ids) if include(k[0], k[1]) }, missing_types, got_all @@ -685,10 +527,23 @@ class StateStore(SQLBaseStore): @defer.inlineCallbacks def _get_state_for_groups(self, groups, types=None): - """Given list of groups returns dict of group -> list of state events - with matching types. `types` is a list of `(type, state_key)`, where - a `state_key` of None matches all state_keys. If `types` is None then - all events are returned. + """Gets the state at each of a list of state groups, optionally + filtering by type/state_key + + Args: + groups (iterable[int]): list of state groups for which we want + to get the state. + types (None|iterable[(str, None|str)]): + indicates the state type/keys required. If None, the whole + state is fetched and returned. + + Otherwise, each entry should be a `(type, state_key)` tuple to + include in the response. A `state_key` of None is a wildcard + meaning that we require all state with that type. + + Returns: + Deferred[dict[int, dict[(type, state_key), EventBase]]] + a dictionary mapping from state group to state dictionary. """ if types: types = frozenset(types) @@ -697,7 +552,7 @@ class StateStore(SQLBaseStore): if types is not None: for group in set(groups): state_dict_ids, _, got_all = self._get_some_state_from_cache( - group, types + group, types, ) results[group] = state_dict_ids @@ -718,32 +573,266 @@ class StateStore(SQLBaseStore): # Okay, so we have some missing_types, lets fetch them. cache_seq_num = self._state_group_cache.sequence + # the DictionaryCache knows if it has *all* the state, but + # does not know if it has all of the keys of a particular type, + # which makes wildcard lookups expensive unless we have a complete + # cache. Hence, if we are doing a wildcard lookup, populate the + # cache fully so that we can do an efficient lookup next time. + + if types and any(k is None for (t, k) in types): + types_to_fetch = None + else: + types_to_fetch = types + group_to_state_dict = yield self._get_state_groups_from_groups( - missing_groups, types + missing_groups, types_to_fetch, ) - # Now we want to update the cache with all the things we fetched - # from the database. - for group, group_state_dict in group_to_state_dict.iteritems(): + for group, group_state_dict in iteritems(group_to_state_dict): state_dict = results[group] - state_dict.update( - ((intern_string(k[0]), intern_string(k[1])), to_ascii(v)) - for k, v in group_state_dict.iteritems() - ) + # update the result, filtering by `types`. + if types: + for k, v in iteritems(group_state_dict): + (typ, _) = k + if k in types or (typ, None) in types: + state_dict[k] = v + else: + state_dict.update(group_state_dict) + # update the cache with all the things we fetched from the + # database. self._state_group_cache.update( cache_seq_num, key=group, - value=state_dict, - full=(types is None), - known_absent=types, + value=group_state_dict, + fetched_keys=types_to_fetch, ) defer.returnValue(results) - def get_next_state_group(self): - return self._state_groups_id_gen.get_next() + def store_state_group(self, event_id, room_id, prev_group, delta_ids, + current_state_ids): + """Store a new set of state, returning a newly assigned state group. + + Args: + event_id (str): The event ID for which the state was calculated + room_id (str) + prev_group (int|None): A previous state group for the room, optional. + delta_ids (dict|None): The delta between state at `prev_group` and + `current_state_ids`, if `prev_group` was given. Same format as + `current_state_ids`. + current_state_ids (dict): The state to store. Map of (type, state_key) + to event_id. + + Returns: + Deferred[int]: The state group ID + """ + def _store_state_group_txn(txn): + if current_state_ids is None: + # AFAIK, this can never happen + raise Exception("current_state_ids cannot be None") + + state_group = self.database_engine.get_next_state_group_id(txn) + + self._simple_insert_txn( + txn, + table="state_groups", + values={ + "id": state_group, + "room_id": room_id, + "event_id": event_id, + }, + ) + + # We persist as a delta if we can, while also ensuring the chain + # of deltas isn't tooo long, as otherwise read performance degrades. + if prev_group: + is_in_db = self._simple_select_one_onecol_txn( + txn, + table="state_groups", + keyvalues={"id": prev_group}, + retcol="id", + allow_none=True, + ) + if not is_in_db: + raise Exception( + "Trying to persist state with unpersisted prev_group: %r" + % (prev_group,) + ) + + potential_hops = self._count_state_group_hops_txn( + txn, prev_group + ) + if prev_group and potential_hops < MAX_STATE_DELTA_HOPS: + self._simple_insert_txn( + txn, + table="state_group_edges", + values={ + "state_group": state_group, + "prev_state_group": prev_group, + }, + ) + + self._simple_insert_many_txn( + txn, + table="state_groups_state", + values=[ + { + "state_group": state_group, + "room_id": room_id, + "type": key[0], + "state_key": key[1], + "event_id": state_id, + } + for key, state_id in iteritems(delta_ids) + ], + ) + else: + self._simple_insert_many_txn( + txn, + table="state_groups_state", + values=[ + { + "state_group": state_group, + "room_id": room_id, + "type": key[0], + "state_key": key[1], + "event_id": state_id, + } + for key, state_id in iteritems(current_state_ids) + ], + ) + + # Prefill the state group cache with this group. + # It's fine to use the sequence like this as the state group map + # is immutable. (If the map wasn't immutable then this prefill could + # race with another update) + txn.call_after( + self._state_group_cache.update, + self._state_group_cache.sequence, + key=state_group, + value=dict(current_state_ids), + ) + + return state_group + + return self.runInteraction("store_state_group", _store_state_group_txn) + + def _count_state_group_hops_txn(self, txn, state_group): + """Given a state group, count how many hops there are in the tree. + + This is used to ensure the delta chains don't get too long. + """ + if isinstance(self.database_engine, PostgresEngine): + sql = (""" + WITH RECURSIVE state(state_group) AS ( + VALUES(?::bigint) + UNION ALL + SELECT prev_state_group FROM state_group_edges e, state s + WHERE s.state_group = e.state_group + ) + SELECT count(*) FROM state; + """) + + txn.execute(sql, (state_group,)) + row = txn.fetchone() + if row and row[0]: + return row[0] + else: + return 0 + else: + # We don't use WITH RECURSIVE on sqlite3 as there are distributions + # that ship with an sqlite3 version that doesn't support it (e.g. wheezy) + next_group = state_group + count = 0 + + while next_group: + next_group = self._simple_select_one_onecol_txn( + txn, + table="state_group_edges", + keyvalues={"state_group": next_group}, + retcol="prev_state_group", + allow_none=True, + ) + if next_group: + count += 1 + + return count + + +class StateStore(StateGroupWorkerStore, BackgroundUpdateStore): + """ Keeps track of the state at a given event. + + This is done by the concept of `state groups`. Every event is a assigned + a state group (identified by an arbitrary string), which references a + collection of state events. The current state of an event is then the + collection of state events referenced by the event's state group. + + Hence, every change in the current state causes a new state group to be + generated. However, if no change happens (e.g., if we get a message event + with only one parent it inherits the state group from its parent.) + + There are three tables: + * `state_groups`: Stores group name, first event with in the group and + room id. + * `event_to_state_groups`: Maps events to state groups. + * `state_groups_state`: Maps state group to state events. + """ + + STATE_GROUP_DEDUPLICATION_UPDATE_NAME = "state_group_state_deduplication" + STATE_GROUP_INDEX_UPDATE_NAME = "state_group_state_type_index" + CURRENT_STATE_INDEX_UPDATE_NAME = "current_state_members_idx" + + def __init__(self, db_conn, hs): + super(StateStore, self).__init__(db_conn, hs) + self.register_background_update_handler( + self.STATE_GROUP_DEDUPLICATION_UPDATE_NAME, + self._background_deduplicate_state, + ) + self.register_background_update_handler( + self.STATE_GROUP_INDEX_UPDATE_NAME, + self._background_index_state, + ) + self.register_background_index_update( + self.CURRENT_STATE_INDEX_UPDATE_NAME, + index_name="current_state_events_member_index", + table="current_state_events", + columns=["state_key"], + where_clause="type='m.room.member'", + ) + + def _store_event_state_mappings_txn(self, txn, events_and_contexts): + state_groups = {} + for event, context in events_and_contexts: + if event.internal_metadata.is_outlier(): + continue + + # if the event was rejected, just give it the same state as its + # predecessor. + if context.rejected: + state_groups[event.event_id] = context.prev_group + continue + + state_groups[event.event_id] = context.state_group + + self._simple_insert_many_txn( + txn, + table="event_to_state_groups", + values=[ + { + "state_group": state_group_id, + "event_id": event_id, + } + for event_id, state_group_id in iteritems(state_groups) + ], + ) + + for event_id, state_group_id in iteritems(state_groups): + txn.call_after( + self._get_state_group_for_event.prefill, + (event_id,), state_group_id + ) @defer.inlineCallbacks def _background_deduplicate_state(self, progress, batch_size): @@ -767,7 +856,7 @@ class StateStore(SQLBaseStore): def reindex_txn(txn): new_last_state_group = last_state_group - for count in xrange(batch_size): + for count in range(batch_size): txn.execute( "SELECT id, room_id FROM state_groups" " WHERE ? < id AND id <= ?" @@ -825,7 +914,7 @@ class StateStore(SQLBaseStore): # of keys delta_state = { - key: value for key, value in curr_state.iteritems() + key: value for key, value in iteritems(curr_state) if prev_state.get(key, None) != value } @@ -865,7 +954,7 @@ class StateStore(SQLBaseStore): "state_key": key[1], "event_id": state_id, } - for key, state_id in delta_state.iteritems() + for key, state_id in iteritems(delta_state) ], ) diff --git a/synapse/storage/stream.py b/synapse/storage/stream.py index dddd5fc0e..66856342f 100644 --- a/synapse/storage/stream.py +++ b/synapse/storage/stream.py @@ -33,17 +33,20 @@ what sort order was used: and stream ordering columns respectively. """ +import abc +import logging +from collections import namedtuple + +from six.moves import range + from twisted.internet import defer -from ._base import SQLBaseStore -from synapse.util.caches.descriptors import cached -from synapse.api.constants import EventTypes +from synapse.storage._base import SQLBaseStore +from synapse.storage.engines import PostgresEngine +from synapse.storage.events import EventsWorkerStore from synapse.types import RoomStreamToken -from synapse.util.logcontext import preserve_fn, preserve_context_over_deferred -from synapse.storage.engines import PostgresEngine, Sqlite3Engine - -import logging - +from synapse.util.caches.stream_change_cache import StreamChangeCache +from synapse.util.logcontext import make_deferred_yieldable, run_in_background logger = logging.getLogger(__name__) @@ -55,6 +58,12 @@ _STREAM_TOKEN = "stream" _TOPOLOGICAL_TOKEN = "topological" +# Used as return values for pagination APIs +_EventDictReturn = namedtuple("_EventDictReturn", ( + "event_id", "topological_ordering", "stream_ordering", +)) + + def lower_bound(token, engine, inclusive=False): inclusive = "=" if inclusive else "" if token.topological is None: @@ -143,81 +152,41 @@ def filter_to_clause(event_filter): return " AND ".join(clauses), args -class StreamStore(SQLBaseStore): - @defer.inlineCallbacks - def get_appservice_room_stream(self, service, from_key, to_key, limit=0): - # NB this lives here instead of appservice.py so we can reuse the - # 'private' StreamToken class in this file. - if limit: - limit = max(limit, MAX_STREAM_SIZE) - else: - limit = MAX_STREAM_SIZE +class StreamWorkerStore(EventsWorkerStore, SQLBaseStore): + """This is an abstract base class where subclasses must implement + `get_room_max_stream_ordering` and `get_room_min_stream_ordering` + which can be called in the initializer. + """ - # From and to keys should be integers from ordering. - from_id = RoomStreamToken.parse_stream_token(from_key) - to_id = RoomStreamToken.parse_stream_token(to_key) + __metaclass__ = abc.ABCMeta - if from_key == to_key: - defer.returnValue(([], to_key)) - return + def __init__(self, db_conn, hs): + super(StreamWorkerStore, self).__init__(db_conn, hs) - # select all the events between from/to with a sensible limit - sql = ( - "SELECT e.event_id, e.room_id, e.type, s.state_key, " - "e.stream_ordering FROM events AS e " - "LEFT JOIN state_events as s ON " - "e.event_id = s.event_id " - "WHERE e.stream_ordering > ? AND e.stream_ordering <= ? " - "ORDER BY stream_ordering ASC LIMIT %(limit)d " - ) % { - "limit": limit - } - - def f(txn): - # pull out all the events between the tokens - txn.execute(sql, (from_id.stream, to_id.stream,)) - rows = self.cursor_to_dict(txn) - - # Logic: - # - We want ALL events which match the AS room_id regex - # - We want ALL events which match the rooms represented by the AS - # room_alias regex - # - We want ALL events for rooms that AS users have joined. - # This is currently supported via get_app_service_rooms (which is - # used for the Notifier listener rooms). We can't reasonably make a - # SQL query for these room IDs, so we'll pull all the events between - # from/to and filter in python. - rooms_for_as = self._get_app_service_rooms_txn(txn, service) - room_ids_for_as = [r.room_id for r in rooms_for_as] - - def app_service_interested(row): - if row["room_id"] in room_ids_for_as: - return True - - if row["type"] == EventTypes.Member: - if service.is_interested_in_user(row.get("state_key")): - return True - return False - - return [r for r in rows if app_service_interested(r)] - - rows = yield self.runInteraction("get_appservice_room_stream", f) - - ret = yield self._get_events( - [r["event_id"] for r in rows], - get_prev_content=True + events_max = self.get_room_max_stream_ordering() + event_cache_prefill, min_event_val = self._get_cache_dict( + db_conn, "events", + entity_column="room_id", + stream_column="stream_ordering", + max_value=events_max, + ) + self._events_stream_cache = StreamChangeCache( + "EventsRoomStreamChangeCache", min_event_val, + prefilled_cache=event_cache_prefill, + ) + self._membership_stream_cache = StreamChangeCache( + "MembershipStreamChangeCache", events_max, ) - self._set_before_and_after(ret, rows, topo_order=from_id is None) + self._stream_order_on_start = self.get_room_max_stream_ordering() - if rows: - key = "s%d" % max(r["stream_ordering"] for r in rows) - else: - # Assume we didn't get anything because there was nothing to - # get. - key = to_key + @abc.abstractmethod + def get_room_max_stream_ordering(self): + raise NotImplementedError() - defer.returnValue((ret, key)) + @abc.abstractmethod + def get_room_min_stream_ordering(self): + raise NotImplementedError() @defer.inlineCallbacks def get_room_events_stream_for_rooms(self, room_ids, from_key, to_key, limit=0, @@ -233,13 +202,14 @@ class StreamStore(SQLBaseStore): results = {} room_ids = list(room_ids) - for rm_ids in (room_ids[i:i + 20] for i in xrange(0, len(room_ids), 20)): - res = yield preserve_context_over_deferred(defer.gatherResults([ - preserve_fn(self.get_room_events_stream_for_room)( + for rm_ids in (room_ids[i:i + 20] for i in range(0, len(room_ids), 20)): + res = yield make_deferred_yieldable(defer.gatherResults([ + run_in_background( + self.get_room_events_stream_for_room, room_id, from_key, to_key, limit, order=order, ) for room_id in rm_ids - ])) + ], consumeErrors=True)) results.update(dict(zip(rm_ids, res))) defer.returnValue(results) @@ -261,54 +231,55 @@ class StreamStore(SQLBaseStore): @defer.inlineCallbacks def get_room_events_stream_for_room(self, room_id, from_key, to_key, limit=0, order='DESC'): - # Note: If from_key is None then we return in topological order. This - # is because in that case we're using this as a "get the last few messages - # in a room" function, rather than "get new messages since last sync" - if from_key is not None: - from_id = RoomStreamToken.parse_stream_token(from_key).stream - else: - from_id = None - to_id = RoomStreamToken.parse_stream_token(to_key).stream + """Get new room events in stream ordering since `from_key`. + + Args: + room_id (str) + from_key (str): Token from which no events are returned before + to_key (str): Token from which no events are returned after. (This + is typically the current stream token) + limit (int): Maximum number of events to return + order (str): Either "DESC" or "ASC". Determines which events are + returned when the result is limited. If "DESC" then the most + recent `limit` events are returned, otherwise returns the + oldest `limit` events. + + Returns: + Deferred[tuple[list[FrozenEvent], str]]: Returns the list of + events (in ascending order) and the token from the start of + the chunk of events returned. + """ if from_key == to_key: defer.returnValue(([], from_key)) - if from_id: - has_changed = yield self._events_stream_cache.has_entity_changed( - room_id, from_id - ) + from_id = RoomStreamToken.parse_stream_token(from_key).stream + to_id = RoomStreamToken.parse_stream_token(to_key).stream - if not has_changed: - defer.returnValue(([], from_key)) + has_changed = yield self._events_stream_cache.has_entity_changed( + room_id, from_id + ) + + if not has_changed: + defer.returnValue(([], from_key)) def f(txn): - if from_id is not None: - sql = ( - "SELECT event_id, stream_ordering FROM events WHERE" - " room_id = ?" - " AND not outlier" - " AND stream_ordering > ? AND stream_ordering <= ?" - " ORDER BY stream_ordering %s LIMIT ?" - ) % (order,) - txn.execute(sql, (room_id, from_id, to_id, limit)) - else: - sql = ( - "SELECT event_id, stream_ordering FROM events WHERE" - " room_id = ?" - " AND not outlier" - " AND stream_ordering <= ?" - " ORDER BY topological_ordering %s, stream_ordering %s LIMIT ?" - ) % (order, order,) - txn.execute(sql, (room_id, to_id, limit)) - - rows = self.cursor_to_dict(txn) + sql = ( + "SELECT event_id, stream_ordering FROM events WHERE" + " room_id = ?" + " AND not outlier" + " AND stream_ordering > ? AND stream_ordering <= ?" + " ORDER BY stream_ordering %s LIMIT ?" + ) % (order,) + txn.execute(sql, (room_id, from_id, to_id, limit)) + rows = [_EventDictReturn(row[0], None, row[1]) for row in txn] return rows rows = yield self.runInteraction("get_room_events_stream_for_room", f) ret = yield self._get_events( - [r["event_id"] for r in rows], + [r.event_id for r in rows], get_prev_content=True ) @@ -318,7 +289,7 @@ class StreamStore(SQLBaseStore): ret.reverse() if rows: - key = "s%d" % min(r["stream_ordering"] for r in rows) + key = "s%d" % min(r.stream_ordering for r in rows) else: # Assume we didn't get anything because there was nothing to # get. @@ -328,10 +299,7 @@ class StreamStore(SQLBaseStore): @defer.inlineCallbacks def get_membership_changes_for_user(self, user_id, from_key, to_key): - if from_key is not None: - from_id = RoomStreamToken.parse_stream_token(from_key).stream - else: - from_id = None + from_id = RoomStreamToken.parse_stream_token(from_key).stream to_id = RoomStreamToken.parse_stream_token(to_key).stream if from_key == to_key: @@ -345,34 +313,24 @@ class StreamStore(SQLBaseStore): defer.returnValue([]) def f(txn): - if from_id is not None: - sql = ( - "SELECT m.event_id, stream_ordering FROM events AS e," - " room_memberships AS m" - " WHERE e.event_id = m.event_id" - " AND m.user_id = ?" - " AND e.stream_ordering > ? AND e.stream_ordering <= ?" - " ORDER BY e.stream_ordering ASC" - ) - txn.execute(sql, (user_id, from_id, to_id,)) - else: - sql = ( - "SELECT m.event_id, stream_ordering FROM events AS e," - " room_memberships AS m" - " WHERE e.event_id = m.event_id" - " AND m.user_id = ?" - " AND stream_ordering <= ?" - " ORDER BY stream_ordering ASC" - ) - txn.execute(sql, (user_id, to_id,)) - rows = self.cursor_to_dict(txn) + sql = ( + "SELECT m.event_id, stream_ordering FROM events AS e," + " room_memberships AS m" + " WHERE e.event_id = m.event_id" + " AND m.user_id = ?" + " AND e.stream_ordering > ? AND e.stream_ordering <= ?" + " ORDER BY e.stream_ordering ASC" + ) + txn.execute(sql, (user_id, from_id, to_id,)) + + rows = [_EventDictReturn(row[0], None, row[1]) for row in txn] return rows rows = yield self.runInteraction("get_membership_changes_for_user", f) ret = yield self._get_events( - [r["event_id"] for r in rows], + [r.event_id for r in rows], get_prev_content=True ) @@ -381,96 +339,28 @@ class StreamStore(SQLBaseStore): defer.returnValue(ret) @defer.inlineCallbacks - def paginate_room_events(self, room_id, from_key, to_key=None, - direction='b', limit=-1, event_filter=None): - # Tokens really represent positions between elements, but we use - # the convention of pointing to the event before the gap. Hence - # we have a bit of asymmetry when it comes to equalities. - args = [False, room_id] - if direction == 'b': - order = "DESC" - bounds = upper_bound( - RoomStreamToken.parse(from_key), self.database_engine - ) - if to_key: - bounds = "%s AND %s" % (bounds, lower_bound( - RoomStreamToken.parse(to_key), self.database_engine - )) - else: - order = "ASC" - bounds = lower_bound( - RoomStreamToken.parse(from_key), self.database_engine - ) - if to_key: - bounds = "%s AND %s" % (bounds, upper_bound( - RoomStreamToken.parse(to_key), self.database_engine - )) + def get_recent_events_for_room(self, room_id, limit, end_token): + """Get the most recent events in the room in topological ordering. - filter_clause, filter_args = filter_to_clause(event_filter) + Args: + room_id (str) + limit (int) + end_token (str): The stream token representing now. - if filter_clause: - bounds += " AND " + filter_clause - args.extend(filter_args) + Returns: + Deferred[tuple[list[FrozenEvent], str]]: Returns a list of + events and a token pointing to the start of the returned + events. + The events returned are in ascending order. + """ - if int(limit) > 0: - args.append(int(limit)) - limit_str = " LIMIT ?" - else: - limit_str = "" - - sql = ( - "SELECT * FROM events" - " WHERE outlier = ? AND room_id = ? AND %(bounds)s" - " ORDER BY topological_ordering %(order)s," - " stream_ordering %(order)s %(limit)s" - ) % { - "bounds": bounds, - "order": order, - "limit": limit_str - } - - def f(txn): - txn.execute(sql, args) - - rows = self.cursor_to_dict(txn) - - if rows: - topo = rows[-1]["topological_ordering"] - toke = rows[-1]["stream_ordering"] - if direction == 'b': - # Tokens are positions between events. - # This token points *after* the last event in the chunk. - # We need it to point to the event before it in the chunk - # when we are going backwards so we subtract one from the - # stream part. - toke -= 1 - next_token = str(RoomStreamToken(topo, toke)) - else: - # TODO (erikj): We should work out what to do here instead. - next_token = to_key if to_key else from_key - - return rows, next_token, - - rows, token = yield self.runInteraction("paginate_room_events", f) - - events = yield self._get_events( - [r["event_id"] for r in rows], - get_prev_content=True - ) - - self._set_before_and_after(events, rows) - - defer.returnValue((events, token)) - - @defer.inlineCallbacks - def get_recent_events_for_room(self, room_id, limit, end_token, from_token=None): rows, token = yield self.get_recent_event_ids_for_room( - room_id, limit, end_token, from_token + room_id, limit, end_token, ) logger.debug("stream before") events = yield self._get_events( - [r["event_id"] for r in rows], + [r.event_id for r in rows], get_prev_content=True ) logger.debug("stream after") @@ -479,59 +369,62 @@ class StreamStore(SQLBaseStore): defer.returnValue((events, token)) - @cached(num_args=4) - def get_recent_event_ids_for_room(self, room_id, limit, end_token, from_token=None): - end_token = RoomStreamToken.parse_stream_token(end_token) + @defer.inlineCallbacks + def get_recent_event_ids_for_room(self, room_id, limit, end_token): + """Get the most recent events in the room in topological ordering. - if from_token is None: + Args: + room_id (str) + limit (int) + end_token (str): The stream token representing now. + + Returns: + Deferred[tuple[list[_EventDictReturn], str]]: Returns a list of + _EventDictReturn and a token pointing to the start of the returned + events. + The events returned are in ascending order. + """ + # Allow a zero limit here, and no-op. + if limit == 0: + defer.returnValue(([], end_token)) + + end_token = RoomStreamToken.parse(end_token) + + rows, token = yield self.runInteraction( + "get_recent_event_ids_for_room", self._paginate_room_events_txn, + room_id, from_token=end_token, limit=limit, + ) + + # We want to return the results in ascending order. + rows.reverse() + + defer.returnValue((rows, token)) + + def get_room_event_after_stream_ordering(self, room_id, stream_ordering): + """Gets details of the first event in a room at or after a stream ordering + + Args: + room_id (str): + stream_ordering (int): + + Returns: + Deferred[(int, int, str)]: + (stream ordering, topological ordering, event_id) + """ + def _f(txn): sql = ( "SELECT stream_ordering, topological_ordering, event_id" " FROM events" - " WHERE room_id = ? AND stream_ordering <= ? AND outlier = ?" - " ORDER BY topological_ordering DESC, stream_ordering DESC" - " LIMIT ?" + " WHERE room_id = ? AND stream_ordering >= ?" + " AND NOT outlier" + " ORDER BY stream_ordering" + " LIMIT 1" ) - else: - from_token = RoomStreamToken.parse_stream_token(from_token) - sql = ( - "SELECT stream_ordering, topological_ordering, event_id" - " FROM events" - " WHERE room_id = ? AND stream_ordering > ?" - " AND stream_ordering <= ? AND outlier = ?" - " ORDER BY topological_ordering DESC, stream_ordering DESC" - " LIMIT ?" - ) - - def get_recent_events_for_room_txn(txn): - if from_token is None: - txn.execute(sql, (room_id, end_token.stream, False, limit,)) - else: - txn.execute(sql, ( - room_id, from_token.stream, end_token.stream, False, limit - )) - - rows = self.cursor_to_dict(txn) - - rows.reverse() # As we selected with reverse ordering - - if rows: - # Tokens are positions between events. - # This token points *after* the last event in the chunk. - # We need it to point to the event before it in the chunk - # since we are going backwards so we subtract one from the - # stream part. - topo = rows[0]["topological_ordering"] - toke = rows[0]["stream_ordering"] - 1 - start_token = str(RoomStreamToken(topo, toke)) - - token = (start_token, str(end_token)) - else: - token = (str(end_token), str(end_token)) - - return rows, token + txn.execute(sql, (room_id, stream_ordering, )) + return txn.fetchone() return self.runInteraction( - "get_recent_events_for_room", get_recent_events_for_room_txn + "get_room_event_after_stream_ordering", _f, ) @defer.inlineCallbacks @@ -542,7 +435,7 @@ class StreamStore(SQLBaseStore): `room_id` causes it to return the current room specific topological token. """ - token = yield self._stream_id_gen.get_current_token() + token = yield self.get_room_max_stream_ordering() if room_id is None: defer.returnValue("s%d" % (token,)) else: @@ -552,12 +445,6 @@ class StreamStore(SQLBaseStore): ) defer.returnValue("t%d-%d" % (topo, token)) - def get_room_max_stream_ordering(self): - return self._stream_id_gen.get_current_token() - - def get_room_min_stream_ordering(self): - return self._backfill_id_gen.get_current_token() - def get_stream_token_for_event(self, event_id): """The stream token for an event Args: @@ -615,10 +502,20 @@ class StreamStore(SQLBaseStore): @staticmethod def _set_before_and_after(events, rows, topo_order=True): + """Inserts ordering information to events' internal metadata from + the DB rows. + + Args: + events (list[FrozenEvent]) + rows (list[_EventDictReturn]) + topo_order (bool): Whether the events were ordered topologically + or by stream ordering. If true then all rows should have a non + null topological_ordering. + """ for event, row in zip(events, rows): - stream = row["stream_ordering"] - if topo_order: - topo = event.depth + stream = row.stream_ordering + if topo_order and row.topological_ordering: + topo = row.topological_ordering else: topo = None internal = event.internal_metadata @@ -690,87 +587,27 @@ class StreamStore(SQLBaseStore): retcols=["stream_ordering", "topological_ordering"], ) - token = RoomStreamToken( + # Paginating backwards includes the event at the token, but paginating + # forward doesn't. + before_token = RoomStreamToken( + results["topological_ordering"] - 1, + results["stream_ordering"], + ) + + after_token = RoomStreamToken( results["topological_ordering"], results["stream_ordering"], ) - if isinstance(self.database_engine, Sqlite3Engine): - # SQLite3 doesn't optimise ``(x < a) OR (x = a AND y < b)`` - # So we give pass it to SQLite3 as the UNION ALL of the two queries. + rows, start_token = self._paginate_room_events_txn( + txn, room_id, before_token, direction='b', limit=before_limit, + ) + events_before = [r.event_id for r in rows] - query_before = ( - "SELECT topological_ordering, stream_ordering, event_id FROM events" - " WHERE room_id = ? AND topological_ordering < ?" - " UNION ALL" - " SELECT topological_ordering, stream_ordering, event_id FROM events" - " WHERE room_id = ? AND topological_ordering = ? AND stream_ordering < ?" - " ORDER BY topological_ordering DESC, stream_ordering DESC LIMIT ?" - ) - before_args = ( - room_id, token.topological, - room_id, token.topological, token.stream, - before_limit, - ) - - query_after = ( - "SELECT topological_ordering, stream_ordering, event_id FROM events" - " WHERE room_id = ? AND topological_ordering > ?" - " UNION ALL" - " SELECT topological_ordering, stream_ordering, event_id FROM events" - " WHERE room_id = ? AND topological_ordering = ? AND stream_ordering > ?" - " ORDER BY topological_ordering ASC, stream_ordering ASC LIMIT ?" - ) - after_args = ( - room_id, token.topological, - room_id, token.topological, token.stream, - after_limit, - ) - else: - query_before = ( - "SELECT topological_ordering, stream_ordering, event_id FROM events" - " WHERE room_id = ? AND %s" - " ORDER BY topological_ordering DESC, stream_ordering DESC LIMIT ?" - ) % (upper_bound(token, self.database_engine, inclusive=False),) - - before_args = (room_id, before_limit) - - query_after = ( - "SELECT topological_ordering, stream_ordering, event_id FROM events" - " WHERE room_id = ? AND %s" - " ORDER BY topological_ordering ASC, stream_ordering ASC LIMIT ?" - ) % (lower_bound(token, self.database_engine, inclusive=False),) - - after_args = (room_id, after_limit) - - txn.execute(query_before, before_args) - - rows = self.cursor_to_dict(txn) - events_before = [r["event_id"] for r in rows] - - if rows: - start_token = str(RoomStreamToken( - rows[0]["topological_ordering"], - rows[0]["stream_ordering"] - 1, - )) - else: - start_token = str(RoomStreamToken( - token.topological, - token.stream - 1, - )) - - txn.execute(query_after, after_args) - - rows = self.cursor_to_dict(txn) - events_after = [r["event_id"] for r in rows] - - if rows: - end_token = str(RoomStreamToken( - rows[-1]["topological_ordering"], - rows[-1]["stream_ordering"], - )) - else: - end_token = str(token) + rows, end_token = self._paginate_room_events_txn( + txn, room_id, after_token, direction='f', limit=after_limit, + ) + events_after = [r.event_id for r in rows] return { "before": { @@ -832,3 +669,139 @@ class StreamStore(SQLBaseStore): def has_room_changed_since(self, room_id, stream_id): return self._events_stream_cache.has_entity_changed(room_id, stream_id) + + def _paginate_room_events_txn(self, txn, room_id, from_token, to_token=None, + direction='b', limit=-1, event_filter=None): + """Returns list of events before or after a given token. + + Args: + txn + room_id (str) + from_token (RoomStreamToken): The token used to stream from + to_token (RoomStreamToken|None): A token which if given limits the + results to only those before + direction(char): Either 'b' or 'f' to indicate whether we are + paginating forwards or backwards from `from_key`. + limit (int): The maximum number of events to return. + event_filter (Filter|None): If provided filters the events to + those that match the filter. + + Returns: + Deferred[tuple[list[_EventDictReturn], str]]: Returns the results + as a list of _EventDictReturn and a token that points to the end + of the result set. + """ + + assert int(limit) >= 0 + + # Tokens really represent positions between elements, but we use + # the convention of pointing to the event before the gap. Hence + # we have a bit of asymmetry when it comes to equalities. + args = [False, room_id] + if direction == 'b': + order = "DESC" + bounds = upper_bound( + from_token, self.database_engine + ) + if to_token: + bounds = "%s AND %s" % (bounds, lower_bound( + to_token, self.database_engine + )) + else: + order = "ASC" + bounds = lower_bound( + from_token, self.database_engine + ) + if to_token: + bounds = "%s AND %s" % (bounds, upper_bound( + to_token, self.database_engine + )) + + filter_clause, filter_args = filter_to_clause(event_filter) + + if filter_clause: + bounds += " AND " + filter_clause + args.extend(filter_args) + + args.append(int(limit)) + + sql = ( + "SELECT event_id, topological_ordering, stream_ordering" + " FROM events" + " WHERE outlier = ? AND room_id = ? AND %(bounds)s" + " ORDER BY topological_ordering %(order)s," + " stream_ordering %(order)s LIMIT ?" + ) % { + "bounds": bounds, + "order": order, + } + + txn.execute(sql, args) + + rows = [_EventDictReturn(row[0], row[1], row[2]) for row in txn] + + if rows: + topo = rows[-1].topological_ordering + toke = rows[-1].stream_ordering + if direction == 'b': + # Tokens are positions between events. + # This token points *after* the last event in the chunk. + # We need it to point to the event before it in the chunk + # when we are going backwards so we subtract one from the + # stream part. + toke -= 1 + next_token = RoomStreamToken(topo, toke) + else: + # TODO (erikj): We should work out what to do here instead. + next_token = to_token if to_token else from_token + + return rows, str(next_token), + + @defer.inlineCallbacks + def paginate_room_events(self, room_id, from_key, to_key=None, + direction='b', limit=-1, event_filter=None): + """Returns list of events before or after a given token. + + Args: + room_id (str) + from_key (str): The token used to stream from + to_key (str|None): A token which if given limits the results to + only those before + direction(char): Either 'b' or 'f' to indicate whether we are + paginating forwards or backwards from `from_key`. + limit (int): The maximum number of events to return. Zero or less + means no limit. + event_filter (Filter|None): If provided filters the events to + those that match the filter. + + Returns: + tuple[list[dict], str]: Returns the results as a list of dicts and + a token that points to the end of the result set. The dicts have + the keys "event_id", "topological_ordering" and "stream_orderign". + """ + + from_key = RoomStreamToken.parse(from_key) + if to_key: + to_key = RoomStreamToken.parse(to_key) + + rows, token = yield self.runInteraction( + "paginate_room_events", self._paginate_room_events_txn, + room_id, from_key, to_key, direction, limit, event_filter, + ) + + events = yield self._get_events( + [r.event_id for r in rows], + get_prev_content=True + ) + + self._set_before_and_after(events, rows) + + defer.returnValue((events, token)) + + +class StreamStore(StreamWorkerStore): + def get_room_max_stream_ordering(self): + return self._stream_id_gen.get_current_token() + + def get_room_min_stream_ordering(self): + return self._backfill_id_gen.get_current_token() diff --git a/synapse/storage/tags.py b/synapse/storage/tags.py index bff73f3f0..0f657b2bd 100644 --- a/synapse/storage/tags.py +++ b/synapse/storage/tags.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2014-2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -13,25 +14,21 @@ # See the License for the specific language governing permissions and # limitations under the License. -from ._base import SQLBaseStore -from synapse.util.caches.descriptors import cached +import logging + +from six.moves import range + +from canonicaljson import json + from twisted.internet import defer -import ujson as json -import logging +from synapse.storage.account_data import AccountDataWorkerStore +from synapse.util.caches.descriptors import cached logger = logging.getLogger(__name__) -class TagsStore(SQLBaseStore): - def get_max_account_data_stream_id(self): - """Get the current max stream id for the private user data stream - - Returns: - A deferred int. - """ - return self._account_data_id_gen.get_current_token() - +class TagsWorkerStore(AccountDataWorkerStore): @cached() def get_tags_for_user(self, user_id): """Get all the tags for a user. @@ -104,7 +101,7 @@ class TagsStore(SQLBaseStore): batch_size = 50 results = [] - for i in xrange(0, len(tag_ids), batch_size): + for i in range(0, len(tag_ids), batch_size): tags = yield self.runInteraction( "get_all_updated_tag_content", get_tag_content, @@ -170,6 +167,8 @@ class TagsStore(SQLBaseStore): row["tag"]: json.loads(row["content"]) for row in rows }) + +class TagsStore(TagsWorkerStore): @defer.inlineCallbacks def add_tag_to_room(self, user_id, room_id, tag, content): """Add a tag to a room for a user. diff --git a/synapse/storage/transactions.py b/synapse/storage/transactions.py index 809fdd311..c3bc94f56 100644 --- a/synapse/storage/transactions.py +++ b/synapse/storage/transactions.py @@ -13,17 +13,25 @@ # See the License for the specific language governing permissions and # limitations under the License. -from ._base import SQLBaseStore -from synapse.util.caches.descriptors import cached +import logging +from collections import namedtuple + +import six + +from canonicaljson import encode_canonical_json, json from twisted.internet import defer -from canonicaljson import encode_canonical_json +from synapse.util.caches.descriptors import cached -from collections import namedtuple +from ._base import SQLBaseStore -import logging -import ujson as json +# py2 sqlite has buffer hardcoded as only binary type, so we must use it, +# despite being deprecated and removed in favor of memoryview +if six.PY2: + db_binary_type = buffer +else: + db_binary_type = memoryview logger = logging.getLogger(__name__) @@ -46,8 +54,8 @@ class TransactionStore(SQLBaseStore): """A collection of queries for handling PDUs. """ - def __init__(self, hs): - super(TransactionStore, self).__init__(hs) + def __init__(self, db_conn, hs): + super(TransactionStore, self).__init__(db_conn, hs) self._clock.looping_call(self._cleanup_transactions, 30 * 60 * 1000) @@ -110,7 +118,7 @@ class TransactionStore(SQLBaseStore): "transaction_id": transaction_id, "origin": origin, "response_code": code, - "response_json": buffer(encode_canonical_json(response_dict)), + "response_json": db_binary_type(encode_canonical_json(response_dict)), "ts": self._clock.time_msec(), }, or_ignore=True, diff --git a/synapse/storage/user_directory.py b/synapse/storage/user_directory.py index 2a4db3f03..a8781b0e5 100644 --- a/synapse/storage/user_directory.py +++ b/synapse/storage/user_directory.py @@ -13,17 +13,19 @@ # See the License for the specific language governing permissions and # limitations under the License. +import logging +import re + +from six import iteritems + from twisted.internet import defer -from ._base import SQLBaseStore - -from synapse.util.caches.descriptors import cached, cachedInlineCallbacks from synapse.api.constants import EventTypes, JoinRules from synapse.storage.engines import PostgresEngine, Sqlite3Engine from synapse.types import get_domain_from_id, get_localpart_from_id +from synapse.util.caches.descriptors import cached, cachedInlineCallbacks -import re -import logging +from ._base import SQLBaseStore logger = logging.getLogger(__name__) @@ -63,7 +65,7 @@ class UserDirectoryStore(SQLBaseStore): user_ids (list(str)): Users to add """ yield self._simple_insert_many( - table="users_in_pubic_room", + table="users_in_public_rooms", values=[ { "user_id": user_id, @@ -100,7 +102,7 @@ class UserDirectoryStore(SQLBaseStore): user_id, get_localpart_from_id(user_id), get_domain_from_id(user_id), profile.display_name, ) - for user_id, profile in users_with_profile.iteritems() + for user_id, profile in iteritems(users_with_profile) ) elif isinstance(self.database_engine, Sqlite3Engine): sql = """ @@ -112,7 +114,7 @@ class UserDirectoryStore(SQLBaseStore): user_id, "%s %s" % (user_id, p.display_name,) if p.display_name else user_id ) - for user_id, p in users_with_profile.iteritems() + for user_id, p in iteritems(users_with_profile) ) else: # This should be unreachable. @@ -130,7 +132,7 @@ class UserDirectoryStore(SQLBaseStore): "display_name": profile.display_name, "avatar_url": profile.avatar_url, } - for user_id, profile in users_with_profile.iteritems() + for user_id, profile in iteritems(users_with_profile) ] ) for user_id in users_with_profile: @@ -164,7 +166,7 @@ class UserDirectoryStore(SQLBaseStore): ) if isinstance(self.database_engine, PostgresEngine): - # We weight the loclpart most highly, then display name and finally + # We weight the localpart most highly, then display name and finally # server name if new_entry: sql = """ @@ -219,7 +221,7 @@ class UserDirectoryStore(SQLBaseStore): @defer.inlineCallbacks def update_user_in_public_user_list(self, user_id, room_id): yield self._simple_update_one( - table="users_in_pubic_room", + table="users_in_public_rooms", keyvalues={"user_id": user_id}, updatevalues={"room_id": room_id}, desc="update_user_in_public_user_list", @@ -240,7 +242,7 @@ class UserDirectoryStore(SQLBaseStore): ) self._simple_delete_txn( txn, - table="users_in_pubic_room", + table="users_in_public_rooms", keyvalues={"user_id": user_id}, ) txn.call_after( @@ -256,18 +258,18 @@ class UserDirectoryStore(SQLBaseStore): @defer.inlineCallbacks def remove_from_user_in_public_room(self, user_id): yield self._simple_delete( - table="users_in_pubic_room", + table="users_in_public_rooms", keyvalues={"user_id": user_id}, desc="remove_from_user_in_public_room", ) self.get_user_in_public_room.invalidate((user_id,)) def get_users_in_public_due_to_room(self, room_id): - """Get all user_ids that are in the room directory becuase they're + """Get all user_ids that are in the room directory because they're in the given room_id """ return self._simple_select_onecol( - table="users_in_pubic_room", + table="users_in_public_rooms", keyvalues={"room_id": room_id}, retcol="user_id", desc="get_users_in_public_due_to_room", @@ -275,7 +277,7 @@ class UserDirectoryStore(SQLBaseStore): @defer.inlineCallbacks def get_users_in_dir_due_to_room(self, room_id): - """Get all user_ids that are in the room directory becuase they're + """Get all user_ids that are in the room directory because they're in the given room_id """ user_ids_dir = yield self._simple_select_onecol( @@ -286,7 +288,7 @@ class UserDirectoryStore(SQLBaseStore): ) user_ids_pub = yield self._simple_select_onecol( - table="users_in_pubic_room", + table="users_in_public_rooms", keyvalues={"room_id": room_id}, retcol="user_id", desc="get_users_in_dir_due_to_room", @@ -317,6 +319,16 @@ class UserDirectoryStore(SQLBaseStore): rows = yield self._execute("get_all_rooms", None, sql) defer.returnValue([room_id for room_id, in rows]) + @defer.inlineCallbacks + def get_all_local_users(self): + """Get all local users + """ + sql = """ + SELECT name FROM users + """ + rows = yield self._execute("get_all_local_users", None, sql) + defer.returnValue([name for name, in rows]) + def add_users_who_share_room(self, room_id, share_private, user_id_tuples): """Insert entries into the users_who_share_rooms table. The first user should be a local user. @@ -514,7 +526,7 @@ class UserDirectoryStore(SQLBaseStore): def _delete_all_from_user_dir_txn(txn): txn.execute("DELETE FROM user_directory") txn.execute("DELETE FROM user_directory_search") - txn.execute("DELETE FROM users_in_pubic_room") + txn.execute("DELETE FROM users_in_public_rooms") txn.execute("DELETE FROM users_who_share_rooms") txn.call_after(self.get_user_in_directory.invalidate_all) txn.call_after(self.get_user_in_public_room.invalidate_all) @@ -537,7 +549,7 @@ class UserDirectoryStore(SQLBaseStore): @cached() def get_user_in_public_room(self, user_id): return self._simple_select_one( - table="users_in_pubic_room", + table="users_in_public_rooms", keyvalues={"user_id": user_id}, retcols=("room_id",), allow_none=True, @@ -629,6 +641,25 @@ class UserDirectoryStore(SQLBaseStore): ] } """ + + if self.hs.config.user_directory_search_all_users: + # make s.user_id null to keep the ordering algorithm happy + join_clause = """ + CROSS JOIN (SELECT NULL as user_id) AS s + """ + join_args = () + where_clause = "1=1" + else: + join_clause = """ + LEFT JOIN users_in_public_rooms AS p USING (user_id) + LEFT JOIN ( + SELECT other_user_id AS user_id FROM users_who_share_rooms + WHERE user_id = ? AND share_private + ) AS s USING (user_id) + """ + join_args = (user_id,) + where_clause = "(s.user_id IS NOT NULL OR p.user_id IS NOT NULL)" + if isinstance(self.database_engine, PostgresEngine): full_query, exact_query, prefix_query = _parse_query_postgres(search_term) @@ -638,16 +669,12 @@ class UserDirectoryStore(SQLBaseStore): # The array of numbers are the weights for the various part of the # search: (domain, _, display name, localpart) sql = """ - SELECT d.user_id, display_name, avatar_url + SELECT d.user_id AS user_id, display_name, avatar_url FROM user_directory_search INNER JOIN user_directory AS d USING (user_id) - LEFT JOIN users_in_pubic_room AS p USING (user_id) - LEFT JOIN ( - SELECT other_user_id AS user_id FROM users_who_share_rooms - WHERE user_id = ? AND share_private - ) AS s USING (user_id) + %s WHERE - (s.user_id IS NOT NULL OR p.user_id IS NOT NULL) + %s AND vector @@ to_tsquery('english', ?) ORDER BY (CASE WHEN s.user_id IS NOT NULL THEN 4.0 ELSE 1.0 END) @@ -671,30 +698,26 @@ class UserDirectoryStore(SQLBaseStore): display_name IS NULL, avatar_url IS NULL LIMIT ? - """ - args = (user_id, full_query, exact_query, prefix_query, limit + 1,) + """ % (join_clause, where_clause) + args = join_args + (full_query, exact_query, prefix_query, limit + 1,) elif isinstance(self.database_engine, Sqlite3Engine): search_query = _parse_query_sqlite(search_term) sql = """ - SELECT d.user_id, display_name, avatar_url + SELECT d.user_id AS user_id, display_name, avatar_url FROM user_directory_search INNER JOIN user_directory AS d USING (user_id) - LEFT JOIN users_in_pubic_room AS p USING (user_id) - LEFT JOIN ( - SELECT other_user_id AS user_id FROM users_who_share_rooms - WHERE user_id = ? AND share_private - ) AS s USING (user_id) + %s WHERE - (s.user_id IS NOT NULL OR p.user_id IS NOT NULL) + %s AND value MATCH ? ORDER BY rank(matchinfo(user_directory_search)) DESC, display_name IS NULL, avatar_url IS NULL LIMIT ? - """ - args = (user_id, search_query, limit + 1) + """ % (join_clause, where_clause) + args = join_args + (search_query, limit + 1) else: # This should be unreachable. raise Exception("Unrecognized database engine") @@ -723,7 +746,7 @@ def _parse_query_sqlite(search_term): # Pull out the individual words, discarding any non-word characters. results = re.findall(r"([\w\-]+)", search_term, re.UNICODE) - return " & ".join("(%s* | %s)" % (result, result,) for result in results) + return " & ".join("(%s* OR %s)" % (result, result,) for result in results) def _parse_query_postgres(search_term): diff --git a/synapse/storage/user_erasure_store.py b/synapse/storage/user_erasure_store.py new file mode 100644 index 000000000..be013f442 --- /dev/null +++ b/synapse/storage/user_erasure_store.py @@ -0,0 +1,103 @@ +# -*- coding: utf-8 -*- +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import operator + +from twisted.internet import defer + +from synapse.storage._base import SQLBaseStore +from synapse.util.caches.descriptors import cached, cachedList + + +class UserErasureWorkerStore(SQLBaseStore): + @cached() + def is_user_erased(self, user_id): + """ + Check if the given user id has requested erasure + + Args: + user_id (str): full user id to check + + Returns: + Deferred[bool]: True if the user has requested erasure + """ + return self._simple_select_onecol( + table="erased_users", + keyvalues={"user_id": user_id}, + retcol="1", + desc="is_user_erased", + ).addCallback(operator.truth) + + @cachedList( + cached_method_name="is_user_erased", + list_name="user_ids", + inlineCallbacks=True, + ) + def are_users_erased(self, user_ids): + """ + Checks which users in a list have requested erasure + + Args: + user_ids (iterable[str]): full user id to check + + Returns: + Deferred[dict[str, bool]]: + for each user, whether the user has requested erasure. + """ + # this serves the dual purpose of (a) making sure we can do len and + # iterate it multiple times, and (b) avoiding duplicates. + user_ids = tuple(set(user_ids)) + + def _get_erased_users(txn): + txn.execute( + "SELECT user_id FROM erased_users WHERE user_id IN (%s)" % ( + ",".join("?" * len(user_ids)) + ), + user_ids, + ) + return set(r[0] for r in txn) + + erased_users = yield self.runInteraction( + "are_users_erased", _get_erased_users, + ) + res = dict((u, u in erased_users) for u in user_ids) + defer.returnValue(res) + + +class UserErasureStore(UserErasureWorkerStore): + def mark_user_erased(self, user_id): + """Indicate that user_id wishes their message history to be erased. + + Args: + user_id (str): full user_id to be erased + """ + def f(txn): + # first check if they are already in the list + txn.execute( + "SELECT 1 FROM erased_users WHERE user_id = ?", + (user_id, ) + ) + if txn.fetchone(): + return + + # they are not already there: do the insert. + txn.execute( + "INSERT INTO erased_users (user_id) VALUES (?)", + (user_id, ) + ) + + self._invalidate_cache_and_stream( + txn, self.is_user_erased, (user_id,) + ) + return self.runInteraction("mark_user_erased", f) diff --git a/synapse/storage/util/id_generators.py b/synapse/storage/util/id_generators.py index 95031dc9e..d6160d5e4 100644 --- a/synapse/storage/util/id_generators.py +++ b/synapse/storage/util/id_generators.py @@ -13,9 +13,9 @@ # See the License for the specific language governing permissions and # limitations under the License. -from collections import deque import contextlib import threading +from collections import deque class IdGenerator(object): diff --git a/synapse/streams/config.py b/synapse/streams/config.py index 4f089bfb9..451e4fa44 100644 --- a/synapse/streams/config.py +++ b/synapse/streams/config.py @@ -13,11 +13,11 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.api.errors import SynapseError -from synapse.types import StreamToken - import logging +from synapse.api.errors import SynapseError +from synapse.http.servlet import parse_integer, parse_string +from synapse.types import StreamToken logger = logging.getLogger(__name__) @@ -57,48 +57,33 @@ class PaginationConfig(object): @classmethod def from_request(cls, request, raise_invalid_params=True, default_limit=None): - def get_param(name, default=None): - lst = request.args.get(name, []) - if len(lst) > 1: - raise SynapseError( - 400, "%s must be specified only once" % (name,) - ) - elif len(lst) == 1: - return lst[0] - else: - return default + direction = parse_string(request, "dir", default='f', allowed_values=['f', 'b']) - direction = get_param("dir", 'f') - if direction not in ['f', 'b']: - raise SynapseError(400, "'dir' parameter is invalid.") - - from_tok = get_param("from") - to_tok = get_param("to") + from_tok = parse_string(request, "from") + to_tok = parse_string(request, "to") try: if from_tok == "END": from_tok = None # For backwards compat. elif from_tok: from_tok = StreamToken.from_string(from_tok) - except: + except Exception: raise SynapseError(400, "'from' paramater is invalid") try: if to_tok: to_tok = StreamToken.from_string(to_tok) - except: + except Exception: raise SynapseError(400, "'to' paramater is invalid") - limit = get_param("limit", None) - if limit is not None and not limit.isdigit(): - raise SynapseError(400, "'limit' parameter must be an integer.") + limit = parse_integer(request, "limit", default=default_limit) - if limit is None: - limit = default_limit + if limit and limit < 0: + raise SynapseError(400, "Limit must be 0 or above") try: return PaginationConfig(from_tok, to_tok, direction, limit) - except: + except Exception: logger.exception("Failed to create pagination config") raise SynapseError(400, "Invalid request.") diff --git a/synapse/streams/events.py b/synapse/streams/events.py index 91a59b0ba..e5220132a 100644 --- a/synapse/streams/events.py +++ b/synapse/streams/events.py @@ -15,13 +15,12 @@ from twisted.internet import defer -from synapse.types import StreamToken - +from synapse.handlers.account_data import AccountDataEventSource from synapse.handlers.presence import PresenceEventSource +from synapse.handlers.receipts import ReceiptEventSource from synapse.handlers.room import RoomEventSource from synapse.handlers.typing import TypingNotificationEventSource -from synapse.handlers.receipts import ReceiptEventSource -from synapse.handlers.account_data import AccountDataEventSource +from synapse.types import StreamToken class EventSources(object): @@ -45,6 +44,7 @@ class EventSources(object): push_rules_key, _ = self.store.get_push_rules_stream_token() to_device_key = self.store.get_to_device_stream_token() device_list_key = self.store.get_device_stream_token() + groups_key = self.store.get_group_stream_token() token = StreamToken( room_key=( @@ -65,6 +65,7 @@ class EventSources(object): push_rules_key=push_rules_key, to_device_key=to_device_key, device_list_key=device_list_key, + groups_key=groups_key, ) defer.returnValue(token) @@ -73,6 +74,7 @@ class EventSources(object): push_rules_key, _ = self.store.get_push_rules_stream_token() to_device_key = self.store.get_to_device_stream_token() device_list_key = self.store.get_device_stream_token() + groups_key = self.store.get_group_stream_token() token = StreamToken( room_key=( @@ -93,5 +95,6 @@ class EventSources(object): push_rules_key=push_rules_key, to_device_key=to_device_key, device_list_key=device_list_key, + groups_key=groups_key, ) defer.returnValue(token) diff --git a/synapse/types.py b/synapse/types.py index 111948540..08f058f71 100644 --- a/synapse/types.py +++ b/synapse/types.py @@ -12,26 +12,65 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. +import string +from collections import namedtuple from synapse.api.errors import SynapseError -from collections import namedtuple - -Requester = namedtuple("Requester", [ +class Requester(namedtuple("Requester", [ "user", "access_token_id", "is_guest", "device_id", "app_service", -]) -""" -Represents the user making a request +])): + """ + Represents the user making a request -Attributes: - user (UserID): id of the user making the request - access_token_id (int|None): *ID* of the access token used for this - request, or None if it came via the appservice API or similar - is_guest (bool): True if the user making this request is a guest user - device_id (str|None): device_id which was set at authentication time - app_service (ApplicationService|None): the AS requesting on behalf of the user -""" + Attributes: + user (UserID): id of the user making the request + access_token_id (int|None): *ID* of the access token used for this + request, or None if it came via the appservice API or similar + is_guest (bool): True if the user making this request is a guest user + device_id (str|None): device_id which was set at authentication time + app_service (ApplicationService|None): the AS requesting on behalf of the user + """ + + def serialize(self): + """Converts self to a type that can be serialized as JSON, and then + deserialized by `deserialize` + + Returns: + dict + """ + return { + "user_id": self.user.to_string(), + "access_token_id": self.access_token_id, + "is_guest": self.is_guest, + "device_id": self.device_id, + "app_server_id": self.app_service.id if self.app_service else None, + } + + @staticmethod + def deserialize(store, input): + """Converts a dict that was produced by `serialize` back into a + Requester. + + Args: + store (DataStore): Used to convert AS ID to AS object + input (dict): A dict produced by `serialize` + + Returns: + Requester + """ + appservice = None + if input["app_server_id"]: + appservice = store.get_app_service_by_id(input["app_server_id"]) + + return Requester( + user=UserID.from_string(input["user_id"]), + access_token_id=input["access_token_id"], + is_guest=input["is_guest"], + device_id=input["device_id"], + app_service=appservice, + ) def create_requester(user_id, access_token_id=None, is_guest=False, @@ -126,14 +165,10 @@ class DomainSpecificString( try: cls.from_string(s) return True - except: + except Exception: return False - __str__ = to_string - - @classmethod - def create(cls, localpart, domain,): - return cls(localpart=localpart, domain=domain) + __repr__ = to_string class UserID(DomainSpecificString): @@ -156,6 +191,43 @@ class EventID(DomainSpecificString): SIGIL = "$" +class GroupID(DomainSpecificString): + """Structure representing a group ID.""" + SIGIL = "+" + + @classmethod + def from_string(cls, s): + group_id = super(GroupID, cls).from_string(s) + if not group_id.localpart: + raise SynapseError( + 400, + "Group ID cannot be empty", + ) + + if contains_invalid_mxid_characters(group_id.localpart): + raise SynapseError( + 400, + "Group ID can only contain characters a-z, 0-9, or '=_-./'", + ) + + return group_id + + +mxid_localpart_allowed_characters = set("_-./=" + string.ascii_lowercase + string.digits) + + +def contains_invalid_mxid_characters(localpart): + """Check for characters not allowed in an mxid or groupid localpart + + Args: + localpart (basestring): the localpart to be checked + + Returns: + bool: True if there are any naughty characters + """ + return any(c not in mxid_localpart_allowed_characters for c in localpart) + + class StreamToken( namedtuple("Token", ( "room_key", @@ -166,6 +238,7 @@ class StreamToken( "push_rules_key", "to_device_key", "device_list_key", + "groups_key", )) ): _SEPARATOR = "_" @@ -178,7 +251,7 @@ class StreamToken( # i.e. old token from before receipt_key keys.append("0") return cls(*keys) - except: + except Exception: raise SynapseError(400, "Invalid Token") def to_string(self): @@ -204,6 +277,7 @@ class StreamToken( or (int(other.push_rules_key) < int(self.push_rules_key)) or (int(other.to_device_key) < int(self.to_device_key)) or (int(other.device_list_key) < int(self.device_list_key)) + or (int(other.groups_key) < int(self.groups_key)) ) def copy_and_advance(self, key, new_value): @@ -263,7 +337,7 @@ class RoomStreamToken(namedtuple("_StreamToken", "topological stream")): if string[0] == 't': parts = string[1:].split('-', 1) return cls(topological=int(parts[0]), stream=int(parts[1])) - except: + except Exception: pass raise SynapseError(400, "Invalid token %r" % (string,)) @@ -272,7 +346,7 @@ class RoomStreamToken(namedtuple("_StreamToken", "topological stream")): try: if string[0] == 's': return cls(topological=None, stream=int(string[1:])) - except: + except Exception: pass raise SynapseError(400, "Invalid token %r" % (string,)) diff --git a/synapse/util/__init__.py b/synapse/util/__init__.py index 2a2360ab5..680ea928c 100644 --- a/synapse/util/__init__.py +++ b/synapse/util/__init__.py @@ -13,38 +13,45 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.api.errors import SynapseError +import logging +from itertools import islice + +import attr + +from twisted.internet import defer, task + from synapse.util.logcontext import PreserveLoggingContext -from twisted.internet import defer, reactor, task - -import time -import logging - logger = logging.getLogger(__name__) -class DeferredTimedOutError(SynapseError): - def __init__(self): - super(DeferredTimedOutError, self).__init__(504, "Timed out") - - def unwrapFirstError(failure): # defer.gatherResults and DeferredLists wrap failures. failure.trap(defer.FirstError) return failure.value.subFailure +@attr.s class Clock(object): - """A small utility that obtains current time-of-day so that time may be - mocked during unit-tests. - - TODO(paul): Also move the sleep() functionality into it """ + A Clock wraps a Twisted reactor and provides utilities on top of it. + + Args: + reactor: The Twisted reactor to use. + """ + _reactor = attr.ib() + + @defer.inlineCallbacks + def sleep(self, seconds): + d = defer.Deferred() + with PreserveLoggingContext(): + self._reactor.callLater(seconds, d.callback, seconds) + res = yield d + defer.returnValue(res) def time(self): """Returns the current system time in seconds since epoch.""" - return time.time() + return self._reactor.seconds() def time_msec(self): """Returns the current system time in miliseconds since epoch.""" @@ -59,9 +66,10 @@ class Clock(object): f(function): The function to call repeatedly. msec(float): How long to wait between calls in milliseconds. """ - l = task.LoopingCall(f) - l.start(msec / 1000.0, now=False) - return l + call = task.LoopingCall(f) + call.clock = self._reactor + call.start(msec / 1000.0, now=False) + return call def call_later(self, delay, callback, *args, **kwargs): """Call something later @@ -77,61 +85,27 @@ class Clock(object): callback(*args, **kwargs) with PreserveLoggingContext(): - return reactor.callLater(delay, wrapped_callback, *args, **kwargs) + return self._reactor.callLater(delay, wrapped_callback, *args, **kwargs) def cancel_call_later(self, timer, ignore_errs=False): try: timer.cancel() - except: + except Exception: if not ignore_errs: raise - def time_bound_deferred(self, given_deferred, time_out): - if given_deferred.called: - return given_deferred - ret_deferred = defer.Deferred() +def batch_iter(iterable, size): + """batch an iterable up into tuples with a maximum size - def timed_out_fn(): - e = DeferredTimedOutError() + Args: + iterable (iterable): the iterable to slice + size (int): the maximum batch size - try: - ret_deferred.errback(e) - except: - pass - - try: - given_deferred.cancel() - except: - pass - - timer = None - - def cancel(res): - try: - self.cancel_call_later(timer) - except: - pass - return res - - ret_deferred.addBoth(cancel) - - def success(res): - try: - ret_deferred.callback(res) - except: - pass - - return res - - def err(res): - try: - ret_deferred.errback(res) - except: - pass - - given_deferred.addCallbacks(callback=success, errback=err) - - timer = self.call_later(time_out, timed_out_fn) - - return ret_deferred + Returns: + an iterator over the chunks + """ + # make sure we can deal with iterables like lists too + sourceiter = iter(iterable) + # call islice until it returns an empty tuple + return iter(lambda: tuple(islice(sourceiter, size)), ()) diff --git a/synapse/util/async.py b/synapse/util/async.py index 1453faf0e..a7094e2fb 100644 --- a/synapse/util/async.py +++ b/synapse/util/async.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2014-2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -12,38 +13,27 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. - - -from twisted.internet import defer, reactor - -from .logcontext import ( - PreserveLoggingContext, preserve_fn, preserve_context_over_deferred, -) -from synapse.util import unwrapFirstError - +import collections +import logging from contextlib import contextmanager -import logging +from six.moves import range + +from twisted.internet import defer +from twisted.internet.defer import CancelledError +from twisted.python import failure + +from synapse.util import Clock, logcontext, unwrapFirstError + +from .logcontext import ( + PreserveLoggingContext, + make_deferred_yieldable, + run_in_background, +) logger = logging.getLogger(__name__) -@defer.inlineCallbacks -def sleep(seconds): - d = defer.Deferred() - with PreserveLoggingContext(): - reactor.callLater(seconds, d.callback, seconds) - res = yield d - defer.returnValue(res) - - -def run_on_reactor(): - """ This will cause the rest of the function to be invoked upon the next - iteration of the main loop - """ - return sleep(0) - - class ObservableDeferred(object): """Wraps a deferred object so that we can add observer deferreds. These observer deferreds do not affect the callback chain of the original @@ -53,6 +43,11 @@ class ObservableDeferred(object): Cancelling or otherwise resolving an observer will not affect the original ObservableDeferred. + + NB that it does not attempt to do anything with logcontexts; in general + you should probably make_deferred_yieldable the deferreds + returned by `observe`, and ensure that the original deferred runs its + callbacks in the sentinel logcontext. """ __slots__ = ["_deferred", "_observers", "_result"] @@ -68,7 +63,7 @@ class ObservableDeferred(object): try: # TODO: Handle errors here. self._observers.pop().callback(r) - except: + except Exception: pass return r @@ -78,7 +73,7 @@ class ObservableDeferred(object): try: # TODO: Handle errors here. self._observers.pop().errback(f) - except: + except Exception: pass if consumeErrors: @@ -151,77 +146,19 @@ def concurrently_execute(func, args, limit): def _concurrently_execute_inner(): try: while True: - yield func(it.next()) + yield func(next(it)) except StopIteration: pass - return preserve_context_over_deferred(defer.gatherResults([ - preserve_fn(_concurrently_execute_inner)() - for _ in xrange(limit) + return logcontext.make_deferred_yieldable(defer.gatherResults([ + run_in_background(_concurrently_execute_inner) + for _ in range(limit) ], consumeErrors=True)).addErrback(unwrapFirstError) class Linearizer(object): - """Linearizes access to resources based on a key. Useful to ensure only one - thing is happening at a time on a given resource. - - Example: - - with (yield linearizer.queue("test_key")): - # do some work. - - """ - def __init__(self, name=None): - if name is None: - self.name = id(self) - else: - self.name = name - self.key_to_defer = {} - - @defer.inlineCallbacks - def queue(self, key): - # If there is already a deferred in the queue, we pull it out so that - # we can wait on it later. - # Then we replace it with a deferred that we resolve *after* the - # context manager has exited. - # We only return the context manager after the previous deferred has - # resolved. - # This all has the net effect of creating a chain of deferreds that - # wait for the previous deferred before starting their work. - current_defer = self.key_to_defer.get(key) - - new_defer = defer.Deferred() - self.key_to_defer[key] = new_defer - - if current_defer: - logger.info( - "Waiting to acquire linearizer lock %r for key %r", self.name, key - ) - try: - with PreserveLoggingContext(): - yield current_defer - except: - logger.exception("Unexpected exception in Linearizer") - - logger.info("Acquired linearizer lock %r for key %r", self.name, key) - - @contextmanager - def _ctx_manager(): - try: - yield - finally: - logger.info("Releasing linearizer lock %r for key %r", self.name, key) - new_defer.callback(None) - current_d = self.key_to_defer.get(key) - if current_d is new_defer: - self.key_to_defer.pop(key, None) - - defer.returnValue(_ctx_manager()) - - -class Limiter(object): """Limits concurrent access to resources based on a key. Useful to ensure - only a few thing happen at a time on a given resource. + only a few things happen at a time on a given resource. Example: @@ -229,22 +166,31 @@ class Limiter(object): # do some work. """ - def __init__(self, max_count): + def __init__(self, name=None, max_count=1, clock=None): """ Args: - max_count(int): The maximum number of concurrent access + max_count(int): The maximum number of concurrent accesses """ + if name is None: + self.name = id(self) + else: + self.name = name + + if not clock: + from twisted.internet import reactor + clock = Clock(reactor) + self._clock = clock self.max_count = max_count # key_to_defer is a map from the key to a 2 element list where - # the first element is the number of things executing - # the second element is a list of deferreds for the things blocked from - # executing. + # the first element is the number of things executing, and + # the second element is an OrderedDict, where the keys are deferreds for the + # things blocked from executing. self.key_to_defer = {} @defer.inlineCallbacks def queue(self, key): - entry = self.key_to_defer.setdefault(key, [0, []]) + entry = self.key_to_defer.setdefault(key, [0, collections.OrderedDict()]) # If the number of things executing is greater than the maximum # then add a deferred to the list of blocked items @@ -252,27 +198,71 @@ class Limiter(object): # this item so that it can continue executing. if entry[0] >= self.max_count: new_defer = defer.Deferred() - entry[1].append(new_defer) - with PreserveLoggingContext(): - yield new_defer + entry[1][new_defer] = 1 - entry[0] += 1 + logger.info( + "Waiting to acquire linearizer lock %r for key %r", self.name, key, + ) + try: + yield make_deferred_yieldable(new_defer) + except Exception as e: + if isinstance(e, CancelledError): + logger.info( + "Cancelling wait for linearizer lock %r for key %r", + self.name, key, + ) + else: + logger.warn( + "Unexpected exception waiting for linearizer lock %r for key %r", + self.name, key, + ) + + # we just have to take ourselves back out of the queue. + del entry[1][new_defer] + raise + + logger.info("Acquired linearizer lock %r for key %r", self.name, key) + entry[0] += 1 + + # if the code holding the lock completes synchronously, then it + # will recursively run the next claimant on the list. That can + # relatively rapidly lead to stack exhaustion. This is essentially + # the same problem as http://twistedmatrix.com/trac/ticket/9304. + # + # In order to break the cycle, we add a cheeky sleep(0) here to + # ensure that we fall back to the reactor between each iteration. + # + # (This needs to happen while we hold the lock, and the context manager's exit + # code must be synchronous, so this is the only sensible place.) + yield self._clock.sleep(0) + + else: + logger.info( + "Acquired uncontended linearizer lock %r for key %r", self.name, key, + ) + entry[0] += 1 @contextmanager def _ctx_manager(): try: yield finally: + logger.info("Releasing linearizer lock %r for key %r", self.name, key) + # We've finished executing so check if there are any things # blocked waiting to execute and start one of them entry[0] -= 1 - try: - entry[1].pop(0).callback(None) - except IndexError: - # If nothing else is executing for this key then remove it - # from the map - if entry[0] == 0: - self.key_to_defer.pop(key, None) + + if entry[1]: + (next_def, _) = entry[1].popitem(last=False) + + # we need to run the next thing in the sentinel context. + with PreserveLoggingContext(): + next_def.callback(None) + elif entry[0] == 0: + # We were the last thing for this key: remove it from the + # map. + del self.key_to_defer[key] defer.returnValue(_ctx_manager()) @@ -316,7 +306,7 @@ class ReadWriteLock(object): # We wait for the latest writer to finish writing. We can safely ignore # any existing readers... as they're readers. - yield curr_writer + yield make_deferred_yieldable(curr_writer) @contextmanager def _ctx_manager(): @@ -345,7 +335,7 @@ class ReadWriteLock(object): curr_readers.clear() self.key_to_current_writer[key] = new_defer - yield preserve_context_over_deferred(defer.gatherResults(to_wait_on)) + yield make_deferred_yieldable(defer.gatherResults(to_wait_on)) @contextmanager def _ctx_manager(): @@ -357,3 +347,69 @@ class ReadWriteLock(object): self.key_to_current_writer.pop(key) defer.returnValue(_ctx_manager()) + + +class DeferredTimeoutError(Exception): + """ + This error is raised by default when a L{Deferred} times out. + """ + + +def add_timeout_to_deferred(deferred, timeout, reactor, on_timeout_cancel=None): + """ + Add a timeout to a deferred by scheduling it to be cancelled after + timeout seconds. + + This is essentially a backport of deferred.addTimeout, which was introduced + in twisted 16.5. + + If the deferred gets timed out, it errbacks with a DeferredTimeoutError, + unless a cancelable function was passed to its initialization or unless + a different on_timeout_cancel callable is provided. + + Args: + deferred (defer.Deferred): deferred to be timed out + timeout (Number): seconds to time out after + reactor (twisted.internet.reactor): the Twisted reactor to use + + on_timeout_cancel (callable): A callable which is called immediately + after the deferred times out, and not if this deferred is + otherwise cancelled before the timeout. + + It takes an arbitrary value, which is the value of the deferred at + that exact point in time (probably a CancelledError Failure), and + the timeout. + + The default callable (if none is provided) will translate a + CancelledError Failure into a DeferredTimeoutError. + """ + timed_out = [False] + + def time_it_out(): + timed_out[0] = True + deferred.cancel() + + delayed_call = reactor.callLater(timeout, time_it_out) + + def convert_cancelled(value): + if timed_out[0]: + to_call = on_timeout_cancel or _cancelled_to_timed_out_error + return to_call(value, timeout) + return value + + deferred.addBoth(convert_cancelled) + + def cancel_timeout(result): + # stop the pending call to cancel the deferred if it's been fired + if delayed_call.active(): + delayed_call.cancel() + return result + + deferred.addBoth(cancel_timeout) + + +def _cancelled_to_timed_out_error(value, timeout): + if isinstance(value, failure.Failure): + value.trap(CancelledError) + raise DeferredTimeoutError(timeout, "Deferred") + return value diff --git a/synapse/util/caches/__init__.py b/synapse/util/caches/__init__.py index 4adae9668..7b065b195 100644 --- a/synapse/util/caches/__init__.py +++ b/synapse/util/caches/__init__.py @@ -13,28 +13,87 @@ # See the License for the specific language governing permissions and # limitations under the License. -import synapse.metrics import os +import six +from six.moves import intern + +from prometheus_client.core import REGISTRY, Gauge, GaugeMetricFamily + CACHE_SIZE_FACTOR = float(os.environ.get("SYNAPSE_CACHE_FACTOR", 0.5)) -metrics = synapse.metrics.get_metrics_for("synapse.util.caches") + +def get_cache_factor_for(cache_name): + env_var = "SYNAPSE_CACHE_FACTOR_" + cache_name.upper() + factor = os.environ.get(env_var) + if factor: + return float(factor) + + return CACHE_SIZE_FACTOR + caches_by_name = {} -# cache_counter = metrics.register_cache( -# "cache", -# lambda: {(name,): len(caches_by_name[name]) for name in caches_by_name.keys()}, -# labels=["name"], -# ) +collectors_by_name = {} + +cache_size = Gauge("synapse_util_caches_cache:size", "", ["name"]) +cache_hits = Gauge("synapse_util_caches_cache:hits", "", ["name"]) +cache_evicted = Gauge("synapse_util_caches_cache:evicted_size", "", ["name"]) +cache_total = Gauge("synapse_util_caches_cache:total", "", ["name"]) + +response_cache_size = Gauge("synapse_util_caches_response_cache:size", "", ["name"]) +response_cache_hits = Gauge("synapse_util_caches_response_cache:hits", "", ["name"]) +response_cache_evicted = Gauge( + "synapse_util_caches_response_cache:evicted_size", "", ["name"] +) +response_cache_total = Gauge("synapse_util_caches_response_cache:total", "", ["name"]) -def register_cache(name, cache): - caches_by_name[name] = cache - return metrics.register_cache( - "cache", - lambda: len(cache), - name, - ) +def register_cache(cache_type, cache_name, cache): + + # Check if the metric is already registered. Unregister it, if so. + # This usually happens during tests, as at runtime these caches are + # effectively singletons. + metric_name = "cache_%s_%s" % (cache_type, cache_name) + if metric_name in collectors_by_name.keys(): + REGISTRY.unregister(collectors_by_name[metric_name]) + + class CacheMetric(object): + + hits = 0 + misses = 0 + evicted_size = 0 + + def inc_hits(self): + self.hits += 1 + + def inc_misses(self): + self.misses += 1 + + def inc_evictions(self, size=1): + self.evicted_size += size + + def describe(self): + return [] + + def collect(self): + if cache_type == "response_cache": + response_cache_size.labels(cache_name).set(len(cache)) + response_cache_hits.labels(cache_name).set(self.hits) + response_cache_evicted.labels(cache_name).set(self.evicted_size) + response_cache_total.labels(cache_name).set(self.hits + self.misses) + else: + cache_size.labels(cache_name).set(len(cache)) + cache_hits.labels(cache_name).set(self.hits) + cache_evicted.labels(cache_name).set(self.evicted_size) + cache_total.labels(cache_name).set(self.hits + self.misses) + + yield GaugeMetricFamily("__unused", "") + + metric = CacheMetric() + REGISTRY.register(metric) + caches_by_name[cache_name] = cache + collectors_by_name[metric_name] = metric + return metric KNOWN_KEYS = { @@ -66,7 +125,9 @@ def intern_string(string): return None try: - string = string.encode("ascii") + if six.PY2: + string = string.encode("ascii") + return intern(string) except UnicodeEncodeError: return string diff --git a/synapse/util/caches/descriptors.py b/synapse/util/caches/descriptors.py index af65bfe7b..f8a07df6b 100644 --- a/synapse/util/caches/descriptors.py +++ b/synapse/util/caches/descriptors.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2015, 2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -12,25 +13,26 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. +import functools +import inspect import logging +import threading +from collections import namedtuple +import six +from six import itervalues, string_types + +from twisted.internet import defer + +from synapse.util import logcontext, unwrapFirstError from synapse.util.async import ObservableDeferred -from synapse.util import unwrapFirstError, logcontext -from synapse.util.caches import CACHE_SIZE_FACTOR +from synapse.util.caches import get_cache_factor_for from synapse.util.caches.lrucache import LruCache from synapse.util.caches.treecache import TreeCache, iterate_tree_cache_entry from synapse.util.stringutils import to_ascii from . import register_cache -from twisted.internet import defer -from collections import namedtuple - -import functools -import inspect -import threading - - logger = logging.getLogger(__name__) @@ -39,12 +41,11 @@ _CacheSentinel = object() class CacheEntry(object): __slots__ = [ - "deferred", "sequence", "callbacks", "invalidated" + "deferred", "callbacks", "invalidated" ] - def __init__(self, deferred, sequence, callbacks): + def __init__(self, deferred, callbacks): self.deferred = deferred - self.sequence = sequence self.callbacks = set(callbacks) self.invalidated = False @@ -62,7 +63,6 @@ class Cache(object): "max_entries", "name", "keylen", - "sequence", "thread", "metrics", "_pending_deferred_cache", @@ -75,13 +75,16 @@ class Cache(object): self.cache = LruCache( max_size=max_entries, keylen=keylen, cache_type=cache_type, size_callback=(lambda d: len(d)) if iterable else None, + evicted_callback=self._on_evicted, ) self.name = name self.keylen = keylen - self.sequence = 0 self.thread = None - self.metrics = register_cache(name, self.cache) + self.metrics = register_cache("cache", name, self.cache) + + def _on_evicted(self, evicted_count): + self.metrics.inc_evictions(evicted_count) def check_thread(self): expected_thread = self.thread @@ -109,11 +112,10 @@ class Cache(object): callbacks = [callback] if callback else [] val = self._pending_deferred_cache.get(key, _CacheSentinel) if val is not _CacheSentinel: - if val.sequence == self.sequence: - val.callbacks.update(callbacks) - if update_metrics: - self.metrics.inc_hits() - return val.deferred + val.callbacks.update(callbacks) + if update_metrics: + self.metrics.inc_hits() + return val.deferred val = self.cache.get(key, _CacheSentinel, callbacks=callbacks) if val is not _CacheSentinel: @@ -133,12 +135,9 @@ class Cache(object): self.check_thread() entry = CacheEntry( deferred=value, - sequence=self.sequence, callbacks=callbacks, ) - entry.callbacks.update(callbacks) - existing_entry = self._pending_deferred_cache.pop(key, None) if existing_entry: existing_entry.invalidate() @@ -146,13 +145,25 @@ class Cache(object): self._pending_deferred_cache[key] = entry def shuffle(result): - if self.sequence == entry.sequence: - existing_entry = self._pending_deferred_cache.pop(key, None) - if existing_entry is entry: - self.cache.set(key, result, entry.callbacks) - else: - entry.invalidate() + existing_entry = self._pending_deferred_cache.pop(key, None) + if existing_entry is entry: + self.cache.set(key, result, entry.callbacks) else: + # oops, the _pending_deferred_cache has been updated since + # we started our query, so we are out of date. + # + # Better put back whatever we took out. (We do it this way + # round, rather than peeking into the _pending_deferred_cache + # and then removing on a match, to make the common case faster) + if existing_entry is not None: + self._pending_deferred_cache[key] = existing_entry + + # we're not going to put this entry into the cache, so need + # to make sure that the invalidation callbacks are called. + # That was probably done when _pending_deferred_cache was + # updated, but it's possible that `set` was called without + # `invalidate` being previously called, in which case it may + # not have been. Either way, let's double-check now. entry.invalidate() return result @@ -164,25 +175,29 @@ class Cache(object): def invalidate(self, key): self.check_thread() + self.cache.pop(key, None) - # Increment the sequence number so that any SELECT statements that - # raced with the INSERT don't update the cache (SYN-369) - self.sequence += 1 + # if we have a pending lookup for this key, remove it from the + # _pending_deferred_cache, which will (a) stop it being returned + # for future queries and (b) stop it being persisted as a proper entry + # in self.cache. entry = self._pending_deferred_cache.pop(key, None) + + # run the invalidation callbacks now, rather than waiting for the + # deferred to resolve. if entry: entry.invalidate() - self.cache.pop(key, None) - def invalidate_many(self, key): self.check_thread() if not isinstance(key, tuple): raise TypeError( "The cache key must be a tuple not %r" % (type(key),) ) - self.sequence += 1 self.cache.del_multi(key) + # if we have a pending lookup for this key, remove it from the + # _pending_deferred_cache, as above entry_dict = self._pending_deferred_cache.pop(key, None) if entry_dict is not None: for entry in iterate_tree_cache_entry(entry_dict): @@ -190,8 +205,10 @@ class Cache(object): def invalidate_all(self): self.check_thread() - self.sequence += 1 self.cache.clear() + for entry in itervalues(self._pending_deferred_cache): + entry.invalidate() + self._pending_deferred_cache.clear() class _CacheDescriptorBase(object): @@ -294,7 +311,7 @@ class CacheDescriptor(_CacheDescriptorBase): orig, num_args=num_args, inlineCallbacks=inlineCallbacks, cache_context=cache_context) - max_entries = int(max_entries * CACHE_SIZE_FACTOR) + max_entries = int(max_entries * get_cache_factor_for(orig.__name__)) self.max_entries = max_entries self.tree = tree @@ -376,9 +393,10 @@ class CacheDescriptor(_CacheDescriptorBase): ret.addErrback(onErr) - # If our cache_key is a string, try to convert to ascii to save - # a bit of space in large caches - if isinstance(cache_key, basestring): + # If our cache_key is a string on py2, try to convert to ascii + # to save a bit of space in large caches. Py3 does this + # internally automatically. + if six.PY2 and isinstance(cache_key, string_types): cache_key = to_ascii(cache_key) result_d = ObservableDeferred(ret, consumeErrors=True) @@ -549,7 +567,7 @@ class CacheListDescriptor(_CacheDescriptorBase): return results return logcontext.make_deferred_yieldable(defer.gatherResults( - cached_defers.values(), + list(cached_defers.values()), consumeErrors=True, ).addCallback(update_results_dict).addErrback( unwrapFirstError diff --git a/synapse/util/caches/dictionary_cache.py b/synapse/util/caches/dictionary_cache.py index d4105822b..6c0b5a409 100644 --- a/synapse/util/caches/dictionary_cache.py +++ b/synapse/util/caches/dictionary_cache.py @@ -13,12 +13,13 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.util.caches.lrucache import LruCache -from collections import namedtuple -from . import register_cache -import threading import logging +import threading +from collections import namedtuple +from synapse.util.caches.lrucache import LruCache + +from . import register_cache logger = logging.getLogger(__name__) @@ -55,7 +56,7 @@ class DictionaryCache(object): __slots__ = [] self.sentinel = Sentinel() - self.metrics = register_cache(name, self.cache) + self.metrics = register_cache("dictionary", name, self.cache) def check_thread(self): expected_thread = self.thread @@ -107,34 +108,37 @@ class DictionaryCache(object): self.sequence += 1 self.cache.clear() - def update(self, sequence, key, value, full=False, known_absent=None): + def update(self, sequence, key, value, fetched_keys=None): """Updates the entry in the cache Args: sequence - key - value (dict): The value to update the cache with. - full (bool): Whether the given value is the full dict, or just a - partial subset there of. If not full then any existing entries - for the key will be updated. - known_absent (set): Set of keys that we know don't exist in the full - dict. + key (K) + value (dict[X,Y]): The value to update the cache with. + fetched_keys (None|set[X]): All of the dictionary keys which were + fetched from the database. + + If None, this is the complete value for key K. Otherwise, it + is used to infer a list of keys which we know don't exist in + the full dict. """ self.check_thread() if self.sequence == sequence: # Only update the cache if the caches sequence number matches the # number that the cache had before the SELECT was started (SYN-369) - if known_absent is None: - known_absent = set() - if full: - self._insert(key, value, known_absent) + if fetched_keys is None: + self._insert(key, value, set()) else: - self._update_or_insert(key, value, known_absent) + self._update_or_insert(key, value, fetched_keys) def _update_or_insert(self, key, value, known_absent): - entry = self.cache.setdefault(key, DictionaryEntry(False, set(), {})) + # We pop and reinsert as we need to tell the cache the size may have + # changed + + entry = self.cache.pop(key, DictionaryEntry(False, set(), {})) entry.value.update(value) entry.known_absent.update(known_absent) + self.cache[key] = entry def _insert(self, key, value, known_absent): self.cache[key] = DictionaryEntry(True, known_absent, value) diff --git a/synapse/util/caches/expiringcache.py b/synapse/util/caches/expiringcache.py index 6ad53a639..465adc54a 100644 --- a/synapse/util/caches/expiringcache.py +++ b/synapse/util/caches/expiringcache.py @@ -13,11 +13,11 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.util.caches import register_cache - -from collections import OrderedDict import logging +from collections import OrderedDict +from synapse.metrics.background_process_metrics import run_as_background_process +from synapse.util.caches import register_cache logger = logging.getLogger(__name__) @@ -52,19 +52,22 @@ class ExpiringCache(object): self._cache = OrderedDict() - self.metrics = register_cache(cache_name, self) - self.iterable = iterable self._size_estimate = 0 + self.metrics = register_cache("expiring", cache_name, self) + def start(self): if not self._expiry_ms: # Don't bother starting the loop if things never expire return def f(): - self._prune_cache() + run_as_background_process( + "prune_cache_%s" % self._cache_name, + self._prune_cache, + ) self._clock.looping_call(f, self._expiry_ms / 2) @@ -79,7 +82,11 @@ class ExpiringCache(object): while self._max_len and len(self) > self._max_len: _key, value = self._cache.popitem(last=False) if self.iterable: - self._size_estimate -= len(value.value) + removed_len = len(value.value) + self.metrics.inc_evictions(removed_len) + self._size_estimate -= removed_len + else: + self.metrics.inc_evictions() def __getitem__(self, key): try: diff --git a/synapse/util/caches/lrucache.py b/synapse/util/caches/lrucache.py index cf5fbb679..b684f24e7 100644 --- a/synapse/util/caches/lrucache.py +++ b/synapse/util/caches/lrucache.py @@ -14,8 +14,8 @@ # limitations under the License. -from functools import wraps import threading +from functools import wraps from synapse.util.caches.treecache import TreeCache @@ -49,7 +49,24 @@ class LruCache(object): Can also set callbacks on objects when getting/setting which are fired when that key gets invalidated/evicted. """ - def __init__(self, max_size, keylen=1, cache_type=dict, size_callback=None): + def __init__(self, max_size, keylen=1, cache_type=dict, size_callback=None, + evicted_callback=None): + """ + Args: + max_size (int): + + keylen (int): + + cache_type (type): + type of underlying cache to be used. Typically one of dict + or TreeCache. + + size_callback (func(V) -> int | None): + + evicted_callback (func(int)|None): + if not None, called on eviction with the size of the evicted + entry + """ cache = cache_type() self.cache = cache # Used for introspection. list_root = _Node(None, None, None, None) @@ -61,8 +78,10 @@ class LruCache(object): def evict(): while cache_len() > max_size: todelete = list_root.prev_node - delete_node(todelete) + evicted_len = delete_node(todelete) cache.pop(todelete.key, None) + if evicted_callback: + evicted_callback(evicted_len) def synchronized(f): @wraps(f) @@ -111,12 +130,15 @@ class LruCache(object): prev_node.next_node = next_node next_node.prev_node = prev_node + deleted_len = 1 if size_callback: - cached_cache_len[0] -= size_callback(node.value) + deleted_len = size_callback(node.value) + cached_cache_len[0] -= deleted_len for cb in node.callbacks: cb() node.callbacks.clear() + return deleted_len @synchronized def cache_get(key, default=None, callbacks=[]): @@ -132,14 +154,21 @@ class LruCache(object): def cache_set(key, value, callbacks=[]): node = cache.get(key, None) if node is not None: - if value != node.value: + # We sometimes store large objects, e.g. dicts, which cause + # the inequality check to take a long time. So let's only do + # the check if we have some callbacks to call. + if node.callbacks and value != node.value: for cb in node.callbacks: cb() node.callbacks.clear() - if size_callback: - cached_cache_len[0] -= size_callback(node.value) - cached_cache_len[0] += size_callback(value) + # We don't bother to protect this by value != node.value as + # generally size_callback will be cheap compared with equality + # checks. (For example, taking the size of two dicts is quicker + # than comparing them for equality.) + if size_callback: + cached_cache_len[0] -= size_callback(node.value) + cached_cache_len[0] += size_callback(value) node.callbacks.update(callbacks) diff --git a/synapse/util/caches/response_cache.py b/synapse/util/caches/response_cache.py index 00af53988..a8491b42d 100644 --- a/synapse/util/caches/response_cache.py +++ b/synapse/util/caches/response_cache.py @@ -12,8 +12,15 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. +import logging + +from twisted.internet import defer from synapse.util.async import ObservableDeferred +from synapse.util.caches import register_cache +from synapse.util.logcontext import make_deferred_yieldable, run_in_background + +logger = logging.getLogger(__name__) class ResponseCache(object): @@ -24,20 +31,69 @@ class ResponseCache(object): used rather than trying to compute a new response. """ - def __init__(self, hs, timeout_ms=0): + def __init__(self, hs, name, timeout_ms=0): self.pending_result_cache = {} # Requests that haven't finished yet. self.clock = hs.get_clock() self.timeout_sec = timeout_ms / 1000. + self._name = name + self._metrics = register_cache( + "response_cache", name, self + ) + + def size(self): + return len(self.pending_result_cache) + + def __len__(self): + return self.size() + def get(self, key): + """Look up the given key. + + Can return either a new Deferred (which also doesn't follow the synapse + logcontext rules), or, if the request has completed, the actual + result. You will probably want to make_deferred_yieldable the result. + + If there is no entry for the key, returns None. It is worth noting that + this means there is no way to distinguish a completed result of None + from an absent cache entry. + + Args: + key (hashable): + + Returns: + twisted.internet.defer.Deferred|None|E: None if there is no entry + for this key; otherwise either a deferred result or the result + itself. + """ result = self.pending_result_cache.get(key) if result is not None: + self._metrics.inc_hits() return result.observe() else: + self._metrics.inc_misses() return None def set(self, key, deferred): + """Set the entry for the given key to the given deferred. + + *deferred* should run its callbacks in the sentinel logcontext (ie, + you should wrap normal synapse deferreds with + logcontext.run_in_background). + + Can return either a new Deferred (which also doesn't follow the synapse + logcontext rules), or, if *deferred* was already complete, the actual + result. You will probably want to make_deferred_yieldable the result. + + Args: + key (hashable): + deferred (twisted.internet.defer.Deferred[T): + + Returns: + twisted.internet.defer.Deferred[T]|T: a new deferred, or the actual + result. + """ result = ObservableDeferred(deferred, consumeErrors=True) self.pending_result_cache[key] = result @@ -53,3 +109,52 @@ class ResponseCache(object): result.addBoth(remove) return result.observe() + + def wrap(self, key, callback, *args, **kwargs): + """Wrap together a *get* and *set* call, taking care of logcontexts + + First looks up the key in the cache, and if it is present makes it + follow the synapse logcontext rules and returns it. + + Otherwise, makes a call to *callback(*args, **kwargs)*, which should + follow the synapse logcontext rules, and adds the result to the cache. + + Example usage: + + @defer.inlineCallbacks + def handle_request(request): + # etc + defer.returnValue(result) + + result = yield response_cache.wrap( + key, + handle_request, + request, + ) + + Args: + key (hashable): key to get/set in the cache + + callback (callable): function to call if the key is not found in + the cache + + *args: positional parameters to pass to the callback, if it is used + + **kwargs: named paramters to pass to the callback, if it is used + + Returns: + twisted.internet.defer.Deferred: yieldable result + """ + result = self.get(key) + if not result: + logger.info("[%s]: no cached result for [%s], calculating new one", + self._name, key) + d = run_in_background(callback, *args, **kwargs) + result = self.set(key, d) + elif not isinstance(result, defer.Deferred) or result.called: + logger.info("[%s]: using completed cached result for [%s]", + self._name, key) + else: + logger.info("[%s]: using incomplete cached result for [%s]", + self._name, key) + return make_deferred_yieldable(result) diff --git a/synapse/util/caches/stream_change_cache.py b/synapse/util/caches/stream_change_cache.py index 941d873ab..f2bde74dc 100644 --- a/synapse/util/caches/stream_change_cache.py +++ b/synapse/util/caches/stream_change_cache.py @@ -13,12 +13,11 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.util.caches import register_cache, CACHE_SIZE_FACTOR - - -from blist import sorteddict import logging +from sortedcontainers import SortedDict + +from synapse.util import caches logger = logging.getLogger(__name__) @@ -32,16 +31,18 @@ class StreamChangeCache(object): entities that may have changed since that position. If position key is too old then the cache will simply return all given entities. """ - def __init__(self, name, current_stream_pos, max_size=10000, prefilled_cache={}): - self._max_size = int(max_size * CACHE_SIZE_FACTOR) + + def __init__(self, name, current_stream_pos, max_size=10000, prefilled_cache=None): + self._max_size = int(max_size * caches.CACHE_SIZE_FACTOR) self._entity_to_key = {} - self._cache = sorteddict() + self._cache = SortedDict() self._earliest_known_stream_pos = current_stream_pos self.name = name - self.metrics = register_cache(self.name, self._cache) + self.metrics = caches.register_cache("cache", self.name, self._cache) - for entity, stream_pos in prefilled_cache.items(): - self.entity_has_changed(entity, stream_pos) + if prefilled_cache: + for entity, stream_pos in prefilled_cache.items(): + self.entity_has_changed(entity, stream_pos) def has_entity_changed(self, entity, stream_pos): """Returns True if the entity may have been updated since stream_pos @@ -65,22 +66,25 @@ class StreamChangeCache(object): return False def get_entities_changed(self, entities, stream_pos): - """Returns subset of entities that have had new things since the - given position. If the position is too old it will just return the given list. + """ + Returns subset of entities that have had new things since the given + position. Entities unknown to the cache will be returned. If the + position is too old it will just return the given list. """ assert type(stream_pos) is int if stream_pos >= self._earliest_known_stream_pos: - keys = self._cache.keys() - i = keys.bisect_right(stream_pos) + changed_entities = { + self._cache[k] for k in self._cache.islice( + start=self._cache.bisect_right(stream_pos), + ) + } - result = set( - self._cache[k] for k in keys[i:] - ).intersection(entities) + result = changed_entities.intersection(entities) self.metrics.inc_hits() else: - result = entities + result = set(entities) self.metrics.inc_misses() return result @@ -90,12 +94,13 @@ class StreamChangeCache(object): """ assert type(stream_pos) is int + if not self._cache: + # If we have no cache, nothing can have changed. + return False + if stream_pos >= self._earliest_known_stream_pos: self.metrics.inc_hits() - keys = self._cache.keys() - i = keys.bisect_right(stream_pos) - - return i < len(keys) + return self._cache.bisect_right(stream_pos) < len(self._cache) else: self.metrics.inc_misses() return True @@ -107,10 +112,8 @@ class StreamChangeCache(object): assert type(stream_pos) is int if stream_pos >= self._earliest_known_stream_pos: - keys = self._cache.keys() - i = keys.bisect_right(stream_pos) - - return [self._cache[k] for k in keys[i:]] + return [self._cache[k] for k in self._cache.islice( + start=self._cache.bisect_right(stream_pos))] else: return None @@ -129,8 +132,10 @@ class StreamChangeCache(object): self._entity_to_key[entity] = stream_pos while len(self._cache) > self._max_size: - k, r = self._cache.popitem() - self._earliest_known_stream_pos = max(k, self._earliest_known_stream_pos) + k, r = self._cache.popitem(0) + self._earliest_known_stream_pos = max( + k, self._earliest_known_stream_pos, + ) self._entity_to_key.pop(r, None) def get_max_pos_of_last_change(self, entity): diff --git a/synapse/util/caches/treecache.py b/synapse/util/caches/treecache.py index fcc341a6b..dd4c9e606 100644 --- a/synapse/util/caches/treecache.py +++ b/synapse/util/caches/treecache.py @@ -1,3 +1,5 @@ +from six import itervalues + SENTINEL = object() @@ -49,7 +51,7 @@ class TreeCache(object): if popped is SENTINEL: return default - node_and_keys = zip(nodes, key) + node_and_keys = list(zip(nodes, key)) node_and_keys.reverse() node_and_keys.append((self.root, None)) @@ -76,7 +78,7 @@ def iterate_tree_cache_entry(d): can contain dicts. """ if isinstance(d, dict): - for value_d in d.itervalues(): + for value_d in itervalues(d): for value in iterate_tree_cache_entry(value_d): yield value else: diff --git a/synapse/util/distributor.py b/synapse/util/distributor.py index e68f94ce7..194da8763 100644 --- a/synapse/util/distributor.py +++ b/synapse/util/distributor.py @@ -13,32 +13,22 @@ # See the License for the specific language governing permissions and # limitations under the License. -from twisted.internet import defer - -from synapse.util.logcontext import ( - PreserveLoggingContext, preserve_context_over_fn -) - -from synapse.util import unwrapFirstError - import logging +from twisted.internet import defer + +from synapse.metrics.background_process_metrics import run_as_background_process +from synapse.util.logcontext import make_deferred_yieldable, run_in_background logger = logging.getLogger(__name__) def user_left_room(distributor, user, room_id): - return preserve_context_over_fn( - distributor.fire, - "user_left_room", user=user, room_id=room_id - ) + distributor.fire("user_left_room", user=user, room_id=room_id) def user_joined_room(distributor, user, room_id): - return preserve_context_over_fn( - distributor.fire, - "user_joined_room", user=user, room_id=room_id - ) + distributor.fire("user_joined_room", user=user, room_id=room_id) class Distributor(object): @@ -52,9 +42,7 @@ class Distributor(object): model will do for today. """ - def __init__(self, suppress_failures=True): - self.suppress_failures = suppress_failures - + def __init__(self): self.signals = {} self.pre_registration = {} @@ -64,7 +52,6 @@ class Distributor(object): self.signals[name] = Signal( name, - suppress_failures=self.suppress_failures, ) if name in self.pre_registration: @@ -83,10 +70,18 @@ class Distributor(object): self.pre_registration[name].append(observer) def fire(self, name, *args, **kwargs): + """Dispatches the given signal to the registered observers. + + Runs the observers as a background process. Does not return a deferred. + """ if name not in self.signals: raise KeyError("%r does not have a signal named %s" % (self, name)) - return self.signals[name].fire(*args, **kwargs) + run_as_background_process( + name, + self.signals[name].fire, + *args, **kwargs + ) class Signal(object): @@ -99,9 +94,8 @@ class Signal(object): method into all of the observers. """ - def __init__(self, name, suppress_failures): + def __init__(self, name): self.name = name - self.suppress_failures = suppress_failures self.observers = [] def observe(self, observer): @@ -111,7 +105,6 @@ class Signal(object): Each observer callable may return a Deferred.""" self.observers.append(observer) - @defer.inlineCallbacks def fire(self, *args, **kwargs): """Invokes every callable in the observer list, passing in the args and kwargs. Exceptions thrown by observers are logged but ignored. It is @@ -129,22 +122,17 @@ class Signal(object): failure.type, failure.value, failure.getTracebackObject())) - if not self.suppress_failures: - return failure return defer.maybeDeferred(observer, *args, **kwargs).addErrback(eb) - with PreserveLoggingContext(): - deferreds = [ - do(observer) - for observer in self.observers - ] + deferreds = [ + run_in_background(do, o) + for o in self.observers + ] - res = yield defer.gatherResults( - deferreds, consumeErrors=True - ).addErrback(unwrapFirstError) - - defer.returnValue(res) + return make_deferred_yieldable(defer.gatherResults( + deferreds, consumeErrors=True, + )) def __repr__(self): return "" % (self.name,) diff --git a/synapse/util/file_consumer.py b/synapse/util/file_consumer.py new file mode 100644 index 000000000..629ed4414 --- /dev/null +++ b/synapse/util/file_consumer.py @@ -0,0 +1,147 @@ +# -*- coding: utf-8 -*- +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from six.moves import queue + +from twisted.internet import threads + +from synapse.util.logcontext import make_deferred_yieldable, run_in_background + + +class BackgroundFileConsumer(object): + """A consumer that writes to a file like object. Supports both push + and pull producers + + Args: + file_obj (file): The file like object to write to. Closed when + finished. + reactor (twisted.internet.reactor): the Twisted reactor to use + """ + + # For PushProducers pause if we have this many unwritten slices + _PAUSE_ON_QUEUE_SIZE = 5 + # And resume once the size of the queue is less than this + _RESUME_ON_QUEUE_SIZE = 2 + + def __init__(self, file_obj, reactor): + self._file_obj = file_obj + + self._reactor = reactor + + # Producer we're registered with + self._producer = None + + # True if PushProducer, false if PullProducer + self.streaming = False + + # For PushProducers, indicates whether we've paused the producer and + # need to call resumeProducing before we get more data. + self._paused_producer = False + + # Queue of slices of bytes to be written. When producer calls + # unregister a final None is sent. + self._bytes_queue = queue.Queue() + + # Deferred that is resolved when finished writing + self._finished_deferred = None + + # If the _writer thread throws an exception it gets stored here. + self._write_exception = None + + def registerProducer(self, producer, streaming): + """Part of IConsumer interface + + Args: + producer (IProducer) + streaming (bool): True if push based producer, False if pull + based. + """ + if self._producer: + raise Exception("registerProducer called twice") + + self._producer = producer + self.streaming = streaming + self._finished_deferred = run_in_background( + threads.deferToThreadPool, + self._reactor, + self._reactor.getThreadPool(), + self._writer, + ) + if not streaming: + self._producer.resumeProducing() + + def unregisterProducer(self): + """Part of IProducer interface + """ + self._producer = None + if not self._finished_deferred.called: + self._bytes_queue.put_nowait(None) + + def write(self, bytes): + """Part of IProducer interface + """ + if self._write_exception: + raise self._write_exception + + if self._finished_deferred.called: + raise Exception("consumer has closed") + + self._bytes_queue.put_nowait(bytes) + + # If this is a PushProducer and the queue is getting behind + # then we pause the producer. + if self.streaming and self._bytes_queue.qsize() >= self._PAUSE_ON_QUEUE_SIZE: + self._paused_producer = True + self._producer.pauseProducing() + + def _writer(self): + """This is run in a background thread to write to the file. + """ + try: + while self._producer or not self._bytes_queue.empty(): + # If we've paused the producer check if we should resume the + # producer. + if self._producer and self._paused_producer: + if self._bytes_queue.qsize() <= self._RESUME_ON_QUEUE_SIZE: + self._reactor.callFromThread(self._resume_paused_producer) + + bytes = self._bytes_queue.get() + + # If we get a None (or empty list) then that's a signal used + # to indicate we should check if we should stop. + if bytes: + self._file_obj.write(bytes) + + # If its a pull producer then we need to explicitly ask for + # more stuff. + if not self.streaming and self._producer: + self._reactor.callFromThread(self._producer.resumeProducing) + except Exception as e: + self._write_exception = e + raise + finally: + self._file_obj.close() + + def wait(self): + """Returns a deferred that resolves when finished writing to file + """ + return make_deferred_yieldable(self._finished_deferred) + + def _resume_paused_producer(self): + """Gets called if we should resume producing after being paused + """ + if self._paused_producer and self._producer: + self._paused_producer = False + self._producer.resumeProducing() diff --git a/synapse/util/frozenutils.py b/synapse/util/frozenutils.py index 6322f0f55..581c6052a 100644 --- a/synapse/util/frozenutils.py +++ b/synapse/util/frozenutils.py @@ -13,18 +13,20 @@ # See the License for the specific language governing permissions and # limitations under the License. +from six import string_types + +from canonicaljson import json from frozendict import frozendict def freeze(o): - t = type(o) - if t is dict: + if isinstance(o, dict): return frozendict({k: freeze(v) for k, v in o.items()}) - if t is frozendict: + if isinstance(o, frozendict): return o - if t is str or t is unicode: + if isinstance(o, string_types): return o try: @@ -36,11 +38,10 @@ def freeze(o): def unfreeze(o): - t = type(o) - if t is dict or t is frozendict: + if isinstance(o, (dict, frozendict)): return dict({k: unfreeze(v) for k, v in o.items()}) - if t is str or t is unicode: + if isinstance(o, string_types): return o try: @@ -49,3 +50,21 @@ def unfreeze(o): pass return o + + +def _handle_frozendict(obj): + """Helper for EventEncoder. Makes frozendicts serializable by returning + the underlying dict + """ + if type(obj) is frozendict: + # fishing the protected dict out of the object is a bit nasty, + # but we don't really want the overhead of copying the dict. + return obj._dict + raise TypeError('Object of type %s is not JSON serializable' % + obj.__class__.__name__) + + +# A JSONEncoder which is capable of encoding frozendics without barfing +frozendict_json_encoder = json.JSONEncoder( + default=_handle_frozendict, +) diff --git a/synapse/util/httpresourcetree.py b/synapse/util/httpresourcetree.py index 45be47159..2d7ddc1cb 100644 --- a/synapse/util/httpresourcetree.py +++ b/synapse/util/httpresourcetree.py @@ -12,10 +12,10 @@ # See the License for the specific language governing permissions and # limitations under the License. -from twisted.web.resource import Resource - import logging +from twisted.web.resource import NoResource + logger = logging.getLogger(__name__) @@ -40,12 +40,15 @@ def create_resource_tree(desired_tree, root_resource): # extra resources to existing nodes. See self._resource_id for the key. resource_mappings = {} for full_path, res in desired_tree.items(): + # twisted requires all resources to be bytes + full_path = full_path.encode("utf-8") + logger.info("Attaching %s to path %s", res, full_path) last_resource = root_resource - for path_seg in full_path.split('/')[1:-1]: + for path_seg in full_path.split(b'/')[1:-1]: if path_seg not in last_resource.listNames(): # resource doesn't exist, so make a "dummy resource" - child_resource = Resource() + child_resource = NoResource() last_resource.putChild(path_seg, child_resource) res_id = _resource_id(last_resource, path_seg) resource_mappings[res_id] = child_resource @@ -57,7 +60,7 @@ def create_resource_tree(desired_tree, root_resource): # =========================== # now attach the actual desired resource - last_path_seg = full_path.split('/')[-1] + last_path_seg = full_path.split(b'/')[-1] # if there is already a resource here, thieve its children and # replace it diff --git a/synapse/util/logcontext.py b/synapse/util/logcontext.py index 990216145..8dcae50b3 100644 --- a/synapse/util/logcontext.py +++ b/synapse/util/logcontext.py @@ -22,10 +22,10 @@ them. See doc/log_contexts.rst for details on how this works. """ -from twisted.internet import defer - -import threading import logging +import threading + +from twisted.internet import defer logger = logging.getLogger(__name__) @@ -42,23 +42,128 @@ try: def get_thread_resource_usage(): return resource.getrusage(RUSAGE_THREAD) -except: +except Exception: # If the system doesn't support resource.getrusage(RUSAGE_THREAD) then we # won't track resource usage by returning None. def get_thread_resource_usage(): return None -class LoggingContext(object): - """Additional context for log formatting. Contexts are scoped within a - "with" block. - Args: - name (str): Name for the context for debugging. +class ContextResourceUsage(object): + """Object for tracking the resources used by a log context + + Attributes: + ru_utime (float): user CPU time (in seconds) + ru_stime (float): system CPU time (in seconds) + db_txn_count (int): number of database transactions done + db_sched_duration_sec (float): amount of time spent waiting for a + database connection + db_txn_duration_sec (float): amount of time spent doing database + transactions (excluding scheduling time) + evt_db_fetch_count (int): number of events requested from the database """ __slots__ = [ - "previous_context", "name", "usage_start", "usage_end", "main_thread", - "__dict__", "tag", "alive", + "ru_stime", "ru_utime", + "db_txn_count", "db_txn_duration_sec", "db_sched_duration_sec", + "evt_db_fetch_count", + ] + + def __init__(self, copy_from=None): + """Create a new ContextResourceUsage + + Args: + copy_from (ContextResourceUsage|None): if not None, an object to + copy stats from + """ + if copy_from is None: + self.reset() + else: + self.ru_utime = copy_from.ru_utime + self.ru_stime = copy_from.ru_stime + self.db_txn_count = copy_from.db_txn_count + + self.db_txn_duration_sec = copy_from.db_txn_duration_sec + self.db_sched_duration_sec = copy_from.db_sched_duration_sec + self.evt_db_fetch_count = copy_from.evt_db_fetch_count + + def copy(self): + return ContextResourceUsage(copy_from=self) + + def reset(self): + self.ru_stime = 0. + self.ru_utime = 0. + self.db_txn_count = 0 + + self.db_txn_duration_sec = 0 + self.db_sched_duration_sec = 0 + self.evt_db_fetch_count = 0 + + def __repr__(self): + return ("") % ( + self.ru_stime, + self.ru_utime, + self.db_txn_count, + self.db_txn_duration_sec, + self.db_sched_duration_sec, + self.evt_db_fetch_count,) + + def __iadd__(self, other): + """Add another ContextResourceUsage's stats to this one's. + + Args: + other (ContextResourceUsage): the other resource usage object + """ + self.ru_utime += other.ru_utime + self.ru_stime += other.ru_stime + self.db_txn_count += other.db_txn_count + self.db_txn_duration_sec += other.db_txn_duration_sec + self.db_sched_duration_sec += other.db_sched_duration_sec + self.evt_db_fetch_count += other.evt_db_fetch_count + return self + + def __isub__(self, other): + self.ru_utime -= other.ru_utime + self.ru_stime -= other.ru_stime + self.db_txn_count -= other.db_txn_count + self.db_txn_duration_sec -= other.db_txn_duration_sec + self.db_sched_duration_sec -= other.db_sched_duration_sec + self.evt_db_fetch_count -= other.evt_db_fetch_count + return self + + def __add__(self, other): + res = ContextResourceUsage(copy_from=self) + res += other + return res + + def __sub__(self, other): + res = ContextResourceUsage(copy_from=self) + res -= other + return res + + +class LoggingContext(object): + """Additional context for log formatting. Contexts are scoped within a + "with" block. + + If a parent is given when creating a new context, then: + - logging fields are copied from the parent to the new context on entry + - when the new context exits, the cpu usage stats are copied from the + child to the parent + + Args: + name (str): Name for the context for debugging. + parent_context (LoggingContext|None): The parent of the new context + """ + + __slots__ = [ + "previous_context", "name", "parent_context", + "_resource_usage", + "usage_start", + "main_thread", "alive", + "request", "tag", ] thread_local = threading.local() @@ -80,32 +185,49 @@ class LoggingContext(object): def stop(self): pass - def add_database_transaction(self, duration_ms): + def add_database_transaction(self, duration_sec): + pass + + def add_database_scheduled(self, sched_sec): + pass + + def record_event_fetch(self, event_count): pass def __nonzero__(self): return False + __bool__ = __nonzero__ # python3 sentinel = Sentinel() - def __init__(self, name=None): + def __init__(self, name=None, parent_context=None): self.previous_context = LoggingContext.current_context() self.name = name - self.ru_stime = 0. - self.ru_utime = 0. - self.db_txn_count = 0 - self.db_txn_duration = 0. + + # track the resources used by this context so far + self._resource_usage = ContextResourceUsage() + + # If alive has the thread resource usage when the logcontext last + # became active. self.usage_start = None + self.main_thread = threading.current_thread() + self.request = None self.tag = "" self.alive = True + self.parent_context = parent_context + def __str__(self): return "%s@%x" % (self.name, id(self)) @classmethod def current_context(cls): - """Get the current logging context from thread local storage""" + """Get the current logging context from thread local storage + + Returns: + LoggingContext: the current logging context + """ return getattr(cls.thread_local, "current_context", cls.sentinel) @classmethod @@ -133,18 +255,22 @@ class LoggingContext(object): self.previous_context, old_context ) self.alive = True + + if self.parent_context is not None: + self.parent_context.copy_to(self) + return self def __exit__(self, type, value, traceback): """Restore the logging context in thread local storage to the state it was before this context was entered. Returns: - None to avoid suppressing any exeptions that were thrown. + None to avoid suppressing any exceptions that were thrown. """ current = self.set_current_context(self.previous_context) if current is not self: if current is self.sentinel: - logger.debug("Expected logging context %s has been lost", self) + logger.warn("Expected logging context %s has been lost", self) else: logger.warn( "Current logging context %s is not expected context %s", @@ -154,47 +280,91 @@ class LoggingContext(object): self.previous_context = None self.alive = False - def copy_to(self, record): - """Copy fields from this context to the record""" - for key, value in self.__dict__.items(): - setattr(record, key, value) + # if we have a parent, pass our CPU usage stats on + if self.parent_context is not None: + self.parent_context._resource_usage += self._resource_usage - record.ru_utime, record.ru_stime = self.get_resource_usage() + # reset them in case we get entered again + self._resource_usage.reset() + + def copy_to(self, record): + """Copy logging fields from this context to a log record or + another LoggingContext + """ + + # 'request' is the only field we currently use in the logger, so that's + # all we need to copy + record.request = self.request def start(self): if threading.current_thread() is not self.main_thread: + logger.warning("Started logcontext %s on different thread", self) return - if self.usage_start and self.usage_end: - self.ru_utime += self.usage_end.ru_utime - self.usage_start.ru_utime - self.ru_stime += self.usage_end.ru_stime - self.usage_start.ru_stime - self.usage_start = None - self.usage_end = None - + # If we haven't already started record the thread resource usage so + # far if not self.usage_start: self.usage_start = get_thread_resource_usage() def stop(self): if threading.current_thread() is not self.main_thread: + logger.warning("Stopped logcontext %s on different thread", self) return - if self.usage_start: - self.usage_end = get_thread_resource_usage() + # When we stop, let's record the cpu used since we started + if not self.usage_start: + logger.warning( + "Called stop on logcontext %s without calling start", self, + ) + return + + usage_end = get_thread_resource_usage() + + self._resource_usage.ru_utime += usage_end.ru_utime - self.usage_start.ru_utime + self._resource_usage.ru_stime += usage_end.ru_stime - self.usage_start.ru_stime + + self.usage_start = None def get_resource_usage(self): - ru_utime = self.ru_utime - ru_stime = self.ru_stime + """Get resources used by this logcontext so far. - if self.usage_start and threading.current_thread() is self.main_thread: + Returns: + ContextResourceUsage: a *copy* of the object tracking resource + usage so far + """ + # we always return a copy, for consistency + res = self._resource_usage.copy() + + # If we are on the correct thread and we're currently running then we + # can include resource usage so far. + is_main_thread = threading.current_thread() is self.main_thread + if self.alive and self.usage_start and is_main_thread: current = get_thread_resource_usage() - ru_utime += current.ru_utime - self.usage_start.ru_utime - ru_stime += current.ru_stime - self.usage_start.ru_stime + res.ru_utime += current.ru_utime - self.usage_start.ru_utime + res.ru_stime += current.ru_stime - self.usage_start.ru_stime - return ru_utime, ru_stime + return res - def add_database_transaction(self, duration_ms): - self.db_txn_count += 1 - self.db_txn_duration += duration_ms / 1000. + def add_database_transaction(self, duration_sec): + self._resource_usage.db_txn_count += 1 + self._resource_usage.db_txn_duration_sec += duration_sec + + def add_database_scheduled(self, sched_sec): + """Record a use of the database pool + + Args: + sched_sec (float): number of seconds it took us to get a + connection + """ + self._resource_usage.db_sched_duration_sec += sched_sec + + def record_event_fetch(self, event_count): + """Record a number of events being fetched from the db + + Args: + event_count (int): number of events being fetched + """ + self._resource_usage.evt_db_fetch_count += event_count class LoggingContextFilter(logging.Filter): @@ -248,7 +418,7 @@ class PreserveLoggingContext(object): context = LoggingContext.set_current_context(self.current_context) if context != self.new_context: - logger.debug( + logger.warn( "Unexpected logging context: %s is not %s", context, self.new_context, ) @@ -261,105 +431,62 @@ class PreserveLoggingContext(object): ) -class _PreservingContextDeferred(defer.Deferred): - """A deferred that ensures that all callbacks and errbacks are called with - the given logging context. - """ - def __init__(self, context): - self._log_context = context - defer.Deferred.__init__(self) - - def addCallbacks(self, callback, errback=None, - callbackArgs=None, callbackKeywords=None, - errbackArgs=None, errbackKeywords=None): - callback = self._wrap_callback(callback) - errback = self._wrap_callback(errback) - return defer.Deferred.addCallbacks( - self, callback, - errback=errback, - callbackArgs=callbackArgs, - callbackKeywords=callbackKeywords, - errbackArgs=errbackArgs, - errbackKeywords=errbackKeywords, - ) - - def _wrap_callback(self, f): - def g(res, *args, **kwargs): - with PreserveLoggingContext(self._log_context): - res = f(res, *args, **kwargs) - return res - return g - - -def preserve_context_over_fn(fn, *args, **kwargs): - """Takes a function and invokes it with the given arguments, but removes - and restores the current logging context while doing so. - - If the result is a deferred, call preserve_context_over_deferred before - returning it. - """ - with PreserveLoggingContext(): - res = fn(*args, **kwargs) - - if isinstance(res, defer.Deferred): - return preserve_context_over_deferred(res) - else: - return res - - -def preserve_context_over_deferred(deferred, context=None): - """Given a deferred wrap it such that any callbacks added later to it will - be invoked with the current context. - - Deprecated: this almost certainly doesn't do want you want, ie make - the deferred follow the synapse logcontext rules: try - ``make_deferred_yieldable`` instead. - """ - if context is None: - context = LoggingContext.current_context() - d = _PreservingContextDeferred(context) - deferred.chainDeferred(d) - return d - - def preserve_fn(f): - """Wraps a function, to ensure that the current context is restored after - return from the function, and that the sentinel context is set once the - deferred returned by the funtion completes. - - Useful for wrapping functions that return a deferred which you don't yield - on. - """ - def reset_context(result): - LoggingContext.set_current_context(LoggingContext.sentinel) - return result - + """Function decorator which wraps the function with run_in_background""" def g(*args, **kwargs): - current = LoggingContext.current_context() - res = f(*args, **kwargs) - if isinstance(res, defer.Deferred) and not res.called: - # The function will have reset the context before returning, so - # we need to restore it now. - LoggingContext.set_current_context(current) - - # The original context will be restored when the deferred - # completes, but there is nothing waiting for it, so it will - # get leaked into the reactor or some other function which - # wasn't expecting it. We therefore need to reset the context - # here. - # - # (If this feels asymmetric, consider it this way: we are - # effectively forking a new thread of execution. We are - # probably currently within a ``with LoggingContext()`` block, - # which is supposed to have a single entry and exit point. But - # by spawning off another deferred, we are effectively - # adding a new exit point.) - res.addBoth(reset_context) - return res + return run_in_background(f, *args, **kwargs) return g -@defer.inlineCallbacks +def run_in_background(f, *args, **kwargs): + """Calls a function, ensuring that the current context is restored after + return from the function, and that the sentinel context is set once the + deferred returned by the function completes. + + Useful for wrapping functions that return a deferred which you don't yield + on (for instance because you want to pass it to deferred.gatherResults()). + + Note that if you completely discard the result, you should make sure that + `f` doesn't raise any deferred exceptions, otherwise a scary-looking + CRITICAL error about an unhandled error will be logged without much + indication about where it came from. + """ + current = LoggingContext.current_context() + try: + res = f(*args, **kwargs) + except: # noqa: E722 + # the assumption here is that the caller doesn't want to be disturbed + # by synchronous exceptions, so let's turn them into Failures. + return defer.fail() + + if not isinstance(res, defer.Deferred): + return res + + if res.called and not res.paused: + # The function should have maintained the logcontext, so we can + # optimise out the messing about + return res + + # The function may have reset the context before returning, so + # we need to restore it now. + ctx = LoggingContext.set_current_context(current) + + # The original context will be restored when the deferred + # completes, but there is nothing waiting for it, so it will + # get leaked into the reactor or some other function which + # wasn't expecting it. We therefore need to reset the context + # here. + # + # (If this feels asymmetric, consider it this way: we are + # effectively forking a new thread of execution. We are + # probably currently within a ``with LoggingContext()`` block, + # which is supposed to have a single entry and exit point. But + # by spawning off another deferred, we are effectively + # adding a new exit point.) + res.addBoth(_set_context_cb, ctx) + return res + + def make_deferred_yieldable(deferred): """Given a deferred, make it follow the Synapse logcontext rules: @@ -371,11 +498,27 @@ def make_deferred_yieldable(deferred): returning a deferred. Then, when the deferred completes, restores the current logcontext before running callbacks/errbacks. - (This is more-or-less the opposite operation to preserve_fn.) + (This is more-or-less the opposite operation to run_in_background.) """ - with PreserveLoggingContext(): - r = yield deferred - defer.returnValue(r) + if not isinstance(deferred, defer.Deferred): + return deferred + + if deferred.called and not deferred.paused: + # it looks like this deferred is ready to run any callbacks we give it + # immediately. We may as well optimise out the logcontext faffery. + return deferred + + # ok, we can't be sure that a yield won't block, so let's reset the + # logcontext, and add a callback to the deferred to restore it. + prev_context = LoggingContext.set_current_context(LoggingContext.sentinel) + deferred.addBoth(_set_context_cb, prev_context) + return deferred + + +def _set_context_cb(result, context): + """A callback function which just sets the logging context""" + LoggingContext.set_current_context(context) + return result # modules to ignore in `logcontext_tracer` diff --git a/synapse/util/logformatter.py b/synapse/util/logformatter.py new file mode 100644 index 000000000..a46bc47ce --- /dev/null +++ b/synapse/util/logformatter.py @@ -0,0 +1,52 @@ +# -*- coding: utf-8 -*- +# Copyright 2017 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + + +import logging +import traceback + +from six import StringIO + + +class LogFormatter(logging.Formatter): + """Log formatter which gives more detail for exceptions + + This is the same as the standard log formatter, except that when logging + exceptions [typically via log.foo("msg", exc_info=1)], it prints the + sequence that led up to the point at which the exception was caught. + (Normally only stack frames between the point the exception was raised and + where it was caught are logged). + """ + def __init__(self, *args, **kwargs): + super(LogFormatter, self).__init__(*args, **kwargs) + + def formatException(self, ei): + sio = StringIO() + (typ, val, tb) = ei + + # log the stack above the exception capture point if possible, but + # check that we actually have an f_back attribute to work around + # https://twistedmatrix.com/trac/ticket/9305 + + if tb and hasattr(tb.tb_frame, 'f_back'): + sio.write("Capture point (most recent call last):\n") + traceback.print_stack(tb.tb_frame.f_back, None, sio) + + traceback.print_exception(typ, val, tb, None, sio) + s = sio.getvalue() + sio.close() + if s[-1:] == "\n": + s = s[:-1] + return s diff --git a/synapse/util/logutils.py b/synapse/util/logutils.py index 3a83828d2..62a00189c 100644 --- a/synapse/util/logutils.py +++ b/synapse/util/logutils.py @@ -14,13 +14,11 @@ # limitations under the License. -from inspect import getcallargs -from functools import wraps - -import logging import inspect +import logging import time - +from functools import wraps +from inspect import getcallargs _TIME_FUNC_ID = 0 @@ -96,7 +94,7 @@ def time_function(f): id = _TIME_FUNC_ID _TIME_FUNC_ID += 1 - start = time.clock() * 1000 + start = time.clock() try: _log_debug_as_f( @@ -107,10 +105,10 @@ def time_function(f): r = f(*args, **kwargs) finally: - end = time.clock() * 1000 + end = time.clock() _log_debug_as_f( f, - "[FUNC END] {%s-%d} %f", + "[FUNC END] {%s-%d} %.3f sec", (func_name, id, end - start,), ) diff --git a/synapse/util/manhole.py b/synapse/util/manhole.py index 97e0f00b6..14be3c739 100644 --- a/synapse/util/manhole.py +++ b/synapse/util/manhole.py @@ -12,11 +12,11 @@ # See the License for the specific language governing permissions and # limitations under the License. -from twisted.conch.manhole import ColoredManhole -from twisted.conch.insults import insults from twisted.conch import manhole_ssh -from twisted.cred import checkers, portal +from twisted.conch.insults import insults +from twisted.conch.manhole import ColoredManhole from twisted.conch.ssh.keys import Key +from twisted.cred import checkers, portal PUBLIC_KEY = ( "ssh-rsa AAAAB3NzaC1yc2EAAAABIwAAAGEArzJx8OYOnJmzf4tfBEvLi8DVPrJ3/c9k2I/Az" diff --git a/synapse/util/metrics.py b/synapse/util/metrics.py index 4ea930d3e..97f126738 100644 --- a/synapse/util/metrics.py +++ b/synapse/util/metrics.py @@ -13,40 +13,37 @@ # See the License for the specific language governing permissions and # limitations under the License. +import logging +from functools import wraps + +from prometheus_client import Counter + from twisted.internet import defer from synapse.util.logcontext import LoggingContext -import synapse.metrics - -from functools import wraps -import logging - logger = logging.getLogger(__name__) +block_counter = Counter("synapse_util_metrics_block_count", "", ["block_name"]) -metrics = synapse.metrics.get_metrics_for(__name__) +block_timer = Counter("synapse_util_metrics_block_time_seconds", "", ["block_name"]) -block_timer = metrics.register_distribution( - "block_timer", - labels=["block_name"] -) +block_ru_utime = Counter( + "synapse_util_metrics_block_ru_utime_seconds", "", ["block_name"]) -block_ru_utime = metrics.register_distribution( - "block_ru_utime", labels=["block_name"] -) +block_ru_stime = Counter( + "synapse_util_metrics_block_ru_stime_seconds", "", ["block_name"]) -block_ru_stime = metrics.register_distribution( - "block_ru_stime", labels=["block_name"] -) +block_db_txn_count = Counter( + "synapse_util_metrics_block_db_txn_count", "", ["block_name"]) -block_db_txn_count = metrics.register_distribution( - "block_db_txn_count", labels=["block_name"] -) +# seconds spent waiting for db txns, excluding scheduling time, in this block +block_db_txn_duration = Counter( + "synapse_util_metrics_block_db_txn_duration_seconds", "", ["block_name"]) -block_db_txn_duration = metrics.register_distribution( - "block_db_txn_duration", labels=["block_name"] -) +# seconds spent waiting for a db connection, in this block +block_db_sched_duration = Counter( + "synapse_util_metrics_block_db_sched_duration_seconds", "", ["block_name"]) def measure_func(name): @@ -63,8 +60,9 @@ def measure_func(name): class Measure(object): __slots__ = [ - "clock", "name", "start_context", "start", "new_context", "ru_utime", - "ru_stime", "db_txn_count", "db_txn_duration", "created_context" + "clock", "name", "start_context", "start", + "created_context", + "start_usage", ] def __init__(self, clock, name): @@ -75,23 +73,23 @@ class Measure(object): self.created_context = False def __enter__(self): - self.start = self.clock.time_msec() + self.start = self.clock.time() self.start_context = LoggingContext.current_context() if not self.start_context: self.start_context = LoggingContext("Measure") self.start_context.__enter__() self.created_context = True - self.ru_utime, self.ru_stime = self.start_context.get_resource_usage() - self.db_txn_count = self.start_context.db_txn_count - self.db_txn_duration = self.start_context.db_txn_duration + self.start_usage = self.start_context.get_resource_usage() def __exit__(self, exc_type, exc_val, exc_tb): if isinstance(exc_type, Exception) or not self.start_context: return - duration = self.clock.time_msec() - self.start - block_timer.inc_by(duration, self.name) + duration = self.clock.time() - self.start + + block_counter.labels(self.name).inc() + block_timer.labels(self.name).inc(duration) context = LoggingContext.current_context() @@ -106,16 +104,19 @@ class Measure(object): logger.warn("Expected context. (%r)", self.name) return - ru_utime, ru_stime = context.get_resource_usage() - - block_ru_utime.inc_by(ru_utime - self.ru_utime, self.name) - block_ru_stime.inc_by(ru_stime - self.ru_stime, self.name) - block_db_txn_count.inc_by( - context.db_txn_count - self.db_txn_count, self.name - ) - block_db_txn_duration.inc_by( - context.db_txn_duration - self.db_txn_duration, self.name - ) + current = context.get_resource_usage() + usage = current - self.start_usage + try: + block_ru_utime.labels(self.name).inc(usage.ru_utime) + block_ru_stime.labels(self.name).inc(usage.ru_stime) + block_db_txn_count.labels(self.name).inc(usage.db_txn_count) + block_db_txn_duration.labels(self.name).inc(usage.db_txn_duration_sec) + block_db_sched_duration.labels(self.name).inc(usage.db_sched_duration_sec) + except ValueError: + logger.warn( + "Failed to save metrics! OLD: %r, NEW: %r", + self.start_usage, current + ) if self.created_context: self.start_context.__exit__(exc_type, exc_val, exc_tb) diff --git a/synapse/util/module_loader.py b/synapse/util/module_loader.py new file mode 100644 index 000000000..4288312b8 --- /dev/null +++ b/synapse/util/module_loader.py @@ -0,0 +1,42 @@ +# -*- coding: utf-8 -*- +# Copyright 2017 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import importlib + +from synapse.config._base import ConfigError + + +def load_module(provider): + """ Loads a module with its config + Take a dict with keys 'module' (the module name) and 'config' + (the config dict). + + Returns + Tuple of (provider class, parsed config object) + """ + # We need to import the module, and then pick the class out of + # that, so we split based on the last dot. + module, clz = provider['module'].rsplit(".", 1) + module = importlib.import_module(module) + provider_class = getattr(module, clz) + + try: + provider_config = provider_class.parse_config(provider["config"]) + except Exception as e: + raise ConfigError( + "Failed to parse config for %r: %r" % (provider['module'], e) + ) + + return provider_class, provider_config diff --git a/synapse/util/msisdn.py b/synapse/util/msisdn.py index 607161e7f..a6c30e526 100644 --- a/synapse/util/msisdn.py +++ b/synapse/util/msisdn.py @@ -14,6 +14,7 @@ # limitations under the License. import phonenumbers + from synapse.api.errors import SynapseError diff --git a/synapse/util/ratelimitutils.py b/synapse/util/ratelimitutils.py index 1101881a2..7deb38f2a 100644 --- a/synapse/util/ratelimitutils.py +++ b/synapse/util/ratelimitutils.py @@ -13,17 +13,18 @@ # See the License for the specific language governing permissions and # limitations under the License. -from twisted.internet import defer - -from synapse.api.errors import LimitExceededError - -from synapse.util.async import sleep -from synapse.util.logcontext import preserve_fn - import collections import contextlib import logging +from twisted.internet import defer + +from synapse.api.errors import LimitExceededError +from synapse.util.logcontext import ( + PreserveLoggingContext, + make_deferred_yieldable, + run_in_background, +) logger = logging.getLogger(__name__) @@ -91,13 +92,22 @@ class _PerHostRatelimiter(object): self.window_size = window_size self.sleep_limit = sleep_limit - self.sleep_msec = sleep_msec + self.sleep_sec = sleep_msec / 1000.0 self.reject_limit = reject_limit self.concurrent_requests = concurrent_requests + # request_id objects for requests which have been slept self.sleeping_requests = set() + + # map from request_id object to Deferred for requests which are ready + # for processing but have been queued self.ready_request_queue = collections.OrderedDict() + + # request id objects for requests which are in progress self.current_processing = set() + + # times at which we have recently (within the last window_size ms) + # received requests. self.request_times = [] @contextlib.contextmanager @@ -116,11 +126,15 @@ class _PerHostRatelimiter(object): def _on_enter(self, request_id): time_now = self.clock.time_msec() + + # remove any entries from request_times which aren't within the window self.request_times[:] = [ r for r in self.request_times if time_now - r < self.window_size ] + # reject the request if we already have too many queued up (either + # sleeping or in the ready queue). queue_size = len(self.ready_request_queue) + len(self.sleeping_requests) if queue_size > self.reject_limit: raise LimitExceededError( @@ -133,9 +147,13 @@ class _PerHostRatelimiter(object): def queue_request(): if len(self.current_processing) > self.concurrent_requests: - logger.debug("Ratelimit [%s]: Queue req", id(request_id)) queue_defer = defer.Deferred() self.ready_request_queue[request_id] = queue_defer + logger.info( + "Ratelimiter: queueing request (queue now %i items)", + len(self.ready_request_queue), + ) + return queue_defer else: return defer.succeed(None) @@ -147,10 +165,9 @@ class _PerHostRatelimiter(object): if len(self.request_times) > self.sleep_limit: logger.debug( - "Ratelimit [%s]: sleeping req", - id(request_id), + "Ratelimiter: sleeping request for %f sec", self.sleep_sec, ) - ret_defer = preserve_fn(sleep)(self.sleep_msec / 1000.0) + ret_defer = run_in_background(self.clock.sleep, self.sleep_sec) self.sleeping_requests.add(request_id) @@ -176,6 +193,9 @@ class _PerHostRatelimiter(object): return r def on_err(r): + # XXX: why is this necessary? this is called before we start + # processing the request so why would the request be in + # current_processing? self.current_processing.discard(request_id) return r @@ -187,7 +207,7 @@ class _PerHostRatelimiter(object): ret_defer.addCallbacks(on_start, on_err) ret_defer.addBoth(on_both) - return ret_defer + return make_deferred_yieldable(ret_defer) def _on_exit(self, request_id): logger.debug( @@ -196,8 +216,10 @@ class _PerHostRatelimiter(object): ) self.current_processing.discard(request_id) try: - request_id, deferred = self.ready_request_queue.popitem() - self.current_processing.add(request_id) - deferred.callback(None) + # start processing the next item on the queue. + _, deferred = self.ready_request_queue.popitem(last=False) + + with PreserveLoggingContext(): + deferred.callback(None) except KeyError: pass diff --git a/synapse/util/retryutils.py b/synapse/util/retryutils.py index 4fa9d1a03..8a3a06fd7 100644 --- a/synapse/util/retryutils.py +++ b/synapse/util/retryutils.py @@ -12,20 +12,31 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. -import synapse.util.logcontext -from twisted.internet import defer - -from synapse.api.errors import CodeMessageException - import logging import random +from twisted.internet import defer + +import synapse.util.logcontext +from synapse.api.errors import CodeMessageException logger = logging.getLogger(__name__) class NotRetryingDestination(Exception): def __init__(self, retry_last_ts, retry_interval, destination): + """Raised by the limiter (and federation client) to indicate that we are + are deliberately not attempting to contact a given server. + + Args: + retry_last_ts (int): the unix ts in milliseconds of our last attempt + to contact the server. 0 indicates that the last attempt was + successful or that we've never actually attempted to connect. + retry_interval (int): the time in milliseconds to wait until the next + attempt. + destination (str): the domain in question + """ + msg = "Not retrying server %s." % (destination,) super(NotRetryingDestination, self).__init__(msg) @@ -189,10 +200,10 @@ class RetryDestinationLimiter(object): yield self.store.set_destination_retry_timings( self.destination, retry_last_ts, self.retry_interval ) - except: + except Exception: logger.exception( - "Failed to store set_destination_retry_timings", + "Failed to store destination_retry_timings", ) # we deliberately do this in the background. - synapse.util.logcontext.preserve_fn(store_retry_timings)() + synapse.util.logcontext.run_in_background(store_retry_timings) diff --git a/synapse/util/rlimit.py b/synapse/util/rlimit.py index f4a9abf83..6c0f2bb0c 100644 --- a/synapse/util/rlimit.py +++ b/synapse/util/rlimit.py @@ -13,9 +13,8 @@ # See the License for the specific language governing permissions and # limitations under the License. -import resource import logging - +import resource logger = logging.getLogger("synapse.app.homeserver") diff --git a/synapse/util/stringutils.py b/synapse/util/stringutils.py index 95a6168e1..43d9db67e 100644 --- a/synapse/util/stringutils.py +++ b/synapse/util/stringutils.py @@ -16,18 +16,20 @@ import random import string +from six.moves import range + _string_with_symbols = ( string.digits + string.ascii_letters + ".,;:^&*-_+=#~@" ) def random_string(length): - return ''.join(random.choice(string.ascii_letters) for _ in xrange(length)) + return ''.join(random.choice(string.ascii_letters) for _ in range(length)) def random_string_with_symbols(length): return ''.join( - random.choice(_string_with_symbols) for _ in xrange(length) + random.choice(_string_with_symbols) for _ in range(length) ) diff --git a/synapse/util/threepids.py b/synapse/util/threepids.py new file mode 100644 index 000000000..75efa0117 --- /dev/null +++ b/synapse/util/threepids.py @@ -0,0 +1,48 @@ +# -*- coding: utf-8 -*- +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging +import re + +logger = logging.getLogger(__name__) + + +def check_3pid_allowed(hs, medium, address): + """Checks whether a given format of 3PID is allowed to be used on this HS + + Args: + hs (synapse.server.HomeServer): server + medium (str): 3pid medium - e.g. email, msisdn + address (str): address within that medium (e.g. "wotan@matrix.org") + msisdns need to first have been canonicalised + Returns: + bool: whether the 3PID medium/address is allowed to be added to this HS + """ + + if hs.config.allowed_local_3pids: + for constraint in hs.config.allowed_local_3pids: + logger.debug( + "Checking 3PID %s (%s) against %s (%s)", + address, medium, constraint['pattern'], constraint['medium'], + ) + if ( + medium == constraint['medium'] and + re.match(constraint['pattern'], address) + ): + return True + else: + return True + + return False diff --git a/synapse/util/versionstring.py b/synapse/util/versionstring.py index 52086df46..1fbcd4111 100644 --- a/synapse/util/versionstring.py +++ b/synapse/util/versionstring.py @@ -14,9 +14,9 @@ # See the License for the specific language governing permissions and # limitations under the License. -import subprocess -import os import logging +import os +import subprocess logger = logging.getLogger(__name__) diff --git a/synapse/util/wheel_timer.py b/synapse/util/wheel_timer.py index 7412fc57a..7a9e45aca 100644 --- a/synapse/util/wheel_timer.py +++ b/synapse/util/wheel_timer.py @@ -13,6 +13,8 @@ # See the License for the specific language governing permissions and # limitations under the License. +from six.moves import range + class _Entry(object): __slots__ = ["end_key", "queue"] @@ -68,7 +70,7 @@ class WheelTimer(object): # Add empty entries between the end of the current list and when we want # to insert. This ensures there are no gaps. self.entries.extend( - _Entry(key) for key in xrange(last_key, then_key + 1) + _Entry(key) for key in range(last_key, then_key + 1) ) self.entries[-1].queue.append(obj) @@ -91,7 +93,4 @@ class WheelTimer(object): return ret def __len__(self): - l = 0 - for entry in self.entries: - l += len(entry.queue) - return l + return sum(len(entry.queue) for entry in self.entries) diff --git a/synapse/visibility.py b/synapse/visibility.py index c4dd9ae2c..d4680863d 100644 --- a/synapse/visibility.py +++ b/synapse/visibility.py @@ -13,14 +13,17 @@ # See the License for the specific language governing permissions and # limitations under the License. +import logging +import operator + +from six import iteritems, itervalues +from six.moves import map + from twisted.internet import defer -from synapse.api.constants import Membership, EventTypes - -from synapse.util.logcontext import preserve_fn, preserve_context_over_deferred - -import logging - +from synapse.api.constants import EventTypes, Membership +from synapse.events.utils import prune_event +from synapse.types import get_domain_from_id logger = logging.getLogger(__name__) @@ -43,53 +46,66 @@ MEMBERSHIP_PRIORITY = ( @defer.inlineCallbacks -def filter_events_for_clients(store, user_tuples, events, event_id_to_state): - """ Returns dict of user_id -> list of events that user is allowed to - see. +def filter_events_for_client(store, user_id, events, is_peeking=False, + always_include_ids=frozenset()): + """ + Check which events a user is allowed to see Args: - user_tuples (str, bool): (user id, is_peeking) for each user to be - checked. is_peeking should be true if: - * the user is not currently a member of the room, and: - * the user has not been a member of the room since the - given events - events ([synapse.events.EventBase]): list of events to filter - """ - forgotten = yield preserve_context_over_deferred(defer.gatherResults([ - defer.maybeDeferred( - preserve_fn(store.who_forgot_in_room), - room_id, - ) - for room_id in frozenset(e.room_id for e in events) - ], consumeErrors=True)) + store (synapse.storage.DataStore): our datastore (can also be a worker + store) + user_id(str): user id to be checked + events(list[synapse.events.EventBase]): sequence of events to be checked + is_peeking(bool): should be True if: + * the user is not currently a member of the room, and: + * the user has not been a member of the room since the given + events + always_include_ids (set(event_id)): set of event ids to specifically + include (unless sender is ignored) - # Set of membership event_ids that have been forgotten - event_id_forgotten = frozenset( - row["event_id"] for rows in forgotten for row in rows + Returns: + Deferred[list[synapse.events.EventBase]] + """ + types = ( + (EventTypes.RoomHistoryVisibility, ""), + (EventTypes.Member, user_id), + ) + event_id_to_state = yield store.get_state_for_events( + frozenset(e.event_id for e in events), + types=types, ) - ignore_dict_content = yield store.get_global_account_data_by_type_for_users( - "m.ignored_user_list", user_ids=[user_id for user_id, _ in user_tuples] + ignore_dict_content = yield store.get_global_account_data_by_type_for_user( + "m.ignored_user_list", user_id, ) # FIXME: This will explode if people upload something incorrect. - ignore_dict = { - user_id: frozenset( - content.get("ignored_users", {}).keys() if content else [] - ) - for user_id, content in ignore_dict_content.items() - } + ignore_list = frozenset( + ignore_dict_content.get("ignored_users", {}).keys() + if ignore_dict_content else [] + ) - def allowed(event, user_id, is_peeking, ignore_list): + erased_senders = yield store.are_users_erased((e.sender for e in events)) + + def allowed(event): """ Args: event (synapse.events.EventBase): event to check - user_id (str) - is_peeking (bool) - ignore_list (list): list of users to ignore + + Returns: + None|EventBase: + None if the user cannot see this event at all + + a redacted copy of the event if they can only see a redacted + version + + the original event if they can see it as normal. """ if not event.is_state() and event.sender in ignore_list: - return False + return None + + if event.event_id in always_include_ids: + return event state = event_id_to_state[event.event_id] @@ -103,10 +119,6 @@ def filter_events_for_clients(store, user_tuples, events, event_id_to_state): if visibility not in VISIBILITY_PRIORITY: visibility = "shared" - # if it was world_readable, it's easy: everyone can read it - if visibility == "world_readable": - return True - # Always allow history visibility events on boundaries. This is done # by setting the effective visibility to the least restrictive # of the old vs new. @@ -140,7 +152,7 @@ def filter_events_for_clients(store, user_tuples, events, event_id_to_state): if membership == "leave" and ( prev_membership == "join" or prev_membership == "invite" ): - return True + return event new_priority = MEMBERSHIP_PRIORITY.index(membership) old_priority = MEMBERSHIP_PRIORITY.index(prev_membership) @@ -151,87 +163,203 @@ def filter_events_for_clients(store, user_tuples, events, event_id_to_state): if membership is None: membership_event = state.get((EventTypes.Member, user_id), None) if membership_event: - if membership_event.event_id not in event_id_forgotten: - membership = membership_event.membership + membership = membership_event.membership # if the user was a member of the room at the time of the event, # they can see it. if membership == Membership.JOIN: - return True + return event + + # otherwise, it depends on the room visibility. if visibility == "joined": # we weren't a member at the time of the event, so we can't # see this event. - return False + return None elif visibility == "invited": # user can also see the event if they were *invited* at the time # of the event. - return membership == Membership.INVITE + return ( + event if membership == Membership.INVITE else None + ) - else: - # visibility is shared: user can also see the event if they have - # become a member since the event + elif visibility == "shared" and is_peeking: + # if the visibility is shared, users cannot see the event unless + # they have *subequently* joined the room (or were members at the + # time, of course) # # XXX: if the user has subsequently joined and then left again, # ideally we would share history up to the point they left. But - # we don't know when they left. - return not is_peeking + # we don't know when they left. We just treat it as though they + # never joined, and restrict access. + return None - defer.returnValue({ - user_id: [ - event - for event in events - if allowed(event, user_id, is_peeking, ignore_dict.get(user_id, [])) - ] - for user_id, is_peeking in user_tuples - }) + # the visibility is either shared or world_readable, and the user was + # not a member at the time. We allow it, provided the original sender + # has not requested their data to be erased, in which case, we return + # a redacted version. + if erased_senders[event.sender]: + return prune_event(event) + + return event + + # check each event: gives an iterable[None|EventBase] + filtered_events = map(allowed, events) + + # remove the None entries + filtered_events = filter(operator.truth, filtered_events) + + # we turn it into a list before returning it. + defer.returnValue(list(filtered_events)) @defer.inlineCallbacks -def filter_events_for_clients_context(store, user_tuples, events, event_id_to_context): - user_ids = set(u[0] for u in user_tuples) - event_id_to_state = {} - for event_id, context in event_id_to_context.items(): - state = yield store.get_events([ - e_id - for key, e_id in context.current_state_ids.iteritems() - if key == (EventTypes.RoomHistoryVisibility, "") - or (key[0] == EventTypes.Member and key[1] in user_ids) - ]) - event_id_to_state[event_id] = state - - res = yield filter_events_for_clients( - store, user_tuples, events, event_id_to_state +def filter_events_for_server(store, server_name, events): + # Whatever else we do, we need to check for senders which have requested + # erasure of their data. + erased_senders = yield store.are_users_erased( + e.sender for e in events, ) - defer.returnValue(res) + def redact_disallowed(event, state): + # if the sender has been gdpr17ed, always return a redacted + # copy of the event. + if erased_senders[event.sender]: + logger.info( + "Sender of %s has been erased, redacting", + event.event_id, + ) + return prune_event(event) -@defer.inlineCallbacks -def filter_events_for_client(store, user_id, events, is_peeking=False): - """ - Check which events a user is allowed to see + # state will be None if we decided we didn't need to filter by + # room membership. + if not state: + return event - Args: - user_id(str): user id to be checked - events([synapse.events.EventBase]): list of events to be checked - is_peeking(bool): should be True if: - * the user is not currently a member of the room, and: - * the user has not been a member of the room since the given - events + history = state.get((EventTypes.RoomHistoryVisibility, ''), None) + if history: + visibility = history.content.get("history_visibility", "shared") + if visibility in ["invited", "joined"]: + # We now loop through all state events looking for + # membership states for the requesting server to determine + # if the server is either in the room or has been invited + # into the room. + for ev in itervalues(state): + if ev.type != EventTypes.Member: + continue + try: + domain = get_domain_from_id(ev.state_key) + except Exception: + continue - Returns: - [synapse.events.EventBase] - """ - types = ( - (EventTypes.RoomHistoryVisibility, ""), - (EventTypes.Member, user_id), - ) - event_id_to_state = yield store.get_state_for_events( + if domain != server_name: + continue + + memtype = ev.membership + if memtype == Membership.JOIN: + return event + elif memtype == Membership.INVITE: + if visibility == "invited": + return event + else: + # server has no users in the room: redact + return prune_event(event) + + return event + + # Next lets check to see if all the events have a history visibility + # of "shared" or "world_readable". If thats the case then we don't + # need to check membership (as we know the server is in the room). + event_to_state_ids = yield store.get_state_ids_for_events( frozenset(e.event_id for e in events), - types=types + types=( + (EventTypes.RoomHistoryVisibility, ""), + ) ) - res = yield filter_events_for_clients( - store, [(user_id, is_peeking)], events, event_id_to_state + + visibility_ids = set() + for sids in itervalues(event_to_state_ids): + hist = sids.get((EventTypes.RoomHistoryVisibility, "")) + if hist: + visibility_ids.add(hist) + + # If we failed to find any history visibility events then the default + # is "shared" visiblity. + if not visibility_ids: + all_open = True + else: + event_map = yield store.get_events(visibility_ids) + all_open = all( + e.content.get("history_visibility") in (None, "shared", "world_readable") + for e in itervalues(event_map) + ) + + if all_open: + # all the history_visibility state affecting these events is open, so + # we don't need to filter by membership state. We *do* need to check + # for user erasure, though. + if erased_senders: + events = [ + redact_disallowed(e, None) + for e in events + ] + + defer.returnValue(events) + + # Ok, so we're dealing with events that have non-trivial visibility + # rules, so we need to also get the memberships of the room. + + # first, for each event we're wanting to return, get the event_ids + # of the history vis and membership state at those events. + event_to_state_ids = yield store.get_state_ids_for_events( + frozenset(e.event_id for e in events), + types=( + (EventTypes.RoomHistoryVisibility, ""), + (EventTypes.Member, None), + ) ) - defer.returnValue(res.get(user_id, [])) + + # We only want to pull out member events that correspond to the + # server's domain. + # + # event_to_state_ids contains lots of duplicates, so it turns out to be + # cheaper to build a complete set of unique + # ((type, state_key), event_id) tuples, and then filter out the ones we + # don't want. + # + state_key_to_event_id_set = { + e + for key_to_eid in itervalues(event_to_state_ids) + for e in key_to_eid.items() + } + + def include(typ, state_key): + if typ != EventTypes.Member: + return True + + # we avoid using get_domain_from_id here for efficiency. + idx = state_key.find(":") + if idx == -1: + return False + return state_key[idx + 1:] == server_name + + event_map = yield store.get_events([ + e_id + for key, e_id in state_key_to_event_id_set + if include(key[0], key[1]) + ]) + + event_to_state = { + e_id: { + key: event_map[inner_e_id] + for key, inner_e_id in iteritems(key_to_eid) + if inner_e_id in event_map + } + for e_id, key_to_eid in iteritems(event_to_state_ids) + } + + defer.returnValue([ + redact_disallowed(e, event_to_state[e.event_id]) + for e in events + ]) diff --git a/tests/__init__.py b/tests/__init__.py index bfebb0f64..24006c949 100644 --- a/tests/__init__.py +++ b/tests/__init__.py @@ -12,3 +12,7 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. + +from twisted.trial import util + +util.DEFAULT_TIMEOUT_DURATION = 10 diff --git a/tests/api/test_auth.py b/tests/api/test_auth.py index 4575dd983..5f158ec4b 100644 --- a/tests/api/test_auth.py +++ b/tests/api/test_auth.py @@ -13,16 +13,19 @@ # See the License for the specific language governing permissions and # limitations under the License. -import pymacaroons from mock import Mock + +import pymacaroons + from twisted.internet import defer import synapse.handlers.auth from synapse.api.auth import Auth from synapse.api.errors import AuthError from synapse.types import UserID + from tests import unittest -from tests.utils import setup_test_homeserver, mock_getRawHeaders +from tests.utils import mock_getRawHeaders, setup_test_homeserver class TestHandlers(object): @@ -86,16 +89,53 @@ class AuthTestCase(unittest.TestCase): @defer.inlineCallbacks def test_get_user_by_req_appservice_valid_token(self): - app_service = Mock(token="foobar", url="a_url", sender=self.test_user) + app_service = Mock( + token="foobar", url="a_url", sender=self.test_user, + ip_range_whitelist=None, + ) self.store.get_app_service_by_token = Mock(return_value=app_service) self.store.get_user_by_access_token = Mock(return_value=None) request = Mock(args={}) + request.getClientIP.return_value = "127.0.0.1" request.args["access_token"] = [self.test_token] request.requestHeaders.getRawHeaders = mock_getRawHeaders() requester = yield self.auth.get_user_by_req(request) self.assertEquals(requester.user.to_string(), self.test_user) + @defer.inlineCallbacks + def test_get_user_by_req_appservice_valid_token_good_ip(self): + from netaddr import IPSet + app_service = Mock( + token="foobar", url="a_url", sender=self.test_user, + ip_range_whitelist=IPSet(["192.168/16"]), + ) + self.store.get_app_service_by_token = Mock(return_value=app_service) + self.store.get_user_by_access_token = Mock(return_value=None) + + request = Mock(args={}) + request.getClientIP.return_value = "192.168.10.10" + request.args["access_token"] = [self.test_token] + request.requestHeaders.getRawHeaders = mock_getRawHeaders() + requester = yield self.auth.get_user_by_req(request) + self.assertEquals(requester.user.to_string(), self.test_user) + + def test_get_user_by_req_appservice_valid_token_bad_ip(self): + from netaddr import IPSet + app_service = Mock( + token="foobar", url="a_url", sender=self.test_user, + ip_range_whitelist=IPSet(["192.168/16"]), + ) + self.store.get_app_service_by_token = Mock(return_value=app_service) + self.store.get_user_by_access_token = Mock(return_value=None) + + request = Mock(args={}) + request.getClientIP.return_value = "131.111.8.42" + request.args["access_token"] = [self.test_token] + request.requestHeaders.getRawHeaders = mock_getRawHeaders() + d = self.auth.get_user_by_req(request) + self.failureResultOf(d, AuthError) + def test_get_user_by_req_appservice_bad_token(self): self.store.get_app_service_by_token = Mock(return_value=None) self.store.get_user_by_access_token = Mock(return_value=None) @@ -119,12 +159,16 @@ class AuthTestCase(unittest.TestCase): @defer.inlineCallbacks def test_get_user_by_req_appservice_valid_token_valid_user_id(self): masquerading_user_id = "@doppelganger:matrix.org" - app_service = Mock(token="foobar", url="a_url", sender=self.test_user) + app_service = Mock( + token="foobar", url="a_url", sender=self.test_user, + ip_range_whitelist=None, + ) app_service.is_interested_in_user = Mock(return_value=True) self.store.get_app_service_by_token = Mock(return_value=app_service) self.store.get_user_by_access_token = Mock(return_value=None) request = Mock(args={}) + request.getClientIP.return_value = "127.0.0.1" request.args["access_token"] = [self.test_token] request.args["user_id"] = [masquerading_user_id] request.requestHeaders.getRawHeaders = mock_getRawHeaders() @@ -133,12 +177,16 @@ class AuthTestCase(unittest.TestCase): def test_get_user_by_req_appservice_valid_token_bad_user_id(self): masquerading_user_id = "@doppelganger:matrix.org" - app_service = Mock(token="foobar", url="a_url", sender=self.test_user) + app_service = Mock( + token="foobar", url="a_url", sender=self.test_user, + ip_range_whitelist=None, + ) app_service.is_interested_in_user = Mock(return_value=False) self.store.get_app_service_by_token = Mock(return_value=app_service) self.store.get_user_by_access_token = Mock(return_value=None) request = Mock(args={}) + request.getClientIP.return_value = "127.0.0.1" request.args["access_token"] = [self.test_token] request.args["user_id"] = [masquerading_user_id] request.requestHeaders.getRawHeaders = mock_getRawHeaders() diff --git a/tests/api/test_filtering.py b/tests/api/test_filtering.py index dcceca7f3..836a23fb5 100644 --- a/tests/api/test_filtering.py +++ b/tests/api/test_filtering.py @@ -13,20 +13,19 @@ # See the License for the specific language governing permissions and # limitations under the License. -from tests import unittest -from twisted.internet import defer - from mock import Mock -from tests.utils import ( - MockHttpResource, DeferredMockCallable, setup_test_homeserver -) - -from synapse.api.filtering import Filter -from synapse.events import FrozenEvent -from synapse.api.errors import SynapseError import jsonschema +from twisted.internet import defer + +from synapse.api.errors import SynapseError +from synapse.api.filtering import Filter +from synapse.events import FrozenEvent + +from tests import unittest +from tests.utils import DeferredMockCallable, MockHttpResource, setup_test_homeserver + user_localpart = "test_user" diff --git a/tests/appservice/test_appservice.py b/tests/appservice/test_appservice.py index 7586ea905..891e0cc97 100644 --- a/tests/appservice/test_appservice.py +++ b/tests/appservice/test_appservice.py @@ -12,14 +12,15 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. -from synapse.appservice import ApplicationService +import re + +from mock import Mock from twisted.internet import defer -from mock import Mock -from tests import unittest +from synapse.appservice import ApplicationService -import re +from tests import unittest def _regex(regex, exclusive=True): @@ -36,6 +37,7 @@ class ApplicationServiceTestCase(unittest.TestCase): id="unique_identifier", url="some_url", token="some_token", + hostname="matrix.org", # only used by get_groups_for_user namespaces={ ApplicationService.NS_USERS: [], ApplicationService.NS_ROOMS: [], diff --git a/tests/appservice/test_scheduler.py b/tests/appservice/test_scheduler.py index e5a902f73..b9f4863e9 100644 --- a/tests/appservice/test_scheduler.py +++ b/tests/appservice/test_scheduler.py @@ -12,15 +12,22 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. +from mock import Mock + +from twisted.internet import defer + from synapse.appservice import ApplicationServiceState from synapse.appservice.scheduler import ( - _ServiceQueuer, _TransactionController, _Recoverer + _Recoverer, + _ServiceQueuer, + _TransactionController, ) -from twisted.internet import defer -from ..utils import MockClock -from mock import Mock +from synapse.util.logcontext import make_deferred_yieldable + from tests import unittest +from ..utils import MockClock + class ApplicationServiceSchedulerTransactionCtrlTestCase(unittest.TestCase): @@ -204,7 +211,9 @@ class ApplicationServiceSchedulerQueuerTestCase(unittest.TestCase): def test_send_single_event_with_queue(self): d = defer.Deferred() - self.txn_ctrl.send = Mock(return_value=d) + self.txn_ctrl.send = Mock( + side_effect=lambda x, y: make_deferred_yieldable(d), + ) service = Mock(id=4) event = Mock(event_id="first") event2 = Mock(event_id="second") @@ -235,7 +244,10 @@ class ApplicationServiceSchedulerQueuerTestCase(unittest.TestCase): srv_2_event2 = Mock(event_id="srv2b") send_return_list = [srv_1_defer, srv_2_defer] - self.txn_ctrl.send = Mock(side_effect=lambda x, y: send_return_list.pop(0)) + + def do_send(x, y): + return make_deferred_yieldable(send_return_list.pop(0)) + self.txn_ctrl.send = Mock(side_effect=do_send) # send events for different ASes and make sure they are sent self.queuer.enqueue(srv1, srv_1_event) diff --git a/tests/config/test_generate.py b/tests/config/test_generate.py index 8f57fbeb2..eb7f0ab12 100644 --- a/tests/config/test_generate.py +++ b/tests/config/test_generate.py @@ -12,10 +12,14 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. + import os.path +import re import shutil import tempfile + from synapse.config.homeserver import HomeServerConfig + from tests import unittest @@ -23,7 +27,6 @@ class ConfigGenerationTestCase(unittest.TestCase): def setUp(self): self.dir = tempfile.mkdtemp() - print self.dir self.file = os.path.join(self.dir, "homeserver.yaml") def tearDown(self): @@ -48,3 +51,16 @@ class ConfigGenerationTestCase(unittest.TestCase): ]), set(os.listdir(self.dir)) ) + + self.assert_log_filename_is( + os.path.join(self.dir, "lemurs.win.log.config"), + os.path.join(os.getcwd(), "homeserver.log"), + ) + + def assert_log_filename_is(self, log_config_file, expected): + with open(log_config_file) as f: + config = f.read() + # find the 'filename' line + matches = re.findall("^\s*filename:\s*(.*)$", config, re.M) + self.assertEqual(1, len(matches)) + self.assertEqual(matches[0], expected) diff --git a/tests/config/test_load.py b/tests/config/test_load.py index 161a87d7e..5c422eff3 100644 --- a/tests/config/test_load.py +++ b/tests/config/test_load.py @@ -15,8 +15,11 @@ import os.path import shutil import tempfile + import yaml + from synapse.config.homeserver import HomeServerConfig + from tests import unittest @@ -24,7 +27,7 @@ class ConfigLoadingTestCase(unittest.TestCase): def setUp(self): self.dir = tempfile.mkdtemp() - print self.dir + print(self.dir) self.file = os.path.join(self.dir, "homeserver.yaml") def tearDown(self): diff --git a/tests/crypto/test_event_signing.py b/tests/crypto/test_event_signing.py index 47cb328a0..cd11871b8 100644 --- a/tests/crypto/test_event_signing.py +++ b/tests/crypto/test_event_signing.py @@ -14,15 +14,13 @@ # limitations under the License. -from tests import unittest - -from synapse.events.builder import EventBuilder -from synapse.crypto.event_signing import add_hashes_and_signatures - +import nacl.signing from unpaddedbase64 import decode_base64 -import nacl.signing +from synapse.crypto.event_signing import add_hashes_and_signatures +from synapse.events.builder import EventBuilder +from tests import unittest # Perform these tests using given secret key so we get entirely deterministic # signatures output that we can test against. diff --git a/tests/crypto/test_keyring.py b/tests/crypto/test_keyring.py new file mode 100644 index 000000000..a9d37fe08 --- /dev/null +++ b/tests/crypto/test_keyring.py @@ -0,0 +1,234 @@ +# -*- coding: utf-8 -*- +# Copyright 2017 New Vector Ltd. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import time + +from mock import Mock + +import signedjson.key +import signedjson.sign + +from twisted.internet import defer, reactor + +from synapse.api.errors import SynapseError +from synapse.crypto import keyring +from synapse.util import Clock, logcontext +from synapse.util.logcontext import LoggingContext + +from tests import unittest, utils + + +class MockPerspectiveServer(object): + def __init__(self): + self.server_name = "mock_server" + self.key = signedjson.key.generate_signing_key(0) + + def get_verify_keys(self): + vk = signedjson.key.get_verify_key(self.key) + return { + "%s:%s" % (vk.alg, vk.version): vk, + } + + def get_signed_key(self, server_name, verify_key): + key_id = "%s:%s" % (verify_key.alg, verify_key.version) + res = { + "server_name": server_name, + "old_verify_keys": {}, + "valid_until_ts": time.time() * 1000 + 3600, + "verify_keys": { + key_id: { + "key": signedjson.key.encode_verify_key_base64(verify_key) + } + } + } + signedjson.sign.sign_json(res, self.server_name, self.key) + return res + + +class KeyringTestCase(unittest.TestCase): + @defer.inlineCallbacks + def setUp(self): + self.mock_perspective_server = MockPerspectiveServer() + self.http_client = Mock() + self.hs = yield utils.setup_test_homeserver( + handlers=None, + http_client=self.http_client, + ) + self.hs.config.perspectives = { + self.mock_perspective_server.server_name: + self.mock_perspective_server.get_verify_keys() + } + + def check_context(self, _, expected): + self.assertEquals( + getattr(LoggingContext.current_context(), "request", None), + expected + ) + + @defer.inlineCallbacks + def test_wait_for_previous_lookups(self): + sentinel_context = LoggingContext.current_context() + + kr = keyring.Keyring(self.hs) + + lookup_1_deferred = defer.Deferred() + lookup_2_deferred = defer.Deferred() + + with LoggingContext("one") as context_one: + context_one.request = "one" + + wait_1_deferred = kr.wait_for_previous_lookups( + ["server1"], + {"server1": lookup_1_deferred}, + ) + + # there were no previous lookups, so the deferred should be ready + self.assertTrue(wait_1_deferred.called) + # ... so we should have preserved the LoggingContext. + self.assertIs(LoggingContext.current_context(), context_one) + wait_1_deferred.addBoth(self.check_context, "one") + + with LoggingContext("two") as context_two: + context_two.request = "two" + + # set off another wait. It should block because the first lookup + # hasn't yet completed. + wait_2_deferred = kr.wait_for_previous_lookups( + ["server1"], + {"server1": lookup_2_deferred}, + ) + self.assertFalse(wait_2_deferred.called) + # ... so we should have reset the LoggingContext. + self.assertIs(LoggingContext.current_context(), sentinel_context) + wait_2_deferred.addBoth(self.check_context, "two") + + # let the first lookup complete (in the sentinel context) + lookup_1_deferred.callback(None) + + # now the second wait should complete and restore our + # loggingcontext. + yield wait_2_deferred + + @defer.inlineCallbacks + def test_verify_json_objects_for_server_awaits_previous_requests(self): + clock = Clock(reactor) + key1 = signedjson.key.generate_signing_key(1) + + kr = keyring.Keyring(self.hs) + json1 = {} + signedjson.sign.sign_json(json1, "server10", key1) + + persp_resp = { + "server_keys": [ + self.mock_perspective_server.get_signed_key( + "server10", + signedjson.key.get_verify_key(key1) + ), + ] + } + persp_deferred = defer.Deferred() + + @defer.inlineCallbacks + def get_perspectives(**kwargs): + self.assertEquals( + LoggingContext.current_context().request, "11", + ) + with logcontext.PreserveLoggingContext(): + yield persp_deferred + defer.returnValue(persp_resp) + self.http_client.post_json.side_effect = get_perspectives + + with LoggingContext("11") as context_11: + context_11.request = "11" + + # start off a first set of lookups + res_deferreds = kr.verify_json_objects_for_server( + [("server10", json1), + ("server11", {}) + ] + ) + + # the unsigned json should be rejected pretty quickly + self.assertTrue(res_deferreds[1].called) + try: + yield res_deferreds[1] + self.assertFalse("unsigned json didn't cause a failure") + except SynapseError: + pass + + self.assertFalse(res_deferreds[0].called) + res_deferreds[0].addBoth(self.check_context, None) + + # wait a tick for it to send the request to the perspectives server + # (it first tries the datastore) + yield clock.sleep(1) # XXX find out why this takes so long! + self.http_client.post_json.assert_called_once() + + self.assertIs(LoggingContext.current_context(), context_11) + + context_12 = LoggingContext("12") + context_12.request = "12" + with logcontext.PreserveLoggingContext(context_12): + # a second request for a server with outstanding requests + # should block rather than start a second call + self.http_client.post_json.reset_mock() + self.http_client.post_json.return_value = defer.Deferred() + + res_deferreds_2 = kr.verify_json_objects_for_server( + [("server10", json1)], + ) + yield clock.sleep(1) + self.http_client.post_json.assert_not_called() + res_deferreds_2[0].addBoth(self.check_context, None) + + # complete the first request + with logcontext.PreserveLoggingContext(): + persp_deferred.callback(persp_resp) + self.assertIs(LoggingContext.current_context(), context_11) + + with logcontext.PreserveLoggingContext(): + yield res_deferreds[0] + yield res_deferreds_2[0] + + @defer.inlineCallbacks + def test_verify_json_for_server(self): + kr = keyring.Keyring(self.hs) + + key1 = signedjson.key.generate_signing_key(1) + yield self.hs.datastore.store_server_verify_key( + "server9", "", time.time() * 1000, + signedjson.key.get_verify_key(key1), + ) + json1 = {} + signedjson.sign.sign_json(json1, "server9", key1) + + sentinel_context = LoggingContext.current_context() + + with LoggingContext("one") as context_one: + context_one.request = "one" + + defer = kr.verify_json_for_server("server9", {}) + try: + yield defer + self.fail("should fail on unsigned json") + except SynapseError: + pass + self.assertIs(LoggingContext.current_context(), context_one) + + defer = kr.verify_json_for_server("server9", json1) + self.assertFalse(defer.called) + self.assertIs(LoggingContext.current_context(), sentinel_context) + yield defer + + self.assertIs(LoggingContext.current_context(), context_one) diff --git a/tests/events/test_utils.py b/tests/events/test_utils.py index dfc870066..f51d99419 100644 --- a/tests/events/test_utils.py +++ b/tests/events/test_utils.py @@ -14,11 +14,11 @@ # limitations under the License. -from .. import unittest - from synapse.events import FrozenEvent from synapse.events.utils import prune_event, serialize_event +from .. import unittest + def MockEvent(**kwargs): if "event_id" not in kwargs: diff --git a/tests/federation/__init__.py b/tests/federation/__init__.py new file mode 100644 index 000000000..e69de29bb diff --git a/tests/federation/test_federation_server.py b/tests/federation/test_federation_server.py new file mode 100644 index 000000000..c91e25f54 --- /dev/null +++ b/tests/federation/test_federation_server.py @@ -0,0 +1,58 @@ +# -*- coding: utf-8 -*- +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import logging + +from synapse.events import FrozenEvent +from synapse.federation.federation_server import server_matches_acl_event + +from tests import unittest + + +@unittest.DEBUG +class ServerACLsTestCase(unittest.TestCase): + def test_blacklisted_server(self): + e = _create_acl_event({ + "allow": ["*"], + "deny": ["evil.com"], + }) + logging.info("ACL event: %s", e.content) + + self.assertFalse(server_matches_acl_event("evil.com", e)) + self.assertFalse(server_matches_acl_event("EVIL.COM", e)) + + self.assertTrue(server_matches_acl_event("evil.com.au", e)) + self.assertTrue(server_matches_acl_event("honestly.not.evil.com", e)) + + def test_block_ip_literals(self): + e = _create_acl_event({ + "allow_ip_literals": False, + "allow": ["*"], + }) + logging.info("ACL event: %s", e.content) + + self.assertFalse(server_matches_acl_event("1.2.3.4", e)) + self.assertTrue(server_matches_acl_event("1a.2.3.4", e)) + self.assertFalse(server_matches_acl_event("[1:2::]", e)) + self.assertTrue(server_matches_acl_event("1:2:3:4", e)) + + +def _create_acl_event(content): + return FrozenEvent({ + "room_id": "!a:b", + "event_id": "$a:b", + "type": "m.room.server_acls", + "sender": "@a:b", + "content": content + }) diff --git a/tests/handlers/test_appservice.py b/tests/handlers/test_appservice.py index 7fe88172c..57c0771cf 100644 --- a/tests/handlers/test_appservice.py +++ b/tests/handlers/test_appservice.py @@ -13,13 +13,15 @@ # See the License for the specific language governing permissions and # limitations under the License. +from mock import Mock + from twisted.internet import defer -from .. import unittest -from tests.utils import MockClock from synapse.handlers.appservice import ApplicationServicesHandler -from mock import Mock +from tests.utils import MockClock + +from .. import unittest class AppServiceHandlerTestCase(unittest.TestCase): @@ -31,6 +33,7 @@ class AppServiceHandlerTestCase(unittest.TestCase): self.mock_scheduler = Mock() hs = Mock() hs.get_datastore = Mock(return_value=self.mock_store) + self.mock_store.get_received_ts.return_value = 0 hs.get_application_service_api = Mock(return_value=self.mock_as_api) hs.get_application_service_scheduler = Mock(return_value=self.mock_scheduler) hs.get_clock.return_value = MockClock() @@ -53,7 +56,10 @@ class AppServiceHandlerTestCase(unittest.TestCase): type="m.room.message", room_id="!foo:bar" ) - self.mock_store.get_new_events_for_appservice.return_value = (0, [event]) + self.mock_store.get_new_events_for_appservice.side_effect = [ + (0, [event]), + (0, []) + ] self.mock_as_api.push = Mock() yield self.handler.notify_interested_services(0) self.mock_scheduler.submit_event_for_as.assert_called_once_with( @@ -75,7 +81,10 @@ class AppServiceHandlerTestCase(unittest.TestCase): ) self.mock_as_api.push = Mock() self.mock_as_api.query_user = Mock() - self.mock_store.get_new_events_for_appservice.return_value = (0, [event]) + self.mock_store.get_new_events_for_appservice.side_effect = [ + (0, [event]), + (0, []) + ] yield self.handler.notify_interested_services(0) self.mock_as_api.query_user.assert_called_once_with( services[0], user_id @@ -98,7 +107,10 @@ class AppServiceHandlerTestCase(unittest.TestCase): ) self.mock_as_api.push = Mock() self.mock_as_api.query_user = Mock() - self.mock_store.get_new_events_for_appservice.return_value = (0, [event]) + self.mock_store.get_new_events_for_appservice.side_effect = [ + (0, [event]), + (0, []) + ] yield self.handler.notify_interested_services(0) self.assertFalse( self.mock_as_api.query_user.called, diff --git a/tests/handlers/test_auth.py b/tests/handlers/test_auth.py index 1822dcf1e..2e5e8e4de 100644 --- a/tests/handlers/test_auth.py +++ b/tests/handlers/test_auth.py @@ -14,11 +14,13 @@ # limitations under the License. import pymacaroons + from twisted.internet import defer import synapse import synapse.api.errors from synapse.handlers.auth import AuthHandler + from tests import unittest from tests.utils import setup_test_homeserver diff --git a/tests/handlers/test_device.py b/tests/handlers/test_device.py index 2eaaa8253..633a0b7f3 100644 --- a/tests/handlers/test_device.py +++ b/tests/handlers/test_device.py @@ -17,9 +17,8 @@ from twisted.internet import defer import synapse.api.errors import synapse.handlers.device - import synapse.storage -from synapse import types + from tests import unittest, utils user1 = "@boris:aaa" @@ -179,6 +178,6 @@ class DeviceTestCase(unittest.TestCase): if ip is not None: yield self.store.insert_client_ip( - types.UserID.from_string(user_id), + user_id, access_token, ip, "user_agent", device_id) self.clock.advance_time(1000) diff --git a/tests/handlers/test_directory.py b/tests/handlers/test_directory.py index 571277390..a35307031 100644 --- a/tests/handlers/test_directory.py +++ b/tests/handlers/test_directory.py @@ -14,14 +14,14 @@ # limitations under the License. -from tests import unittest -from twisted.internet import defer - from mock import Mock +from twisted.internet import defer + from synapse.handlers.directory import DirectoryHandler from synapse.types import RoomAlias +from tests import unittest from tests.utils import setup_test_homeserver @@ -35,21 +35,20 @@ class DirectoryTestCase(unittest.TestCase): @defer.inlineCallbacks def setUp(self): - self.mock_federation = Mock(spec=[ - "make_query", - "register_edu_handler", - ]) + self.mock_federation = Mock() + self.mock_registry = Mock() self.query_handlers = {} def register_query_handler(query_type, handler): self.query_handlers[query_type] = handler - self.mock_federation.register_query_handler = register_query_handler + self.mock_registry.register_query_handler = register_query_handler hs = yield setup_test_homeserver( http_client=None, resource_for_federation=Mock(), - replication_layer=self.mock_federation, + federation_client=self.mock_federation, + federation_registry=self.mock_registry, ) hs.handlers = DirectoryHandlers(hs) diff --git a/tests/handlers/test_e2e_keys.py b/tests/handlers/test_e2e_keys.py index 19f5ed6bc..ca1542236 100644 --- a/tests/handlers/test_e2e_keys.py +++ b/tests/handlers/test_e2e_keys.py @@ -14,13 +14,14 @@ # limitations under the License. import mock -from synapse.api import errors + from twisted.internet import defer import synapse.api.errors import synapse.handlers.e2e_keys - import synapse.storage +from synapse.api import errors + from tests import unittest, utils @@ -34,7 +35,7 @@ class E2eKeysHandlerTestCase(unittest.TestCase): def setUp(self): self.hs = yield utils.setup_test_homeserver( handlers=None, - replication_layer=mock.Mock(), + federation_client=mock.Mock(), ) self.handler = synapse.handlers.e2e_keys.E2eKeysHandler(self.hs) @@ -143,7 +144,6 @@ class E2eKeysHandlerTestCase(unittest.TestCase): except errors.SynapseError: pass - @unittest.DEBUG @defer.inlineCallbacks def test_claim_one_time_key(self): local_user = "@boris:" + self.hs.hostname diff --git a/tests/handlers/test_presence.py b/tests/handlers/test_presence.py index de06a6ad3..121ce7863 100644 --- a/tests/handlers/test_presence.py +++ b/tests/handlers/test_presence.py @@ -14,18 +14,22 @@ # limitations under the License. -from tests import unittest - from mock import Mock, call from synapse.api.constants import PresenceState from synapse.handlers.presence import ( - handle_update, handle_timeout, - IDLE_TIMER, SYNC_ONLINE_TIMEOUT, LAST_ACTIVE_GRANULARITY, FEDERATION_TIMEOUT, FEDERATION_PING_INTERVAL, + FEDERATION_TIMEOUT, + IDLE_TIMER, + LAST_ACTIVE_GRANULARITY, + SYNC_ONLINE_TIMEOUT, + handle_timeout, + handle_update, ) from synapse.storage.presence import UserPresenceState +from tests import unittest + class PresenceUpdateTestCase(unittest.TestCase): def test_offline_to_online(self): diff --git a/tests/handlers/test_profile.py b/tests/handlers/test_profile.py index 2a203129c..dc17918a3 100644 --- a/tests/handlers/test_profile.py +++ b/tests/handlers/test_profile.py @@ -14,16 +14,16 @@ # limitations under the License. -from tests import unittest -from twisted.internet import defer - from mock import Mock, NonCallableMock +from twisted.internet import defer + import synapse.types from synapse.api.errors import AuthError from synapse.handlers.profile import ProfileHandler from synapse.types import UserID +from tests import unittest from tests.utils import setup_test_homeserver @@ -37,23 +37,23 @@ class ProfileTestCase(unittest.TestCase): @defer.inlineCallbacks def setUp(self): - self.mock_federation = Mock(spec=[ - "make_query", - "register_edu_handler", - ]) + self.mock_federation = Mock() + self.mock_registry = Mock() self.query_handlers = {} def register_query_handler(query_type, handler): self.query_handlers[query_type] = handler - self.mock_federation.register_query_handler = register_query_handler + self.mock_registry.register_query_handler = register_query_handler hs = yield setup_test_homeserver( http_client=None, handlers=None, resource_for_federation=Mock(), - replication_layer=self.mock_federation, + federation_client=self.mock_federation, + federation_server=Mock(), + federation_registry=self.mock_registry, ratelimiter=NonCallableMock(spec_set=[ "send_message", ]) @@ -62,8 +62,6 @@ class ProfileTestCase(unittest.TestCase): self.ratelimiter = hs.get_ratelimiter() self.ratelimiter.send_message.return_value = (True, 0) - hs.handlers = ProfileHandlers(hs) - self.store = hs.get_datastore() self.frank = UserID.from_string("@1234ABCD:test") @@ -72,7 +70,7 @@ class ProfileTestCase(unittest.TestCase): yield self.store.create_profile(self.frank.localpart) - self.handler = hs.get_handlers().profile_handler + self.handler = hs.get_profile_handler() @defer.inlineCallbacks def test_get_my_name(self): diff --git a/tests/handlers/test_register.py b/tests/handlers/test_register.py index c8cf9a63e..025fa1be8 100644 --- a/tests/handlers/test_register.py +++ b/tests/handlers/test_register.py @@ -13,15 +13,16 @@ # See the License for the specific language governing permissions and # limitations under the License. +from mock import Mock + from twisted.internet import defer -from .. import unittest from synapse.handlers.register import RegistrationHandler from synapse.types import UserID, create_requester from tests.utils import setup_test_homeserver -from mock import Mock +from .. import unittest class RegistrationHandlers(object): @@ -40,13 +41,14 @@ class RegistrationTestCase(unittest.TestCase): self.hs = yield setup_test_homeserver( handlers=None, http_client=None, - expire_access_token=True) + expire_access_token=True, + profile_handler=Mock(), + ) self.macaroon_generator = Mock( generate_access_token=Mock(return_value='secret')) self.hs.get_macaroon_generator = Mock(return_value=self.macaroon_generator) self.hs.handlers = RegistrationHandlers(self.hs) self.handler = self.hs.get_handlers().registration_handler - self.hs.get_handlers().profile_handler = Mock() @defer.inlineCallbacks def test_user_is_created_and_logged_in_if_doesnt_exist(self): diff --git a/tests/handlers/test_typing.py b/tests/handlers/test_typing.py index dbe50383d..b08856f76 100644 --- a/tests/handlers/test_typing.py +++ b/tests/handlers/test_typing.py @@ -14,19 +14,24 @@ # limitations under the License. -from tests import unittest -from twisted.internet import defer - -from mock import Mock, call, ANY import json -from ..utils import ( - MockHttpResource, MockClock, DeferredMockCallable, setup_test_homeserver -) +from mock import ANY, Mock, call + +from twisted.internet import defer from synapse.api.errors import AuthError from synapse.types import UserID +from tests import unittest + +from ..utils import ( + DeferredMockCallable, + MockClock, + MockHttpResource, + setup_test_homeserver, +) + def _expect_edu(destination, edu_type, content, origin="test"): return { @@ -58,7 +63,7 @@ class TypingNotificationsTestCase(unittest.TestCase): self.mock_federation_resource = MockHttpResource() - mock_notifier = Mock(spec=["on_new_event"]) + mock_notifier = Mock() self.on_new_event = mock_notifier.on_new_event self.auth = Mock(spec=[]) @@ -76,9 +81,12 @@ class TypingNotificationsTestCase(unittest.TestCase): "set_received_txn_response", "get_destination_retry_timings", "get_devices_by_remote", + # Bits that user_directory needs + "get_user_directory_stream_pos", + "get_current_state_deltas", ]), state_handler=self.state_handler, - handlers=None, + handlers=Mock(), notifier=mock_notifier, resource_for_client=Mock(), resource_for_federation=self.mock_federation_resource, @@ -122,6 +130,15 @@ class TypingNotificationsTestCase(unittest.TestCase): return set(str(u) for u in self.room_members) self.state_handler.get_current_user_in_room = get_current_user_in_room + self.datastore.get_user_directory_stream_pos.return_value = ( + # we deliberately return a non-None stream pos to avoid doing an initial_spam + defer.succeed(1) + ) + + self.datastore.get_current_state_deltas.return_value = ( + None + ) + self.auth.check_joined_room = check_joined_room self.datastore.get_to_device_stream_token = lambda: 0 diff --git a/tests/http/__init__.py b/tests/http/__init__.py new file mode 100644 index 000000000..e69de29bb diff --git a/tests/http/test_endpoint.py b/tests/http/test_endpoint.py new file mode 100644 index 000000000..60e6a7595 --- /dev/null +++ b/tests/http/test_endpoint.py @@ -0,0 +1,53 @@ +# -*- coding: utf-8 -*- +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +from synapse.http.endpoint import parse_and_validate_server_name, parse_server_name + +from tests import unittest + + +class ServerNameTestCase(unittest.TestCase): + def test_parse_server_name(self): + test_data = { + 'localhost': ('localhost', None), + 'my-example.com:1234': ('my-example.com', 1234), + '1.2.3.4': ('1.2.3.4', None), + '[0abc:1def::1234]': ('[0abc:1def::1234]', None), + '1.2.3.4:1': ('1.2.3.4', 1), + '[0abc:1def::1234]:8080': ('[0abc:1def::1234]', 8080), + } + + for i, o in test_data.items(): + self.assertEqual(parse_server_name(i), o) + + def test_validate_bad_server_names(self): + test_data = [ + "", # empty + "localhost:http", # non-numeric port + "1234]", # smells like ipv6 literal but isn't + "[1234", + "underscore_.com", + "percent%65.com", + "1234:5678:80", # too many colons + ] + for i in test_data: + try: + parse_and_validate_server_name(i) + self.fail( + "Expected parse_and_validate_server_name('%s') to throw" % ( + i, + ), + ) + except ValueError: + pass diff --git a/tests/metrics/test_metric.py b/tests/metrics/test_metric.py deleted file mode 100644 index f85455a5a..000000000 --- a/tests/metrics/test_metric.py +++ /dev/null @@ -1,161 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2015, 2016 OpenMarket Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -from tests import unittest - -from synapse.metrics.metric import ( - CounterMetric, CallbackMetric, DistributionMetric, CacheMetric -) - - -class CounterMetricTestCase(unittest.TestCase): - - def test_scalar(self): - counter = CounterMetric("scalar") - - self.assertEquals(counter.render(), [ - 'scalar 0', - ]) - - counter.inc() - - self.assertEquals(counter.render(), [ - 'scalar 1', - ]) - - counter.inc_by(2) - - self.assertEquals(counter.render(), [ - 'scalar 3' - ]) - - def test_vector(self): - counter = CounterMetric("vector", labels=["method"]) - - # Empty counter doesn't yet know what values it has - self.assertEquals(counter.render(), []) - - counter.inc("GET") - - self.assertEquals(counter.render(), [ - 'vector{method="GET"} 1', - ]) - - counter.inc("GET") - counter.inc("PUT") - - self.assertEquals(counter.render(), [ - 'vector{method="GET"} 2', - 'vector{method="PUT"} 1', - ]) - - -class CallbackMetricTestCase(unittest.TestCase): - - def test_scalar(self): - d = dict() - - metric = CallbackMetric("size", lambda: len(d)) - - self.assertEquals(metric.render(), [ - 'size 0', - ]) - - d["key"] = "value" - - self.assertEquals(metric.render(), [ - 'size 1', - ]) - - def test_vector(self): - vals = dict() - - metric = CallbackMetric("values", lambda: vals, labels=["type"]) - - self.assertEquals(metric.render(), []) - - # Keys have to be tuples, even if they're 1-element - vals[("foo",)] = 1 - vals[("bar",)] = 2 - - self.assertEquals(metric.render(), [ - 'values{type="bar"} 2', - 'values{type="foo"} 1', - ]) - - -class DistributionMetricTestCase(unittest.TestCase): - - def test_scalar(self): - metric = DistributionMetric("thing") - - self.assertEquals(metric.render(), [ - 'thing:count 0', - 'thing:total 0', - ]) - - metric.inc_by(500) - - self.assertEquals(metric.render(), [ - 'thing:count 1', - 'thing:total 500', - ]) - - def test_vector(self): - metric = DistributionMetric("queries", labels=["verb"]) - - self.assertEquals(metric.render(), []) - - metric.inc_by(300, "SELECT") - metric.inc_by(200, "SELECT") - metric.inc_by(800, "INSERT") - - self.assertEquals(metric.render(), [ - 'queries:count{verb="INSERT"} 1', - 'queries:count{verb="SELECT"} 2', - 'queries:total{verb="INSERT"} 800', - 'queries:total{verb="SELECT"} 500', - ]) - - -class CacheMetricTestCase(unittest.TestCase): - - def test_cache(self): - d = dict() - - metric = CacheMetric("cache", lambda: len(d), "cache_name") - - self.assertEquals(metric.render(), [ - 'cache:hits{name="cache_name"} 0', - 'cache:total{name="cache_name"} 0', - 'cache:size{name="cache_name"} 0', - ]) - - metric.inc_misses() - d["key"] = "value" - - self.assertEquals(metric.render(), [ - 'cache:hits{name="cache_name"} 0', - 'cache:total{name="cache_name"} 1', - 'cache:size{name="cache_name"} 1', - ]) - - metric.inc_hits() - - self.assertEquals(metric.render(), [ - 'cache:hits{name="cache_name"} 1', - 'cache:total{name="cache_name"} 2', - 'cache:size{name="cache_name"} 1', - ]) diff --git a/tests/replication/slave/storage/_base.py b/tests/replication/slave/storage/_base.py index 81063f19a..8708c8a19 100644 --- a/tests/replication/slave/storage/_base.py +++ b/tests/replication/slave/storage/_base.py @@ -12,15 +12,20 @@ # See the License for the specific language governing permissions and # limitations under the License. -from twisted.internet import defer, reactor -from tests import unittest +import tempfile from mock import Mock, NonCallableMock -from tests.utils import setup_test_homeserver -from synapse.replication.tcp.resource import ReplicationStreamProtocolFactory + +from twisted.internet import defer, reactor + from synapse.replication.tcp.client import ( - ReplicationClientHandler, ReplicationClientFactory, + ReplicationClientFactory, + ReplicationClientHandler, ) +from synapse.replication.tcp.resource import ReplicationStreamProtocolFactory + +from tests import unittest +from tests.utils import setup_test_homeserver class BaseSlavedStoreTestCase(unittest.TestCase): @@ -29,7 +34,7 @@ class BaseSlavedStoreTestCase(unittest.TestCase): self.hs = yield setup_test_homeserver( "blue", http_client=None, - replication_layer=Mock(), + federation_client=Mock(), ratelimiter=NonCallableMock(spec_set=[ "send_message", ]), @@ -41,7 +46,9 @@ class BaseSlavedStoreTestCase(unittest.TestCase): self.event_id = 0 server_factory = ReplicationStreamProtocolFactory(self.hs) - listener = reactor.listenUNIX("\0xxx", server_factory) + # XXX: mktemp is unsafe and should never be used. but we're just a test. + path = tempfile.mktemp(prefix="base_slaved_store_test_case_socket") + listener = reactor.listenUNIX(path, server_factory) self.addCleanup(listener.stopListening) self.streamer = server_factory.streamer @@ -49,7 +56,7 @@ class BaseSlavedStoreTestCase(unittest.TestCase): client_factory = ReplicationClientFactory( self.hs, "client_name", self.replication_handler ) - client_connector = reactor.connectUNIX("\0xxx", client_factory) + client_connector = reactor.connectUNIX(path, client_factory) self.addCleanup(client_factory.stopTrying) self.addCleanup(client_connector.disconnect) diff --git a/tests/replication/slave/storage/test_account_data.py b/tests/replication/slave/storage/test_account_data.py index da54d478c..adf226404 100644 --- a/tests/replication/slave/storage/test_account_data.py +++ b/tests/replication/slave/storage/test_account_data.py @@ -13,11 +13,11 @@ # limitations under the License. -from ._base import BaseSlavedStoreTestCase +from twisted.internet import defer from synapse.replication.slave.storage.account_data import SlavedAccountDataStore -from twisted.internet import defer +from ._base import BaseSlavedStoreTestCase USER_ID = "@feeling:blue" TYPE = "my.type" @@ -37,10 +37,6 @@ class SlavedAccountDataStoreTestCase(BaseSlavedStoreTestCase): "get_global_account_data_by_type_for_user", [TYPE, USER_ID], {"a": 1} ) - yield self.check( - "get_global_account_data_by_type_for_users", - [TYPE, [USER_ID]], {USER_ID: {"a": 1}} - ) yield self.master_store.add_account_data_for_user( USER_ID, TYPE, {"a": 2} @@ -50,7 +46,3 @@ class SlavedAccountDataStoreTestCase(BaseSlavedStoreTestCase): "get_global_account_data_by_type_for_user", [TYPE, USER_ID], {"a": 2} ) - yield self.check( - "get_global_account_data_by_type_for_users", - [TYPE, [USER_ID]], {USER_ID: {"a": 2}} - ) diff --git a/tests/replication/slave/storage/test_events.py b/tests/replication/slave/storage/test_events.py index 105e1228b..f5b47f5ec 100644 --- a/tests/replication/slave/storage/test_events.py +++ b/tests/replication/slave/storage/test_events.py @@ -12,15 +12,14 @@ # See the License for the specific language governing permissions and # limitations under the License. -from ._base import BaseSlavedStoreTestCase +from twisted.internet import defer from synapse.events import FrozenEvent, _EventInternalMetadata from synapse.events.snapshot import EventContext from synapse.replication.slave.storage.events import SlavedEventStore from synapse.storage.roommember import RoomsForUser -from twisted.internet import defer - +from ._base import BaseSlavedStoreTestCase USER_ID = "@feeling:blue" USER_ID_2 = "@bright:blue" @@ -223,16 +222,21 @@ class SlavedEventStoreTestCase(BaseSlavedStoreTestCase): state_ids = { key: e.event_id for key, e in state.items() } - context = EventContext() - context.current_state_ids = state_ids - context.prev_state_ids = state_ids - elif not backfill: + context = EventContext.with_state( + state_group=None, + current_state_ids=state_ids, + prev_state_ids=state_ids + ) + else: state_handler = self.hs.get_state_handler() context = yield state_handler.compute_event_context(event) - else: - context = EventContext() - context.push_actions = push_actions + yield self.master_store.add_push_actions_to_staging( + event.event_id, { + user_id: actions + for user_id, actions in push_actions + }, + ) ordering = None if backfill: diff --git a/tests/replication/slave/storage/test_receipts.py b/tests/replication/slave/storage/test_receipts.py index 6624fe4ee..e6d670cc1 100644 --- a/tests/replication/slave/storage/test_receipts.py +++ b/tests/replication/slave/storage/test_receipts.py @@ -12,11 +12,11 @@ # See the License for the specific language governing permissions and # limitations under the License. -from ._base import BaseSlavedStoreTestCase +from twisted.internet import defer from synapse.replication.slave.storage.receipts import SlavedReceiptsStore -from twisted.internet import defer +from ._base import BaseSlavedStoreTestCase USER_ID = "@feeling:blue" ROOM_ID = "!room:blue" diff --git a/tests/rest/client/test_transactions.py b/tests/rest/client/test_transactions.py index d7cea3026..34e68ae82 100644 --- a/tests/rest/client/test_transactions.py +++ b/tests/rest/client/test_transactions.py @@ -1,7 +1,11 @@ -from synapse.rest.client.transactions import HttpTransactionCache -from synapse.rest.client.transactions import CLEANUP_PERIOD_MS -from twisted.internet import defer from mock import Mock, call + +from twisted.internet import defer, reactor + +from synapse.rest.client.transactions import CLEANUP_PERIOD_MS, HttpTransactionCache +from synapse.util import Clock +from synapse.util.logcontext import LoggingContext + from tests import unittest from tests.utils import MockClock @@ -10,7 +14,10 @@ class HttpTransactionCacheTestCase(unittest.TestCase): def setUp(self): self.clock = MockClock() - self.cache = HttpTransactionCache(self.clock) + self.hs = Mock() + self.hs.get_clock = Mock(return_value=self.clock) + self.hs.get_auth = Mock() + self.cache = HttpTransactionCache(self.hs) self.mock_http_response = (200, "GOOD JOB!") self.mock_key = "foo" @@ -39,6 +46,78 @@ class HttpTransactionCacheTestCase(unittest.TestCase): # expect only a single call to do the work cb.assert_called_once_with("some_arg", keyword="arg", changing_args=0) + @defer.inlineCallbacks + def test_logcontexts_with_async_result(self): + @defer.inlineCallbacks + def cb(): + yield Clock(reactor).sleep(0) + defer.returnValue("yay") + + @defer.inlineCallbacks + def test(): + with LoggingContext("c") as c1: + res = yield self.cache.fetch_or_execute(self.mock_key, cb) + self.assertIs(LoggingContext.current_context(), c1) + self.assertEqual(res, "yay") + + # run the test twice in parallel + d = defer.gatherResults([test(), test()]) + self.assertIs(LoggingContext.current_context(), LoggingContext.sentinel) + yield d + self.assertIs(LoggingContext.current_context(), LoggingContext.sentinel) + + @defer.inlineCallbacks + def test_does_not_cache_exceptions(self): + """Checks that, if the callback throws an exception, it is called again + for the next request. + """ + called = [False] + + def cb(): + if called[0]: + # return a valid result the second time + return defer.succeed(self.mock_http_response) + + called[0] = True + raise Exception("boo") + + with LoggingContext("test") as test_context: + try: + yield self.cache.fetch_or_execute(self.mock_key, cb) + except Exception as e: + self.assertEqual(e.message, "boo") + self.assertIs(LoggingContext.current_context(), test_context) + + res = yield self.cache.fetch_or_execute(self.mock_key, cb) + self.assertEqual(res, self.mock_http_response) + self.assertIs(LoggingContext.current_context(), test_context) + + @defer.inlineCallbacks + def test_does_not_cache_failures(self): + """Checks that, if the callback returns a failure, it is called again + for the next request. + """ + called = [False] + + def cb(): + if called[0]: + # return a valid result the second time + return defer.succeed(self.mock_http_response) + + called[0] = True + return defer.fail(Exception("boo")) + + with LoggingContext("test") as test_context: + try: + yield self.cache.fetch_or_execute(self.mock_key, cb) + except Exception as e: + self.assertEqual(e.message, "boo") + self.assertIs(LoggingContext.current_context(), test_context) + + res = yield self.cache.fetch_or_execute(self.mock_key, cb) + self.assertEqual(res, self.mock_http_response) + self.assertIs(LoggingContext.current_context(), test_context) + @defer.inlineCallbacks def test_cleans_up(self): cb = Mock( diff --git a/tests/rest/client/v1/test_admin.py b/tests/rest/client/v1/test_admin.py new file mode 100644 index 000000000..8c9014560 --- /dev/null +++ b/tests/rest/client/v1/test_admin.py @@ -0,0 +1,305 @@ +# -*- coding: utf-8 -*- +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import hashlib +import hmac +import json + +from mock import Mock + +from synapse.http.server import JsonResource +from synapse.rest.client.v1.admin import register_servlets +from synapse.util import Clock + +from tests import unittest +from tests.server import ( + ThreadedMemoryReactorClock, + make_request, + render, + setup_test_homeserver, +) + + +class UserRegisterTestCase(unittest.TestCase): + def setUp(self): + + self.clock = ThreadedMemoryReactorClock() + self.hs_clock = Clock(self.clock) + self.url = "/_matrix/client/r0/admin/register" + + self.registration_handler = Mock() + self.identity_handler = Mock() + self.login_handler = Mock() + self.device_handler = Mock() + self.device_handler.check_device_registered = Mock(return_value="FAKE") + + self.datastore = Mock(return_value=Mock()) + self.datastore.get_current_state_deltas = Mock(return_value=[]) + + self.secrets = Mock() + + self.hs = setup_test_homeserver( + http_client=None, clock=self.hs_clock, reactor=self.clock + ) + + self.hs.config.registration_shared_secret = u"shared" + + self.hs.get_media_repository = Mock() + self.hs.get_deactivate_account_handler = Mock() + + self.resource = JsonResource(self.hs) + register_servlets(self.hs, self.resource) + + def test_disabled(self): + """ + If there is no shared secret, registration through this method will be + prevented. + """ + self.hs.config.registration_shared_secret = None + + request, channel = make_request("POST", self.url, b'{}') + render(request, self.resource, self.clock) + + self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual( + 'Shared secret registration is not enabled', channel.json_body["error"] + ) + + def test_get_nonce(self): + """ + Calling GET on the endpoint will return a randomised nonce, using the + homeserver's secrets provider. + """ + secrets = Mock() + secrets.token_hex = Mock(return_value="abcd") + + self.hs.get_secrets = Mock(return_value=secrets) + + request, channel = make_request("GET", self.url) + render(request, self.resource, self.clock) + + self.assertEqual(channel.json_body, {"nonce": "abcd"}) + + def test_expired_nonce(self): + """ + Calling GET on the endpoint will return a randomised nonce, which will + only last for SALT_TIMEOUT (60s). + """ + request, channel = make_request("GET", self.url) + render(request, self.resource, self.clock) + nonce = channel.json_body["nonce"] + + # 59 seconds + self.clock.advance(59) + + body = json.dumps({"nonce": nonce}) + request, channel = make_request("POST", self.url, body.encode('utf8')) + render(request, self.resource, self.clock) + + self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual('username must be specified', channel.json_body["error"]) + + # 61 seconds + self.clock.advance(2) + + request, channel = make_request("POST", self.url, body.encode('utf8')) + render(request, self.resource, self.clock) + + self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual('unrecognised nonce', channel.json_body["error"]) + + def test_register_incorrect_nonce(self): + """ + Only the provided nonce can be used, as it's checked in the MAC. + """ + request, channel = make_request("GET", self.url) + render(request, self.resource, self.clock) + nonce = channel.json_body["nonce"] + + want_mac = hmac.new(key=b"shared", digestmod=hashlib.sha1) + want_mac.update(b"notthenonce\x00bob\x00abc123\x00admin") + want_mac = want_mac.hexdigest() + + body = json.dumps( + { + "nonce": nonce, + "username": "bob", + "password": "abc123", + "admin": True, + "mac": want_mac, + } + ).encode('utf8') + request, channel = make_request("POST", self.url, body.encode('utf8')) + render(request, self.resource, self.clock) + + self.assertEqual(403, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("HMAC incorrect", channel.json_body["error"]) + + def test_register_correct_nonce(self): + """ + When the correct nonce is provided, and the right key is provided, the + user is registered. + """ + request, channel = make_request("GET", self.url) + render(request, self.resource, self.clock) + nonce = channel.json_body["nonce"] + + want_mac = hmac.new(key=b"shared", digestmod=hashlib.sha1) + want_mac.update(nonce.encode('ascii') + b"\x00bob\x00abc123\x00admin") + want_mac = want_mac.hexdigest() + + body = json.dumps( + { + "nonce": nonce, + "username": "bob", + "password": "abc123", + "admin": True, + "mac": want_mac, + } + ).encode('utf8') + request, channel = make_request("POST", self.url, body.encode('utf8')) + render(request, self.resource, self.clock) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("@bob:test", channel.json_body["user_id"]) + + def test_nonce_reuse(self): + """ + A valid unrecognised nonce. + """ + request, channel = make_request("GET", self.url) + render(request, self.resource, self.clock) + nonce = channel.json_body["nonce"] + + want_mac = hmac.new(key=b"shared", digestmod=hashlib.sha1) + want_mac.update(nonce.encode('ascii') + b"\x00bob\x00abc123\x00admin") + want_mac = want_mac.hexdigest() + + body = json.dumps( + { + "nonce": nonce, + "username": "bob", + "password": "abc123", + "admin": True, + "mac": want_mac, + } + ).encode('utf8') + request, channel = make_request("POST", self.url, body.encode('utf8')) + render(request, self.resource, self.clock) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("@bob:test", channel.json_body["user_id"]) + + # Now, try and reuse it + request, channel = make_request("POST", self.url, body.encode('utf8')) + render(request, self.resource, self.clock) + + self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual('unrecognised nonce', channel.json_body["error"]) + + def test_missing_parts(self): + """ + Synapse will complain if you don't give nonce, username, password, and + mac. Admin is optional. Additional checks are done for length and + type. + """ + def nonce(): + request, channel = make_request("GET", self.url) + render(request, self.resource, self.clock) + return channel.json_body["nonce"] + + # + # Nonce check + # + + # Must be present + body = json.dumps({}) + request, channel = make_request("POST", self.url, body.encode('utf8')) + render(request, self.resource, self.clock) + + self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual('nonce must be specified', channel.json_body["error"]) + + # + # Username checks + # + + # Must be present + body = json.dumps({"nonce": nonce()}) + request, channel = make_request("POST", self.url, body.encode('utf8')) + render(request, self.resource, self.clock) + + self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual('username must be specified', channel.json_body["error"]) + + # Must be a string + body = json.dumps({"nonce": nonce(), "username": 1234}) + request, channel = make_request("POST", self.url, body.encode('utf8')) + render(request, self.resource, self.clock) + + self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual('Invalid username', channel.json_body["error"]) + + # Must not have null bytes + body = json.dumps({"nonce": nonce(), "username": b"abcd\x00"}) + request, channel = make_request("POST", self.url, body.encode('utf8')) + render(request, self.resource, self.clock) + + self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual('Invalid username', channel.json_body["error"]) + + # Must not have null bytes + body = json.dumps({"nonce": nonce(), "username": "a" * 1000}) + request, channel = make_request("POST", self.url, body.encode('utf8')) + render(request, self.resource, self.clock) + + self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual('Invalid username', channel.json_body["error"]) + + # + # Username checks + # + + # Must be present + body = json.dumps({"nonce": nonce(), "username": "a"}) + request, channel = make_request("POST", self.url, body.encode('utf8')) + render(request, self.resource, self.clock) + + self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual('password must be specified', channel.json_body["error"]) + + # Must be a string + body = json.dumps({"nonce": nonce(), "username": "a", "password": 1234}) + request, channel = make_request("POST", self.url, body.encode('utf8')) + render(request, self.resource, self.clock) + + self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual('Invalid password', channel.json_body["error"]) + + # Must not have null bytes + body = json.dumps({"nonce": nonce(), "username": "a", "password": b"abcd\x00"}) + request, channel = make_request("POST", self.url, body.encode('utf8')) + render(request, self.resource, self.clock) + + self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual('Invalid password', channel.json_body["error"]) + + # Super long + body = json.dumps({"nonce": nonce(), "username": "a", "password": "A" * 1000}) + request, channel = make_request("POST", self.url, body.encode('utf8')) + render(request, self.resource, self.clock) + + self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual('Invalid password', channel.json_body["error"]) diff --git a/tests/rest/client/v1/test_events.py b/tests/rest/client/v1/test_events.py index e9698bfdc..50418153f 100644 --- a/tests/rest/client/v1/test_events.py +++ b/tests/rest/client/v1/test_events.py @@ -14,107 +14,35 @@ # limitations under the License. """ Tests REST events for /events paths.""" -from tests import unittest -# twisted imports +from mock import Mock, NonCallableMock +from six import PY3 + from twisted.internet import defer -import synapse.rest.client.v1.events -import synapse.rest.client.v1.register -import synapse.rest.client.v1.room - - from ....utils import MockHttpResource, setup_test_homeserver from .utils import RestTestCase -from mock import Mock, NonCallableMock - - PATH_PREFIX = "/_matrix/client/api/v1" -class EventStreamPaginationApiTestCase(unittest.TestCase): - """ Tests event streaming query parameters and start/end keys used in the - Pagination stream API. """ - user_id = "sid1" - - def setUp(self): - # configure stream and inject items - pass - - def tearDown(self): - pass - - def TODO_test_long_poll(self): - # stream from 'end' key, send (self+other) message, expect message. - - # stream from 'END', send (self+other) message, expect message. - - # stream from 'end' key, send (self+other) topic, expect topic. - - # stream from 'END', send (self+other) topic, expect topic. - - # stream from 'end' key, send (self+other) invite, expect invite. - - # stream from 'END', send (self+other) invite, expect invite. - - pass - - def TODO_test_stream_forward(self): - # stream from START, expect injected items - - # stream from 'start' key, expect same content - - # stream from 'end' key, expect nothing - - # stream from 'END', expect nothing - - # The following is needed for cases where content is removed e.g. you - # left a room, so the token you're streaming from is > the one that - # would be returned naturally from START>END. - # stream from very new token (higher than end key), expect same token - # returned as end key - pass - - def TODO_test_limits(self): - # stream from a key, expect limit_num items - - # stream from START, expect limit_num items - - pass - - def TODO_test_range(self): - # stream from key to key, expect X items - - # stream from key to END, expect X items - - # stream from START to key, expect X items - - # stream from START to END, expect all items - pass - - def TODO_test_direction(self): - # stream from END to START and fwds, expect newest first - - # stream from END to START and bwds, expect oldest first - - # stream from START to END and fwds, expect oldest first - - # stream from START to END and bwds, expect newest first - - pass - - class EventStreamPermissionsTestCase(RestTestCase): """ Tests event streaming (GET /events). """ + if PY3: + skip = "Skip on Py3 until ported to use not V1 only register." + @defer.inlineCallbacks def setUp(self): + import synapse.rest.client.v1.events + import synapse.rest.client.v1_only.register + import synapse.rest.client.v1.room + self.mock_resource = MockHttpResource(prefix=PATH_PREFIX) hs = yield setup_test_homeserver( http_client=None, - replication_layer=Mock(), + federation_client=Mock(), ratelimiter=NonCallableMock(spec_set=[ "send_message", ]), @@ -123,10 +51,11 @@ class EventStreamPermissionsTestCase(RestTestCase): self.ratelimiter.send_message.return_value = (True, 0) hs.config.enable_registration_captcha = False hs.config.enable_registration = True + hs.config.auto_join_rooms = [] hs.get_handlers().federation_handler = Mock() - synapse.rest.client.v1.register.register_servlets(hs, self.mock_resource) + synapse.rest.client.v1_only.register.register_servlets(hs, self.mock_resource) synapse.rest.client.v1.events.register_servlets(hs, self.mock_resource) synapse.rest.client.v1.room.register_servlets(hs, self.mock_resource) @@ -147,11 +76,16 @@ class EventStreamPermissionsTestCase(RestTestCase): @defer.inlineCallbacks def test_stream_basic_permissions(self): - # invalid token, expect 403 + # invalid token, expect 401 + # note: this is in violation of the original v1 spec, which expected + # 403. However, since the v1 spec no longer exists and the v1 + # implementation is now part of the r0 implementation, the newer + # behaviour is used instead to be consistent with the r0 spec. + # see issue #2602 (code, response) = yield self.mock_resource.trigger_get( "/events?access_token=%s" % ("invalid" + self.token, ) ) - self.assertEquals(403, code, msg=str(response)) + self.assertEquals(401, code, msg=str(response)) # valid token, expect content (code, response) = yield self.mock_resource.trigger_get( diff --git a/tests/rest/client/v1/test_profile.py b/tests/rest/client/v1/test_profile.py index 1e95e9753..d71cc8e0d 100644 --- a/tests/rest/client/v1/test_profile.py +++ b/tests/rest/client/v1/test_profile.py @@ -15,12 +15,15 @@ """Tests REST events for /profile paths.""" from mock import Mock + from twisted.internet import defer import synapse.types -from synapse.api.errors import SynapseError, AuthError +from synapse.api.errors import AuthError, SynapseError from synapse.rest.client.v1 import profile + from tests import unittest + from ....utils import MockHttpResource, setup_test_homeserver myid = "@1234ABCD:test" @@ -45,15 +48,14 @@ class ProfileTestCase(unittest.TestCase): http_client=None, resource_for_client=self.mock_resource, federation=Mock(), - replication_layer=Mock(), + federation_client=Mock(), + profile_handler=self.mock_handler ) def _get_user_by_req(request=None, allow_guest=False): return synapse.types.create_requester(myid) - hs.get_v1auth().get_user_by_req = _get_user_by_req - - hs.get_handlers().profile_handler = self.mock_handler + hs.get_auth().get_user_by_req = _get_user_by_req profile.register_servlets(hs, self.mock_resource) diff --git a/tests/rest/client/v1/test_register.py b/tests/rest/client/v1/test_register.py index a6a4e2ffe..83a23cd8f 100644 --- a/tests/rest/client/v1/test_register.py +++ b/tests/rest/client/v1/test_register.py @@ -13,26 +13,29 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.rest.client.v1.register import CreateUserRestServlet -from twisted.internet import defer -from mock import Mock -from tests import unittest -from tests.utils import mock_getRawHeaders import json +from mock import Mock +from six import PY3 + +from twisted.test.proto_helpers import MemoryReactorClock + +from synapse.http.server import JsonResource +from synapse.rest.client.v1_only.register import register_servlets +from synapse.util import Clock + +from tests import unittest +from tests.server import make_request, setup_test_homeserver + class CreateUserServletTestCase(unittest.TestCase): + """ + Tests for CreateUserRestServlet. + """ + if PY3: + skip = "Not ported to Python 3." def setUp(self): - # do the dance to hook up request data to self.request_data - self.request_data = "" - self.request = Mock( - content=Mock(read=Mock(side_effect=lambda: self.request_data)), - path='/_matrix/client/api/v1/createUser' - ) - self.request.args = {} - self.request.requestHeaders.getRawHeaders = mock_getRawHeaders() - self.registration_handler = Mock() self.appservice = Mock(sender="@as:test") @@ -40,39 +43,49 @@ class CreateUserServletTestCase(unittest.TestCase): get_app_service_by_token=Mock(return_value=self.appservice) ) - # do the dance to hook things up to the hs global - handlers = Mock( - registration_handler=self.registration_handler, + handlers = Mock(registration_handler=self.registration_handler) + self.clock = MemoryReactorClock() + self.hs_clock = Clock(self.clock) + + self.hs = self.hs = setup_test_homeserver( + http_client=None, clock=self.hs_clock, reactor=self.clock ) - self.hs = Mock() - self.hs.hostname = "superbig~testing~thing.com" self.hs.get_datastore = Mock(return_value=self.datastore) self.hs.get_handlers = Mock(return_value=handlers) - self.servlet = CreateUserRestServlet(self.hs) - @defer.inlineCallbacks def test_POST_createuser_with_valid_user(self): + + res = JsonResource(self.hs) + register_servlets(self.hs, res) + + request_data = json.dumps( + { + "localpart": "someone", + "displayname": "someone interesting", + "duration_seconds": 200, + } + ) + + url = b'/_matrix/client/api/v1/createUser?access_token=i_am_an_app_service' + user_id = "@someone:interesting" token = "my token" - self.request.args = { - "access_token": "i_am_an_app_service" - } - self.request_data = json.dumps({ - "localpart": "someone", - "displayname": "someone interesting", - "duration_seconds": 200 - }) self.registration_handler.get_or_create_user = Mock( return_value=(user_id, token) ) - (code, result) = yield self.servlet.on_POST(self.request) - self.assertEquals(code, 200) + request, channel = make_request(b"POST", url, request_data) + request.render(res) + + # Advance the clock because it waits + self.clock.advance(1) + + self.assertEquals(channel.result["code"], b"200") det_data = { "user_id": user_id, "access_token": token, - "home_server": self.hs.hostname + "home_server": self.hs.hostname, } - self.assertDictContainsSubset(det_data, result) + self.assertDictContainsSubset(det_data, json.loads(channel.result["body"])) diff --git a/tests/rest/client/v1/test_rooms.py b/tests/rest/client/v1/test_rooms.py index d746ea856..00fc79678 100644 --- a/tests/rest/client/v1/test_rooms.py +++ b/tests/rest/client/v1/test_rooms.py @@ -15,963 +15,782 @@ """Tests REST events for /rooms paths.""" -# twisted imports +import json + +from mock import Mock, NonCallableMock +from six.moves.urllib import parse as urlparse + from twisted.internet import defer import synapse.rest.client.v1.room from synapse.api.constants import Membership - +from synapse.http.server import JsonResource from synapse.types import UserID +from synapse.util import Clock -import json -import urllib +from tests import unittest +from tests.server import ( + ThreadedMemoryReactorClock, + make_request, + render, + setup_test_homeserver, +) -from ....utils import MockHttpResource, setup_test_homeserver -from .utils import RestTestCase +from .utils import RestHelper -from mock import Mock, NonCallableMock - -PATH_PREFIX = "/_matrix/client/api/v1" +PATH_PREFIX = b"/_matrix/client/api/v1" -class RoomPermissionsTestCase(RestTestCase): - """ Tests room permissions. """ - user_id = "@sid1:red" - rmcreator_id = "@notme:red" +class RoomBase(unittest.TestCase): + rmcreator_id = None - @defer.inlineCallbacks def setUp(self): - self.mock_resource = MockHttpResource(prefix=PATH_PREFIX) - hs = yield setup_test_homeserver( + self.clock = ThreadedMemoryReactorClock() + self.hs_clock = Clock(self.clock) + + self.hs = setup_test_homeserver( "red", http_client=None, - replication_layer=Mock(), + clock=self.hs_clock, + reactor=self.clock, + federation_client=Mock(), ratelimiter=NonCallableMock(spec_set=["send_message"]), ) - self.ratelimiter = hs.get_ratelimiter() + self.ratelimiter = self.hs.get_ratelimiter() self.ratelimiter.send_message.return_value = (True, 0) - hs.get_handlers().federation_handler = Mock() + self.hs.get_federation_handler = Mock(return_value=Mock()) def get_user_by_access_token(token=None, allow_guest=False): return { - "user": UserID.from_string(self.auth_user_id), + "user": UserID.from_string(self.helper.auth_user_id), "token_id": 1, "is_guest": False, } - hs.get_v1auth().get_user_by_access_token = get_user_by_access_token + + def get_user_by_req(request, allow_guest=False, rights="access"): + return synapse.types.create_requester( + UserID.from_string(self.helper.auth_user_id), 1, False, None + ) + + self.hs.get_auth().get_user_by_req = get_user_by_req + self.hs.get_auth().get_user_by_access_token = get_user_by_access_token + self.hs.get_auth().get_access_token_from_request = Mock(return_value=b"1234") def _insert_client_ip(*args, **kwargs): return defer.succeed(None) - hs.get_datastore().insert_client_ip = _insert_client_ip - self.auth_user_id = self.rmcreator_id + self.hs.get_datastore().insert_client_ip = _insert_client_ip - synapse.rest.client.v1.room.register_servlets(hs, self.mock_resource) + self.resource = JsonResource(self.hs) + synapse.rest.client.v1.room.register_servlets(self.hs, self.resource) + synapse.rest.client.v1.room.register_deprecated_servlets(self.hs, self.resource) + self.helper = RestHelper(self.hs, self.resource, self.user_id) - self.auth = hs.get_v1auth() +class RoomPermissionsTestCase(RoomBase): + """ Tests room permissions. """ + + user_id = b"@sid1:red" + rmcreator_id = b"@notme:red" + + def setUp(self): + + super(RoomPermissionsTestCase, self).setUp() + + self.helper.auth_user_id = self.rmcreator_id # create some rooms under the name rmcreator_id self.uncreated_rmid = "!aa:test" - - self.created_rmid = yield self.create_room_as(self.rmcreator_id, - is_public=False) - - self.created_public_rmid = yield self.create_room_as(self.rmcreator_id, - is_public=True) + self.created_rmid = self.helper.create_room_as( + self.rmcreator_id, is_public=False + ) + self.created_public_rmid = self.helper.create_room_as( + self.rmcreator_id, is_public=True + ) # send a message in one of the rooms self.created_rmid_msg_path = ( - "/rooms/%s/send/m.room.message/a1" % (self.created_rmid) - ) - (code, response) = yield self.mock_resource.trigger( - "PUT", + "rooms/%s/send/m.room.message/a1" % (self.created_rmid) + ).encode('ascii') + request, channel = make_request( + b"PUT", self.created_rmid_msg_path, - '{"msgtype":"m.text","body":"test msg"}' + b'{"msgtype":"m.text","body":"test msg"}', ) - self.assertEquals(200, code, msg=str(response)) + render(request, self.resource, self.clock) + self.assertEquals(channel.result["code"], b"200", channel.result) # set topic for public room - (code, response) = yield self.mock_resource.trigger( - "PUT", - "/rooms/%s/state/m.room.topic" % self.created_public_rmid, - '{"topic":"Public Room Topic"}' + request, channel = make_request( + b"PUT", + ("rooms/%s/state/m.room.topic" % self.created_public_rmid).encode('ascii'), + b'{"topic":"Public Room Topic"}', ) - self.assertEquals(200, code, msg=str(response)) + render(request, self.resource, self.clock) + self.assertEquals(channel.result["code"], b"200", channel.result) # auth as user_id now - self.auth_user_id = self.user_id + self.helper.auth_user_id = self.user_id - def tearDown(self): - pass - - @defer.inlineCallbacks def test_send_message(self): - msg_content = '{"msgtype":"m.text","body":"hello"}' - send_msg_path = ( - "/rooms/%s/send/m.room.message/mid1" % (self.created_rmid,) - ) + msg_content = b'{"msgtype":"m.text","body":"hello"}' + + seq = iter(range(100)) + + def send_msg_path(): + return b"/rooms/%s/send/m.room.message/mid%s" % ( + self.created_rmid, + str(next(seq)).encode('ascii'), + ) # send message in uncreated room, expect 403 - (code, response) = yield self.mock_resource.trigger( - "PUT", - "/rooms/%s/send/m.room.message/mid2" % (self.uncreated_rmid,), - msg_content + request, channel = make_request( + b"PUT", + b"/rooms/%s/send/m.room.message/mid2" % (self.uncreated_rmid,), + msg_content, ) - self.assertEquals(403, code, msg=str(response)) + render(request, self.resource, self.clock) + self.assertEquals(403, int(channel.result["code"]), msg=channel.result["body"]) # send message in created room not joined (no state), expect 403 - (code, response) = yield self.mock_resource.trigger( - "PUT", - send_msg_path, - msg_content - ) - self.assertEquals(403, code, msg=str(response)) + request, channel = make_request(b"PUT", send_msg_path(), msg_content) + render(request, self.resource, self.clock) + self.assertEquals(403, int(channel.result["code"]), msg=channel.result["body"]) # send message in created room and invited, expect 403 - yield self.invite( - room=self.created_rmid, - src=self.rmcreator_id, - targ=self.user_id - ) - (code, response) = yield self.mock_resource.trigger( - "PUT", - send_msg_path, - msg_content - ) - self.assertEquals(403, code, msg=str(response)) - - # send message in created room and joined, expect 200 - yield self.join(room=self.created_rmid, user=self.user_id) - (code, response) = yield self.mock_resource.trigger( - "PUT", - send_msg_path, - msg_content - ) - self.assertEquals(200, code, msg=str(response)) - - # send message in created room and left, expect 403 - yield self.leave(room=self.created_rmid, user=self.user_id) - (code, response) = yield self.mock_resource.trigger( - "PUT", - send_msg_path, - msg_content - ) - self.assertEquals(403, code, msg=str(response)) - - @defer.inlineCallbacks - def test_topic_perms(self): - topic_content = '{"topic":"My Topic Name"}' - topic_path = "/rooms/%s/state/m.room.topic" % self.created_rmid - - # set/get topic in uncreated room, expect 403 - (code, response) = yield self.mock_resource.trigger( - "PUT", "/rooms/%s/state/m.room.topic" % self.uncreated_rmid, - topic_content - ) - self.assertEquals(403, code, msg=str(response)) - (code, response) = yield self.mock_resource.trigger_get( - "/rooms/%s/state/m.room.topic" % self.uncreated_rmid - ) - self.assertEquals(403, code, msg=str(response)) - - # set/get topic in created PRIVATE room not joined, expect 403 - (code, response) = yield self.mock_resource.trigger( - "PUT", topic_path, topic_content - ) - self.assertEquals(403, code, msg=str(response)) - (code, response) = yield self.mock_resource.trigger_get(topic_path) - self.assertEquals(403, code, msg=str(response)) - - # set topic in created PRIVATE room and invited, expect 403 - yield self.invite( + self.helper.invite( room=self.created_rmid, src=self.rmcreator_id, targ=self.user_id ) - (code, response) = yield self.mock_resource.trigger( - "PUT", topic_path, topic_content + request, channel = make_request(b"PUT", send_msg_path(), msg_content) + render(request, self.resource, self.clock) + self.assertEquals(403, int(channel.result["code"]), msg=channel.result["body"]) + + # send message in created room and joined, expect 200 + self.helper.join(room=self.created_rmid, user=self.user_id) + request, channel = make_request(b"PUT", send_msg_path(), msg_content) + render(request, self.resource, self.clock) + self.assertEquals(200, int(channel.result["code"]), msg=channel.result["body"]) + + # send message in created room and left, expect 403 + self.helper.leave(room=self.created_rmid, user=self.user_id) + request, channel = make_request(b"PUT", send_msg_path(), msg_content) + render(request, self.resource, self.clock) + self.assertEquals(403, int(channel.result["code"]), msg=channel.result["body"]) + + def test_topic_perms(self): + topic_content = b'{"topic":"My Topic Name"}' + topic_path = b"/rooms/%s/state/m.room.topic" % self.created_rmid + + # set/get topic in uncreated room, expect 403 + request, channel = make_request( + b"PUT", b"/rooms/%s/state/m.room.topic" % self.uncreated_rmid, topic_content ) - self.assertEquals(403, code, msg=str(response)) + render(request, self.resource, self.clock) + self.assertEquals(403, int(channel.result["code"]), msg=channel.result["body"]) + request, channel = make_request( + b"GET", "/rooms/%s/state/m.room.topic" % self.uncreated_rmid + ) + render(request, self.resource, self.clock) + self.assertEquals(403, int(channel.result["code"]), msg=channel.result["body"]) + + # set/get topic in created PRIVATE room not joined, expect 403 + request, channel = make_request(b"PUT", topic_path, topic_content) + render(request, self.resource, self.clock) + self.assertEquals(403, int(channel.result["code"]), msg=channel.result["body"]) + request, channel = make_request(b"GET", topic_path) + render(request, self.resource, self.clock) + self.assertEquals(403, int(channel.result["code"]), msg=channel.result["body"]) + + # set topic in created PRIVATE room and invited, expect 403 + self.helper.invite( + room=self.created_rmid, src=self.rmcreator_id, targ=self.user_id + ) + request, channel = make_request(b"PUT", topic_path, topic_content) + render(request, self.resource, self.clock) + self.assertEquals(403, int(channel.result["code"]), msg=channel.result["body"]) # get topic in created PRIVATE room and invited, expect 403 - (code, response) = yield self.mock_resource.trigger_get(topic_path) - self.assertEquals(403, code, msg=str(response)) + request, channel = make_request(b"GET", topic_path) + render(request, self.resource, self.clock) + self.assertEquals(403, int(channel.result["code"]), msg=channel.result["body"]) # set/get topic in created PRIVATE room and joined, expect 200 - yield self.join(room=self.created_rmid, user=self.user_id) + self.helper.join(room=self.created_rmid, user=self.user_id) # Only room ops can set topic by default - self.auth_user_id = self.rmcreator_id - (code, response) = yield self.mock_resource.trigger( - "PUT", topic_path, topic_content - ) - self.assertEquals(200, code, msg=str(response)) - self.auth_user_id = self.user_id + self.helper.auth_user_id = self.rmcreator_id + request, channel = make_request(b"PUT", topic_path, topic_content) + render(request, self.resource, self.clock) + self.assertEquals(200, int(channel.result["code"]), msg=channel.result["body"]) + self.helper.auth_user_id = self.user_id - (code, response) = yield self.mock_resource.trigger_get(topic_path) - self.assertEquals(200, code, msg=str(response)) - self.assert_dict(json.loads(topic_content), response) + request, channel = make_request(b"GET", topic_path) + render(request, self.resource, self.clock) + self.assertEquals(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assert_dict(json.loads(topic_content), channel.json_body) # set/get topic in created PRIVATE room and left, expect 403 - yield self.leave(room=self.created_rmid, user=self.user_id) - (code, response) = yield self.mock_resource.trigger( - "PUT", topic_path, topic_content - ) - self.assertEquals(403, code, msg=str(response)) - (code, response) = yield self.mock_resource.trigger_get(topic_path) - self.assertEquals(200, code, msg=str(response)) + self.helper.leave(room=self.created_rmid, user=self.user_id) + request, channel = make_request(b"PUT", topic_path, topic_content) + render(request, self.resource, self.clock) + self.assertEquals(403, int(channel.result["code"]), msg=channel.result["body"]) + request, channel = make_request(b"GET", topic_path) + render(request, self.resource, self.clock) + self.assertEquals(200, int(channel.result["code"]), msg=channel.result["body"]) # get topic in PUBLIC room, not joined, expect 403 - (code, response) = yield self.mock_resource.trigger_get( - "/rooms/%s/state/m.room.topic" % self.created_public_rmid + request, channel = make_request( + b"GET", b"/rooms/%s/state/m.room.topic" % self.created_public_rmid ) - self.assertEquals(403, code, msg=str(response)) + render(request, self.resource, self.clock) + self.assertEquals(403, int(channel.result["code"]), msg=channel.result["body"]) # set topic in PUBLIC room, not joined, expect 403 - (code, response) = yield self.mock_resource.trigger( - "PUT", - "/rooms/%s/state/m.room.topic" % self.created_public_rmid, - topic_content + request, channel = make_request( + b"PUT", + b"/rooms/%s/state/m.room.topic" % self.created_public_rmid, + topic_content, ) - self.assertEquals(403, code, msg=str(response)) + render(request, self.resource, self.clock) + self.assertEquals(403, int(channel.result["code"]), msg=channel.result["body"]) - @defer.inlineCallbacks def _test_get_membership(self, room=None, members=[], expect_code=None): for member in members: - path = "/rooms/%s/state/m.room.member/%s" % (room, member) - (code, response) = yield self.mock_resource.trigger_get(path) - self.assertEquals(expect_code, code) + path = b"/rooms/%s/state/m.room.member/%s" % (room, member) + request, channel = make_request(b"GET", path) + render(request, self.resource, self.clock) + self.assertEquals(expect_code, int(channel.result["code"])) - @defer.inlineCallbacks def test_membership_basic_room_perms(self): # === room does not exist === room = self.uncreated_rmid # get membership of self, get membership of other, uncreated room # expect all 403s - yield self._test_get_membership( - members=[self.user_id, self.rmcreator_id], - room=room, expect_code=403) + self._test_get_membership( + members=[self.user_id, self.rmcreator_id], room=room, expect_code=403 + ) # trying to invite people to this room should 403 - yield self.invite(room=room, src=self.user_id, targ=self.rmcreator_id, - expect_code=403) + self.helper.invite( + room=room, src=self.user_id, targ=self.rmcreator_id, expect_code=403 + ) # set [invite/join/left] of self, set [invite/join/left] of other, # expect all 404s because room doesn't exist on any server for usr in [self.user_id, self.rmcreator_id]: - yield self.join(room=room, user=usr, expect_code=404) - yield self.leave(room=room, user=usr, expect_code=404) + self.helper.join(room=room, user=usr, expect_code=404) + self.helper.leave(room=room, user=usr, expect_code=404) - @defer.inlineCallbacks def test_membership_private_room_perms(self): room = self.created_rmid # get membership of self, get membership of other, private room + invite # expect all 403s - yield self.invite(room=room, src=self.rmcreator_id, - targ=self.user_id) - yield self._test_get_membership( - members=[self.user_id, self.rmcreator_id], - room=room, expect_code=403) + self.helper.invite(room=room, src=self.rmcreator_id, targ=self.user_id) + self._test_get_membership( + members=[self.user_id, self.rmcreator_id], room=room, expect_code=403 + ) # get membership of self, get membership of other, private room + joined # expect all 200s - yield self.join(room=room, user=self.user_id) - yield self._test_get_membership( - members=[self.user_id, self.rmcreator_id], - room=room, expect_code=200) + self.helper.join(room=room, user=self.user_id) + self._test_get_membership( + members=[self.user_id, self.rmcreator_id], room=room, expect_code=200 + ) # get membership of self, get membership of other, private room + left # expect all 200s - yield self.leave(room=room, user=self.user_id) - yield self._test_get_membership( - members=[self.user_id, self.rmcreator_id], - room=room, expect_code=200) + self.helper.leave(room=room, user=self.user_id) + self._test_get_membership( + members=[self.user_id, self.rmcreator_id], room=room, expect_code=200 + ) - @defer.inlineCallbacks def test_membership_public_room_perms(self): room = self.created_public_rmid # get membership of self, get membership of other, public room + invite # expect 403 - yield self.invite(room=room, src=self.rmcreator_id, - targ=self.user_id) - yield self._test_get_membership( - members=[self.user_id, self.rmcreator_id], - room=room, expect_code=403) + self.helper.invite(room=room, src=self.rmcreator_id, targ=self.user_id) + self._test_get_membership( + members=[self.user_id, self.rmcreator_id], room=room, expect_code=403 + ) # get membership of self, get membership of other, public room + joined # expect all 200s - yield self.join(room=room, user=self.user_id) - yield self._test_get_membership( - members=[self.user_id, self.rmcreator_id], - room=room, expect_code=200) + self.helper.join(room=room, user=self.user_id) + self._test_get_membership( + members=[self.user_id, self.rmcreator_id], room=room, expect_code=200 + ) # get membership of self, get membership of other, public room + left # expect 200. - yield self.leave(room=room, user=self.user_id) - yield self._test_get_membership( - members=[self.user_id, self.rmcreator_id], - room=room, expect_code=200) + self.helper.leave(room=room, user=self.user_id) + self._test_get_membership( + members=[self.user_id, self.rmcreator_id], room=room, expect_code=200 + ) - @defer.inlineCallbacks def test_invited_permissions(self): room = self.created_rmid - yield self.invite(room=room, src=self.rmcreator_id, targ=self.user_id) + self.helper.invite(room=room, src=self.rmcreator_id, targ=self.user_id) # set [invite/join/left] of other user, expect 403s - yield self.invite(room=room, src=self.user_id, targ=self.rmcreator_id, - expect_code=403) - yield self.change_membership(room=room, src=self.user_id, - targ=self.rmcreator_id, - membership=Membership.JOIN, - expect_code=403) - yield self.change_membership(room=room, src=self.user_id, - targ=self.rmcreator_id, - membership=Membership.LEAVE, - expect_code=403) - - @defer.inlineCallbacks - def test_joined_permissions(self): - room = self.created_rmid - yield self.invite(room=room, src=self.rmcreator_id, targ=self.user_id) - yield self.join(room=room, user=self.user_id) - - # set invited of self, expect 403 - yield self.invite(room=room, src=self.user_id, targ=self.user_id, - expect_code=403) - - # set joined of self, expect 200 (NOOP) - yield self.join(room=room, user=self.user_id) - - other = "@burgundy:red" - # set invited of other, expect 200 - yield self.invite(room=room, src=self.user_id, targ=other, - expect_code=200) - - # set joined of other, expect 403 - yield self.change_membership(room=room, src=self.user_id, - targ=other, - membership=Membership.JOIN, - expect_code=403) - - # set left of other, expect 403 - yield self.change_membership(room=room, src=self.user_id, - targ=other, - membership=Membership.LEAVE, - expect_code=403) - - # set left of self, expect 200 - yield self.leave(room=room, user=self.user_id) - - @defer.inlineCallbacks - def test_leave_permissions(self): - room = self.created_rmid - yield self.invite(room=room, src=self.rmcreator_id, targ=self.user_id) - yield self.join(room=room, user=self.user_id) - yield self.leave(room=room, user=self.user_id) - - # set [invite/join/left] of self, set [invite/join/left] of other, - # expect all 403s - for usr in [self.user_id, self.rmcreator_id]: - yield self.change_membership( - room=room, - src=self.user_id, - targ=usr, - membership=Membership.INVITE, - expect_code=403 - ) - - yield self.change_membership( - room=room, - src=self.user_id, - targ=usr, - membership=Membership.JOIN, - expect_code=403 - ) - - # It is always valid to LEAVE if you've already left (currently.) - yield self.change_membership( + self.helper.invite( + room=room, src=self.user_id, targ=self.rmcreator_id, expect_code=403 + ) + self.helper.change_membership( + room=room, + src=self.user_id, + targ=self.rmcreator_id, + membership=Membership.JOIN, + expect_code=403, + ) + self.helper.change_membership( room=room, src=self.user_id, targ=self.rmcreator_id, membership=Membership.LEAVE, - expect_code=403 + expect_code=403, + ) + + def test_joined_permissions(self): + room = self.created_rmid + self.helper.invite(room=room, src=self.rmcreator_id, targ=self.user_id) + self.helper.join(room=room, user=self.user_id) + + # set invited of self, expect 403 + self.helper.invite( + room=room, src=self.user_id, targ=self.user_id, expect_code=403 + ) + + # set joined of self, expect 200 (NOOP) + self.helper.join(room=room, user=self.user_id) + + other = "@burgundy:red" + # set invited of other, expect 200 + self.helper.invite(room=room, src=self.user_id, targ=other, expect_code=200) + + # set joined of other, expect 403 + self.helper.change_membership( + room=room, + src=self.user_id, + targ=other, + membership=Membership.JOIN, + expect_code=403, + ) + + # set left of other, expect 403 + self.helper.change_membership( + room=room, + src=self.user_id, + targ=other, + membership=Membership.LEAVE, + expect_code=403, + ) + + # set left of self, expect 200 + self.helper.leave(room=room, user=self.user_id) + + def test_leave_permissions(self): + room = self.created_rmid + self.helper.invite(room=room, src=self.rmcreator_id, targ=self.user_id) + self.helper.join(room=room, user=self.user_id) + self.helper.leave(room=room, user=self.user_id) + + # set [invite/join/left] of self, set [invite/join/left] of other, + # expect all 403s + for usr in [self.user_id, self.rmcreator_id]: + self.helper.change_membership( + room=room, + src=self.user_id, + targ=usr, + membership=Membership.INVITE, + expect_code=403, + ) + + self.helper.change_membership( + room=room, + src=self.user_id, + targ=usr, + membership=Membership.JOIN, + expect_code=403, + ) + + # It is always valid to LEAVE if you've already left (currently.) + self.helper.change_membership( + room=room, + src=self.user_id, + targ=self.rmcreator_id, + membership=Membership.LEAVE, + expect_code=403, ) -class RoomsMemberListTestCase(RestTestCase): +class RoomsMemberListTestCase(RoomBase): """ Tests /rooms/$room_id/members/list REST events.""" - user_id = "@sid1:red" - @defer.inlineCallbacks - def setUp(self): - self.mock_resource = MockHttpResource(prefix=PATH_PREFIX) + user_id = b"@sid1:red" - hs = yield setup_test_homeserver( - "red", - http_client=None, - replication_layer=Mock(), - ratelimiter=NonCallableMock(spec_set=["send_message"]), - ) - self.ratelimiter = hs.get_ratelimiter() - self.ratelimiter.send_message.return_value = (True, 0) - - hs.get_handlers().federation_handler = Mock() - - self.auth_user_id = self.user_id - - def get_user_by_access_token(token=None, allow_guest=False): - return { - "user": UserID.from_string(self.auth_user_id), - "token_id": 1, - "is_guest": False, - } - hs.get_v1auth().get_user_by_access_token = get_user_by_access_token - - def _insert_client_ip(*args, **kwargs): - return defer.succeed(None) - hs.get_datastore().insert_client_ip = _insert_client_ip - - synapse.rest.client.v1.room.register_servlets(hs, self.mock_resource) - - def tearDown(self): - pass - - @defer.inlineCallbacks def test_get_member_list(self): - room_id = yield self.create_room_as(self.user_id) - (code, response) = yield self.mock_resource.trigger_get( - "/rooms/%s/members" % room_id - ) - self.assertEquals(200, code, msg=str(response)) + room_id = self.helper.create_room_as(self.user_id) + request, channel = make_request(b"GET", b"/rooms/%s/members" % room_id) + render(request, self.resource, self.clock) + self.assertEquals(200, int(channel.result["code"]), msg=channel.result["body"]) - @defer.inlineCallbacks def test_get_member_list_no_room(self): - (code, response) = yield self.mock_resource.trigger_get( - "/rooms/roomdoesnotexist/members" - ) - self.assertEquals(403, code, msg=str(response)) + request, channel = make_request(b"GET", b"/rooms/roomdoesnotexist/members") + render(request, self.resource, self.clock) + self.assertEquals(403, int(channel.result["code"]), msg=channel.result["body"]) - @defer.inlineCallbacks def test_get_member_list_no_permission(self): - room_id = yield self.create_room_as("@some_other_guy:red") - (code, response) = yield self.mock_resource.trigger_get( - "/rooms/%s/members" % room_id - ) - self.assertEquals(403, code, msg=str(response)) + room_id = self.helper.create_room_as(b"@some_other_guy:red") + request, channel = make_request(b"GET", b"/rooms/%s/members" % room_id) + render(request, self.resource, self.clock) + self.assertEquals(403, int(channel.result["code"]), msg=channel.result["body"]) - @defer.inlineCallbacks def test_get_member_list_mixed_memberships(self): - room_creator = "@some_other_guy:red" - room_id = yield self.create_room_as(room_creator) - room_path = "/rooms/%s/members" % room_id - yield self.invite(room=room_id, src=room_creator, - targ=self.user_id) + room_creator = b"@some_other_guy:red" + room_id = self.helper.create_room_as(room_creator) + room_path = b"/rooms/%s/members" % room_id + self.helper.invite(room=room_id, src=room_creator, targ=self.user_id) # can't see list if you're just invited. - (code, response) = yield self.mock_resource.trigger_get(room_path) - self.assertEquals(403, code, msg=str(response)) + request, channel = make_request(b"GET", room_path) + render(request, self.resource, self.clock) + self.assertEquals(403, int(channel.result["code"]), msg=channel.result["body"]) - yield self.join(room=room_id, user=self.user_id) + self.helper.join(room=room_id, user=self.user_id) # can see list now joined - (code, response) = yield self.mock_resource.trigger_get(room_path) - self.assertEquals(200, code, msg=str(response)) + request, channel = make_request(b"GET", room_path) + render(request, self.resource, self.clock) + self.assertEquals(200, int(channel.result["code"]), msg=channel.result["body"]) - yield self.leave(room=room_id, user=self.user_id) + self.helper.leave(room=room_id, user=self.user_id) # can see old list once left - (code, response) = yield self.mock_resource.trigger_get(room_path) - self.assertEquals(200, code, msg=str(response)) + request, channel = make_request(b"GET", room_path) + render(request, self.resource, self.clock) + self.assertEquals(200, int(channel.result["code"]), msg=channel.result["body"]) -class RoomsCreateTestCase(RestTestCase): +class RoomsCreateTestCase(RoomBase): """ Tests /rooms and /rooms/$room_id REST events. """ - user_id = "@sid1:red" - @defer.inlineCallbacks - def setUp(self): - self.mock_resource = MockHttpResource(prefix=PATH_PREFIX) - self.auth_user_id = self.user_id + user_id = b"@sid1:red" - hs = yield setup_test_homeserver( - "red", - http_client=None, - replication_layer=Mock(), - ratelimiter=NonCallableMock(spec_set=["send_message"]), - ) - self.ratelimiter = hs.get_ratelimiter() - self.ratelimiter.send_message.return_value = (True, 0) - - hs.get_handlers().federation_handler = Mock() - - def get_user_by_access_token(token=None, allow_guest=False): - return { - "user": UserID.from_string(self.auth_user_id), - "token_id": 1, - "is_guest": False, - } - hs.get_v1auth().get_user_by_access_token = get_user_by_access_token - - def _insert_client_ip(*args, **kwargs): - return defer.succeed(None) - hs.get_datastore().insert_client_ip = _insert_client_ip - - synapse.rest.client.v1.room.register_servlets(hs, self.mock_resource) - - def tearDown(self): - pass - - @defer.inlineCallbacks def test_post_room_no_keys(self): # POST with no config keys, expect new room id - (code, response) = yield self.mock_resource.trigger("POST", - "/createRoom", - "{}") - self.assertEquals(200, code, response) - self.assertTrue("room_id" in response) + request, channel = make_request(b"POST", b"/createRoom", b"{}") + + render(request, self.resource, self.clock) + self.assertEquals(200, int(channel.result["code"]), channel.result) + self.assertTrue("room_id" in channel.json_body) - @defer.inlineCallbacks def test_post_room_visibility_key(self): # POST with visibility config key, expect new room id - (code, response) = yield self.mock_resource.trigger( - "POST", - "/createRoom", - '{"visibility":"private"}') - self.assertEquals(200, code) - self.assertTrue("room_id" in response) + request, channel = make_request( + b"POST", b"/createRoom", b'{"visibility":"private"}' + ) + render(request, self.resource, self.clock) + self.assertEquals(200, int(channel.result["code"])) + self.assertTrue("room_id" in channel.json_body) - @defer.inlineCallbacks def test_post_room_custom_key(self): # POST with custom config keys, expect new room id - (code, response) = yield self.mock_resource.trigger( - "POST", - "/createRoom", - '{"custom":"stuff"}') - self.assertEquals(200, code) - self.assertTrue("room_id" in response) + request, channel = make_request(b"POST", b"/createRoom", b'{"custom":"stuff"}') + render(request, self.resource, self.clock) + self.assertEquals(200, int(channel.result["code"])) + self.assertTrue("room_id" in channel.json_body) - @defer.inlineCallbacks def test_post_room_known_and_unknown_keys(self): # POST with custom + known config keys, expect new room id - (code, response) = yield self.mock_resource.trigger( - "POST", - "/createRoom", - '{"visibility":"private","custom":"things"}') - self.assertEquals(200, code) - self.assertTrue("room_id" in response) + request, channel = make_request( + b"POST", b"/createRoom", b'{"visibility":"private","custom":"things"}' + ) + render(request, self.resource, self.clock) + self.assertEquals(200, int(channel.result["code"])) + self.assertTrue("room_id" in channel.json_body) - @defer.inlineCallbacks def test_post_room_invalid_content(self): # POST with invalid content / paths, expect 400 - (code, response) = yield self.mock_resource.trigger( - "POST", - "/createRoom", - '{"visibili') - self.assertEquals(400, code) + request, channel = make_request(b"POST", b"/createRoom", b'{"visibili') + render(request, self.resource, self.clock) + self.assertEquals(400, int(channel.result["code"])) - (code, response) = yield self.mock_resource.trigger( - "POST", - "/createRoom", - '["hello"]') - self.assertEquals(400, code) + request, channel = make_request(b"POST", b"/createRoom", b'["hello"]') + render(request, self.resource, self.clock) + self.assertEquals(400, int(channel.result["code"])) -class RoomTopicTestCase(RestTestCase): +class RoomTopicTestCase(RoomBase): """ Tests /rooms/$room_id/topic REST events. """ - user_id = "@sid1:red" - @defer.inlineCallbacks + user_id = b"@sid1:red" + def setUp(self): - self.mock_resource = MockHttpResource(prefix=PATH_PREFIX) - self.auth_user_id = self.user_id - hs = yield setup_test_homeserver( - "red", - http_client=None, - replication_layer=Mock(), - ratelimiter=NonCallableMock(spec_set=["send_message"]), - ) - self.ratelimiter = hs.get_ratelimiter() - self.ratelimiter.send_message.return_value = (True, 0) - - hs.get_handlers().federation_handler = Mock() - - def get_user_by_access_token(token=None, allow_guest=False): - return { - "user": UserID.from_string(self.auth_user_id), - "token_id": 1, - "is_guest": False, - } - - hs.get_v1auth().get_user_by_access_token = get_user_by_access_token - - def _insert_client_ip(*args, **kwargs): - return defer.succeed(None) - hs.get_datastore().insert_client_ip = _insert_client_ip - - synapse.rest.client.v1.room.register_servlets(hs, self.mock_resource) + super(RoomTopicTestCase, self).setUp() # create the room - self.room_id = yield self.create_room_as(self.user_id) - self.path = "/rooms/%s/state/m.room.topic" % (self.room_id,) + self.room_id = self.helper.create_room_as(self.user_id) + self.path = b"/rooms/%s/state/m.room.topic" % (self.room_id,) - def tearDown(self): - pass - - @defer.inlineCallbacks def test_invalid_puts(self): # missing keys or invalid json - (code, response) = yield self.mock_resource.trigger( - "PUT", self.path, '{}' - ) - self.assertEquals(400, code, msg=str(response)) + request, channel = make_request(b"PUT", self.path, '{}') + render(request, self.resource, self.clock) + self.assertEquals(400, int(channel.result["code"]), msg=channel.result["body"]) - (code, response) = yield self.mock_resource.trigger( - "PUT", self.path, '{"_name":"bob"}' - ) - self.assertEquals(400, code, msg=str(response)) + request, channel = make_request(b"PUT", self.path, '{"_name":"bob"}') + render(request, self.resource, self.clock) + self.assertEquals(400, int(channel.result["code"]), msg=channel.result["body"]) - (code, response) = yield self.mock_resource.trigger( - "PUT", self.path, '{"nao' - ) - self.assertEquals(400, code, msg=str(response)) + request, channel = make_request(b"PUT", self.path, '{"nao') + render(request, self.resource, self.clock) + self.assertEquals(400, int(channel.result["code"]), msg=channel.result["body"]) - (code, response) = yield self.mock_resource.trigger( - "PUT", self.path, '[{"_name":"bob"},{"_name":"jill"}]' + request, channel = make_request( + b"PUT", self.path, '[{"_name":"bob"},{"_name":"jill"}]' ) - self.assertEquals(400, code, msg=str(response)) + render(request, self.resource, self.clock) + self.assertEquals(400, int(channel.result["code"]), msg=channel.result["body"]) - (code, response) = yield self.mock_resource.trigger( - "PUT", self.path, 'text only' - ) - self.assertEquals(400, code, msg=str(response)) + request, channel = make_request(b"PUT", self.path, 'text only') + render(request, self.resource, self.clock) + self.assertEquals(400, int(channel.result["code"]), msg=channel.result["body"]) - (code, response) = yield self.mock_resource.trigger( - "PUT", self.path, '' - ) - self.assertEquals(400, code, msg=str(response)) + request, channel = make_request(b"PUT", self.path, '') + render(request, self.resource, self.clock) + self.assertEquals(400, int(channel.result["code"]), msg=channel.result["body"]) # valid key, wrong type content = '{"topic":["Topic name"]}' - (code, response) = yield self.mock_resource.trigger( - "PUT", self.path, content - ) - self.assertEquals(400, code, msg=str(response)) + request, channel = make_request(b"PUT", self.path, content) + render(request, self.resource, self.clock) + self.assertEquals(400, int(channel.result["code"]), msg=channel.result["body"]) - @defer.inlineCallbacks def test_rooms_topic(self): # nothing should be there - (code, response) = yield self.mock_resource.trigger_get(self.path) - self.assertEquals(404, code, msg=str(response)) + request, channel = make_request(b"GET", self.path) + render(request, self.resource, self.clock) + self.assertEquals(404, int(channel.result["code"]), msg=channel.result["body"]) # valid put content = '{"topic":"Topic name"}' - (code, response) = yield self.mock_resource.trigger( - "PUT", self.path, content - ) - self.assertEquals(200, code, msg=str(response)) + request, channel = make_request(b"PUT", self.path, content) + render(request, self.resource, self.clock) + self.assertEquals(200, int(channel.result["code"]), msg=channel.result["body"]) # valid get - (code, response) = yield self.mock_resource.trigger_get(self.path) - self.assertEquals(200, code, msg=str(response)) - self.assert_dict(json.loads(content), response) + request, channel = make_request(b"GET", self.path) + render(request, self.resource, self.clock) + self.assertEquals(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assert_dict(json.loads(content), channel.json_body) - @defer.inlineCallbacks def test_rooms_topic_with_extra_keys(self): # valid put with extra keys content = '{"topic":"Seasons","subtopic":"Summer"}' - (code, response) = yield self.mock_resource.trigger( - "PUT", self.path, content - ) - self.assertEquals(200, code, msg=str(response)) + request, channel = make_request(b"PUT", self.path, content) + render(request, self.resource, self.clock) + self.assertEquals(200, int(channel.result["code"]), msg=channel.result["body"]) # valid get - (code, response) = yield self.mock_resource.trigger_get(self.path) - self.assertEquals(200, code, msg=str(response)) - self.assert_dict(json.loads(content), response) + request, channel = make_request(b"GET", self.path) + render(request, self.resource, self.clock) + self.assertEquals(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assert_dict(json.loads(content), channel.json_body) -class RoomMemberStateTestCase(RestTestCase): +class RoomMemberStateTestCase(RoomBase): """ Tests /rooms/$room_id/members/$user_id/state REST events. """ - user_id = "@sid1:red" - @defer.inlineCallbacks + user_id = b"@sid1:red" + def setUp(self): - self.mock_resource = MockHttpResource(prefix=PATH_PREFIX) - self.auth_user_id = self.user_id - hs = yield setup_test_homeserver( - "red", - http_client=None, - replication_layer=Mock(), - ratelimiter=NonCallableMock(spec_set=["send_message"]), - ) - self.ratelimiter = hs.get_ratelimiter() - self.ratelimiter.send_message.return_value = (True, 0) - - hs.get_handlers().federation_handler = Mock() - - def get_user_by_access_token(token=None, allow_guest=False): - return { - "user": UserID.from_string(self.auth_user_id), - "token_id": 1, - "is_guest": False, - } - hs.get_v1auth().get_user_by_access_token = get_user_by_access_token - - def _insert_client_ip(*args, **kwargs): - return defer.succeed(None) - hs.get_datastore().insert_client_ip = _insert_client_ip - - synapse.rest.client.v1.room.register_servlets(hs, self.mock_resource) - - self.room_id = yield self.create_room_as(self.user_id) + super(RoomMemberStateTestCase, self).setUp() + self.room_id = self.helper.create_room_as(self.user_id) def tearDown(self): pass - @defer.inlineCallbacks def test_invalid_puts(self): path = "/rooms/%s/state/m.room.member/%s" % (self.room_id, self.user_id) # missing keys or invalid json - (code, response) = yield self.mock_resource.trigger("PUT", path, '{}') - self.assertEquals(400, code, msg=str(response)) + request, channel = make_request(b"PUT", path, '{}') + render(request, self.resource, self.clock) + self.assertEquals(400, int(channel.result["code"]), msg=channel.result["body"]) - (code, response) = yield self.mock_resource.trigger( - "PUT", path, '{"_name":"bob"}' - ) - self.assertEquals(400, code, msg=str(response)) + request, channel = make_request(b"PUT", path, '{"_name":"bob"}') + render(request, self.resource, self.clock) + self.assertEquals(400, int(channel.result["code"]), msg=channel.result["body"]) - (code, response) = yield self.mock_resource.trigger( - "PUT", path, '{"nao' - ) - self.assertEquals(400, code, msg=str(response)) + request, channel = make_request(b"PUT", path, '{"nao') + render(request, self.resource, self.clock) + self.assertEquals(400, int(channel.result["code"]), msg=channel.result["body"]) - (code, response) = yield self.mock_resource.trigger( - "PUT", path, '[{"_name":"bob"},{"_name":"jill"}]' + request, channel = make_request( + b"PUT", path, b'[{"_name":"bob"},{"_name":"jill"}]' ) - self.assertEquals(400, code, msg=str(response)) + render(request, self.resource, self.clock) + self.assertEquals(400, int(channel.result["code"]), msg=channel.result["body"]) - (code, response) = yield self.mock_resource.trigger( - "PUT", path, 'text only' - ) - self.assertEquals(400, code, msg=str(response)) + request, channel = make_request(b"PUT", path, 'text only') + render(request, self.resource, self.clock) + self.assertEquals(400, int(channel.result["code"]), msg=channel.result["body"]) - (code, response) = yield self.mock_resource.trigger( - "PUT", path, '' - ) - self.assertEquals(400, code, msg=str(response)) + request, channel = make_request(b"PUT", path, '') + render(request, self.resource, self.clock) + self.assertEquals(400, int(channel.result["code"]), msg=channel.result["body"]) # valid keys, wrong types - content = ('{"membership":["%s","%s","%s"]}' % ( - Membership.INVITE, Membership.JOIN, Membership.LEAVE - )) - (code, response) = yield self.mock_resource.trigger("PUT", path, content) - self.assertEquals(400, code, msg=str(response)) + content = '{"membership":["%s","%s","%s"]}' % ( + Membership.INVITE, + Membership.JOIN, + Membership.LEAVE, + ) + request, channel = make_request(b"PUT", path, content.encode('ascii')) + render(request, self.resource, self.clock) + self.assertEquals(400, int(channel.result["code"]), msg=channel.result["body"]) - @defer.inlineCallbacks def test_rooms_members_self(self): path = "/rooms/%s/state/m.room.member/%s" % ( - urllib.quote(self.room_id), self.user_id + urlparse.quote(self.room_id), + self.user_id, ) # valid join message (NOOP since we made the room) content = '{"membership":"%s"}' % Membership.JOIN - (code, response) = yield self.mock_resource.trigger("PUT", path, content) - self.assertEquals(200, code, msg=str(response)) + request, channel = make_request(b"PUT", path, content.encode('ascii')) + render(request, self.resource, self.clock) + self.assertEquals(200, int(channel.result["code"]), msg=channel.result["body"]) - (code, response) = yield self.mock_resource.trigger("GET", path, None) - self.assertEquals(200, code, msg=str(response)) + request, channel = make_request(b"GET", path, None) + render(request, self.resource, self.clock) + self.assertEquals(200, int(channel.result["code"]), msg=channel.result["body"]) - expected_response = { - "membership": Membership.JOIN, - } - self.assertEquals(expected_response, response) + expected_response = {"membership": Membership.JOIN} + self.assertEquals(expected_response, channel.json_body) - @defer.inlineCallbacks def test_rooms_members_other(self): self.other_id = "@zzsid1:red" path = "/rooms/%s/state/m.room.member/%s" % ( - urllib.quote(self.room_id), self.other_id + urlparse.quote(self.room_id), + self.other_id, ) # valid invite message content = '{"membership":"%s"}' % Membership.INVITE - (code, response) = yield self.mock_resource.trigger("PUT", path, content) - self.assertEquals(200, code, msg=str(response)) + request, channel = make_request(b"PUT", path, content) + render(request, self.resource, self.clock) + self.assertEquals(200, int(channel.result["code"]), msg=channel.result["body"]) - (code, response) = yield self.mock_resource.trigger("GET", path, None) - self.assertEquals(200, code, msg=str(response)) - self.assertEquals(json.loads(content), response) + request, channel = make_request(b"GET", path, None) + render(request, self.resource, self.clock) + self.assertEquals(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEquals(json.loads(content), channel.json_body) - @defer.inlineCallbacks def test_rooms_members_other_custom_keys(self): self.other_id = "@zzsid1:red" path = "/rooms/%s/state/m.room.member/%s" % ( - urllib.quote(self.room_id), self.other_id + urlparse.quote(self.room_id), + self.other_id, ) # valid invite message with custom key - content = ('{"membership":"%s","invite_text":"%s"}' % ( - Membership.INVITE, "Join us!" - )) - (code, response) = yield self.mock_resource.trigger("PUT", path, content) - self.assertEquals(200, code, msg=str(response)) + content = '{"membership":"%s","invite_text":"%s"}' % ( + Membership.INVITE, + "Join us!", + ) + request, channel = make_request(b"PUT", path, content) + render(request, self.resource, self.clock) + self.assertEquals(200, int(channel.result["code"]), msg=channel.result["body"]) - (code, response) = yield self.mock_resource.trigger("GET", path, None) - self.assertEquals(200, code, msg=str(response)) - self.assertEquals(json.loads(content), response) + request, channel = make_request(b"GET", path, None) + render(request, self.resource, self.clock) + self.assertEquals(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEquals(json.loads(content), channel.json_body) -class RoomMessagesTestCase(RestTestCase): +class RoomMessagesTestCase(RoomBase): """ Tests /rooms/$room_id/messages/$user_id/$msg_id REST events. """ + user_id = "@sid1:red" - @defer.inlineCallbacks def setUp(self): - self.mock_resource = MockHttpResource(prefix=PATH_PREFIX) - self.auth_user_id = self.user_id + super(RoomMessagesTestCase, self).setUp() - hs = yield setup_test_homeserver( - "red", - http_client=None, - replication_layer=Mock(), - ratelimiter=NonCallableMock(spec_set=["send_message"]), - ) - self.ratelimiter = hs.get_ratelimiter() - self.ratelimiter.send_message.return_value = (True, 0) + self.room_id = self.helper.create_room_as(self.user_id) - hs.get_handlers().federation_handler = Mock() - - def get_user_by_access_token(token=None, allow_guest=False): - return { - "user": UserID.from_string(self.auth_user_id), - "token_id": 1, - "is_guest": False, - } - hs.get_v1auth().get_user_by_access_token = get_user_by_access_token - - def _insert_client_ip(*args, **kwargs): - return defer.succeed(None) - hs.get_datastore().insert_client_ip = _insert_client_ip - - synapse.rest.client.v1.room.register_servlets(hs, self.mock_resource) - - self.room_id = yield self.create_room_as(self.user_id) - - def tearDown(self): - pass - - @defer.inlineCallbacks def test_invalid_puts(self): - path = "/rooms/%s/send/m.room.message/mid1" % ( - urllib.quote(self.room_id)) + path = "/rooms/%s/send/m.room.message/mid1" % (urlparse.quote(self.room_id)) # missing keys or invalid json - (code, response) = yield self.mock_resource.trigger( - "PUT", path, '{}' - ) - self.assertEquals(400, code, msg=str(response)) + request, channel = make_request(b"PUT", path, '{}') + render(request, self.resource, self.clock) + self.assertEquals(400, int(channel.result["code"]), msg=channel.result["body"]) - (code, response) = yield self.mock_resource.trigger( - "PUT", path, '{"_name":"bob"}' - ) - self.assertEquals(400, code, msg=str(response)) + request, channel = make_request(b"PUT", path, '{"_name":"bob"}') + render(request, self.resource, self.clock) + self.assertEquals(400, int(channel.result["code"]), msg=channel.result["body"]) - (code, response) = yield self.mock_resource.trigger( - "PUT", path, '{"nao' - ) - self.assertEquals(400, code, msg=str(response)) + request, channel = make_request(b"PUT", path, '{"nao') + render(request, self.resource, self.clock) + self.assertEquals(400, int(channel.result["code"]), msg=channel.result["body"]) - (code, response) = yield self.mock_resource.trigger( - "PUT", path, '[{"_name":"bob"},{"_name":"jill"}]' + request, channel = make_request( + b"PUT", path, '[{"_name":"bob"},{"_name":"jill"}]' ) - self.assertEquals(400, code, msg=str(response)) + render(request, self.resource, self.clock) + self.assertEquals(400, int(channel.result["code"]), msg=channel.result["body"]) - (code, response) = yield self.mock_resource.trigger( - "PUT", path, 'text only' - ) - self.assertEquals(400, code, msg=str(response)) + request, channel = make_request(b"PUT", path, 'text only') + render(request, self.resource, self.clock) + self.assertEquals(400, int(channel.result["code"]), msg=channel.result["body"]) - (code, response) = yield self.mock_resource.trigger( - "PUT", path, '' - ) - self.assertEquals(400, code, msg=str(response)) + request, channel = make_request(b"PUT", path, '') + render(request, self.resource, self.clock) + self.assertEquals(400, int(channel.result["code"]), msg=channel.result["body"]) - @defer.inlineCallbacks def test_rooms_messages_sent(self): - path = "/rooms/%s/send/m.room.message/mid1" % ( - urllib.quote(self.room_id)) + path = "/rooms/%s/send/m.room.message/mid1" % (urlparse.quote(self.room_id)) content = '{"body":"test","msgtype":{"type":"a"}}' - (code, response) = yield self.mock_resource.trigger("PUT", path, content) - self.assertEquals(400, code, msg=str(response)) + request, channel = make_request(b"PUT", path, content) + render(request, self.resource, self.clock) + self.assertEquals(400, int(channel.result["code"]), msg=channel.result["body"]) # custom message types content = '{"body":"test","msgtype":"test.custom.text"}' - (code, response) = yield self.mock_resource.trigger("PUT", path, content) - self.assertEquals(200, code, msg=str(response)) - -# (code, response) = yield self.mock_resource.trigger("GET", path, None) -# self.assertEquals(200, code, msg=str(response)) -# self.assert_dict(json.loads(content), response) + request, channel = make_request(b"PUT", path, content) + render(request, self.resource, self.clock) + self.assertEquals(200, int(channel.result["code"]), msg=channel.result["body"]) # m.text message type - path = "/rooms/%s/send/m.room.message/mid2" % ( - urllib.quote(self.room_id)) + path = "/rooms/%s/send/m.room.message/mid2" % (urlparse.quote(self.room_id)) content = '{"body":"test2","msgtype":"m.text"}' - (code, response) = yield self.mock_resource.trigger("PUT", path, content) - self.assertEquals(200, code, msg=str(response)) + request, channel = make_request(b"PUT", path, content) + render(request, self.resource, self.clock) + self.assertEquals(200, int(channel.result["code"]), msg=channel.result["body"]) -class RoomInitialSyncTestCase(RestTestCase): +class RoomInitialSyncTestCase(RoomBase): """ Tests /rooms/$room_id/initialSync. """ + user_id = "@sid1:red" - @defer.inlineCallbacks def setUp(self): - self.mock_resource = MockHttpResource(prefix=PATH_PREFIX) - self.auth_user_id = self.user_id - - hs = yield setup_test_homeserver( - "red", - http_client=None, - replication_layer=Mock(), - ratelimiter=NonCallableMock(spec_set=[ - "send_message", - ]), - ) - self.ratelimiter = hs.get_ratelimiter() - self.ratelimiter.send_message.return_value = (True, 0) - - hs.get_handlers().federation_handler = Mock() - - def get_user_by_access_token(token=None, allow_guest=False): - return { - "user": UserID.from_string(self.auth_user_id), - "token_id": 1, - "is_guest": False, - } - hs.get_v1auth().get_user_by_access_token = get_user_by_access_token - - def _insert_client_ip(*args, **kwargs): - return defer.succeed(None) - hs.get_datastore().insert_client_ip = _insert_client_ip - - synapse.rest.client.v1.room.register_servlets(hs, self.mock_resource) + super(RoomInitialSyncTestCase, self).setUp() # create the room - self.room_id = yield self.create_room_as(self.user_id) + self.room_id = self.helper.create_room_as(self.user_id) - @defer.inlineCallbacks def test_initial_sync(self): - (code, response) = yield self.mock_resource.trigger_get( - "/rooms/%s/initialSync" % self.room_id - ) - self.assertEquals(200, code) + request, channel = make_request(b"GET", "/rooms/%s/initialSync" % self.room_id) + render(request, self.resource, self.clock) + self.assertEquals(200, int(channel.result["code"])) - self.assertEquals(self.room_id, response["room_id"]) - self.assertEquals("join", response["membership"]) + self.assertEquals(self.room_id, channel.json_body["room_id"]) + self.assertEquals("join", channel.json_body["membership"]) # Room state is easier to assert on if we unpack it into a dict state = {} - for event in response["state"]: + for event in channel.json_body["state"]: if "state_key" not in event: continue t = event["type"] @@ -981,75 +800,48 @@ class RoomInitialSyncTestCase(RestTestCase): self.assertTrue("m.room.create" in state) - self.assertTrue("messages" in response) - self.assertTrue("chunk" in response["messages"]) - self.assertTrue("end" in response["messages"]) + self.assertTrue("messages" in channel.json_body) + self.assertTrue("chunk" in channel.json_body["messages"]) + self.assertTrue("end" in channel.json_body["messages"]) - self.assertTrue("presence" in response) + self.assertTrue("presence" in channel.json_body) presence_by_user = { - e["content"]["user_id"]: e for e in response["presence"] + e["content"]["user_id"]: e for e in channel.json_body["presence"] } self.assertTrue(self.user_id in presence_by_user) self.assertEquals("m.presence", presence_by_user[self.user_id]["type"]) -class RoomMessageListTestCase(RestTestCase): +class RoomMessageListTestCase(RoomBase): """ Tests /rooms/$room_id/messages REST events. """ + user_id = "@sid1:red" - @defer.inlineCallbacks def setUp(self): - self.mock_resource = MockHttpResource(prefix=PATH_PREFIX) - self.auth_user_id = self.user_id + super(RoomMessageListTestCase, self).setUp() + self.room_id = self.helper.create_room_as(self.user_id) - hs = yield setup_test_homeserver( - "red", - http_client=None, - replication_layer=Mock(), - ratelimiter=NonCallableMock(spec_set=["send_message"]), - ) - self.ratelimiter = hs.get_ratelimiter() - self.ratelimiter.send_message.return_value = (True, 0) - - hs.get_handlers().federation_handler = Mock() - - def get_user_by_access_token(token=None, allow_guest=False): - return { - "user": UserID.from_string(self.auth_user_id), - "token_id": 1, - "is_guest": False, - } - hs.get_v1auth().get_user_by_access_token = get_user_by_access_token - - def _insert_client_ip(*args, **kwargs): - return defer.succeed(None) - hs.get_datastore().insert_client_ip = _insert_client_ip - - synapse.rest.client.v1.room.register_servlets(hs, self.mock_resource) - - self.room_id = yield self.create_room_as(self.user_id) - - @defer.inlineCallbacks def test_topo_token_is_accepted(self): - token = "t1-0_0_0_0_0_0_0_0" - (code, response) = yield self.mock_resource.trigger_get( - "/rooms/%s/messages?access_token=x&from=%s" % - (self.room_id, token)) - self.assertEquals(200, code) - self.assertTrue("start" in response) - self.assertEquals(token, response['start']) - self.assertTrue("chunk" in response) - self.assertTrue("end" in response) + token = "t1-0_0_0_0_0_0_0_0_0" + request, channel = make_request( + b"GET", "/rooms/%s/messages?access_token=x&from=%s" % (self.room_id, token) + ) + render(request, self.resource, self.clock) + self.assertEquals(200, int(channel.result["code"])) + self.assertTrue("start" in channel.json_body) + self.assertEquals(token, channel.json_body['start']) + self.assertTrue("chunk" in channel.json_body) + self.assertTrue("end" in channel.json_body) - @defer.inlineCallbacks def test_stream_token_is_accepted_for_fwd_pagianation(self): - token = "s0_0_0_0_0_0_0_0" - (code, response) = yield self.mock_resource.trigger_get( - "/rooms/%s/messages?access_token=x&from=%s" % - (self.room_id, token)) - self.assertEquals(200, code) - self.assertTrue("start" in response) - self.assertEquals(token, response['start']) - self.assertTrue("chunk" in response) - self.assertTrue("end" in response) + token = "s0_0_0_0_0_0_0_0_0" + request, channel = make_request( + b"GET", "/rooms/%s/messages?access_token=x&from=%s" % (self.room_id, token) + ) + render(request, self.resource, self.clock) + self.assertEquals(200, int(channel.result["code"])) + self.assertTrue("start" in channel.json_body) + self.assertEquals(token, channel.json_body['start']) + self.assertTrue("chunk" in channel.json_body) + self.assertTrue("end" in channel.json_body) diff --git a/tests/rest/client/v1/test_typing.py b/tests/rest/client/v1/test_typing.py index a269e6f56..bddb3302e 100644 --- a/tests/rest/client/v1/test_typing.py +++ b/tests/rest/client/v1/test_typing.py @@ -15,18 +15,17 @@ """Tests REST events for /rooms paths.""" +from mock import Mock, NonCallableMock + # twisted imports from twisted.internet import defer import synapse.rest.client.v1.room from synapse.types import UserID -from ....utils import MockHttpResource, MockClock, setup_test_homeserver +from ....utils import MockClock, MockHttpResource, setup_test_homeserver from .utils import RestTestCase -from mock import Mock, NonCallableMock - - PATH_PREFIX = "/_matrix/client/api/v1" @@ -47,7 +46,7 @@ class RoomTypingTestCase(RestTestCase): "red", clock=self.clock, http_client=None, - replication_layer=Mock(), + federation_client=Mock(), ratelimiter=NonCallableMock(spec_set=[ "send_message", ]), @@ -68,7 +67,7 @@ class RoomTypingTestCase(RestTestCase): "is_guest": False, } - hs.get_v1auth().get_user_by_access_token = get_user_by_access_token + hs.get_auth().get_user_by_access_token = get_user_by_access_token def _insert_client_ip(*args, **kwargs): return defer.succeed(None) @@ -95,7 +94,7 @@ class RoomTypingTestCase(RestTestCase): else: if remotedomains is not None: remotedomains.add(member.domain) - hs.get_handlers().room_member_handler.fetch_room_distributions_into = ( + hs.get_room_member_handler().fetch_room_distributions_into = ( fetch_room_distributions_into ) diff --git a/tests/rest/client/v1/utils.py b/tests/rest/client/v1/utils.py index 3bb1dd003..41de8e076 100644 --- a/tests/rest/client/v1/utils.py +++ b/tests/rest/client/v1/utils.py @@ -13,16 +13,17 @@ # See the License for the specific language governing permissions and # limitations under the License. -# twisted imports -from twisted.internet import defer +import json +import time -# trial imports -from tests import unittest +import attr + +from twisted.internet import defer from synapse.api.constants import Membership -import json -import time +from tests import unittest +from tests.server import make_request, wait_until_result class RestTestCase(unittest.TestCase): @@ -133,3 +134,113 @@ class RestTestCase(unittest.TestCase): for key in required: self.assertEquals(required[key], actual[key], msg="%s mismatch. %s" % (key, actual)) + + +@attr.s +class RestHelper(object): + """Contains extra helper functions to quickly and clearly perform a given + REST action, which isn't the focus of the test. + """ + + hs = attr.ib() + resource = attr.ib() + auth_user_id = attr.ib() + + def create_room_as(self, room_creator, is_public=True, tok=None): + temp_id = self.auth_user_id + self.auth_user_id = room_creator + path = b"/_matrix/client/r0/createRoom" + content = {} + if not is_public: + content["visibility"] = "private" + if tok: + path = path + b"?access_token=%s" % tok.encode('ascii') + + request, channel = make_request(b"POST", path, json.dumps(content).encode('utf8')) + request.render(self.resource) + wait_until_result(self.hs.get_reactor(), channel) + + assert channel.result["code"] == b"200", channel.result + self.auth_user_id = temp_id + return channel.json_body["room_id"] + + def invite(self, room=None, src=None, targ=None, expect_code=200, tok=None): + self.change_membership( + room=room, + src=src, + targ=targ, + tok=tok, + membership=Membership.INVITE, + expect_code=expect_code, + ) + + def join(self, room=None, user=None, expect_code=200, tok=None): + self.change_membership( + room=room, + src=user, + targ=user, + tok=tok, + membership=Membership.JOIN, + expect_code=expect_code, + ) + + def leave(self, room=None, user=None, expect_code=200, tok=None): + self.change_membership( + room=room, + src=user, + targ=user, + tok=tok, + membership=Membership.LEAVE, + expect_code=expect_code, + ) + + def change_membership(self, room, src, targ, membership, tok=None, expect_code=200): + temp_id = self.auth_user_id + self.auth_user_id = src + + path = "/_matrix/client/r0/rooms/%s/state/m.room.member/%s" % (room, targ) + if tok: + path = path + "?access_token=%s" % tok + + data = {"membership": membership} + + request, channel = make_request( + b"PUT", path.encode('ascii'), json.dumps(data).encode('utf8') + ) + + request.render(self.resource) + wait_until_result(self.hs.get_reactor(), channel) + + assert int(channel.result["code"]) == expect_code, ( + "Expected: %d, got: %d, resp: %r" + % (expect_code, int(channel.result["code"]), channel.result["body"]) + ) + + self.auth_user_id = temp_id + + @defer.inlineCallbacks + def register(self, user_id): + (code, response) = yield self.mock_resource.trigger( + "POST", + "/_matrix/client/r0/register", + json.dumps( + {"user": user_id, "password": "test", "type": "m.login.password"} + ), + ) + self.assertEquals(200, code) + defer.returnValue(response) + + @defer.inlineCallbacks + def send(self, room_id, body=None, txn_id=None, tok=None, expect_code=200): + if txn_id is None: + txn_id = "m%s" % (str(time.time())) + if body is None: + body = "body_text_here" + + path = "/_matrix/client/r0/rooms/%s/send/m.room.message/%s" % (room_id, txn_id) + content = '{"msgtype":"m.text","body":"%s"}' % body + if tok: + path = path + "?access_token=%s" % tok + + (code, response) = yield self.mock_resource.trigger("PUT", path, content) + self.assertEquals(expect_code, code, msg=str(response)) diff --git a/tests/rest/client/v2_alpha/__init__.py b/tests/rest/client/v2_alpha/__init__.py index 5170217d9..e69de29bb 100644 --- a/tests/rest/client/v2_alpha/__init__.py +++ b/tests/rest/client/v2_alpha/__init__.py @@ -1,62 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2015, 2016 OpenMarket Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -from tests import unittest - -from mock import Mock - -from ....utils import MockHttpResource, setup_test_homeserver - -from synapse.types import UserID - -from twisted.internet import defer - - -PATH_PREFIX = "/_matrix/client/v2_alpha" - - -class V2AlphaRestTestCase(unittest.TestCase): - # Consumer must define - # USER_ID = - # TO_REGISTER = [] - - @defer.inlineCallbacks - def setUp(self): - self.mock_resource = MockHttpResource(prefix=PATH_PREFIX) - - hs = yield setup_test_homeserver( - datastore=self.make_datastore_mock(), - http_client=None, - resource_for_client=self.mock_resource, - resource_for_federation=self.mock_resource, - ) - - def get_user_by_access_token(token=None, allow_guest=False): - return { - "user": UserID.from_string(self.USER_ID), - "token_id": 1, - "is_guest": False, - } - hs.get_auth().get_user_by_access_token = get_user_by_access_token - - for r in self.TO_REGISTER: - r.register_servlets(hs, self.mock_resource) - - def make_datastore_mock(self): - store = Mock(spec=[ - "insert_client_ip", - ]) - store.get_app_service_by_token = Mock(return_value=None) - return store diff --git a/tests/rest/client/v2_alpha/test_filter.py b/tests/rest/client/v2_alpha/test_filter.py index 76b833e11..e890f0fea 100644 --- a/tests/rest/client/v2_alpha/test_filter.py +++ b/tests/rest/client/v2_alpha/test_filter.py @@ -13,38 +13,37 @@ # See the License for the specific language governing permissions and # limitations under the License. -from twisted.internet import defer +import synapse.types +from synapse.api.errors import Codes +from synapse.http.server import JsonResource +from synapse.rest.client.v2_alpha import filter +from synapse.types import UserID +from synapse.util import Clock from tests import unittest - -from synapse.rest.client.v2_alpha import filter - -from synapse.api.errors import Codes - -import synapse.types - -from synapse.types import UserID - -from ....utils import MockHttpResource, setup_test_homeserver +from tests.server import ( + ThreadedMemoryReactorClock as MemoryReactorClock, + make_request, + setup_test_homeserver, + wait_until_result, +) PATH_PREFIX = "/_matrix/client/v2_alpha" class FilterTestCase(unittest.TestCase): - USER_ID = "@apple:test" + USER_ID = b"@apple:test" EXAMPLE_FILTER = {"room": {"timeline": {"types": ["m.room.message"]}}} - EXAMPLE_FILTER_JSON = '{"room": {"timeline": {"types": ["m.room.message"]}}}' + EXAMPLE_FILTER_JSON = b'{"room": {"timeline": {"types": ["m.room.message"]}}}' TO_REGISTER = [filter] - @defer.inlineCallbacks def setUp(self): - self.mock_resource = MockHttpResource(prefix=PATH_PREFIX) + self.clock = MemoryReactorClock() + self.hs_clock = Clock(self.clock) - self.hs = yield setup_test_homeserver( - http_client=None, - resource_for_client=self.mock_resource, - resource_for_federation=self.mock_resource, + self.hs = setup_test_homeserver( + http_client=None, clock=self.hs_clock, reactor=self.clock ) self.auth = self.hs.get_auth() @@ -58,82 +57,103 @@ class FilterTestCase(unittest.TestCase): def get_user_by_req(request, allow_guest=False, rights="access"): return synapse.types.create_requester( - UserID.from_string(self.USER_ID), 1, False, None) + UserID.from_string(self.USER_ID), 1, False, None + ) self.auth.get_user_by_access_token = get_user_by_access_token self.auth.get_user_by_req = get_user_by_req self.store = self.hs.get_datastore() self.filtering = self.hs.get_filtering() + self.resource = JsonResource(self.hs) for r in self.TO_REGISTER: - r.register_servlets(self.hs, self.mock_resource) + r.register_servlets(self.hs, self.resource) - @defer.inlineCallbacks def test_add_filter(self): - (code, response) = yield self.mock_resource.trigger( - "POST", "/user/%s/filter" % (self.USER_ID), self.EXAMPLE_FILTER_JSON + request, channel = make_request( + b"POST", + b"/_matrix/client/r0/user/%s/filter" % (self.USER_ID), + self.EXAMPLE_FILTER_JSON, ) - self.assertEquals(200, code) - self.assertEquals({"filter_id": "0"}, response) - filter = yield self.store.get_user_filter( - user_localpart='apple', - filter_id=0, - ) - self.assertEquals(filter, self.EXAMPLE_FILTER) + request.render(self.resource) + wait_until_result(self.clock, channel) + + self.assertEqual(channel.result["code"], b"200") + self.assertEqual(channel.json_body, {"filter_id": "0"}) + filter = self.store.get_user_filter(user_localpart="apple", filter_id=0) + self.clock.advance(0) + self.assertEquals(filter.result, self.EXAMPLE_FILTER) - @defer.inlineCallbacks def test_add_filter_for_other_user(self): - (code, response) = yield self.mock_resource.trigger( - "POST", "/user/%s/filter" % ('@watermelon:test'), self.EXAMPLE_FILTER_JSON + request, channel = make_request( + b"POST", + b"/_matrix/client/r0/user/%s/filter" % (b"@watermelon:test"), + self.EXAMPLE_FILTER_JSON, ) - self.assertEquals(403, code) - self.assertEquals(response['errcode'], Codes.FORBIDDEN) + request.render(self.resource) + wait_until_result(self.clock, channel) + + self.assertEqual(channel.result["code"], b"403") + self.assertEquals(channel.json_body["errcode"], Codes.FORBIDDEN) - @defer.inlineCallbacks def test_add_filter_non_local_user(self): _is_mine = self.hs.is_mine self.hs.is_mine = lambda target_user: False - (code, response) = yield self.mock_resource.trigger( - "POST", "/user/%s/filter" % (self.USER_ID), self.EXAMPLE_FILTER_JSON + request, channel = make_request( + b"POST", + b"/_matrix/client/r0/user/%s/filter" % (self.USER_ID), + self.EXAMPLE_FILTER_JSON, ) + request.render(self.resource) + wait_until_result(self.clock, channel) + self.hs.is_mine = _is_mine - self.assertEquals(403, code) - self.assertEquals(response['errcode'], Codes.FORBIDDEN) + self.assertEqual(channel.result["code"], b"403") + self.assertEquals(channel.json_body["errcode"], Codes.FORBIDDEN) - @defer.inlineCallbacks def test_get_filter(self): - filter_id = yield self.filtering.add_user_filter( - user_localpart='apple', - user_filter=self.EXAMPLE_FILTER + filter_id = self.filtering.add_user_filter( + user_localpart="apple", user_filter=self.EXAMPLE_FILTER ) - (code, response) = yield self.mock_resource.trigger_get( - "/user/%s/filter/%s" % (self.USER_ID, filter_id) + self.clock.advance(1) + filter_id = filter_id.result + request, channel = make_request( + b"GET", b"/_matrix/client/r0/user/%s/filter/%s" % (self.USER_ID, filter_id) ) - self.assertEquals(200, code) - self.assertEquals(self.EXAMPLE_FILTER, response) + request.render(self.resource) + wait_until_result(self.clock, channel) + + self.assertEqual(channel.result["code"], b"200") + self.assertEquals(channel.json_body, self.EXAMPLE_FILTER) - @defer.inlineCallbacks def test_get_filter_non_existant(self): - (code, response) = yield self.mock_resource.trigger_get( - "/user/%s/filter/12382148321" % (self.USER_ID) + request, channel = make_request( + b"GET", "/_matrix/client/r0/user/%s/filter/12382148321" % (self.USER_ID) ) - self.assertEquals(400, code) - self.assertEquals(response['errcode'], Codes.NOT_FOUND) + request.render(self.resource) + wait_until_result(self.clock, channel) + + self.assertEqual(channel.result["code"], b"400") + self.assertEquals(channel.json_body["errcode"], Codes.NOT_FOUND) # Currently invalid params do not have an appropriate errcode # in errors.py - @defer.inlineCallbacks def test_get_filter_invalid_id(self): - (code, response) = yield self.mock_resource.trigger_get( - "/user/%s/filter/foobar" % (self.USER_ID) + request, channel = make_request( + b"GET", "/_matrix/client/r0/user/%s/filter/foobar" % (self.USER_ID) ) - self.assertEquals(400, code) + request.render(self.resource) + wait_until_result(self.clock, channel) + + self.assertEqual(channel.result["code"], b"400") # No ID also returns an invalid_id error - @defer.inlineCallbacks def test_get_filter_no_id(self): - (code, response) = yield self.mock_resource.trigger_get( - "/user/%s/filter/" % (self.USER_ID) + request, channel = make_request( + b"GET", "/_matrix/client/r0/user/%s/filter/" % (self.USER_ID) ) - self.assertEquals(400, code) + request.render(self.resource) + wait_until_result(self.clock, channel) + + self.assertEqual(channel.result["code"], b"400") diff --git a/tests/rest/client/v2_alpha/test_register.py b/tests/rest/client/v2_alpha/test_register.py index b6173ab2e..e004d8fc7 100644 --- a/tests/rest/client/v2_alpha/test_register.py +++ b/tests/rest/client/v2_alpha/test_register.py @@ -1,158 +1,193 @@ -from synapse.rest.client.v2_alpha.register import RegisterRestServlet -from synapse.api.errors import SynapseError -from twisted.internet import defer -from mock import Mock -from tests import unittest -from tests.utils import mock_getRawHeaders import json +from mock import Mock + +from twisted.python import failure +from twisted.test.proto_helpers import MemoryReactorClock + +from synapse.api.errors import InteractiveAuthIncompleteError +from synapse.http.server import JsonResource +from synapse.rest.client.v2_alpha.register import register_servlets +from synapse.util import Clock + +from tests import unittest +from tests.server import make_request, setup_test_homeserver, wait_until_result + class RegisterRestServletTestCase(unittest.TestCase): - def setUp(self): - # do the dance to hook up request data to self.request_data - self.request_data = "" - self.request = Mock( - content=Mock(read=Mock(side_effect=lambda: self.request_data)), - path='/_matrix/api/v2_alpha/register' - ) - self.request.args = {} - self.request.requestHeaders.getRawHeaders = mock_getRawHeaders() + + self.clock = MemoryReactorClock() + self.hs_clock = Clock(self.clock) + self.url = b"/_matrix/client/r0/register" self.appservice = None - self.auth = Mock(get_appservice_by_req=Mock( - side_effect=lambda x: self.appservice) + self.auth = Mock( + get_appservice_by_req=Mock(side_effect=lambda x: self.appservice) ) - self.auth_result = (False, None, None, None) + self.auth_result = failure.Failure(InteractiveAuthIncompleteError(None)) self.auth_handler = Mock( check_auth=Mock(side_effect=lambda x, y, z: self.auth_result), - get_session_data=Mock(return_value=None) + get_session_data=Mock(return_value=None), ) self.registration_handler = Mock() self.identity_handler = Mock() self.login_handler = Mock() self.device_handler = Mock() + self.device_handler.check_device_registered = Mock(return_value="FAKE") + + self.datastore = Mock(return_value=Mock()) + self.datastore.get_current_state_deltas = Mock(return_value=[]) # do the dance to hook it up to the hs global self.handlers = Mock( registration_handler=self.registration_handler, identity_handler=self.identity_handler, - login_handler=self.login_handler + login_handler=self.login_handler, + ) + self.hs = setup_test_homeserver( + http_client=None, clock=self.hs_clock, reactor=self.clock ) - self.hs = Mock() - self.hs.hostname = "superbig~testing~thing.com" self.hs.get_auth = Mock(return_value=self.auth) self.hs.get_handlers = Mock(return_value=self.handlers) self.hs.get_auth_handler = Mock(return_value=self.auth_handler) self.hs.get_device_handler = Mock(return_value=self.device_handler) + self.hs.get_datastore = Mock(return_value=self.datastore) self.hs.config.enable_registration = True + self.hs.config.registrations_require_3pid = [] + self.hs.config.auto_join_rooms = [] - # init the thing we're testing - self.servlet = RegisterRestServlet(self.hs) + self.resource = JsonResource(self.hs) + register_servlets(self.hs, self.resource) - @defer.inlineCallbacks def test_POST_appservice_registration_valid(self): user_id = "@kermit:muppet" token = "kermits_access_token" - self.request.args = { - "access_token": "i_am_an_app_service" - } - self.request_data = json.dumps({ - "username": "kermit" - }) - self.appservice = { - "id": "1234" - } - self.registration_handler.appservice_register = Mock( - return_value=user_id - ) - self.auth_handler.get_access_token_for_user_id = Mock( - return_value=token - ) + self.appservice = {"id": "1234"} + self.registration_handler.appservice_register = Mock(return_value=user_id) + self.auth_handler.get_access_token_for_user_id = Mock(return_value=token) + request_data = json.dumps({"username": "kermit"}) - (code, result) = yield self.servlet.on_POST(self.request) - self.assertEquals(code, 200) + request, channel = make_request( + b"POST", self.url + b"?access_token=i_am_an_app_service", request_data + ) + request.render(self.resource) + wait_until_result(self.clock, channel) + + self.assertEquals(channel.result["code"], b"200", channel.result) det_data = { "user_id": user_id, "access_token": token, - "home_server": self.hs.hostname + "home_server": self.hs.hostname, } - self.assertDictContainsSubset(det_data, result) + self.assertDictContainsSubset(det_data, json.loads(channel.result["body"])) - @defer.inlineCallbacks def test_POST_appservice_registration_invalid(self): - self.request.args = { - "access_token": "i_am_an_app_service" - } - self.request_data = json.dumps({ - "username": "kermit" - }) self.appservice = None # no application service exists - result = yield self.servlet.on_POST(self.request) - self.assertEquals(result, (401, None)) + request_data = json.dumps({"username": "kermit"}) + request, channel = make_request( + b"POST", self.url + b"?access_token=i_am_an_app_service", request_data + ) + request.render(self.resource) + wait_until_result(self.clock, channel) + + self.assertEquals(channel.result["code"], b"401", channel.result) def test_POST_bad_password(self): - self.request_data = json.dumps({ - "username": "kermit", - "password": 666 - }) - d = self.servlet.on_POST(self.request) - return self.assertFailure(d, SynapseError) + request_data = json.dumps({"username": "kermit", "password": 666}) + request, channel = make_request(b"POST", self.url, request_data) + request.render(self.resource) + wait_until_result(self.clock, channel) + + self.assertEquals(channel.result["code"], b"400", channel.result) + self.assertEquals( + json.loads(channel.result["body"])["error"], "Invalid password" + ) def test_POST_bad_username(self): - self.request_data = json.dumps({ - "username": 777, - "password": "monkey" - }) - d = self.servlet.on_POST(self.request) - return self.assertFailure(d, SynapseError) + request_data = json.dumps({"username": 777, "password": "monkey"}) + request, channel = make_request(b"POST", self.url, request_data) + request.render(self.resource) + wait_until_result(self.clock, channel) + + self.assertEquals(channel.result["code"], b"400", channel.result) + self.assertEquals( + json.loads(channel.result["body"])["error"], "Invalid username" + ) - @defer.inlineCallbacks def test_POST_user_valid(self): user_id = "@kermit:muppet" token = "kermits_access_token" device_id = "frogfone" - self.request_data = json.dumps({ - "username": "kermit", - "password": "monkey", - "device_id": device_id, - }) - self.registration_handler.check_username = Mock(return_value=True) - self.auth_result = (True, None, { - "username": "kermit", - "password": "monkey" - }, None) - self.registration_handler.register = Mock(return_value=(user_id, None)) - self.auth_handler.get_access_token_for_user_id = Mock( - return_value=token + request_data = json.dumps( + {"username": "kermit", "password": "monkey", "device_id": device_id} ) - self.device_handler.check_device_registered = \ - Mock(return_value=device_id) + self.registration_handler.check_username = Mock(return_value=True) + self.auth_result = (None, {"username": "kermit", "password": "monkey"}, None) + self.registration_handler.register = Mock(return_value=(user_id, None)) + self.auth_handler.get_access_token_for_user_id = Mock(return_value=token) + self.device_handler.check_device_registered = Mock(return_value=device_id) + + request, channel = make_request(b"POST", self.url, request_data) + request.render(self.resource) + wait_until_result(self.clock, channel) - (code, result) = yield self.servlet.on_POST(self.request) - self.assertEquals(code, 200) det_data = { "user_id": user_id, "access_token": token, "home_server": self.hs.hostname, "device_id": device_id, } - self.assertDictContainsSubset(det_data, result) + self.assertEquals(channel.result["code"], b"200", channel.result) + self.assertDictContainsSubset(det_data, json.loads(channel.result["body"])) self.auth_handler.get_login_tuple_for_user_id( - user_id, device_id=device_id, initial_device_display_name=None) + user_id, device_id=device_id, initial_device_display_name=None + ) def test_POST_disabled_registration(self): self.hs.config.enable_registration = False - self.request_data = json.dumps({ - "username": "kermit", - "password": "monkey" - }) + request_data = json.dumps({"username": "kermit", "password": "monkey"}) self.registration_handler.check_username = Mock(return_value=True) - self.auth_result = (True, None, { - "username": "kermit", - "password": "monkey" - }, None) + self.auth_result = (None, {"username": "kermit", "password": "monkey"}, None) self.registration_handler.register = Mock(return_value=("@user:id", "t")) - d = self.servlet.on_POST(self.request) - return self.assertFailure(d, SynapseError) + + request, channel = make_request(b"POST", self.url, request_data) + request.render(self.resource) + wait_until_result(self.clock, channel) + + self.assertEquals(channel.result["code"], b"403", channel.result) + self.assertEquals( + json.loads(channel.result["body"])["error"], + "Registration has been disabled", + ) + + def test_POST_guest_registration(self): + user_id = "a@b" + self.hs.config.macaroon_secret_key = "test" + self.hs.config.allow_guest_access = True + self.registration_handler.register = Mock(return_value=(user_id, None)) + + request, channel = make_request(b"POST", self.url + b"?kind=guest", b"{}") + request.render(self.resource) + wait_until_result(self.clock, channel) + + det_data = { + "user_id": user_id, + "home_server": self.hs.hostname, + "device_id": "guest_device", + } + self.assertEquals(channel.result["code"], b"200", channel.result) + self.assertDictContainsSubset(det_data, json.loads(channel.result["body"])) + + def test_POST_disabled_guest_registration(self): + self.hs.config.allow_guest_access = False + + request, channel = make_request(b"POST", self.url + b"?kind=guest", b"{}") + request.render(self.resource) + wait_until_result(self.clock, channel) + + self.assertEquals(channel.result["code"], b"403", channel.result) + self.assertEquals( + json.loads(channel.result["body"])["error"], "Guest access is disabled" + ) diff --git a/tests/rest/client/v2_alpha/test_sync.py b/tests/rest/client/v2_alpha/test_sync.py new file mode 100644 index 000000000..03ec3993b --- /dev/null +++ b/tests/rest/client/v2_alpha/test_sync.py @@ -0,0 +1,87 @@ +# -*- coding: utf-8 -*- +# Copyright 2018 New Vector +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import synapse.types +from synapse.http.server import JsonResource +from synapse.rest.client.v2_alpha import sync +from synapse.types import UserID +from synapse.util import Clock + +from tests import unittest +from tests.server import ( + ThreadedMemoryReactorClock as MemoryReactorClock, + make_request, + setup_test_homeserver, + wait_until_result, +) + +PATH_PREFIX = "/_matrix/client/v2_alpha" + + +class FilterTestCase(unittest.TestCase): + + USER_ID = b"@apple:test" + TO_REGISTER = [sync] + + def setUp(self): + self.clock = MemoryReactorClock() + self.hs_clock = Clock(self.clock) + + self.hs = setup_test_homeserver( + http_client=None, clock=self.hs_clock, reactor=self.clock + ) + + self.auth = self.hs.get_auth() + + def get_user_by_access_token(token=None, allow_guest=False): + return { + "user": UserID.from_string(self.USER_ID), + "token_id": 1, + "is_guest": False, + } + + def get_user_by_req(request, allow_guest=False, rights="access"): + return synapse.types.create_requester( + UserID.from_string(self.USER_ID), 1, False, None + ) + + self.auth.get_user_by_access_token = get_user_by_access_token + self.auth.get_user_by_req = get_user_by_req + + self.store = self.hs.get_datastore() + self.filtering = self.hs.get_filtering() + self.resource = JsonResource(self.hs) + + for r in self.TO_REGISTER: + r.register_servlets(self.hs, self.resource) + + def test_sync_argless(self): + request, channel = make_request(b"GET", b"/_matrix/client/r0/sync") + request.render(self.resource) + wait_until_result(self.clock, channel) + + self.assertEqual(channel.result["code"], b"200") + self.assertTrue( + set( + [ + "next_batch", + "rooms", + "presence", + "account_data", + "to_device", + "device_lists", + ] + ).issubset(set(channel.json_body.keys())) + ) diff --git a/tests/rest/media/__init__.py b/tests/rest/media/__init__.py new file mode 100644 index 000000000..a354d38ca --- /dev/null +++ b/tests/rest/media/__init__.py @@ -0,0 +1,14 @@ +# -*- coding: utf-8 -*- +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. diff --git a/tests/rest/media/v1/__init__.py b/tests/rest/media/v1/__init__.py new file mode 100644 index 000000000..a354d38ca --- /dev/null +++ b/tests/rest/media/v1/__init__.py @@ -0,0 +1,14 @@ +# -*- coding: utf-8 -*- +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. diff --git a/tests/rest/media/v1/test_media_storage.py b/tests/rest/media/v1/test_media_storage.py new file mode 100644 index 000000000..bf254a260 --- /dev/null +++ b/tests/rest/media/v1/test_media_storage.py @@ -0,0 +1,87 @@ +# -*- coding: utf-8 -*- +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + + +import os +import shutil +import tempfile + +from mock import Mock + +from twisted.internet import defer, reactor + +from synapse.rest.media.v1._base import FileInfo +from synapse.rest.media.v1.filepath import MediaFilePaths +from synapse.rest.media.v1.media_storage import MediaStorage +from synapse.rest.media.v1.storage_provider import FileStorageProviderBackend + +from tests import unittest + + +class MediaStorageTests(unittest.TestCase): + def setUp(self): + self.test_dir = tempfile.mkdtemp(prefix="synapse-tests-") + + self.primary_base_path = os.path.join(self.test_dir, "primary") + self.secondary_base_path = os.path.join(self.test_dir, "secondary") + + hs = Mock() + hs.get_reactor = Mock(return_value=reactor) + hs.config.media_store_path = self.primary_base_path + + storage_providers = [FileStorageProviderBackend( + hs, self.secondary_base_path + )] + + self.filepaths = MediaFilePaths(self.primary_base_path) + self.media_storage = MediaStorage( + hs, self.primary_base_path, self.filepaths, storage_providers, + ) + + def tearDown(self): + shutil.rmtree(self.test_dir) + + @defer.inlineCallbacks + def test_ensure_media_is_in_local_cache(self): + media_id = "some_media_id" + test_body = "Test\n" + + # First we create a file that is in a storage provider but not in the + # local primary media store + rel_path = self.filepaths.local_media_filepath_rel(media_id) + secondary_path = os.path.join(self.secondary_base_path, rel_path) + + os.makedirs(os.path.dirname(secondary_path)) + + with open(secondary_path, "w") as f: + f.write(test_body) + + # Now we run ensure_media_is_in_local_cache, which should copy the file + # to the local cache. + file_info = FileInfo(None, media_id) + local_path = yield self.media_storage.ensure_media_is_in_local_cache(file_info) + + self.assertTrue(os.path.exists(local_path)) + + # Asserts the file is under the expected local cache directory + self.assertEquals( + os.path.commonprefix([self.primary_base_path, local_path]), + self.primary_base_path, + ) + + with open(local_path) as f: + body = f.read() + + self.assertEqual(test_body, body) diff --git a/tests/server.py b/tests/server.py new file mode 100644 index 000000000..c611dd605 --- /dev/null +++ b/tests/server.py @@ -0,0 +1,193 @@ +import json +from io import BytesIO + +from six import text_type + +import attr + +from twisted.internet import threads +from twisted.internet.defer import Deferred +from twisted.python.failure import Failure +from twisted.test.proto_helpers import MemoryReactorClock + +from synapse.http.site import SynapseRequest + +from tests.utils import setup_test_homeserver as _sth + + +@attr.s +class FakeChannel(object): + """ + A fake Twisted Web Channel (the part that interfaces with the + wire). + """ + + result = attr.ib(default=attr.Factory(dict)) + + @property + def json_body(self): + if not self.result: + raise Exception("No result yet.") + return json.loads(self.result["body"]) + + def writeHeaders(self, version, code, reason, headers): + self.result["version"] = version + self.result["code"] = code + self.result["reason"] = reason + self.result["headers"] = headers + + def write(self, content): + if "body" not in self.result: + self.result["body"] = b"" + + self.result["body"] += content + + def requestDone(self, _self): + self.result["done"] = True + + def getPeer(self): + return None + + def getHost(self): + return None + + @property + def transport(self): + return self + + +class FakeSite: + """ + A fake Twisted Web Site, with mocks of the extra things that + Synapse adds. + """ + + server_version_string = b"1" + site_tag = "test" + + @property + def access_logger(self): + class FakeLogger: + def info(self, *args, **kwargs): + pass + + return FakeLogger() + + +def make_request(method, path, content=b""): + """ + Make a web request using the given method and path, feed it the + content, and return the Request and the Channel underneath. + """ + + # Decorate it to be the full path + if not path.startswith(b"/_matrix"): + path = b"/_matrix/client/r0/" + path + path = path.replace("//", "/") + + if isinstance(content, text_type): + content = content.encode('utf8') + + site = FakeSite() + channel = FakeChannel() + + req = SynapseRequest(site, channel) + req.process = lambda: b"" + req.content = BytesIO(content) + req.requestReceived(method, path, b"1.1") + + return req, channel + + +def wait_until_result(clock, channel, timeout=100): + """ + Wait until the channel has a result. + """ + clock.run() + x = 0 + + while not channel.result: + x += 1 + + if x > timeout: + raise Exception("Timed out waiting for request to finish.") + + clock.advance(0.1) + + +def render(request, resource, clock): + request.render(resource) + wait_until_result(clock, request._channel) + + +class ThreadedMemoryReactorClock(MemoryReactorClock): + """ + A MemoryReactorClock that supports callFromThread. + """ + def callFromThread(self, callback, *args, **kwargs): + """ + Make the callback fire in the next reactor iteration. + """ + d = Deferred() + d.addCallback(lambda x: callback(*args, **kwargs)) + self.callLater(0, d.callback, True) + return d + + +def setup_test_homeserver(*args, **kwargs): + """ + Set up a synchronous test server, driven by the reactor used by + the homeserver. + """ + d = _sth(*args, **kwargs).result + + # Make the thread pool synchronous. + clock = d.get_clock() + pool = d.get_db_pool() + + def runWithConnection(func, *args, **kwargs): + return threads.deferToThreadPool( + pool._reactor, + pool.threadpool, + pool._runWithConnection, + func, + *args, + **kwargs + ) + + def runInteraction(interaction, *args, **kwargs): + return threads.deferToThreadPool( + pool._reactor, + pool.threadpool, + pool._runInteraction, + interaction, + *args, + **kwargs + ) + + pool.runWithConnection = runWithConnection + pool.runInteraction = runInteraction + + class ThreadPool: + """ + Threadless thread pool. + """ + def start(self): + pass + + def callInThreadWithCallback(self, onResult, function, *args, **kwargs): + def _(res): + if isinstance(res, Failure): + onResult(False, res) + else: + onResult(True, res) + + d = Deferred() + d.addCallback(lambda x: function(*args, **kwargs)) + d.addBoth(_) + clock._reactor.callLater(0, d.callback, True) + return d + + clock.threadpool = ThreadPool() + pool.threadpool = ThreadPool() + return d diff --git a/tests/storage/event_injector.py b/tests/storage/event_injector.py deleted file mode 100644 index 024ac1506..000000000 --- a/tests/storage/event_injector.py +++ /dev/null @@ -1,76 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2015, 2016 OpenMarket Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - - -from twisted.internet import defer - -from synapse.api.constants import EventTypes - - -class EventInjector: - def __init__(self, hs): - self.hs = hs - self.store = hs.get_datastore() - self.message_handler = hs.get_handlers().message_handler - self.event_builder_factory = hs.get_event_builder_factory() - - @defer.inlineCallbacks - def create_room(self, room, user): - builder = self.event_builder_factory.new({ - "type": EventTypes.Create, - "sender": user.to_string(), - "room_id": room.to_string(), - "content": {}, - }) - - event, context = yield self.message_handler._create_new_client_event( - builder - ) - - yield self.store.persist_event(event, context) - - @defer.inlineCallbacks - def inject_room_member(self, room, user, membership): - builder = self.event_builder_factory.new({ - "type": EventTypes.Member, - "sender": user.to_string(), - "state_key": user.to_string(), - "room_id": room.to_string(), - "content": {"membership": membership}, - }) - - event, context = yield self.message_handler._create_new_client_event( - builder - ) - - yield self.store.persist_event(event, context) - - defer.returnValue(event) - - @defer.inlineCallbacks - def inject_message(self, room, user, body): - builder = self.event_builder_factory.new({ - "type": EventTypes.Message, - "sender": user.to_string(), - "state_key": user.to_string(), - "room_id": room.to_string(), - "content": {"body": body, "msgtype": u"message"}, - }) - - event, context = yield self.message_handler._create_new_client_event( - builder - ) - - yield self.store.persist_event(event, context) diff --git a/tests/storage/test__base.py b/tests/storage/test__base.py index 3cfa21c9f..6d6f00c5c 100644 --- a/tests/storage/test__base.py +++ b/tests/storage/test__base.py @@ -14,15 +14,15 @@ # limitations under the License. -from tests import unittest -from twisted.internet import defer - from mock import Mock -from synapse.util.async import ObservableDeferred +from twisted.internet import defer +from synapse.util.async import ObservableDeferred from synapse.util.caches.descriptors import Cache, cached +from tests import unittest + class CacheTestCase(unittest.TestCase): diff --git a/tests/storage/test_appservice.py b/tests/storage/test_appservice.py index 9e98d0e33..099861b27 100644 --- a/tests/storage/test_appservice.py +++ b/tests/storage/test_appservice.py @@ -12,22 +12,26 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. -import tempfile -from synapse.config._base import ConfigError -from tests import unittest -from twisted.internet import defer - -from tests.utils import setup_test_homeserver -from synapse.appservice import ApplicationService, ApplicationServiceState -from synapse.storage.appservice import ( - ApplicationServiceStore, ApplicationServiceTransactionStore -) - import json import os -import yaml +import tempfile + from mock import Mock +import yaml + +from twisted.internet import defer + +from synapse.appservice import ApplicationService, ApplicationServiceState +from synapse.config._base import ConfigError +from synapse.storage.appservice import ( + ApplicationServiceStore, + ApplicationServiceTransactionStore, +) + +from tests import unittest +from tests.utils import setup_test_homeserver + class ApplicationServiceStoreTestCase(unittest.TestCase): @@ -42,7 +46,7 @@ class ApplicationServiceStoreTestCase(unittest.TestCase): hs = yield setup_test_homeserver( config=config, federation_sender=Mock(), - replication_layer=Mock(), + federation_client=Mock(), ) self.as_token = "token1" @@ -58,14 +62,14 @@ class ApplicationServiceStoreTestCase(unittest.TestCase): self._add_appservice("token2", "as2", "some_url", "some_hs_token", "bob") self._add_appservice("token3", "as3", "some_url", "some_hs_token", "bob") # must be done after inserts - self.store = ApplicationServiceStore(hs) + self.store = ApplicationServiceStore(None, hs) def tearDown(self): # TODO: suboptimal that we need to create files for tests! for f in self.as_yaml_files: try: os.remove(f) - except: + except Exception: pass def _add_appservice(self, as_token, id, url, hs_token, sender): @@ -119,7 +123,7 @@ class ApplicationServiceTransactionStoreTestCase(unittest.TestCase): hs = yield setup_test_homeserver( config=config, federation_sender=Mock(), - replication_layer=Mock(), + federation_client=Mock(), ) self.db_pool = hs.get_db_pool() @@ -150,7 +154,7 @@ class ApplicationServiceTransactionStoreTestCase(unittest.TestCase): self.as_yaml_files = [] - self.store = TestTransactionStore(hs) + self.store = TestTransactionStore(None, hs) def _add_service(self, url, as_token, id): as_yaml = dict(url=url, as_token=as_token, hs_token="something", @@ -420,8 +424,8 @@ class ApplicationServiceTransactionStoreTestCase(unittest.TestCase): class TestTransactionStore(ApplicationServiceTransactionStore, ApplicationServiceStore): - def __init__(self, hs): - super(TestTransactionStore, self).__init__(hs) + def __init__(self, db_conn, hs): + super(TestTransactionStore, self).__init__(db_conn, hs) class ApplicationServiceStoreConfigTestCase(unittest.TestCase): @@ -455,10 +459,10 @@ class ApplicationServiceStoreConfigTestCase(unittest.TestCase): config=config, datastore=Mock(), federation_sender=Mock(), - replication_layer=Mock(), + federation_client=Mock(), ) - ApplicationServiceStore(hs) + ApplicationServiceStore(None, hs) @defer.inlineCallbacks def test_duplicate_ids(self): @@ -473,16 +477,16 @@ class ApplicationServiceStoreConfigTestCase(unittest.TestCase): config=config, datastore=Mock(), federation_sender=Mock(), - replication_layer=Mock(), + federation_client=Mock(), ) with self.assertRaises(ConfigError) as cm: - ApplicationServiceStore(hs) + ApplicationServiceStore(None, hs) e = cm.exception - self.assertIn(f1, e.message) - self.assertIn(f2, e.message) - self.assertIn("id", e.message) + self.assertIn(f1, str(e)) + self.assertIn(f2, str(e)) + self.assertIn("id", str(e)) @defer.inlineCallbacks def test_duplicate_as_tokens(self): @@ -497,13 +501,13 @@ class ApplicationServiceStoreConfigTestCase(unittest.TestCase): config=config, datastore=Mock(), federation_sender=Mock(), - replication_layer=Mock(), + federation_client=Mock(), ) with self.assertRaises(ConfigError) as cm: - ApplicationServiceStore(hs) + ApplicationServiceStore(None, hs) e = cm.exception - self.assertIn(f1, e.message) - self.assertIn(f2, e.message) - self.assertIn("as_token", e.message) + self.assertIn(f1, str(e)) + self.assertIn(f2, str(e)) + self.assertIn("as_token", str(e)) diff --git a/tests/storage/test_background_update.py b/tests/storage/test_background_update.py index 1286b4ce2..ab1f31057 100644 --- a/tests/storage/test_background_update.py +++ b/tests/storage/test_background_update.py @@ -1,10 +1,10 @@ -from tests import unittest +from mock import Mock + from twisted.internet import defer +from tests import unittest from tests.utils import setup_test_homeserver -from mock import Mock - class BackgroundUpdateTestCase(unittest.TestCase): diff --git a/tests/storage/test_base.py b/tests/storage/test_base.py index 91e971190..1d1234ee3 100644 --- a/tests/storage/test_base.py +++ b/tests/storage/test_base.py @@ -14,18 +14,18 @@ # limitations under the License. -from tests import unittest -from twisted.internet import defer +from collections import OrderedDict from mock import Mock -from collections import OrderedDict +from twisted.internet import defer from synapse.server import HomeServer - from synapse.storage._base import SQLBaseStore from synapse.storage.engines import create_engine +from tests import unittest + class SQLBaseStoreTestCase(unittest.TestCase): """ Test the "simple" SQL generating methods in SQLBaseStore. """ @@ -56,7 +56,7 @@ class SQLBaseStoreTestCase(unittest.TestCase): database_engine=create_engine(config.database_config), ) - self.datastore = SQLBaseStore(hs) + self.datastore = SQLBaseStore(None, hs) @defer.inlineCallbacks def test_insert_1col(self): diff --git a/tests/storage/test_client_ips.py b/tests/storage/test_client_ips.py index 03df69757..bd6fda6cb 100644 --- a/tests/storage/test_client_ips.py +++ b/tests/storage/test_client_ips.py @@ -15,9 +15,6 @@ from twisted.internet import defer -import synapse.server -import synapse.storage -import synapse.types import tests.unittest import tests.utils @@ -39,7 +36,7 @@ class ClientIpStoreTestCase(tests.unittest.TestCase): self.clock.now = 12345678 user_id = "@user:id" yield self.store.insert_client_ip( - synapse.types.UserID.from_string(user_id), + user_id, "access_token", "ip", "user_agent", "device_id", ) diff --git a/tests/storage/test_devices.py b/tests/storage/test_devices.py index f8725acea..a54cc6bc3 100644 --- a/tests/storage/test_devices.py +++ b/tests/storage/test_devices.py @@ -16,6 +16,7 @@ from twisted.internet import defer import synapse.api.errors + import tests.unittest import tests.utils diff --git a/tests/storage/test_directory.py b/tests/storage/test_directory.py index b087892e0..129ebaf34 100644 --- a/tests/storage/test_directory.py +++ b/tests/storage/test_directory.py @@ -14,12 +14,12 @@ # limitations under the License. -from tests import unittest from twisted.internet import defer from synapse.storage.directory import DirectoryStore -from synapse.types import RoomID, RoomAlias +from synapse.types import RoomAlias, RoomID +from tests import unittest from tests.utils import setup_test_homeserver @@ -29,7 +29,7 @@ class DirectoryStoreTestCase(unittest.TestCase): def setUp(self): hs = yield setup_test_homeserver() - self.store = DirectoryStore(hs) + self.store = DirectoryStore(None, hs) self.room = RoomID.from_string("!abcde:test") self.alias = RoomAlias.from_string("#my-room:test") diff --git a/tests/storage/test_event_federation.py b/tests/storage/test_event_federation.py new file mode 100644 index 000000000..30683e788 --- /dev/null +++ b/tests/storage/test_event_federation.py @@ -0,0 +1,68 @@ +# -*- coding: utf-8 -*- +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the 'License'); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an 'AS IS' BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from twisted.internet import defer + +import tests.unittest +import tests.utils + + +class EventFederationWorkerStoreTestCase(tests.unittest.TestCase): + @defer.inlineCallbacks + def setUp(self): + hs = yield tests.utils.setup_test_homeserver() + self.store = hs.get_datastore() + + @defer.inlineCallbacks + def test_get_prev_events_for_room(self): + room_id = '@ROOM:local' + + # add a bunch of events and hashes to act as forward extremities + def insert_event(txn, i): + event_id = '$event_%i:local' % i + + txn.execute(( + "INSERT INTO events (" + " room_id, event_id, type, depth, topological_ordering," + " content, processed, outlier) " + "VALUES (?, ?, 'm.test', ?, ?, 'test', ?, ?)" + ), (room_id, event_id, i, i, True, False)) + + txn.execute(( + 'INSERT INTO event_forward_extremities (room_id, event_id) ' + 'VALUES (?, ?)' + ), (room_id, event_id)) + + txn.execute(( + 'INSERT INTO event_reference_hashes ' + '(event_id, algorithm, hash) ' + "VALUES (?, 'sha256', ?)" + ), (event_id, 'ffff')) + + for i in range(0, 11): + yield self.store.runInteraction("insert", insert_event, i) + + # this should get the last five and five others + r = yield self.store.get_prev_events_for_room(room_id) + self.assertEqual(10, len(r)) + for i in range(0, 5): + el = r[i] + depth = el[2] + self.assertEqual(10 - i, depth) + + for i in range(5, 5): + el = r[i] + depth = el[2] + self.assertLessEqual(5, depth) diff --git a/tests/storage/test_event_push_actions.py b/tests/storage/test_event_push_actions.py index 313548835..8430fc7ba 100644 --- a/tests/storage/test_event_push_actions.py +++ b/tests/storage/test_event_push_actions.py @@ -13,11 +13,12 @@ # See the License for the specific language governing permissions and # limitations under the License. +from mock import Mock + from twisted.internet import defer import tests.unittest import tests.utils -from mock import Mock USER_ID = "@user:example.com" @@ -55,13 +56,14 @@ class EventPushActionsStoreTestCase(tests.unittest.TestCase): def _assert_counts(noitf_count, highlight_count): counts = yield self.store.runInteraction( "", self.store._get_unread_counts_by_pos_txn, - room_id, user_id, 0, 0 + room_id, user_id, 0 ) self.assertEquals( counts, {"notify_count": noitf_count, "highlight_count": highlight_count} ) + @defer.inlineCallbacks def _inject_actions(stream, action): event = Mock() event.room_id = room_id @@ -69,11 +71,12 @@ class EventPushActionsStoreTestCase(tests.unittest.TestCase): event.internal_metadata.stream_ordering = stream event.depth = stream - tuples = [(user_id, action)] - - return self.store.runInteraction( + yield self.store.add_push_actions_to_staging( + event.event_id, {user_id: action}, + ) + yield self.store.runInteraction( "", self.store._set_push_actions_for_event_and_users_txn, - event, tuples + [(event, None)], [(event, None)], ) def _rotate(stream): @@ -84,7 +87,7 @@ class EventPushActionsStoreTestCase(tests.unittest.TestCase): def _mark_read(stream, depth): return self.store.runInteraction( "", self.store._remove_old_push_actions_before_txn, - room_id, user_id, depth, stream + room_id, user_id, stream ) yield _assert_counts(0, 0) @@ -125,3 +128,69 @@ class EventPushActionsStoreTestCase(tests.unittest.TestCase): yield _assert_counts(1, 1) yield _rotate(10) yield _assert_counts(1, 1) + + @defer.inlineCallbacks + def test_find_first_stream_ordering_after_ts(self): + def add_event(so, ts): + return self.store._simple_insert("events", { + "stream_ordering": so, + "received_ts": ts, + "event_id": "event%i" % so, + "type": "", + "room_id": "", + "content": "", + "processed": True, + "outlier": False, + "topological_ordering": 0, + "depth": 0, + }) + + # start with the base case where there are no events in the table + r = yield self.store.find_first_stream_ordering_after_ts(11) + self.assertEqual(r, 0) + + # now with one event + yield add_event(2, 10) + r = yield self.store.find_first_stream_ordering_after_ts(9) + self.assertEqual(r, 2) + r = yield self.store.find_first_stream_ordering_after_ts(10) + self.assertEqual(r, 2) + r = yield self.store.find_first_stream_ordering_after_ts(11) + self.assertEqual(r, 3) + + # add a bunch of dummy events to the events table + for (stream_ordering, ts) in ( + (3, 110), + (4, 120), + (5, 120), + (10, 130), + (20, 140), + ): + yield add_event(stream_ordering, ts) + + r = yield self.store.find_first_stream_ordering_after_ts(110) + self.assertEqual(r, 3, + "First event after 110ms should be 3, was %i" % r) + + # 4 and 5 are both after 120: we want 4 rather than 5 + r = yield self.store.find_first_stream_ordering_after_ts(120) + self.assertEqual(r, 4, + "First event after 120ms should be 4, was %i" % r) + + r = yield self.store.find_first_stream_ordering_after_ts(129) + self.assertEqual(r, 10, + "First event after 129ms should be 10, was %i" % r) + + # check we can get the last event + r = yield self.store.find_first_stream_ordering_after_ts(140) + self.assertEqual(r, 20, + "First event after 14ms should be 20, was %i" % r) + + # off the end + r = yield self.store.find_first_stream_ordering_after_ts(160) + self.assertEqual(r, 21) + + # check we can find an event at ordering zero + yield add_event(0, 5) + r = yield self.store.find_first_stream_ordering_after_ts(1) + self.assertEqual(r, 0) diff --git a/tests/storage/test_keys.py b/tests/storage/test_keys.py index 0be790d8f..3a3d00278 100644 --- a/tests/storage/test_keys.py +++ b/tests/storage/test_keys.py @@ -14,6 +14,7 @@ # limitations under the License. import signedjson.key + from twisted.internet import defer import tests.unittest diff --git a/tests/storage/test_presence.py b/tests/storage/test_presence.py index 63203cea3..3276b3950 100644 --- a/tests/storage/test_presence.py +++ b/tests/storage/test_presence.py @@ -14,13 +14,13 @@ # limitations under the License. -from tests import unittest from twisted.internet import defer from synapse.storage.presence import PresenceStore from synapse.types import UserID -from tests.utils import setup_test_homeserver, MockClock +from tests import unittest +from tests.utils import MockClock, setup_test_homeserver class PresenceStoreTestCase(unittest.TestCase): @@ -29,7 +29,7 @@ class PresenceStoreTestCase(unittest.TestCase): def setUp(self): hs = yield setup_test_homeserver(clock=MockClock()) - self.store = PresenceStore(hs) + self.store = PresenceStore(None, hs) self.u_apple = UserID.from_string("@apple:test") self.u_banana = UserID.from_string("@banana:test") diff --git a/tests/storage/test_profile.py b/tests/storage/test_profile.py index 24118bbc8..2c95e5e95 100644 --- a/tests/storage/test_profile.py +++ b/tests/storage/test_profile.py @@ -14,12 +14,12 @@ # limitations under the License. -from tests import unittest from twisted.internet import defer from synapse.storage.profile import ProfileStore from synapse.types import UserID +from tests import unittest from tests.utils import setup_test_homeserver @@ -29,7 +29,7 @@ class ProfileStoreTestCase(unittest.TestCase): def setUp(self): hs = yield setup_test_homeserver() - self.store = ProfileStore(hs) + self.store = ProfileStore(None, hs) self.u_frank = UserID.from_string("@frank:test") diff --git a/tests/storage/test_redaction.py b/tests/storage/test_redaction.py index 6afaca3a6..475ec900c 100644 --- a/tests/storage/test_redaction.py +++ b/tests/storage/test_redaction.py @@ -14,16 +14,16 @@ # limitations under the License. -from tests import unittest +from mock import Mock + from twisted.internet import defer from synapse.api.constants import EventTypes, Membership -from synapse.types import UserID, RoomID +from synapse.types import RoomID, UserID +from tests import unittest from tests.utils import setup_test_homeserver -from mock import Mock - class RedactionTestCase(unittest.TestCase): @@ -36,8 +36,7 @@ class RedactionTestCase(unittest.TestCase): self.store = hs.get_datastore() self.event_builder_factory = hs.get_event_builder_factory() - self.handlers = hs.get_handlers() - self.message_handler = self.handlers.message_handler + self.event_creation_handler = hs.get_event_creation_handler() self.u_alice = UserID.from_string("@alice:test") self.u_bob = UserID.from_string("@bob:test") @@ -59,7 +58,7 @@ class RedactionTestCase(unittest.TestCase): "content": content, }) - event, context = yield self.message_handler._create_new_client_event( + event, context = yield self.event_creation_handler.create_new_client_event( builder ) @@ -79,7 +78,7 @@ class RedactionTestCase(unittest.TestCase): "content": {"body": body, "msgtype": u"message"}, }) - event, context = yield self.message_handler._create_new_client_event( + event, context = yield self.event_creation_handler.create_new_client_event( builder ) @@ -98,7 +97,7 @@ class RedactionTestCase(unittest.TestCase): "redacts": event_id, }) - event, context = yield self.message_handler._create_new_client_event( + event, context = yield self.event_creation_handler.create_new_client_event( builder ) diff --git a/tests/storage/test_registration.py b/tests/storage/test_registration.py index 316ecdb32..7821ea3fa 100644 --- a/tests/storage/test_registration.py +++ b/tests/storage/test_registration.py @@ -14,9 +14,9 @@ # limitations under the License. -from tests import unittest from twisted.internet import defer +from tests import unittest from tests.utils import setup_test_homeserver @@ -42,9 +42,15 @@ class RegistrationStoreTestCase(unittest.TestCase): yield self.store.register(self.user_id, self.tokens[0], self.pwhash) self.assertEquals( - # TODO(paul): Surely this field should be 'user_id', not 'name' - # Additionally surely it shouldn't come in a 1-element list - {"name": self.user_id, "password_hash": self.pwhash, "is_guest": 0}, + { + # TODO(paul): Surely this field should be 'user_id', not 'name' + "name": self.user_id, + "password_hash": self.pwhash, + "is_guest": 0, + "consent_version": None, + "consent_server_notice_sent": None, + "appservice_id": None, + }, (yield self.store.get_user_by_id(self.user_id)) ) @@ -86,7 +92,8 @@ class RegistrationStoreTestCase(unittest.TestCase): # now delete some yield self.store.user_delete_access_tokens( - self.user_id, device_id=self.device_id, delete_refresh_tokens=True) + self.user_id, device_id=self.device_id, + ) # check they were deleted user = yield self.store.get_user_by_access_token(self.tokens[1]) @@ -97,8 +104,7 @@ class RegistrationStoreTestCase(unittest.TestCase): self.assertEqual(self.user_id, user["name"]) # now delete the rest - yield self.store.user_delete_access_tokens( - self.user_id, delete_refresh_tokens=True) + yield self.store.user_delete_access_tokens(self.user_id) user = yield self.store.get_user_by_access_token(self.tokens[0]) self.assertIsNone(user, diff --git a/tests/storage/test_room.py b/tests/storage/test_room.py index ef8a4d234..ae8ae94b6 100644 --- a/tests/storage/test_room.py +++ b/tests/storage/test_room.py @@ -14,12 +14,12 @@ # limitations under the License. -from tests import unittest from twisted.internet import defer from synapse.api.constants import EventTypes -from synapse.types import UserID, RoomID, RoomAlias +from synapse.types import RoomAlias, RoomID, UserID +from tests import unittest from tests.utils import setup_test_homeserver diff --git a/tests/storage/test_roommember.py b/tests/storage/test_roommember.py index 1be7d932f..c5fd54f67 100644 --- a/tests/storage/test_roommember.py +++ b/tests/storage/test_roommember.py @@ -14,16 +14,16 @@ # limitations under the License. -from tests import unittest +from mock import Mock + from twisted.internet import defer from synapse.api.constants import EventTypes, Membership -from synapse.types import UserID, RoomID +from synapse.types import RoomID, UserID +from tests import unittest from tests.utils import setup_test_homeserver -from mock import Mock - class RoomMemberStoreTestCase(unittest.TestCase): @@ -37,8 +37,7 @@ class RoomMemberStoreTestCase(unittest.TestCase): # storage logic self.store = hs.get_datastore() self.event_builder_factory = hs.get_event_builder_factory() - self.handlers = hs.get_handlers() - self.message_handler = self.handlers.message_handler + self.event_creation_handler = hs.get_event_creation_handler() self.u_alice = UserID.from_string("@alice:test") self.u_bob = UserID.from_string("@bob:test") @@ -58,7 +57,7 @@ class RoomMemberStoreTestCase(unittest.TestCase): "content": {"membership": membership}, }) - event, context = yield self.message_handler._create_new_client_event( + event, context = yield self.event_creation_handler.create_new_client_event( builder ) diff --git a/tests/storage/test_user_directory.py b/tests/storage/test_user_directory.py new file mode 100644 index 000000000..23fad12bc --- /dev/null +++ b/tests/storage/test_user_directory.py @@ -0,0 +1,89 @@ +# -*- coding: utf-8 -*- +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from twisted.internet import defer + +from synapse.storage import UserDirectoryStore +from synapse.storage.roommember import ProfileInfo + +from tests import unittest +from tests.utils import setup_test_homeserver + +ALICE = "@alice:a" +BOB = "@bob:b" +BOBBY = "@bobby:a" + + +class UserDirectoryStoreTestCase(unittest.TestCase): + @defer.inlineCallbacks + def setUp(self): + self.hs = yield setup_test_homeserver() + self.store = UserDirectoryStore(None, self.hs) + + # alice and bob are both in !room_id. bobby is not but shares + # a homeserver with alice. + yield self.store.add_profiles_to_user_dir( + "!room:id", + { + ALICE: ProfileInfo(None, "alice"), + BOB: ProfileInfo(None, "bob"), + BOBBY: ProfileInfo(None, "bobby") + }, + ) + yield self.store.add_users_to_public_room( + "!room:id", + [ALICE, BOB], + ) + yield self.store.add_users_who_share_room( + "!room:id", + False, + ( + (ALICE, BOB), + (BOB, ALICE), + ), + ) + + @defer.inlineCallbacks + def test_search_user_dir(self): + # normally when alice searches the directory she should just find + # bob because bobby doesn't share a room with her. + r = yield self.store.search_user_dir(ALICE, "bob", 10) + self.assertFalse(r["limited"]) + self.assertEqual(1, len(r["results"])) + self.assertDictEqual(r["results"][0], { + "user_id": BOB, + "display_name": "bob", + "avatar_url": None, + }) + + @defer.inlineCallbacks + def test_search_user_dir_all_users(self): + self.hs.config.user_directory_search_all_users = True + try: + r = yield self.store.search_user_dir(ALICE, "bob", 10) + self.assertFalse(r["limited"]) + self.assertEqual(2, len(r["results"])) + self.assertDictEqual(r["results"][0], { + "user_id": BOB, + "display_name": "bob", + "avatar_url": None, + }) + self.assertDictEqual(r["results"][1], { + "user_id": BOBBY, + "display_name": "bobby", + "avatar_url": None, + }) + finally: + self.hs.config.user_directory_search_all_users = False diff --git a/tests/test_distributor.py b/tests/test_distributor.py index acebcf4a8..71d11cda7 100644 --- a/tests/test_distributor.py +++ b/tests/test_distributor.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2014-2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -13,13 +14,11 @@ # See the License for the specific language governing permissions and # limitations under the License. -from . import unittest -from twisted.internet import defer - from mock import Mock, patch from synapse.util.distributor import Distributor -from synapse.util.async import run_on_reactor + +from . import unittest class DistributorTestCase(unittest.TestCase): @@ -27,42 +26,19 @@ class DistributorTestCase(unittest.TestCase): def setUp(self): self.dist = Distributor() - @defer.inlineCallbacks def test_signal_dispatch(self): self.dist.declare("alert") observer = Mock() self.dist.observe("alert", observer) - d = self.dist.fire("alert", 1, 2, 3) - yield d - self.assertTrue(d.called) + self.dist.fire("alert", 1, 2, 3) observer.assert_called_with(1, 2, 3) - @defer.inlineCallbacks - def test_signal_dispatch_deferred(self): - self.dist.declare("whine") - - d_inner = defer.Deferred() - - def observer(): - return d_inner - - self.dist.observe("whine", observer) - - d_outer = self.dist.fire("whine") - - self.assertFalse(d_outer.called) - - d_inner.callback(None) - yield d_outer - self.assertTrue(d_outer.called) - - @defer.inlineCallbacks def test_signal_catch(self): self.dist.declare("alarm") - observers = [Mock() for i in 1, 2] + observers = [Mock() for i in (1, 2)] for o in observers: self.dist.observe("alarm", o) @@ -71,9 +47,7 @@ class DistributorTestCase(unittest.TestCase): with patch( "synapse.util.distributor.logger", spec=["warning"] ) as mock_logger: - d = self.dist.fire("alarm", "Go") - yield d - self.assertTrue(d.called) + self.dist.fire("alarm", "Go") observers[0].assert_called_once_with("Go") observers[1].assert_called_once_with("Go") @@ -83,35 +57,12 @@ class DistributorTestCase(unittest.TestCase): mock_logger.warning.call_args[0][0], str ) - @defer.inlineCallbacks - def test_signal_catch_no_suppress(self): - # Gut-wrenching - self.dist.suppress_failures = False - - self.dist.declare("whail") - - class MyException(Exception): - pass - - @defer.inlineCallbacks - def observer(): - yield run_on_reactor() - raise MyException("Oopsie") - - self.dist.observe("whail", observer) - - d = self.dist.fire("whail") - - yield self.assertFailure(d, MyException) - self.dist.suppress_failures = True - - @defer.inlineCallbacks def test_signal_prereg(self): observer = Mock() self.dist.observe("flare", observer) self.dist.declare("flare") - yield self.dist.fire("flare", 4, 5) + self.dist.fire("flare", 4, 5) observer.assert_called_with(4, 5) diff --git a/tests/test_dns.py b/tests/test_dns.py index c394c57ee..b647d9269 100644 --- a/tests/test_dns.py +++ b/tests/test_dns.py @@ -13,26 +13,27 @@ # See the License for the specific language governing permissions and # limitations under the License. -from . import unittest +from mock import Mock + from twisted.internet import defer from twisted.names import dns, error -from mock import Mock - from synapse.http.endpoint import resolve_service from tests.utils import MockClock +from . import unittest + +@unittest.DEBUG class DnsTestCase(unittest.TestCase): @defer.inlineCallbacks def test_resolve(self): dns_client_mock = Mock() - service_name = "test_service.examle.com" + service_name = "test_service.example.com" host_name = "example.com" - ip_address = "127.0.0.1" answer_srv = dns.RRHeader( type=dns.SRV, @@ -41,16 +42,10 @@ class DnsTestCase(unittest.TestCase): ) ) - answer_a = dns.RRHeader( - type=dns.A, - payload=dns.Record_A( - address=ip_address, - ) + dns_client_mock.lookupService.return_value = defer.succeed( + ([answer_srv], None, None), ) - dns_client_mock.lookupService.return_value = ([answer_srv], None, None) - dns_client_mock.lookupAddress.return_value = ([answer_a], None, None) - cache = {} servers = yield resolve_service( @@ -58,18 +53,17 @@ class DnsTestCase(unittest.TestCase): ) dns_client_mock.lookupService.assert_called_once_with(service_name) - dns_client_mock.lookupAddress.assert_called_once_with(host_name) self.assertEquals(len(servers), 1) self.assertEquals(servers, cache[service_name]) - self.assertEquals(servers[0].host, ip_address) + self.assertEquals(servers[0].host, host_name) @defer.inlineCallbacks def test_from_cache_expired_and_dns_fail(self): dns_client_mock = Mock() dns_client_mock.lookupService.return_value = defer.fail(error.DNSServerError()) - service_name = "test_service.examle.com" + service_name = "test_service.example.com" entry = Mock(spec_set=["expires"]) entry.expires = 0 @@ -94,7 +88,7 @@ class DnsTestCase(unittest.TestCase): dns_client_mock = Mock(spec_set=['lookupService']) dns_client_mock.lookupService = Mock(spec_set=[]) - service_name = "test_service.examle.com" + service_name = "test_service.example.com" entry = Mock(spec_set=["expires"]) entry.expires = 999999999 @@ -118,7 +112,7 @@ class DnsTestCase(unittest.TestCase): dns_client_mock.lookupService.return_value = defer.fail(error.DNSServerError()) - service_name = "test_service.examle.com" + service_name = "test_service.example.com" cache = {} @@ -133,7 +127,7 @@ class DnsTestCase(unittest.TestCase): dns_client_mock.lookupService.return_value = defer.fail(error.DNSNameError()) - service_name = "test_service.examle.com" + service_name = "test_service.example.com" cache = {} diff --git a/tests/test_event_auth.py b/tests/test_event_auth.py new file mode 100644 index 000000000..06112430e --- /dev/null +++ b/tests/test_event_auth.py @@ -0,0 +1,152 @@ +# -*- coding: utf-8 -*- +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import unittest + +from synapse import event_auth +from synapse.api.errors import AuthError +from synapse.events import FrozenEvent + + +class EventAuthTestCase(unittest.TestCase): + def test_random_users_cannot_send_state_before_first_pl(self): + """ + Check that, before the first PL lands, the creator is the only user + that can send a state event. + """ + creator = "@creator:example.com" + joiner = "@joiner:example.com" + auth_events = { + ("m.room.create", ""): _create_event(creator), + ("m.room.member", creator): _join_event(creator), + ("m.room.member", joiner): _join_event(joiner), + } + + # creator should be able to send state + event_auth.check( + _random_state_event(creator), auth_events, + do_sig_check=False, + ) + + # joiner should not be able to send state + self.assertRaises( + AuthError, + event_auth.check, + _random_state_event(joiner), + auth_events, + do_sig_check=False, + ), + + def test_state_default_level(self): + """ + Check that users above the state_default level can send state and + those below cannot + """ + creator = "@creator:example.com" + pleb = "@joiner:example.com" + king = "@joiner2:example.com" + + auth_events = { + ("m.room.create", ""): _create_event(creator), + ("m.room.member", creator): _join_event(creator), + ("m.room.power_levels", ""): _power_levels_event(creator, { + "state_default": "30", + "users": { + pleb: "29", + king: "30", + }, + }), + ("m.room.member", pleb): _join_event(pleb), + ("m.room.member", king): _join_event(king), + } + + # pleb should not be able to send state + self.assertRaises( + AuthError, + event_auth.check, + _random_state_event(pleb), + auth_events, + do_sig_check=False, + ), + + # king should be able to send state + event_auth.check( + _random_state_event(king), auth_events, + do_sig_check=False, + ) + + +# helpers for making events + +TEST_ROOM_ID = "!test:room" + + +def _create_event(user_id): + return FrozenEvent({ + "room_id": TEST_ROOM_ID, + "event_id": _get_event_id(), + "type": "m.room.create", + "sender": user_id, + "content": { + "creator": user_id, + }, + }) + + +def _join_event(user_id): + return FrozenEvent({ + "room_id": TEST_ROOM_ID, + "event_id": _get_event_id(), + "type": "m.room.member", + "sender": user_id, + "state_key": user_id, + "content": { + "membership": "join", + }, + }) + + +def _power_levels_event(sender, content): + return FrozenEvent({ + "room_id": TEST_ROOM_ID, + "event_id": _get_event_id(), + "type": "m.room.power_levels", + "sender": sender, + "state_key": "", + "content": content, + }) + + +def _random_state_event(sender): + return FrozenEvent({ + "room_id": TEST_ROOM_ID, + "event_id": _get_event_id(), + "type": "test.state", + "sender": sender, + "state_key": "", + "content": { + "membership": "join", + }, + }) + + +event_count = 0 + + +def _get_event_id(): + global event_count + c = event_count + event_count += 1 + return "!%i:example.com" % (c, ) diff --git a/tests/test_federation.py b/tests/test_federation.py new file mode 100644 index 000000000..f40ff29b5 --- /dev/null +++ b/tests/test_federation.py @@ -0,0 +1,242 @@ + +from mock import Mock + +from twisted.internet.defer import maybeDeferred, succeed + +from synapse.events import FrozenEvent +from synapse.types import Requester, UserID +from synapse.util import Clock + +from tests import unittest +from tests.server import ThreadedMemoryReactorClock, setup_test_homeserver + + +class MessageAcceptTests(unittest.TestCase): + def setUp(self): + + self.http_client = Mock() + self.reactor = ThreadedMemoryReactorClock() + self.hs_clock = Clock(self.reactor) + self.homeserver = setup_test_homeserver( + http_client=self.http_client, clock=self.hs_clock, reactor=self.reactor + ) + + user_id = UserID("us", "test") + our_user = Requester(user_id, None, False, None, None) + room_creator = self.homeserver.get_room_creation_handler() + room = room_creator.create_room( + our_user, room_creator.PRESETS_DICT["public_chat"], ratelimit=False + ) + self.reactor.advance(0.1) + self.room_id = self.successResultOf(room)["room_id"] + + # Figure out what the most recent event is + most_recent = self.successResultOf( + maybeDeferred( + self.homeserver.datastore.get_latest_event_ids_in_room, self.room_id + ) + )[0] + + join_event = FrozenEvent( + { + "room_id": self.room_id, + "sender": "@baduser:test.serv", + "state_key": "@baduser:test.serv", + "event_id": "$join:test.serv", + "depth": 1000, + "origin_server_ts": 1, + "type": "m.room.member", + "origin": "test.servx", + "content": {"membership": "join"}, + "auth_events": [], + "prev_state": [(most_recent, {})], + "prev_events": [(most_recent, {})], + } + ) + + self.handler = self.homeserver.get_handlers().federation_handler + self.handler.do_auth = lambda *a, **b: succeed(True) + self.client = self.homeserver.get_federation_client() + self.client._check_sigs_and_hash_and_fetch = lambda dest, pdus, **k: succeed( + pdus + ) + + # Send the join, it should return None (which is not an error) + d = self.handler.on_receive_pdu( + "test.serv", join_event, sent_to_us_directly=True + ) + self.reactor.advance(1) + self.assertEqual(self.successResultOf(d), None) + + # Make sure we actually joined the room + self.assertEqual( + self.successResultOf( + maybeDeferred( + self.homeserver.datastore.get_latest_event_ids_in_room, self.room_id + ) + )[0], + "$join:test.serv", + ) + + def test_cant_hide_direct_ancestors(self): + """ + If you send a message, you must be able to provide the direct + prev_events that said event references. + """ + + def post_json(destination, path, data, headers=None, timeout=0): + # If it asks us for new missing events, give them NOTHING + if path.startswith("/_matrix/federation/v1/get_missing_events/"): + return {"events": []} + + self.http_client.post_json = post_json + + # Figure out what the most recent event is + most_recent = self.successResultOf( + maybeDeferred( + self.homeserver.datastore.get_latest_event_ids_in_room, self.room_id + ) + )[0] + + # Now lie about an event + lying_event = FrozenEvent( + { + "room_id": self.room_id, + "sender": "@baduser:test.serv", + "event_id": "one:test.serv", + "depth": 1000, + "origin_server_ts": 1, + "type": "m.room.message", + "origin": "test.serv", + "content": "hewwo?", + "auth_events": [], + "prev_events": [("two:test.serv", {}), (most_recent, {})], + } + ) + + d = self.handler.on_receive_pdu( + "test.serv", lying_event, sent_to_us_directly=True + ) + + # Step the reactor, so the database fetches come back + self.reactor.advance(1) + + # on_receive_pdu should throw an error + failure = self.failureResultOf(d) + self.assertEqual( + failure.value.args[0], + ( + "ERROR 403: Your server isn't divulging details about prev_events " + "referenced in this event." + ), + ) + + # Make sure the invalid event isn't there + extrem = maybeDeferred( + self.homeserver.datastore.get_latest_event_ids_in_room, self.room_id + ) + self.assertEqual(self.successResultOf(extrem)[0], "$join:test.serv") + + def test_cant_hide_past_history(self): + """ + If you send a message, you must be able to provide the direct + prev_events that said event references. + """ + + def post_json(destination, path, data, headers=None, timeout=0): + if path.startswith("/_matrix/federation/v1/get_missing_events/"): + return { + "events": [ + { + "room_id": self.room_id, + "sender": "@baduser:test.serv", + "event_id": "three:test.serv", + "depth": 1000, + "origin_server_ts": 1, + "type": "m.room.message", + "origin": "test.serv", + "content": "hewwo?", + "auth_events": [], + "prev_events": [("four:test.serv", {})], + } + ] + } + + self.http_client.post_json = post_json + + def get_json(destination, path, args, headers=None): + if path.startswith("/_matrix/federation/v1/state_ids/"): + d = self.successResultOf( + self.homeserver.datastore.get_state_ids_for_event("one:test.serv") + ) + + return succeed( + { + "pdu_ids": [ + y + for x, y in d.items() + if x == ("m.room.member", "@us:test") + ], + "auth_chain_ids": list(d.values()), + } + ) + + self.http_client.get_json = get_json + + # Figure out what the most recent event is + most_recent = self.successResultOf( + maybeDeferred( + self.homeserver.datastore.get_latest_event_ids_in_room, self.room_id + ) + )[0] + + # Make a good event + good_event = FrozenEvent( + { + "room_id": self.room_id, + "sender": "@baduser:test.serv", + "event_id": "one:test.serv", + "depth": 1000, + "origin_server_ts": 1, + "type": "m.room.message", + "origin": "test.serv", + "content": "hewwo?", + "auth_events": [], + "prev_events": [(most_recent, {})], + } + ) + + d = self.handler.on_receive_pdu( + "test.serv", good_event, sent_to_us_directly=True + ) + self.reactor.advance(1) + self.assertEqual(self.successResultOf(d), None) + + bad_event = FrozenEvent( + { + "room_id": self.room_id, + "sender": "@baduser:test.serv", + "event_id": "two:test.serv", + "depth": 1000, + "origin_server_ts": 1, + "type": "m.room.message", + "origin": "test.serv", + "content": "hewwo?", + "auth_events": [], + "prev_events": [("one:test.serv", {}), ("three:test.serv", {})], + } + ) + + d = self.handler.on_receive_pdu( + "test.serv", bad_event, sent_to_us_directly=True + ) + self.reactor.advance(1) + + extrem = maybeDeferred( + self.homeserver.datastore.get_latest_event_ids_in_room, self.room_id + ) + self.assertEqual(self.successResultOf(extrem)[0], "two:test.serv") + + state = self.homeserver.get_state_handler().get_current_state_ids(self.room_id) + self.reactor.advance(1) + self.assertIn(("m.room.member", "@us:test"), self.successResultOf(state).keys()) diff --git a/tests/test_preview.py b/tests/test_preview.py index 5bd36c74a..446843367 100644 --- a/tests/test_preview.py +++ b/tests/test_preview.py @@ -13,12 +13,13 @@ # See the License for the specific language governing permissions and # limitations under the License. -from . import unittest - from synapse.rest.media.v1.preview_url_resource import ( - summarize_paragraphs, decode_and_calc_og + decode_and_calc_og, + summarize_paragraphs, ) +from . import unittest + class PreviewTestCase(unittest.TestCase): diff --git a/tests/test_server.py b/tests/test_server.py new file mode 100644 index 000000000..7e063c029 --- /dev/null +++ b/tests/test_server.py @@ -0,0 +1,131 @@ +import json +import re + +from twisted.internet.defer import Deferred +from twisted.test.proto_helpers import MemoryReactorClock + +from synapse.api.errors import Codes, SynapseError +from synapse.http.server import JsonResource +from synapse.util import Clock + +from tests import unittest +from tests.server import make_request, setup_test_homeserver + + +class JsonResourceTests(unittest.TestCase): + def setUp(self): + self.reactor = MemoryReactorClock() + self.hs_clock = Clock(self.reactor) + self.homeserver = setup_test_homeserver( + http_client=None, clock=self.hs_clock, reactor=self.reactor + ) + + def test_handler_for_request(self): + """ + JsonResource.handler_for_request gives correctly decoded URL args to + the callback, while Twisted will give the raw bytes of URL query + arguments. + """ + got_kwargs = {} + + def _callback(request, **kwargs): + got_kwargs.update(kwargs) + return (200, kwargs) + + res = JsonResource(self.homeserver) + res.register_paths( + "GET", [re.compile("^/_matrix/foo/(?P[^/]*)$")], _callback + ) + + request, channel = make_request(b"GET", b"/_matrix/foo/%E2%98%83?a=%E2%98%83") + request.render(res) + + self.assertEqual(request.args, {b'a': [u"\N{SNOWMAN}".encode('utf8')]}) + self.assertEqual(got_kwargs, {u"room_id": u"\N{SNOWMAN}"}) + + def test_callback_direct_exception(self): + """ + If the web callback raises an uncaught exception, it will be translated + into a 500. + """ + + def _callback(request, **kwargs): + raise Exception("boo") + + res = JsonResource(self.homeserver) + res.register_paths("GET", [re.compile("^/_matrix/foo$")], _callback) + + request, channel = make_request(b"GET", b"/_matrix/foo") + request.render(res) + + self.assertEqual(channel.result["code"], b'500') + + def test_callback_indirect_exception(self): + """ + If the web callback raises an uncaught exception in a Deferred, it will + be translated into a 500. + """ + + def _throw(*args): + raise Exception("boo") + + def _callback(request, **kwargs): + d = Deferred() + d.addCallback(_throw) + self.reactor.callLater(1, d.callback, True) + return d + + res = JsonResource(self.homeserver) + res.register_paths("GET", [re.compile("^/_matrix/foo$")], _callback) + + request, channel = make_request(b"GET", b"/_matrix/foo") + request.render(res) + + # No error has been raised yet + self.assertTrue("code" not in channel.result) + + # Advance time, now there's an error + self.reactor.advance(1) + self.assertEqual(channel.result["code"], b'500') + + def test_callback_synapseerror(self): + """ + If the web callback raises a SynapseError, it returns the appropriate + status code and message set in it. + """ + + def _callback(request, **kwargs): + raise SynapseError(403, "Forbidden!!one!", Codes.FORBIDDEN) + + res = JsonResource(self.homeserver) + res.register_paths("GET", [re.compile("^/_matrix/foo$")], _callback) + + request, channel = make_request(b"GET", b"/_matrix/foo") + request.render(res) + + self.assertEqual(channel.result["code"], b'403') + reply_body = json.loads(channel.result["body"]) + self.assertEqual(reply_body["error"], "Forbidden!!one!") + self.assertEqual(reply_body["errcode"], "M_FORBIDDEN") + + def test_no_handler(self): + """ + If there is no handler to process the request, Synapse will return 400. + """ + + def _callback(request, **kwargs): + """ + Not ever actually called! + """ + self.fail("shouldn't ever get here") + + res = JsonResource(self.homeserver) + res.register_paths("GET", [re.compile("^/_matrix/foo$")], _callback) + + request, channel = make_request(b"GET", b"/_matrix/foobar") + request.render(res) + + self.assertEqual(channel.result["code"], b'400') + reply_body = json.loads(channel.result["body"]) + self.assertEqual(reply_body["error"], "Unrecognized request") + self.assertEqual(reply_body["errcode"], "M_UNRECOGNIZED") diff --git a/tests/test_state.py b/tests/test_state.py index feb84f3d4..429a18cbf 100644 --- a/tests/test_state.py +++ b/tests/test_state.py @@ -13,18 +13,18 @@ # See the License for the specific language governing permissions and # limitations under the License. -from tests import unittest -from twisted.internet import defer - -from synapse.events import FrozenEvent -from synapse.api.auth import Auth -from synapse.api.constants import EventTypes, Membership -from synapse.state import StateHandler - -from .utils import MockClock - from mock import Mock +from twisted.internet import defer + +from synapse.api.auth import Auth +from synapse.api.constants import EventTypes, Membership +from synapse.events import FrozenEvent +from synapse.state import StateHandler, StateResolutionHandler + +from tests import unittest + +from .utils import MockClock _next_event_id = 1000 @@ -80,14 +80,14 @@ class StateGroupStore(object): return defer.succeed(groups) - def store_state_groups(self, event, context): - if context.current_state_ids is None: - return + def store_state_group(self, event_id, room_id, prev_group, delta_ids, + current_state_ids): + state_group = self._next_group + self._next_group += 1 - state_events = dict(context.current_state_ids) + self._group_to_state[state_group] = dict(current_state_ids) - self._group_to_state[context.state_group] = state_events - self._event_to_state_group[event.event_id] = context.state_group + return state_group def get_events(self, event_ids, **kwargs): return { @@ -95,10 +95,19 @@ class StateGroupStore(object): if e_id in self._event_id_to_event } + def get_state_group_delta(self, name): + return (None, None) + def register_events(self, events): for e in events: self._event_id_to_event[e.event_id] = e + def register_event_context(self, event, context): + self._event_to_state_group[event.event_id] = context.state_group + + def register_event_id_state_group(self, event_id, state_group): + self._event_to_state_group[event_id] = state_group + class DictObj(dict): def __init__(self, **kwargs): @@ -137,25 +146,16 @@ class Graph(object): class StateTestCase(unittest.TestCase): def setUp(self): - self.store = Mock( - spec_set=[ - "get_state_groups_ids", - "add_event_hashes", - "get_events", - "get_next_state_group", - "get_state_group_delta", - ] - ) + self.store = StateGroupStore() hs = Mock(spec_set=[ "get_datastore", "get_auth", "get_state_handler", "get_clock", + "get_state_resolution_handler", ]) hs.get_datastore.return_value = self.store hs.get_state_handler.return_value = None hs.get_clock.return_value = MockClock() hs.get_auth.return_value = Auth(hs) - - self.store.get_next_state_group.side_effect = Mock - self.store.get_state_group_delta.return_value = (None, None) + hs.get_state_resolution_handler = lambda: StateResolutionHandler(hs) self.state = StateHandler(hs) self.event_id = 0 @@ -195,17 +195,17 @@ class StateTestCase(unittest.TestCase): } ) - store = StateGroupStore() - self.store.get_state_groups_ids.side_effect = store.get_state_groups_ids + self.store.register_events(graph.walk()) context_store = {} for event in graph.walk(): context = yield self.state.compute_event_context(event) - store.store_state_groups(event, context) + self.store.register_event_context(event, context) context_store[event.event_id] = context - self.assertEqual(2, len(context_store["D"].prev_state_ids)) + prev_state_ids = yield context_store["D"].get_prev_state_ids(self.store) + self.assertEqual(2, len(prev_state_ids)) @defer.inlineCallbacks def test_branch_basic_conflict(self): @@ -247,21 +247,20 @@ class StateTestCase(unittest.TestCase): } ) - store = StateGroupStore() - self.store.get_state_groups_ids.side_effect = store.get_state_groups_ids - self.store.get_events = store.get_events - store.register_events(graph.walk()) + self.store.register_events(graph.walk()) context_store = {} for event in graph.walk(): context = yield self.state.compute_event_context(event) - store.store_state_groups(event, context) + self.store.register_event_context(event, context) context_store[event.event_id] = context + prev_state_ids = yield context_store["D"].get_prev_state_ids(self.store) + self.assertSetEqual( {"START", "A", "C"}, - {e_id for e_id in context_store["D"].prev_state_ids.values()} + {e_id for e_id in prev_state_ids.values()} ) @defer.inlineCallbacks @@ -313,21 +312,20 @@ class StateTestCase(unittest.TestCase): } ) - store = StateGroupStore() - self.store.get_state_groups_ids.side_effect = store.get_state_groups_ids - self.store.get_events = store.get_events - store.register_events(graph.walk()) + self.store.register_events(graph.walk()) context_store = {} for event in graph.walk(): context = yield self.state.compute_event_context(event) - store.store_state_groups(event, context) + self.store.register_event_context(event, context) context_store[event.event_id] = context + prev_state_ids = yield context_store["E"].get_prev_state_ids(self.store) + self.assertSetEqual( {"START", "A", "B", "C"}, - {e for e in context_store["E"].prev_state_ids.values()} + {e for e in prev_state_ids.values()} ) @defer.inlineCallbacks @@ -396,21 +394,20 @@ class StateTestCase(unittest.TestCase): self._add_depths(nodes, edges) graph = Graph(nodes, edges) - store = StateGroupStore() - self.store.get_state_groups_ids.side_effect = store.get_state_groups_ids - self.store.get_events = store.get_events - store.register_events(graph.walk()) + self.store.register_events(graph.walk()) context_store = {} for event in graph.walk(): context = yield self.state.compute_event_context(event) - store.store_state_groups(event, context) + self.store.register_event_context(event, context) context_store[event.event_id] = context + prev_state_ids = yield context_store["D"].get_prev_state_ids(self.store) + self.assertSetEqual( {"A1", "A2", "A3", "A5", "B"}, - {e for e in context_store["D"].prev_state_ids.values()} + {e for e in prev_state_ids.values()} ) def _add_depths(self, nodes, edges): @@ -439,8 +436,10 @@ class StateTestCase(unittest.TestCase): event, old_state=old_state ) + current_state_ids = yield context.get_current_state_ids(self.store) + self.assertEqual( - set(e.event_id for e in old_state), set(context.current_state_ids.values()) + set(e.event_id for e in old_state), set(current_state_ids.values()) ) self.assertIsNotNone(context.state_group) @@ -459,13 +458,19 @@ class StateTestCase(unittest.TestCase): event, old_state=old_state ) + prev_state_ids = yield context.get_prev_state_ids(self.store) + self.assertEqual( - set(e.event_id for e in old_state), set(context.prev_state_ids.values()) + set(e.event_id for e in old_state), set(prev_state_ids.values()) ) @defer.inlineCallbacks def test_trivial_annotate_message(self): - event = create_event(type="test_message", name="event") + prev_event_id = "prev_event_id" + event = create_event( + type="test_message", name="event2", + prev_events=[(prev_event_id, {})], + ) old_state = [ create_event(type="test1", state_key="1"), @@ -473,24 +478,30 @@ class StateTestCase(unittest.TestCase): create_event(type="test2", state_key=""), ] - group_name = "group_name_1" - - self.store.get_state_groups_ids.return_value = { - group_name: {(e.type, e.state_key): e.event_id for e in old_state}, - } + group_name = self.store.store_state_group( + prev_event_id, event.room_id, None, None, + {(e.type, e.state_key): e.event_id for e in old_state}, + ) + self.store.register_event_id_state_group(prev_event_id, group_name) context = yield self.state.compute_event_context(event) + current_state_ids = yield context.get_current_state_ids(self.store) + self.assertEqual( set([e.event_id for e in old_state]), - set(context.current_state_ids.values()) + set(current_state_ids.values()) ) self.assertEqual(group_name, context.state_group) @defer.inlineCallbacks def test_trivial_annotate_state(self): - event = create_event(type="state", state_key="", name="event") + prev_event_id = "prev_event_id" + event = create_event( + type="state", state_key="", name="event2", + prev_events=[(prev_event_id, {})], + ) old_state = [ create_event(type="test1", state_key="1"), @@ -498,24 +509,31 @@ class StateTestCase(unittest.TestCase): create_event(type="test2", state_key=""), ] - group_name = "group_name_1" - - self.store.get_state_groups_ids.return_value = { - group_name: {(e.type, e.state_key): e.event_id for e in old_state}, - } + group_name = self.store.store_state_group( + prev_event_id, event.room_id, None, None, + {(e.type, e.state_key): e.event_id for e in old_state}, + ) + self.store.register_event_id_state_group(prev_event_id, group_name) context = yield self.state.compute_event_context(event) + prev_state_ids = yield context.get_prev_state_ids(self.store) + self.assertEqual( set([e.event_id for e in old_state]), - set(context.prev_state_ids.values()) + set(prev_state_ids.values()) ) self.assertIsNotNone(context.state_group) @defer.inlineCallbacks def test_resolve_message_conflict(self): - event = create_event(type="test_message", name="event") + prev_event_id1 = "event_id1" + prev_event_id2 = "event_id2" + event = create_event( + type="test_message", name="event3", + prev_events=[(prev_event_id1, {}), (prev_event_id2, {})], + ) creation = create_event( type=EventTypes.Create, state_key="" @@ -535,20 +553,27 @@ class StateTestCase(unittest.TestCase): create_event(type="test4", state_key=""), ] - store = StateGroupStore() - store.register_events(old_state_1) - store.register_events(old_state_2) - self.store.get_events = store.get_events + self.store.register_events(old_state_1) + self.store.register_events(old_state_2) - context = yield self._get_context(event, old_state_1, old_state_2) + context = yield self._get_context( + event, prev_event_id1, old_state_1, prev_event_id2, old_state_2, + ) - self.assertEqual(len(context.current_state_ids), 6) + current_state_ids = yield context.get_current_state_ids(self.store) + + self.assertEqual(len(current_state_ids), 6) self.assertIsNotNone(context.state_group) @defer.inlineCallbacks def test_resolve_state_conflict(self): - event = create_event(type="test4", state_key="", name="event") + prev_event_id1 = "event_id1" + prev_event_id2 = "event_id2" + event = create_event( + type="test4", state_key="", name="event", + prev_events=[(prev_event_id1, {}), (prev_event_id2, {})], + ) creation = create_event( type=EventTypes.Create, state_key="" @@ -573,15 +598,24 @@ class StateTestCase(unittest.TestCase): store.register_events(old_state_2) self.store.get_events = store.get_events - context = yield self._get_context(event, old_state_1, old_state_2) + context = yield self._get_context( + event, prev_event_id1, old_state_1, prev_event_id2, old_state_2, + ) - self.assertEqual(len(context.current_state_ids), 6) + current_state_ids = yield context.get_current_state_ids(self.store) + + self.assertEqual(len(current_state_ids), 6) self.assertIsNotNone(context.state_group) @defer.inlineCallbacks def test_standard_depth_conflict(self): - event = create_event(type="test4", name="event") + prev_event_id1 = "event_id1" + prev_event_id2 = "event_id2" + event = create_event( + type="test4", name="event", + prev_events=[(prev_event_id1, {}), (prev_event_id2, {})], + ) member_event = create_event( type=EventTypes.Member, @@ -591,6 +625,14 @@ class StateTestCase(unittest.TestCase): } ) + power_levels = create_event( + type=EventTypes.PowerLevels, state_key="", + content={"users": { + "@foo:bar": "100", + "@user_id:example.com": "100", + }} + ) + creation = create_event( type=EventTypes.Create, state_key="", content={"creator": "@foo:bar"} @@ -598,12 +640,14 @@ class StateTestCase(unittest.TestCase): old_state_1 = [ creation, + power_levels, member_event, create_event(type="test1", state_key="1", depth=1), ] old_state_2 = [ creation, + power_levels, member_event, create_event(type="test1", state_key="1", depth=2), ] @@ -613,10 +657,14 @@ class StateTestCase(unittest.TestCase): store.register_events(old_state_2) self.store.get_events = store.get_events - context = yield self._get_context(event, old_state_1, old_state_2) + context = yield self._get_context( + event, prev_event_id1, old_state_1, prev_event_id2, old_state_2, + ) + + current_state_ids = yield context.get_current_state_ids(self.store) self.assertEqual( - old_state_2[2].event_id, context.current_state_ids[("test1", "1")] + old_state_2[3].event_id, current_state_ids[("test1", "1")] ) # Reverse the depth to make sure we are actually using the depths @@ -624,12 +672,14 @@ class StateTestCase(unittest.TestCase): old_state_1 = [ creation, + power_levels, member_event, create_event(type="test1", state_key="1", depth=2), ] old_state_2 = [ creation, + power_levels, member_event, create_event(type="test1", state_key="1", depth=1), ] @@ -637,19 +687,28 @@ class StateTestCase(unittest.TestCase): store.register_events(old_state_1) store.register_events(old_state_2) - context = yield self._get_context(event, old_state_1, old_state_2) - - self.assertEqual( - old_state_1[2].event_id, context.current_state_ids[("test1", "1")] + context = yield self._get_context( + event, prev_event_id1, old_state_1, prev_event_id2, old_state_2, ) - def _get_context(self, event, old_state_1, old_state_2): - group_name_1 = "group_name_1" - group_name_2 = "group_name_2" + current_state_ids = yield context.get_current_state_ids(self.store) - self.store.get_state_groups_ids.return_value = { - group_name_1: {(e.type, e.state_key): e.event_id for e in old_state_1}, - group_name_2: {(e.type, e.state_key): e.event_id for e in old_state_2}, - } + self.assertEqual( + old_state_1[3].event_id, current_state_ids[("test1", "1")] + ) + + def _get_context(self, event, prev_event_id_1, old_state_1, prev_event_id_2, + old_state_2): + sg1 = self.store.store_state_group( + prev_event_id_1, event.room_id, None, None, + {(e.type, e.state_key): e.event_id for e in old_state_1}, + ) + self.store.register_event_id_state_group(prev_event_id_1, sg1) + + sg2 = self.store.store_state_group( + prev_event_id_2, event.room_id, None, None, + {(e.type, e.state_key): e.event_id for e in old_state_2}, + ) + self.store.register_event_id_state_group(prev_event_id_2, sg2) return self.state.compute_event_context(event) diff --git a/tests/test_test_utils.py b/tests/test_test_utils.py index d28bb726b..bc97c1224 100644 --- a/tests/test_test_utils.py +++ b/tests/test_test_utils.py @@ -14,7 +14,6 @@ # limitations under the License. from tests import unittest - from tests.utils import MockClock diff --git a/tests/test_types.py b/tests/test_types.py index 24d61dbe5..729bd676c 100644 --- a/tests/test_types.py +++ b/tests/test_types.py @@ -13,11 +13,11 @@ # See the License for the specific language governing permissions and # limitations under the License. -from tests import unittest - from synapse.api.errors import SynapseError from synapse.server import HomeServer -from synapse.types import UserID, RoomAlias +from synapse.types import GroupID, RoomAlias, UserID + +from tests import unittest mock_homeserver = HomeServer(hostname="my.domain") @@ -60,3 +60,25 @@ class RoomAliasTestCase(unittest.TestCase): room = RoomAlias("channel", "my.domain") self.assertEquals(room.to_string(), "#channel:my.domain") + + +class GroupIDTestCase(unittest.TestCase): + def test_parse(self): + group_id = GroupID.from_string("+group/=_-.123:my.domain") + self.assertEqual("group/=_-.123", group_id.localpart) + self.assertEqual("my.domain", group_id.domain) + + def test_validate(self): + bad_ids = [ + "$badsigil:domain", + "+:empty", + ] + [ + "+group" + c + ":domain" for c in "A%?æ£" + ] + for id_string in bad_ids: + try: + GroupID.from_string(id_string) + self.fail("Parsing '%s' should raise exception" % id_string) + except SynapseError as exc: + self.assertEqual(400, exc.code) + self.assertEqual("M_UNKNOWN", exc.errcode) diff --git a/tests/test_visibility.py b/tests/test_visibility.py new file mode 100644 index 000000000..0dc1a924d --- /dev/null +++ b/tests/test_visibility.py @@ -0,0 +1,324 @@ +# -*- coding: utf-8 -*- +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import logging + +from twisted.internet import defer +from twisted.internet.defer import succeed + +from synapse.events import FrozenEvent +from synapse.visibility import filter_events_for_server + +import tests.unittest +from tests.utils import setup_test_homeserver + +logger = logging.getLogger(__name__) + +TEST_ROOM_ID = "!TEST:ROOM" + + +class FilterEventsForServerTestCase(tests.unittest.TestCase): + @defer.inlineCallbacks + def setUp(self): + self.hs = yield setup_test_homeserver() + self.event_creation_handler = self.hs.get_event_creation_handler() + self.event_builder_factory = self.hs.get_event_builder_factory() + self.store = self.hs.get_datastore() + + @defer.inlineCallbacks + def test_filtering(self): + # + # The events to be filtered consist of 10 membership events (it doesn't + # really matter if they are joins or leaves, so let's make them joins). + # One of those membership events is going to be for a user on the + # server we are filtering for (so we can check the filtering is doing + # the right thing). + # + + # before we do that, we persist some other events to act as state. + self.inject_visibility("@admin:hs", "joined") + for i in range(0, 10): + yield self.inject_room_member("@resident%i:hs" % i) + + events_to_filter = [] + + for i in range(0, 10): + user = "@user%i:%s" % ( + i, "test_server" if i == 5 else "other_server" + ) + evt = yield self.inject_room_member(user, extra_content={"a": "b"}) + events_to_filter.append(evt) + + filtered = yield filter_events_for_server( + self.store, "test_server", events_to_filter, + ) + + # the result should be 5 redacted events, and 5 unredacted events. + for i in range(0, 5): + self.assertEqual(events_to_filter[i].event_id, filtered[i].event_id) + self.assertNotIn("a", filtered[i].content) + + for i in range(5, 10): + self.assertEqual(events_to_filter[i].event_id, filtered[i].event_id) + self.assertEqual(filtered[i].content["a"], "b") + + @tests.unittest.DEBUG + @defer.inlineCallbacks + def test_erased_user(self): + # 4 message events, from erased and unerased users, with a membership + # change in the middle of them. + events_to_filter = [] + + evt = yield self.inject_message("@unerased:local_hs") + events_to_filter.append(evt) + + evt = yield self.inject_message("@erased:local_hs") + events_to_filter.append(evt) + + evt = yield self.inject_room_member("@joiner:remote_hs") + events_to_filter.append(evt) + + evt = yield self.inject_message("@unerased:local_hs") + events_to_filter.append(evt) + + evt = yield self.inject_message("@erased:local_hs") + events_to_filter.append(evt) + + # the erasey user gets erased + self.hs.get_datastore().mark_user_erased("@erased:local_hs") + + # ... and the filtering happens. + filtered = yield filter_events_for_server( + self.store, "test_server", events_to_filter, + ) + + for i in range(0, len(events_to_filter)): + self.assertEqual( + events_to_filter[i].event_id, filtered[i].event_id, + "Unexpected event at result position %i" % (i, ) + ) + + for i in (0, 3): + self.assertEqual( + events_to_filter[i].content["body"], filtered[i].content["body"], + "Unexpected event content at result position %i" % (i,) + ) + + for i in (1, 4): + self.assertNotIn("body", filtered[i].content) + + @defer.inlineCallbacks + def inject_visibility(self, user_id, visibility): + content = {"history_visibility": visibility} + builder = self.event_builder_factory.new({ + "type": "m.room.history_visibility", + "sender": user_id, + "state_key": "", + "room_id": TEST_ROOM_ID, + "content": content, + }) + + event, context = yield self.event_creation_handler.create_new_client_event( + builder + ) + yield self.hs.get_datastore().persist_event(event, context) + defer.returnValue(event) + + @defer.inlineCallbacks + def inject_room_member(self, user_id, membership="join", extra_content={}): + content = {"membership": membership} + content.update(extra_content) + builder = self.event_builder_factory.new({ + "type": "m.room.member", + "sender": user_id, + "state_key": user_id, + "room_id": TEST_ROOM_ID, + "content": content, + }) + + event, context = yield self.event_creation_handler.create_new_client_event( + builder + ) + + yield self.hs.get_datastore().persist_event(event, context) + defer.returnValue(event) + + @defer.inlineCallbacks + def inject_message(self, user_id, content=None): + if content is None: + content = {"body": "testytest"} + builder = self.event_builder_factory.new({ + "type": "m.room.message", + "sender": user_id, + "room_id": TEST_ROOM_ID, + "content": content, + }) + + event, context = yield self.event_creation_handler.create_new_client_event( + builder + ) + + yield self.hs.get_datastore().persist_event(event, context) + defer.returnValue(event) + + @defer.inlineCallbacks + def test_large_room(self): + # see what happens when we have a large room with hundreds of thousands + # of membership events + + # As above, the events to be filtered consist of 10 membership events, + # where one of them is for a user on the server we are filtering for. + + import cProfile + import pstats + import time + + # we stub out the store, because building up all that state the normal + # way is very slow. + test_store = _TestStore() + + # our initial state is 100000 membership events and one + # history_visibility event. + room_state = [] + + history_visibility_evt = FrozenEvent({ + "event_id": "$history_vis", + "type": "m.room.history_visibility", + "sender": "@resident_user_0:test.com", + "state_key": "", + "room_id": TEST_ROOM_ID, + "content": {"history_visibility": "joined"}, + }) + room_state.append(history_visibility_evt) + test_store.add_event(history_visibility_evt) + + for i in range(0, 100000): + user = "@resident_user_%i:test.com" % (i, ) + evt = FrozenEvent({ + "event_id": "$res_event_%i" % (i, ), + "type": "m.room.member", + "state_key": user, + "sender": user, + "room_id": TEST_ROOM_ID, + "content": { + "membership": "join", + "extra": "zzz," + }, + }) + room_state.append(evt) + test_store.add_event(evt) + + events_to_filter = [] + for i in range(0, 10): + user = "@user%i:%s" % ( + i, "test_server" if i == 5 else "other_server" + ) + evt = FrozenEvent({ + "event_id": "$evt%i" % (i, ), + "type": "m.room.member", + "state_key": user, + "sender": user, + "room_id": TEST_ROOM_ID, + "content": { + "membership": "join", + "extra": "zzz", + }, + }) + events_to_filter.append(evt) + room_state.append(evt) + + test_store.add_event(evt) + test_store.set_state_ids_for_event(evt, { + (e.type, e.state_key): e.event_id for e in room_state + }) + + pr = cProfile.Profile() + pr.enable() + + logger.info("Starting filtering") + start = time.time() + filtered = yield filter_events_for_server( + test_store, "test_server", events_to_filter, + ) + logger.info("Filtering took %f seconds", time.time() - start) + + pr.disable() + with open("filter_events_for_server.profile", "w+") as f: + ps = pstats.Stats(pr, stream=f).sort_stats('cumulative') + ps.print_stats() + + # the result should be 5 redacted events, and 5 unredacted events. + for i in range(0, 5): + self.assertEqual(events_to_filter[i].event_id, filtered[i].event_id) + self.assertNotIn("extra", filtered[i].content) + + for i in range(5, 10): + self.assertEqual(events_to_filter[i].event_id, filtered[i].event_id) + self.assertEqual(filtered[i].content["extra"], "zzz") + + test_large_room.skip = "Disabled by default because it's slow" + + +class _TestStore(object): + """Implements a few methods of the DataStore, so that we can test + filter_events_for_server + + """ + def __init__(self): + # data for get_events: a map from event_id to event + self.events = {} + + # data for get_state_ids_for_events mock: a map from event_id to + # a map from (type_state_key) -> event_id for the state at that + # event + self.state_ids_for_events = {} + + def add_event(self, event): + self.events[event.event_id] = event + + def set_state_ids_for_event(self, event, state): + self.state_ids_for_events[event.event_id] = state + + def get_state_ids_for_events(self, events, types): + res = {} + include_memberships = False + for (type, state_key) in types: + if type == "m.room.history_visibility": + continue + if type != "m.room.member" or state_key is not None: + raise RuntimeError( + "Unimplemented: get_state_ids with type (%s, %s)" % + (type, state_key), + ) + include_memberships = True + + if include_memberships: + for event_id in events: + res[event_id] = self.state_ids_for_events[event_id] + + else: + k = ("m.room.history_visibility", "") + for event_id in events: + hve = self.state_ids_for_events[event_id][k] + res[event_id] = {k: hve} + + return succeed(res) + + def get_events(self, events): + return succeed({ + event_id: self.events[event_id] for event_id in events + }) + + def are_users_erased(self, users): + return succeed({u: False for u in users}) diff --git a/tests/unittest.py b/tests/unittest.py index 38715972d..b15b06726 100644 --- a/tests/unittest.py +++ b/tests/unittest.py @@ -13,22 +13,39 @@ # See the License for the specific language governing permissions and # limitations under the License. -from twisted.trial import unittest - import logging -# logging doesn't have a "don't log anything at all EVARRRR setting, -# but since the highest value is 50, 1000000 should do ;) -NEVER = 1000000 +import twisted +import twisted.logger +from twisted.trial import unittest -handler = logging.StreamHandler() -handler.setFormatter(logging.Formatter( - "%(levelname)s:%(name)s:%(message)s [%(pathname)s:%(lineno)d]" -)) -logging.getLogger().addHandler(handler) -logging.getLogger().setLevel(NEVER) -logging.getLogger("synapse.storage.SQL").setLevel(NEVER) -logging.getLogger("synapse.storage.txn").setLevel(NEVER) +from synapse.util.logcontext import LoggingContextFilter + +# Set up putting Synapse's logs into Trial's. +rootLogger = logging.getLogger() + +log_format = ( + "%(asctime)s - %(name)s - %(lineno)d - %(levelname)s - %(request)s - %(message)s" +) + + +class ToTwistedHandler(logging.Handler): + tx_log = twisted.logger.Logger() + + def emit(self, record): + log_entry = self.format(record) + log_level = record.levelname.lower().replace('warning', 'warn') + self.tx_log.emit( + twisted.logger.LogLevel.levelWithName(log_level), + log_entry.replace("{", r"(").replace("}", r")"), + ) + + +handler = ToTwistedHandler() +formatter = logging.Formatter(log_format) +handler.setFormatter(formatter) +handler.addFilter(LoggingContextFilter(request="")) +rootLogger.addHandler(handler) def around(target): @@ -61,10 +78,14 @@ class TestCase(unittest.TestCase): method = getattr(self, methodName) - level = getattr(method, "loglevel", getattr(self, "loglevel", NEVER)) + level = getattr(method, "loglevel", getattr(self, "loglevel", logging.ERROR)) @around(self) def setUp(orig): + # enable debugging of delayed calls - this means that we get a + # traceback when a unit test exits leaving things on the reactor. + twisted.internet.base.DelayedCall.debug = True + old_level = logging.getLogger().level if old_level != level: @@ -88,6 +109,17 @@ class TestCase(unittest.TestCase): except AssertionError as e: raise (type(e))(e.message + " for '.%s'" % key) + def assert_dict(self, required, actual): + """Does a partial assert of a dict. + + Args: + required (dict): The keys and value which MUST be in 'actual'. + actual (dict): The test result. Extra keys will not be checked. + """ + for key in required: + self.assertEquals(required[key], actual[key], + msg="%s mismatch. %s" % (key, actual)) + def DEBUG(target): """A decorator to set the .loglevel attribute to logging.DEBUG. diff --git a/tests/util/caches/test_descriptors.py b/tests/util/caches/test_descriptors.py index 3f14ab503..8176a7dab 100644 --- a/tests/util/caches/test_descriptors.py +++ b/tests/util/caches/test_descriptors.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -13,18 +14,71 @@ # See the License for the specific language governing permissions and # limitations under the License. import logging +from functools import partial import mock + +from twisted.internet import defer, reactor + from synapse.api.errors import SynapseError -from synapse.util import async from synapse.util import logcontext -from twisted.internet import defer from synapse.util.caches import descriptors + from tests import unittest logger = logging.getLogger(__name__) +def run_on_reactor(): + d = defer.Deferred() + reactor.callLater(0, d.callback, 0) + return logcontext.make_deferred_yieldable(d) + + +class CacheTestCase(unittest.TestCase): + def test_invalidate_all(self): + cache = descriptors.Cache("testcache") + + callback_record = [False, False] + + def record_callback(idx): + callback_record[idx] = True + + # add a couple of pending entries + d1 = defer.Deferred() + cache.set("key1", d1, partial(record_callback, 0)) + + d2 = defer.Deferred() + cache.set("key2", d2, partial(record_callback, 1)) + + # lookup should return the deferreds + self.assertIs(cache.get("key1"), d1) + self.assertIs(cache.get("key2"), d2) + + # let one of the lookups complete + d2.callback("result2") + self.assertEqual(cache.get("key2"), "result2") + + # now do the invalidation + cache.invalidate_all() + + # lookup should return none + self.assertIsNone(cache.get("key1", None)) + self.assertIsNone(cache.get("key2", None)) + + # both callbacks should have been callbacked + self.assertTrue( + callback_record[0], "Invalidation callback for key1 not called", + ) + self.assertTrue( + callback_record[1], "Invalidation callback for key2 not called", + ) + + # letting the other lookup complete should do nothing + d1.callback("result1") + self.assertIsNone(cache.get("key1", None)) + + class DescriptorTestCase(unittest.TestCase): @defer.inlineCallbacks def test_cache(self): @@ -149,7 +203,8 @@ class DescriptorTestCase(unittest.TestCase): def fn(self, arg1): @defer.inlineCallbacks def inner_fn(): - yield async.run_on_reactor() + # we want this to behave like an asynchronous function + yield run_on_reactor() raise SynapseError(400, "blah") return inner_fn() @@ -159,7 +214,12 @@ class DescriptorTestCase(unittest.TestCase): with logcontext.LoggingContext() as c1: c1.name = "c1" try: - yield obj.fn(1) + d = obj.fn(1) + self.assertEqual( + logcontext.LoggingContext.current_context(), + logcontext.LoggingContext.sentinel, + ) + yield d self.fail("No exception thrown") except SynapseError: pass diff --git a/tests/util/test_dict_cache.py b/tests/util/test_dict_cache.py index bc92f85fa..26f2fa580 100644 --- a/tests/util/test_dict_cache.py +++ b/tests/util/test_dict_cache.py @@ -14,10 +14,10 @@ # limitations under the License. -from tests import unittest - from synapse.util.caches.dictionary_cache import DictionaryCache +from tests import unittest + class DictCacheTestCase(unittest.TestCase): @@ -32,7 +32,7 @@ class DictCacheTestCase(unittest.TestCase): seq = self.cache.sequence test_value = {"test": "test_simple_cache_hit_full"} - self.cache.update(seq, key, test_value, full=True) + self.cache.update(seq, key, test_value) c = self.cache.get(key) self.assertEqual(test_value, c.value) @@ -44,7 +44,7 @@ class DictCacheTestCase(unittest.TestCase): test_value = { "test": "test_simple_cache_hit_partial" } - self.cache.update(seq, key, test_value, full=True) + self.cache.update(seq, key, test_value) c = self.cache.get(key, ["test"]) self.assertEqual(test_value, c.value) @@ -56,7 +56,7 @@ class DictCacheTestCase(unittest.TestCase): test_value = { "test": "test_simple_cache_miss_partial" } - self.cache.update(seq, key, test_value, full=True) + self.cache.update(seq, key, test_value) c = self.cache.get(key, ["test2"]) self.assertEqual({}, c.value) @@ -70,7 +70,7 @@ class DictCacheTestCase(unittest.TestCase): "test2": "test_simple_cache_hit_miss_partial2", "test3": "test_simple_cache_hit_miss_partial3", } - self.cache.update(seq, key, test_value, full=True) + self.cache.update(seq, key, test_value) c = self.cache.get(key, ["test2"]) self.assertEqual({"test2": "test_simple_cache_hit_miss_partial2"}, c.value) @@ -82,13 +82,13 @@ class DictCacheTestCase(unittest.TestCase): test_value_1 = { "test": "test_simple_cache_hit_miss_partial", } - self.cache.update(seq, key, test_value_1, full=False) + self.cache.update(seq, key, test_value_1, fetched_keys=set("test")) seq = self.cache.sequence test_value_2 = { "test2": "test_simple_cache_hit_miss_partial2", } - self.cache.update(seq, key, test_value_2, full=False) + self.cache.update(seq, key, test_value_2, fetched_keys=set("test2")) c = self.cache.get(key) self.assertEqual( diff --git a/tests/util/test_expiring_cache.py b/tests/util/test_expiring_cache.py index 31d24adb8..d12b5e838 100644 --- a/tests/util/test_expiring_cache.py +++ b/tests/util/test_expiring_cache.py @@ -14,12 +14,12 @@ # limitations under the License. -from .. import unittest - from synapse.util.caches.expiringcache import ExpiringCache from tests.utils import MockClock +from .. import unittest + class ExpiringCacheTestCase(unittest.TestCase): diff --git a/tests/util/test_file_consumer.py b/tests/util/test_file_consumer.py new file mode 100644 index 000000000..7ce5f8c25 --- /dev/null +++ b/tests/util/test_file_consumer.py @@ -0,0 +1,177 @@ +# -*- coding: utf-8 -*- +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + + +import threading + +from mock import NonCallableMock +from six import StringIO + +from twisted.internet import defer, reactor + +from synapse.util.file_consumer import BackgroundFileConsumer + +from tests import unittest + + +class FileConsumerTests(unittest.TestCase): + + @defer.inlineCallbacks + def test_pull_consumer(self): + string_file = StringIO() + consumer = BackgroundFileConsumer(string_file, reactor=reactor) + + try: + producer = DummyPullProducer() + + yield producer.register_with_consumer(consumer) + + yield producer.write_and_wait("Foo") + + self.assertEqual(string_file.getvalue(), "Foo") + + yield producer.write_and_wait("Bar") + + self.assertEqual(string_file.getvalue(), "FooBar") + finally: + consumer.unregisterProducer() + + yield consumer.wait() + + self.assertTrue(string_file.closed) + + @defer.inlineCallbacks + def test_push_consumer(self): + string_file = BlockingStringWrite() + consumer = BackgroundFileConsumer(string_file, reactor=reactor) + + try: + producer = NonCallableMock(spec_set=[]) + + consumer.registerProducer(producer, True) + + consumer.write("Foo") + yield string_file.wait_for_n_writes(1) + + self.assertEqual(string_file.buffer, "Foo") + + consumer.write("Bar") + yield string_file.wait_for_n_writes(2) + + self.assertEqual(string_file.buffer, "FooBar") + finally: + consumer.unregisterProducer() + + yield consumer.wait() + + self.assertTrue(string_file.closed) + + @defer.inlineCallbacks + def test_push_producer_feedback(self): + string_file = BlockingStringWrite() + consumer = BackgroundFileConsumer(string_file, reactor=reactor) + + try: + producer = NonCallableMock(spec_set=["pauseProducing", "resumeProducing"]) + + resume_deferred = defer.Deferred() + producer.resumeProducing.side_effect = lambda: resume_deferred.callback(None) + + consumer.registerProducer(producer, True) + + number_writes = 0 + with string_file.write_lock: + for _ in range(consumer._PAUSE_ON_QUEUE_SIZE): + consumer.write("Foo") + number_writes += 1 + + producer.pauseProducing.assert_called_once() + + yield string_file.wait_for_n_writes(number_writes) + + yield resume_deferred + producer.resumeProducing.assert_called_once() + finally: + consumer.unregisterProducer() + + yield consumer.wait() + + self.assertTrue(string_file.closed) + + +class DummyPullProducer(object): + def __init__(self): + self.consumer = None + self.deferred = defer.Deferred() + + def resumeProducing(self): + d = self.deferred + self.deferred = defer.Deferred() + d.callback(None) + + def write_and_wait(self, bytes): + d = self.deferred + self.consumer.write(bytes) + return d + + def register_with_consumer(self, consumer): + d = self.deferred + self.consumer = consumer + self.consumer.registerProducer(self, False) + return d + + +class BlockingStringWrite(object): + def __init__(self): + self.buffer = "" + self.closed = False + self.write_lock = threading.Lock() + + self._notify_write_deferred = None + self._number_of_writes = 0 + + def write(self, bytes): + with self.write_lock: + self.buffer += bytes + self._number_of_writes += 1 + + reactor.callFromThread(self._notify_write) + + def close(self): + self.closed = True + + def _notify_write(self): + "Called by write to indicate a write happened" + with self.write_lock: + if not self._notify_write_deferred: + return + d = self._notify_write_deferred + self._notify_write_deferred = None + d.callback(None) + + @defer.inlineCallbacks + def wait_for_n_writes(self, n): + "Wait for n writes to have happened" + while True: + with self.write_lock: + if n <= self._number_of_writes: + return + + if not self._notify_write_deferred: + self._notify_write_deferred = defer.Deferred() + + d = self._notify_write_deferred + + yield d diff --git a/tests/util/test_limiter.py b/tests/util/test_limiter.py deleted file mode 100644 index 9c795d9fd..000000000 --- a/tests/util/test_limiter.py +++ /dev/null @@ -1,70 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2016 OpenMarket Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - - -from tests import unittest - -from twisted.internet import defer - -from synapse.util.async import Limiter - - -class LimiterTestCase(unittest.TestCase): - - @defer.inlineCallbacks - def test_limiter(self): - limiter = Limiter(3) - - key = object() - - d1 = limiter.queue(key) - cm1 = yield d1 - - d2 = limiter.queue(key) - cm2 = yield d2 - - d3 = limiter.queue(key) - cm3 = yield d3 - - d4 = limiter.queue(key) - self.assertFalse(d4.called) - - d5 = limiter.queue(key) - self.assertFalse(d5.called) - - with cm1: - self.assertFalse(d4.called) - self.assertFalse(d5.called) - - self.assertTrue(d4.called) - self.assertFalse(d5.called) - - with cm3: - self.assertFalse(d5.called) - - self.assertTrue(d5.called) - - with cm2: - pass - - with (yield d4): - pass - - with (yield d5): - pass - - d6 = limiter.queue(key) - with (yield d6): - pass diff --git a/tests/util/test_linearizer.py b/tests/util/test_linearizer.py index afcba482f..4729bd5a0 100644 --- a/tests/util/test_linearizer.py +++ b/tests/util/test_linearizer.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -13,13 +14,16 @@ # See the License for the specific language governing permissions and # limitations under the License. +from six.moves import range + +from twisted.internet import defer, reactor +from twisted.internet.defer import CancelledError + +from synapse.util import Clock, logcontext +from synapse.util.async import Linearizer from tests import unittest -from twisted.internet import defer - -from synapse.util.async import Linearizer - class LinearizerTestCase(unittest.TestCase): @@ -38,7 +42,104 @@ class LinearizerTestCase(unittest.TestCase): with cm1: self.assertFalse(d2.called) - self.assertTrue(d2.called) - with (yield d2): pass + + def test_lots_of_queued_things(self): + # we have one slow thing, and lots of fast things queued up behind it. + # it should *not* explode the stack. + linearizer = Linearizer() + + @defer.inlineCallbacks + def func(i, sleep=False): + with logcontext.LoggingContext("func(%s)" % i) as lc: + with (yield linearizer.queue("")): + self.assertEqual( + logcontext.LoggingContext.current_context(), lc) + if sleep: + yield Clock(reactor).sleep(0) + + self.assertEqual( + logcontext.LoggingContext.current_context(), lc) + + func(0, sleep=True) + for i in range(1, 100): + func(i) + + return func(1000) + + @defer.inlineCallbacks + def test_multiple_entries(self): + limiter = Linearizer(max_count=3) + + key = object() + + d1 = limiter.queue(key) + cm1 = yield d1 + + d2 = limiter.queue(key) + cm2 = yield d2 + + d3 = limiter.queue(key) + cm3 = yield d3 + + d4 = limiter.queue(key) + self.assertFalse(d4.called) + + d5 = limiter.queue(key) + self.assertFalse(d5.called) + + with cm1: + self.assertFalse(d4.called) + self.assertFalse(d5.called) + + cm4 = yield d4 + self.assertFalse(d5.called) + + with cm3: + self.assertFalse(d5.called) + + cm5 = yield d5 + + with cm2: + pass + + with cm4: + pass + + with cm5: + pass + + d6 = limiter.queue(key) + with (yield d6): + pass + + @defer.inlineCallbacks + def test_cancellation(self): + linearizer = Linearizer() + + key = object() + + d1 = linearizer.queue(key) + cm1 = yield d1 + + d2 = linearizer.queue(key) + self.assertFalse(d2.called) + + d3 = linearizer.queue(key) + self.assertFalse(d3.called) + + d2.cancel() + + with cm1: + pass + + self.assertTrue(d2.called) + try: + yield d2 + self.fail("Expected d2 to raise CancelledError") + except CancelledError: + pass + + with (yield d3): + pass diff --git a/tests/util/test_log_context.py b/tests/util/test_log_context.py deleted file mode 100644 index 9ffe209c4..000000000 --- a/tests/util/test_log_context.py +++ /dev/null @@ -1,96 +0,0 @@ -import twisted.python.failure -from twisted.internet import defer -from twisted.internet import reactor -from .. import unittest - -from synapse.util.async import sleep -from synapse.util import logcontext -from synapse.util.logcontext import LoggingContext - - -class LoggingContextTestCase(unittest.TestCase): - - def _check_test_key(self, value): - self.assertEquals( - LoggingContext.current_context().test_key, value - ) - - def test_with_context(self): - with LoggingContext() as context_one: - context_one.test_key = "test" - self._check_test_key("test") - - @defer.inlineCallbacks - def test_sleep(self): - @defer.inlineCallbacks - def competing_callback(): - with LoggingContext() as competing_context: - competing_context.test_key = "competing" - yield sleep(0) - self._check_test_key("competing") - - reactor.callLater(0, competing_callback) - - with LoggingContext() as context_one: - context_one.test_key = "one" - yield sleep(0) - self._check_test_key("one") - - def _test_preserve_fn(self, function): - sentinel_context = LoggingContext.current_context() - - callback_completed = [False] - - @defer.inlineCallbacks - def cb(): - context_one.test_key = "one" - yield function() - self._check_test_key("one") - - callback_completed[0] = True - - with LoggingContext() as context_one: - context_one.test_key = "one" - - # fire off function, but don't wait on it. - logcontext.preserve_fn(cb)() - - self._check_test_key("one") - - # now wait for the function under test to have run, and check that - # the logcontext is left in a sane state. - d2 = defer.Deferred() - - def check_logcontext(): - if not callback_completed[0]: - reactor.callLater(0.01, check_logcontext) - return - - # make sure that the context was reset before it got thrown back - # into the reactor - try: - self.assertIs(LoggingContext.current_context(), - sentinel_context) - d2.callback(None) - except BaseException: - d2.errback(twisted.python.failure.Failure()) - - reactor.callLater(0.01, check_logcontext) - - # test is done once d2 finishes - return d2 - - def test_preserve_fn_with_blocking_fn(self): - @defer.inlineCallbacks - def blocking_function(): - yield sleep(0) - - return self._test_preserve_fn(blocking_function) - - def test_preserve_fn_with_non_blocking_fn(self): - @defer.inlineCallbacks - def nonblocking_function(): - with logcontext.PreserveLoggingContext(): - yield defer.succeed(None) - - return self._test_preserve_fn(nonblocking_function) diff --git a/tests/util/test_logcontext.py b/tests/util/test_logcontext.py new file mode 100644 index 000000000..c54001f7a --- /dev/null +++ b/tests/util/test_logcontext.py @@ -0,0 +1,179 @@ +import twisted.python.failure +from twisted.internet import defer, reactor + +from synapse.util import Clock, logcontext +from synapse.util.logcontext import LoggingContext + +from .. import unittest + + +class LoggingContextTestCase(unittest.TestCase): + + def _check_test_key(self, value): + self.assertEquals( + LoggingContext.current_context().request, value + ) + + def test_with_context(self): + with LoggingContext() as context_one: + context_one.request = "test" + self._check_test_key("test") + + @defer.inlineCallbacks + def test_sleep(self): + clock = Clock(reactor) + + @defer.inlineCallbacks + def competing_callback(): + with LoggingContext() as competing_context: + competing_context.request = "competing" + yield clock.sleep(0) + self._check_test_key("competing") + + reactor.callLater(0, competing_callback) + + with LoggingContext() as context_one: + context_one.request = "one" + yield clock.sleep(0) + self._check_test_key("one") + + def _test_run_in_background(self, function): + sentinel_context = LoggingContext.current_context() + + callback_completed = [False] + + def test(): + context_one.request = "one" + d = function() + + def cb(res): + self._check_test_key("one") + callback_completed[0] = True + return res + d.addCallback(cb) + + return d + + with LoggingContext() as context_one: + context_one.request = "one" + + # fire off function, but don't wait on it. + logcontext.run_in_background(test) + + self._check_test_key("one") + + # now wait for the function under test to have run, and check that + # the logcontext is left in a sane state. + d2 = defer.Deferred() + + def check_logcontext(): + if not callback_completed[0]: + reactor.callLater(0.01, check_logcontext) + return + + # make sure that the context was reset before it got thrown back + # into the reactor + try: + self.assertIs(LoggingContext.current_context(), + sentinel_context) + d2.callback(None) + except BaseException: + d2.errback(twisted.python.failure.Failure()) + + reactor.callLater(0.01, check_logcontext) + + # test is done once d2 finishes + return d2 + + def test_run_in_background_with_blocking_fn(self): + @defer.inlineCallbacks + def blocking_function(): + yield Clock(reactor).sleep(0) + + return self._test_run_in_background(blocking_function) + + def test_run_in_background_with_non_blocking_fn(self): + @defer.inlineCallbacks + def nonblocking_function(): + with logcontext.PreserveLoggingContext(): + yield defer.succeed(None) + + return self._test_run_in_background(nonblocking_function) + + def test_run_in_background_with_chained_deferred(self): + # a function which returns a deferred which looks like it has been + # called, but is actually paused + def testfunc(): + return logcontext.make_deferred_yieldable( + _chained_deferred_function() + ) + + return self._test_run_in_background(testfunc) + + @defer.inlineCallbacks + def test_make_deferred_yieldable(self): + # a function which retuns an incomplete deferred, but doesn't follow + # the synapse rules. + def blocking_function(): + d = defer.Deferred() + reactor.callLater(0, d.callback, None) + return d + + sentinel_context = LoggingContext.current_context() + + with LoggingContext() as context_one: + context_one.request = "one" + + d1 = logcontext.make_deferred_yieldable(blocking_function()) + # make sure that the context was reset by make_deferred_yieldable + self.assertIs(LoggingContext.current_context(), sentinel_context) + + yield d1 + + # now it should be restored + self._check_test_key("one") + + @defer.inlineCallbacks + def test_make_deferred_yieldable_with_chained_deferreds(self): + sentinel_context = LoggingContext.current_context() + + with LoggingContext() as context_one: + context_one.request = "one" + + d1 = logcontext.make_deferred_yieldable(_chained_deferred_function()) + # make sure that the context was reset by make_deferred_yieldable + self.assertIs(LoggingContext.current_context(), sentinel_context) + + yield d1 + + # now it should be restored + self._check_test_key("one") + + @defer.inlineCallbacks + def test_make_deferred_yieldable_on_non_deferred(self): + """Check that make_deferred_yieldable does the right thing when its + argument isn't actually a deferred""" + + with LoggingContext() as context_one: + context_one.request = "one" + + d1 = logcontext.make_deferred_yieldable("bum") + self._check_test_key("one") + + r = yield d1 + self.assertEqual(r, "bum") + self._check_test_key("one") + + +# a function which returns a deferred which has been "called", but +# which had a function which returned another incomplete deferred on +# its callback list, so won't yet call any other new callbacks. +def _chained_deferred_function(): + d = defer.succeed(None) + + def cb(res): + d2 = defer.Deferred() + reactor.callLater(0, d2.callback, res) + return d2 + d.addCallback(cb) + return d diff --git a/tests/util/test_clock.py b/tests/util/test_logformatter.py similarity index 52% rename from tests/util/test_clock.py rename to tests/util/test_logformatter.py index 967260357..297aebbfb 100644 --- a/tests/util/test_clock.py +++ b/tests/util/test_logformatter.py @@ -1,5 +1,5 @@ # -*- coding: utf-8 -*- -# Copyright 2017 Vector Creations Ltd +# Copyright 2018 New Vector Ltd # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -12,22 +12,28 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. -from synapse import util -from twisted.internet import defer +import sys + +from synapse.util.logformatter import LogFormatter + from tests import unittest -class ClockTestCase(unittest.TestCase): - @defer.inlineCallbacks - def test_time_bound_deferred(self): - # just a deferred which never resolves - slow_deferred = defer.Deferred() +class TestException(Exception): + pass - clock = util.Clock() - time_bound = clock.time_bound_deferred(slow_deferred, 0.001) + +class LogFormatterTestCase(unittest.TestCase): + def test_formatter(self): + formatter = LogFormatter() try: - yield time_bound - self.fail("Expected timedout error, but got nothing") - except util.DeferredTimedOutError: - pass + raise TestException("testytest") + except TestException: + ei = sys.exc_info() + + output = formatter.formatException(ei) + + # check the output looks vaguely sane + self.assertIn("testytest", output) + self.assertIn("Capture point", output) diff --git a/tests/util/test_lrucache.py b/tests/util/test_lrucache.py index dfb78cb8b..9b36ef448 100644 --- a/tests/util/test_lrucache.py +++ b/tests/util/test_lrucache.py @@ -14,12 +14,12 @@ # limitations under the License. -from .. import unittest +from mock import Mock from synapse.util.caches.lrucache import LruCache from synapse.util.caches.treecache import TreeCache -from mock import Mock +from .. import unittest class LruCacheTestCase(unittest.TestCase): diff --git a/tests/util/test_rwlock.py b/tests/util/test_rwlock.py index 1d745ae1a..24194e3b2 100644 --- a/tests/util/test_rwlock.py +++ b/tests/util/test_rwlock.py @@ -14,10 +14,10 @@ # limitations under the License. -from tests import unittest - from synapse.util.async import ReadWriteLock +from tests import unittest + class ReadWriteLockTestCase(unittest.TestCase): diff --git a/tests/util/test_snapshot_cache.py b/tests/util/test_snapshot_cache.py index d3a8630c2..0f5b32fcc 100644 --- a/tests/util/test_snapshot_cache.py +++ b/tests/util/test_snapshot_cache.py @@ -14,10 +14,11 @@ # limitations under the License. -from .. import unittest +from twisted.internet.defer import Deferred from synapse.util.caches.snapshot_cache import SnapshotCache -from twisted.internet.defer import Deferred + +from .. import unittest class SnapshotCacheTestCase(unittest.TestCase): diff --git a/tests/util/test_stream_change_cache.py b/tests/util/test_stream_change_cache.py new file mode 100644 index 000000000..65b0f2e6f --- /dev/null +++ b/tests/util/test_stream_change_cache.py @@ -0,0 +1,215 @@ +from mock import patch + +from synapse.util.caches.stream_change_cache import StreamChangeCache + +from tests import unittest + + +class StreamChangeCacheTests(unittest.TestCase): + """ + Tests for StreamChangeCache. + """ + + def test_prefilled_cache(self): + """ + Providing a prefilled cache to StreamChangeCache will result in a cache + with the prefilled-cache entered in. + """ + cache = StreamChangeCache("#test", 1, prefilled_cache={"user@foo.com": 2}) + self.assertTrue(cache.has_entity_changed("user@foo.com", 1)) + + def test_has_entity_changed(self): + """ + StreamChangeCache.entity_has_changed will mark entities as changed, and + has_entity_changed will observe the changed entities. + """ + cache = StreamChangeCache("#test", 3) + + cache.entity_has_changed("user@foo.com", 6) + cache.entity_has_changed("bar@baz.net", 7) + + # If it's been changed after that stream position, return True + self.assertTrue(cache.has_entity_changed("user@foo.com", 4)) + self.assertTrue(cache.has_entity_changed("bar@baz.net", 4)) + + # If it's been changed at that stream position, return False + self.assertFalse(cache.has_entity_changed("user@foo.com", 6)) + + # If there's no changes after that stream position, return False + self.assertFalse(cache.has_entity_changed("user@foo.com", 7)) + + # If the entity does not exist, return False. + self.assertFalse(cache.has_entity_changed("not@here.website", 7)) + + # If we request before the stream cache's earliest known position, + # return True, whether it's a known entity or not. + self.assertTrue(cache.has_entity_changed("user@foo.com", 0)) + self.assertTrue(cache.has_entity_changed("not@here.website", 0)) + + @patch("synapse.util.caches.CACHE_SIZE_FACTOR", 1.0) + def test_has_entity_changed_pops_off_start(self): + """ + StreamChangeCache.entity_has_changed will respect the max size and + purge the oldest items upon reaching that max size. + """ + cache = StreamChangeCache("#test", 1, max_size=2) + + cache.entity_has_changed("user@foo.com", 2) + cache.entity_has_changed("bar@baz.net", 3) + cache.entity_has_changed("user@elsewhere.org", 4) + + # The cache is at the max size, 2 + self.assertEqual(len(cache._cache), 2) + + # The oldest item has been popped off + self.assertTrue("user@foo.com" not in cache._entity_to_key) + + # If we update an existing entity, it keeps the two existing entities + cache.entity_has_changed("bar@baz.net", 5) + self.assertEqual( + set(["bar@baz.net", "user@elsewhere.org"]), set(cache._entity_to_key) + ) + + def test_get_all_entities_changed(self): + """ + StreamChangeCache.get_all_entities_changed will return all changed + entities since the given position. If the position is before the start + of the known stream, it returns None instead. + """ + cache = StreamChangeCache("#test", 1) + + cache.entity_has_changed("user@foo.com", 2) + cache.entity_has_changed("bar@baz.net", 3) + cache.entity_has_changed("user@elsewhere.org", 4) + + self.assertEqual( + cache.get_all_entities_changed(1), + ["user@foo.com", "bar@baz.net", "user@elsewhere.org"], + ) + self.assertEqual( + cache.get_all_entities_changed(2), ["bar@baz.net", "user@elsewhere.org"] + ) + self.assertEqual(cache.get_all_entities_changed(3), ["user@elsewhere.org"]) + self.assertEqual(cache.get_all_entities_changed(0), None) + + def test_has_any_entity_changed(self): + """ + StreamChangeCache.has_any_entity_changed will return True if any + entities have been changed since the provided stream position, and + False if they have not. If the cache has entries and the provided + stream position is before it, it will return True, otherwise False if + the cache has no entries. + """ + cache = StreamChangeCache("#test", 1) + + # With no entities, it returns False for the past, present, and future. + self.assertFalse(cache.has_any_entity_changed(0)) + self.assertFalse(cache.has_any_entity_changed(1)) + self.assertFalse(cache.has_any_entity_changed(2)) + + # We add an entity + cache.entity_has_changed("user@foo.com", 2) + + # With an entity, it returns True for the past, the stream start + # position, and False for the stream position the entity was changed + # on and ones after it. + self.assertTrue(cache.has_any_entity_changed(0)) + self.assertTrue(cache.has_any_entity_changed(1)) + self.assertFalse(cache.has_any_entity_changed(2)) + self.assertFalse(cache.has_any_entity_changed(3)) + + def test_get_entities_changed(self): + """ + StreamChangeCache.get_entities_changed will return the entities in the + given list that have changed since the provided stream ID. If the + stream position is earlier than the earliest known position, it will + return all of the entities queried for. + """ + cache = StreamChangeCache("#test", 1) + + cache.entity_has_changed("user@foo.com", 2) + cache.entity_has_changed("bar@baz.net", 3) + cache.entity_has_changed("user@elsewhere.org", 4) + + # Query all the entries, but mid-way through the stream. We should only + # get the ones after that point. + self.assertEqual( + cache.get_entities_changed( + ["user@foo.com", "bar@baz.net", "user@elsewhere.org"], stream_pos=2 + ), + set(["bar@baz.net", "user@elsewhere.org"]), + ) + + # Query all the entries mid-way through the stream, but include one + # that doesn't exist in it. We shouldn't get back the one that doesn't + # exist. + self.assertEqual( + cache.get_entities_changed( + [ + "user@foo.com", + "bar@baz.net", + "user@elsewhere.org", + "not@here.website", + ], + stream_pos=2, + ), + set(["bar@baz.net", "user@elsewhere.org"]), + ) + + # Query all the entries, but before the first known point. We will get + # all the entries we queried for, including ones that don't exist. + self.assertEqual( + cache.get_entities_changed( + [ + "user@foo.com", + "bar@baz.net", + "user@elsewhere.org", + "not@here.website", + ], + stream_pos=0, + ), + set( + [ + "user@foo.com", + "bar@baz.net", + "user@elsewhere.org", + "not@here.website", + ] + ), + ) + + # Query a subset of the entries mid-way through the stream. We should + # only get back the subset. + self.assertEqual( + cache.get_entities_changed( + [ + "bar@baz.net", + ], + stream_pos=2, + ), + set( + [ + "bar@baz.net", + ] + ), + ) + + def test_max_pos(self): + """ + StreamChangeCache.get_max_pos_of_last_change will return the most + recent point where the entity could have changed. If the entity is not + known, the stream start is provided instead. + """ + cache = StreamChangeCache("#test", 1) + + cache.entity_has_changed("user@foo.com", 2) + cache.entity_has_changed("bar@baz.net", 3) + cache.entity_has_changed("user@elsewhere.org", 4) + + # Known entities will return the point where they were changed. + self.assertEqual(cache.get_max_pos_of_last_change("user@foo.com"), 2) + self.assertEqual(cache.get_max_pos_of_last_change("bar@baz.net"), 3) + self.assertEqual(cache.get_max_pos_of_last_change("user@elsewhere.org"), 4) + + # Unknown entities will return the stream start position. + self.assertEqual(cache.get_max_pos_of_last_change("not@here.website"), 1) diff --git a/tests/util/test_treecache.py b/tests/util/test_treecache.py index 7ab578a18..a5f226120 100644 --- a/tests/util/test_treecache.py +++ b/tests/util/test_treecache.py @@ -14,10 +14,10 @@ # limitations under the License. -from .. import unittest - from synapse.util.caches.treecache import TreeCache +from .. import unittest + class TreeCacheTestCase(unittest.TestCase): def test_get_set_onelevel(self): diff --git a/tests/util/test_wheel_timer.py b/tests/util/test_wheel_timer.py index c44567e52..03201a4d9 100644 --- a/tests/util/test_wheel_timer.py +++ b/tests/util/test_wheel_timer.py @@ -13,10 +13,10 @@ # See the License for the specific language governing permissions and # limitations under the License. -from .. import unittest - from synapse.util.wheel_timer import WheelTimer +from .. import unittest + class WheelTimerTestCase(unittest.TestCase): def test_single_insert_fetch(self): @@ -33,7 +33,7 @@ class WheelTimerTestCase(unittest.TestCase): self.assertListEqual(wheel.fetch(156), [obj]) self.assertListEqual(wheel.fetch(170), []) - def test_mutli_insert(self): + def test_multi_insert(self): wheel = WheelTimer(bucket_size=5) obj1 = object() @@ -58,7 +58,7 @@ class WheelTimerTestCase(unittest.TestCase): wheel.insert(100, obj, 50) self.assertListEqual(wheel.fetch(120), [obj]) - def test_insert_past_mutli(self): + def test_insert_past_multi(self): wheel = WheelTimer(bucket_size=5) obj1 = object() diff --git a/tests/utils.py b/tests/utils.py index 4f7e32b3a..c3dbff850 100644 --- a/tests/utils.py +++ b/tests/utils.py @@ -13,35 +13,40 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.http.server import HttpServer -from synapse.api.errors import cs_error, CodeMessageException, StoreError -from synapse.api.constants import EventTypes -from synapse.storage.prepare_database import prepare_database -from synapse.storage.engines import create_engine -from synapse.server import HomeServer -from synapse.federation.transport import server -from synapse.util.ratelimitutils import FederationRateLimiter +import hashlib +from inspect import getcallargs -from synapse.util.logcontext import LoggingContext +from mock import Mock, patch +from six.moves.urllib import parse as urlparse from twisted.internet import defer, reactor -from twisted.enterprise.adbapi import ConnectionPool -from collections import namedtuple -from mock import patch, Mock -import hashlib -import urllib -import urlparse +from synapse.api.errors import CodeMessageException, cs_error +from synapse.federation.transport import server +from synapse.http.server import HttpServer +from synapse.server import HomeServer +from synapse.storage import PostgresEngine +from synapse.storage.engines import create_engine +from synapse.storage.prepare_database import prepare_database +from synapse.util.logcontext import LoggingContext +from synapse.util.ratelimitutils import FederationRateLimiter -from inspect import getcallargs +# set this to True to run the tests against postgres instead of sqlite. +# It requires you to have a local postgres database called synapse_test, within +# which ALL TABLES WILL BE DROPPED +USE_POSTGRES_FOR_TESTS = False @defer.inlineCallbacks -def setup_test_homeserver(name="test", datastore=None, config=None, **kargs): +def setup_test_homeserver(name="test", datastore=None, config=None, reactor=None, + **kargs): """Setup a homeserver suitable for running tests against. Keyword arguments are passed to the Homeserver constructor. If no datastore is supplied a datastore backed by an in-memory sqlite db will be given to the HS. """ + if reactor is None: + from twisted.internet import reactor + if config is None: config = Mock() config.signing_key = [MockKey()] @@ -56,34 +61,84 @@ def setup_test_homeserver(name="test", datastore=None, config=None, **kargs): config.worker_replication_url = "" config.worker_app = None config.email_enable_notifs = False + config.block_non_admin_invites = False + config.federation_domain_whitelist = None + config.federation_rc_reject_limit = 10 + config.federation_rc_sleep_limit = 10 + config.federation_rc_sleep_delay = 100 + config.federation_rc_concurrent = 10 + config.filter_timeline_limit = 5000 + config.user_directory_search_all_users = False + config.user_consent_server_notice_content = None + config.block_events_without_consent_error = None + config.media_storage_providers = [] + config.auto_join_rooms = [] + + # disable user directory updates, because they get done in the + # background, which upsets the test runner. + config.update_user_directory = False config.use_frozen_dicts = True - config.database_config = {"name": "sqlite3"} config.ldap_enabled = False if "clock" not in kargs: kargs["clock"] = MockClock() + if USE_POSTGRES_FOR_TESTS: + config.database_config = { + "name": "psycopg2", + "args": { + "database": "synapse_test", + "cp_min": 1, + "cp_max": 5, + }, + } + else: + config.database_config = { + "name": "sqlite3", + "args": { + "database": ":memory:", + "cp_min": 1, + "cp_max": 1, + }, + } + + db_engine = create_engine(config.database_config) + + # we need to configure the connection pool to run the on_new_connection + # function, so that we can test code that uses custom sqlite functions + # (like rank). + config.database_config["args"]["cp_openfun"] = db_engine.on_new_connection + if datastore is None: - db_pool = SQLiteMemoryDbPool() - yield db_pool.prepare() hs = HomeServer( - name, db_pool=db_pool, config=config, + name, config=config, + db_config=config.database_config, version_string="Synapse/tests", - database_engine=create_engine(config.database_config), - get_db_conn=db_pool.get_db_conn, + database_engine=db_engine, room_list_handler=object(), tls_server_context_factory=Mock(), + reactor=reactor, **kargs ) + db_conn = hs.get_db_conn() + # make sure that the database is empty + if isinstance(db_engine, PostgresEngine): + cur = db_conn.cursor() + cur.execute("SELECT tablename FROM pg_tables where schemaname='public'") + rows = cur.fetchall() + for r in rows: + cur.execute("DROP TABLE %s CASCADE" % r[0]) + yield prepare_database(db_conn, db_engine, config) hs.setup() else: hs = HomeServer( name, db_pool=None, datastore=datastore, config=config, version_string="Synapse/tests", - database_engine=create_engine(config.database_config), + database_engine=db_engine, room_list_handler=object(), tls_server_context_factory=Mock(), + reactor=reactor, **kargs ) @@ -172,7 +227,7 @@ class MockHttpResource(HttpServer): headers = {} if federation_auth: - headers["Authorization"] = ["X-Matrix origin=test,key=,sig="] + headers[b"Authorization"] = ["X-Matrix origin=test,key=,sig="] mock_request.requestHeaders.getRawHeaders = mock_getRawHeaders(headers) # return the right path if the event requires it @@ -183,7 +238,7 @@ class MockHttpResource(HttpServer): mock_request.args = urlparse.parse_qs(path.split('?')[1]) mock_request.path = path.split('?')[0] path = mock_request.path - except: + except Exception: pass for (method, pattern, func) in self.callbacks: @@ -194,7 +249,7 @@ class MockHttpResource(HttpServer): if matcher: try: args = [ - urllib.unquote(u).decode("UTF-8") + urlparse.unquote(u).decode("UTF-8") for u in matcher.groups() ] @@ -300,167 +355,6 @@ class MockClock(object): return d -class SQLiteMemoryDbPool(ConnectionPool, object): - def __init__(self): - super(SQLiteMemoryDbPool, self).__init__( - "sqlite3", ":memory:", - cp_min=1, - cp_max=1, - ) - - self.config = Mock() - self.config.database_config = {"name": "sqlite3"} - - def prepare(self): - engine = self.create_engine() - return self.runWithConnection( - lambda conn: prepare_database(conn, engine, self.config) - ) - - def get_db_conn(self): - conn = self.connect() - engine = self.create_engine() - prepare_database(conn, engine, self.config) - return conn - - def create_engine(self): - return create_engine(self.config.database_config) - - -class MemoryDataStore(object): - - Room = namedtuple( - "Room", - ["room_id", "is_public", "creator"] - ) - - def __init__(self): - self.tokens_to_users = {} - self.paths_to_content = {} - - self.members = {} - self.rooms = {} - - self.current_state = {} - self.events = [] - - class Snapshot(namedtuple("Snapshot", "room_id user_id membership_state")): - def fill_out_prev_events(self, event): - pass - - def snapshot_room(self, room_id, user_id, state_type=None, state_key=None): - return self.Snapshot( - room_id, user_id, self.get_room_member(user_id, room_id) - ) - - def register(self, user_id, token, password_hash): - if user_id in self.tokens_to_users.values(): - raise StoreError(400, "User in use.") - self.tokens_to_users[token] = user_id - - def get_user_by_access_token(self, token): - try: - return { - "name": self.tokens_to_users[token], - } - except: - raise StoreError(400, "User does not exist.") - - def get_room(self, room_id): - try: - return self.rooms[room_id] - except: - return None - - def store_room(self, room_id, room_creator_user_id, is_public): - if room_id in self.rooms: - raise StoreError(409, "Conflicting room!") - - room = MemoryDataStore.Room( - room_id=room_id, - is_public=is_public, - creator=room_creator_user_id - ) - self.rooms[room_id] = room - - def get_room_member(self, user_id, room_id): - return self.members.get(room_id, {}).get(user_id) - - def get_room_members(self, room_id, membership=None): - if membership: - return [ - v for k, v in self.members.get(room_id, {}).items() - if v.membership == membership - ] - else: - return self.members.get(room_id, {}).values() - - def get_rooms_for_user_where_membership_is(self, user_id, membership_list): - return [ - m[user_id] for m in self.members.values() - if user_id in m and m[user_id].membership in membership_list - ] - - def get_room_events_stream(self, user_id=None, from_key=None, to_key=None, - limit=0, with_feedback=False): - return ([], from_key) # TODO - - def get_joined_hosts_for_room(self, room_id): - return defer.succeed([]) - - def persist_event(self, event): - if event.type == EventTypes.Member: - room_id = event.room_id - user = event.state_key - self.members.setdefault(room_id, {})[user] = event - - if hasattr(event, "state_key"): - key = (event.room_id, event.type, event.state_key) - self.current_state[key] = event - - self.events.append(event) - - def get_current_state(self, room_id, event_type=None, state_key=""): - if event_type: - key = (room_id, event_type, state_key) - if self.current_state.get(key): - return [self.current_state.get(key)] - return None - else: - return [ - e for e in self.current_state - if e[0] == room_id - ] - - def set_presence_state(self, user_localpart, state): - return defer.succeed({"state": 0}) - - def get_presence_list(self, user_localpart, accepted): - return [] - - def get_room_events_max_id(self): - return "s0" # TODO (erikj) - - def get_send_event_level(self, room_id): - return defer.succeed(0) - - def get_power_level(self, room_id, user_id): - return defer.succeed(0) - - def get_add_state_level(self, room_id): - return defer.succeed(0) - - def get_room_join_rule(self, room_id): - # TODO (erikj): This should be configurable - return defer.succeed("invite") - - def get_ops_levels(self, room_id): - return defer.succeed((5, 5, 5)) - - def insert_client_ip(self, user, access_token, ip, user_agent): - return defer.succeed(None) - - def _format_call(args, kwargs): return ", ".join( ["%r" % (a) for a in args] + @@ -498,7 +392,7 @@ class DeferredMockCallable(object): for _, _, d in self.expectations: try: d.errback(failure) - except: + except Exception: pass raise failure diff --git a/tox.ini b/tox.ini index 39ad30536..ed26644bd 100644 --- a/tox.ini +++ b/tox.ini @@ -1,5 +1,5 @@ [tox] -envlist = packaging, py27, pep8 +envlist = packaging, py27, py36, pep8, check_isort [testenv] deps = @@ -14,12 +14,80 @@ deps = setenv = PYTHONDONTWRITEBYTECODE = no_byte_code - # As of twisted 16.4, trial tries to import the tests as a package, which - # means it needs to be on the pythonpath. - PYTHONPATH = {toxinidir} + commands = - /bin/sh -c "find {toxinidir} -name '*.pyc' -delete ; coverage run {env:COVERAGE_OPTS:} --source={toxinidir}/synapse \ - {envbindir}/trial {env:TRIAL_FLAGS:} {posargs:tests} {env:TOXSUFFIX:}" + /usr/bin/find "{toxinidir}" -name '*.pyc' -delete + coverage run {env:COVERAGE_OPTS:} --source="{toxinidir}/synapse" \ + "{envbindir}/trial" {env:TRIAL_FLAGS:} {posargs:tests} {env:TOXSUFFIX:} + {env:DUMP_COVERAGE_COMMAND:coverage report -m} + +[testenv:py27] + +# As of twisted 16.4, trial tries to import the tests as a package (previously +# it loaded the files explicitly), which means they need to be on the +# pythonpath. Our sdist doesn't include the 'tests' package, so normally it +# doesn't work within the tox virtualenv. +# +# As a workaround, we tell tox to do install with 'pip -e', which just +# creates a symlink to the project directory instead of unpacking the sdist. +# +# (An alternative to this would be to set PYTHONPATH to include the project +# directory. Note two problems with this: +# +# - if you set it via `setenv`, then it is also set during the 'install' +# phase, which inhibits unpacking the sdist, so the virtualenv isn't +# useful for anything else without setting PYTHONPATH similarly. +# +# - `synapse` is also loaded from PYTHONPATH so even if you only set +# PYTHONPATH for the test phase, we're still running the tests against +# the working copy rather than the contents of the sdist. So frankly +# you might as well use -e in the first place. +# +# ) +usedevelop=true + +[testenv:py36] +usedevelop=true +commands = + /usr/bin/find "{toxinidir}" -name '*.pyc' -delete + coverage run {env:COVERAGE_OPTS:} --source="{toxinidir}/synapse" \ + "{envbindir}/trial" {env:TRIAL_FLAGS:} {posargs:tests/config \ + tests/api/test_filtering.py \ + tests/api/test_ratelimiting.py \ + tests/appservice \ + tests/crypto \ + tests/events \ + tests/handlers/test_appservice.py \ + tests/handlers/test_auth.py \ + tests/handlers/test_device.py \ + tests/handlers/test_directory.py \ + tests/handlers/test_e2e_keys.py \ + tests/handlers/test_presence.py \ + tests/handlers/test_profile.py \ + tests/handlers/test_register.py \ + tests/replication/slave/storage/test_account_data.py \ + tests/replication/slave/storage/test_receipts.py \ + tests/storage/test_appservice.py \ + tests/storage/test_background_update.py \ + tests/storage/test_base.py \ + tests/storage/test__base.py \ + tests/storage/test_client_ips.py \ + tests/storage/test_devices.py \ + tests/storage/test_end_to_end_keys.py \ + tests/storage/test_event_push_actions.py \ + tests/storage/test_keys.py \ + tests/storage/test_presence.py \ + tests/storage/test_profile.py \ + tests/storage/test_registration.py \ + tests/storage/test_room.py \ + tests/storage/test_user_directory.py \ + tests/test_distributor.py \ + tests/test_dns.py \ + tests/test_preview.py \ + tests/test_test_utils.py \ + tests/test_types.py \ + tests/util} \ + {env:TOXSUFFIX:} {env:DUMP_COVERAGE_COMMAND:coverage report -m} [testenv:packaging] @@ -34,3 +102,15 @@ basepython = python2.7 deps = flake8 commands = /bin/sh -c "flake8 synapse tests {env:PEP8SUFFIX:}" + +[testenv:check_isort] +skip_install = True +deps = isort +commands = /bin/sh -c "isort -c -sp setup.cfg -rc synapse tests" + +[testenv:check-newsfragment] +skip_install = True +deps = towncrier>=18.6.0rc1 +commands = + python -m towncrier.check --compare-with=origin/develop +basepython = python3.6