These changes add support for provider-side previews of create and
update operations, which allows resource providers to supply output
property values for resources that are being created or updated during a
preview.
If a plugin supports provider-side preview, its create/update methods
will be invoked during previews with the `preview` property set to true.
It is the responsibility of the provider to fill in any output
properties that are known before returning. It is a best practice for
providers to only fill in property values that are guaranteed to be
identical if the preview were instead an update (i.e. only those output
properties whose values can be conclusively determined without
actually performing the create/update operation should be populated).
Providers that support previews must accept unknown values in their
create and update methods.
If a plugin does not support provider-side preview, the inputs to a
create or update operation will be propagated to the outputs as they are
today.
Fixes#4992.
Fixes: https://github.com/pulumi/docs/issues/4340
The deprecated message sits below the examples so it is not clear
to the user that the resource / datasource is deprecated
We're not going to generate language-specific API docs for the Azure NextGen provider, only resource docs. This change makes it so the resource docs do not emit any links to nonexistent API docs.
Non-string provider inputs must be projected as JSON formatted strings. The current codegen simply calls `json.dumps` for such properties, but this does not work for the new input types, which aren't JSON serializable.
To address this, make use of the new `pulumi.runtime.to_json` utility function, which is capable of serializing raw dicts and input types as JSON.
We currently emit array types as `List[T]` for Python, but `List[T]` is invariant, which causes type checkers like mypy to produce errors when values like `["foo", "bar"]` are passed as args typed as `List[pulumi.Input[str]]`. Instead, we should move to using `Sequence[T]` which is covariant, and does not have this problem.
We actually already do this for `Dict` vs. `Mapping`, emitting map types as `Mapping[str, T]` rather than `Dict[str, T]` because `Mapping[str, T]` is covariant. This change makes us consistent for array types.
An extra constructor overload was recently added to pass undefined state from
`get` for resources that do not have any state inputs (notably Kubernetes
resources). This ended up breaking PaC's `validateResourceOfType`, which relies
on type inference in common usage to determine the resource's args type based
on the signature of the constructor.
This constructor overload isn't necessary. Instead, we can remove it and modify
how the constructor is called inside `get`. This also makes it so we're not
exposing details about `get`'s implementation in the public API.
Changed the codegen in 6fd72dc0 but missed a condition that
is causing incorrect code in pulumi-kubernetes. This change
correctly generates inputs in both conditional branches.
If PyName(name) != PyNameLegacy(name), we panic, unless name is in an exclusion list.
Once all providers have been updated to this codegen (and we've added to `useLegacyName` and `excludeFromPanic` as appropriate), we can go back and remove the panic behavior along with the `excludeFromPanic` behavior.
If there's a config subpackage, we had explicitly added it to the map of modules because there *are* config properties, so we know it will be non-empty. This change ensures the config subpackage is imported in the top-level package's __init__.py and that the config subpackage itself has an __init__.py emitted.
Resource doc changes for Python:
- Types are included in constructor/function args
- The property names for input/output types are now always snake_case, regardless of the generated mapping tables, to match the new input/output classes
- Some other minor tweaks to function/constructor signatures (e.g. removed the `__props__` arg, as it's not meant to be used directly; use `@staticmethod` for static `get` methods).
* Use a package map where appropriate
* Fix imports
* handle NoneType for literal value expressions in go program gen
* add operator tests for c#, python, and node
Co-authored-by: evanboyle <evan@pulumi.com>
Rewrites that should produce nested applies due to functions that return
eventual types were instead producing a single top-level apply. These
changes fix that by considering a function that produces an eventual
value as inspecting eventual values.
Just what it says on the tin. Currently it's not possible to create a
valid value of this type because the `bytes` field is unexported. This
constructor fixes that.
The apply rewrite for relative traversals did not consider whether or
not the receiver was eventually-typed, and did not properly check
whether or not the relative traversal itself was eventually-typed. These
changes correct those mistakes.
In general, a package/module name in these targets is derived from the
module portion of a type token. If the type token is not already in an
expected form--namely, all lowercase--the generated package/module names
will also be in unexpected forms. These changes normalize the module
names to lowercase s.t. the generated package/module names conform to
expectations.
- NodeJS: fix get + constructor generation for types with required args
and no state type.
- Go: fix type + import generation for properties in the root module.
Add a new package, `codegen/importer`, that can generate definitions for
resource states in PCL or TS/Python/C#/Go. The pipeline is relatively
simple: given a list of resource states, generate a PCL program in
memory, bind it, and pass it to the language-specific code generator.
This builds upon the existing PCL IR, and can be used with the currently
supported code generators.
Related to #1635.
* Use the correct format for package name when module name is empty.
* Add an exclusion for Docker's Image component resource when generating the Python formal params.
* Prefix the type name with the package name when linking to Python function names.
Instead of requiring a plugin host for package loading in the HCL2
binder, define a much narrower interface that exposes the ability to
fetch the schema for a package at a specific version. This interface is
defined in the schema package, which also exposes a caching loader that
is backed by provider plugins.
These changes also add some convenience methods to `*schema.Package` for
fast access to particular resources and functions.
Related to #1635.
In general, each item in an HCL2 body must be followed by a trailing
newline. The printer did not properly insert these newlines for body
items without any associated tokens/trivia, or with trivia that did not
include a trailing new line.
Related to #1635.
The k8s SDK logs deprecations warnings as part of the
provider logic, and includes a flag to suppress these
warnings. Including them in the SDKs breaks this logic.
The PCL binder has supported resource options for some time, but these
options haven't been used or processed by the various code generators.
These options--particularly the parent and provider options0--are
critical for import codegen. These changes implement the basic set of
options, and add a note about fleshing out the rest as necessary.
One component of these changes is a new rewriter that rewrites property
references into property paths that are understood by the Pulumi engine.
This rewriter is used to preprocess the contents of the `ignoreChanges`
resource option.
These changes also hack around a weakness in the HCL2 type system:
In Go, references to resources should be typed as `hcl2.ResourceType`.
Unfortunately, this breaks the existing collection semantics associated
with resources. Because of this, the Go code generator does not have
enough information to know that it should generate a `[]pulumi.Resource`
for lists of resources. These changes hack around that limitation using
a Go-specific opaque type and some hardcoded comparisons in
`argumentTypeName`.
Fixes#4923.
The docs generator previously assumed that the opts parameter
for every resource was of the CustomResource type. This is
incorrect for the YAML and Helm overlays, which are
ComponentResources. This should be handled more generally
once our schema supports ComponentResources, but this fixes
the docs for now.
For the NodeJS k8s SDK, rather than falling back to default values
using the || operator, use the nullish coalescing operator (??).
This avoid situations where the primary value is set to false,
and then is overridden by the default value.
Use the schema package's Markdown parser and walk its AST to extract
examples.
These changes also rename StripNonRelevantExamples to FilterExamples.
This is preparatory work for #4159 and #4632.
In particular, use the parser to filter and extract examples. This also
sets up support for entity references in documentation that can be used
in order to render language-specific names for resources, functions,
types, and properties.
Related to #4632 and #4159.
Fixes: #4827
Diff:
```
if encrypted is not None:
- warnings.warn("Rename PostgreSQL role resource attribute "encrypted" to "encrypted_password"", DeprecationWarning)
+ warnings.warn("Rename PostgreSQL role resource attribute \"encrypted\" to \"encrypted_password\"", DeprecationWarning)
pulumi.log.warn("encrypted is deprecated: Rename PostgreSQL role resource attribute "encrypted" to "encrypted_password"")
```
If extra files are present, make sure that appropriate modules are
created as necessary and that extra sources are exported from each
module's `index.ts`.
If the type of a configuration variable is a token type (i.e. a type
provided by an overlay) with the underlying type string, call
`config.get` rather than `config.getObject`. This fixes a regression in
the `aws.config.region` variable.
- Split out common functionality into generateModuleContextMap method
- Add LanguageResource and LanguageProperty structs for use by downstream
codegen tools
- Add LanguageResources function for use by downstream codegen tools
Add support for union types that indicate a default type for targets
that do not support unions, or do not support unions in certain
positions (e.g. output properties). The NodeJS backend makes use of this
in combination with a new flag, `disableUnionOutputTypes`, to avoid
generating unions in output types.
These changes also refactor the various module ->
package/module/namespace mapping methods so that these entities can be
fetched by their language name rather than their token.
- Typecheck in all cases where a type may have changed
- Do not perform literal conversions if the type is already correct
- Perform literal conversions before checking to see if a call to
`__convert` is required. This catches cases such as string literals
passed where ints are required. Without this change, that form in
particular generates a bare number literal rather than a number
literal wrapped in a `__convert`.
If the source file for a resource or function definition would collide
with a reserved filename (e.g. `index.ts`), append an underscore to its
filename (e.g. `index_.ts`).
These changes are necessary in order to move tfgen over to the schema
code generator.
The schema generator for tfgen has an annoying behavior for nested types
in which it does not separate input and output types. Worse, the shape
of the type that results from a collision of input and output types is
order-dependent: whichever of the two was observed last wins (note that
the shape is still determenistic, as the order in which types are
recorded is always the same). As a result, the NodeJS code generator
needs to know the set of required properties for the other aspect of the
type in order to generate proper code.
Contributes to
https://github.com/pulumi/pulumi-terraform-bridge/issues/179.
Types like output(T), promise(T), and union(T_0, ..., T_N) should be
assignable from dynamic if they contain an element type that is
assignable from dynamic.
This allows consumers to use the following code to check if some type
behaves like the dynamic type w.r.t. conversions:
```
if t.AssignableFrom(model.DynamicType) {
}
```
Fixes#4703.
Types like output(T), promise(T), and union(T_0, ..., T_N) should be
assignable from dynamic if they contain an element type that is
assignable from dynamic.
This allows consumers to use the following code to check if some type
behaves like the dynamic type w.r.t. conversions:
```
if t.AssignableFrom(model.DynamicType) {
}
```
Fixes#4703.
- Move the implementation of loadPackageSchema into a method on
PackageCache
- Protect the cache with synchronization primitives to enable
concurrency in downstream consumers
- Use jsoniter to deserialize schemas
* Remove code that was hiding Go as a language option for k8s overlay resources. Mark input args type for Functions as optional in Go.
* Show a special note in the Go function's snippet when the function name does not match other languages.
* Make sure the draft PR stays assigned to the original author. Add some more comments.
* Remove logic to use the Go ModuleToPackage map to then lookup the relevant C# namespace. Make getLanguageModuleName a method of modContext since passing in a package is not needed anymore.
Some type tokens contain characters that are not valid in a single
URL path element (`/` in particular). These type tokens may be
path-escaped, and should be unescaped before they are interpreted.
- Add support for constant values in unions (e.g. `"foo" | "bar"`)
- Generate `Input<T | U>` instead of `Input<T> | Input<U>`
- Emit deprecation messages for resources
- Handle nil state inputs
- Allow packages to provide a README
- Remove sync invoke support
These changes are part of
https://github.com/pulumi/pulumi-terraform-bridge/issues/179, and
restore functional parity with the current `tfgen` generator.
* Generate language package details in the index pages.
* Add a new DocLanguageHelper interface method to get to a module per-language.
* Add a new workflow file for automatically creating draft docs PRs for previewing resource docs for AWS and Kubernetes as a result of changes in the resource docs generator.
Add a rewriter that reifies implicit conversions into a call to the
`__convert` intrinsic. Code generators can recognize this intrinsic and
use it to generate appropriate conversion code.
Part of this work involves redesigning the type annotations system.
Annotations are now only applicable to opaque and object types. Instead
of inspecting annotations directly, code generators should use
`hcl2.GetSchemaForType` to extract the `schema.Type` for a `model.Type`.
If we are generating code into an async context (e.g. an async main),
await calls to invoke rather than leaving them as promises. This results
in more idiomatic code withing such contexts.
* Add a new template for examples section. Extract the examples section into a structured format for custom template processing.
* Update the description IFF we were able to extract examples sections from it.
* Update doc comments and add missing file header for the newly added file.
* Make the example description readable. Add a check for empty example sections.
* Add a chooser right below the Example Usage header. Remove javascript as a language.
* Allow an empty new-line between short-codes boundaries.
There are two expressions in HCL2 that are used to iterate over
collections:
- Splat expressions, e.g. `foo.*.bar`, and
- For expressions, e.g. `[for v in foo: v.bar]`
In both of these cases, the parts of the expression that are not the
collection being iterated behave like callbacks, and must be treated as
such by the apply rewriter.
Token detection was broken for conditional and for expressions that
represent template control sequences. The code originally attempted to
determine whether or not a conditional or for expression was a control
sequence by inspecting the expression's parent. Unfortunately, that
approach is unable to distinguish between expressions that are control
sequences and those that are merely template parts. These changes
instead inspect the first token of the expression for a template control
token (i.e. `%{`): if such a token is found, the expression is detected
as a template control sequence.
* Adjust the args type names for k8s overlay resources.
* Revert to using input type link for kubernetes non-overlay resources.
* Use the correct module name to lookup the C# namespace for resources that belong to a module.
* Helm's arg types don't use the version number.
Specifically, handle index and relative traversal expressions, and clean
up the code a little bit.
This should also help us pick up more `pulumi.interpolate` calls in TS.
With the addition of Python examples, doc comments may now contain
triple quotes. These must be escaped in order to avoid malformed source
files.
Fixes#4568.
- Determine variable types for ranged resources by typechecking an
equivalent expression
- Detect top-level await in NodeJS and generate an async main
- Fix `pulumi.all` generation for NodeJS
- Fix a bug in the lowering of relative traversals in Python
* Restore the API type links for C#.
* Also restore them in function.tmpl.
* Add package details to the Functions template as well. Add a global template function to detect if the APIDocLinks has links for a language. Don't generate C# API doc links for k8s.
This code was accidentally removed in a prior PR.
These changes also remove an assert; the situation the assertion guards
against is valid and occurs when dealing with traversals inside of
splat expressions.
Unlike most languages with interpolated strings, Python's formatted
string literals do not allow the nesting of quotes. For example,
this expression is not legal Python:
f"Foo {"bar"} baz"
If an interpolation requires quotes, those quotes nust differ from the
quotes used by the enclosing literal. We can fix the previous example
by rewriting it with single quotes:
f"Foo {'bar'} baz"
However, this presents a problem if there are more than two levels of
nesting, as Python only has two kinds of quotes (four if the outermost
string uses """ or '''): in this case, the expression becomes
unspellable, and must be assigned to a local that is then used in place
of the original expression. So this:
f"Foo {bar[f'index {baz["qux"]}']} zed"
becomes this:
index = "qux"
f"Foo {bar[f'index {baz[index]}']}"
To put it bluntly, Python code generation reqiures register allocation,
but for quotes. These changes implement exactly that.
These changes also include a fix for traversals that access values that
are dictionaries rather than objects, and must use indexers rather than
attributes.
Mapping happens at package scope, not type scope. This is what causes
some of the surprising mixes of camel and snake casing in our packages.
In order to accommodate this, build the case mapping tables up front,
and refer to them when performing case mapping.
The tokens that make up the "key" portion of an index traversal
(e.g. `"foo"` in `a["foo"]`) are structured like those that make up a
block label: an open quote token, a string literal token, and a close
quote token. The token mapper did not account for that fact, and instead
recorded the key token as the open quote. These changes correct that
error, and adjust the code in `literalText` to allow for
properly-escaped and quoted strings where necessary.
If a single process is going to bind and generate multiple programs, it
is useful to be able to cache package schemas in order to avoid the
(large) overhead of deserializing schemas multiple times.
Use strings.Builder instead of raw slices. We could go further here and
collapse to a single, ~right-sized bulder at the cost of complexity in
the `stateAcronym -> stateLowerOrNumber` transition.
Also, eliminate the pyName function, as it is redundant.
* Add a new DocLangHelper interface method to generate function names per language. Use the functionNames override map in Go to correctly generate a function name.
* Add kong to the title lookup map.
* Check if item was found in map.
* Use title case for the nodejs Function args type name. Use title case for the function name in the index page.
* Also fix the result name for a Function in nodejs to use title case.
* Use the module name to lookup the package.
* Use the title-case name for the Function name in the index page. Fix the language value passed to the lang chooser in function.tmpl.
* Use title case for nested type titles.
* Add title lookup for GitHub.
Some of the apply rewriter's assumptions were broken by the richer
expressions available in HCL2. These changes fix those broken
assumptions, in particular the assumption that only scope traversal
expressions are sources of eventual values.
In particular, ensure that they are keyword-safe. This affects
`aws:lambda:*` in particular: prior to these changes, we were generating
code into `pulumi_aws.lamdba`, which is not referencable due to its use
of the `lamdba` keyword. With these changes, we generate code into
`pulumi_aws.lambda_`.
There is also a small fix for multi-line non-formatted strings: these
strings do not need escaped braces.
* Fix output property names in Python codegen.
Some property names are mapped from their `camelCase` Pulumi name to a
`snake_case` Python name. This mapping is irregular, and only occurs for
resources properties and function calls.
Note that there's still more work to do here: this only fixes names on
the output side; the input side is still broken for nested resource
proprerties and function calls.
The underlying design--annotated types in `hcl2/model`--may need some
additional work in the future, but I _believe_ it's good enough for now.
* Fix Python input property names.
- Fix input property names and forms for invokes. Previously we
generated a dict; now we generate properly-named args.
- Fix nested property names for resources.
* PR feedback
- Fix input property names and forms for invokes. Previously we
generated a dict; now we generate properly-named args.
- Fix nested property names for resources.
* [codegen/go] Fix accessors on struct ptr outputs
The accesor methods on nestred struct Ptr outputs were previously not accepting pointer typed inputs as they should, and would thus always panic if used.
The "simple" fix would be to just accept the pointer type and blindly dereference it. But this doesn't seem like the right experience - it would make these accessors very unsafe to use in practice.
Instead, this PR implements the accessors on pointer-typed outputs as nil-coaslescing, always lifting the output type into a pointer type and flowing a nil value into the result type. This ensures the accessor will not nil-deref, and that user code can handle the `nil` value itself (or use `.Apply` directly to implement more specialized behaviour).
Before:
```go
// Name of your S3 bucket.
func (o BuildStorageLocationPtrOutput) Bucket() pulumi.StringOutput {
return o.ApplyT(func(v BuildStorageLocation) string { return v.Bucket }).(pulumi.StringOutput)
}
```
After:
```go
// Name of your S3 bucket.
func (o BuildStorageLocationPtrOutput) Bucket() pulumi.StringPtrOutput {
return o.ApplyT(func(v *BuildStorageLocation) *string {
if v == nil {
return nil
}
return &v.Bucket
}).(pulumi.StringPtrOutput)
}
```
However, due to the decision to have this more usable behaviour, this is a breaking change, as some/many accessors now return a pointer type when they previously did not.
Fixespulumi/pulumi-azure#530.
* Mark nested property types as requiring ptr types
* Add CHANGELOG
* More fixes
Some property names are mapped from their `camelCase` Pulumi name to a
`snake_case` Python name. This mapping is irregular, and only occurs for
resources properties and function calls.
Note that there's still more work to do here: this only fixes names on
the output side; the input side is still broken for nested resource
proprerties and function calls.
The underlying design--annotated types in `hcl2/model`--may need some
additional work in the future, but I _believe_ it's good enough for now.
* Optimize titles and descriptions
* Move regexp to file level variable
* wip
* Add title tags and description to module and pkg index pages
* Remove block_external_search_index from index.tmpl
* Improve resource descriptions
* Add more providers to look up map
* Improve module level descriptions
Clean up
cleanup
* Update the function.tmpl with the auto-generated disclaimer. Pass title tag and meta description for the function template as well.
* Fix whitespace in the package_details template..
* Account for empty module names for package-level resources. Move logic for generating title tag and description into separate methods for Resources and Functions.
* Add a test for the generated resource and function title tag.
Co-authored-by: Praneet Loke <1466314+praneetloke@users.noreply.github.com>