Module utils default path (#20913)

* Make the module_utils path configurable
* Add a config value to define the path site module_utils files
* Handle module_utils that do not have source as an error
* Make an integration test for module_utils envvar working
* Add documentation for the ANSIBLE_MODULE_UTILS config option/envvar
* Add it to the sample ansible.cfg
* Add it to intro_configuration.
* Also modify intro_configuration to place envvars on equal footing with
  the config options (will need to document the envvar names in the
  future)
* Also add the ANSIBLE_LIBRARY use case from
  https://github.com/ansible/ansible/issues/15432 so we can close out
  that bug.

docs/docsite/rst/intro_configuration.rst

@@ -52,8 +52,9 @@ You may wish to consult the `ansible.cfg in source control <https://raw.github.c
 Environmental configuration
 ```````````````````````````
 
-Ansible also allows configuration of settings via environment variables.  If these environment variables are set, they will
-override any setting loaded from the configuration file.  These variables are for brevity not defined here, but look in `constants.py <https://github.com/ansible/ansible/blob/devel/lib/ansible/constants.py>`_ in the source tree if you want to use these.  They are mostly considered to be a legacy system as compared to the config file, but are equally valid.
+Ansible also allows configuration of settings via environment variables.  If
+these environment variables are set, they will override any setting loaded
+from the configuration file.  These variables are defined in `constants.py <https://github.com/ansible/ansible/blob/devel/lib/ansible/constants.py>`_.
 
 .. _config_values_by_section:
 
@@ -521,8 +522,25 @@ This is the default location Ansible looks to find modules::
 
      library = /usr/share/ansible
 
-Ansible knows how to look in multiple locations if you feed it a colon separated path, and it also will look for modules in the
-"./library" directory alongside a playbook.
+Ansible can look in multiple locations if you feed it a colon
+separated path, and it also will look for modules in the :file:`./library`
+directory alongside a playbook.
+
+This can be used to manage modules pulled from several different locations.
+For instance, a site wishing to checkout modules from several different git
+repositories might handle it like this:
+
+.. code-block:: shell-session
+
+    $ mkdir -p /srv/modules
+    $ cd /srv/modules
+    $ git checkout https://vendor_modules .
+    $ git checkout ssh://custom_modules .
+    $ export ANSIBLE_LIBRARY=/srv/modules/custom_modules:/srv/modules/vendor_modules
+    $ ansible [...]
+
+In case of modules with the same name, the library paths are searched in order
+and the first module found with that name is used.
 
 .. _local_tmp:
 
@@ -586,6 +604,31 @@ together.  The same holds true for --skip-tags.
     default value will be True.  After 2.4, the option is going away.
     Multiple --tags and multiple --skip-tags will always be merged together.
 
+.. _module_lang:
+
+module_lang
+===========
+
+This is to set the default language to communicate between the module and the system.
+By default, the value is value `LANG` on the controller or, if unset, `en_US.UTF-8` (it used to be `C` in previous versions)::
+
+    module_lang = en_US.UTF-8
+
+.. note::
+
+    This is only used if :ref:`module_set_locale` is set to True.
+
+.. _module_name:
+
+module_name
+===========
+
+This is the default module name (-m) value for /usr/bin/ansible.  The default is the 'command' module.
+Remember the command module doesn't support shell variables, pipes, or quotes, so you might wish to change
+it to 'shell'::
+
+    module_name = command
+
 .. _module_set_locale:
 
 module_set_locale
@@ -600,27 +643,23 @@ being set when the module is executed on the given remote system.  By default th
     The module_set_locale option was added in Ansible-2.1 and defaulted to
     True.  The default was changed to False in Ansible-2.2
 
-.. _module_lang:
+.. _module_utils:
 
+module_utils
+============
 
-module_lang
-===========
+This is the default location Ansible looks to find module_utils::
 
-This is to set the default language to communicate between the module and the system.
-By default, the value is value `LANG` on the controller or, if unset, `en_US.UTF-8` (it used to be `C` in previous versions)::
+    module_utils = /usr/share/ansible/my_module_utils
 
-    module_lang = en_US.UTF-8
+module_utils are python modules that Ansible is able to combine with Ansible
+modules when sending them to the remote machine.  Having custom module_utils
+is useful for extracting common code when developing a set of site-specific
+modules.
 
-.. _module_name:
-
-module_name
-===========
-
-This is the default module name (-m) value for /usr/bin/ansible.  The default is the 'command' module.
-Remember the command module doesn't support shell variables, pipes, or quotes, so you might wish to change
-it to 'shell'::
-
-    module_name = command
+Ansible can look in multiple locations if you feed it a colon
+separated path, and it also will look for modules in the
+:file:`./module_utils` directory alongside a playbook.
 
 .. _nocolor:
 

examples/ansible.cfg

@@ -13,6 +13,7 @@
 
 #inventory      = /etc/ansible/hosts
 #library        = /usr/share/my_modules/
+#module_utils   = /usr/share/my_module_utils/
 #remote_tmp     = ~/.ansible/tmp
 #local_tmp      = ~/.ansible/tmp
 #forks          = 5

lib/ansible/constants.py

@@ -196,7 +196,6 @@ MERGE_MULTIPLE_CLI_TAGS = get_config(p, DEFAULTS, 'merge_multiple_cli_tags', 'AN
 #### GENERALLY CONFIGURABLE THINGS ####
 DEFAULT_DEBUG             = get_config(p, DEFAULTS, 'debug',            'ANSIBLE_DEBUG',            False, value_type='boolean')
 DEFAULT_HOST_LIST         = get_config(p, DEFAULTS,'inventory', 'ANSIBLE_INVENTORY', DEPRECATED_HOST_LIST, value_type='path')
-DEFAULT_MODULE_PATH       = get_config(p, DEFAULTS, 'library',          'ANSIBLE_LIBRARY',          None, value_type='pathlist')
 DEFAULT_ROLES_PATH        = get_config(p, DEFAULTS, 'roles_path',       'ANSIBLE_ROLES_PATH',       '/etc/ansible/roles', value_type='pathlist', expand_relative_paths=True)
 DEFAULT_REMOTE_TMP        = get_config(p, DEFAULTS, 'remote_tmp',       'ANSIBLE_REMOTE_TEMP',      '~/.ansible/tmp')
 DEFAULT_LOCAL_TMP         = get_config(p, DEFAULTS, 'local_tmp',        'ANSIBLE_LOCAL_TEMP',      '~/.ansible/tmp', value_type='tmppath')
@@ -285,16 +284,20 @@ DEFAULT_BECOME_ASK_PASS   = get_config(p, 'privilege_escalation', 'become_ask_pa
 # (mapping of param: squash field)
 DEFAULT_SQUASH_ACTIONS         = get_config(p, DEFAULTS, 'squash_actions',     'ANSIBLE_SQUASH_ACTIONS', "apk, apt, dnf, homebrew, openbsd_pkg, pacman, pkgng, yum, zypper", value_type='list')
 # paths
+
 DEFAULT_ACTION_PLUGIN_PATH     = get_config(p, DEFAULTS, 'action_plugins',     'ANSIBLE_ACTION_PLUGINS', '~/.ansible/plugins/action:/usr/share/ansible/plugins/action', value_type='pathlist')
 DEFAULT_CACHE_PLUGIN_PATH      = get_config(p, DEFAULTS, 'cache_plugins',      'ANSIBLE_CACHE_PLUGINS', '~/.ansible/plugins/cache:/usr/share/ansible/plugins/cache', value_type='pathlist')
 DEFAULT_CALLBACK_PLUGIN_PATH   = get_config(p, DEFAULTS, 'callback_plugins',   'ANSIBLE_CALLBACK_PLUGINS', '~/.ansible/plugins/callback:/usr/share/ansible/plugins/callback', value_type='pathlist')
 DEFAULT_CONNECTION_PLUGIN_PATH = get_config(p, DEFAULTS, 'connection_plugins', 'ANSIBLE_CONNECTION_PLUGINS', '~/.ansible/plugins/connection:/usr/share/ansible/plugins/connection', value_type='pathlist')
 DEFAULT_LOOKUP_PLUGIN_PATH     = get_config(p, DEFAULTS, 'lookup_plugins',     'ANSIBLE_LOOKUP_PLUGINS', '~/.ansible/plugins/lookup:/usr/share/ansible/plugins/lookup', value_type='pathlist')
+DEFAULT_MODULE_PATH            = get_config(p, DEFAULTS, 'library',            'ANSIBLE_LIBRARY',        None, value_type='pathlist')
+DEFAULT_MODULE_UTILS_PATH      = get_config(p, DEFAULTS, 'module_utils',       'ANSIBLE_MODULE_UTILS',   None, value_type='pathlist')
 DEFAULT_INVENTORY_PLUGIN_PATH  = get_config(p, DEFAULTS, 'inventory_plugins',  'ANSIBLE_INVENTORY_PLUGINS', '~/.ansible/plugins/inventory:/usr/share/ansible/plugins/inventory', value_type='pathlist')
 DEFAULT_VARS_PLUGIN_PATH       = get_config(p, DEFAULTS, 'vars_plugins',       'ANSIBLE_VARS_PLUGINS', '~/.ansible/plugins/vars:/usr/share/ansible/plugins/vars', value_type='pathlist')
 DEFAULT_FILTER_PLUGIN_PATH     = get_config(p, DEFAULTS, 'filter_plugins',     'ANSIBLE_FILTER_PLUGINS', '~/.ansible/plugins/filter:/usr/share/ansible/plugins/filter', value_type='pathlist')
 DEFAULT_TEST_PLUGIN_PATH       = get_config(p, DEFAULTS, 'test_plugins',       'ANSIBLE_TEST_PLUGINS', '~/.ansible/plugins/test:/usr/share/ansible/plugins/test', value_type='pathlist')
 DEFAULT_STRATEGY_PLUGIN_PATH   = get_config(p, DEFAULTS, 'strategy_plugins',   'ANSIBLE_STRATEGY_PLUGINS', '~/.ansible/plugins/strategy:/usr/share/ansible/plugins/strategy', value_type='pathlist')
+
 DEFAULT_STRATEGY               = get_config(p, DEFAULTS, 'strategy',           'ANSIBLE_STRATEGY', 'linear')
 DEFAULT_STDOUT_CALLBACK        = get_config(p, DEFAULTS, 'stdout_callback',    'ANSIBLE_STDOUT_CALLBACK', 'default')
 # cache

lib/ansible/executor/module_common.py

@@ -498,7 +498,20 @@ def recursive_finder(name, data, py_module_names, py_module_cache, zf):
         if module_info is None:
             msg = ['Could not find imported module support code for %s.  Looked for' % name]
             if idx == 2:
-                msg.append('either %s or %s' % (py_module_name[-1], py_module_name[-2]))
+                msg.append('either %s.py or %s.py' % (py_module_name[-1], py_module_name[-2]))
+            else:
+                msg.append(py_module_name[-1])
+            raise AnsibleError(' '.join(msg))
+
+        # Found a byte compiled file rather than source.  We cannot send byte
+        # compiled over the wire as the python version might be different.
+        # imp.find_module seems to prefer to return source packages so we just
+        # error out if imp.find_module returns byte compiled files (This is
+        # fragile as it depends on undocumented imp.find_module behaviour)
+        if module_info[2][2] not in (imp.PY_SOURCE, imp.PKG_DIRECTORY):
+            msg = ['Could not find python source for imported module support code for %s.  Looked for' % name]
+            if idx == 2:
+                msg.append('either %s.py or %s.py' % (py_module_name[-1], py_module_name[-2]))
             else:
                 msg.append(py_module_name[-1])
             raise AnsibleError(' '.join(msg))
@@ -571,7 +584,6 @@ def _find_module_utils(module_name, b_module_data, module_path, module_args, tas
     Given the source of the module, convert it to a Jinja2 template to insert
     module code and return whether it's a new or old style module.
     """
-
     module_substyle = module_style = 'old'
 
     # module_style is something important to calling code (ActionBase).  It

lib/ansible/plugins/__init__.py

@@ -485,7 +485,7 @@ module_loader = PluginLoader(
 module_utils_loader = PluginLoader(
     '',
     'ansible.module_utils',
-    'module_utils',
+    C.DEFAULT_MODULE_UTILS_PATH,
     'module_utils',
 )
 

test/integration/targets/module_utils/library/test_env_override.py

@@ -0,0 +1,8 @@
+#!/usr/bin/python
+from ansible.module_utils.basic import AnsibleModule
+from ansible.module_utils.json_utils import data
+from ansible.module_utils.mork import data as mork_data
+
+results = {"json_utils": data, "mork": mork_data}
+
+AnsibleModule(argument_spec=dict()).exit_json(**results)

test/integration/targets/module_utils/module_utils/service.py

@@ -0,0 +1 @@
+sysv_is_enabled = 'sysv_is_enabled'

test/integration/targets/module_utils/module_utils_envvar.yml

@@ -0,0 +1,51 @@
+- hosts: localhost
+  gather_facts: no
+  tasks:
+  - name: Use a specially crafted module to see if things were imported correctly
+    test:
+    register: result
+
+  - name: Check that these are all loaded from playbook dir's module_utils
+    assert:
+      that:
+        - 'result["abcdefgh"] == "abcdefgh"'
+        - 'result["bar0"] == "bar0"'
+        - 'result["bar1"] == "bar1"'
+        - 'result["bar2"] == "bar2"'
+        - 'result["baz1"] == "baz1"'
+        - 'result["baz2"] == "baz2"'
+        - 'result["foo0"] == "foo0"'
+        - 'result["foo1"] == "foo1"'
+        - 'result["foo2"] == "foo2"'
+        - 'result["qux1"] == "qux1"'
+        - 'result["qux2"] == ["qux2:quux", "qux2:quuz"]'
+        - 'result["spam1"] == "spam1"'
+        - 'result["spam2"] == "spam2"'
+        - 'result["spam3"] == "spam3"'
+        - 'result["spam4"] == "spam4"'
+        - 'result["spam5"] == ["spam5:bacon", "spam5:eggs"]'
+        - 'result["spam6"] == ["spam6:bacon", "spam6:eggs"]'
+        - 'result["spam7"] == ["spam7:bacon", "spam7:eggs"]'
+        - 'result["spam8"] == ["spam8:bacon", "spam8:eggs"]'
+
+  # Test that overriding something in module_utils with something in the local library works
+  - name: Test that playbook dir's module_utils overrides facts.py
+    test_override:
+    register: result
+
+  - name: Make sure the we used the local facts.py, not the one shipped with ansible
+    assert:
+      that:
+        - 'result["data"] == "overridden facts.py"'
+
+  - name: Test that importing something from the module_utils in the env_vars works
+    test_env_override:
+    register: result
+
+  - name: Make sure we used the module_utils from the env_var for these
+    assert:
+      that:
+        # Override of shipped module_utils
+        - 'result["json_utils"] == "overridden json_utils"'
+        # Only i nthe env vars directory
+        - 'result["mork"] == "mork"'

test/integration/targets/module_utils/module_utils_test.yml

@@ -43,8 +43,9 @@
     ignore_errors: True
     register: result
 
+  - debug: var=result
   - name: Make sure we failed in AnsiBallZ
     assert:
       that:
         - 'result["failed"] == True'
-        - '"Could not find imported module support code for test_failure.  Looked for either foo or zebra" == result["msg"]'
+        - '"Could not find imported module support code for test_failure.  Looked for either foo.py or zebra.py" == result["msg"]'

test/integration/targets/module_utils/other_mu_dir/a/b/c/d/e/f/g/h/__init__.py

@@ -0,0 +1 @@
+data = 'should not be visible abcdefgh'

test/integration/targets/module_utils/other_mu_dir/facts.py

@@ -0,0 +1 @@
+data = 'should not be visible facts.py'

test/integration/targets/module_utils/other_mu_dir/json_utils.py

@@ -0,0 +1 @@
+data = 'overridden json_utils'

test/integration/targets/module_utils/other_mu_dir/mork.py

@@ -0,0 +1 @@
+data = 'mork'

test/integration/targets/module_utils/runme.sh

@@ -3,3 +3,4 @@
 set -eux
 
 ansible-playbook module_utils_test.yml -i ../../inventory -v "$@"
+ANSIBLE_MODULE_UTILS=$(pwd)/other_mu_dir ansible-playbook module_utils_envvar.yml -i ../../inventory -v "$@"