A py.test plugin providing fixtures to simplify inmanta modules testing.
Project description
pytest-inmanta
A pytest plugin to test inmanta modules
Installation
pip install pytest-inmanta
If you want to use pytest-inmanta
to test a v2 module, make sure to install the module:
inmanta module install -e .
Usage
This plugin provides a test fixture that can compile, export and deploy code without running an actual inmanta server.
def test_compile(project):
"""
Test compiling a simple model that uses std
"""
project.compile("""
host = std::Host(name="server", os=std::linux)
file = std::ConfigFile(host=host, path="/tmp/test", content="1234")
""")
The fixture also provides access to the model internals
assert len(project.get_instances("std::Host")) == 1
assert project.get_instances("std::Host")[0].name == "server"
To the exported resources
f = project.get_resource("std::ConfigFile")
assert f.permissions == 644
To compiler output and mock filesystem
def test_template(project):
"""
Test the evaluation of a template
"""
project.add_mock_file("templates", "test.tmpl", "{{ value }}")
project.compile("""import unittest
value = "1234"
std::print(std::template("unittest/test.tmpl"))
""")
assert project.get_stdout() == "1234\n"
And allows deployment of specific resources
# perform deploy
result = project.deploy_resource_v2("std::ConfigFile", expected_status=inmanta.const.ResourceState.deployed)
# assert the deploy performed no changes
result.assert_no_changes()
# assert the deploy produced specific log lines
result.assert_has_logline("Calling read_resource")
And dryrun
result = project.dryrun_resource_v2("testmodule::Resource")
assert result.changes == {"value": {'current': 'read', 'desired': 'write'}}
# Or dryrun all resources at once
result = project.dryrun_all()
It is also possible to deploy all resources at once:
results = project.deploy_all()
assert results.get_context_for("std::ConfigFile", path="/tmp/test").status == ResourceState.deployed
The dryrun_all
and deploy_all
functions return a Result
object with
some helpful auxiliary functions to assert some sanity checks.
We can check if every resource on the result has the correct state:
results = project.dryrun_all()
results.assert_all(ResourceState.dry)
It is possible to determine if every resource has the attribute purged
in its changes.
This is helpful to assert if the resources are to be created (purged
set to True) or deleted (purged
set to False):
results = project.dryrun_all()
results.assert_resources_have_purged()
The same applies to a deploy_all
:
results = project.deploy_all()
results.assert_all(ResourceState.deployed)
To check if a deploy is successful and we achieved the desired state, it is possible to do a dryrun after the deploy and check if there are no changes:
results = project.deploy_all()
results.assert_all(ResourceState.deployed)
results = project.dryrun_all()
results.assert_has_no_changes()
For convenience, it is also possible to dryrun and deploy all resources at once.
This method also asserts that the dryruns and deploys pass the sanity checks above.
It returns a DeployResultCollection
that aggregates the Results
from the dryruns and the deploy.
resutls = project.dryrun_and_deploy_all(assert_create_or_delete=True)
results.first_dryrun.assert_all(ResourceState.dry)
results.deploy.assert_all(ResourceState.deployed)
results.last_dryrun.assert_all(ResourceState.dry)
Testing functions and classes defined in a v1 module is also possible
using the inmanta_plugins
fixture. The fixture exposes inmanta modules as its attributes
and imports them dynamically when accessed. For v2 modules, the recommended approach is to
just use top-level imports instead of using the fixture.
def test_example(inmanta_plugins):
inmanta_plugins.testmodule.regular_function("example")
This dynamism is required because the compiler resets module imports when project.compile
is called. As a result, if you store a module in a local variable, it will not survive a
compilation. Therefore you are advised to access modules in the inmanta_plugins
package
in a fully qualified manner (using the fixture). The following example demonstrates this.
def test_module_inequality(project, inmanta_plugins):
cached_module = inmanta_plugins.testmodule
assert cached_module is inmanta_plugins.testmodule
project.compile("import testmodule")
assert cached_module is not inmanta_plugins.testmodule
While you could import from the inmanta_plugins
package directly, the fixture makes abstraction
of module reloading. Without the fixture you would be required to reimport after project.compile
.
Testing plugins
Take the following plugin as an example:
# <module-name>/plugins/__init__.py
from inmanta.plugins import plugin
@plugin
def hostname(fqdn: "string") -> "string":
"""
Return the hostname part of the fqdn
"""
return fqdn.split(".")[0]
A test case, to test this plugin looks like this:
# <module-name>/tests/test_hostname.py
def test_hostname(project):
host = "test"
fqdn = f"{host}.something.com"
assert project.get_plugin_function("hostname")(fqdn) == host
- Line 3: Creates a pytest test case, which requires the
project
fixture. - Line 6: Calls the function
project.get_plugin_function(plugin_name: str): FunctionType
, which returns the plugin function namedplugin_name
. As such, this line tests whetherhost
is returned when the plugin functionhostname
is called with the parameterfqdn
.
Advanced usage
Because pytest-inmanta keeps inmanta_plugins
submodule objects alive to support top-level imports, any stateful modules
(modules that keep state on global Python variables in the module's namespace) must define cleanup logic to reset state between
compiles. Pytest-inmanta expects such cleanup functions to be synchronous functions that live in the top-level scope (defined
on the module object, not in a class) of a inmanta_plugins
submodule (of any depth). Their name should start with
"inmanta_reset_state" and they should not take any parameters. For example:
# <module-name>/plugins/state.py
MY_STATE = set()
def inmanta_reset_state() -> None:
global MY_STATE
MY_STATE = set()
Multiple cleanup functions may be defined, in which case no guaranteed call order is defined.
Options
The following options are available.
--venv
: folder in which to place the virtual env for tests (will be shared by all tests), overridesINMANTA_TEST_ENV
. This options depends on symlink support. This does not work on all windows versions. On windows 10 you need to run pytest in an admin shell. Using a fixed virtual environment can speed up running the tests.--pip-index-url
: pip index to install dependencies from. Can be specified multiple times to add multiple indexes. When set, it will overwrite the system index-url even ifpip-use-system-config
is set. (overridesPIP_INDEX_URL
, defaults to[]
)--pip-pre
,--no-pip-pre
Allow installation of pre-release package by pip or not? (overridesPIP_PRE
, defaults to--install-mode != release
)--pip-use-system-config
Allow pytest-inmanta to use the system pip config or not? (overridesINMANTA_PIP_USE_SYSTEM_CONFIG
, defaults toFalse
)--use-module-in-place
: makes inmanta add the parent directory of your module directory to it's directory path, instead of copying your module to a temporary libs directory. It allows testing the current module against specific versions of dependent modules. Using this option can speed up the tests, because the module dependencies are not downloaded multiple times.--module-repo
: location to download v1 modules from, overridesINMANTA_MODULE_REPO
. The default value is the inmanta github organisation. Multiple repos can be passed by space-separating them or by passing the parameter multiple times.--install-mode
: install mode to use for v1 modules downloaded during this test, overridesINMANTA_INSTALL_MODE
.--no-load-plugins
: Don't load plugins in the Project class. OverridesINMANTA_NO_LOAD_PLUGINS
. When not using this option during the testing of plugins with theproject.get_plugin_function
method, it's possible that the module'splugin/__init__.py
is loaded multiple times, which can cause issues when it has side effects, as they are executed multiple times as well.--no-strict-deps-check
: option to run pytest-inmanta using the legacy check(less strict) on requirements. By default the new strict will be used.
Use the generic pytest options --log-cli-level
to show Inmanta logger to see any setup or cleanup warnings. For example,
--log-cli-level=INFO
Compatibility with pytest-cov
The --use-module-in-place
option should be set when pytest-inmanta is used in combination with the pytest-cov
pytest plugin. Without the --use-module-in-place
option, the reported test coverage will be incorrect.
Using the pytest option framework
The pytest-inmanta
extension contains a framework to help create pytest options to use in your test suite or test extension. Options/parameters created with the framework will automatically be registered and picked up by pytest.
Each option can be set via cli argument or via environment variable. If both are set, the cli argument value takes precedence over the environment variable.
When creating a new option, pay attention to place it in a place that will always be loaded by pytest, e.g. the conftest.py
file.
The different type of test parameters that can be used are shown here: pytest_inmanta/test_parameters
. The currently supported types are:
BooleanTestParameter
EnumTestParameter
FloatTestParameter
IntegerTestParameter
ListTestParameter
PathTestParameter
StringTestParameter
You can of course add and use your own option type, as long as it extends the base class TestParameter
properly.
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.