pytest-console-scripts 1.1.0

Pytest plugin for testing console scripts

pytest-console-scripts

Pytest-console-scripts is a pytest plugin for running python scripts from within tests. It's quite similar to subprocess.run(), but it also has an in-process mode, where the scripts are executed by the interpreter that's running pytest (using some amount of sandboxing).

In-process mode significantly reduces the run time of the test suites that run many external scripts. This is speeds up development. In the CI environment subprocess mode can be used to make sure the scripts also work (and behave the same) when run by a fresh interpreter.

Requirements

• Python 3.5+, or PyPy3,
• Pytest 4.0 or newer.

Installation

You can install "pytest-console-scripts" via pip from PyPI:

$ pip install pytest-console-scripts

Normally you would add it as a test dependency in tox.ini (see tox documentation).

Usage

This plugin will run scripts that are installed via console_scripts entry point in setup.py, python files in current directory (or anywhere else, if given the path), and Python scripts anywhere else in the path. It will also run executables that are not Python scripts, but only in subprocess mode (there's no benefit in using pytest-console-scripts for this, you should just use subprocess.run).

Here's an example with console_scripts entry point. Imagine we have a python package foo with the following setup.py:

setup(
    name='foo',
    version='0.0.1',
    py_modules=['foo'],
    entry_points={
        'console_scripts': ['foobar=foo:bar']
    },
)

We could use pytest-console-scripts to test the foobar script:

def test_foo_bar(script_runner):
    ret = script_runner.run('foobar', '--version')
    assert ret.success
    assert ret.stdout == '3.2.1\n'
    assert ret.stderr == ''

This would use the script_runner fixture provided by the plugin to run the script and capture its output.

The arguments of script_runner.run are the command name of the script and any command line arguments that should be passed to it. Additionally the following keyword arguments can be used:

• cwd - set the working directory of the script under test.
• env - a dictionary with environment variables to use instead of the current environment.
• stdin - a file-like object that will be piped to standard input of the script.

Configuring script execution mode

In the example above the foobar script would run in in-process mode (which is the default). This is fast and good for quick iteration during development. After we're happy with the functionality, it's time to run the script in subprocess mode to simulate real invocation more closely. There are several ways to do this. We can configure it via pytest configuration (for example in tox.ini):

 [pytest]
 script_launch_mode = subprocess

We can give a command line option to pytest (this will override the configuration file):

$ pytest --script-launch-mode=subprocess test_foobar.py

We can also mark individual tests to run in a specific mode:

@pytest.mark.script_launch_mode('subprocess')
def test_foobar(script_runner):
    ...

Between these three methods the marking of the tests has priority before the command line option that in turn overrides the configuration setting. All three can take three possible values: "inprocess", "subprocess", and "both" (which will cause the test to be run twice: in in-process and in subprocess modes).

Suppressing the printing of script run results

When tests involving pytest-console-scripts fail, it tends to be quite useful to see the output of the scripts that were executed in them. We try to be helpful and print it out just before returning the result from script_runner.run(). Normally PyTest captures all the output during a test run and it's not shown to the user unless some tests fail. This is exactly what we want.

However, in some cases it might be useful to disable the output capturing and PyTest provides ways to do it. When capturing is disabled, all test run results will be printed out and this might make it harder to inspect the other output of the tests. To deal with this, pytest-console-scripts has an option to disable the printing of script run results:

$ pytest --hide-run-results test_foobar.py

It's also possible to disable it just for one script run:

ret = script_runner.run('foobar', print_result=False)

When printing of script run results is disabled, script output won't be visisble even when the test fails. Unfortunately there's no easy way to print it only if the test fails because by the time a script run completes we don't yet know whether the test will fail or not. In any case, with this option and capturing control we can configure what output gets displayed so it should be possible to get to the bottom of things.

Package installation and testing during development

Since pytest-console-scripts relies on the scripts being located in the path, it can only run the console scripts from packages that have been installed (if you are interested in working on removing this limitation, take a look at this ticket and in particular this comment). If you want to run the tests quickly during development, the additional installation step would add a significant overhead and slow you down.

There's a way around this: install your package in development mode using python setup.py develop. If you use tox, you can take one of its existing virtualenvs (they live in .tox/). Otherwise create a virtualenv just for development, activate it and run python setup.py develop to install your package in development mode. You will need to re-install every time you add a new console script, but otherwise all the changes to your code will be immediately picked up by the tests.

Contributing

Contributions are very welcome. Tests can be run with tox, please ensure the coverage at least stays the same before you submit a pull request.

License

Distributed under the terms of the MIT license, "pytest-console-scripts" is free and open source software.

Issues

If you encounter any problems, please file an issue along with a detailed description.

---

Pytest-console-scripts was initially generated with Cookiecutter along with @hackebrot's Cookiecutter-pytest-plugin template.