unit test function with click decorators

[wyardley]

I realize the right fix might be to restructure the program, but is there any way to test a function with the click decorators on it without pytest's own command line flags getting passed in / throwing an error?

Simple self-contained example:

#!/usr/bin/env python3

import os
import sys
import click
import pytest

@click.command()
@click.option("--some-option", is_flag=True)
@click.option("--other-option", is_flag=True)
@click.option("--dry-run", is_flag=True)
def main(some_option: bool, other_option: bool, dry_run: bool):

    foobar = os.environ.get("FOO_BAR")
    if not foobar:
        sys.stderr.write("FOO_BAR not set\n")
        sys.exit(1)

# Normally not in the same place / file, but keeping it self-contained here
def test_main(capsys):
    with pytest.raises(SystemExit) as e_info:
        main()

    _, err = capsys.readouterr()
    assert err.startswith("FOO_BAR not set")
    assert e_info.value.code == 1

if __name__ == "__main__":
    main()

If this is run in a directory and the file is named test_test.py, running pytest will succeed, but pytest -v will fail, because the arguments to pytest are getting captured by click and passed into the function.

E       assert False
E        +  where False = <built-in method startswith of str object at 0x1042df4b0>('FOO_BAR not set')
E        +    where <built-in method startswith of str object at 0x1042df4b0> = "Usage: pytest [OPTIONS]\nTry 'pytest --help' for help.\n\nError: Got unexpected extra argument (testtest.py)\n".startswith

[davidism]

Yes, see the documentation on testing: https://click.palletsprojects.com/en/8.1.x/testing/

[wyardley]

Thanks - will try this. I’m familiar with this and have used it before; just didn’t think to in this context. Thanks!

[wyardley]

Something like this (for the above example code) is not working (even if I pass arguments explicitly to main() vs passing []):

    runner = CliRunner()
    with pytest.raises(SystemExit):
        result = runner.invoke(main([]))
    # assertion has to happen here to not be evergreen, but result is out of scope
    assert result.exit_code == 0

Even if I raise click.UsageError() instead:

def main(some_option: bool, other_option: bool, dry_run: bool):
    foobar = os.environ.get("FOO_BAR")
    if not foobar:
        raise click.UsageError("FOO_BAR not set")

def test_main(capsys):
    runner = CliRunner()
    result = runner.invoke(main(["--some-option", "--other-option", "--dry-run"]))
    assert result.exit_code == 2
    # This doesn't work either
    assert result.exception

I still get a test failure:

>               sys.exit(e.exit_code)
E               SystemExit: 2

/opt/homebrew/lib/python3.11/site-packages/click/core.py:1096: SystemExit
---------------------------------- Captured stderr call ----------------------------------
Usage: pytest [OPTIONS]
Try 'pytest --help' for help.

Error: FOO_BAR not set
================================ short test summary info =================================
FAILED testtest.py::test_main - SystemExit: 2

and it still looks like the options are getting grabbed by pytest?

Doing it this way is evergreen

    runner = CliRunner()
    with pytest.raises(SystemExit):
        result = runner.invoke(main(["--some-option", "--other-option", "--dry-run"]))
        # This will pass regardless of specified exit code
        assert result.exit_code == 9999

[davidism]

You don't call main, you pass it to runner.invoke.

[wyardley]

Got it. Yes, works - thanks so much! 🙏