gh-129061: Fix FORCE_COLOR and NO_COLOR when empty strings

[hugovk]

First refactor the tests into subtests, which makes it easier compare the env var inputs and expected outputs.

• Issue: Empty values of FORCE_COLOR and NO_COLOR are not ignored #129061

[bedevere-bot]

🤖 New build scheduled with the buildbot fleet by @hugovk for commit ec36e22 🤖

Results will be shown at:

https://buildbot.python.org/all/#/grid?branch=refs%2Fpull%2F30617%2Fmerge

If you want to schedule another build, you need to add the 🔨 test-with-buildbots label again.

[hugovk]

Like #129137 (comment), test_external_inspection failed on Fedora and RHEL8 buildbots and is unrelated.

[serhiy-storchaka]

We only need to test pairs of environment variables that set conflicting values.
For example, NO_COLOR=1 and TERM=dumb should be tested with FORCE_COLOR=1 and PYTHON_COLOR=1. Tests for a single environment variable which forces color (FORCE_COLOR and PYTHON_COLOR) should be performed when the fallback value determined by isatty or _supports_virtual_terminal is False.

We need also to test can_colorize() when stderr is a TTY, but stdout is not, and with an explicit file argument. Including corner cases: no file.fileno, no file.isatty, file.isatty() and os.isatty(file.fileno()) return different results. BTW, I think that file.isatty() should perhaps be tested before os.isatty(file.fileno()).

This is a large issue and can be done in a separate PR.

[hugovk]

Some failures on Windows for:

                ({"PYTHON_COLORS": "true"}, True),
                ({"PYTHON_COLORS": "2"}, True),
                ({"PYTHON_COLORS": ""}, True),

That's because according to the docs:

If this variable is set to 1, the interpreter will colorize various kinds of output. Setting it to 0 deactivates this behavior.

Therefore everything that is not "0" or "1" is ignored, and the result depends on other conditions.

---

This also fails on Windows:

                ({"NO_COLOR": ""}, True),

That's because an empty value means we ignore it, and again the result depends on other conditions.

---

Shall we remove these four, or use them in combination with some other variables?

[serhiy-storchaka]

See improved tests in #129234. They cover all corner cases. I propose to push it first, and then add tests for empty values.

[serhiy-storchaka]

Only the first two checks are necessary. Others do not add anything useful. They look out of place and make the intention of the tests less clear.

The first group tests the different values of the variables. The second group tests relative priority of the variables (it does not repeat tests for different values). Then the same is done for PYTHON_COLORS -- first the different values of PYTHON_COLORS (including the empty value), then its relative priority to other variables.

[hugovk]

Okay, I've removed all but the first two.

[serhiy-storchaka]

LGTM! 👍

[miss-islington-app]

Thanks @hugovk for the PR 🌮🎉.. I'm working now to backport this PR to: 3.13.
🐍🍒⛏🤖

[bedevere-app]

GH-129360 is a backport of this pull request to the 3.13 branch.