Move flake8 settings into pyproject.toml

[cclauss]

Move flake8 linter settings from the file .flake8 into pyproject.toml by using Flake8-pyproject.

% uv run --with=Flake8-pyproject flake8 isort/ tests/

Fixes: #2356

• Remove .flake8 config #2356

Related to:

• Lint Python code with ruff #2359

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 99.16%. Comparing base (7190d1d) to head (7de0a2d).
Report is 3 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #2360   +/-   ##
=======================================
  Coverage   99.16%   99.16%           
=======================================
  Files          40       40           
  Lines        3098     3098           
  Branches      680      680           
=======================================
  Hits         3072     3072           
  Misses         15       15           
  Partials       11       11           

[DanielNoord]

I think @staticdev pointed in #2356 (comment) towards just replacing flake8 so that we would also no longer need this.

[cclauss]

Some code imports flake8 code…

• https://github.com/search?q=repo%3APyCQA%2Fisort%20flake8.options

[DanielNoord]

That's just a string in a test case. The test itself might be dependent on flake8 being in the environment but we can 1) fix that and 2) even if we didn't we wouldn't need to config for flake8 anymore.

[cclauss]

• Remove .flake8 config #2356 is not about removing flake8 so this pull request is merely about moving flake8 config to align with PEP 621.

Several flake8 plugins require stable interactions between flake8 and isort.
https://github.com/orgs/wemake-services/repositories?q=flake8 has three flake8 plugins that rely on isort.profile = wecare and there are other flake8 plugins that depend on sorting imports like:

• https://pypi.org/project/flake8-isort
• https://pypi.org/project/flake8-imports

I am unsure what I am breaking by removing flake8 so none of my pull requests have attempted that but if someone else wants to that would be great.

[staticdev]

@cclauss just curious why you added max-complexity?
This one was not in original file.

[cclauss]

https://docs.astral.sh/ruff/rules/complex-structure

This function is 9 times more complex than the suggested code complexity setting. This is also what DeepSource is complaining about but they make a different calculation than flake8 and ruff.

It is useful to set upper limits on code complexity so that if someone makes this code even more complex (i.e. less understandable and more difficult to maintain) then the changing of this setting should set off a discussion in the code review about taking time to refactor or at least add more tests.

% uvx flake8 --select=C9 --max-complexity=90 **/*.py

isort/core.py:32:1: C901 'process' is too complex (91)

This function is more than 450 lines long.

[staticdev]

Yes, not the best software engineering example of logic. But I don't see value adding a super high complexity to it.

I would prefer to set a normal recommended value and ignore this method with a TODO to reduce the complexity and remove the ignore.

[cclauss]

Suggested change

max-complexity = 91

Let’s remove it here and I will do as you say in a separate code complexity PR using ruff after that lands.

• Document areas of code complexity #2370

[staticdev]

I think @staticdev pointed in #2356 (comment) towards just replacing flake8 so that we would also no longer need this.

That would be indeed another way, but since @cclauss already put the effort. I think it is valid and better to maintain the settings in a single file.

[staticdev]

Thanks @cclauss LGTM.

[cclauss]

Failing GitHub Actions tests fixed in:

• GitHub Actions: Use uv cache, not pip cache #2363 or
• Upgrade to uv>=0.6.0 and enable-cache #2364