Replace black, flake8, pyupgrade with ruff for Python linting and formatting

[flodolo]

No description provided.

[flodolo]

Thoughts on adding ruff on top of black? Does it need to be called out in contributing.rst? Should we drop black alltogether?

[flodolo]

Looking at AMO, they completely dropped black for ruff, and use some configuration options
https://github.com/eviljeff/olympia/blob/9c570b5180b02a6e5d9e0d1d1dc550d76b3c249e/pyproject.toml

I guess it should be replacing both flake8 and black at this point 🤔
https://docs.astral.sh/ruff/

[mathjazz]

Seems like ruff would let us drop black, flake8 and even pyupgrade.

All that while being faster.

Sounds like a win-win to me.

[flodolo]

Oh, I missed pyupgrade.

Let me play with it a bit more, especially with the configuration part, and see how far I can go.

[flodolo]

There are quite a few changes running ruff format to replace black:

• Space after the class definition: formatter removes blank line after class definition astral-sh/ruff#8566
• Some are due to the fact that we were using black 22. The latest version already reformatted 17 files (e.g. adding an empty line after the docstring).

Not completely sure what we want to add/enable in the TOML file.

EDIT: I suppose that we could exclude */migrations/* from listing/formatting, but is there a reason for doing it?

[mathjazz]

Space after the class definition: formatter removes blank line after class definition astral-sh/ruff#8566

Disable that?

https://docs.astral.sh/ruff/rules/one-blank-line-after-class/

[flodolo]

Space after the class definition: formatter removes blank line after class definition astral-sh/ruff#8566

Disable that?

https://docs.astral.sh/ruff/rules/one-blank-line-after-class/

I don't see a way to disable it in the format part (while it would be possible in the linting), very much like black wasn't configurable.

[flodolo]

See also astral-sh/ruff#9745

[mathjazz]

I don't see a way to disable it in the format part (while it would be possible in the linting), very much like black wasn't configurable.

Oh, I see...

We talked about this PR at the call today and we all agreed we should go ahead.

[mathjazz]

Also, nobody expressed strong opinions the new line after block open and/or about excluding migration files, so let's leave that as it is for now.

Eemeli suggested we also look into https://pycqa.github.io/isort/.

[flodolo]

Eemeli suggested we also look into https://pycqa.github.io/isort/.

In case, looks like it can be fine-tuned for Django (see AMO file)