Return NothingChanged if non-Python cell magic is detected, to avoid tokenize error

[danielsparing]

Description

Fixes #2627 , a non-Python cell magic such as %%writeline can legitimately contain "incorrect" indentation, however this causes tokenize-rt to return an error. To avoid this, validate_cell should early detect cell magics (just like it detects TransformerManager transformations).

Test added too, in the shape of a "badly indented" %%writefile within test_non_python_magics.

Checklist - did you ...

• Add a CHANGELOG entry if necessary?
• Add / update tests if necessary?
• Add new / update outdated documentation?

[MarcoGorelli]

Thanks @danielsparing for this PR!

I'm not a black maintainer, but I did make the Jupyter PR, so am leaving a couple of comments:

1 - does this mean that lines

black/src/black/handle_ipynb_magics.py

Lines 233 to 234 in 9a73bb8

	if cell_magic_finder.cell_magic.name in NON_PYTHON_CELL_MAGICS:
	raise NothingChanged

can be removed?

2. Perhaps this should be done the other way round? I.e. rather than checking if the magic name is in a blocklist, and skipping the cell if that's the case, black should check if the magic name is in an allow list, and only format the cell if it is. Thoughts on this?

[danielsparing]

@MarcoGorelli Thanks for the quick review!

I agree that the allowlist is cleaner. This also necessitates something I should have done earlier, to exactly check for the first word in the cell for the cell magic.

This does mean that I needed to remove the test with %%t and make it a %%timeit as it is a custom magic, so shouldn't be in the allowlist.

I also removed the lines you mentioned.

What do you think?

[MarcoGorelli]

Thanks for updating @danielsparing !

Generally looks good to me, tagging @JelleZijlstra and @ichard26 (who reviewed the original PR extensively) for further comments

[MarcoGorelli]

I think this can be made more explicit - can you make it clearer what the behaviour was before, and what it is now, so it's clear that black is now intentionally more timid about processing cell magics?

[danielsparing]

addressed in 1e8a0b1

[MarcoGorelli]

Awesome - just got a minor comment left, will handover to the black maintainers now

[ichard26]

Thank you so much for your contribution! This project is only possible by contributions like these 🖤. Seriously, the core team barely has any experience with Jupyter notebooks so we rely on third-party contributions to keep this feature of Black alive and well. You're awesome, @danielsparing - congrats on your first PR to psf/black 🎉

I'm also going to take this moment to thank the defacto core dev on black's jupyter support: @MarcoGorelli!

Y'all are great :)