feat: Improve DataFrame.__getitem__ consistency

[MarcoGorelli]

closes #2405

What type of PR is this? (check all applicable)

• 💾 Refactor
• ✨ Feature
• 🐛 Bug Fix
• 🔧 Optimization
• 📝 Documentation
• ✅ Test
• 🐳 Other

Related issues

• Related issue #<issue number>
• Closes #<issue number>

Checklist

• Code follows style guide (ruff)
• Tests added
• Documented the changes

If you have comments or can explain your changes, please do so below

[dangotbanned]

Yes please 😍

I'm pretty sure these were on the chopping block from

• [internal]: Adopting/following more ruff complexity rules #2376

[dangotbanned]

Feel like I didn't do a good enough job documenting this pattern

• https://github.com/narwhals-dev/narwhals/pull/2315/files#diff-76bea174dd4497eac2915b830209c9b96f3b9004d6be41b9d2405fcf8e2e45a2

narwhals/narwhals/_compliant/namespace.py

Lines 146 to 158 in 26d97e5

	@overload
	def from_native(self, data: NativeFrameT, /) -> EagerDataFrameT: ...
	@overload
	def from_native(self, data: NativeSeriesT, /) -> EagerSeriesT: ...
	def from_native(
	self, data: NativeFrameT | NativeSeriesT | Any, /
	) -> EagerDataFrameT | EagerSeriesT:
	if self._dataframe._is_native(data):
	return self._dataframe.from_native(data, context=self)
	elif self._series._is_native(data):
	return self._series.from_native(data, context=self)
	msg = f"Unsupported type: {type(data).__name__!r}"
	raise TypeError(msg)

[lorentzenchr]

@MarcoGorelli I think this PR will be super useful for scikit-learn.
Do you also envision to allow for boolean masks, especially for row indexing? (Not a must, but - again - useful.)

[MarcoGorelli]

Thanks for your feedback!

Do you also envision to allow for boolean masks, especially for row indexing?

Maybe, in the meantime you can use filter for that

[MarcoGorelli]

thanks all for comments 🙏 really appreciate it!

especially:

• @lorentzenchr for highlighting this area of improvement - we'd added __getitem__ a while ago but then hadn't touched it in some time, really glad to have revisited it
• @sjdenny for the hypothesis tests, which helped avoid regressing on some edge cases
• @dangotbanned for your patient reviews and help with annotations, which make everything SO much better

---

EDIT: I stripped down the commit message to avoid it containing 97 commits (to avoid git log becoming unreadable), but in doing so accidentally removed the co-author information - sorry about that, I'll be more careful in the future