Support type stub generation for staticmethod

[WeilerMarcel]

Fixes #13574

This PR fixes the generation of type hints for static methods of pybind11 classes. The code changes are based on the suggestions in #13574.

The fix introduces an additional check if the property under inspection is of type staticmethod. If it is, the type information is read from the staticmethod's __func__ attribute, instead of the staticmethod instance itself.

I added a test for C++ classes with static methods bound using pybind11. Both, an overloaded and a non-overloaded static method are tested.

[WeilerMarcel]

Hi @sobolevn and @AlexWaygood, I updated this PR to work with the latest changes on master. Could you please review the current changes and merge the PR if everything is fine?

[WeilerMarcel]

Hey @sobolevn and @AlexWaygood, what is the status of this pull request? I would really like to see it merged into MyPy. Do you need anything from my side to proceed?

[d4n1elchen]

Looking forward to this getting merged as well +1

[WeilerMarcel]

Hi @sobolevn and @AlexWaygood, please give me an update on this PR. Is there a reason why it has not been merged? As far as I can see, I can not merge it myself since I don't have the necessary permissions.

[sobolevn]

I can't merge this PR, because I don't fully understand the C++ code, but I think that @JukkaL or @msullivan can help us :)

[AlexWaygood]

I can't merge this PR because I'm not a maintainer

[WeilerMarcel]

Thanks for the info :) I was under the impression that you could merge, since you were reviewing my changes.
I will wait for a reply from @JukkaL and @msullivan.

[sobolevn]

In the meantime you can resolve conflicts :)

[WeilerMarcel]

Hi everyone, I integrated my changes into the new master branch and from my perspective it is ready for merging.

I noticed, though, that the mypyc runtime tests with py39-macos seem to be unstable. Since I did not find a way to rerun tests for a commit, I added two empty dummy commits to get the pipeline to succeed. See #16420.

[JukkaL]

You can ignore the mypyc runtime test failures, since it's a known issue happening on master.

[AlexWaygood]

Why has this been # noqa'd? Wouldn't it be cleaner to do this, as the linter suggests?

Suggested change

	raw_lookup = getattr(class_info.cls, "__dict__") # noqa: B009
	raw_lookup = class_info.cls.__dict__

[WeilerMarcel]

To be honest, I wrote it that way because there are two other places in stubgenc.py where this pattern was used:

• mypy/mypy/stubgenc.py

  Lines 453 to 455 in 8121e3c

  	def get_members(self, obj: object) -> list[tuple[str, Any]]:
  	obj_dict: Mapping[str, Any] = getattr(obj, "__dict__") # noqa: B009
  	results = []

• mypy/mypy/stubgenc.py

  Lines 718 to 725 in 8121e3c

  	def generate_class_stub(self, class_name: str, cls: type, output: list[str]) -> None:
  	"""Generate stub for a single class using runtime introspection.
  	The result lines will be appended to 'output'. If necessary, any
  	required names will be added to 'imports'.
  	"""
  	raw_lookup = getattr(cls, "__dict__") # noqa: B009
  	items = self.get_members(cls)

  
  I will check the file history to see if I can find a specific reason why getattr is used.

[WeilerMarcel]

The getattr call originally comes from commit 6bcaf40 and was added to make typeshed happy:

mypy/mypy/stubgenc.py

Lines 140 to 142 in 6bcaf40

	# typeshed gives obj.__dict__ the not quite correct type Dict[str, Any]
	# (it could be a mappingproxy!), which makes mypyc mad, so obfuscate it.
	obj_dict = getattr(obj, '__dict__') # type: Mapping[str, Any]

The # noqa was added in 4287af4:

mypy/mypy/stubgenc.py

Line 246 in 4287af4

obj_dict = getattr(obj, '__dict__') # type: Mapping[str, Any] # noqa

[WeilerMarcel]

Hi @JukkaL and @msullivan, I resolved conflicts with the current master branch. Can this PR be merged?

[WeilerMarcel]

Happy new year everyone! 🥳
Are there plans to merge this PR yet? I would really like to get this off my mind.

[JukkaL]

I can merge this PR if somebody who has more context on the relevant code than I have gives a quick thumbs up.

[JukkaL]

more context on the relevant code than I have

Or somebody who has any familiarity with the code :-)

[hauntsaninja]

Not that familiar with stubgen, but this mostly looks good to me. I had one question

[hauntsaninja]

Thank you! :-)