Correct type annotations on NetworkX DiGraphs

[charmoniumQ]

Two fixes:

• in_degree is a cached_property that is InDegreeView or InMultiDegreeView (for a MultiDiGraph). InDegreeView can then return an int or itself when called. However, the Graph's property in_degree is never itself an int, always an InDegreeView (source).
• Calling a NodeView returns not just any Iterator, but a NodeView itself. See example in documentation,

[jenshnielsen]

Thanks, this closes #14592

[github-actions]

Diff from mypy_primer, showing the effect of this PR on open source code:

discord.py (https://github.com/Rapptz/discord.py)
- discord/ext/commands/hybrid.py:508: error: Overlap between argument names and ** TypedDict items: "description", "name"  [misc]
+ discord/ext/commands/hybrid.py:508: error: Overlap between argument names and ** TypedDict items: "name", "description"  [misc]
- discord/ext/commands/hybrid.py:629: error: Overlap between argument names and ** TypedDict items: "description", "name"  [misc]
+ discord/ext/commands/hybrid.py:629: error: Overlap between argument names and ** TypedDict items: "name", "description"  [misc]

[Avasam]

I believe @srittau 's comment in #14595 (comment) only applied to the first overload
When data is True, it's not returning self

As-is, this overload is useless since only the params differ (could be a Union, but I think it was correct to have an overload, see:)

Suggested change

	def __call__(self, data: Literal[False] = False, default=None) -> Self: ...
	@overload
	def __call__(self, data: Literal[True] | str, default=None) -> NodeDataView[_Node]: ...
	def data(self, data: bool | str = True, default=None) -> NodeDataView[_Node]: ...
	def __call__(self, data: Literal[True] | str, default=None) -> Self: ...
	def data(self, data: bool | str = True, default=None) -> Self: ...
	@overload
	def __call__(self, data: Literal[False] = False, default=None) -> Self: ...
	@overload
	def __call__(self, data: Literal[True] | str, default=None) -> NodeDataView[_Node]: ...
	@overload
	def data(self, data: Literal[False], default=None) -> Self: ...
	@overload
	def data(self, data: Literal[True] | str = True, default=None) -> NodeDataView[_Node]: ...

[charmoniumQ]

You are right; I misread NodeView as NodeDataView in a haste.

[Avasam]

The overlap error is correct here. I also think it's inverted since nbunch=None should return self

Suggested change

	@overload
	def __call__(self, nbunch: None = None, weight: None | bool | str = None) -> int: ... # type: ignore[overload-overlap]
	@overload
	def __call__(self, nbunch: None | Iterable[_Node], weight: None | bool | str = None) -> Self: ...
	@overload
	def __call__(self, nbunch: None = None, weight: None | bool | str = None) -> Self: ...
	@overload
	def __call__(self, nbunch: Iterable[_Node], weight: None | bool | str = None) -> int: ...

[charmoniumQ]

You're right; the overload should be specialized to _Node (which still needs to ignore possible overlap between None/NoneType and _Node, as is done in OutEdgeView).

None and Iterable[_Node] return Self (first and last return), but _Node returns int (middle two returns)

(ignoring, like we have been, that it could actually return float in the case of graphs whose edge weights are float)

[Avasam]

Oh I was too slow on the review and this got merged ^^"

No worries, I'll propose the same changes in #14597