DOC: Fix NaT and NA API docs

[meet-vasita]

Set __module__ = "pandas" on NaTType and NAType to fix incorrect autodoc "alias of" messages.
Closes #62185.

[meet-vasita]

Hi @rhshadrach, I've implemented your suggested fix by adding __module__ = "pandas" to both NAType and NaTType classes.

Would you be able to review this PR when you have a chance? Thanks for identifying the root cause and providing the solution direction.

[simonjayhawkins]

/docbuild

[simonjayhawkins]

/preview

[github-actions]

Website preview of this PR available at: https://pandas.pydata.org/preview/pandas-dev/pandas/62313/

[simonjayhawkins]

@meet-vasita thanks for the PR.

i'm not seeing any difference at https://pandas.pydata.org/preview/pandas-dev/pandas/62313/docs/reference/api/pandas.NA.html or https://pandas.pydata.org/preview/pandas-dev/pandas/62313/docs/reference/api/pandas.NaT.html

[rhshadrach]

Thanks @meet-vasita and @simonjayhawkins - I think the remaining issue is that NA and NaT are not classes, they are instances of classes. I don't believe instances play well with autodoc. I get the "desired" output by changing https://github.com/pandas-dev/pandas/blob/main/doc/source/reference/missing_value.rst to use

api.typing.NAType

and

api.typing.NaTType

But there is the less-than desired effect of having the class show up in the API page. Open to thoughts here.

[wjandrea]

@rhshadrach I'm out of my depth on Sphinx, but the SO page you linked had these comments - do they help?

• "If you do this then you break inspect.getsource(Foo)" — Eric
• "Do this in the setup function in conf.py and then you won't mess with anything in the original source code. For example: github.com/slundberg/shap/blob/…" — slund

[rhshadrach]

@wjandrea - no, we'd still face the issue I identified in my previous comment.

[jorisvandenbossche]

Since this worked fine in the past, this might be caused by c8e7a98, which while deduplicating the entry also added a :template: autosummary/class_without_autosummary.rst (originally, the first occurence did not have that).

Testing it out locally, removing that line from doc/source/reference/missing_value.rst seems to fix it

[jorisvandenbossche]

(and we can keep the __module__ changes, I think, since they are useful anyway regardless of the doc issue)

[rhshadrach]

lgtm

[rhshadrach]

Thanks @meet-vasita!

[meet-vasita]

You’re welcome @rhshadrach, I’m glad it was helpful.