REF: define arithmetic methods non-dynamically

[jbrockmendel]

• closes #xxxx (Replace xxxx with the GitHub issue number)
• Tests added and passed if fixing a bug or adding a new feature
• All code checks passed.
• Added type annotations to new arguments/methods/functions.
• Added an entry in the latest doc/source/whatsnew/vX.X.X.rst file if fixing a bug or adding a new feature.

Avoid runtime imports, get stuff out of ops.__init__, which was the goal 4-5 years ago when ops was first madea directory.

Move relevant helper methods from frame/series files . Decided on a new mixin rather than dumping everything directly in DataFrame/Series bc those files are too bug as it is. If others feel strongly i'd be OK putting it all in DataFrame/Series.

[jbrockmendel]

@Dr-Irv how do i get pyright to stop complaining about this?

  /home/runner/work/pandas/pandas/pandas/core/ops/methods.py:785:20 - error: Cannot assign member "name" for type "tuple[Series, Series]"
    Member "name" is unknown (reportGeneralTypeIssues)

[twoertwein]

I removed the ignore comment locally and run pre-commit run -av --hook-stage manual pyright_reportGeneralTypeIssues but did not get an error

(could also bump pyright to 1.1.296)

[Dr-Irv]

Was thinking this could be a pyright bug, but might be difficult to create a reproducible example. At least when I've seen that kind of error, it's usually been a pyright bug

[twoertwein]

I assume sub-classes implement this method? Would it be worth making this an abstractmethod to ensure that sub-classes definitely have it (and to make the type checkers happy)?

[jbrockmendel]

yah this is only inherited by DataFrame.

might just put this all in Series/DataFrame instead of having a subclass to avoid this typing gymnastics

[mroeschke]

+1 to have these directly on DataFrame/Series, less indirection too

[jbrockmendel]

moved + green

[jbrockmendel]

The failing doctest looks like the doctest is wrong. Must just not be getting in the status quo?

>>> a = pd.Series([1, 1, 1, np.nan], index=['a', 'b', 'c', 'd'])
>>> b = pd.Series([1, np.nan, 1, np.nan], index=['a', 'b', 'd', 'e'])
>>> a.divmod(b, fill_value=0)
(a    1.0
- b    NaN
- c    NaN
+ b    inf
+ c    inf
 d    0.0
 e    NaN
 dtype: float64,
 a    0.0
 b    NaN
 c    NaN
 d    0.0
 e    NaN
 dtype: float64)

[mroeschke]

The failing doctest looks like the doctest is wrong. Must just not be getting in the status quo?

Hmm is this the same for DataFrame too and/or a recent change?

For label b, I would expect this to follow:

In [10]: 1 / np.nan
Out[10]: nan

For label c, since there's no corresponding values in series b I would expect it to be filled with nan too?

[jbrockmendel]

I would expect it to be filled with nan too?

At the reindex stage it is, but then fill_value=0 part is done after the reindex.

[mroeschke]

Thanks @jbrockmendel