Fixed metadata propagation in DataFrame.__getitem__

[TomAugspurger]

xref #28283

[jreback]

can you merge master (i think should be good then)

[TomAugspurger]

There's a deeper problem here. NDFrame.__finalize__ iterates through self._metadata copying items from other to new. When we have a method with a type like f(self: DataFrame, ...) -> Series, we end up trying to copy other.name to the result. That causes problem with a DataFrame with a name column:

In [10]: other = pd.DataFrame({"name": [1, 2]})

In [11]: self.__finalize__(other)
---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-11-0a67385a6694> in <module>
----> 1 self.__finalize__(other)

~/Envs/dask-dev/lib/python3.8/site-packages/pandas/core/generic.py in __finalize__(self, other, method, **kwargs)
   5115             for name in self._metadata:
   5116                 assert isinstance(name, str)
-> 5117                 object.__setattr__(self, name, getattr(other, name, None))
   5118         return self
   5119

~/Envs/dask-dev/lib/python3.8/site-packages/pandas/core/series.py in name(self, value)
    493     def name(self, value: Label) -> None:
    494         if not is_hashable(value):
--> 495             raise TypeError("Series.name must be a hashable type")
    496         object.__setattr__(self, "_name", value)
    497

TypeError: Series.name must be a hashable type

I'll think a bit about this.

[github-actions]

This pull request is stale because it has been open for thirty days with no activity. Please update or respond to this comment if you're still interested in working on this.

[TomAugspurger]

df.eval(...) now propagates attrs with engine="python", but engine="numexpr" doesn't. We can fix numexpr as a followup.

[jreback]

thanks @TomAugspurger