Fix Series.map default for na_action

[hamdanal]

• Tests added (Please use assert_type() to assert the type of any return value)

na_action default to None

[cmp0xff]

I tried your idea myself and found one more unaddressed case.

[hamdanal]

Looks like the overloads are not actually needed. I simplified the stub and added the tests you suggested.

[cmp0xff]

Looks like the overloads are not actually needed. I simplified the stub and added the tests you suggested.

Removing the overloads is a bit alerting. I've come up with the following examples.

>>> import pandas as pd
>>> pd.Series([1, 2, 3]).map({1: "a", 2: 1.5})
0      a
1    1.5
2    NaN
dtype: object
>>> pd.Series([1, "a"]).map({1: 1.0, "a": 2.0})
0    1.0
1    2.0
dtype: float64

The first case is pd.Series[int] becoming pd.Series[Any]. The second case is pd.Series[Any] becoming pd.Series[float]. Could you include them in the test?

[hamdanal]

The first case is pd.Series[int] becoming pd.Series[Any].

pyright thinks it is pd.Series[str | float], mypy pd.Series[Any] but this is the same on main so I will not work on it here

[cmp0xff]

This is not a good behaviour, because unknown_series.map({1: True, 0: False, None: None}) gives dtype=object at runtime. We would like to have Series[Any] in such case.

[cmp0xff]

unknown_series.map({1: True, 0: False, None: None}, na_action="ignore") also gives dtype=object at runtime. We would like to have Series[Any] in such case.

[cmp0xff]

Suggested change

	arg: Callable[[Any], object] | Mapping[Any, object] | Series[Any],
	arg: Callable[[Any], Any] | Mapping[Any, Any] | Series,

1. Currently we prefer Any over object to denote "arbitrary" types
2. Series defaults to Series[Any], because S1 defaults to Any.