BUG: The count() API does not consider inf as NA (with pandas.options.mode.use_inf_as_na = True) when dtype is "double[pyarrow]"

[tinadu0806]

Pandas version checks

• I have checked that this issue has not already been reported.

• I have confirmed this bug exists on the latest version of pandas.

• I have confirmed this bug exists on the main branch of pandas.

Reproducible Example

>>>import pandas as pd
>>>pd.options.mode.use_inf_as_na = True
>>>s = pd.Series([1.1, float("inf"), float("-inf"), 2.3], dtype="double[pyarrow]")
>>>s
0    1.1
1    NaN
2    NaN
3    2.3
dtype: double[pyarrow]
>>>s.count()
4
>>>type(s.count())
<class 'numpy.int64'>

Issue Description

When create a pandas series with pyarrow type "double[pyarrow]", after setting pd.options.mode.use_inf_as_na = True, the inf values are still counted. Also it returns a numpy int64 type scalar instead of pyarrow double scalar.

Expected Behavior

After setting pd.options.mode.use_inf_as_na = True, above s.count() should return 2 as result.

Installed Versions

Replace this line with the output of pd.show_versions()

[mroeschke]

Thanks for the report. Yeah generally no extension array types will respect this option.

We are also considering deprecating the use_inf_as_na option #51684 in favor of replacing inf with na first.

[mroeschke]

Given that this was deprecated in 2.1, closing as a won't fix