BUG: Partial support only for extension DTypes in `pandas.core.dtypes.common`

[erezinman]

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
import pandas.core.dtypes.common as pdc

pdc.is_bool_dtype(pd.BooleanDtype)

Issue Description

The above code returns a property and not bool.
The bug here is actually because of pandas.core.dtypes.common.get_dtype that doesn't deal with extension DTypes (I guess it should return the extension DType, but I'm unsure). On the other hand, if the get_dtype method should not work with extension DTypes (for some reason), then it should check if this in an extension DType prior to calling get_dtype.

Anyway, is_bool_dtype should return a boolean and it doesn't necessarily do so (and I'm sure that it's a problem with many other functions in this module).

This is also the issue with not inferring strings that represent extension DTypes. Calling

pdc.is_bool_dtype('boolean')

yields

AttributeError: 'str' object has no attribute '_is_boolean'

Thanks

p.s. Regarding checking the bug's existence, I encountered it in my python v.3.6.9 code (so it can't be updated to the latest), but I checked the code in the repo, and I'm quite sure that it should reproduce.

Expected Behavior

1. get_dtype(ExtensionDtype) should return that DType.
2. is_bool_dtype and similar should always return boolean values.
3. is_bool_dtype('boolean') should not raise an exception.
4. is_bool_dtype(pd.BooleanDtype) should return True (like pdc.is_bool_dtype(pd.Series([], dtype='boolean')) does)

Installed Versions

INSTALLED VERSIONS

commit : 2a7d332
python : 3.6.9.final.0
python-bits : 64
OS : Linux
OS-release : 5.4.0-77-generic
Version : #86~18.04.1-Ubuntu SMP Fri Jun 18 01:23:22 UTC 2021
machine : x86_64
processor : x86_64
byteorder : little
LC_ALL : None
LANG : en_US.UTF-8
LOCALE : en_US.UTF-8
pandas : 1.1.2
numpy : 1.19.5
pytz : 2018.3
dateutil : 2.8.1
pip : 21.1.3
setuptools : 45.2.0
Cython : 0.29.21
pytest : 5.3.5
hypothesis : None
sphinx : None
blosc : None
feather : None
xlsxwriter : None
lxml.etree : 4.5.0
html5lib : 0.999999999
pymysql : None
psycopg2 : None
jinja2 : 2.11.1
IPython : 7.16.1
pandas_datareader: None
bs4 : 4.9.3
bottleneck : None
fsspec : 0.8.7
fastparquet : 0.4.1
gcsfs : None
matplotlib : 3.3.4
numexpr : 2.7.2
odfpy : None
openpyxl : 3.0.7
pandas_gbq : None
pyarrow : 4.0.1
pytables : None
pyxlsb : None
s3fs : None
scipy : 1.4.1
sqlalchemy : 1.4.17
tables : 3.6.1
tabulate : 0.8.7
xarray : 0.16.0
xlrd : 1.2.0
xlwt : None
numba : 0.48.0

[jbrockmendel]

Best guess is that in pdc.pandas_dtype we return the class instead of giving an instance. I think changing that would solve this, not sure if the current behavior is intentional. I think changing this would also fix #31356

[erezinman]

@jbrockmendel Yes, passing pd.BooleanDtype() and not pd.BooleanDtype does solve this issue, but this is not all that there is to it - there's also the exception raised by pdc.is_bool_dtype('boolean'). Also, all the internal dtypes in pandas are immutable, so this behavior of not recognizing pd.BooleanDtype as a valid DType is unexpected.

[erezinman]

Moreover, is_bool_dtype should always return a boolean value or raise a TypeError.

[jreback]

the is_ only take instances and not classes themselves

so this is intentional and works correctly for non EA types

linked PR to address a better error message