BUG: Pandas DataFrame replace() doesn't work if the dataframe has nullable boolean columns

[ilCatania]

• 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 master branch of pandas.

Reproducible Example

import pandas as pd
df = pd.DataFrame({"a": ["x", "y", "z", "w"], "b": [True, False, False, True]}).astype({"b":"boolean"})
df.replace("x", "X")

Issue Description

When trying to use pandas.DataFrame.replace() on a dataframe with several columns including boolean columns where the boolean (i.e. nullable) dtype has been set explicitly beforehand. When running the above snippet I get the following error:

File "C:\users\el_caminooooo\playground\venv/lib/python3.7/site-packages/pandas/core/arrays/boolean.py", line 265, in __init__
    "values should be boolean numpy array. Use "
TypeError: values should be boolean numpy array. Use the 'pd.array' function instead

Is this a pandas bug?. Also please note both of the following work fine:

import pandas as pd
df = pd.DataFrame({"a": ["x", "y", "z", "w"], "b": [True, False, False, True]}).astype({"b":"bool"})  # notice 'bool' instead of 'boolean'
df.replace("x", "X")

and:

df = pd.DataFrame({"a": ["x", "y", "z", "w"], "b": [True, False, False, True]})
df.replace("x", "X")

Expected Behavior

the replace should complete successfully

Installed Versions

INSTALLED VERSIONS ------------------ commit : 7c48ff4 python : 3.7.10.final.0 python-bits : 64 OS : Linux OS-release : 3.10.0-957.27.2.el7.x86_64 Version : #1 SMP Mon Jul 29 17:46:05 UTC 2019 machine : x86_64 processor : x86_64 byteorder : little LC_ALL : None LANG : en_US.UTF-8 LOCALE : en_US.UTF-8 pandas : 1.3.4 numpy : 1.21.0 pytz : 2021.1 dateutil : 2.8.1 pip : 21.3.1 setuptools : 47.1.0 Cython : 0.29.23 pytest : 6.2.5 hypothesis : None sphinx : None blosc : None feather : None xlsxwriter : None lxml.etree : 4.6.3 html5lib : None pymysql : None psycopg2 : None jinja2 : 3.0.1 IPython : None pandas_datareader: 0.10.0 bs4 : None bottleneck : None fsspec : 2021.08.1 fastparquet : 0.7.0 gcsfs : None matplotlib : 3.4.2 numexpr : None odfpy : None openpyxl : 3.0.9 pandas_gbq : None pyarrow : 4.0.0 pyxlsb : None s3fs : None scipy : 1.6.3 sqlalchemy : None tables : None tabulate : 0.8.9 xarray : None xlrd : None xlwt : None numba : None

[ilCatania]

Also featured as a stackoverflow question: https://stackoverflow.com/q/70004591/902755

[mzeitlin11]

Thanks for reporting this @ilCatania! Definitely a pandas bug, looks to trace back to comparing a nullable boolean column to a string.