BUG: pandas 2.2.0 DataFrame.groupby regression when group labels of type pd.Timestamp

[maread99]

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
assert pd.__version__ == "2.2.0"

index = pd.date_range("2024-02-01", "2024-02-03", freq="8h", tz="UTC")

# index
# DatetimeIndex(['2024-02-01 00:00:00+00:00', '2024-02-01 08:00:00+00:00',
#                '2024-02-01 16:00:00+00:00', '2024-02-02 00:00:00+00:00',
#                '2024-02-02 08:00:00+00:00', '2024-02-02 16:00:00+00:00',
#                '2024-02-03 00:00:00+00:00'],
#               dtype='datetime64[ns, UTC]', freq='8h')

df = pd.DataFrame([1] * len(index), index)

def f(ts: pd.Timestamp) -> pd.Timestamp:
    """Group by date, i.e. as timezone naive timestamp."""
    return ts.normalize().tz_convert(None)

grouped = df.groupby(by=f)
keys = list(grouped.groups.keys())
print(keys)

# [Timestamp('2024-02-01 00:00:00+0000', tz='UTC'),
#  Timestamp('2024-02-02 00:00:00+0000', tz='UTC'),
#  Timestamp('2024-02-03 00:00:00+0000', tz='UTC')]

Issue Description

Although the return from the 'by' function is timezone naive, the group labels are timezone aware (UTC).

(I've tried every combination of the as_index and group_keys options, always get the same group keys).

Expected Behavior

Would have expected behaviour as pandas 2.1.4 and prior, i.e,. group keys as returned from the 'by' function...

import pandas as pd
assert pd.__version__ == "2.1.4"

index = pd.date_range("2024-02-01", "2024-02-03", freq="8h", tz="UTC")
df = pd.DataFrame([1] * len(index), index)

def f(ts: pd.Timestamp) -> pd.Timestamp:
    return ts.normalize().tz_convert(None)

grouped = df.groupby(by=f)
keys = list(grouped.groups.keys())
print(keys)

[Timestamp('2024-02-01 00:00:00'),
 Timestamp('2024-02-02 00:00:00'),
 Timestamp('2024-02-03 00:00:00')]

Installed Versions

INSTALLED VERSIONS

commit : f538741
python : 3.9.13.final.0
python-bits : 64
OS : Windows
OS-release : 10
Version : 10.0.22631
machine : AMD64
processor : Intel64 Family 6 Model 141 Stepping 1, GenuineIntel
byteorder : little
LC_ALL : None
LANG : None
LOCALE : English_United Kingdom.1252

pandas : 2.2.0
numpy : 1.26.3
pytz : 2023.4
dateutil : 2.8.2
setuptools : 58.1.0
pip : 23.3.2
Cython : None
pytest : 8.0.0
hypothesis : 6.97.3
sphinx : None
blosc : None
feather : None
xlsxwriter : None
lxml.etree : 4.9.4
html5lib : None
pymysql : None
psycopg2 : None
jinja2 : None
IPython : 8.14.0
pandas_datareader : None
adbc-driver-postgresql: None
adbc-driver-sqlite : None
bs4 : 4.12.3
bottleneck : None
dataframe-api-compat : None
fastparquet : None
fsspec : None
gcsfs : None
matplotlib : None
numba : None
numexpr : 2.9.0
odfpy : None
openpyxl : None
pandas_gbq : None
pyarrow : 15.0.0
pyreadstat : None
python-calamine : None
pyxlsb : None
s3fs : None
scipy : None
sqlalchemy : None
tables : 3.9.2
tabulate : None
xarray : None
xlrd : None
zstandard : None
tzdata : 2023.4
qtpy : None
pyqt5 : None

[rhshadrach]

Thanks for the report. Result of a git bisect:

commit 746e5eee860b6e143c33c9b985e095dac2e42677
Author: jbrockmendel
Date:   Mon Oct 16 11:50:28 2023 -0700

    ENH: EA._from_scalars (#53089)

cc @jbrockmendel

[jbrockmendel]

yah i think thats a problem in DTA._from_scalars. _from_scalars itself was a mistake xref #56430

[rhshadrach]

This can be reproduced just using index.map, so taking off the groupby label.