PERF: datetime index getters functions are 10 times slower with ZoneInfo vs pytz timezone

[ndclt]

Pandas version checks

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

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

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

Reproducible Example

Hello,

Function allowing to get attributes on time index like .month, .year are 10 times slower when the time index is built from a ZoneInfo instead of a pytz.timezone.

With pandas 2.2.3, 3.0.1 or from main, with the use of ZoneInfo:

$ python -m timeit -n 200 -s 'import pandas as pd; from zoneinfo import ZoneInfo' 'pd.date_range("2024-01-01", "2024-12-31 23:59:59", freq="1min", tz=ZoneInfo("Europe/Paris")).month'
200 loops, best of 5: 119 msec per loop

With pandas 2.2.3, 3.0.1 or from main with pytz and dztada installed, with the use of pytz timezone:

$ python -m timeit -n 1000 -s 'import pandas as pd;from pytz import timezone' 'pd.date_range("2024-01-01", "2024-12-31 23:59:59", freq="1min", tz=timezone("Europe/Paris")).month'
1000 loops, best of 5: 11.3 msec per loop

Same behavior with year, day, hour and minute but I didn't check others.

Installed Versions

Details

INSTALLED VERSIONS

commit : 0691c5c
python : 3.11.14
python-bits : 64
OS : Linux
OS-release : 6.17.0-14-generic
Version : #14~24.04.1-Ubuntu SMP PREEMPT_DYNAMIC Thu Jan 15 15:52:10 UTC 2
machine : x86_64
processor : x86_64
byteorder : little
LC_ALL : None
LANG : en_US.UTF-8
LOCALE : en_US.UTF-8

pandas : 2.2.3
numpy : 2.4.2
pytz : 2025.2
dateutil : 2.9.0.post0
pip : 24.0
Cython : None
sphinx : None
IPython : None
adbc-driver-postgresql: None
adbc-driver-sqlite : None
bs4 : None
blosc : None
bottleneck : None
dataframe-api-compat : None
fastparquet : None
fsspec : None
html5lib : None
hypothesis : None
gcsfs : None
jinja2 : None
lxml.etree : None
matplotlib : None
numba : None
numexpr : None
odfpy : None
openpyxl : None
pandas_gbq : None
psycopg2 : None
pymysql : None
pyarrow : 22.0.0
pyreadstat : None
pytest : 8.3.5
python-calamine : None
pyxlsb : None
s3fs : None
scipy : None
sqlalchemy : 2.0.25
tables : None
tabulate : None
xarray : None
xlrd : None
xlsxwriter : None
zstandard : None
tzdata : 2025.3
qtpy : None
pyqt5 : None

Prior Performance

No response

[jbrockmendel]

Can you do some profiling to see where the slowdown occurs?

[kjmin622]

In Localizer class, pytz caches transition arrays upfront and uses C-level binary search to find offsets, whereas ZoneInfo takes the use_tzlocal path, which creates a Python datetime object and calls tz.utcoffset() for every single timestamp. This is what causes the performance difference. If ZoneInfo were handled the same way as pytz, it should achieve comparable performance.
If this analysis is correct, I'd like to take this issue.

[jbrockmendel]

That analysis is correct. Give it a go.

[ndclt]

Hello,
thanks for your quick answer @jbrockmendel and for taking care of it @kjmin622 .

Here the result of the following script with the cProfile.

from zoneinfo import ZoneInfo

import pandas as pd

time_range = pd.date_range(
    "2024-01-01", "2024-12-31 23:59:59", freq="1min", tz=ZoneInfo("Europe/Paris")
)
time_range.month

The file is generated with the following command python -m cProfile -o zoneinfo profile_zoneinfo.py. I had to zip for uploading.
With the same script pycharm profiler shows me that the function _local_timestamps is taking almost all the time.

zoneinfo.zip