Apply pendulum to entire date column

[aawalker9]

• [ X] I have searched the issues of this repo and believe that this is not a duplicate.

Issue

Hello! I have tried many things to convert an entire column to a pendulum datetime type to be able to perform operations on it. I have tried in Polars and Pandas. I have converted it to a string and a timestamp with pendulum commands, but neither of these allows me to do operations on it. See example below. How do I perform pendulum operations on entire columns?

[kokoc30]

The key detail in your screenshot is 'Timestamp' object has no attribute 'add'. Your first callback creates Pendulum objects, but pandas infers a datetime dtype from the results. The next callback therefore receives pandas Timestamp objects, which do not have Pendulum's .add(days=...) method. There is no native “Pendulum column” dtype in that operation.

For the string input shown in the screenshot, perform the Pendulum operation inside the same callback, before pandas collects its results:

import pandas as pd
import pendulum

source = pd_df["scheduledEventTimesInUtc"]
pd_df["plus_7_days"] = source.apply(
    lambda value: (
        pd.NaT
        if pd.isna(value)
        else pendulum.parse(value).add(days=7)
    )
)

The final column may again be a pandas datetime column; that is fine if the calculation is complete. Pendulum's add/subtract methods return new values, so remember to assign the result.

If you need to store actual Pendulum objects for several later Pendulum-specific operations, explicitly construct an object Series from those objects:

values = [
    None if pd.isna(value) else pendulum.parse(value)
    for value in source
]
pd_df["pendulum"] = pd.Series(
    values, index=pd_df.index, dtype=object
)

pd_df["plus_7_days"] = pd.Series(
    [None if value is None else value.add(days=7) for value in values],
    index=pd_df.index,
    dtype=object,
)

Specify dtype=object when constructing from the Pendulum objects; changing an already-converted Timestamp column to object dtype does not recreate Pendulum instances. The Series constructor documentation explains dtype inference and explicitly requesting object dtype.

For only the UTC +7-day calculation, pandas also has a direct vectorized expression:

pd_df["plus_7_days"] = pd.to_datetime(source, utc=True) + pd.Timedelta(days=7)

That example adds seven elapsed days to UTC timestamps. Calendar operations in another timezone can have different daylight-saving semantics. I also omitted .naive() intentionally: removing timezone information is a separate decision and is unnecessary to solve the missing-method error.

These examples address the pandas traceback shown. The object/callback approach is Python-level work rather than a vectorized Pendulum engine. I checked the documented APIs but have not executed your data.