DOC: to_pylist returns a pandas.Timestamp instead of datetime.datetime when the type is timestamp[ns]

[mroeschke]

Describe the enhancement requested

In [2]: import pyarrow as pa; from datetime import datetime

In [4]: pa.array([datetime(2023, 1, 1)], type=pa.timestamp("ns")).to_pylist()
Out[4]: [Timestamp('2023-01-01 00:00:00')]

From the docs, I assumed Convert to a list of native Python objects. means I should always expected a datetime.datetimeobject, but since I suppose it cannot handle nanosecond components this method returns a pandas.Timestamp when the unit is "ns"

Component(s)

Python

[jorisvandenbossche]

but since I suppose it cannot handle nanosecond components this method returns a pandas.Timestamp when the unit is "ns"

That's a correct assumption. I don't think there is anything to do about that? (except for raising an error instead, given that datetime.datetime can't handle nanoseconds)

[mroeschke]

Yeah would be good to document this behavior. Or as you mentioned potentially raising. Or warning that nanosecond components would be dropped if storing in a datetime.datetime. I don't have a strong preference either way.