BUG: nanoseconds argument for DateOffset results in wrong output

[GYHHAHA]

>>>ts = pd.Timestamp('2020-1-1')
>>>ts + pd.DateOffset(nanoseconds=1000)
Timestamp('2020-01-02 00:00:00')

This is related to offset = timedelta(1) in _determine_offset from pd._libs.tslibs.offsets.

[phofl]

Thanks for your report.

Should we raise in _determine_offset if the only item in kwds is nanoseconds/s?

[GYHHAHA]

Yes I think so. Users can use nano in Timedelta to handle this instead of DateOffset.

Should I open a PR? I'm not sure if other developers have other opinion.

[jbrockmendel]

@tushushu i think this was closed by #43968. can you confirm?

[tushushu]

@tushushu i think this was closed by #43968. can you confirm?

Yes, confirmed with that.