BUG: indexing regression with datetime index

[0x0L]

Sample code:

import pandas as pd

idx = pd.date_range('2008', '2009')
idx.to_series()['2008']

in 0.25.3 we get

2008-01-01   2008-01-01
2008-01-02   2008-01-02
2008-01-03   2008-01-03
2008-01-04   2008-01-04
2008-01-05   2008-01-05
                ...    
2008-12-27   2008-12-27
2008-12-28   2008-12-28
2008-12-29   2008-12-29
2008-12-30   2008-12-30
2008-12-31   2008-12-31
Freq: D, Length: 366, dtype: datetime64[ns]

in >= 1.0.0 (up to 1.0.4 at least) we get

2008   2008-01-01
2008   2008-01-02
2008   2008-01-03
2008   2008-01-04
2008   2008-01-05
          ...    
2008   2008-12-27
2008   2008-12-28
2008   2008-12-29
2008   2008-12-30
2008   2008-12-31
Length: 366, dtype: datetime64[ns]

The index has dtype obj

Variantations like

idx.to_series().astype(str)['2008']
idx.to_series().loc['2008']

look unaffected

[TomAugspurger]

Working as expected on master

Out[1]:
2008-01-01   2008-01-01
2008-01-02   2008-01-02
2008-01-03   2008-01-03
2008-01-04   2008-01-04
2008-01-05   2008-01-05
                ...
2008-12-27   2008-12-27
2008-12-28   2008-12-28
2008-12-29   2008-12-29
2008-12-30   2008-12-30
2008-12-31   2008-12-31
Freq: D, Length: 366, dtype: datetime64[ns]

I don't recall which commit fixed it, but it'd be good to ensure that we have test coverage for this in pandas/tests/series/indexing/test_datetime.py.

[0x0L]

@TomAugspurger
Thanks. I'll check to find the exact commit: it is fixed in master but not in 1.0.5
I'll add my example as a test case if you think that's a good idea

[faraz16iqbal]

Interested in working on this issue!

[0x0L]

Hi @faraz16iqbal !
Feel free to submit a pull request. The contributing guide is a good starting point.