BUG: to_datetime with unit mapping ignores utc and box arguments

[mroeschke]

In [12]: pd.__version__
Out[12]: '0.24.0.dev0+1040.g24bce1a5f'

In [13]: df = pd.DataFrame({'year': [2015, 2016],
    ...:                        'month': [2, 3],
    ...:                        'day': [4, 5]})
    ...:

In [14]: pd.to_datetime(df, box=False)
Out[14]:
0   2015-02-04
1   2016-03-05
dtype: datetime64[ns]

In [15]: pd.to_datetime(df, utc=True)
Out[15]:
0   2015-02-04
1   2016-03-05
dtype: datetime64[ns]

These arguments need to be handled directly in this function or the output

pandas/pandas/core/tools/datetimes.py

Line 574 in 24bce1a

result = _assemble_from_unit_mappings(arg, errors=errors)

Expected Output

In [16]: pd.to_datetime(df, box=False).values
Out[16]:
array(['2015-02-04T00:00:00.000000000', '2016-03-05T00:00:00.000000000'],
      dtype='datetime64[ns]')

In [17]: pd.to_datetime(df, utc=True).dt.tz_localize('UTC')
Out[17]:
0   2015-02-04 00:00:00+00:00
1   2016-03-05 00:00:00+00:00
dtype: datetime64[ns, UTC]

[evangelinehl]

Hi,

Can my team and I contribute to this issue?

Thanks.

[mroeschke]

Sure, go for it @evangelineliu