get_level_values() method for MultiIndex containing dates

[aldanor]

I was wondering if it was possible to elegantly extract the values of a MultiIndex which contains dates as datetime (and not an ndarray of datetime64[ns] which screws things up)?

>>> print df

                  numtrades
date       id              
2012-01-03 11891          1
           14259          2
2012-01-05 14259          1
2012-01-06 11891          7
2012-01-09 11891          2
2012-01-10 11891         12
           14259          4
           10304          2
2012-01-11 11891         11
           14259          8

>>>  df.index[0]

(<Timestamp: 2012-01-03 00:00:00>, 11891)

>>> df.index.get_level_values('date')

array([1970-01-16 16:00:00, 1970-01-16 16:00:00, 1970-01-16 64:00:00,
       1970-01-16 88:00:00, 1970-01-16 160:00:00, 1970-01-16 184:00:00,
       1970-01-16 184:00:00, 1970-01-16 184:00:00, 1970-01-16 208:00:00,
       1970-01-16 208:00:00], dtype=datetime64[ns])

Of course, we can always use to_datetime and then to_pydatetime, is that the preferred method? Am I missing something?

[wesm]

Yeah, this should be reasonably simple to work around

[aldanor]

Yea. It just looks a bit inconsistent, I guess.