Period slicing with period_range returns error

[dsimpson1980]

Slicing a TimeSeries object with a PeriodIndex fails with an index out of bounds error

Example

dates = pd.period_range(start='01-Apr-11',end='31-Mar-12 23:00',freq='h')
ts = pd.TimeSeries(0,dates)
ts[ts.index[4:10]]

Traceback (most recent call last):

File "", line 3, in
ts[ts.index[4:10]]

File "C:\dev\bin\Python27\lib\site-packages\pandas\core\series.py", line 506, in getitem
return self._get_with(key)

File "C:\dev\bin\Python27\lib\site-packages\pandas\core\series.py", line 544, in _get_with
return self._get_values(key)

File "C:\dev\bin\Python27\lib\site-packages\pandas\core\series.py", line 575, in _get_values
return self.values[indexer]

IndexError: index 361564 out of bounds 0<=index<8784

It appears the PeriodIndex is being inferred as 'integer' in series._get_with() using lib.infer_dtype() whereas self.index.inferred_type returns 'period'

        if isinstance(key, Index):
            key_type = lib.infer_dtype(key.values)
        else:
            key_type = lib.infer_dtype(key)

        if key_type == 'integer':
            if self.index.inferred_type == 'integer':
                return self.reindex(key)
            else:
                return self._get_values(key)
        elif key_type == 'boolean':
            return self._get_values(key)

[stephenwlin]

changing

            if isinstance(key, Index):
                key_type = lib.infer_dtype(key.values)
            else:
                key_type = lib.infer_dtype(key)

to

            if isinstance(key, Index): 
                key_type = key.inferred_type 
            else: 
                key_type = lib.infer_dtype(key)

does the trick and doesn't break any tests, but it seems kind of odd that it was written the way it was.

[stephenwlin]

the fix is merged into master #2785