Lookup using datetimes does not work with hierarchical indices containing periods

[SleepingPills]

Lookup in a PeriodIndex using a datetime works as expected (the period in which the timestamp falls will be returned). However, when the PeriodIndex is part of a hierarchy, this functionality fails in a non-obvious way:

>>> s = pd.Series([1,2,3,4,5], pd.MultiIndex.from_arrays([["a", "a", "a", "b", "b"], pd.period_range("2012-01", periods=5, freq="M")]))
>>> s.loc["a", datetime(2012,1,1)]
Traceback (most recent call last):
  File "C:\VirtualEnvs\test\lib\site-packages\ipython-1.0.dev-py2.6.egg\IPython\core\interactiveshell.py", line 2837, in run_code
    exec code_obj in self.user_global_ns, self.user_ns
  File "<ipython-input-18-9e6cd34eee66>", line 1, in <module>
    a.loc["a", datetime(2012,1,1)]
  File "C:\VirtualEnvs\test\lib\site-packages\pandas-0.12.0-py2.6-win32.egg\pandas\core\indexing.py", line 697, in __getitem__
    return self._getitem_tuple(key)
  File "C:\VirtualEnvs\test\lib\site-packages\pandas-0.12.0-py2.6-win32.egg\pandas\core\indexing.py", line 258, in _getitem_tuple
    self._has_valid_tuple(tup)
  File "C:\VirtualEnvs\test\lib\site-packages\pandas-0.12.0-py2.6-win32.egg\pandas\core\indexing.py", line 691, in _has_valid_tuple
    raise ValueError('Too many indexers')
ValueError: Too many indexers

Using a period works just fine:

>>> s.loc["a", pd.Period("2012-01")]
1

[SleepingPills]

A possibly related issue (happens when using a MultiIndex containing periods), is that when querying with a label that is not in the index, a ValueError("Too many indexers") will be raised instead of a KeyError.

[jbrockmendel]

Works in 0.23.2, needs test.