MultiIndexed unstack with tuple names fails with KeyError

[TomAugspurger]

In [8]: idx = pd.MultiIndex.from_product([['a', 'b', 'c'], [1, 2, 3]], names=[('A', 'a'), ('B', 'b')])

In [9]: s = pd.Series(1, index=idx)

In [10]: s
Out[10]:
(A, a)  (B, b)
a       1         1
        2         1
        3         1
b       1         1
        2         1
        3         1
c       1         1
        2         1
        3         1
dtype: int64

In [11]: s.unstack(("A", "a"))
---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
~/Envs/pandas-dev/lib/python3.6/site-packages/pandas/pandas/core/indexes/multi.py in _get_level_number(self, level)
    749                                  'level number' % level)
--> 750             level = self.names.index(level)
    751         except ValueError:

ValueError: 'A' is not in list

During handling of the above exception, another exception occurred:

KeyError                                  Traceback (most recent call last)
<ipython-input-11-1ce241b42d82> in <module>()
----> 1 s.unstack(("A", "a"))

~/Envs/pandas-dev/lib/python3.6/site-packages/pandas/pandas/core/series.py in unstack(self, level, fill_value)
   2231         """
   2232         from pandas.core.reshape.reshape import unstack
-> 2233         return unstack(self, level, fill_value)
   2234
   2235     # ----------------------------------------------------------------------

~/Envs/pandas-dev/lib/python3.6/site-packages/pandas/pandas/core/reshape/reshape.py in unstack(obj, level, fill_value)
    481             # _unstack_multiple only handles MultiIndexes,
    482             # and isn't needed for a single level
--> 483             return _unstack_multiple(obj, level, fill_value=fill_value)
    484         else:
    485             level = level[0]

~/Envs/pandas-dev/lib/python3.6/site-packages/pandas/pandas/core/reshape/reshape.py in _unstack_multiple(data, clocs, fill_value)
    315     index = data.index
    316
--> 317     clocs = [index._get_level_number(i) for i in clocs]
    318
    319     rlocs = [i for i in range(index.nlevels) if i not in clocs]

~/Envs/pandas-dev/lib/python3.6/site-packages/pandas/pandas/core/reshape/reshape.py in <listcomp>(.0)
    315     index = data.index
    316
--> 317     clocs = [index._get_level_number(i) for i in clocs]
    318
    319     rlocs = [i for i in range(index.nlevels) if i not in clocs]

~/Envs/pandas-dev/lib/python3.6/site-packages/pandas/pandas/core/indexes/multi.py in _get_level_number(self, level)
    751         except ValueError:
    752             if not isinstance(level, int):
--> 753                 raise KeyError('Level %s not found' % str(level))
    754             elif level < 0:
    755                 level += self.nlevels

KeyError: 'Level A not found'

cc @ibrahimsharaf, @toobaz does this look difficult?

[toobaz]

Reminds me of #18321 probably unrelated, and probably simpler... one of these cases in which we probably only need to tell unstack that tuples are not lists.

[ibrahimsharaf]

Hi @TomAugspurger, I'm willing to help here, do you have any initial ideas on how to fix this? (before I dive in the code)

[TomAugspurger]

in reshape/reshape, the line here

--> 317 clocs = [index._get_level_number(i) for i in clocs]

is incorrect. It's assuming that all iterables are referring to multiple levels. In this case, we have a single level whose name happens to be an iterable. We want to do

In [8]: s.index._get_level_number(('A', 'a'))
Out[8]: 0

but there may be other problems later on. You'll probably run into ambiguous situations as well (e.g. the index names are [('A', 'a'), 'A', 'a']. Not sure how best to handle those.