BUG: Series.rename treats Series as list-like instead of mapping

[TomAugspurger]

#### Code Sample, a copy-pastable example if possible

In [6]: s = pd.Series([0, 1])

In [5]: s.rename(pd.Series({0: 10, 1: 20}))
Out[5]:
0    0
1    1
Name: [10, 20], dtype: int64

Problem is here using an isinstance(,MutableMapping) instead of com.is_dict_like. rename_axis should be fine, but will add tests for both.

[jreback]

Indexes are NOT dict_like though (Series ARE). so something else happening.

In [3]: pandas.core.common.is_dict_like(pd.RangeIndex(10))
Out[3]: False

In [4]: pandas.core.common.is_dict_like(pd.Index([1,2,3]))
Out[4]: False

[TomAugspurger]

Ah.... my bad. I thought I didn't reproduce it with a non-RangeIndexed series, but I must have been mistaken.

In [52]: s.rename(pd.Series({'a': 'A', 'b': 'B'}))
Out[52]:
a    0
b    1
Name: [A, B], dtype: int64

I'll update.

[jreback]

somewhat related to #12610 (IOW setting the .name with a list should just fail), but you don't need to explicity check for this (well, you do but for a different reason, namely that you are doing something different with it)

[TomAugspurger]

Updated, can't believe I missed this. Will have a fix tonight.