No implicit casting of int -> float on full indexing of multiindex

[toobaz]

Code Sample, a copy-pastable example if possible

In [2]: mi = pd.MultiIndex.from_product([[0., 1.], ['begin', 'end']])

In [3]: mi.get_loc(1)
Out[3]: slice(2, 4, None)

In [4]: mi.get_indexer_for([(1, 'end')])
Out[4]: array([3])

In [5]: mi.get_loc((1, 'end'))
---------------------------------------------------------------------------
KeyError                                  Traceback (most recent call last)
<ipython-input-5-eb387fc9b362> in <module>()
----> 1 mi.get_loc((1, 'end'))

/home/pietro/nobackup/repo/pandas/pandas/indexes/multi.py in get_loc(self, key, method)
   1962             key = _values_from_object(key)
   1963             key = tuple(map(_maybe_str_to_time_stamp, key, self.levels))
-> 1964             return self._engine.get_loc(key)
   1965 
   1966         # -- partial selection or non-unique index

/home/pietro/nobackup/repo/pandas/pandas/_libs/index.pyx in pandas._libs.index.MultiIndexEngine.get_loc (pandas/_libs/index.c:13171)()

/home/pietro/nobackup/repo/pandas/pandas/_libs/index.pyx in pandas._libs.index.MultiIndexEngine.get_loc (pandas/_libs/index.c:13018)()

/home/pietro/nobackup/repo/pandas/pandas/_libs/hashtable_class_helper.pxi in pandas._libs.hashtable.MultiIndexHashTable.get_item (pandas/_libs/hashtable.c:23625)()

/home/pietro/nobackup/repo/pandas/pandas/_libs/hashtable_class_helper.pxi in pandas._libs.hashtable.MultiIndexHashTable.get_item (pandas/_libs/hashtable.c:23578)()

KeyError: (1, 'end')

Problem description

We implicitly cast if indexing only one level, and if asking a list of (complete) keys: we should do the same for a single complete key.

Expected Output

In [6]: mi.get_loc((1., 'end'))
Out[6]: 3

Output of pd.show_versions()

INSTALLED VERSIONS

commit: None
python: 3.5.3.final.0
python-bits: 64
OS: Linux
OS-release: 4.7.0-1-amd64
machine: x86_64
processor:
byteorder: little
LC_ALL: None
LANG: it_IT.utf8
LOCALE: it_IT.UTF-8

pandas: 0.19.0+783.gcd35d22a0.dirty
pytest: 3.0.6
pip: 9.0.1
setuptools: 33.1.1
Cython: 0.25.2
numpy: 1.12.0
scipy: 0.18.1
xarray: 0.9.1
IPython: 5.1.0.dev
sphinx: 1.4.9
patsy: 0.3.0-dev
dateutil: 2.5.3
pytz: 2016.7
blosc: None
bottleneck: 1.2.0
tables: 3.3.0
numexpr: 2.6.1
feather: 0.3.1
matplotlib: 2.0.0
openpyxl: 2.3.0
xlrd: 1.0.0
xlwt: 1.1.2
xlsxwriter: 0.9.6
lxml: 3.7.1
bs4: 4.5.3
html5lib: 0.999999999
sqlalchemy: 1.0.15
pymysql: None
psycopg2: None
jinja2: 2.8
s3fs: None
pandas_gbq: None
pandas_datareader: 0.2.1

[toobaz]

Forgot to mention: this emerged in some code which ran fine some months ago, so I think it is a regression (where "I think" means "I'm not 100% sure the difference in behaviour comes from the portion of my code affected by this bug").

[jorisvandenbossche]

In any case, the isolated example you show is clearly a regression:

In [5]: pd.__version__
Out[5]: '0.19.2'

In [6]: mi = pd.MultiIndex.from_product([[0., 1.], ['begin', 'end']])

In [7]: mi.get_loc((1, 'end'))
Out[7]: 3

In [19]: pd.__version__
Out[19]: '0.19.0+783.g7453b63'

In [20]: mi = pd.MultiIndex.from_product([[0., 1.], ['begin', 'end']])

In [21]: mi.get_loc((1, 'end'))
...
KeyError: (1, 'end')

[toobaz]

In any case, the isolated example you show is clearly a regression:

Indeed, git bisect points at e351ed0