CLN/BUG: indexing fixes (GH5725, GH5727)

doc/source/release.rst

@@ -826,6 +826,7 @@ Bug Fixes
   - Bug in repeated indexing of object with resultant non-unique index (:issue:`5678`)
   - Bug in fillna with Series and a passed series/dict (:issue:`5703`)
   - Bug in groupby transform with a datetime-like grouper (:issue:`5712`)
+  - Bug in multi-index selection in PY3 when using certain keys (:issue:`5725`)
 
 pandas 0.12.0
 -------------

pandas/core/generic.py

@@ -721,13 +721,13 @@ def __setstate__(self, state):
                 # to avoid definitional recursion
                 # e.g. say fill_value needing _data to be
                 # defined
-                for k in self._internal_names:
+                for k in self._internal_names_set:
                     if k in state:
                         v = state[k]
                         object.__setattr__(self, k, v)
 
                 for k, v in state.items():
-                    if k not in self._internal_names:
+                    if k not in self._internal_names_set:
                         object.__setattr__(self, k, v)
 
             else:
@@ -938,15 +938,22 @@ def to_clipboard(self, excel=None, sep=None, **kwargs):
     @classmethod
     def _create_indexer(cls, name, indexer):
         """ create an indexer like _name in the class """
-        iname = '_%s' % name
-        setattr(cls, iname, None)
 
-        def _indexer(self):
-            if getattr(self, iname, None) is None:
-                setattr(self, iname, indexer(self, name))
-            return getattr(self, iname)
+        if getattr(cls, name, None) is None:
+            iname = '_%s' % name
+            setattr(cls, iname, None)
 
-        setattr(cls, name, property(_indexer))
+            def _indexer(self):
+                i = getattr(self, iname)
+                if i is None:
+                    i = indexer(self, name)
+                    setattr(self, iname, i)
+                return i
+
+            setattr(cls, name, property(_indexer))
+
+            # add to our internal names set
+            cls._internal_names_set.add(iname)
 
     def get(self, key, default=None):
         """
@@ -1831,9 +1838,9 @@ def fillna(self, value=None, method=None, axis=0, inplace=False,
             pad / ffill: propagate last valid observation forward to next valid
             backfill / bfill: use NEXT valid observation to fill gap
         value : scalar, dict, or Series
-            Value to use to fill holes (e.g. 0), alternately a dict/Series of 
-            values specifying which value to use for each index (for a Series) or 
-            column (for a DataFrame). (values not in the dict/Series will not be 
+            Value to use to fill holes (e.g. 0), alternately a dict/Series of
+            values specifying which value to use for each index (for a Series) or
+            column (for a DataFrame). (values not in the dict/Series will not be
             filled). This value cannot be a list.
         axis : {0, 1}, default 0
             0: fill column-by-column
@@ -1845,8 +1852,8 @@ def fillna(self, value=None, method=None, axis=0, inplace=False,
         limit : int, default None
             Maximum size gap to forward or backward fill
         downcast : dict, default is None
-             a dict of item->dtype of what to downcast if possible, 
-            or the string 'infer' which will try to downcast to an appropriate 
+             a dict of item->dtype of what to downcast if possible,
+            or the string 'infer' which will try to downcast to an appropriate
             equal type (e.g. float64 to int64 if possible)
 
         See also

pandas/core/index.py

@@ -2300,8 +2300,14 @@ def _try_mi(k):
 
             # a Timestamp will raise a TypeError in a multi-index
             # rather than a KeyError, try it here
+            # note that a string that 'looks' like a Timestamp will raise
+            # a KeyError! (GH5725)
             if isinstance(key, (datetime.datetime, np.datetime64)) or (
                     compat.PY3 and isinstance(key, compat.string_types)):
+                try:
+                    return _try_mi(key)
+                except:
+                    pass
                 try:
                     return _try_mi(Timestamp(key))
                 except:

pandas/tests/test_indexing.py

@@ -3,7 +3,7 @@
 import itertools
 import warnings
 
-from pandas.compat import range, lrange, StringIO, lmap, map
+from pandas.compat import range, lrange, lzip, StringIO, lmap, map
 from numpy import random, nan
 from numpy.random import randn
 import numpy as np
@@ -249,6 +249,15 @@ def _print(result, error = None):
                         k2 = key2
                         _eq(t, o, a, obj, key1, k2)
 
+    def test_indexer_caching(self):
+        # GH5727
+        # make sure that indexers are in the _internal_names_set
+        n = 1000001
+        arrays = [lrange(n), lrange(n)]
+        index = MultiIndex.from_tuples(lzip(*arrays))
+        s = Series(np.zeros(n), index=index)
+        str(s)
+
     def test_at_and_iat_get(self):
 
         def _check(f, func, values = False):
@@ -830,6 +839,20 @@ def test_xs_multiindex(self):
         expected.columns = expected.columns.droplevel('lvl1')
         assert_frame_equal(result, expected)
 
+    def test_getitem_multiindex(self):
+
+        # GH 5725
+        # the 'A' happens to be a valid Timestamp so the doesn't raise the appropriate
+        # error, only in PY3 of course!
+        index = MultiIndex(levels=[['A', 'B', 'C'], [0, 26, 27, 37, 57, 67, 75, 82]],
+                           labels=[[0, 0, 0, 1, 2, 2, 2, 2, 2, 2], [1, 3, 4, 6, 0, 2, 2, 3, 5, 7]],
+                           names=['tag', 'day'])
+        arr = np.random.randn(len(index),1)
+        df = DataFrame(arr,index=index,columns=['val'])
+        result = df.val['A']
+        expected = Series(arr.ravel()[0:3],name='val',index=Index([26,37,57],name='day'))
+        assert_series_equal(result,expected)
+
     def test_setitem_dtype_upcast(self):
 
         # GH3216