rename and clarify hashing of the indexing key

Author: jreback

pandas/indexes/multi.py

@@ -681,12 +681,6 @@ def is_monotonic(self):
     def is_unique(self):
         return not self.duplicated().any()
 
-    @cache_readonly
-    def _hashed_values(self):
-        """ return a uint64 ndarray of my hashed values """
-        from pandas.tools.hashing import hash_tuples
-        return hash_tuples(self)
-
     @cache_readonly
     def _have_mixed_levels(self):
         """ return a boolean list indicated if we have mixed levels """
@@ -697,15 +691,35 @@ def _inferred_type_levels(self):
         """ return a list of the inferred types, one for each level """
         return [i.inferred_type for i in self.levels]
 
-    def _as_valid_indexing_key(self, key):
+    @cache_readonly
+    def _hashed_values(self):
+        """ return a uint64 ndarray of my hashed values """
+        from pandas.tools.hashing import hash_tuples
+        return hash_tuples(self)
+
+    def _hashed_indexing_key(self, key):
         """
-        validate and return our key
+        validate and return the hash for the provided key
+
+        *this is internal for use for the cython routines*
+
+        Paramters
+        ---------
+        key : string or tuple
+
+        Returns
+        -------
+        np.uint64
+
+        Notes
+        -----
         we need to stringify if we have mixed levels
 
-        this is internal for use for the cython routines
         """
+        from pandas.tools.hashing import hash_tuples
+
         if not isinstance(key, tuple):
-            return key
+            return hash_tuples(key)
 
         if not len(key) == self.nlevels:
             raise KeyError
@@ -714,8 +728,9 @@ def f(k, stringify):
             if stringify and not isinstance(k, compat.string_types):
                 k = str(k)
             return k
-        return tuple([f(k, stringify)
-                      for k, stringify in zip(key, self._have_mixed_levels)])
+        key = tuple([f(k, stringify)
+                     for k, stringify in zip(key, self._have_mixed_levels)])
+        return hash_tuples(key)
 
     @deprecate_kwarg('take_last', 'keep', mapping={True: 'last',
                                                    False: 'first'})

pandas/src/hashtable_class_helper.pxi.in

@@ -864,22 +864,16 @@ cdef class MultiIndexHashTable(HashTable):
             khiter_t k
             uint64_t value
 
-        from pandas.tools.hashing import hash_tuples
-        key = self.mi._as_valid_indexing_key(key)
-        value = hash_tuples(key)
-        k = kh_get_uint64(self.table, key)
+        value = self.mi._hashed_indexing_key(key)
+        k = kh_get_uint64(self.table, value)
         return k != self.table.n_buckets
 
     cpdef get_item(self, object key):
         cdef:
             khiter_t k
             uint64_t value
 
-        from pandas.tools.hashing import hash_tuples
-        key = self.mi._as_valid_indexing_key(key)
-        value = hash_tuples(key)
-
-        # should check nlevels == len tuple
+        value = self.mi._hashed_indexing_key(key)
         k = kh_get_uint64(self.table, value)
         if k != self.table.n_buckets:
             return self.table.vals[k]