[Backport #14767] ERR: raise on python in object hashing, only supporting strings, nulls

xref #14729

Author: Jeff Reback <jeff@reback.net>

Closes #14767 from jreback/hashing_object and squashes the following commits:

9a5a5d4 [Jeff Reback] ERR: raise on python in object hashing, only supporting strings, nulls

(cherry picked from commit de1132d878)

pandas/src/hash.pyx

@@ -7,6 +7,7 @@ cimport numpy as cnp
 import numpy as np
 from numpy cimport ndarray, uint8_t, uint32_t, uint64_t
 
+from util cimport _checknull
 from cpython cimport (PyString_Check,
                       PyBytes_Check,
                       PyUnicode_Check)
@@ -29,6 +30,11 @@ def hash_object_array(ndarray[object] arr, object key, object encoding='utf8'):
     -------
     1-d uint64 ndarray of hashes
 
+    Notes
+    -----
+    allowed values must be strings, or nulls
+    mixed array types will raise TypeError
+
     """
     cdef:
         Py_ssize_t i, l, n
@@ -60,10 +66,14 @@ def hash_object_array(ndarray[object] arr, object key, object encoding='utf8'):
             data = <bytes>val
         elif PyUnicode_Check(val):
             data = <bytes>val.encode(encoding)
-        else:
-            # non-strings
+        elif _checknull(val):
+            # null, stringify and encode
             data = <bytes>str(val).encode(encoding)
 
+        else:
+            raise TypeError("{} of type {} is not a valid type for "
+                            "hashing, must be string or null".format(val, type(val)))
+
         l = len(data)
         lens[i] = l
         cdata = data

pandas/tools/tests/test_hashing.py

@@ -63,6 +63,7 @@ class TestHashing(tm.TestCase):
                     Series([1.0, 1.5, 3.2], index=[1.5, 1.1, 3.3]),
                     Series(['a', 'b', 'c']),
                     Series(['a', np.nan, 'c']),
+                    Series(['a', None, 'c']),
                     Series([True, False, True]),
                     Index([1, 2, 3]),
                     Index([True, False, True]),
@@ -71,9 +72,7 @@ class TestHashing(tm.TestCase):
                     tm.makeMixedDataFrame(),
                     tm.makeTimeDataFrame(),
                     tm.makeTimeSeries(),
-                    tm.makeTimedeltaIndex(),
-                    Series([1, 2, 3], index=pd.MultiIndex.from_tuples(
-                        [('a', 1), ('a', 2), ('b', 1)]))]:
+                    tm.makeTimedeltaIndex()]:
             self.check_equal(obj)
             self.check_not_equal_with_index(obj)
 
@@ -115,16 +114,22 @@ class TestHashing(tm.TestCase):
             hash_pandas_object(Series(list('abc')), hash_key='foo')
         self.assertRaises(ValueError, f)
 
-    def test_mixed(self):
-        # mixed objects
-        obj = Series(['1', 2, 3])
-        self.check_equal(obj)
-        self.check_not_equal_with_index(obj)
+    def test_unsupported_objects(self):
 
-        # mixed are actually equal when stringified
-        a = hash_pandas_object(obj)
-        b = hash_pandas_object(Series(list('123')))
-        self.assert_series_equal(a, b)
+        # mixed objects are not supported
+        obj = Series(['1', 2, 3])
+
+        def f():
+            hash_pandas_object(obj)
+        self.assertRaises(TypeError, f)
+
+        # MultiIndex are represented as tuples
+        obj = Series([1, 2, 3], index=pd.MultiIndex.from_tuples(
+            [('a', 1), ('a', 2), ('b', 1)]))
+
+        def f():
+            hash_pandas_object(obj)
+        self.assertRaises(TypeError, f)
 
     def test_alread_encoded(self):
         # if already encoded then ok