PYTHON-977 - Fix __hash__ method on BSON types that inherit from Python builtin types.

In Python 2, objects automatically inherit the __hash__ of their parent class. In Python 3, objects that override __eq__ do not automatically inherit __hash__, so these objects were not hashable under Python 3. Additionally, mutable BSON types and types that overide __eq__ but did not explicitly define __hash__ had broken __hash__ methods under Python 2. This commit unifies the hashing behavior between Python versions and fixes the __hash__ methods such that two BSON objects hash the same only if they are equal. N.B.: bson.code.Code and bson.regex.Regex are no longer hashable under Python 2 because they are mutable.

Author: Luke Lovett

bson/binary.py

@@ -171,6 +171,9 @@ def __eq__(self, other):
  # subclass of str...
  return False
 
+ def __hash__(self):
+ return super(Binary, self).__hash__() ^ hash(self.__subtype)
+
  def __ne__(self, other):
  return not self == other
 

bson/code.py

@@ -77,5 +77,7 @@ def __eq__(self, other):
  return (self.__scope, str(self)) == (other.__scope, str(other))
  return False
 
+ __hash__ = None
+
  def __ne__(self, other):
  return not self == other

bson/max_key.py

@@ -28,6 +28,9 @@ class MaxKey(object):
  def __eq__(self, other):
  return isinstance(other, MaxKey)
 
+ def __hash__(self):
+ return hash(self._type_marker)
+
  def __ne__(self, other):
  return not self == other
 

bson/min_key.py

@@ -28,6 +28,9 @@ class MinKey(object):
  def __eq__(self, other):
  return isinstance(other, MinKey)
 
+ def __hash__(self):
+ return hash(self._type_marker)
+
  def __ne__(self, other):
  return not self == other
 

bson/regex.py

@@ -104,6 +104,8 @@ def __eq__(self, other):
  else:
  return NotImplemented
 
+ __hash__ = None
+
  def __ne__(self, other):
  return not self == other
 

bson/timestamp.py

@@ -23,6 +23,7 @@
 
 UPPERBOUND = 4294967296
 
+
 class Timestamp(object):
  """MongoDB internal timestamps used in the opLog.
  """
@@ -81,6 +82,9 @@ def __eq__(self, other):
  else:
  return NotImplemented
 
+ def __hash__(self):
+ return hash(self.time) ^ hash(self.inc)
+
  def __ne__(self, other):
  return not self == other
 

test/test_binary.py

@@ -129,6 +129,13 @@ def test_repr(self):
  self.assertEqual(repr(five),
  "Binary(%s, 100)" % (repr(b"test"),))
 
+ def test_hash(self):
+ one = Binary(b"hello world")
+ two = Binary(b"hello world", 42)
+ self.assertEqual(hash(Binary(b"hello world")), hash(one))
+ self.assertNotEqual(hash(one), hash(two))
+ self.assertEqual(hash(Binary(b"hello world", 42)), hash(two))
+
  def test_legacy_java_uuid(self):
  # Test decoding
  data = self.java_data

test/test_bson.py

@@ -759,6 +759,9 @@ def test_regex_from_native(self):
  unicode_regex = re.compile('', re.U)
  self.assertEqual(re.U, Regex.from_native(unicode_regex).flags)
 
+ def test_regex_hash(self):
+ self.assertRaises(TypeError, hash, Regex('hello'))
+
  def test_exception_wrapping(self):
  # No matter what exception is raised while trying to decode BSON,
  # the final exception always matches InvalidBSON.
@@ -815,6 +818,11 @@ def test_minkey_maxkey_comparison(self):
  self.assertTrue(MaxKey() != MinKey())
  self.assertFalse(MaxKey() == MinKey())
 
+ def test_minkey_maxkey_hash(self):
+ self.assertEqual(hash(MaxKey()), hash(MaxKey()))
+ self.assertEqual(hash(MinKey()), hash(MinKey()))
+ self.assertNotEqual(hash(MaxKey()), hash(MinKey()))
+
  def test_timestamp_comparison(self):
  # Timestamp is initialized with time, inc. Time is the more
  # significant comparand.

test/test_code.py

@@ -77,6 +77,9 @@ def test_equality(self):
  self.assertFalse(b != Code("hello"))
  self.assertFalse(b != Code("hello", {}))
 
+ def test_hash(self):
+ self.assertRaises(TypeError, hash, Code("hello world"))
+
  def test_scope_preserved(self):
  a = Code("hello")
  b = Code("hello", {"foo": 5})

test/test_timestamp.py

@@ -69,6 +69,11 @@ def test_equality(self):
  # Explicitly test inequality
  self.assertFalse(t != Timestamp(1, 1))
 
+ def test_hash(self):
+ self.assertEqual(hash(Timestamp(1, 2)), hash(Timestamp(1, 2)))
+ self.assertNotEqual(hash(Timestamp(1, 2)), hash(Timestamp(1, 3)))
+ self.assertNotEqual(hash(Timestamp(1, 2)), hash(Timestamp(2, 2)))
+
  def test_repr(self):
  t = Timestamp(0, 0)
  self.assertEqual(repr(t), "Timestamp(0, 0)")