make EmbeddedDocument not hashable by default

docs/changelog.rst

@@ -6,6 +6,11 @@ Development
 ===========
 - (Fill this out as you fix issues and develop your features).
 
+Changes in 0.13.0
+=================
+- POTENTIAL BREAKING CHANGE: Added Unicode support to the `EmailField`, see
+  docs/upgrade.rst for details.
+
 Changes in 0.12.0
 =================
 - POTENTIAL BREAKING CHANGE: Fixed limit/skip/hint/batch_size chaining #1476

docs/upgrade.rst

@@ -6,6 +6,17 @@ Development
 ***********
 (Fill this out whenever you introduce breaking changes to MongoEngine)
 
+0.13.0
+******
+This release adds Unicode support to the `EmailField` and changes its
+structure significantly. Previously, email addresses containing Unicode
+characters didn't work at all. Starting with v0.13.0, domains with Unicode
+characters are supported out of the box, meaning some emails that previously
+didn't pass validation now do. Make sure the rest of your application can
+accept such email addresses. Additionally, if you subclassed the `EmailField`
+in your application and overrode `EmailField.EMAIL_REGEX`, you will have to
+adjust your code to override `EmailField.USER_REGEX`, `EmailField.DOMAIN_REGEX`,
+and potentially `EmailField.UTF8_USER_REGEX`.
 
 0.12.0
 ******

mongoengine/__init__.py

@@ -23,7 +23,7 @@ __all__ = (list(document.__all__) + list(fields.__all__) +
            list(signals.__all__) + list(errors.__all__))
 
 
-VERSION = (0, 12, 0)
+VERSION = (0, 13, 0)
 
 
 def get_version():

mongoengine/base/document.py

@@ -272,13 +272,6 @@ class BaseDocument(object):
     def __ne__(self, other):
         return not self.__eq__(other)
 
-    def __hash__(self):
-        if getattr(self, 'pk', None) is None:
-            # For new object
-            return super(BaseDocument, self).__hash__()
-        else:
-            return hash(self.pk)
-
     def clean(self):
         """
         Hook for doing document level data cleaning before validation is run.

mongoengine/document.py

@@ -60,6 +60,12 @@ class EmbeddedDocument(BaseDocument):
     my_metaclass = DocumentMetaclass
     __metaclass__ = DocumentMetaclass
 
+    # A generic embedded document doesn't have any immutable properties
+    # that describe it uniquely, hence it shouldn't be hashable. You can
+    # define your own __hash__ method on a subclass if you need your
+    # embedded documents to be hashable.
+    __hash__ = None
+
     def __init__(self, *args, **kwargs):
         super(EmbeddedDocument, self).__init__(*args, **kwargs)
         self._instance = None
@@ -160,6 +166,15 @@ class Document(BaseDocument):
         """Set the primary key."""
         return setattr(self, self._meta['id_field'], value)
 
+    def __hash__(self):
+        """Return the hash based on the PK of this document. If it's new
+        and doesn't have a PK yet, return the default object hash instead.
+        """
+        if self.pk is None:
+            return super(BaseDocument, self).__hash__()
+        else:
+            return hash(self.pk)
+
     @classmethod
     def _get_db(cls):
         """Some Model using other db_alias"""

tests/document/instance.py

@@ -2164,7 +2164,7 @@ class InstanceTest(unittest.TestCase):
         class BlogPost(Document):
             pass
 
-        # Clear old datas
+        # Clear old data
         User.drop_collection()
         BlogPost.drop_collection()
 
@@ -2176,17 +2176,18 @@ class InstanceTest(unittest.TestCase):
         b1 = BlogPost.objects.create()
         b2 = BlogPost.objects.create()
 
-        # in List
+        # Make sure docs are properly identified in a list (__eq__ is used
+        # for the comparison).
         all_user_list = list(User.objects.all())
-
         self.assertTrue(u1 in all_user_list)
         self.assertTrue(u2 in all_user_list)
         self.assertTrue(u3 in all_user_list)
-        self.assertFalse(u4 in all_user_list)  # New object
+        self.assertTrue(u4 not in all_user_list)  # New object
-        self.assertFalse(b1 in all_user_list)  # Other object
+        self.assertTrue(b1 not in all_user_list)  # Other object
-        self.assertFalse(b2 in all_user_list)  # Other object
+        self.assertTrue(b2 not in all_user_list)  # Other object
 
-        # in Dict
+        # Make sure docs can be used as keys in a dict (__hash__ is used
+        # for hashing the docs).
         all_user_dic = {}
         for u in User.objects.all():
             all_user_dic[u] = "OK"
@@ -2198,9 +2199,20 @@ class InstanceTest(unittest.TestCase):
         self.assertEqual(all_user_dic.get(b1, False), False)  # Other object
         self.assertEqual(all_user_dic.get(b2, False), False)  # Other object
 
-        # in Set
+        # Make sure docs are properly identified in a set (__hash__ is used
+        # for hashing the docs).
         all_user_set = set(User.objects.all())
         self.assertTrue(u1 in all_user_set)
+        self.assertTrue(u4 not in all_user_set)
+        self.assertTrue(b1 not in all_user_list)
+        self.assertTrue(b2 not in all_user_list)
+
+        # Make sure duplicate docs aren't accepted in the set
+        self.assertEqual(len(all_user_set), 3)
+        all_user_set.add(u1)
+        all_user_set.add(u2)
+        all_user_set.add(u3)
+        self.assertEqual(len(all_user_set), 3)
 
     def test_picklable(self):
         pickle_doc = PickleTest(number=1, string="One", lists=['1', '2'])