Add continue_on_error optional kwarg to QuerySet.insert

README.rst

@@ -57,7 +57,7 @@ Some simple examples of what MongoEngine code looks like:
 
     class BlogPost(Document):
         title = StringField(required=True, max_length=200)
-        posted = DateTimeField(default=datetime.datetime.now)
+        posted = DateTimeField(default=datetime.datetime.utcnow)
         tags = ListField(StringField(max_length=50))
         meta = {'allow_inheritance': True}
 
@@ -87,17 +87,18 @@ Some simple examples of what MongoEngine code looks like:
     ...     print
     ...
 
-    >>> len(BlogPost.objects)
+    # Count all blog posts and its subtypes
+    >>> BlogPost.objects.count()
     2
-    >>> len(TextPost.objects)
+    >>> TextPost.objects.count()
     1
-    >>> len(LinkPost.objects)
+    >>> LinkPost.objects.count()
     1
 
-    # Find tagged posts
-    >>> len(BlogPost.objects(tags='mongoengine'))
+    # Count tagged posts
+    >>> BlogPost.objects(tags='mongoengine').count()
     2
-    >>> len(BlogPost.objects(tags='mongodb'))
+    >>> BlogPost.objects(tags='mongodb').count()
     1
 
 Tests
@@ -130,8 +131,7 @@ Community
   <http://groups.google.com/group/mongoengine-users>`_
 - `MongoEngine Developers mailing list
   <http://groups.google.com/group/mongoengine-dev>`_
-- `#mongoengine IRC channel <http://webchat.freenode.net/?channels=mongoengine>`_
 
 Contributing
 ============
-We welcome contributions! see  the `Contribution guidelines <https://github.com/MongoEngine/mongoengine/blob/master/CONTRIBUTING.rst>`_
+We welcome contributions! See the `Contribution guidelines <https://github.com/MongoEngine/mongoengine/blob/master/CONTRIBUTING.rst>`_

docs/changelog.rst

@@ -13,6 +13,7 @@ Changes in 0.11.0
 - BREAKING CHANGE: Renamed `ConnectionError` to `MongoEngineConnectionError` since the former is a built-in exception name in Python v3.x. #1428
 - BREAKING CHANGE: Dropped Python 2.6 support. #1428
 - BREAKING CHANGE: `from mongoengine.base import ErrorClass` won't work anymore for any error from `mongoengine.errors` (e.g. `ValidationError`). Use `from mongoengine.errors import ErrorClass instead`. #1428
+- BREAKING CHANGE: Accessing a broken reference will raise a `DoesNotExist` error. In the past it used to return `None`. #1334
 - Fixed absent rounding for DecimalField when `force_string` is set. #1103
 
 Changes in 0.10.8

docs/guide/connecting.rst

@@ -33,7 +33,7 @@ the :attr:`host` to
     corresponding parameters in :func:`~mongoengine.connect`: ::
 
         connect(
-            name='test',
+            db='test',
             username='user',
             password='12345',
             host='mongodb://admin:qwerty@localhost/production'

mongoengine/base/common.py

@@ -5,7 +5,7 @@ __all__ = ('UPDATE_OPERATORS', 'get_document', '_document_registry')
 
 UPDATE_OPERATORS = set(['set', 'unset', 'inc', 'dec', 'pop', 'push',
                         'push_all', 'pull', 'pull_all', 'add_to_set',
-                        'set_on_insert', 'min', 'max'])
+                        'set_on_insert', 'min', 'max', 'rename'])
 
 
 _document_registry = {}

mongoengine/base/datastructures.py

@@ -429,7 +429,7 @@ class StrictDict(object):
     def __eq__(self, other):
         return self.items() == other.items()
 
-    def __neq__(self, other):
+    def __ne__(self, other):
         return self.items() != other.items()
 
     @classmethod

mongoengine/base/document.py

@@ -402,9 +402,11 @@ class BaseDocument(object):
             raise ValidationError(message, errors=errors)
 
     def to_json(self, *args, **kwargs):
-        """Converts a document to JSON.
-        :param use_db_field: Set to True by default but enables the output of the json structure with the field names
-            and not the mongodb store db_names in case of set to False
+        """Convert this document to JSON.
+
+        :param use_db_field: Serialize field names as they appear in
+            MongoDB (as opposed to attribute names on this document).
+            Defaults to True.
         """
         use_db_field = kwargs.pop('use_db_field', True)
         return json_util.dumps(self.to_mongo(use_db_field), *args, **kwargs)

mongoengine/base/fields.py

@@ -41,7 +41,7 @@ class BaseField(object):
         """
         :param db_field: The database field to store this field in
             (defaults to the name of the field)
-        :param name: Depreciated - use db_field
+        :param name: Deprecated - use db_field
         :param required: If the field is required. Whether it has to have a
             value or not. Defaults to False.
         :param default: (optional) The default value for this field if no value
@@ -81,6 +81,17 @@ class BaseField(object):
         self.sparse = sparse
         self._owner_document = None
 
+        # Validate the db_field
+        if isinstance(self.db_field, six.string_types) and (
+            '.' in self.db_field or
+            '\0' in self.db_field or
+            self.db_field.startswith('$')
+        ):
+            raise ValueError(
+                'field names cannot contain dots (".") or null characters '
+                '("\\0"), and they must not start with a dollar sign ("$").'
+            )
+
         # Detect and report conflicts between metadata and base properties.
         conflicts = set(dir(self)) & set(kwargs)
         if conflicts:

mongoengine/document.py

@@ -332,68 +332,20 @@ class Document(BaseDocument):
         signals.pre_save_post_validation.send(self.__class__, document=self,
                                               created=created, **signal_kwargs)
 
-        try:
-            collection = self._get_collection()
         if self._meta.get('auto_create_index', True):
             self.ensure_indexes()
+
+        try:
+            # Save a new document or update an existing one
             if created:
-                if force_insert:
-                    object_id = collection.insert(doc, **write_concern)
+                object_id = self._save_create(doc, force_insert, write_concern)
             else:
-                    object_id = collection.save(doc, **write_concern)
-                    # In PyMongo 3.0, the save() call calls internally the _update() call
-                    # but they forget to return the _id value passed back, therefore getting it back here
-                    # Correct behaviour in 2.X and in 3.0.1+ versions
-                    if not object_id and pymongo.version_tuple == (3, 0):
-                        pk_as_mongo_obj = self._fields.get(self._meta['id_field']).to_mongo(self.pk)
-                        object_id = (
-                            self._qs.filter(pk=pk_as_mongo_obj).first() and
-                            self._qs.filter(pk=pk_as_mongo_obj).first().pk
-                        )  # TODO doesn't this make 2 queries?
-            else:
-                object_id = doc['_id']
-                updates, removals = self._delta()
-                # Need to add shard key to query, or you get an error
-                if save_condition is not None:
-                    select_dict = transform.query(self.__class__,
-                                                  **save_condition)
-                else:
-                    select_dict = {}
-                select_dict['_id'] = object_id
-                shard_key = self._meta.get('shard_key', tuple())
-                for k in shard_key:
-                    path = self._lookup_field(k.split('.'))
-                    actual_key = [p.db_field for p in path]
-                    val = doc
-                    for ak in actual_key:
-                        val = val[ak]
-                    select_dict['.'.join(actual_key)] = val
-
-                def is_new_object(last_error):
-                    if last_error is not None:
-                        updated = last_error.get('updatedExisting')
-                        if updated is not None:
-                            return not updated
-                    return created
-
-                update_query = {}
-
-                if updates:
-                    update_query['$set'] = updates
-                if removals:
-                    update_query['$unset'] = removals
-                if updates or removals:
-                    upsert = save_condition is None
-                    last_error = collection.update(select_dict, update_query,
-                                                   upsert=upsert, **write_concern)
-                    if not upsert and last_error['n'] == 0:
-                        raise SaveConditionError('Race condition preventing'
-                                                 ' document update detected')
-                    created = is_new_object(last_error)
+                object_id, created = self._save_update(doc, save_condition,
+                                                       write_concern)
 
             if cascade is None:
-                cascade = self._meta.get(
-                    'cascade', False) or cascade_kwargs is not None
+                cascade = (self._meta.get('cascade', False) or
+                           cascade_kwargs is not None)
 
             if cascade:
                 kwargs = {
@@ -406,6 +358,7 @@ class Document(BaseDocument):
                     kwargs.update(cascade_kwargs)
                 kwargs['_refs'] = _refs
                 self.cascade_save(**kwargs)
+
         except pymongo.errors.DuplicateKeyError as err:
             message = u'Tried to save duplicate unique keys (%s)'
             raise NotUniqueError(message % six.text_type(err))
@@ -418,16 +371,91 @@ class Document(BaseDocument):
                 raise NotUniqueError(message % six.text_type(err))
             raise OperationError(message % six.text_type(err))
 
+        # Make sure we store the PK on this document now that it's saved
         id_field = self._meta['id_field']
         if created or id_field not in self._meta.get('shard_key', []):
             self[id_field] = self._fields[id_field].to_python(object_id)
 
         signals.post_save.send(self.__class__, document=self,
                                created=created, **signal_kwargs)
+
         self._clear_changed_fields()
         self._created = False
+
         return self
 
+    def _save_create(self, doc, force_insert, write_concern):
+        """Save a new document.
+
+        Helper method, should only be used inside save().
+        """
+        collection = self._get_collection()
+
+        if force_insert:
+            return collection.insert(doc, **write_concern)
+
+        object_id = collection.save(doc, **write_concern)
+
+        # In PyMongo 3.0, the save() call calls internally the _update() call
+        # but they forget to return the _id value passed back, therefore getting it back here
+        # Correct behaviour in 2.X and in 3.0.1+ versions
+        if not object_id and pymongo.version_tuple == (3, 0):
+            pk_as_mongo_obj = self._fields.get(self._meta['id_field']).to_mongo(self.pk)
+            object_id = (
+                self._qs.filter(pk=pk_as_mongo_obj).first() and
+                self._qs.filter(pk=pk_as_mongo_obj).first().pk
+            )  # TODO doesn't this make 2 queries?
+
+        return object_id
+
+    def _save_update(self, doc, save_condition, write_concern):
+        """Update an existing document.
+
+        Helper method, should only be used inside save().
+        """
+        collection = self._get_collection()
+        object_id = doc['_id']
+        created = False
+
+        select_dict = {}
+        if save_condition is not None:
+            select_dict = transform.query(self.__class__, **save_condition)
+
+        select_dict['_id'] = object_id
+
+        # Need to add shard key to query, or you get an error
+        shard_key = self._meta.get('shard_key', tuple())
+        for k in shard_key:
+            path = self._lookup_field(k.split('.'))
+            actual_key = [p.db_field for p in path]
+            val = doc
+            for ak in actual_key:
+                val = val[ak]
+            select_dict['.'.join(actual_key)] = val
+
+        updates, removals = self._delta()
+        update_query = {}
+        if updates:
+            update_query['$set'] = updates
+        if removals:
+            update_query['$unset'] = removals
+        if updates or removals:
+            upsert = save_condition is None
+            last_error = collection.update(select_dict, update_query,
+                                           upsert=upsert, **write_concern)
+            if not upsert and last_error['n'] == 0:
+                raise SaveConditionError('Race condition preventing'
+                                         ' document update detected')
+            if last_error is not None:
+                updated_existing = last_error.get('updatedExisting')
+                if updated_existing is False:
+                    created = True
+                    # !!! This is bad, means we accidentally created a new,
+                    # potentially corrupted document. See
+                    # https://github.com/MongoEngine/mongoengine/issues/564
+
+        return object_id, created
+
     def cascade_save(self, **kwargs):
         """Recursively save any references and generic references on the
         document.

mongoengine/errors.py

@@ -50,8 +50,8 @@ class FieldDoesNotExist(Exception):
     or an :class:`~mongoengine.EmbeddedDocument`.
 
     To avoid this behavior on data loading,
-    you should the :attr:`strict` to ``False``
-    in the :attr:`meta` dictionnary.
+    you should set the :attr:`strict` to ``False``
+    in the :attr:`meta` dictionary.
     """
 
 

mongoengine/queryset/base.py

@@ -296,22 +296,25 @@ class BaseQuerySet(object):
             result = None
         return result
 
-    def insert(self, doc_or_docs, load_bulk=True,
-               write_concern=None, signal_kwargs=None):
+    def insert(self, doc_or_docs, load_bulk=True, write_concern=None,
+               signal_kwargs=None, continue_on_error=None):
         """bulk insert documents
 
         :param doc_or_docs: a document or list of documents to be inserted
         :param load_bulk (optional): If True returns the list of document
             instances
-        :param write_concern: Extra keyword arguments are passed down to
-                :meth:`~pymongo.collection.Collection.insert`
-                which will be used as options for the resultant
-                ``getLastError`` command.  For example,
-                ``insert(..., {w: 2, fsync: True})`` will wait until at least
-                two servers have recorded the write and will force an fsync on
-                each server being written to.
+        :param write_concern: Optional keyword argument passed down to
+                :meth:`~pymongo.collection.Collection.insert`, representing
+                the write concern. For example,
+                ``insert(..., write_concert={w: 2, fsync: True})`` will
+                wait until at least two servers have recorded the write
+                and will force an fsync on each server being written to.
         :parm signal_kwargs: (optional) kwargs dictionary to be passed to
             the signal calls.
+        :param continue_on_error: Optional keyword argument passed down to
+                :meth:`~pymongo.collection.Collection.insert`. Defines what
+                to do when a document cannot be inserted (e.g. due to
+                duplicate IDs). Read PyMongo's docs for more info.
 
         By default returns document instances, set ``load_bulk`` to False to
         return just ``ObjectIds``
@@ -322,12 +325,10 @@ class BaseQuerySet(object):
         """
         Document = _import_class('Document')
 
-        if write_concern is None:
-            write_concern = {}
-
+        # Determine if we're inserting one doc or more
         docs = doc_or_docs
         return_one = False
-        if isinstance(docs, Document) or issubclass(docs.__class__, Document):
+        if isinstance(docs, Document):
             return_one = True
             docs = [docs]
 
@@ -344,9 +345,16 @@ class BaseQuerySet(object):
         signals.pre_bulk_insert.send(self._document,
                                      documents=docs, **signal_kwargs)
 
+        # Resolve optional insert kwargs
+        insert_kwargs = {}
+        if write_concern is not None:
+            insert_kwargs.update(write_concern)
+        if continue_on_error is not None:
+            insert_kwargs['continue_on_error'] = continue_on_error
+
         raw = [doc.to_mongo() for doc in docs]
         try:
-            ids = self._collection.insert(raw, **write_concern)
+            ids = self._collection.insert(raw, **insert_kwargs)
         except pymongo.errors.DuplicateKeyError as err:
             message = 'Could not save document (%s)'
             raise NotUniqueError(message % six.text_type(err))
@@ -933,7 +941,20 @@ class BaseQuerySet(object):
             key = '.'.join(parts)
             cleaned_fields.append((key, value))
 
-        fields = sorted(cleaned_fields, key=operator.itemgetter(1))
+        # Sort fields by their values, explicitly excluded fields first, then
+        # explicitly included, and then more complicated operators such as
+        # $slice.
+        def _sort_key(field_tuple):
+            key, value = field_tuple
+            if isinstance(value, (int)):
+                return value  # 0 for exclusion, 1 for inclusion
+            else:
+                return 2  # so that complex values appear last
+
+        fields = sorted(cleaned_fields, key=_sort_key)
+
+        # Clone the queryset, group all fields by their value, convert
+        # each of them to db_fields, and set the queryset's _loaded_fields
         queryset = self.clone()
         for value, group in itertools.groupby(fields, lambda x: x[1]):
             fields = [field for field, value in group]

tests/document/instance.py

@@ -1232,6 +1232,19 @@ class InstanceTest(unittest.TestCase):
         self.assertEqual(person.name, None)
         self.assertEqual(person.age, None)
 
+    def test_update_rename_operator(self):
+        """Test the $rename operator."""
+        coll = self.Person._get_collection()
+        doc = self.Person(name='John').save()
+        raw_doc = coll.find_one({'_id': doc.pk})
+        self.assertEqual(set(raw_doc.keys()), set(['_id', '_cls', 'name']))
+
+        doc.update(rename__name='first_name')
+        raw_doc = coll.find_one({'_id': doc.pk})
+        self.assertEqual(set(raw_doc.keys()),
+                         set(['_id', '_cls', 'first_name']))
+        self.assertEqual(raw_doc['first_name'], 'John')
+
     def test_inserts_if_you_set_the_pk(self):
         p1 = self.Person(name='p1', id=bson.ObjectId()).save()
         p2 = self.Person(name='p2')

tests/fields/fields.py

@@ -306,6 +306,24 @@ class FieldTest(unittest.TestCase):
         person.id = '497ce96f395f2f052a494fd4'
         person.validate()
 
+    def test_db_field_validation(self):
+        """Ensure that db_field doesn't accept invalid values."""
+
+        # dot in the name
+        with self.assertRaises(ValueError):
+            class User(Document):
+                name = StringField(db_field='user.name')
+
+        # name starting with $
+        with self.assertRaises(ValueError):
+            class User(Document):
+                name = StringField(db_field='$name')
+
+        # name containing a null character
+        with self.assertRaises(ValueError):
+            class User(Document):
+                name = StringField(db_field='name\0')
+
     def test_string_validation(self):
         """Ensure that invalid values cannot be assigned to string fields.
         """
@@ -1967,7 +1985,7 @@ class FieldTest(unittest.TestCase):
         self.assertEqual(content, User.objects.first().groups[0].content)
 
     def test_reference_miss(self):
-        """Ensure an exception is raised when dereferencing unknow document
+        """Ensure an exception is raised when dereferencing unknown document
         """
 
         class Foo(Document):
@@ -3973,30 +3991,25 @@ class FieldTest(unittest.TestCase):
         """Tests if a `FieldDoesNotExist` exception is raised when trying to
         instanciate a document with a field that's not defined.
         """
-
         class Doc(Document):
-            foo = StringField(db_field='f')
+            foo = StringField()
 
-        def test():
+        with self.assertRaises(FieldDoesNotExist):
             Doc(bar='test')
 
-        self.assertRaises(FieldDoesNotExist, test)
 
     def test_undefined_field_exception_with_strict(self):
         """Tests if a `FieldDoesNotExist` exception is raised when trying to
         instanciate a document with a field that's not defined,
         even when strict is set to False.
         """
-
         class Doc(Document):
-            foo = StringField(db_field='f')
+            foo = StringField()
             meta = {'strict': False}
 
-        def test():
+        with self.assertRaises(FieldDoesNotExist):
             Doc(bar='test')
 
-        self.assertRaises(FieldDoesNotExist, test)
-
     def test_long_field_is_considered_as_int64(self):
         """
         Tests that long fields are stored as long in mongo, even if long value

tests/queryset/field_list.py

@@ -141,6 +141,16 @@ class OnlyExcludeAllTest(unittest.TestCase):
         self.assertEqual(qs._loaded_fields.as_dict(),
                          {'b': {'$slice': 5}})
 
+    def test_mix_slice_with_other_fields(self):
+        class MyDoc(Document):
+            a = ListField()
+            b = ListField()
+            c = ListField()
+
+        qs = MyDoc.objects.fields(a=1, b=0, slice__c=2)
+        self.assertEqual(qs._loaded_fields.as_dict(),
+                         {'c': {'$slice': 2}, 'a': 1})
+
     def test_only(self):
         """Ensure that QuerySet.only only returns the requested fields.
         """

tests/queryset/queryset.py

@@ -766,8 +766,7 @@ class QuerySetTest(unittest.TestCase):
         self.assertEqual(record.embed.field, 2)
 
     def test_bulk_insert(self):
-        """Ensure that bulk insert works
-        """
+        """Ensure that bulk insert works."""
 
         class Comment(EmbeddedDocument):
             name = StringField()
@@ -885,9 +884,37 @@ class QuerySetTest(unittest.TestCase):
 
         self.assertEqual(Blog.objects.count(), 2)
 
-        Blog.objects.insert([blog2, blog3],
-                            write_concern={"w": 0, 'continue_on_error': True})
-        self.assertEqual(Blog.objects.count(), 3)
+    def test_bulk_insert_continue_on_error(self):
+        """Ensure that bulk insert works with the continue_on_error option."""
+
+        class Person(Document):
+            email = EmailField(unique=True)
+
+        Person.drop_collection()
+
+        Person.objects.insert([
+            Person(email='alice@example.com'),
+            Person(email='bob@example.com')
+        ])
+        self.assertEqual(Person.objects.count(), 2)
+
+        new_docs = [
+            Person(email='alice@example.com'),  # dupe
+            Person(email='bob@example.com'),  # dupe
+            Person(email='steve@example.com')  # new one
+        ]
+
+        # By default inserting dupe docs should fail and no new docs should
+        # be inserted.
+        with self.assertRaises(NotUniqueError):
+            Person.objects.insert(new_docs)
+        self.assertEqual(Person.objects.count(), 2)
+
+        # With continue_on_error, new doc should be inserted, even though we
+        # still get a NotUniqueError caused by the other 2 dupes.
+        with self.assertRaises(NotUniqueError):
+            Person.objects.insert(new_docs, continue_on_error=True)
+        self.assertEqual(Person.objects.count(), 3)
 
     def test_get_changed_fields_query_count(self):
 

tests/test_connection.py

@@ -296,6 +296,19 @@ class ConnectionTest(unittest.TestCase):
         conn = get_connection('t2')
         self.assertFalse(get_tz_awareness(conn))
 
+    def test_write_concern(self):
+        """Ensure write concern can be specified in connect() via
+        a kwarg or as part of the connection URI.
+        """
+        conn1 = connect(alias='conn1', host='mongodb://localhost/testing?w=1&j=true')
+        conn2 = connect('testing', alias='conn2', w=1, j=True)
+        if IS_PYMONGO_3:
+            self.assertEqual(conn1.write_concern.document, {'w': 1, 'j': True})
+            self.assertEqual(conn2.write_concern.document, {'w': 1, 'j': True})
+        else:
+            self.assertEqual(dict(conn1.write_concern), {'w': 1, 'j': True})
+            self.assertEqual(dict(conn2.write_concern), {'w': 1, 'j': True})
+
     def test_datetime(self):
         connect('mongoenginetest', tz_aware=True)
         d = datetime.datetime(2010, 5, 5, tzinfo=utc)