Merge branch 'dev'

Ensures defined documents are valid and users don't override core methods
by accident.

fixes #275

.gitignore

@@ -1,3 +1,5 @@
+.*
+!.gitignore
 *.pyc
 .*.swp
 *.egg
@@ -6,4 +8,8 @@ docs/_build
 build/
 dist/
 mongoengine.egg-info/
-env/
+env/
+.settings
+.project
+.pydevproject
+tests/bugfix.py

AUTHORS

@@ -1,5 +1,69 @@
+The PRIMARY AUTHORS are (and/or have been):
+
 Harry Marr <harry@hmarr.com>
 Matt Dennewitz <mattdennewitz@gmail.com>
 Deepak Thukral <iapain@yahoo.com>
 Florian Schlachter <flori@n-schlachter.de>
 Steve Challis <steve@stevechallis.com>
+Ross Lawley <ross.lawley@gmail.com>
+Wilson Júnior <wilsonpjunior@gmail.com>
+Dan Crosta https://github.com/dcrosta
+
+CONTRIBUTORS
+
+Dervived from the git logs, inevitably incomplete but all of whom and others
+have submitted patches, reported bugs and generally helped make MongoEngine
+that much better:
+
+ * Harry Marr
+ * Ross Lawley
+ * blackbrrr
+ * Florian Schlachter
+ * Vincent Driessen
+ * Steve Challis
+ * flosch
+ * Deepak Thukral
+ * Colin Howe
+ * Wilson Júnior
+ * Alistair Roche
+ * Dan Crosta
+ * Viktor Kerkez
+ * Stephan Jaekel
+ * Rached Ben Mustapha
+ * Greg Turner
+ * Daniel Hasselrot
+ * Mircea Pasoi
+ * Matt Chisholm
+ * James Punteney
+ * TimothéePeignier
+ * Stuart Rackham
+ * Serge Matveenko
+ * Matt Dennewitz
+ * Don Spaulding
+ * Ales Zoulek
+ * sshwsfc
+ * sib
+ * Samuel Clay
+ * Nick Vlku
+ * martin
+ * Flavio Amieiro
+ * Анхбаяр Лхагвадорж
+ * Zak Johnson
+ * Victor Farazdagi
+ * vandersonmota
+ * Theo Julienne
+ * sp
+ * Slavi Pantaleev
+ * Richard Henry
+ * Nicolas Perriault
+ * Nick Vlku Jr
+ * Michael Henson
+ * Leo Honkanen
+ * kuno
+ * Josh Ourisman
+ * Jaime
+ * Igor Ivanov
+ * Gregg Lind
+ * Gareth Lloyd
+ * Albert Choi
+ * John Arnfield

docs/apireference.rst

@@ -41,6 +41,8 @@ Fields
 
 .. autoclass:: mongoengine.URLField
 
+.. autoclass:: mongoengine.EmailField
+
 .. autoclass:: mongoengine.IntField
 
 .. autoclass:: mongoengine.FloatField
@@ -51,12 +53,16 @@ Fields
 
 .. autoclass:: mongoengine.DateTimeField
 
+.. autoclass:: mongoengine.ComplexDateTimeField
+
 .. autoclass:: mongoengine.EmbeddedDocumentField
 
 .. autoclass:: mongoengine.DictField
 
 .. autoclass:: mongoengine.ListField
 
+.. autoclass:: mongoengine.SortedListField
+
 .. autoclass:: mongoengine.BinaryField
 
 .. autoclass:: mongoengine.ObjectIdField

docs/changelog.rst

@@ -2,6 +2,72 @@
 Changelog
 =========
 
+Changes in dev
+==============
+
+- Added InvalidDocumentError - so Document core methods can't be overwritten
+- Added GenericEmbeddedDocument - so you can embed any type of embeddable document
+- Added within_polygon support - for those with mongodb 1.9
+- Updated sum / average to use map_reduce as db.eval doesn't work in sharded environments
+- Added where() - filter to allowing users to specify query expressions as Javascript
+- Added SequenceField - for creating sequential counters
+- Added update() convenience method to a document
+- Added cascading saves - so changes to Referenced documents are saved on .save()
+- Added select_related() support
+- Added support for the positional operator
+- Updated geo index checking to be recursive and check in embedded documents
+- Updated default collection naming convention
+- Added Document Mixin support
+- Fixed queryet __repr__ mid iteration
+- Added hint() support, so cantell Mongo the proper index to use for the query
+- Fixed issue with inconsitent setting of _cls breaking inherited referencing
+- Added help_text and verbose_name to fields to help with some form libs
+- Updated item_frequencies to handle embedded document lookups
+- Added delta tracking now only sets / unsets explicitly changed fields
+- Fixed saving so sets updated values rather than overwrites
+- Added ComplexDateTimeField - Handles datetimes correctly with microseconds
+- Added ComplexBaseField - for improved flexibility and performance
+- Added get_FIELD_display() method for easy choice field displaying
+- Added queryset.slave_okay(enabled) method
+- Updated queryset.timeout(enabled) and queryset.snapshot(enabled) to be chainable
+- Added insert method for bulk inserts
+- Added blinker signal support
+- Added query_counter context manager for tests
+- Added map_reduce method item_frequencies and set as default (as db.eval doesn't work in sharded environments)
+- Added inline_map_reduce option to map_reduce
+- Updated connection exception so it provides more info on the cause.
+- Added searching multiple levels deep in ``DictField``
+- Added ``DictField`` entries containing strings to use matching operators
+- Added ``MapField``, similar to ``DictField``
+- Added Abstract Base Classes
+- Added Custom Objects Managers
+- Added sliced subfields updating
+- Added ``NotRegistered`` exception if dereferencing ``Document`` not in the registry
+- Added a write concern for ``save``, ``update``, ``update_one`` and ``get_or_create``
+- Added slicing / subarray fetching controls
+- Fixed various unique index and other index issues
+- Fixed threaded connection issues
+- Added spherical geospatial query operators
+- Updated queryset to handle latest version of pymongo
+  map_reduce now requires an output.
+- Added ``Document`` __hash__, __ne__ for pickling
+- Added ``FileField`` optional size arg for read method
+- Fixed ``FileField`` seek and tell methods for reading files
+- Added ``QuerySet.clone`` to support copying querysets
+- Fixed item_frequencies when using name thats the same as a native js function
+- Added reverse delete rules
+- Fixed issue with unset operation
+- Fixed Q-object bug
+- Added ``QuerySet.all_fields`` resets previous .only() and .exclude()
+- Added ``QuerySet.exclude``
+- Added django style choices
+- Fixed order and filter issue
+- Added ``QuerySet.only`` subfield support
+- Added creation_counter to ``BaseField`` allowing fields to be sorted in the
+  way the user has specified them
+- Fixed various errors
+- Added many tests
+
 Changes in v0.4
 ===============
 - Added ``GridFSStorage`` Django storage backend
@@ -32,7 +98,7 @@ Changes in v0.3
 ===============
 - Added MapReduce support
 - Added ``contains``, ``startswith`` and ``endswith`` query operators (and
-  case-insensitive versions that are prefixed with 'i') 
+  case-insensitive versions that are prefixed with 'i')
 - Deprecated fields' ``name`` parameter, replaced with ``db_field``
 - Added ``QuerySet.only`` for only retrieving specific fields
 - Added ``QuerySet.in_bulk()`` for bulk querying using ids
@@ -79,7 +145,7 @@ Changes in v0.2
 ===============
 - Added ``Q`` class for building advanced queries
 - Added ``QuerySet`` methods for atomic updates to documents
-- Fields may now specify ``unique=True`` to enforce uniqueness across a 
+- Fields may now specify ``unique=True`` to enforce uniqueness across a
   collection
 - Added option for default document ordering
 - Fixed bug in index definitions
@@ -87,7 +153,7 @@ Changes in v0.2
 Changes in v0.1.3
 =================
 - Added Django authentication backend
-- Added ``Document.meta`` support for indexes, which are ensured just before 
+- Added ``Document.meta`` support for indexes, which are ensured just before
   querying takes place
 - A few minor bugfixes
 

docs/conf.py

@@ -38,7 +38,7 @@ master_doc = 'index'
 
 # General information about the project.
 project = u'MongoEngine'
-copyright = u'2009-2010, Harry Marr'
+copyright = u'2009-2011, Harry Marr'
 
 # The version info for the project you're documenting, acts as replacement for
 # |version| and |release|, also used in various other places throughout the

docs/django.rst

@@ -49,10 +49,11 @@ Storage
 =======
 With MongoEngine's support for GridFS via the :class:`~mongoengine.FileField`,
 it is useful to have a Django file storage backend that wraps this. The new
-storage module is called :class:`~mongoengine.django.GridFSStorage`. Using it
-is very similar to using the default FileSystemStorage.::
-
-    fs = mongoengine.django.GridFSStorage()
+storage module is called :class:`~mongoengine.django.storage.GridFSStorage`. 
+Using it is very similar to using the default FileSystemStorage.::
+    
+    from mongoengine.django.storage import GridFSStorage
+    fs = GridFSStorage()
 
     filename = fs.save('hello.txt', 'Hello, World!')
 

docs/guide/defining-documents.rst

@@ -4,14 +4,14 @@ Defining documents
 In MongoDB, a **document** is roughly equivalent to a **row** in an RDBMS. When
 working with relational databases, rows are stored in **tables**, which have a
 strict **schema** that the rows follow. MongoDB stores documents in
-**collections** rather than tables - the principle difference is that no schema 
-is enforced at a database level. 
+**collections** rather than tables - the principle difference is that no schema
+is enforced at a database level.
 
 Defining a document's schema
 ============================
 MongoEngine allows you to define schemata for documents as this helps to reduce
 coding errors, and allows for utility methods to be defined on fields which may
-be present. 
+be present.
 
 To define a schema for a document, create a class that inherits from
 :class:`~mongoengine.Document`. Fields are specified by adding **field
@@ -19,7 +19,7 @@ objects** as class attributes to the document class::
 
     from mongoengine import *
     import datetime
-    
+
     class Page(Document):
         title = StringField(max_length=200, required=True)
         date_modified = DateTimeField(default=datetime.datetime.now)
@@ -31,31 +31,35 @@ By default, fields are not required. To make a field mandatory, set the
 validation constraints available (such as :attr:`max_length` in the example
 above). Fields may also take default values, which will be used if a value is
 not provided. Default values may optionally be a callable, which will be called
-to retrieve the value (such as in the above example). The field types available 
+to retrieve the value (such as in the above example). The field types available
 are as follows:
 
 * :class:`~mongoengine.StringField`
 * :class:`~mongoengine.URLField`
+* :class:`~mongoengine.EmailField`
 * :class:`~mongoengine.IntField`
 * :class:`~mongoengine.FloatField`
 * :class:`~mongoengine.DecimalField`
 * :class:`~mongoengine.DateTimeField`
+* :class:`~mongoengine.ComplexDateTimeField`
 * :class:`~mongoengine.ListField`
+* :class:`~mongoengine.SortedListField`
 * :class:`~mongoengine.DictField`
+* :class:`~mongoengine.MapField`
 * :class:`~mongoengine.ObjectIdField`
-* :class:`~mongoengine.EmbeddedDocumentField`
 * :class:`~mongoengine.ReferenceField`
 * :class:`~mongoengine.GenericReferenceField`
+* :class:`~mongoengine.EmbeddedDocumentField`
+* :class:`~mongoengine.GenericEmbeddedDocumentField`
 * :class:`~mongoengine.BooleanField`
 * :class:`~mongoengine.FileField`
-* :class:`~mongoengine.EmailField`
-* :class:`~mongoengine.SortedListField`
 * :class:`~mongoengine.BinaryField`
 * :class:`~mongoengine.GeoPointField`
+* :class:`~mongoengine.SequenceField`
 
 Field arguments
 ---------------
-Each field type can be customized by keyword arguments.  The following keyword 
+Each field type can be customized by keyword arguments.  The following keyword
 arguments can be set on all fields:
 
 :attr:`db_field` (Default: None)
@@ -74,7 +78,7 @@ arguments can be set on all fields:
 
     The definion of default parameters follow `the general rules on Python
     <http://docs.python.org/reference/compound_stmts.html#function-definitions>`__,
-    which means that some care should be taken when dealing with default mutable objects 
+    which means that some care should be taken when dealing with default mutable objects
     (like in :class:`~mongoengine.ListField` or :class:`~mongoengine.DictField`)::
 
         class ExampleFirst(Document):
@@ -89,7 +93,7 @@ arguments can be set on all fields:
             # This can make an .append call to  add values to the default (and all the following objects),
             # instead to just an object
             values = ListField(IntField(), default=[1,2,3])
-    
+
 
 :attr:`unique` (Default: False)
     When True, no documents in the collection will have the same value for this
@@ -104,7 +108,13 @@ arguments can be set on all fields:
 
 :attr:`choices` (Default: None)
     An iterable of choices to which the value of this field should be limited.
-    
+
+:attr:`help_text` (Default: None)
+    Optional help text to output with the field - used by form libraries
+
+:attr:`verbose` (Default: None)
+    Optional human-readable name for the field - used by form libraries
+
 
 List fields
 -----------
@@ -121,7 +131,7 @@ Embedded documents
 MongoDB has the ability to embed documents within other documents. Schemata may
 be defined for these embedded documents, just as they may be for regular
 documents. To create an embedded document, just define a document as usual, but
-inherit from :class:`~mongoengine.EmbeddedDocument` rather than 
+inherit from :class:`~mongoengine.EmbeddedDocument` rather than
 :class:`~mongoengine.Document`::
 
     class Comment(EmbeddedDocument):
@@ -134,8 +144,8 @@ document class as the first argument::
     class Page(Document):
         comments = ListField(EmbeddedDocumentField(Comment))
 
-    comment1 = Comment('Good work!')
-    comment2 = Comment('Nice article!')
+    comment1 = Comment(content='Good work!')
+    comment2 = Comment(content='Nice article!')
     page = Page(comments=[comment1, comment2])
 
 Dictionary Fields
@@ -144,7 +154,7 @@ Often, an embedded document may be used instead of a dictionary -- generally
 this is recommended as dictionaries don't support validation or custom field
 types. However, sometimes you will not know the structure of what you want to
 store; in this situation a :class:`~mongoengine.DictField` is appropriate::
-    
+
     class SurveyResponse(Document):
         date = DateTimeField()
         user = ReferenceField(User)
@@ -152,16 +162,19 @@ store; in this situation a :class:`~mongoengine.DictField` is appropriate::
 
     survey_response = SurveyResponse(date=datetime.now(), user=request.user)
     response_form = ResponseForm(request.POST)
-    survey_response.answers = response_form.cleaned_data()   
+    survey_response.answers = response_form.cleaned_data()
     survey_response.save()
 
+Dictionaries can store complex data, other dictionaries, lists, references to
+other objects, so are the most flexible field type available.
+
 Reference fields
 ----------------
 References may be stored to other documents in the database using the
 :class:`~mongoengine.ReferenceField`. Pass in another document class as the
 first argument to the constructor, then simply assign document objects to the
 field::
-    
+
     class User(Document):
         name = StringField()
 
@@ -193,19 +206,72 @@ as the constructor's argument::
     class ProfilePage(Document):
         content = StringField()
 
+
+Dealing with deletion of referred documents
+'''''''''''''''''''''''''''''''''''''''''''
+By default, MongoDB doesn't check the integrity of your data, so deleting
+documents that other documents still hold references to will lead to consistency
+issues.  Mongoengine's :class:`ReferenceField` adds some functionality to
+safeguard against these kinds of database integrity problems, providing each
+reference with a delete rule specification.  A delete rule is specified by
+supplying the :attr:`reverse_delete_rule` attributes on the
+:class:`ReferenceField` definition, like this::
+
+    class Employee(Document):
+        ...
+        profile_page = ReferenceField('ProfilePage', reverse_delete_rule=mongoengine.NULLIFY)
+
+The declaration in this example means that when an :class:`Employee` object is
+removed, the :class:`ProfilePage` that belongs to that employee is removed as
+well.  If a whole batch of employees is removed, all profile pages that are
+linked are removed as well.
+
+Its value can take any of the following constants:
+
+:const:`mongoengine.DO_NOTHING`
+  This is the default and won't do anything.  Deletes are fast, but may cause
+  database inconsistency or dangling references.
+:const:`mongoengine.DENY`
+  Deletion is denied if there still exist references to the object being
+  deleted.
+:const:`mongoengine.NULLIFY`
+  Any object's fields still referring to the object being deleted are removed
+  (using MongoDB's "unset" operation), effectively nullifying the relationship.
+:const:`mongoengine.CASCADE`
+  Any object containing fields that are refererring to the object being deleted
+  are deleted first.
+
+
+.. warning::
+   A safety note on setting up these delete rules!  Since the delete rules are
+   not recorded on the database level by MongoDB itself, but instead at runtime,
+   in-memory, by the MongoEngine module, it is of the upmost importance
+   that the module that declares the relationship is loaded **BEFORE** the
+   delete is invoked.
+
+   If, for example, the :class:`Employee` object lives in the
+   :mod:`payroll` app, and the :class:`ProfilePage` in the :mod:`people`
+   app, it is extremely important that the :mod:`people` app is loaded
+   before any employee is removed, because otherwise, MongoEngine could
+   never know this relationship exists.
+
+   In Django, be sure to put all apps that have such delete rule declarations in
+   their :file:`models.py` in the :const:`INSTALLED_APPS` tuple.
+
+
 Generic reference fields
 ''''''''''''''''''''''''
 A second kind of reference field also exists,
 :class:`~mongoengine.GenericReferenceField`. This allows you to reference any
-kind of :class:`~mongoengine.Document`, and hence doesn't take a 
+kind of :class:`~mongoengine.Document`, and hence doesn't take a
 :class:`~mongoengine.Document` subclass as a constructor argument::
 
     class Link(Document):
         url = StringField()
-        
+
     class Post(Document):
         title = StringField()
-        
+
     class Bookmark(Document):
         bookmark_object = GenericReferenceField()
 
@@ -219,9 +285,10 @@ kind of :class:`~mongoengine.Document`, and hence doesn't take a
     Bookmark(bookmark_object=post).save()
 
 .. note::
+
    Using :class:`~mongoengine.GenericReferenceField`\ s is slightly less
    efficient than the standard :class:`~mongoengine.ReferenceField`\ s, so if
-   you will only be referencing one document type, prefer the standard 
+   you will only be referencing one document type, prefer the standard
    :class:`~mongoengine.ReferenceField`.
 
 Uniqueness constraints
@@ -229,7 +296,7 @@ Uniqueness constraints
 MongoEngine allows you to specify that a field should be unique across a
 collection by providing ``unique=True`` to a :class:`~mongoengine.Field`\ 's
 constructor. If you try to save a document that has the same value for a unique
-field as a document that is already in the database, a 
+field as a document that is already in the database, a
 :class:`~mongoengine.OperationError` will be raised. You may also specify
 multi-field uniqueness constraints by using :attr:`unique_with`, which may be
 either a single field name, or a list or tuple of field names::
@@ -241,14 +308,14 @@ either a single field name, or a list or tuple of field names::
 
 Skipping Document validation on save
 ------------------------------------
-You can also skip the whole document validation process by setting 
-``validate=False`` when caling the :meth:`~mongoengine.document.Document.save` 
+You can also skip the whole document validation process by setting
+``validate=False`` when caling the :meth:`~mongoengine.document.Document.save`
 method::
 
     class Recipient(Document):
         name = StringField()
         email = EmailField()
-    
+
     recipient = Recipient(name='admin', email='root@localhost')
     recipient.save()               # will raise a ValidationError while
     recipient.save(validate=False) # won't
@@ -276,7 +343,7 @@ A :class:`~mongoengine.Document` may use a **Capped Collection** by specifying
 stored in the collection, and :attr:`max_size` is the maximum size of the
 collection in bytes. If :attr:`max_size` is not specified and
 :attr:`max_documents` is, :attr:`max_size` defaults to 10000000 bytes (10MB).
-The following example shows a :class:`Log` document that will be limited to 
+The following example shows a :class:`Log` document that will be limited to
 1000 entries and 2MB of disk space::
 
     class Log(Document):
@@ -288,9 +355,10 @@ Indexes
 You can specify indexes on collections to make querying faster. This is done
 by creating a list of index specifications called :attr:`indexes` in the
 :attr:`~mongoengine.Document.meta` dictionary, where an index specification may
-either be a single field name, or a tuple containing multiple field names. A
-direction may be specified on fields by prefixing the field name with a **+**
-or a **-** sign. Note that direction only matters on multi-field indexes. ::
+either be a single field name, a tuple containing multiple field names, or a
+dictionary containing a full index definition. A direction may be specified on
+fields by prefixing the field name with a **+** or a **-** sign. Note that
+direction only matters on multi-field indexes. ::
 
     class Page(Document):
         title = StringField()
@@ -299,10 +367,26 @@ or a **-** sign. Note that direction only matters on multi-field indexes. ::
             'indexes': ['title', ('title', '-rating')]
         }
 
+If a dictionary is passed then the following options are available:
+
+:attr:`fields` (Default: None)
+    The fields to index. Specified in the same format as described above.
+
+:attr:`types` (Default: True)
+    Whether the index should have the :attr:`_types` field added automatically
+    to the start of the index.
+
+:attr:`sparse` (Default: False)
+    Whether the index should be sparse.
+
+:attr:`unique` (Default: False)
+    Whether the index should be sparse.
+
 .. note::
-   Geospatial indexes will be automatically created for all 
+
+   Geospatial indexes will be automatically created for all
    :class:`~mongoengine.GeoPointField`\ s
-        
+
 Ordering
 ========
 A default ordering can be specified for your
@@ -324,7 +408,7 @@ subsequent calls to :meth:`~mongoengine.queryset.QuerySet.order_by`. ::
     blog_post_1 = BlogPost(title="Blog Post #1")
     blog_post_1.published_date = datetime(2010, 1, 5, 0, 0 ,0)
 
-    blog_post_2 = BlogPost(title="Blog Post #2") 
+    blog_post_2 = BlogPost(title="Blog Post #2")
     blog_post_2.published_date = datetime(2010, 1, 6, 0, 0 ,0)
 
     blog_post_3 = BlogPost(title="Blog Post #3")
@@ -336,7 +420,7 @@ subsequent calls to :meth:`~mongoengine.queryset.QuerySet.order_by`. ::
 
     # get the "first" BlogPost using default ordering
     # from BlogPost.meta.ordering
-    latest_post = BlogPost.objects.first() 
+    latest_post = BlogPost.objects.first()
     assert latest_post.title == "Blog Post #3"
 
     # override default ordering, order BlogPosts by "published_date"
@@ -365,7 +449,7 @@ Working with existing data
 To enable correct retrieval of documents involved in this kind of heirarchy,
 two extra attributes are stored on each document in the database: :attr:`_cls`
 and :attr:`_types`. These are hidden from the user through the MongoEngine
-interface, but may not be present if you are trying to use MongoEngine with 
+interface, but may not be present if you are trying to use MongoEngine with
 an existing database. For this reason, you may disable this inheritance
 mechansim, removing the dependency of :attr:`_cls` and :attr:`_types`, enabling
 you to work with existing databases. To disable inheritance on a document

docs/guide/document-instances.rst

@@ -4,12 +4,12 @@ Documents instances
 To create a new document object, create an instance of the relevant document
 class, providing values for its fields as its constructor keyword arguments.
 You may provide values for any of the fields on the document::
-    
+
     >>> page = Page(title="Test Page")
     >>> page.title
     'Test Page'
 
-You may also assign values to the document's fields using standard object 
+You may also assign values to the document's fields using standard object
 attribute syntax::
 
     >>> page.title = "Example Page"
@@ -18,10 +18,22 @@ attribute syntax::
 
 Saving and deleting documents
 =============================
-To save the document to the database, call the
-:meth:`~mongoengine.Document.save` method. If the document does not exist in
-the database, it will be created. If it does already exist, it will be
-updated.
+MongoEngine tracks changes to documents to provide efficient saving.  To save
+the document to the database, call the :meth:`~mongoengine.Document.save` method.
+If the document does not exist in the database, it will be created. If it does
+already exist, then any changes will be updated atomically.  For example::
+
+    >>> page = Page(title="Test Page")
+    >>> page.save()  # Performs an insert
+    >>> page.title = "My Page"
+    >>> page.save()  # Performs an atomic set on the title field.
+
+.. note::
+
+    Changes to documents are tracked and on the whole perform `set` operations.
+
+    * ``list_field.pop(0)`` - *sets* the resulting list
+    * ``del(list_field)``   - *unsets* whole list
 
 To delete a document, call the :meth:`~mongoengine.Document.delete` method.
 Note that this will only work if the document exists in the database and has a
@@ -67,6 +79,7 @@ is an alias to :attr:`id`::
     >>> page.id == page.pk
 
 .. note::
+
    If you define your own primary key field, the field implicitly becomes
    required, so a :class:`ValidationError` will be thrown if you don't provide
    it.

docs/guide/gridfs.rst

@@ -66,6 +66,7 @@ Deleting stored files is achieved with the :func:`delete` method::
     marmot.photo.delete()
 
 .. note::
+
     The FileField in a Document actually only stores the ID of a file in a
     separate GridFS collection. This means that deleting a document
     with a defined FileField does not actually delete the file. You must be

docs/guide/index.rst

@@ -11,3 +11,4 @@ User Guide
    document-instances
    querying
    gridfs
+   signals

docs/guide/installing.rst

@@ -1,31 +1,31 @@
 ======================
 Installing MongoEngine
 ======================
+
 To use MongoEngine, you will need to download `MongoDB <http://mongodb.org/>`_
 and ensure it is running in an accessible location. You will also need
 `PyMongo <http://api.mongodb.org/python>`_ to use MongoEngine, but if you
 install MongoEngine using setuptools, then the dependencies will be handled for
 you.
 
-MongoEngine is available on PyPI, so to use it you can use 
-:program:`easy_install`:
-    
+MongoEngine is available on PyPI, so to use it you can use :program:`pip`:
+
 .. code-block:: console
 
-    # easy_install mongoengine
+    $ pip install mongoengine
 
-Alternatively, if you don't have setuptools installed, `download it from PyPi 
+Alternatively, if you don't have setuptools installed, `download it from PyPi
 <http://pypi.python.org/pypi/mongoengine/>`_ and run
 
 .. code-block:: console
 
-    # python setup.py install
+    $ python setup.py install
 
 To use the bleeding-edge version of MongoEngine, you can get the source from
 `GitHub <http://github.com/hmarr/mongoengine/>`_ and install it as above:
-    
+
 .. code-block:: console
 
-    # git clone git://github.com/hmarr/mongoengine
-    # cd mongoengine
-    # python setup.py install
+    $ git clone git://github.com/hmarr/mongoengine
+    $ cd mongoengine
+    $ python setup.py install

docs/guide/querying.rst

@@ -5,8 +5,8 @@ Querying the database
 is used for accessing the objects in the database associated with the class.
 The :attr:`objects` attribute is actually a
 :class:`~mongoengine.queryset.QuerySetManager`, which creates and returns a new
-a new :class:`~mongoengine.queryset.QuerySet` object on access. The
-:class:`~mongoengine.queryset.QuerySet` object may may be iterated over to
+:class:`~mongoengine.queryset.QuerySet` object on access. The
+:class:`~mongoengine.queryset.QuerySet` object may be iterated over to
 fetch documents from the database::
 
     # Prints out the names of all the users in the database
@@ -14,6 +14,7 @@ fetch documents from the database::
         print user.name
 
 .. note::
+
    Once the iteration finishes (when :class:`StopIteration` is raised),
    :meth:`~mongoengine.queryset.QuerySet.rewind` will be called so that the
    :class:`~mongoengine.queryset.QuerySet` may be iterated over again. The
@@ -23,7 +24,7 @@ fetch documents from the database::
 Filtering queries
 =================
 The query may be filtered by calling the
-:class:`~mongoengine.queryset.QuerySet` object with field lookup keyword 
+:class:`~mongoengine.queryset.QuerySet` object with field lookup keyword
 arguments. The keys in the keyword arguments correspond to fields on the
 :class:`~mongoengine.Document` you are querying::
 
@@ -39,29 +40,6 @@ syntax::
     # been written by a user whose 'country' field is set to 'uk'
     uk_pages = Page.objects(author__country='uk')
 
-Querying lists
---------------
-On most fields, this syntax will look up documents where the field specified
-matches the given value exactly, but when the field refers to a
-:class:`~mongoengine.ListField`, a single item may be provided, in which case
-lists that contain that item will be matched::
-
-    class Page(Document):
-        tags = ListField(StringField())
-
-    # This will match all pages that have the word 'coding' as an item in the
-    # 'tags' list
-    Page.objects(tags='coding')
-
-Raw queries
------------
-It is possible to provide a raw PyMongo query as a query parameter, which will
-be integrated directly into the query. This is done using the ``__raw__``
-keyword argument::
-
-    Page.objects(__raw__={'tags': 'coding'})
-
-.. versionadded:: 0.4
 
 Query operators
 ===============
@@ -84,7 +62,7 @@ Available operators are as follows:
 * ``nin`` -- value is not in list (a list of values should be provided)
 * ``mod`` -- ``value % x == y``, where ``x`` and ``y`` are two provided values
 * ``all`` -- every item in list of values provided is in array
-* ``size`` -- the size of the array is 
+* ``size`` -- the size of the array is
 * ``exists`` -- value for field exists
 
 The following operators are available as shortcuts to querying with regular
@@ -99,26 +77,67 @@ expressions:
 * ``endswith`` -- string field ends with value
 * ``iendswith`` -- string field ends with value (case insensitive)
 
-.. versionadded:: 0.3
-
 There are a few special operators for performing geographical queries, that
 may used with :class:`~mongoengine.GeoPointField`\ s:
 
 * ``within_distance`` -- provide a list containing a point and a maximum
   distance (e.g. [(41.342, -87.653), 5])
+* ``within_spherical_distance`` -- Same as above but using the spherical geo model
+  (e.g. [(41.342, -87.653), 5/earth_radius])
+* ``near`` -- order the documents by how close they are to a given point
+* ``near_sphere`` -- Same as above but using the spherical geo model
 * ``within_box`` -- filter documents to those within a given bounding box (e.g.
   [(35.0, -125.0), (40.0, -100.0)])
-* ``near`` -- order the documents by how close they are to a given point
+* ``within_polygon`` -- filter documents to those within a given polygon (e.g.
+  [(41.91,-87.69), (41.92,-87.68), (41.91,-87.65), (41.89,-87.65)]).
+  .. note:: Requires Mongo Server 2.0
 
-.. versionadded:: 0.4
 
-Querying by position
-====================
+Querying lists
+--------------
+On most fields, this syntax will look up documents where the field specified
+matches the given value exactly, but when the field refers to a
+:class:`~mongoengine.ListField`, a single item may be provided, in which case
+lists that contain that item will be matched::
+
+    class Page(Document):
+        tags = ListField(StringField())
+
+    # This will match all pages that have the word 'coding' as an item in the
+    # 'tags' list
+    Page.objects(tags='coding')
+
 It is possible to query by position in a list by using a numerical value as a
 query operator. So if you wanted to find all pages whose first tag was ``db``,
 you could use the following query::
 
-    BlogPost.objects(tags__0='db')
+    Page.objects(tags__0='db')
+
+If you only want to fetch part of a list eg: you want to paginate a list, then
+the `slice` operator is required::
+
+    # comments - skip 5, limit 10
+    Page.objects.fields(slice__comments=[5, 10])
+
+For updating documents, if you don't know the position in a list, you can use
+the $ positional operator ::
+
+    Post.objects(comments__by="joe").update(**{'inc__comments__$__votes': 1})
+
+However, this doesn't map well to the syntax so you can also use a capital S instead ::
+
+    Post.objects(comments__by="joe").update(inc__comments__S__votes=1)
+
+    .. note:: Due to Mongo currently the $ operator only applies to the first matched item in the query.
+
+
+Raw queries
+-----------
+It is possible to provide a raw PyMongo query as a query parameter, which will
+be integrated directly into the query. This is done using the ``__raw__``
+keyword argument::
+
+    Page.objects(__raw__={'tags': 'coding'})
 
 .. versionadded:: 0.4
 
@@ -163,9 +182,9 @@ To retrieve a result that should be unique in the collection, use
 and :class:`~mongoengine.queryset.MultipleObjectsReturned` if more than one
 document matched the query.
 
-A variation of this method exists, 
+A variation of this method exists,
 :meth:`~mongoengine.queryset.Queryset.get_or_create`, that will create a new
-document with the query arguments if no documents match the query. An 
+document with the query arguments if no documents match the query. An
 additional keyword argument, :attr:`defaults` may be provided, which will be
 used as default values for the new document, in the case that it should need
 to be created::
@@ -175,6 +194,22 @@ to be created::
     >>> a.name == b.name and a.age == b.age
     True
 
+Dereferencing results
+---------------------
+When iterating the results of :class:`~mongoengine.ListField` or
+:class:`~mongoengine.DictField` we automatically dereference any
+:class:`~pymongo.dbref.DBRef` objects as efficiently as possible, reducing the
+number the queries to mongo.
+
+There are times when that efficiency is not enough, documents that have
+:class:`~mongoengine.ReferenceField` objects or
+:class:`~mongoengine.GenericReferenceField` objects at the top level are
+expensive as the number of queries to MongoDB can quickly rise.
+
+To limit the number of queries use
+:func:`~mongoengine.queryset.QuerySet.select_related` which converts the
+QuerySet to a list and dereferences as efficiently as possible.
+
 Default Document queries
 ========================
 By default, the objects :attr:`~mongoengine.Document.objects` attribute on a
@@ -240,7 +275,7 @@ Javascript code that is executed on the database server.
 Counting results
 ----------------
 Just as with limiting and skipping results, there is a method on
-:class:`~mongoengine.queryset.QuerySet` objects -- 
+:class:`~mongoengine.queryset.QuerySet` objects --
 :meth:`~mongoengine.queryset.QuerySet.count`, but there is also a more Pythonic
 way of achieving this::
 
@@ -254,6 +289,7 @@ You may sum over the values of a specific field on documents using
     yearly_expense = Employee.objects.sum('salary')
 
 .. note::
+
    If the field isn't present on a document, that document will be ignored from
    the sum.
 
@@ -302,6 +338,11 @@ will be given::
     >>> f.rating # default value
     3
 
+.. note::
+
+    The :meth:`~mongoengine.queryset.QuerySet.exclude` is the opposite of
+    :meth:`~mongoengine.queryset.QuerySet.only` if you want to exclude a field.
+
 If you later need the missing fields, just call
 :meth:`~mongoengine.Document.reload` on your document.
 
@@ -309,11 +350,11 @@ Advanced queries
 ================
 Sometimes calling a :class:`~mongoengine.queryset.QuerySet` object with keyword
 arguments can't fully express the query you want to use -- for example if you
-need to combine a number of constraints using *and* and *or*. This is made 
+need to combine a number of constraints using *and* and *or*. This is made
 possible in MongoEngine through the :class:`~mongoengine.queryset.Q` class.
 A :class:`~mongoengine.queryset.Q` object represents part of a query, and
 can be initialised using the same keyword-argument syntax you use to query
-documents. To build a complex query, you may combine 
+documents. To build a complex query, you may combine
 :class:`~mongoengine.queryset.Q` objects using the ``&`` (and) and ``|`` (or)
 operators. To use a :class:`~mongoengine.queryset.Q` object, pass it in as the
 first positional argument to :attr:`Document.objects` when you filter it by
@@ -325,11 +366,66 @@ calling it with keyword arguments::
     # Get top posts
     Post.objects((Q(featured=True) & Q(hits__gte=1000)) | Q(hits__gte=5000))
 
-.. warning::
-   Only use these advanced queries if absolutely necessary as they will execute
-   significantly slower than regular queries. This is because they are not
-   natively supported by MongoDB -- they are compiled to Javascript and sent
-   to the server for execution.
+.. _guide-atomic-updates:
+
+Atomic updates
+==============
+Documents may be updated atomically by using the
+:meth:`~mongoengine.queryset.QuerySet.update_one` and
+:meth:`~mongoengine.queryset.QuerySet.update` methods on a
+:meth:`~mongoengine.queryset.QuerySet`. There are several different "modifiers"
+that you may use with these methods:
+
+* ``set`` -- set a particular value
+* ``unset`` -- delete a particular value (since MongoDB v1.3+)
+* ``inc`` -- increment a value by a given amount
+* ``dec`` -- decrement a value by a given amount
+* ``pop`` -- remove the last item from a list
+* ``push`` -- append a value to a list
+* ``push_all`` -- append several values to a list
+* ``pop`` -- remove the first or last element of a list
+* ``pull`` -- remove a value from a list
+* ``pull_all`` -- remove several values from a list
+* ``add_to_set`` -- add value to a list only if its not in the list already
+
+The syntax for atomic updates is similar to the querying syntax, but the
+modifier comes before the field, not after it::
+
+    >>> post = BlogPost(title='Test', page_views=0, tags=['database'])
+    >>> post.save()
+    >>> BlogPost.objects(id=post.id).update_one(inc__page_views=1)
+    >>> post.reload()  # the document has been changed, so we need to reload it
+    >>> post.page_views
+    1
+    >>> BlogPost.objects(id=post.id).update_one(set__title='Example Post')
+    >>> post.reload()
+    >>> post.title
+    'Example Post'
+    >>> BlogPost.objects(id=post.id).update_one(push__tags='nosql')
+    >>> post.reload()
+    >>> post.tags
+    ['database', 'nosql']
+
+.. note ::
+
+    In version 0.5 the :meth:`~mongoengine.Document.save` runs atomic updates
+    on changed documents by tracking changes to that document.
+
+The positional operator allows you to update list items without knowing the
+index position, therefore making the update a single atomic operation.  As we
+cannot use the `$` syntax in keyword arguments it has been mapped to `S`::
+
+    >>> post = BlogPost(title='Test', page_views=0, tags=['database', 'mongo'])
+    >>> post.save()
+    >>> BlogPost.objects(id=post.id, tags='mongo').update(set__tags__S='mongodb')
+    >>> post.reload()
+    >>> post.tags
+    ['database', 'mongodb']
+
+.. note ::
+    Currently only top level lists are handled, future versions of mongodb /
+    pymongo plan to support nested positional operators.  See `The $ positional
+    operator <http://www.mongodb.org/display/DOCS/Updating#Updating-The%24positionaloperator>`_.
 
 Server-side javascript execution
 ================================
@@ -433,43 +529,3 @@ following example shows how the substitutions are made::
         return comments;
     }
     """)
-
-.. _guide-atomic-updates:
-
-Atomic updates
-==============
-Documents may be updated atomically by using the
-:meth:`~mongoengine.queryset.QuerySet.update_one` and
-:meth:`~mongoengine.queryset.QuerySet.update` methods on a 
-:meth:`~mongoengine.queryset.QuerySet`. There are several different "modifiers"
-that you may use with these methods:
-
-* ``set`` -- set a particular value
-* ``unset`` -- delete a particular value (since MongoDB v1.3+)
-* ``inc`` -- increment a value by a given amount
-* ``dec`` -- decrement a value by a given amount
-* ``pop`` -- remove the last item from a list
-* ``push`` -- append a value to a list
-* ``push_all`` -- append several values to a list
-* ``pop`` -- remove the first or last element of a list
-* ``pull`` -- remove a value from a list
-* ``pull_all`` -- remove several values from a list
-* ``add_to_set`` -- add value to a list only if its not in the list already
-
-The syntax for atomic updates is similar to the querying syntax, but the 
-modifier comes before the field, not after it::
-
-    >>> post = BlogPost(title='Test', page_views=0, tags=['database'])
-    >>> post.save()
-    >>> BlogPost.objects(id=post.id).update_one(inc__page_views=1)
-    >>> post.reload()  # the document has been changed, so we need to reload it
-    >>> post.page_views
-    1
-    >>> BlogPost.objects(id=post.id).update_one(set__title='Example Post')
-    >>> post.reload()
-    >>> post.title
-    'Example Post'
-    >>> BlogPost.objects(id=post.id).update_one(push__tags='nosql')
-    >>> post.reload()
-    >>> post.tags
-    ['database', 'nosql']

docs/guide/signals.rst

@@ -0,0 +1,49 @@
+.. _signals:
+
+Signals
+=======
+
+.. versionadded:: 0.5
+
+Signal support is provided by the excellent `blinker`_ library and
+will gracefully fall back if it is not available.
+
+
+The following document signals exist in MongoEngine and are pretty self explaintary:
+
+  * `mongoengine.signals.pre_init`
+  * `mongoengine.signals.post_init`
+  * `mongoengine.signals.pre_save`
+  * `mongoengine.signals.post_save`
+  * `mongoengine.signals.pre_delete`
+  * `mongoengine.signals.post_delete`
+
+Example usage::
+
+    from mongoengine import *
+    from mongoengine import signals
+
+    class Author(Document):
+        name = StringField()
+
+        def __unicode__(self):
+            return self.name
+
+        @classmethod
+        def pre_save(cls, sender, document, **kwargs):
+            logging.debug("Pre Save: %s" % document.name)
+
+        @classmethod
+        def post_save(cls, sender, document, **kwargs):
+            logging.debug("Post Save: %s" % document.name)
+            if 'created' in kwargs:
+                if kwargs['created']:
+                    logging.debug("Created")
+                else:
+                    logging.debug("Updated")
+
+        signals.pre_save.connect(Author.pre_save, sender=Author)
+        signals.post_save.connect(Author.post_save, sender=Author)
+
+
+.. _blinker: http://pypi.python.org/pypi/blinker

docs/index.rst

@@ -2,34 +2,62 @@
 MongoEngine User Documentation
 ==============================
 
-MongoEngine is an Object-Document Mapper, written in Python for working with 
+**MongoEngine** is an Object-Document Mapper, written in Python for working with
 MongoDB. To install it, simply run
 
 .. code-block:: console
 
     # pip install -U mongoengine
 
-The source is available on `GitHub <http://github.com/hmarr/mongoengine>`_.
+:doc:`tutorial`
+  Start here for a quick overview.
+
+:doc:`guide/index`
+  The Full guide to MongoEngine
+
+:doc:`apireference`
+  The complete API documentation.
+
+:doc:`django`
+  Using MongoEngine and Django
+
+Community
+---------
 
 To get help with using MongoEngine, use the `MongoEngine Users mailing list
 <http://groups.google.com/group/mongoengine-users>`_ or come chat on the
 `#mongoengine IRC channel <irc://irc.freenode.net/mongoengine>`_.
 
-If you are interested in contributing, join the developers' `mailing list 
+Contributing
+------------
+
+The source is available on `GitHub <http://github.com/hmarr/mongoengine>`_ and
+contributions are always encouraged. Contributions can be as simple as
+minor tweaks to this documentation. To contribute, fork the project on
+`GitHub <http://github.com/hmarr/mongoengine>`_ and send a
+pull request.
+
+Also, you can join the developers' `mailing list
 <http://groups.google.com/group/mongoengine-dev>`_.
 
+Changes
+-------
+See the :doc:`changelog` for a full list of changes to MongoEngine.
+
 .. toctree::
-   :maxdepth: 2
+   :hidden:
 
    tutorial
    guide/index
    apireference
    django
    changelog
+   upgrade
 
 Indices and tables
-==================
+------------------
 
 * :ref:`genindex`
+* :ref:`modindex`
 * :ref:`search`
 

docs/tutorial.rst

@@ -22,7 +22,7 @@ function. The only argument we need to provide is the name of the MongoDB
 database to use::
 
     from mongoengine import *
-    
+
     connect('tumblelog')
 
 For more information about connecting to MongoDB see :ref:`guide-connecting`.
@@ -112,7 +112,7 @@ link table, we can just store a list of tags in each post. So, for both
 efficiency and simplicity's sake, we'll store the tags as strings directly
 within the post, rather than storing references to tags in a separate
 collection. Especially as tags are generally very short (often even shorter
-than a document's id), this denormalisation won't impact very strongly on the 
+than a document's id), this denormalisation won't impact very strongly on the
 size of our database. So let's take a look that the code our modified
 :class:`Post` class::
 
@@ -152,6 +152,21 @@ We can then store a list of comment documents in our post document::
         tags = ListField(StringField(max_length=30))
         comments = ListField(EmbeddedDocumentField(Comment))
 
+Handling deletions of references
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+The :class:`~mongoengine.ReferenceField` object takes a keyword
+`reverse_delete_rule` for handling deletion rules if the reference is deleted.
+To delete all the posts if a user is deleted set the rule::
+
+    class Post(Document):
+        title = StringField(max_length=120, required=True)
+        author = ReferenceField(User, reverse_delete_rule=CASCADE)
+        tags = ListField(StringField(max_length=30))
+        comments = ListField(EmbeddedDocumentField(Comment))
+
+See :class:`~mongoengine.ReferenceField` for more information.
+
 Adding data to our Tumblelog
 ============================
 Now that we've defined how our documents will be structured, let's start adding
@@ -250,5 +265,5 @@ the first matched by the query you provide. Aggregation functions may also be
 used on :class:`~mongoengine.queryset.QuerySet` objects::
 
     num_posts = Post.objects(tags='mongodb').count()
-    print 'Found % posts with tag "mongodb"' % num_posts
-    
+    print 'Found %d posts with tag "mongodb"' % num_posts
+

docs/upgrade.rst

@@ -0,0 +1,97 @@
+=========
+Upgrading
+=========
+
+0.4 to 0.5
+===========
+
+There have been the following backwards incompatibilities from 0.4 to 0.5.  The
+main areas of changed are: choices in fields, map_reduce and collection names.
+
+Choice options:
+--------------
+
+Are now expected to be an iterable of tuples, with  the first element in each
+tuple being the actual value to be stored. The second element is the
+human-readable name for the option.
+
+
+PyMongo / MongoDB
+-----------------
+
+map reduce now requires pymongo 1.11+- The pymongo merge_output and reduce_output
+parameters, have been depreciated.
+
+More methods now use map_reduce as db.eval is not supported for sharding as such
+the following have been changed:
+
+    * :meth:`~mongoengine.queryset.QuerySet.sum`
+    * :meth:`~mongoengine.queryset.QuerySet.average`
+    * :meth:`~mongoengine.queryset.QuerySet.item_frequencies`
+
+
+Default collection naming
+-------------------------
+
+Previously it was just lowercase, its now much more pythonic and readable as its
+lowercase and underscores, previously ::
+
+    class MyAceDocument(Document):
+        pass
+
+    MyAceDocument._meta['collection'] == myacedocument
+
+In 0.5 this will change to ::
+
+    class MyAceDocument(Document):
+        pass
+
+    MyAceDocument._get_collection_name() == my_ace_document
+
+To upgrade use a Mixin class to set meta like so ::
+
+    class BaseMixin(object):
+        meta = {
+            'collection': lambda c: c.__name__.lower()
+        }
+
+    class MyAceDocument(Document, BaseMixin):
+        pass
+
+    MyAceDocument._get_collection_name() == myacedocument
+
+Alternatively, you can rename your collections eg ::
+
+    from mongoengine.connection import _get_db
+    from mongoengine.base import _document_registry
+
+    def rename_collections():
+        db = _get_db()
+
+        failure = False
+
+        collection_names = [d._get_collection_name() for d in _document_registry.values()]
+
+        for new_style_name in collection_names:
+            if not new_style_name:  # embedded documents don't have collections
+                continue
+            old_style_name = new_style_name.replace('_', '')
+
+            if old_style_name == new_style_name:
+                continue  # Nothing to do
+
+            existing = db.collection_names()
+            if old_style_name in existing:
+                if new_style_name in existing:
+                    failure = True
+                    print "FAILED to rename: %s to %s (already exists)" % (
+                        old_style_name, new_style_name)
+                else:
+                    db[old_style_name].rename(new_style_name)
+                    print "Renamed:  %s to %s" % (old_style_name, new_style_name)
+
+        if failure:
+            print "Upgrading  collection names failed"
+        else:
+            print "Upgraded collection names"
+

mongoengine/__init__.py

@@ -6,13 +6,16 @@ import connection
 from connection import *
 import queryset
 from queryset import *
+import signals
+from signals import *
 
 __all__ = (document.__all__ + fields.__all__ + connection.__all__ +
-           queryset.__all__)
+           queryset.__all__ + signals.__all__)
 
 __author__ = 'Harry Marr'
 
-VERSION = (0, 4, 0)
+VERSION = (0, 5, 0)
+
 
 def get_version():
     version = '%s.%s' % (VERSION[0], VERSION[1])
@@ -21,4 +24,3 @@ def get_version():
     return version
 
 __version__ = get_version()
-

mongoengine/base.py

@@ -1,33 +1,68 @@
 from queryset import QuerySet, QuerySetManager
 from queryset import DoesNotExist, MultipleObjectsReturned
+from queryset import DO_NOTHING
 
+from mongoengine import signals
+
+import weakref
 import sys
 import pymongo
 import pymongo.objectid
+import operator
+from functools import partial
 
 
-_document_registry = {}
+class NotRegistered(Exception):
+    pass
 
-def get_document(name):
-    return _document_registry[name]
 
+class InvalidDocumentError(Exception):
+    pass
 
 class ValidationError(Exception):
     pass
 
 
+_document_registry = {}
+
+
+def get_document(name):
+    doc = _document_registry.get(name, None)
+    if not doc:
+        # Possible old style names
+        end = ".%s" % name
+        possible_match = [k for k in _document_registry.keys() if k.endswith(end)]
+        if len(possible_match) == 1:
+            doc = _document_registry.get(possible_match.pop(), None)
+    if not doc:
+        raise NotRegistered("""
+            `%s` has not been registered in the document registry.
+            Importing the document class automatically registers it, has it
+            been imported?
+        """.strip() % name)
+    return doc
+
+
 class BaseField(object):
     """A base class for fields in a MongoDB document. Instances of this class
     may be added to subclasses of `Document` to define a document's schema.
+
+    .. versionchanged:: 0.5 - added verbose and help text
     """
 
-    # Fields may have _types inserted into indexes by default 
+    # Fields may have _types inserted into indexes by default
     _index_with_types = True
     _geo_index = False
 
-    def __init__(self, db_field=None, name=None, required=False, default=None, 
+    # These track each time a Field instance is created. Used to retain order.
+    # The auto_creation_counter is used for fields that MongoEngine implicitly
+    # creates, creation_counter is used for all user-specified fields.
+    creation_counter = 0
+    auto_creation_counter = -1
+
+    def __init__(self, db_field=None, name=None, required=False, default=None,
                  unique=False, unique_with=None, primary_key=False,
-                 validation=None, choices=None):
+                 validation=None, choices=None, verbose_name=None, help_text=None):
         self.db_field = (db_field or name) if not primary_key else '_id'
         if name:
             import warnings
@@ -41,9 +76,19 @@ class BaseField(object):
         self.primary_key = primary_key
         self.validation = validation
         self.choices = choices
+        self.verbose_name = verbose_name
+        self.help_text = help_text
+
+        # Adjust the appropriate creation counter, and save our local copy.
+        if self.db_field == '_id':
+            self.creation_counter = BaseField.auto_creation_counter
+            BaseField.auto_creation_counter -= 1
+        else:
+            self.creation_counter = BaseField.creation_counter
+            BaseField.creation_counter += 1
 
     def __get__(self, instance, owner):
-        """Descriptor for retrieving a value from a field in a document. Do 
+        """Descriptor for retrieving a value from a field in a document. Do
         any necessary conversion between Python and MongoDB types.
         """
         if instance is None:
@@ -57,12 +102,19 @@ class BaseField(object):
             # Allow callable default values
             if callable(value):
                 value = value()
+
+        # Convert lists / values so we can watch for any changes on them
+        if isinstance(value, (list, tuple)) and not isinstance(value, BaseList):
+            value = BaseList(value, instance=instance, name=self.name)
+        elif isinstance(value, dict) and not isinstance(value, BaseDict):
+            value = BaseDict(value, instance=instance, name=self.name)
         return value
 
     def __set__(self, instance, value):
         """Descriptor for assigning a value to a field in a document.
         """
         instance._data[self.name] = value
+        instance._mark_as_changed(self.name)
 
     def to_python(self, value):
         """Convert a MongoDB-compatible type to a Python type.
@@ -87,9 +139,9 @@ class BaseField(object):
     def _validate(self, value):
         # check choices
         if self.choices is not None:
-            if value not in self.choices:
-                raise ValidationError("Value must be one of %s."
-                    % unicode(self.choices))
+            option_keys = [option_key for option_key, option_value in self.choices]
+            if value not in option_keys:
+                raise ValidationError("Value must be one of %s." % unicode(option_keys))
 
         # check validation argument
         if self.validation is not None:
@@ -102,13 +154,159 @@ class BaseField(object):
 
         self.validate(value)
 
+
+class ComplexBaseField(BaseField):
+    """Handles complex fields, such as lists / dictionaries.
+
+    Allows for nesting of embedded documents inside complex types.
+    Handles the lazy dereferencing of a queryset by lazily dereferencing all
+    items in a list / dict rather than one at a time.
+
+    .. versionadded:: 0.5
+    """
+
+    field = None
+
+    def __get__(self, instance, owner):
+        """Descriptor to automatically dereference references.
+        """
+        if instance is None:
+            # Document class being used rather than a document object
+            return self
+
+        from dereference import dereference
+        instance._data[self.name] = dereference(
+            instance._data.get(self.name), max_depth=1, instance=instance, name=self.name, get=True
+        )
+        return super(ComplexBaseField, self).__get__(instance, owner)
+
+    def to_python(self, value):
+        """Convert a MongoDB-compatible type to a Python type.
+        """
+        from mongoengine import Document
+
+        if isinstance(value, basestring):
+            return value
+
+        if hasattr(value, 'to_python'):
+            return value.to_python()
+
+        is_list = False
+        if not hasattr(value, 'items'):
+            try:
+                is_list = True
+                value = dict([(k,v) for k,v in enumerate(value)])
+            except TypeError:  # Not iterable return the value
+                return value
+
+        if self.field:
+            value_dict = dict([(key, self.field.to_python(item)) for key, item in value.items()])
+        else:
+            value_dict = {}
+            for k,v in value.items():
+                if isinstance(v, Document):
+                    # We need the id from the saved object to create the DBRef
+                    if v.pk is None:
+                        raise ValidationError('You can only reference documents once '
+                                      'they have been saved to the database')
+                    collection = v._get_collection_name()
+                    value_dict[k] = pymongo.dbref.DBRef(collection, v.pk)
+                elif hasattr(v, 'to_python'):
+                    value_dict[k] = v.to_python()
+                else:
+                    value_dict[k] = self.to_python(v)
+
+        if is_list:  # Convert back to a list
+            return [v for k,v in sorted(value_dict.items(), key=operator.itemgetter(0))]
+        return value_dict
+
+    def to_mongo(self, value):
+        """Convert a Python type to a MongoDB-compatible type.
+        """
+        from mongoengine import Document
+
+        if isinstance(value, basestring):
+            return value
+
+        if hasattr(value, 'to_mongo'):
+            return value.to_mongo()
+
+        is_list = False
+        if not hasattr(value, 'items'):
+            try:
+                is_list = True
+                value = dict([(k,v) for k,v in enumerate(value)])
+            except TypeError:  # Not iterable return the value
+                return value
+
+        if self.field:
+            value_dict = dict([(key, self.field.to_mongo(item)) for key, item in value.items()])
+        else:
+            value_dict = {}
+            for k,v in value.items():
+                if isinstance(v, Document):
+                    # We need the id from the saved object to create the DBRef
+                    if v.pk is None:
+                        raise ValidationError('You can only reference documents once '
+                                      'they have been saved to the database')
+
+                    # If its a document that is not inheritable it won't have
+                    # _types / _cls data so make it a generic reference allows
+                    # us to dereference
+                    meta = getattr(v, 'meta', getattr(v, '_meta', {}))
+                    if meta and not meta['allow_inheritance'] and not self.field:
+                        from fields import GenericReferenceField
+                        value_dict[k] = GenericReferenceField().to_mongo(v)
+                    else:
+                        collection = v._get_collection_name()
+                        value_dict[k] = pymongo.dbref.DBRef(collection, v.pk)
+                elif hasattr(v, 'to_mongo'):
+                    value_dict[k] = v.to_mongo()
+                else:
+                    value_dict[k] = self.to_mongo(v)
+
+        if is_list:  # Convert back to a list
+            return [v for k,v in sorted(value_dict.items(), key=operator.itemgetter(0))]
+        return value_dict
+
+    def validate(self, value):
+        """If field provided ensure the value is valid.
+        """
+        if self.field:
+            try:
+                if hasattr(value, 'iteritems'):
+                    [self.field.validate(v) for k,v in value.iteritems()]
+                else:
+                    [self.field.validate(v) for v in value]
+            except Exception, err:
+                raise ValidationError('Invalid %s item (%s)' % (
+                        self.field.__class__.__name__, str(v)))
+
+    def prepare_query_value(self, op, value):
+        return self.to_mongo(value)
+
+    def lookup_member(self, member_name):
+        if self.field:
+            return self.field.lookup_member(member_name)
+        return None
+
+    def _set_owner_document(self, owner_document):
+        if self.field:
+            self.field.owner_document = owner_document
+        self._owner_document = owner_document
+
+    def _get_owner_document(self, owner_document):
+        self._owner_document = owner_document
+
+    owner_document = property(_get_owner_document, _set_owner_document)
+
+
 class ObjectIdField(BaseField):
     """An field wrapper around MongoDB's ObjectIds.
     """
 
     def to_python(self, value):
         return value
-        # return unicode(value)
 
     def to_mongo(self, value):
         if not isinstance(value, pymongo.objectid.ObjectId):
@@ -143,25 +341,30 @@ class DocumentMetaclass(type):
         class_name = [name]
         superclasses = {}
         simple_class = True
+
         for base in bases:
             # Include all fields present in superclasses
             if hasattr(base, '_fields'):
                 doc_fields.update(base._fields)
-                class_name.append(base._class_name)
                 # Get superclasses from superclass
                 superclasses[base._class_name] = base
                 superclasses.update(base._superclasses)
+            else:  # Add any mixin fields
+                attrs.update(dict([(k,v) for k,v in base.__dict__.items()
+                                    if issubclass(v.__class__, BaseField)]))
 
-            if hasattr(base, '_meta'):
-                # Ensure that the Document class may be subclassed - 
-                # inheritance may be disabled to remove dependency on 
+            if hasattr(base, '_meta') and not base._meta.get('abstract'):
+                # Ensure that the Document class may be subclassed -
+                # inheritance may be disabled to remove dependency on
                 # additional fields _cls and _types
+                class_name.append(base._class_name)
                 if base._meta.get('allow_inheritance', True) == False:
                     raise ValueError('Document %s may not be subclassed' %
                                      base.__name__)
                 else:
                     simple_class = False
 
+        doc_class_name = '.'.join(reversed(class_name))
         meta = attrs.get('_meta', attrs.get('meta', {}))
 
         if 'allow_inheritance' not in meta:
@@ -169,12 +372,11 @@ class DocumentMetaclass(type):
 
         # Only simple classes - direct subclasses of Document - may set
         # allow_inheritance to False
-        if not simple_class and not meta['allow_inheritance']:
+        if not simple_class and not meta['allow_inheritance'] and not meta['abstract']:
             raise ValueError('Only direct subclasses of Document may set '
                              '"allow_inheritance" to False')
         attrs['_meta'] = meta
-
-        attrs['_class_name'] = '.'.join(reversed(class_name))
+        attrs['_class_name'] = doc_class_name
         attrs['_superclasses'] = superclasses
 
         # Add the document's fields to the _fields attribute
@@ -186,26 +388,37 @@ class DocumentMetaclass(type):
                     attr_value.db_field = attr_name
                 doc_fields[attr_name] = attr_value
         attrs['_fields'] = doc_fields
+        attrs['_db_field_map'] = dict([(k, v.db_field) for k, v in doc_fields.items() if k!=v.db_field])
+        attrs['_reverse_db_field_map'] = dict([(v, k) for k, v in attrs['_db_field_map'].items()])
+
+        from mongoengine import Document
 
         new_class = super_new(cls, name, bases, attrs)
         for field in new_class._fields.values():
             field.owner_document = new_class
+            delete_rule = getattr(field, 'reverse_delete_rule', DO_NOTHING)
+            if delete_rule != DO_NOTHING:
+                field.document_type.register_delete_rule(new_class, field.name,
+                        delete_rule)
+
+            if field.name and hasattr(Document, field.name):
+                raise InvalidDocumentError("%s is a document method and not a valid field name" % field.name)
 
         module = attrs.get('__module__')
 
-        base_excs = tuple(base.DoesNotExist for base in bases 
+        base_excs = tuple(base.DoesNotExist for base in bases
                           if hasattr(base, 'DoesNotExist')) or (DoesNotExist,)
         exc = subclass_exception('DoesNotExist', base_excs, module)
         new_class.add_to_class('DoesNotExist', exc)
 
-        base_excs = tuple(base.MultipleObjectsReturned for base in bases 
+        base_excs = tuple(base.MultipleObjectsReturned for base in bases
                           if hasattr(base, 'MultipleObjectsReturned'))
         base_excs = base_excs or (MultipleObjectsReturned,)
         exc = subclass_exception('MultipleObjectsReturned', base_excs, module)
         new_class.add_to_class('MultipleObjectsReturned', exc)
 
         global _document_registry
-        _document_registry[name] = new_class
+        _document_registry[doc_class_name] = new_class
 
         return new_class
 
@@ -220,15 +433,24 @@ class TopLevelDocumentMetaclass(DocumentMetaclass):
 
     def __new__(cls, name, bases, attrs):
         super_new = super(TopLevelDocumentMetaclass, cls).__new__
-        # Classes defined in this package are abstract and should not have 
+        # Classes defined in this package are abstract and should not have
         # their own metadata with DB collection, etc.
-        # __metaclass__ is only set on the class with the __metaclass__ 
+        # __metaclass__ is only set on the class with the __metaclass__
         # attribute (i.e. it is not set on subclasses). This differentiates
         # 'real' documents from the 'Document' class
-        if attrs.get('__metaclass__') == TopLevelDocumentMetaclass:
+        #
+        # Also assume a class is abstract if it has abstract set to True in
+        # its meta dictionary. This allows custom Document superclasses.
+        if (attrs.get('__metaclass__') == TopLevelDocumentMetaclass or
+            ('meta' in attrs and attrs['meta'].get('abstract', False))):
+            # Make sure no base class was non-abstract
+            non_abstract_bases = [b for b in bases
+                if hasattr(b,'_meta') and not b._meta.get('abstract', False)]
+            if non_abstract_bases:
+                raise ValueError("Abstract document cannot have non-abstract base")
             return super_new(cls, name, bases, attrs)
 
-        collection = name.lower()
+        collection = ''.join('_%s' % c if c.isupper() else c for c in name).strip('_').lower()
 
         id_field = None
         base_indexes = []
@@ -236,28 +458,45 @@ class TopLevelDocumentMetaclass(DocumentMetaclass):
 
         # Subclassed documents inherit collection from superclass
         for base in bases:
-            if hasattr(base, '_meta') and 'collection' in base._meta:
-                collection = base._meta['collection']
-
+            if hasattr(base, '_meta'):
+                if 'collection' in attrs.get('meta', {}) and not base._meta.get('abstract', False):
+                    import warnings
+                    msg = "Trying to set a collection on a subclass (%s)" % name
+                    warnings.warn(msg, SyntaxWarning)
+                    del(attrs['meta']['collection'])
+                if base._get_collection_name():
+                    collection = base._get_collection_name()
                 # Propagate index options.
                 for key in ('index_background', 'index_drop_dups', 'index_opts'):
-                   if key in base._meta:
-                      base_meta[key] = base._meta[key]
+                    if key in base._meta:
+                        base_meta[key] = base._meta[key]
 
                 id_field = id_field or base._meta.get('id_field')
                 base_indexes += base._meta.get('indexes', [])
+                # Propagate 'allow_inheritance'
+                if 'allow_inheritance' in base._meta:
+                    base_meta['allow_inheritance'] = base._meta['allow_inheritance']
+                if 'queryset_class' in base._meta:
+                    base_meta['queryset_class'] = base._meta['queryset_class']
+            try:
+                base_meta['objects'] = base.__getattribute__(base, 'objects')
+            except AttributeError:
+                pass
 
         meta = {
+            'abstract': False,
             'collection': collection,
             'max_documents': None,
             'max_size': None,
-            'ordering': [], # default ordering applied at runtime
-            'indexes': [], # indexes to be ensured at runtime
+            'ordering': [],  # default ordering applied at runtime
+            'indexes': [],  # indexes to be ensured at runtime
             'id_field': id_field,
             'index_background': False,
             'index_drop_dups': False,
             'index_opts': {},
             'queryset_class': QuerySet,
+            'delete_rules': {},
+            'allow_inheritance': True
         }
         meta.update(base_meta)
 
@@ -269,14 +508,44 @@ class TopLevelDocumentMetaclass(DocumentMetaclass):
         # DocumentMetaclass before instantiating CollectionManager object
         new_class = super_new(cls, name, bases, attrs)
 
+        collection = attrs['_meta'].get('collection', None)
+        if callable(collection):
+            new_class._meta['collection'] = collection(new_class)
+
         # Provide a default queryset unless one has been manually provided
-        if not hasattr(new_class, 'objects'):
-            new_class.objects = QuerySetManager()
+        manager = attrs.get('objects', meta.get('objects', QuerySetManager()))
+        if hasattr(manager, 'queryset_class'):
+            meta['queryset_class'] = manager.queryset_class
+        new_class.objects = manager
 
         user_indexes = [QuerySet._build_index_spec(new_class, spec)
                         for spec in meta['indexes']] + base_indexes
         new_class._meta['indexes'] = user_indexes
 
+        unique_indexes = cls._unique_with_indexes(new_class)
+        new_class._meta['unique_indexes'] = unique_indexes
+
+        for field_name, field in new_class._fields.items():
+            # Check for custom primary key
+            if field.primary_key:
+                current_pk = new_class._meta['id_field']
+                if current_pk and current_pk != field_name:
+                    raise ValueError('Cannot override primary key field')
+
+                if not current_pk:
+                    new_class._meta['id_field'] = field_name
+                    # Make 'Document.id' an alias to the real primary key field
+                    new_class.id = field
+
+        if not new_class._meta['id_field']:
+            new_class._meta['id_field'] = 'id'
+            new_class._fields['id'] = ObjectIdField(db_field='_id')
+            new_class.id = new_class._fields['id']
+
+        return new_class
+
+    @classmethod
+    def _unique_with_indexes(cls, new_class, namespace=""):
         unique_indexes = []
         for field_name, field in new_class._fields.items():
             # Generate a list of indexes needed by uniqueness constraints
@@ -302,52 +571,50 @@ class TopLevelDocumentMetaclass(DocumentMetaclass):
                     unique_fields += unique_with
 
                 # Add the new index to the list
-                index = [(f, pymongo.ASCENDING) for f in unique_fields]
+                index = [("%s%s" % (namespace, f), pymongo.ASCENDING) for f in unique_fields]
                 unique_indexes.append(index)
 
-            # Check for custom primary key
-            if field.primary_key:
-                current_pk = new_class._meta['id_field']
-                if current_pk and current_pk != field_name:
-                    raise ValueError('Cannot override primary key field')
+            # Grab any embedded document field unique indexes
+            if field.__class__.__name__ == "EmbeddedDocumentField":
+                field_namespace = "%s." % field_name
+                unique_indexes += cls._unique_with_indexes(field.document_type,
+                                    field_namespace)
 
-                if not current_pk:
-                    new_class._meta['id_field'] = field_name
-                    # Make 'Document.id' an alias to the real primary key field
-                    new_class.id = field
-
-        new_class._meta['unique_indexes'] = unique_indexes
-
-        if not new_class._meta['id_field']:
-            new_class._meta['id_field'] = 'id'
-            new_class._fields['id'] = ObjectIdField(db_field='_id')
-            new_class.id = new_class._fields['id']
-
-        return new_class
+        return unique_indexes
 
 
 class BaseDocument(object):
 
     def __init__(self, **values):
+        signals.pre_init.send(self.__class__, document=self, values=values)
+
         self._data = {}
+        self._initialised = False
         # Assign default values to instance
-        for attr_name in self._fields.keys():
-            # Use default value if present
+        for attr_name, field in self._fields.items():
             value = getattr(self, attr_name, None)
             setattr(self, attr_name, value)
+
         # Assign initial values to instance
         for attr_name in values.keys():
             try:
-                setattr(self, attr_name, values.pop(attr_name))
+                value = values.pop(attr_name)
+                setattr(self, attr_name, value)
             except AttributeError:
                 pass
 
+        # Set any get_fieldname_display methods
+        self.__set_field_display()
+        # Flag initialised
+        self._initialised = True
+        signals.post_init.send(self.__class__, document=self)
+
     def validate(self):
         """Ensure that all fields' values are valid and that required fields
         are present.
         """
         # Get a list of tuples of field names and their current values
-        fields = [(field, getattr(self, name)) 
+        fields = [(field, getattr(self, name))
                   for name, field in self._fields.items()]
 
         # Ensure that each field is matched to a valid value
@@ -356,11 +623,44 @@ class BaseDocument(object):
                 try:
                     field._validate(value)
                 except (ValueError, AttributeError, AssertionError), e:
-                    raise ValidationError('Invalid value for field of type "%s": %s'
-                                          % (field.__class__.__name__, value))
+                    raise ValidationError('Invalid value for field named "%s" of type "%s": %s'
+                                          % (field.name, field.__class__.__name__, value))
             elif field.required:
                 raise ValidationError('Field "%s" is required' % field.name)
 
+    @apply
+    def pk():
+        """Primary key alias
+        """
+        def fget(self):
+            return getattr(self, self._meta['id_field'])
+        def fset(self, value):
+            return setattr(self, self._meta['id_field'], value)
+        return property(fget, fset)
+
+    def to_mongo(self):
+        """Return data dictionary ready for use with MongoDB.
+        """
+        data = {}
+        for field_name, field in self._fields.items():
+            value = getattr(self, field_name, None)
+            if value is not None:
+                data[field.db_field] = field.to_mongo(value)
+        # Only add _cls and _types if allow_inheritance is not False
+        if not (hasattr(self, '_meta') and
+                self._meta.get('allow_inheritance', True) == False):
+            data['_cls'] = self._class_name
+            data['_types'] = self._superclasses.keys() + [self._class_name]
+        if '_id' in data and data['_id'] is None:
+            del data['_id']
+        return data
+
+    @classmethod
+    def _get_collection_name(cls):
+        """Returns the collection name for this class.
+        """
+        return cls._meta.get('collection', None)
+
     @classmethod
     def _get_subclasses(cls):
         """Return a dictionary of all subclasses (found recursively).
@@ -376,15 +676,184 @@ class BaseDocument(object):
             all_subclasses.update(subclass._get_subclasses())
         return all_subclasses
 
-    @apply
-    def pk():
-        """Primary key alias
+    @classmethod
+    def _from_son(cls, son):
+        """Create an instance of a Document (subclass) from a PyMongo SON.
         """
-        def fget(self):
-            return getattr(self, self._meta['id_field'])
-        def fset(self, value):
-            return setattr(self, self._meta['id_field'], value)
-        return property(fget, fset)
+        # get the class name from the document, falling back to the given
+        # class if unavailable
+        class_name = son.get(u'_cls', cls._class_name)
+        data = dict((str(key), value) for key, value in son.items())
+
+        if '_types' in data:
+            del data['_types']
+
+        if '_cls' in data:
+            del data['_cls']
+
+        # Return correct subclass for document type
+        if class_name != cls._class_name:
+            subclasses = cls._get_subclasses()
+            if class_name not in subclasses:
+                # Type of document is probably more generic than the class
+                # that has been queried to return this SON
+                raise NotRegistered("""
+                        `%s` has not been registered in the document registry.
+                        Importing the document class automatically registers it,
+                        has it been imported?
+                    """.strip() % class_name)
+            cls = subclasses[class_name]
+
+        present_fields = data.keys()
+        for field_name, field in cls._fields.items():
+            if field.db_field in data:
+                value = data[field.db_field]
+                data[field_name] = (value if value is None
+                                    else field.to_python(value))
+
+        obj = cls(**data)
+        obj._changed_fields = []
+        return obj
+
+    def _mark_as_changed(self, key):
+        """Marks a key as explicitly changed by the user
+        """
+        if not key:
+            return
+        key = self._db_field_map.get(key, key)
+        if hasattr(self, '_changed_fields') and key not in self._changed_fields:
+            self._changed_fields.append(key)
+
+    def _get_changed_fields(self, key=''):
+        """Returns a list of all fields that have explicitly been changed.
+        """
+        from mongoengine import EmbeddedDocument
+        _changed_fields = []
+        _changed_fields += getattr(self, '_changed_fields', [])
+        for field_name in self._fields:
+            db_field_name = self._db_field_map.get(field_name, field_name)
+            key = '%s.' % db_field_name
+            field = getattr(self, field_name, None)
+            if isinstance(field, EmbeddedDocument) and db_field_name not in _changed_fields:  # Grab all embedded fields that have been changed
+                _changed_fields += ["%s%s" % (key, k) for k in field._get_changed_fields(key) if k]
+            elif isinstance(field, (list, tuple, dict)) and db_field_name not in _changed_fields:  # Loop list / dict fields as they contain documents
+                # Determine the iterator to use
+                if not hasattr(field, 'items'):
+                    iterator = enumerate(field)
+                else:
+                    iterator = field.iteritems()
+                for index, value in iterator:
+                    if not hasattr(value, '_get_changed_fields'):
+                        continue
+                    list_key = "%s%s." % (key, index)
+                    _changed_fields += ["%s%s" % (list_key, k) for k in value._get_changed_fields(list_key) if k]
+
+        return _changed_fields
+
+    def _delta(self):
+        """Returns the delta (set, unset) of the changes for a document.
+        Gets any values that have been explicitly changed.
+        """
+        # Handles cases where not loaded from_son but has _id
+        doc = self.to_mongo()
+        set_fields = self._get_changed_fields()
+        set_data = {}
+        unset_data = {}
+        if hasattr(self, '_changed_fields'):
+            set_data = {}
+            # Fetch each set item from its path
+            for path in set_fields:
+                parts = path.split('.')
+                d = doc
+                for p in parts:
+                    if hasattr(d, '__getattr__'):
+                        d = getattr(p, d)
+                    elif p.isdigit():
+                        d = d[int(p)]
+                    else:
+                        d = d.get(p)
+                set_data[path] = d
+        else:
+            set_data = doc
+            if '_id' in set_data:
+                del(set_data['_id'])
+
+        # Determine if any changed items were actually unset.
+        for path, value in set_data.items():
+            if value:
+                continue
+
+            # If we've set a value that ain't the default value dont unset it.
+            default = None
+
+            if path in self._fields:
+                default = self._fields[path].default
+            else:  # Perform a full lookup for lists / embedded lookups
+                d = self
+                parts = path.split('.')
+                db_field_name = parts.pop()
+                for p in parts:
+                    if p.isdigit():
+                        d = d[int(p)]
+                    elif hasattr(d, '__getattribute__') and not isinstance(d, dict):
+                        real_path = d._reverse_db_field_map.get(p, p)
+                        d = getattr(d, real_path)
+                    else:
+                        d = d.get(p)
+
+                if hasattr(d, '_fields'):
+                    field_name = d._reverse_db_field_map.get(db_field_name,
+                                                             db_field_name)
+
+                    default = d._fields[field_name].default
+
+            if default is not None:
+                if callable(default):
+                    default = default()
+            if default != value:
+                continue
+
+            del(set_data[path])
+            unset_data[path] = 1
+        return set_data, unset_data
+
+    @classmethod
+    def _geo_indices(cls, inspected_classes=None):
+        inspected_classes = inspected_classes or []
+        geo_indices = []
+        inspected_classes.append(cls)
+        for field in cls._fields.values():
+            if hasattr(field, 'document_type'):
+                field_cls = field.document_type
+                if field_cls in inspected_classes:
+                    continue
+                if hasattr(field_cls, '_geo_indices'):
+                    geo_indices += field_cls._geo_indices(inspected_classes)
+            elif field._geo_index:
+                geo_indices.append(field)
+        return geo_indices
+
+    def __getstate__(self):
+        self_dict = self.__dict__
+        removals = ["get_%s_display" % k for k,v in self._fields.items() if v.choices]
+        for k in removals:
+            if hasattr(self, k):
+                delattr(self, k)
+        return self.__dict__
+
+    def __setstate__(self, __dict__):
+        self.__dict__ = __dict__
+        self.__set_field_display()
+
+    def __set_field_display(self):
+        for attr_name, field in self._fields.items():
+            if field.choices:  # dynamically adds a way to get the display value for a field with choices
+                setattr(self, 'get_%s_display' % attr_name, partial(self.__get_field_display, field=field))
+
+    def __get_field_display(self, field):
+        """Returns the display value for a choice field"""
+        value = getattr(self, field.name)
+        return dict(field.choices).get(value, value)
 
     def __iter__(self):
         return iter(self._fields)
@@ -429,69 +898,125 @@ class BaseDocument(object):
             return unicode(self).encode('utf-8')
         return '%s object' % self.__class__.__name__
 
-    def to_mongo(self):
-        """Return data dictionary ready for use with MongoDB.
-        """
-        data = {}
-        for field_name, field in self._fields.items():
-            value = getattr(self, field_name, None)
-            if value is not None:
-                data[field.db_field] = field.to_mongo(value)
-        # Only add _cls and _types if allow_inheritance is not False
-        if not (hasattr(self, '_meta') and
-                self._meta.get('allow_inheritance', True) == False):
-            data['_cls'] = self._class_name
-            data['_types'] = self._superclasses.keys() + [self._class_name]
-        if data.has_key('_id') and not data['_id']:
-            del data['_id']
-        return data
-
-    @classmethod
-    def _from_son(cls, son):
-        """Create an instance of a Document (subclass) from a PyMongo SON.
-        """
-        # get the class name from the document, falling back to the given
-        # class if unavailable
-        class_name = son.get(u'_cls', cls._class_name)
-
-        data = dict((str(key), value) for key, value in son.items())
-
-        if '_types' in data:
-            del data['_types']
-
-        if '_cls' in data:
-            del data['_cls']
-
-        # Return correct subclass for document type
-        if class_name != cls._class_name:
-            subclasses = cls._get_subclasses()
-            if class_name not in subclasses:
-                # Type of document is probably more generic than the class
-                # that has been queried to return this SON
-                return None
-            cls = subclasses[class_name]
-
-        present_fields = data.keys()
-
-        for field_name, field in cls._fields.items():
-            if field.db_field in data:
-                value = data[field.db_field]
-                data[field_name] = (value if value is None
-                                    else field.to_python(value))
-
-        obj = cls(**data)
-        obj._present_fields = present_fields
-        return obj
-
     def __eq__(self, other):
         if isinstance(other, self.__class__) and hasattr(other, 'id'):
             if self.id == other.id:
                 return True
         return False
 
+    def __ne__(self, other):
+        return not self.__eq__(other)
+
+    def __hash__(self):
+        if self.pk is None:
+            # For new object
+            return super(BaseDocument,self).__hash__()
+        else:
+            return hash(self.pk)
+
+
+class BaseList(list):
+    """A special list so we can watch any changes
+    """
+
+    def __init__(self, list_items, instance, name):
+        self.instance = instance
+        self.name = name
+        super(BaseList, self).__init__(list_items)
+
+    def __setitem__(self, *args, **kwargs):
+        self._mark_as_changed()
+        super(BaseList, self).__setitem__(*args, **kwargs)
+
+    def __delitem__(self, *args, **kwargs):
+        self._mark_as_changed()
+        super(BaseList, self).__delitem__(*args, **kwargs)
+
+    def append(self, *args, **kwargs):
+        self._mark_as_changed()
+        return super(BaseList, self).append(*args, **kwargs)
+
+    def extend(self, *args, **kwargs):
+        self._mark_as_changed()
+        return super(BaseList, self).extend(*args, **kwargs)
+
+    def insert(self, *args, **kwargs):
+        self._mark_as_changed()
+        return super(BaseList, self).insert(*args, **kwargs)
+
+    def pop(self, *args, **kwargs):
+        self._mark_as_changed()
+        return super(BaseList, self).pop(*args, **kwargs)
+
+    def remove(self, *args, **kwargs):
+        self._mark_as_changed()
+        return super(BaseList, self).remove(*args, **kwargs)
+
+    def reverse(self, *args, **kwargs):
+        self._mark_as_changed()
+        return super(BaseList, self).reverse(*args, **kwargs)
+
+    def sort(self, *args, **kwargs):
+        self._mark_as_changed()
+        return super(BaseList, self).sort(*args, **kwargs)
+
+    def _mark_as_changed(self):
+        """Marks a list as changed if has an instance and a name"""
+        if hasattr(self, 'instance') and hasattr(self, 'name'):
+            self.instance._mark_as_changed(self.name)
+
+
+class BaseDict(dict):
+    """A special dict so we can watch any changes
+    """
+
+    def __init__(self, dict_items, instance, name):
+        self.instance = instance
+        self.name = name
+        super(BaseDict, self).__init__(dict_items)
+
+    def __setitem__(self, *args, **kwargs):
+        self._mark_as_changed()
+        super(BaseDict, self).__setitem__(*args, **kwargs)
+
+    def __setattr__(self, *args, **kwargs):
+        self._mark_as_changed()
+        super(BaseDict, self).__setattr__(*args, **kwargs)
+
+    def __delete__(self, *args, **kwargs):
+        self._mark_as_changed()
+        super(BaseDict, self).__delete__(*args, **kwargs)
+
+    def __delitem__(self, *args, **kwargs):
+        self._mark_as_changed()
+        super(BaseDict, self).__delitem__(*args, **kwargs)
+
+    def __delattr__(self, *args, **kwargs):
+        self._mark_as_changed()
+        super(BaseDict, self).__delattr__(*args, **kwargs)
+
+    def clear(self, *args, **kwargs):
+        self._mark_as_changed()
+        super(BaseDict, self).clear(*args, **kwargs)
+
+    def pop(self, *args, **kwargs):
+        self._mark_as_changed()
+        super(BaseDict, self).clear(*args, **kwargs)
+
+    def popitem(self, *args, **kwargs):
+        self._mark_as_changed()
+        super(BaseDict, self).clear(*args, **kwargs)
+
+    def _mark_as_changed(self):
+        """Marks a dict as changed if has an instance and a name"""
+        if hasattr(self, 'instance') and hasattr(self, 'name'):
+            self.instance._mark_as_changed(self.name)
+
 if sys.version_info < (2, 5):
     # Prior to Python 2.5, Exception was an old-style class
+    import types
     def subclass_exception(name, parents, unused):
+        import types
         return types.ClassType(name, parents, {})
 else:
     def subclass_exception(name, parents, module):

mongoengine/connection.py

@@ -1,5 +1,6 @@
 from pymongo import Connection
 import multiprocessing
+import threading
 
 __all__ = ['ConnectionError', 'connect']
 
@@ -22,17 +23,22 @@ class ConnectionError(Exception):
 
 
 def _get_connection(reconnect=False):
+    """Handles the connection to the database
+    """
     global _connection
     identity = get_identity()
     # Connect to the database if not already connected
     if _connection.get(identity) is None or reconnect:
         try:
             _connection[identity] = Connection(**_connection_settings)
-        except:
-            raise ConnectionError('Cannot connect to the database')
+        except Exception, e:
+            raise ConnectionError("Cannot connect to the database:\n%s" % e)
     return _connection[identity]
 
 def _get_db(reconnect=False):
+    """Handles database connections and authentication based on the current
+    identity
+    """
     global _db, _connection
     identity = get_identity()
     # Connect if not already connected
@@ -52,12 +58,17 @@ def _get_db(reconnect=False):
     return _db[identity]
 
 def get_identity():
+    """Creates an identity key based on the current process and thread
+    identity.
+    """
     identity = multiprocessing.current_process()._identity
     identity = 0 if not identity else identity[0]
+
+    identity = (identity, threading.current_thread().ident)
     return identity
-    
+
 def connect(db, username=None, password=None, **kwargs):
-    """Connect to the database specified by the 'db' argument. Connection 
+    """Connect to the database specified by the 'db' argument. Connection
     settings may be provided here as well if the database is not running on
     the default port on localhost. If authentication is needed, provide
     username and password arguments as well.

mongoengine/dereference.py

@@ -0,0 +1,184 @@
+import operator
+
+import pymongo
+
+from base import BaseDict, BaseList, get_document, TopLevelDocumentMetaclass
+from fields import ReferenceField
+from connection import _get_db
+from queryset import QuerySet
+from document import Document
+
+
+class DeReference(object):
+
+    def __call__(self, items, max_depth=1, instance=None, name=None, get=False):
+        """
+        Cheaply dereferences the items to a set depth.
+        Also handles the convertion of complex data types.
+
+        :param items: The iterable (dict, list, queryset) to be dereferenced.
+        :param max_depth: The maximum depth to recurse to
+        :param instance: The owning instance used for tracking changes by
+            :class:`~mongoengine.base.ComplexBaseField`
+        :param name: The name of the field, used for tracking changes by
+            :class:`~mongoengine.base.ComplexBaseField`
+        :param get: A boolean determining if being called by __get__
+        """
+        if items is None or isinstance(items, basestring):
+            return items
+
+        # cheapest way to convert a queryset to a list
+        # list(queryset) uses a count() query to determine length
+        if isinstance(items, QuerySet):
+            items = [i for i in items]
+
+        self.max_depth = max_depth
+        
+        doc_type = None
+        if instance and instance._fields:
+            doc_type = instance._fields[name].field
+    
+            if isinstance(doc_type, ReferenceField):
+                doc_type = doc_type.document_type
+
+        self.reference_map = self._find_references(items)
+        self.object_map = self._fetch_objects(doc_type=doc_type)
+        return self._attach_objects(items, 0, instance, name, get)
+
+    def _find_references(self, items, depth=0):
+        """
+        Recursively finds all db references to be dereferenced
+
+        :param items: The iterable (dict, list, queryset)
+        :param depth: The current depth of recursion
+        """
+        reference_map = {}
+        if not items:
+            return reference_map
+
+        # Determine the iterator to use
+        if not hasattr(items, 'items'):
+            iterator = enumerate(items)
+        else:
+            iterator = items.iteritems()
+
+        # Recursively find dbreferences
+        for k, item in iterator:
+            if hasattr(item, '_fields'):
+                for field_name, field in item._fields.iteritems():
+                    v = item._data.get(field_name, None)
+                    if isinstance(v, (pymongo.dbref.DBRef)):
+                        reference_map.setdefault(field.document_type, []).append(v.id)
+                    elif isinstance(v, (dict, pymongo.son.SON)) and '_ref' in v:
+                        reference_map.setdefault(get_document(v['_cls']), []).append(v['_ref'].id)
+                    elif isinstance(v, (dict, list, tuple)) and depth <= self.max_depth:
+                        field_cls = getattr(getattr(field, 'field', None), 'document_type', None)
+                        references = self._find_references(v, depth)
+                        for key, refs in references.iteritems():
+                            if isinstance(field_cls, (Document, TopLevelDocumentMetaclass)):
+                                key = field_cls
+                            reference_map.setdefault(key, []).extend(refs)
+            elif isinstance(item, (pymongo.dbref.DBRef)):
+                reference_map.setdefault(item.collection, []).append(item.id)
+            elif isinstance(item, (dict, pymongo.son.SON)) and '_ref' in item:
+                reference_map.setdefault(get_document(item['_cls']), []).append(item['_ref'].id)
+            elif isinstance(item, (dict, list, tuple)) and depth <= self.max_depth:
+                references = self._find_references(item, depth)
+                for key, refs in references.iteritems():
+                    reference_map.setdefault(key, []).extend(refs)
+        depth += 1
+        return reference_map
+
+    def _fetch_objects(self, doc_type=None):
+        """Fetch all references and convert to their document objects
+        """
+        object_map = {}
+        for col, dbrefs in self.reference_map.iteritems():
+            keys = object_map.keys()
+            refs = list(set([dbref for dbref in dbrefs if str(dbref) not in keys]))
+            if hasattr(col, 'objects'):  # We have a document class for the refs
+                references = col.objects.in_bulk(refs)
+                for key, doc in references.iteritems():
+                    object_map[key] = doc
+            else:  # Generic reference: use the refs data to convert to document
+                references = _get_db()[col].find({'_id': {'$in': refs}})
+                for ref in references:
+                    if '_cls' in ref:
+                        doc = get_document(ref['_cls'])._from_son(ref)
+                    else:
+                        doc = doc_type._from_son(ref)
+                    object_map[doc.id] = doc
+        return object_map
+
+    def _attach_objects(self, items, depth=0, instance=None, name=None, get=False):
+        """
+        Recursively finds all db references to be dereferenced
+
+        :param items: The iterable (dict, list, queryset)
+        :param depth: The current depth of recursion
+        :param instance: The owning instance used for tracking changes by
+            :class:`~mongoengine.base.ComplexBaseField`
+        :param name: The name of the field, used for tracking changes by
+            :class:`~mongoengine.base.ComplexBaseField`
+        :param get: A boolean determining if being called by __get__
+        """
+        if not items:
+            if isinstance(items, (BaseDict, BaseList)):
+                return items
+
+            if instance:
+                if isinstance(items, dict):
+                    return BaseDict(items, instance=instance, name=name)
+                else:
+                    return BaseList(items, instance=instance, name=name)
+
+        if isinstance(items, (dict, pymongo.son.SON)):
+            if '_ref' in items:
+                return self.object_map.get(items['_ref'].id, items)
+            elif '_types' in items and '_cls' in items:
+                doc = get_document(items['_cls'])._from_son(items)
+                if not get:
+                    doc._data = self._attach_objects(doc._data, depth, doc, name, get)
+                return doc
+
+        if not hasattr(items, 'items'):
+            is_list = True
+            iterator = enumerate(items)
+            data = []
+        else:
+            is_list = False
+            iterator = items.iteritems()
+            data = {}
+
+        for k, v in iterator:
+            if is_list:
+                data.append(v)
+            else:
+                data[k] = v
+
+            if k in self.object_map:
+                data[k] = self.object_map[k]
+            elif hasattr(v, '_fields'):
+                for field_name, field in v._fields.iteritems():
+                    v = data[k]._data.get(field_name, None)
+                    if isinstance(v, (pymongo.dbref.DBRef)):
+                        data[k]._data[field_name] = self.object_map.get(v.id, v)
+                    elif isinstance(v, (dict, pymongo.son.SON)) and '_ref' in v:
+                        data[k]._data[field_name] = self.object_map.get(v['_ref'].id, v)
+                    elif isinstance(v, dict) and depth < self.max_depth:
+                        data[k]._data[field_name] = self._attach_objects(v, depth, instance=instance, name=name, get=get)
+                    elif isinstance(v, (list, tuple)):
+                        data[k]._data[field_name] = self._attach_objects(v, depth, instance=instance, name=name, get=get)
+            elif isinstance(v, (dict, list, tuple)) and depth < self.max_depth:
+                data[k] = self._attach_objects(v, depth, instance=instance, name=name, get=get)
+            elif hasattr(v, 'id'):
+                data[k] = self.object_map.get(v.id, v)
+
+        if instance and name:
+            if is_list:
+                return BaseList(data, instance=instance, name=name)
+            return BaseDict(data, instance=instance, name=name)
+        depth += 1
+        return data
+
+dereference = DeReference()

mongoengine/django/auth.py

@@ -3,6 +3,7 @@ from mongoengine import *
 from django.utils.hashcompat import md5_constructor, sha_constructor
 from django.utils.encoding import smart_str
 from django.contrib.auth.models import AnonymousUser
+from django.utils.translation import ugettext_lazy as _
 
 import datetime
 
@@ -21,16 +22,38 @@ class User(Document):
     """A User document that aims to mirror most of the API specified by Django
     at http://docs.djangoproject.com/en/dev/topics/auth/#users
     """
-    username = StringField(max_length=30, required=True)
-    first_name = StringField(max_length=30)
-    last_name = StringField(max_length=30)
-    email = StringField()
-    password = StringField(max_length=128)
-    is_staff = BooleanField(default=False)
-    is_active = BooleanField(default=True)
-    is_superuser = BooleanField(default=False)
-    last_login = DateTimeField(default=datetime.datetime.now)
-    date_joined = DateTimeField(default=datetime.datetime.now)
+    username = StringField(max_length=30, required=True,
+                           verbose_name=_('username'),
+                           help_text=_("Required. 30 characters or fewer. Letters, numbers and @/./+/-/_ characters"))
+
+    first_name = StringField(max_length=30,
+                             verbose_name=_('first name'))
+
+    last_name = StringField(max_length=30,
+                            verbose_name=_('last name'))
+    email = EmailField(verbose_name=_('e-mail address'))
+    password = StringField(max_length=128,
+                           verbose_name=_('password'),
+                           help_text=_("Use '[algo]$[salt]$[hexdigest]' or use the <a href=\"password/\">change password form</a>."))
+    is_staff = BooleanField(default=False,
+                            verbose_name=_('staff status'),
+                            help_text=_("Designates whether the user can log into this admin site."))
+    is_active = BooleanField(default=True,
+                             verbose_name=_('active'),
+                             help_text=_("Designates whether this user should be treated as active. Unselect this instead of deleting accounts."))
+    is_superuser = BooleanField(default=False,
+                                verbose_name=_('superuser status'),
+                                help_text=_("Designates that this user has all permissions without explicitly assigning them."))
+    last_login = DateTimeField(default=datetime.datetime.now,
+                               verbose_name=_('last login'))
+    date_joined = DateTimeField(default=datetime.datetime.now,
+                                verbose_name=_('date joined'))
+
+    meta = {
+        'indexes': [
+            {'fields': ['username'], 'unique': True}
+        ]
+    }
 
     def __unicode__(self):
         return self.username
@@ -86,7 +109,7 @@ class User(Document):
             else:
                 email = '@'.join([email_name, domain_part.lower()])
 
-        user = User(username=username, email=email, date_joined=now)
+        user = cls(username=username, email=email, date_joined=now)
         user.set_password(password)
         user.save()
         return user
@@ -99,6 +122,10 @@ class MongoEngineBackend(object):
     """Authenticate using MongoEngine and mongoengine.django.auth.User.
     """
 
+    supports_object_permissions = False
+    supports_anonymous_user = False
+    supports_inactive_user = False
+
     def authenticate(self, username=None, password=None):
         user = User.objects(username=username).first()
         if user:

mongoengine/django/shortcuts.py

@@ -1,6 +1,7 @@
 from django.http import Http404
 from mongoengine.queryset import QuerySet
 from mongoengine.base import BaseDocument
+from mongoengine.base import ValidationError
 
 def _get_queryset(cls):
     """Inspired by django.shortcuts.*"""
@@ -25,7 +26,7 @@ def get_document_or_404(cls, *args, **kwargs):
     queryset = _get_queryset(cls)
     try:
         return queryset.get(*args, **kwargs)
-    except queryset._document.DoesNotExist:
+    except (queryset._document.DoesNotExist, ValidationError):
         raise Http404('No %s matches the given query.' % queryset._document._class_name)
 
 def get_list_or_404(cls, *args, **kwargs):

mongoengine/document.py

@@ -1,12 +1,17 @@
+from mongoengine import signals
 from base import (DocumentMetaclass, TopLevelDocumentMetaclass, BaseDocument,
-                  ValidationError)
+                  ValidationError, BaseDict, BaseList)
 from queryset import OperationError
 from connection import _get_db
 
 import pymongo
 
+__all__ = ['Document', 'EmbeddedDocument', 'ValidationError',
+           'OperationError', 'InvalidCollectionError']
 
-__all__ = ['Document', 'EmbeddedDocument', 'ValidationError', 'OperationError']
+
+class InvalidCollectionError(Exception):
+    pass
 
 
 class EmbeddedDocument(BaseDocument):
@@ -18,6 +23,18 @@ class EmbeddedDocument(BaseDocument):
 
     __metaclass__ = DocumentMetaclass
 
+    def __delattr__(self, *args, **kwargs):
+        """Handle deletions of fields"""
+        field_name = args[0]
+        if field_name in self._fields:
+            default = self._fields[field_name].default
+            if callable(default):
+                default = default()
+            setattr(self, field_name, default)
+        else:
+            super(EmbeddedDocument, self).__delattr__(*args, **kwargs)
+
+
 
 class Document(BaseDocument):
     """The base class used for defining the structure and properties of
@@ -40,44 +57,125 @@ class Document(BaseDocument):
     presence of `_cls` and `_types`, set :attr:`allow_inheritance` to
     ``False`` in the :attr:`meta` dictionary.
 
-    A :class:`~mongoengine.Document` may use a **Capped Collection** by 
+    A :class:`~mongoengine.Document` may use a **Capped Collection** by
     specifying :attr:`max_documents` and :attr:`max_size` in the :attr:`meta`
     dictionary. :attr:`max_documents` is the maximum number of documents that
-    is allowed to be stored in the collection, and :attr:`max_size` is the 
-    maximum size of the collection in bytes. If :attr:`max_size` is not 
-    specified and :attr:`max_documents` is, :attr:`max_size` defaults to 
+    is allowed to be stored in the collection, and :attr:`max_size` is the
+    maximum size of the collection in bytes. If :attr:`max_size` is not
+    specified and :attr:`max_documents` is, :attr:`max_size` defaults to
     10000000 bytes (10MB).
 
     Indexes may be created by specifying :attr:`indexes` in the :attr:`meta`
-    dictionary. The value should be a list of field names or tuples of field 
+    dictionary. The value should be a list of field names or tuples of field
     names. Index direction may be specified by prefixing the field names with
     a **+** or **-** sign.
-    """
 
+    By default, _types will be added to the start of every index (that
+    doesn't contain a list) if allow_inheritence is True. This can be
+    disabled by either setting types to False on the specific index or
+    by setting index_types to False on the meta dictionary for the document.
+    """
     __metaclass__ = TopLevelDocumentMetaclass
 
-    def save(self, safe=True, force_insert=False, validate=True):
+    @classmethod
+    def _get_collection(self):
+        """Returns the collection for the document."""
+        db = _get_db()
+        collection_name = self._get_collection_name()
+
+        if not hasattr(self, '_collection') or self._collection is None:
+            # Create collection as a capped collection if specified
+            if self._meta['max_size'] or self._meta['max_documents']:
+                # Get max document limit and max byte size from meta
+                max_size = self._meta['max_size'] or 10000000  # 10MB default
+                max_documents = self._meta['max_documents']
+
+                if collection_name in db.collection_names():
+                    self._collection = db[collection_name]
+                    # The collection already exists, check if its capped
+                    # options match the specified capped options
+                    options = self._collection.options()
+                    if options.get('max') != max_documents or \
+                       options.get('size') != max_size:
+                        msg = ('Cannot create collection "%s" as a capped '
+                               'collection as it already exists') % self._collection
+                        raise InvalidCollectionError(msg)
+                else:
+                    # Create the collection as a capped collection
+                    opts = {'capped': True, 'size': max_size}
+                    if max_documents:
+                        opts['max'] = max_documents
+                    self._collection = db.create_collection(
+                        collection_name, **opts
+                    )
+            else:
+                self._collection = db[collection_name]
+        return self._collection
+
+    def save(self, safe=True, force_insert=False, validate=True, write_options=None, _refs=None):
         """Save the :class:`~mongoengine.Document` to the database. If the
         document already exists, it will be updated, otherwise it will be
         created.
 
-        If ``safe=True`` and the operation is unsuccessful, an 
+        If ``safe=True`` and the operation is unsuccessful, an
         :class:`~mongoengine.OperationError` will be raised.
 
         :param safe: check if the operation succeeded before returning
-        :param force_insert: only try to create a new document, don't allow 
+        :param force_insert: only try to create a new document, don't allow
             updates of existing documents
-        :param validate: validates the document; set to ``False`` for skiping
+        :param validate: validates the document; set to ``False`` to skip.
+        :param write_options: Extra keyword arguments are passed down to
+                :meth:`~pymongo.collection.Collection.save` OR
+                :meth:`~pymongo.collection.Collection.insert`
+                which will be used as options for the resultant ``getLastError`` command.
+                For example, ``save(..., 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.
+
+        .. versionchanged:: 0.5
+            In existing documents it only saves changed fields using set / unset
+            Saves are cascaded and any :class:`~pymongo.dbref.DBRef` objects
+            that have changes are saved as well.
         """
+        from fields import ReferenceField, GenericReferenceField
+
+        signals.pre_save.send(self.__class__, document=self)
+
         if validate:
             self.validate()
+
+        if not write_options:
+            write_options = {}
+
         doc = self.to_mongo()
+
+        created = '_id' in doc
+        creation_mode = force_insert or not created
         try:
             collection = self.__class__.objects._collection
-            if force_insert:
-                object_id = collection.insert(doc, safe=safe)
+            if creation_mode:
+                if force_insert:
+                    object_id = collection.insert(doc, safe=safe, **write_options)
+                else:
+                    object_id = collection.save(doc, safe=safe, **write_options)
             else:
-                object_id = collection.save(doc, safe=safe)
+                object_id = doc['_id']
+                updates, removals = self._delta()
+                if updates:
+                    collection.update({'_id': object_id}, {"$set": updates}, upsert=True, safe=safe, **write_options)
+                if removals:
+                    collection.update({'_id': object_id}, {"$unset": removals}, upsert=True, safe=safe, **write_options)
+
+            # Save any references / generic references
+            _refs = _refs or []
+            for name, cls in self._fields.items():
+                if isinstance(cls, (ReferenceField, GenericReferenceField)):
+                    ref = getattr(self, name)
+                    if ref and str(ref) not in _refs:
+                        _refs.append(str(ref))
+                        ref.save(safe=safe, force_insert=force_insert,
+                                 validate=validate, write_options=write_options,
+                                 _refs=_refs)
+
         except pymongo.errors.OperationFailure, err:
             message = 'Could not save document (%s)'
             if u'duplicate key' in unicode(err):
@@ -86,12 +184,42 @@ class Document(BaseDocument):
         id_field = self._meta['id_field']
         self[id_field] = self._fields[id_field].to_python(object_id)
 
+        def reset_changed_fields(doc, inspected_docs=None):
+            """Loop through and reset changed fields lists"""
+
+            inspected_docs = inspected_docs or []
+            inspected_docs.append(doc)
+            if hasattr(doc, '_changed_fields'):
+                doc._changed_fields = []
+
+            for field_name in doc._fields:
+                field = getattr(doc, field_name)
+                if field not in inspected_docs and hasattr(field, '_changed_fields'):
+                    reset_changed_fields(field, inspected_docs)
+
+        reset_changed_fields(self)
+        signals.post_save.send(self.__class__, document=self, created=creation_mode)
+
+    def update(self, **kwargs):
+        """Performs an update on the :class:`~mongoengine.Document`
+        A convenience wrapper to :meth:`~mongoengine.QuerySet.update`.
+
+        Raises :class:`OperationError` if called on an object that has not yet
+        been saved.
+        """
+        if not self.pk:
+            raise OperationError('attempt to update a document not yet saved')
+
+        return self.__class__.objects(pk=self.pk).update_one(**kwargs)
+
     def delete(self, safe=False):
         """Delete the :class:`~mongoengine.Document` from the database. This
         will only take effect if the document has been previously saved.
 
         :param safe: check if the operation succeeded before returning
         """
+        signals.pre_delete.send(self.__class__, document=self)
+
         id_field = self._meta['id_field']
         object_id = self._fields[id_field].to_mongo(self[id_field])
         try:
@@ -100,6 +228,18 @@ class Document(BaseDocument):
             message = u'Could not delete document (%s)' % err.message
             raise OperationError(message)
 
+        signals.post_delete.send(self.__class__, document=self)
+
+    def select_related(self, max_depth=1):
+        """Handles dereferencing of :class:`~pymongo.dbref.DBRef` objects to
+        a maximum depth in order to cut down the number queries to mongodb.
+
+        .. versionadded:: 0.5
+        """
+        from dereference import dereference
+        self._data = dereference(self._data, max_depth)
+        return self
+
     def reload(self):
         """Reloads all attributes from the database.
 
@@ -108,7 +248,37 @@ class Document(BaseDocument):
         id_field = self._meta['id_field']
         obj = self.__class__.objects(**{id_field: self[id_field]}).first()
         for field in self._fields:
-            setattr(self, field, obj[field])
+            setattr(self, field, self._reload(field, obj[field]))
+        self._changed_fields = []
+
+    def _reload(self, key, value):
+        """Used by :meth:`~mongoengine.Document.reload` to ensure the
+        correct instance is linked to self.
+        """
+        if isinstance(value, BaseDict):
+            value = [(k, self._reload(k,v)) for k,v in value.items()]
+            value = BaseDict(value, instance=self, name=key)
+        elif isinstance(value, BaseList):
+            value = [self._reload(key, v) for v in value]
+            value = BaseList(value, instance=self, name=key)
+        elif isinstance(value, EmbeddedDocument):
+            value._changed_fields = []
+        return value
+
+    def to_dbref(self):
+        """Returns an instance of :class:`~pymongo.dbref.DBRef` useful in
+        `__raw__` queries."""
+        if not self.pk:
+            msg = "Only saved documents can have a valid dbref"
+            raise OperationError(msg)
+        return pymongo.dbref.DBRef(self.__class__._get_collection_name(), self.pk)
+
+    @classmethod
+    def register_delete_rule(cls, document_cls, field_name, rule):
+        """This method registers the delete rules to apply when removing this
+        object.
+        """
+        cls._meta['delete_rules'][(document_cls, field_name)] = rule
 
     @classmethod
     def drop_collection(cls):
@@ -116,16 +286,16 @@ class Document(BaseDocument):
         :class:`~mongoengine.Document` type from the database.
         """
         db = _get_db()
-        db.drop_collection(cls._meta['collection'])
+        db.drop_collection(cls._get_collection_name())
 
 
 class MapReduceDocument(object):
     """A document returned from a map/reduce query.
 
     :param collection: An instance of :class:`~pymongo.Collection`
-    :param key: Document/result key, often an instance of 
-                :class:`~pymongo.objectid.ObjectId`. If supplied as 
-                an ``ObjectId`` found in the given ``collection``, 
+    :param key: Document/result key, often an instance of
+                :class:`~pymongo.objectid.ObjectId`. If supplied as
+                an ``ObjectId`` found in the given ``collection``,
                 the object can be accessed via the ``object`` property.
     :param value: The result(s) for this key.
 
@@ -140,7 +310,7 @@ class MapReduceDocument(object):
 
     @property
     def object(self):
-        """Lazy-load the object referenced by ``self.key``. ``self.key`` 
+        """Lazy-load the object referenced by ``self.key``. ``self.key``
         should be the ``primary_key``.
         """
         id_field = self._document()._meta['id_field']

mongoengine/fields.py

@@ -1,4 +1,6 @@
-from base import BaseField, ObjectIdField, ValidationError, get_document
+from base import (BaseField, ComplexBaseField, ObjectIdField,
+                  ValidationError, get_document)
+from queryset import DO_NOTHING
 from document import Document, EmbeddedDocument
 from connection import _get_db
 from operator import itemgetter
@@ -8,18 +10,18 @@ import pymongo
 import pymongo.dbref
 import pymongo.son
 import pymongo.binary
-import datetime
+import datetime, time
 import decimal
 import gridfs
-import warnings
-import types
 
 
 __all__ = ['StringField', 'IntField', 'FloatField', 'BooleanField',
            'DateTimeField', 'EmbeddedDocumentField', 'ListField', 'DictField',
-           'ObjectIdField', 'ReferenceField', 'ValidationError',
-           'DecimalField', 'URLField', 'GenericReferenceField', 'FileField',
-           'BinaryField', 'SortedListField', 'EmailField', 'GeoPointField']
+           'ObjectIdField', 'ReferenceField', 'ValidationError', 'MapField',
+           'DecimalField', 'ComplexDateTimeField', 'URLField',
+           'GenericReferenceField', 'FileField', 'BinaryField',
+           'SortedListField', 'EmailField', 'GeoPointField',
+           'SequenceField', 'GenericEmbeddedDocumentField']
 
 RECURSIVE_REFERENCE_CONSTANT = 'self'
 
@@ -118,8 +120,8 @@ class EmailField(StringField):
 
     EMAIL_REGEX = re.compile(
         r"(^[-!#$%&'*+/=?^_`{}|~0-9A-Z]+(\.[-!#$%&'*+/=?^_`{}|~0-9A-Z]+)*"  # dot-atom
-        r'|^"([\001-\010\013\014\016-\037!#-\[\]-\177]|\\[\001-011\013\014\016-\177])*"' # quoted-string
-        r')@(?:[A-Z0-9](?:[A-Z0-9-]{0,61}[A-Z0-9])?\.)+[A-Z]{2,6}\.?$', re.IGNORECASE # domain
+        r'|^"([\001-\010\013\014\016-\037!#-\[\]-\177]|\\[\001-011\013\014\016-\177])*"'  # quoted-string
+        r')@(?:[A-Z0-9](?:[A-Z0-9-]{0,61}[A-Z0-9])?\.)+[A-Z]{2,6}\.?$', re.IGNORECASE  # domain
     )
 
     def validate(self, value):
@@ -150,6 +152,9 @@ class IntField(BaseField):
         if self.max_value is not None and value > self.max_value:
             raise ValidationError('Integer value is too large')
 
+    def prepare_query_value(self, op, value):
+        return int(value)
+
 
 class FloatField(BaseField):
     """An floating point number field.
@@ -173,6 +178,10 @@ class FloatField(BaseField):
         if self.max_value is not None and value > self.max_value:
             raise ValidationError('Float value is too large')
 
+    def prepare_query_value(self, op, value):
+        return float(value)
+
+
 class DecimalField(BaseField):
     """A fixed-point decimal number field.
 
@@ -222,15 +231,151 @@ class BooleanField(BaseField):
 
 class DateTimeField(BaseField):
     """A datetime field.
+
+    Note: Microseconds are rounded to the nearest millisecond.
+      Pre UTC microsecond support is effecively broken.
+      Use :class:`~mongoengine.fields.ComplexDateTimeField` if you
+      need accurate microsecond support.
     """
 
     def validate(self, value):
-        assert isinstance(value, datetime.datetime)
+        assert isinstance(value, (datetime.datetime, datetime.date))
+
+    def to_mongo(self, value):
+        return self.prepare_query_value(None, value)
+
+    def prepare_query_value(self, op, value):
+        if value is None:
+            return value
+        if isinstance(value, datetime.datetime):
+            return value
+        if isinstance(value, datetime.date):
+            return datetime.datetime(value.year, value.month, value.day)
+
+        # Attempt to parse a datetime:
+        # value = smart_str(value)
+        # split usecs, because they are not recognized by strptime.
+        if '.' in value:
+            try:
+                value, usecs = value.split('.')
+                usecs = int(usecs)
+            except ValueError:
+                return None
+        else:
+            usecs = 0
+        kwargs = {'microsecond': usecs}
+        try:  # Seconds are optional, so try converting seconds first.
+            return datetime.datetime(*time.strptime(value, '%Y-%m-%d %H:%M:%S')[:6],
+                                     **kwargs)
+        except ValueError:
+            try:  # Try without seconds.
+                return datetime.datetime(*time.strptime(value, '%Y-%m-%d %H:%M')[:5],
+                                         **kwargs)
+            except ValueError:  # Try without hour/minutes/seconds.
+                try:
+                    return datetime.datetime(*time.strptime(value, '%Y-%m-%d')[:3],
+                                             **kwargs)
+                except ValueError:
+                    return None
+
+
+class ComplexDateTimeField(StringField):
+    """
+    ComplexDateTimeField handles microseconds exactly instead of rounding
+    like DateTimeField does.
+
+    Derives from a StringField so you can do `gte` and `lte` filtering by
+    using lexicographical comparison when filtering / sorting strings.
+
+    The stored string has the following format:
+
+        YYYY,MM,DD,HH,MM,SS,NNNNNN
+
+    Where NNNNNN is the number of microseconds of the represented `datetime`.
+    The `,` as the separator can be easily modified by passing the `separator`
+    keyword when initializing the field.
+
+    .. versionadded:: 0.5
+    """
+
+    def __init__(self, separator=',', **kwargs):
+        self.names = ['year', 'month', 'day', 'hour', 'minute', 'second',
+                      'microsecond']
+        self.separtor = separator
+        super(ComplexDateTimeField, self).__init__(**kwargs)
+
+    def _leading_zero(self, number):
+        """
+        Converts the given number to a string.
+
+        If it has only one digit, a leading zero so as it has always at least
+        two digits.
+        """
+        if int(number) < 10:
+            return "0%s" % number
+        else:
+            return str(number)
+
+    def _convert_from_datetime(self, val):
+        """
+        Convert a `datetime` object to a string representation (which will be
+        stored in MongoDB). This is the reverse function of
+        `_convert_from_string`.
+
+        >>> a = datetime(2011, 6, 8, 20, 26, 24, 192284)
+        >>> RealDateTimeField()._convert_from_datetime(a)
+        '2011,06,08,20,26,24,192284'
+        """
+        data = []
+        for name in self.names:
+            data.append(self._leading_zero(getattr(val, name)))
+        return ','.join(data)
+
+    def _convert_from_string(self, data):
+        """
+        Convert a string representation to a `datetime` object (the object you
+        will manipulate). This is the reverse function of
+        `_convert_from_datetime`.
+
+        >>> a = '2011,06,08,20,26,24,192284'
+        >>> ComplexDateTimeField()._convert_from_string(a)
+        datetime.datetime(2011, 6, 8, 20, 26, 24, 192284)
+        """
+        data = data.split(',')
+        data = map(int, data)
+        values = {}
+        for i in range(7):
+            values[self.names[i]] = data[i]
+        return datetime.datetime(**values)
+
+    def __get__(self, instance, owner):
+        data = super(ComplexDateTimeField, self).__get__(instance, owner)
+        if data == None:
+            return datetime.datetime.now()
+        return self._convert_from_string(data)
+
+    def __set__(self, instance, value):
+        value = self._convert_from_datetime(value)
+        return super(ComplexDateTimeField, self).__set__(instance, value)
+
+    def validate(self, value):
+        if not isinstance(value, datetime.datetime):
+            raise ValidationError('Only datetime objects may used in a \
+                                   ComplexDateTimeField')
+
+    def to_python(self, value):
+        return self._convert_from_string(value)
+
+    def to_mongo(self, value):
+        return self._convert_from_datetime(value)
+
+    def prepare_query_value(self, op, value):
+        return self._convert_from_datetime(value)
 
 
 class EmbeddedDocumentField(BaseField):
-    """An embedded document field. Only valid values are subclasses of
-    :class:`~mongoengine.EmbeddedDocument`.
+    """An embedded document field - with a declared document_type.
+    Only valid values are subclasses of :class:`~mongoengine.EmbeddedDocument`.
     """
 
     def __init__(self, document_type, **kwargs):
@@ -256,6 +401,8 @@ class EmbeddedDocumentField(BaseField):
         return value
 
     def to_mongo(self, value):
+        if not isinstance(value, self.document_type):
+            return value
         return self.document_type.to_mongo(value)
 
     def validate(self, value):
@@ -275,7 +422,41 @@ class EmbeddedDocumentField(BaseField):
         return self.to_mongo(value)
 
 
-class ListField(BaseField):
+class GenericEmbeddedDocumentField(BaseField):
+    """A generic embedded document field - allows any
+    :class:`~mongoengine.EmbeddedDocument` to be stored.
+
+    Only valid values are subclasses of :class:`~mongoengine.EmbeddedDocument`.
+    """
+
+    def prepare_query_value(self, op, value):
+        return self.to_mongo(value)
+
+    def to_python(self, value):
+        if isinstance(value, dict):
+            doc_cls = get_document(value['_cls'])
+            value = doc_cls._from_son(value)
+
+        return value
+
+    def validate(self, value):
+        if not isinstance(value, EmbeddedDocument):
+            raise ValidationError('Invalid embedded document instance '
+                                  'provided to an GenericEmbeddedDocumentField')
+
+        value.validate()
+
+    def to_mongo(self, document):
+        if document is None:
+            return None
+
+        data = document.to_mongo()
+        if not '_cls' in data:
+            data['_cls'] = document._class_name
+        return data
+
+
+class ListField(ComplexBaseField):
     """A list field that wraps a standard field, allowing multiple instances
     of the field to be used as a list in the database.
     """
@@ -283,84 +464,26 @@ class ListField(BaseField):
     # ListFields cannot be indexed with _types - MongoDB doesn't support this
     _index_with_types = False
 
-    def __init__(self, field, **kwargs):
-        if not isinstance(field, BaseField):
-            raise ValidationError('Argument to ListField constructor must be '
-                                  'a valid field')
+    def __init__(self, field=None, **kwargs):
         self.field = field
         kwargs.setdefault('default', lambda: [])
         super(ListField, self).__init__(**kwargs)
 
-    def __get__(self, instance, owner):
-        """Descriptor to automatically dereference references.
-        """
-        if instance is None:
-            # Document class being used rather than a document object
-            return self
-
-        if isinstance(self.field, ReferenceField):
-            referenced_type = self.field.document_type
-            # Get value from document instance if available 
-            value_list = instance._data.get(self.name)
-            if value_list:
-                deref_list = []
-                for value in value_list:
-                    # Dereference DBRefs
-                    if isinstance(value, (pymongo.dbref.DBRef)):
-                        value = _get_db().dereference(value)
-                        deref_list.append(referenced_type._from_son(value))
-                    else:
-                        deref_list.append(value)
-                instance._data[self.name] = deref_list
-
-        if isinstance(self.field, GenericReferenceField):
-            value_list = instance._data.get(self.name)
-            if value_list:
-                deref_list = []
-                for value in value_list:
-                    # Dereference DBRefs
-                    if isinstance(value, (dict, pymongo.son.SON)):
-                        deref_list.append(self.field.dereference(value))
-                    else:
-                        deref_list.append(value)
-                instance._data[self.name] = deref_list
-
-        return super(ListField, self).__get__(instance, owner)
-
-    def to_python(self, value):
-        return [self.field.to_python(item) for item in value]
-
-    def to_mongo(self, value):
-        return [self.field.to_mongo(item) for item in value]
-
     def validate(self, value):
         """Make sure that a list of valid fields is being used.
         """
         if not isinstance(value, (list, tuple)):
             raise ValidationError('Only lists and tuples may be used in a '
                                   'list field')
-
-        try:
-            [self.field.validate(item) for item in value]
-        except Exception, err:
-            raise ValidationError('Invalid ListField item (%s)' % str(item))
+        super(ListField, self).validate(value)
 
     def prepare_query_value(self, op, value):
-        if op in ('set', 'unset'):
-            return [self.field.prepare_query_value(op, v) for v in value]
-        return self.field.prepare_query_value(op, value)
-
-    def lookup_member(self, member_name):
-        return self.field.lookup_member(member_name)
-
-    def _set_owner_document(self, owner_document):
-        self.field.owner_document = owner_document
-        self._owner_document = owner_document
-
-    def _get_owner_document(self, owner_document):
-        self._owner_document = owner_document
-
-    owner_document = property(_get_owner_document, _set_owner_document)
+        if self.field:
+            if op in ('set', 'unset') and (not isinstance(value, basestring)
+                and hasattr(value, '__iter__')):
+                return [self.field.prepare_query_value(op, v) for v in value]
+            return self.field.prepare_query_value(op, value)
+        return super(ListField, self).prepare_query_value(op, value)
 
 
 class SortedListField(ListField):
@@ -379,20 +502,22 @@ class SortedListField(ListField):
         super(SortedListField, self).__init__(field, **kwargs)
 
     def to_mongo(self, value):
+        value = super(SortedListField, self).to_mongo(value)
         if self._ordering is not None:
-            return sorted([self.field.to_mongo(item) for item in value],
-                          key=itemgetter(self._ordering))
-        return sorted([self.field.to_mongo(item) for item in value])
+            return sorted(value, key=itemgetter(self._ordering))
+        return sorted(value)
 
 
-class DictField(BaseField):
+class DictField(ComplexBaseField):
     """A dictionary field that wraps a standard Python dictionary. This is
     similar to an embedded document, but the structure is not defined.
 
     .. versionadded:: 0.3
+    .. versionchanged:: 0.5 - Can now handle complex / varying types of data
     """
 
-    def __init__(self, basecls=None, *args, **kwargs):
+    def __init__(self, basecls=None, field=None, *args, **kwargs):
+        self.field = field
         self.basecls = basecls or BaseField
         assert issubclass(self.basecls, BaseField)
         kwargs.setdefault('default', lambda: {})
@@ -408,21 +533,67 @@ class DictField(BaseField):
         if any(('.' in k or '$' in k) for k in value):
             raise ValidationError('Invalid dictionary key name - keys may not '
                                   'contain "." or "$" characters')
+        super(DictField, self).validate(value)
 
     def lookup_member(self, member_name):
-        return self.basecls(db_field=member_name)
+        return DictField(basecls=self.basecls, db_field=member_name)
+
+    def prepare_query_value(self, op, value):
+        match_operators = ['contains', 'icontains', 'startswith',
+                           'istartswith', 'endswith', 'iendswith',
+                           'exact', 'iexact']
+
+        if op in match_operators and isinstance(value, basestring):
+            return StringField().prepare_query_value(op, value)
+
+        return super(DictField, self).prepare_query_value(op, value)
+
+
+class MapField(DictField):
+    """A field that maps a name to a specified field type. Similar to
+    a DictField, except the 'value' of each item must match the specified
+    field type.
+
+    .. versionadded:: 0.5
+    """
+
+    def __init__(self, field=None, *args, **kwargs):
+        if not isinstance(field, BaseField):
+            raise ValidationError('Argument to MapField constructor must be '
+                                  'a valid field')
+        super(MapField, self).__init__(field=field, *args, **kwargs)
+
+
 
 class ReferenceField(BaseField):
     """A reference to a document that will be automatically dereferenced on
     access (lazily).
+
+    Use the `reverse_delete_rule` to handle what should happen if the document
+    the field is referencing is deleted.
+
+    The options are:
+
+      * DO_NOTHING  - don't do anything (default).
+      * NULLIFY     - Updates the reference to null.
+      * CASCADE     - Deletes the documents associated with the reference.
+      * DENY        - Prevent the deletion of the reference object.
+
+    .. versionchanged:: 0.5 added `reverse_delete_rule`
     """
 
-    def __init__(self, document_type, **kwargs):
+    def __init__(self, document_type, reverse_delete_rule=DO_NOTHING, **kwargs):
+        """Initialises the Reference Field.
+
+        :param reverse_delete_rule: Determines what to do when the referring
+          object is deleted
+        """
         if not isinstance(document_type, basestring):
             if not issubclass(document_type, (Document, basestring)):
                 raise ValidationError('Argument to ReferenceField constructor '
                                       'must be a document class or a string')
         self.document_type_obj = document_type
+        self.reverse_delete_rule = reverse_delete_rule
         super(ReferenceField, self).__init__(**kwargs)
 
     @property
@@ -465,7 +636,7 @@ class ReferenceField(BaseField):
             id_ = document
 
         id_ = id_field.to_mongo(id_)
-        collection = self.document_type._meta['collection']
+        collection = self.document_type._get_collection_name()
         return pymongo.dbref.DBRef(collection, id_)
 
     def prepare_query_value(self, op, value):
@@ -474,6 +645,11 @@ class ReferenceField(BaseField):
     def validate(self, value):
         assert isinstance(value, (self.document_type, pymongo.dbref.DBRef))
 
+        if isinstance(value, Document) and value.id is None:
+            raise ValidationError('You can only reference documents once '
+                                  'they have been saved to the database')
+
+
     def lookup_member(self, member_name):
         return self.document_type._fields.get(member_name)
 
@@ -482,6 +658,9 @@ class GenericReferenceField(BaseField):
     """A reference to *any* :class:`~mongoengine.document.Document` subclass
     that will be automatically dereferenced on access (lazily).
 
+    ..note ::  Any documents used as a generic reference must be registered in the
+    document registry.  Importing the model will automatically register it.
+
     .. versionadded:: 0.3
     """
 
@@ -495,6 +674,15 @@ class GenericReferenceField(BaseField):
 
         return super(GenericReferenceField, self).__get__(instance, owner)
 
+    def validate(self, value):
+        if not isinstance(value, (Document, pymongo.dbref.DBRef)):
+            raise ValidationError('GenericReferences can only contain documents')
+
+        # We need the id from the saved object to create the DBRef
+        if isinstance(value, Document) and value.id is None:
+            raise ValidationError('You can only reference documents once '
+                                  'they have been saved to the database')
+
     def dereference(self, value):
         doc_cls = get_document(value['_cls'])
         reference = value['_ref']
@@ -504,6 +692,9 @@ class GenericReferenceField(BaseField):
         return doc
 
     def to_mongo(self, document):
+        if document is None:
+            return None
+
         id_field_name = document.__class__._meta['id_field']
         id_field = document.__class__._fields[id_field_name]
 
@@ -517,9 +708,9 @@ class GenericReferenceField(BaseField):
             id_ = document
 
         id_ = id_field.to_mongo(id_)
-        collection = document._meta['collection']
+        collection = document._get_collection_name()
         ref = pymongo.dbref.DBRef(collection, id_)
-        return {'_cls': document.__class__.__name__, '_ref': ref}
+        return {'_cls': document._class_name, '_ref': ref}
 
     def prepare_query_value(self, op, value):
         return self.to_mongo(value)
@@ -555,12 +746,16 @@ class GridFSProxy(object):
     """Proxy object to handle writing and reading of files to and from GridFS
 
     .. versionadded:: 0.4
+    .. versionchanged:: 0.5 - added optional size param to read
     """
 
-    def __init__(self, grid_id=None):
+    def __init__(self, grid_id=None, key=None, instance=None):
         self.fs = gridfs.GridFS(_get_db())  # Filesystem instance
         self.newfile = None                 # Used for partial writes
         self.grid_id = grid_id              # Store GridFS id for file
+        self.gridout = None
+        self.key = key
+        self.instance = instance
 
     def __getattr__(self, name):
         obj = self.get()
@@ -571,11 +766,18 @@ class GridFSProxy(object):
     def __get__(self, instance, value):
         return self
 
+    def __nonzero__(self):
+        return bool(self.grid_id)
+
     def get(self, id=None):
         if id:
             self.grid_id = id
+        if self.grid_id is None:
+            return None
         try:
-            return self.fs.get(id or self.grid_id)
+            if self.gridout is None:
+                self.gridout = self.fs.get(self.grid_id)
+            return self.gridout
         except:
             # File has been deleted
             return None
@@ -584,11 +786,12 @@ class GridFSProxy(object):
         self.newfile = self.fs.new_file(**kwargs)
         self.grid_id = self.newfile._id
 
-    def put(self, file, **kwargs):
+    def put(self, file_obj, **kwargs):
         if self.grid_id:
             raise GridFSError('This document already has a file. Either delete '
                               'it or call replace to overwrite it')
-        self.grid_id = self.fs.put(file, **kwargs)
+        self.grid_id = self.fs.put(file_obj, **kwargs)
+        self._mark_as_changed()
 
     def write(self, string):
         if self.grid_id:
@@ -603,11 +806,11 @@ class GridFSProxy(object):
         if not self.newfile:
             self.new_file()
             self.grid_id = self.newfile._id
-        self.newfile.writelines(lines) 
+        self.newfile.writelines(lines)
 
-    def read(self):
+    def read(self, size=-1):
         try:
-            return self.get().read()
+            return self.get().read(size)
         except:
             return None
 
@@ -615,23 +818,28 @@ class GridFSProxy(object):
         # Delete file from GridFS, FileField still remains
         self.fs.delete(self.grid_id)
         self.grid_id = None
+        self.gridout = None
+        self._mark_as_changed()
 
-    def replace(self, file, **kwargs):
+    def replace(self, file_obj, **kwargs):
         self.delete()
-        self.put(file, **kwargs)
+        self.put(file_obj, **kwargs)
 
     def close(self):
         if self.newfile:
             self.newfile.close()
-        else:
-            msg = "The close() method is only necessary after calling write()"
-            warnings.warn(msg)
+
+    def _mark_as_changed(self):
+        """Inform the instance that `self.key` has been changed"""
+        if self.instance:
+            self.instance._mark_as_changed(self.key)
 
 
 class FileField(BaseField):
     """A GridFS storage field.
 
     .. versionadded:: 0.4
+    .. versionchanged:: 0.5 added optional size param for read
     """
 
     def __init__(self, **kwargs):
@@ -644,11 +852,15 @@ class FileField(BaseField):
         # Check if a file already exists for this model
         grid_file = instance._data.get(self.name)
         self.grid_file = grid_file
-        if self.grid_file:
+        if isinstance(self.grid_file, GridFSProxy):
+            if not self.grid_file.key:
+                self.grid_file.key = self.name
+                self.grid_file.instance = instance
             return self.grid_file
-        return GridFSProxy()
+        return GridFSProxy(key=self.name, instance=instance)
 
     def __set__(self, instance, value):
+        key = self.name
         if isinstance(value, file) or isinstance(value, str):
             # using "FileField() = file/string" notation
             grid_file = instance._data.get(self.name)
@@ -662,10 +874,12 @@ class FileField(BaseField):
                 grid_file.put(value)
             else:
                 # Create a new proxy object as we don't already have one
-                instance._data[self.name] = GridFSProxy()
-                instance._data[self.name].put(value)
+                instance._data[key] = GridFSProxy(key=key, instance=instance)
+                instance._data[key].put(value)
         else:
-            instance._data[self.name] = value
+            instance._data[key] = value
+
+        instance._mark_as_changed(key)
 
     def to_mongo(self, value):
         # Store the GridFS file id in MongoDB
@@ -703,3 +917,61 @@ class GeoPointField(BaseField):
         if (not isinstance(value[0], (float, int)) and
             not isinstance(value[1], (float, int))):
             raise ValidationError('Both values in point must be float or int.')
+
+
+class SequenceField(IntField):
+    """Provides a sequental counter.
+
+    ..note:: Although traditional databases often use increasing sequence
+             numbers for primary keys. In MongoDB, the preferred approach is to
+             use Object IDs instead.  The concept is that in a very large
+             cluster of machines, it is easier to create an object ID than have
+             global, uniformly increasing sequence numbers.
+
+    .. versionadded:: 0.5
+    """
+    def __init__(self, collection_name=None, *args, **kwargs):
+        self.collection_name = collection_name or 'mongoengine.counters'
+        return super(SequenceField, self).__init__(*args, **kwargs)
+
+    def generate_new_value(self):
+        """
+        Generate and Increment the counter
+        """
+        sequence_id = "{0}.{1}".format(self.owner_document._get_collection_name(),
+                                       self.name)
+        collection = _get_db()[self.collection_name]
+        counter = collection.find_and_modify(query={"_id": sequence_id},
+                                             update={"$inc": {"next": 1}},
+                                             new=True,
+                                             upsert=True)
+        return counter['next']
+
+    def __get__(self, instance, owner):
+
+        if instance is None:
+            return self
+
+        if not instance._data:
+            return
+
+        value = instance._data.get(self.name)
+
+        if not value and instance._initialised:
+            value = self.generate_new_value()
+            instance._data[self.name] = value
+            instance._mark_as_changed(self.name)
+
+        return value
+
+    def __set__(self, instance, value):
+
+        if value is None and instance._initialised:
+            value = self.generate_new_value()
+
+        return super(SequenceField, self).__set__(instance, value)
+
+    def to_python(self, value):
+        if value is None:
+            value = self.generate_new_value()
+        return value

mongoengine/signals.py

@@ -0,0 +1,44 @@
+# -*- coding: utf-8 -*-
+
+__all__ = ['pre_init', 'post_init', 'pre_save', 'post_save',
+           'pre_delete', 'post_delete']
+
+signals_available = False
+try:
+    from blinker import Namespace
+    signals_available = True
+except ImportError:
+    class Namespace(object):
+        def signal(self, name, doc=None):
+            return _FakeSignal(name, doc)
+
+    class _FakeSignal(object):
+        """If blinker is unavailable, create a fake class with the same
+        interface that allows sending of signals but will fail with an
+        error on anything else.  Instead of doing anything on send, it
+        will just ignore the arguments and do nothing instead.
+        """
+
+        def __init__(self, name, doc=None):
+            self.name = name
+            self.__doc__ = doc
+
+        def _fail(self, *args, **kwargs):
+            raise RuntimeError('signalling support is unavailable '
+                               'because the blinker library is '
+                               'not installed.')
+        send = lambda *a, **kw: None
+        connect = disconnect = has_receivers_for = receivers_for = \
+            temporarily_connected_to = _fail
+        del _fail
+
+# the namespace for code signals.  If you are not mongoengine code, do
+# not put signals in here.  Create your own namespace instead.
+_signals = Namespace()
+
+pre_init = _signals.signal('pre_init')
+post_init = _signals.signal('post_init')
+pre_save = _signals.signal('pre_save')
+post_save = _signals.signal('post_save')
+pre_delete = _signals.signal('pre_delete')
+post_delete = _signals.signal('post_delete')

mongoengine/tests.py

@@ -0,0 +1,59 @@
+from mongoengine.connection import _get_db
+
+
+class query_counter(object):
+    """ Query_counter contextmanager to get the number of queries. """
+
+    def __init__(self):
+        """ Construct the query_counter. """
+        self.counter = 0
+        self.db = _get_db()
+
+    def __enter__(self):
+        """ On every with block we need to drop the profile collection. """
+        self.db.set_profiling_level(0)
+        self.db.system.profile.drop()
+        self.db.set_profiling_level(2)
+        return self
+
+    def __exit__(self, t, value, traceback):
+        """ Reset the profiling level. """
+        self.db.set_profiling_level(0)
+
+    def __eq__(self, value):
+        """ == Compare querycounter. """
+        return value == self._get_count()
+
+    def __ne__(self, value):
+        """ != Compare querycounter. """
+        return not self.__eq__(value)
+
+    def __lt__(self, value):
+        """ < Compare querycounter. """
+        return self._get_count() < value
+
+    def __le__(self, value):
+        """ <= Compare querycounter. """
+        return self._get_count() <= value
+
+    def __gt__(self, value):
+        """ > Compare querycounter. """
+        return self._get_count() > value
+
+    def __ge__(self, value):
+        """ >= Compare querycounter. """
+        return self._get_count() >= value
+
+    def __int__(self):
+        """ int representation. """
+        return self._get_count()
+
+    def __repr__(self):
+        """ repr query_counter as the number of queries. """
+        return u"%s" % self._get_count()
+
+    def _get_count(self):
+        """ Get the number of queries. """
+        count = self.db.system.profile.find().count() - self.counter
+        self.counter += 1
+        return count

setup.py

@@ -15,7 +15,7 @@ def get_version(version_tuple):
         version = '%s.%s' % (version, version_tuple[2])
     return version
 
-# Dirty hack to get version number from monogengine/__init__.py - we can't 
+# Dirty hack to get version number from monogengine/__init__.py - we can't
 # import it as it depends on PyMongo and PyMongo isn't installed until this
 # file is read
 init = os.path.join(os.path.dirname(__file__), 'mongoengine', '__init__.py')
@@ -47,4 +47,5 @@ setup(name='mongoengine',
       classifiers=CLASSIFIERS,
       install_requires=['pymongo'],
       test_suite='tests',
+      tests_require=['blinker', 'django==1.3']
 )

tests/dereference.py

@@ -0,0 +1,658 @@
+import unittest
+
+from mongoengine import *
+from mongoengine.connection import _get_db
+from mongoengine.tests import query_counter
+
+
+class FieldTest(unittest.TestCase):
+
+    def setUp(self):
+        connect(db='mongoenginetest')
+        self.db = _get_db()
+
+    def test_list_item_dereference(self):
+        """Ensure that DBRef items in ListFields are dereferenced.
+        """
+        class User(Document):
+            name = StringField()
+
+        class Group(Document):
+            members = ListField(ReferenceField(User))
+
+        User.drop_collection()
+        Group.drop_collection()
+
+        for i in xrange(1, 51):
+            user = User(name='user %s' % i)
+            user.save()
+
+        group = Group(members=User.objects)
+        group.save()
+
+        group = Group(members=User.objects)
+        group.save()
+
+        with query_counter() as q:
+            self.assertEqual(q, 0)
+
+            group_obj = Group.objects.first()
+            self.assertEqual(q, 1)
+
+            [m for m in group_obj.members]
+            self.assertEqual(q, 2)
+
+        # Document select_related
+        with query_counter() as q:
+            self.assertEqual(q, 0)
+
+            group_obj = Group.objects.first().select_related()
+            self.assertEqual(q, 2)
+            [m for m in group_obj.members]
+            self.assertEqual(q, 2)
+
+        # Queryset select_related
+        with query_counter() as q:
+            self.assertEqual(q, 0)
+            group_objs = Group.objects.select_related()
+            self.assertEqual(q, 2)
+            for group_obj in group_objs:
+                [m for m in group_obj.members]
+                self.assertEqual(q, 2)
+
+        User.drop_collection()
+        Group.drop_collection()
+
+    def test_recursive_reference(self):
+        """Ensure that ReferenceFields can reference their own documents.
+        """
+        class Employee(Document):
+            name = StringField()
+            boss = ReferenceField('self')
+            friends = ListField(ReferenceField('self'))
+
+        Employee.drop_collection()
+
+        bill = Employee(name='Bill Lumbergh')
+        bill.save()
+
+        michael = Employee(name='Michael Bolton')
+        michael.save()
+
+        samir = Employee(name='Samir Nagheenanajar')
+        samir.save()
+
+        friends = [michael, samir]
+        peter = Employee(name='Peter Gibbons', boss=bill, friends=friends)
+        peter.save()
+
+        Employee(name='Funky Gibbon', boss=bill, friends=friends).save()
+        Employee(name='Funky Gibbon', boss=bill, friends=friends).save()
+        Employee(name='Funky Gibbon', boss=bill, friends=friends).save()
+
+        with query_counter() as q:
+            self.assertEqual(q, 0)
+
+            peter = Employee.objects.with_id(peter.id)
+            self.assertEqual(q, 1)
+
+            peter.boss
+            self.assertEqual(q, 2)
+
+            peter.friends
+            self.assertEqual(q, 3)
+
+        # Document select_related
+        with query_counter() as q:
+            self.assertEqual(q, 0)
+
+            peter = Employee.objects.with_id(peter.id).select_related()
+            self.assertEqual(q, 2)
+
+            self.assertEquals(peter.boss, bill)
+            self.assertEqual(q, 2)
+
+            self.assertEquals(peter.friends, friends)
+            self.assertEqual(q, 2)
+
+        # Queryset select_related
+        with query_counter() as q:
+            self.assertEqual(q, 0)
+
+            employees = Employee.objects(boss=bill).select_related()
+            self.assertEqual(q, 2)
+
+            for employee in employees:
+                self.assertEquals(employee.boss, bill)
+                self.assertEqual(q, 2)
+
+                self.assertEquals(employee.friends, friends)
+                self.assertEqual(q, 2)
+
+    def test_generic_reference(self):
+
+        class UserA(Document):
+            name = StringField()
+
+        class UserB(Document):
+            name = StringField()
+
+        class UserC(Document):
+            name = StringField()
+
+        class Group(Document):
+            members = ListField(GenericReferenceField())
+
+        UserA.drop_collection()
+        UserB.drop_collection()
+        UserC.drop_collection()
+        Group.drop_collection()
+
+        members = []
+        for i in xrange(1, 51):
+            a = UserA(name='User A %s' % i)
+            a.save()
+
+            b = UserB(name='User B %s' % i)
+            b.save()
+
+            c = UserC(name='User C %s' % i)
+            c.save()
+
+            members += [a, b, c]
+
+        group = Group(members=members)
+        group.save()
+
+        group = Group(members=members)
+        group.save()
+
+        with query_counter() as q:
+            self.assertEqual(q, 0)
+
+            group_obj = Group.objects.first()
+            self.assertEqual(q, 1)
+
+            [m for m in group_obj.members]
+            self.assertEqual(q, 4)
+
+            [m for m in group_obj.members]
+            self.assertEqual(q, 4)
+
+            for m in group_obj.members:
+                self.assertTrue('User' in m.__class__.__name__)
+
+        # Document select_related
+        with query_counter() as q:
+            self.assertEqual(q, 0)
+
+            group_obj = Group.objects.first().select_related()
+            self.assertEqual(q, 4)
+
+            [m for m in group_obj.members]
+            self.assertEqual(q, 4)
+
+            [m for m in group_obj.members]
+            self.assertEqual(q, 4)
+
+            for m in group_obj.members:
+                self.assertTrue('User' in m.__class__.__name__)
+
+        # Queryset select_related
+        with query_counter() as q:
+            self.assertEqual(q, 0)
+
+            group_objs = Group.objects.select_related()
+            self.assertEqual(q, 4)
+
+            for group_obj in group_objs:
+                [m for m in group_obj.members]
+                self.assertEqual(q, 4)
+
+                [m for m in group_obj.members]
+                self.assertEqual(q, 4)
+
+                for m in group_obj.members:
+                    self.assertTrue('User' in m.__class__.__name__)
+
+        UserA.drop_collection()
+        UserB.drop_collection()
+        UserC.drop_collection()
+        Group.drop_collection()
+
+    def test_list_field_complex(self):
+
+        class UserA(Document):
+            name = StringField()
+
+        class UserB(Document):
+            name = StringField()
+
+        class UserC(Document):
+            name = StringField()
+
+        class Group(Document):
+            members = ListField()
+
+        UserA.drop_collection()
+        UserB.drop_collection()
+        UserC.drop_collection()
+        Group.drop_collection()
+
+        members = []
+        for i in xrange(1, 51):
+            a = UserA(name='User A %s' % i)
+            a.save()
+
+            b = UserB(name='User B %s' % i)
+            b.save()
+
+            c = UserC(name='User C %s' % i)
+            c.save()
+
+            members += [a, b, c]
+
+        group = Group(members=members)
+        group.save()
+
+        group = Group(members=members)
+        group.save()
+
+        with query_counter() as q:
+            self.assertEqual(q, 0)
+
+            group_obj = Group.objects.first()
+            self.assertEqual(q, 1)
+
+            [m for m in group_obj.members]
+            self.assertEqual(q, 4)
+
+            [m for m in group_obj.members]
+            self.assertEqual(q, 4)
+
+            for m in group_obj.members:
+                self.assertTrue('User' in m.__class__.__name__)
+
+        # Document select_related
+        with query_counter() as q:
+            self.assertEqual(q, 0)
+
+            group_obj = Group.objects.first().select_related()
+            self.assertEqual(q, 4)
+
+            [m for m in group_obj.members]
+            self.assertEqual(q, 4)
+
+            [m for m in group_obj.members]
+            self.assertEqual(q, 4)
+
+            for m in group_obj.members:
+                self.assertTrue('User' in m.__class__.__name__)
+
+        # Queryset select_related
+        with query_counter() as q:
+            self.assertEqual(q, 0)
+
+            group_objs = Group.objects.select_related()
+            self.assertEqual(q, 4)
+
+            for group_obj in group_objs:
+                [m for m in group_obj.members]
+                self.assertEqual(q, 4)
+
+                [m for m in group_obj.members]
+                self.assertEqual(q, 4)
+
+                for m in group_obj.members:
+                    self.assertTrue('User' in m.__class__.__name__)
+
+        UserA.drop_collection()
+        UserB.drop_collection()
+        UserC.drop_collection()
+        Group.drop_collection()
+
+    def test_map_field_reference(self):
+
+        class User(Document):
+            name = StringField()
+
+        class Group(Document):
+            members = MapField(ReferenceField(User))
+
+        User.drop_collection()
+        Group.drop_collection()
+
+        members = []
+        for i in xrange(1, 51):
+            user = User(name='user %s' % i)
+            user.save()
+            members.append(user)
+
+        group = Group(members=dict([(str(u.id), u) for u in members]))
+        group.save()
+
+        group = Group(members=dict([(str(u.id), u) for u in members]))
+        group.save()
+
+        with query_counter() as q:
+            self.assertEqual(q, 0)
+
+            group_obj = Group.objects.first()
+            self.assertEqual(q, 1)
+
+            [m for m in group_obj.members]
+            self.assertEqual(q, 2)
+
+            for k, m in group_obj.members.iteritems():
+                self.assertTrue(isinstance(m, User))
+
+        # Document select_related
+        with query_counter() as q:
+            self.assertEqual(q, 0)
+
+            group_obj = Group.objects.first().select_related()
+            self.assertEqual(q, 2)
+
+            [m for m in group_obj.members]
+            self.assertEqual(q, 2)
+
+            for k, m in group_obj.members.iteritems():
+                self.assertTrue(isinstance(m, User))
+
+       # Queryset select_related
+        with query_counter() as q:
+            self.assertEqual(q, 0)
+
+            group_objs = Group.objects.select_related()
+            self.assertEqual(q, 2)
+
+            for group_obj in group_objs:
+                [m for m in group_obj.members]
+                self.assertEqual(q, 2)
+
+                for k, m in group_obj.members.iteritems():
+                    self.assertTrue(isinstance(m, User))
+
+        User.drop_collection()
+        Group.drop_collection()
+
+    def test_dict_field(self):
+
+        class UserA(Document):
+            name = StringField()
+
+        class UserB(Document):
+            name = StringField()
+
+        class UserC(Document):
+            name = StringField()
+
+        class Group(Document):
+            members = DictField()
+
+        UserA.drop_collection()
+        UserB.drop_collection()
+        UserC.drop_collection()
+        Group.drop_collection()
+
+        members = []
+        for i in xrange(1, 51):
+            a = UserA(name='User A %s' % i)
+            a.save()
+
+            b = UserB(name='User B %s' % i)
+            b.save()
+
+            c = UserC(name='User C %s' % i)
+            c.save()
+
+            members += [a, b, c]
+
+        group = Group(members=dict([(str(u.id), u) for u in members]))
+        group.save()
+        group = Group(members=dict([(str(u.id), u) for u in members]))
+        group.save()
+
+        with query_counter() as q:
+            self.assertEqual(q, 0)
+
+            group_obj = Group.objects.first()
+            self.assertEqual(q, 1)
+
+            [m for m in group_obj.members]
+            self.assertEqual(q, 4)
+
+            [m for m in group_obj.members]
+            self.assertEqual(q, 4)
+
+            for k, m in group_obj.members.iteritems():
+                self.assertTrue('User' in m.__class__.__name__)
+
+        # Document select_related
+        with query_counter() as q:
+            self.assertEqual(q, 0)
+
+            group_obj = Group.objects.first().select_related()
+            self.assertEqual(q, 4)
+
+            [m for m in group_obj.members]
+            self.assertEqual(q, 4)
+
+            [m for m in group_obj.members]
+            self.assertEqual(q, 4)
+
+            for k, m in group_obj.members.iteritems():
+                self.assertTrue('User' in m.__class__.__name__)
+
+        # Queryset select_related
+        with query_counter() as q:
+            self.assertEqual(q, 0)
+
+            group_objs = Group.objects.select_related()
+            self.assertEqual(q, 4)
+
+            for group_obj in group_objs:
+                [m for m in group_obj.members]
+                self.assertEqual(q, 4)
+
+                [m for m in group_obj.members]
+                self.assertEqual(q, 4)
+
+                for k, m in group_obj.members.iteritems():
+                    self.assertTrue('User' in m.__class__.__name__)
+
+        Group.objects.delete()
+        Group().save()
+
+        with query_counter() as q:
+            self.assertEqual(q, 0)
+
+            group_obj = Group.objects.first()
+            self.assertEqual(q, 1)
+
+            [m for m in group_obj.members]
+            self.assertEqual(q, 1)
+            self.assertEqual(group_obj.members, {})
+
+        UserA.drop_collection()
+        UserB.drop_collection()
+        UserC.drop_collection()
+        Group.drop_collection()
+
+    def test_dict_field_no_field_inheritance(self):
+
+        class UserA(Document):
+            name = StringField()
+            meta = {'allow_inheritance': False}
+
+        class Group(Document):
+            members = DictField()
+
+        UserA.drop_collection()
+        Group.drop_collection()
+
+        members = []
+        for i in xrange(1, 51):
+            a = UserA(name='User A %s' % i)
+            a.save()
+
+            members += [a]
+
+        group = Group(members=dict([(str(u.id), u) for u in members]))
+        group.save()
+
+        group = Group(members=dict([(str(u.id), u) for u in members]))
+        group.save()
+
+        with query_counter() as q:
+            self.assertEqual(q, 0)
+
+            group_obj = Group.objects.first()
+            self.assertEqual(q, 1)
+
+            [m for m in group_obj.members]
+            self.assertEqual(q, 2)
+
+            [m for m in group_obj.members]
+            self.assertEqual(q, 2)
+
+            for k, m in group_obj.members.iteritems():
+                self.assertTrue(isinstance(m, UserA))
+
+        # Document select_related
+        with query_counter() as q:
+            self.assertEqual(q, 0)
+
+            group_obj = Group.objects.first().select_related()
+            self.assertEqual(q, 2)
+
+            [m for m in group_obj.members]
+            self.assertEqual(q, 2)
+
+            [m for m in group_obj.members]
+            self.assertEqual(q, 2)
+
+            for k, m in group_obj.members.iteritems():
+                self.assertTrue(isinstance(m, UserA))
+
+        # Queryset select_related
+        with query_counter() as q:
+            self.assertEqual(q, 0)
+
+            group_objs = Group.objects.select_related()
+            self.assertEqual(q, 2)
+
+            for group_obj in group_objs:
+                [m for m in group_obj.members]
+                self.assertEqual(q, 2)
+
+                [m for m in group_obj.members]
+                self.assertEqual(q, 2)
+
+                for k, m in group_obj.members.iteritems():
+                    self.assertTrue(isinstance(m, UserA))
+
+        UserA.drop_collection()
+        Group.drop_collection()
+
+    def test_generic_reference_map_field(self):
+
+        class UserA(Document):
+            name = StringField()
+
+        class UserB(Document):
+            name = StringField()
+
+        class UserC(Document):
+            name = StringField()
+
+        class Group(Document):
+            members = MapField(GenericReferenceField())
+
+        UserA.drop_collection()
+        UserB.drop_collection()
+        UserC.drop_collection()
+        Group.drop_collection()
+
+        members = []
+        for i in xrange(1, 51):
+            a = UserA(name='User A %s' % i)
+            a.save()
+
+            b = UserB(name='User B %s' % i)
+            b.save()
+
+            c = UserC(name='User C %s' % i)
+            c.save()
+
+            members += [a, b, c]
+
+        group = Group(members=dict([(str(u.id), u) for u in members]))
+        group.save()
+        group = Group(members=dict([(str(u.id), u) for u in members]))
+        group.save()
+
+        with query_counter() as q:
+            self.assertEqual(q, 0)
+
+            group_obj = Group.objects.first()
+            self.assertEqual(q, 1)
+
+            [m for m in group_obj.members]
+            self.assertEqual(q, 4)
+
+            [m for m in group_obj.members]
+            self.assertEqual(q, 4)
+
+            for k, m in group_obj.members.iteritems():
+                self.assertTrue('User' in m.__class__.__name__)
+
+        # Document select_related
+        with query_counter() as q:
+            self.assertEqual(q, 0)
+
+            group_obj = Group.objects.first().select_related()
+            self.assertEqual(q, 4)
+
+            [m for m in group_obj.members]
+            self.assertEqual(q, 4)
+
+            [m for m in group_obj.members]
+            self.assertEqual(q, 4)
+
+            for k, m in group_obj.members.iteritems():
+                self.assertTrue('User' in m.__class__.__name__)
+
+        # Queryset select_related
+        with query_counter() as q:
+            self.assertEqual(q, 0)
+
+            group_objs = Group.objects.select_related()
+            self.assertEqual(q, 4)
+
+            for group_obj in group_objs:
+                [m for m in group_obj.members]
+                self.assertEqual(q, 4)
+
+                [m for m in group_obj.members]
+                self.assertEqual(q, 4)
+
+                for k, m in group_obj.members.iteritems():
+                    self.assertTrue('User' in m.__class__.__name__)
+
+        Group.objects.delete()
+        Group().save()
+
+        with query_counter() as q:
+            self.assertEqual(q, 0)
+
+            group_obj = Group.objects.first()
+            self.assertEqual(q, 1)
+
+            [m for m in group_obj.members]
+            self.assertEqual(q, 1)
+
+        UserA.drop_collection()
+        UserB.drop_collection()
+        UserC.drop_collection()
+        Group.drop_collection()

tests/django_tests.py

@@ -0,0 +1,69 @@
+# -*- coding: utf-8 -*-
+
+import unittest
+
+from mongoengine import *
+from mongoengine.django.shortcuts import get_document_or_404
+
+from django.http import Http404
+from django.template import Context, Template
+from django.conf import settings
+settings.configure()
+
+class QuerySetTest(unittest.TestCase):
+
+    def setUp(self):
+        connect(db='mongoenginetest')
+
+        class Person(Document):
+            name = StringField()
+            age = IntField()
+        self.Person = Person
+
+    def test_order_by_in_django_template(self):
+        """Ensure that QuerySets are properly ordered in Django template.
+        """
+        self.Person.drop_collection()
+
+        self.Person(name="A", age=20).save()
+        self.Person(name="D", age=10).save()
+        self.Person(name="B", age=40).save()
+        self.Person(name="C", age=30).save()
+
+        t = Template("{% for o in ol %}{{ o.name }}-{{ o.age }}:{% endfor %}")
+
+        d = {"ol": self.Person.objects.order_by('-name')}
+        self.assertEqual(t.render(Context(d)), u'D-10:C-30:B-40:A-20:')
+        d = {"ol": self.Person.objects.order_by('+name')}
+        self.assertEqual(t.render(Context(d)), u'A-20:B-40:C-30:D-10:')
+        d = {"ol": self.Person.objects.order_by('-age')}
+        self.assertEqual(t.render(Context(d)), u'B-40:C-30:A-20:D-10:')
+        d = {"ol": self.Person.objects.order_by('+age')}
+        self.assertEqual(t.render(Context(d)), u'D-10:A-20:C-30:B-40:')
+
+        self.Person.drop_collection()
+
+    def test_q_object_filter_in_template(self):
+
+        self.Person.drop_collection()
+
+        self.Person(name="A", age=20).save()
+        self.Person(name="D", age=10).save()
+        self.Person(name="B", age=40).save()
+        self.Person(name="C", age=30).save()
+
+        t = Template("{% for o in ol %}{{ o.name }}-{{ o.age }}:{% endfor %}")
+
+        d = {"ol": self.Person.objects.filter(Q(age=10) | Q(name="C"))}
+        self.assertEqual(t.render(Context(d)), 'D-10:C-30:')
+
+        # Check double rendering doesn't throw an error
+        self.assertEqual(t.render(Context(d)), 'D-10:C-30:')
+
+    def test_get_document_or_404(self):
+        p = self.Person(name="G404")
+        p.save()
+
+        self.assertRaises(Http404, get_document_or_404, self.Person, pk='1234')
+        self.assertEqual(p, get_document_or_404(self.Person, pk=p.pk))
+

tests/fixtures.py

@@ -0,0 +1,25 @@
+from datetime import datetime
+import pymongo
+
+from mongoengine import *
+from mongoengine.base import BaseField
+from mongoengine.connection import _get_db
+
+
+class PickleEmbedded(EmbeddedDocument):
+    date = DateTimeField(default=datetime.now)
+
+
+class PickleTest(Document):
+    number = IntField()
+    string = StringField(choices=(('One', '1'), ('Two', '2')))
+    embedded = EmbeddedDocumentField(PickleEmbedded)
+    lists = ListField(StringField())
+
+
+class Mixin(object):
+    name = StringField()
+
+
+class Base(Document):
+    pass

tests/signals.py

@@ -0,0 +1,181 @@
+# -*- coding: utf-8 -*-
+import unittest
+
+from mongoengine import *
+from mongoengine import signals
+
+signal_output = []
+
+
+class SignalTests(unittest.TestCase):
+    """
+    Testing signals before/after saving and deleting.
+    """
+
+    def get_signal_output(self, fn, *args, **kwargs):
+        # Flush any existing signal output
+        global signal_output
+        signal_output = []
+        fn(*args, **kwargs)
+        return signal_output
+
+    def setUp(self):
+        connect(db='mongoenginetest')
+        class Author(Document):
+            name = StringField()
+
+            def __unicode__(self):
+                return self.name
+
+            @classmethod
+            def pre_init(cls, sender, document, *args, **kwargs):
+                signal_output.append('pre_init signal, %s' % cls.__name__)
+                signal_output.append(str(kwargs['values']))
+
+            @classmethod
+            def post_init(cls, sender, document, **kwargs):
+                signal_output.append('post_init signal, %s' % document)
+
+            @classmethod
+            def pre_save(cls, sender, document, **kwargs):
+                signal_output.append('pre_save signal, %s' % document)
+
+            @classmethod
+            def post_save(cls, sender, document, **kwargs):
+                signal_output.append('post_save signal, %s' % document)
+                if 'created' in kwargs:
+                    if kwargs['created']:
+                        signal_output.append('Is created')
+                    else:
+                        signal_output.append('Is updated')
+
+            @classmethod
+            def pre_delete(cls, sender, document, **kwargs):
+                signal_output.append('pre_delete signal, %s' % document)
+
+            @classmethod
+            def post_delete(cls, sender, document, **kwargs):
+                signal_output.append('post_delete signal, %s' % document)
+        self.Author = Author
+
+
+        class Another(Document):
+            name = StringField()
+
+            def __unicode__(self):
+                return self.name
+
+            @classmethod
+            def pre_init(cls, sender, document, **kwargs):
+                signal_output.append('pre_init Another signal, %s' % cls.__name__)
+                signal_output.append(str(kwargs['values']))
+
+            @classmethod
+            def post_init(cls, sender, document, **kwargs):
+                signal_output.append('post_init Another signal, %s' % document)
+
+            @classmethod
+            def pre_save(cls, sender, document, **kwargs):
+                signal_output.append('pre_save Another signal, %s' % document)
+
+            @classmethod
+            def post_save(cls, sender, document, **kwargs):
+                signal_output.append('post_save Another signal, %s' % document)
+                if 'created' in kwargs:
+                    if kwargs['created']:
+                        signal_output.append('Is created')
+                    else:
+                        signal_output.append('Is updated')
+
+            @classmethod
+            def pre_delete(cls, sender, document, **kwargs):
+                signal_output.append('pre_delete Another signal, %s' % document)
+
+            @classmethod
+            def post_delete(cls, sender, document, **kwargs):
+                signal_output.append('post_delete Another signal, %s' % document)
+
+        self.Another = Another
+        # Save up the number of connected signals so that we can check at the end
+        # that all the signals we register get properly unregistered
+        self.pre_signals = (
+            len(signals.pre_init.receivers),
+            len(signals.post_init.receivers),
+            len(signals.pre_save.receivers),
+            len(signals.post_save.receivers),
+            len(signals.pre_delete.receivers),
+            len(signals.post_delete.receivers)
+        )
+
+        signals.pre_init.connect(Author.pre_init, sender=Author)
+        signals.post_init.connect(Author.post_init, sender=Author)
+        signals.pre_save.connect(Author.pre_save, sender=Author)
+        signals.post_save.connect(Author.post_save, sender=Author)
+        signals.pre_delete.connect(Author.pre_delete, sender=Author)
+        signals.post_delete.connect(Author.post_delete, sender=Author)
+
+        signals.pre_init.connect(Another.pre_init, sender=Another)
+        signals.post_init.connect(Another.post_init, sender=Another)
+        signals.pre_save.connect(Another.pre_save, sender=Another)
+        signals.post_save.connect(Another.post_save, sender=Another)
+        signals.pre_delete.connect(Another.pre_delete, sender=Another)
+        signals.post_delete.connect(Another.post_delete, sender=Another)
+
+    def tearDown(self):
+        signals.pre_init.disconnect(self.Author.pre_init)
+        signals.post_init.disconnect(self.Author.post_init)
+        signals.post_delete.disconnect(self.Author.post_delete)
+        signals.pre_delete.disconnect(self.Author.pre_delete)
+        signals.post_save.disconnect(self.Author.post_save)
+        signals.pre_save.disconnect(self.Author.pre_save)
+
+        signals.pre_init.disconnect(self.Another.pre_init)
+        signals.post_init.disconnect(self.Another.post_init)
+        signals.post_delete.disconnect(self.Another.post_delete)
+        signals.pre_delete.disconnect(self.Another.pre_delete)
+        signals.post_save.disconnect(self.Another.post_save)
+        signals.pre_save.disconnect(self.Another.pre_save)
+
+        # Check that all our signals got disconnected properly.
+        post_signals = (
+            len(signals.pre_init.receivers),
+            len(signals.post_init.receivers),
+            len(signals.pre_save.receivers),
+            len(signals.post_save.receivers),
+            len(signals.pre_delete.receivers),
+            len(signals.post_delete.receivers)
+        )
+
+        self.assertEqual(self.pre_signals, post_signals)
+
+    def test_model_signals(self):
+        """ Model saves should throw some signals. """
+
+        def create_author():
+            a1 = self.Author(name='Bill Shakespeare')
+
+        self.assertEqual(self.get_signal_output(create_author), [
+            "pre_init signal, Author",
+            "{'name': 'Bill Shakespeare'}",
+            "post_init signal, Bill Shakespeare",
+        ])
+
+        a1 = self.Author(name='Bill Shakespeare')
+        self.assertEqual(self.get_signal_output(a1.save), [
+            "pre_save signal, Bill Shakespeare",
+            "post_save signal, Bill Shakespeare",
+            "Is created"
+        ])
+
+        a1.reload()
+        a1.name='William Shakespeare'
+        self.assertEqual(self.get_signal_output(a1.save), [
+            "pre_save signal, William Shakespeare",
+            "post_save signal, William Shakespeare",
+            "Is updated"
+        ])
+
+        self.assertEqual(self.get_signal_output(a1.delete), [
+            'pre_delete signal, William Shakespeare',
+            'post_delete signal, William Shakespeare',
+        ])