An extremely thin ORM-ish wrapper over pymongo.
Clone or download


mongorm is an extremely thin ODM layer on top of pymongo that allows you to create classes that represent MongoDB documents.

It's designed to give you all the flexibility of pymongo, with a few convenience features, such as attribute-style ( access to fields.

Getting Started

The recommended way to install mongorm is to install via pip, pip install mongorm

mongorm only has a single class for you to import:

>>> from mongorm import Database

You can connect to a database either via a MongoDB URI:

>>> db = Database(uri='mongodb://localhost:27017/some_db')

or with a host-port-db combination:

>>> db = Database(host='localhost', port=27017, db='some_db')

If any of the keyword arguments aren't matched, or if the URI is missing a database name, the following are used as defaults:

  • host: 'localhost'
  • port: 27017
  • db: 'test'

Database Class

The Database class has the following methods:

  • authenticate: Works the same as pymongo's
  • drop: drops a database
  • drop_collection: drops a collection
  • get_collections: gets a list of collections in the database

and the following (read-only) properties:

  • host: MongoDB host
  • port: MongoDB port
  • name: database name

You can access the pymongo MongoClient with db.__client__ and the pymongo.database instance with db.__db__. Eventually, common operations will be accessible from the db object itself.


The DotDict class is a wrapper around python's default dict that allows attribute-style access to dict key-value pairs. In other words, the following accesses are the same:

>>> d = DotDict({'hello': 'world'})
>>> print d['hello']
>>> print d.hello

mongorm.Documents inherit from it to gain this feature. If you'd like to be able to refer to your nested documents with an attribute-style access, declare them as mongorm.DotDicts instead of {}s.

Defining Models

With a configured Database, as above, you can declare models as:

class SomeClass(db.Document):

These models will inherit the database connection from the db instance.

The following demonstrates some of the features of the Document class.

from mongorm import Field

class User(db.Document):
	# Override the collection name
	# Defaults to the underscored version of the class name
	__collection__ = 'auth_user'

	# Enforce validation on certain fields
	# All fields in this dict are considered required
	__fields__ = {

		# user.username is a required field of type str, without a default
		'username': Field.required(str),

		# user.age is a required field, with a default value
		'age': Field.required(int, 12)

		# is an optional field
		'name': Field.optional(str),

		# Nested document
		'nested': {
			'key_a': Field.required(str),
			'key_b': Field.optional(int)

		# List. Note that list elements are ALWAYS treated as optional
		'a_list': [ Field.optional(int) ]

		# List of objects
		'b_list' = [ {
			'key_a': Field.required(str),
			'key_b': Field.optional(int)
		} ]


	# Specify indices
	# These are directly passed to pymongo's collection.ensure_index
	__indices__ = [

		# Normal index over name field

		# Descending index over age
		Index([('age': pymongo.DESCENDING)]),

		# Compound index
		Index([('age', pymongo.DESCENDING), ('name', pymongo.ASCENDING)]),


	# Override the validate function
	# This gets called before a save operation
	# Error conditions should throw exceptions
	def validate(self):
		if self.age < 18:
			raise CannotLegallyDrinkError

The Document class also has some useful/essential methods:

  • dump_dict: returns a dict with keys that have camelCased names
  • dump_json: dumps the above dict as JSON
  • load_dict: updates self from a dict; it converts all keys to underscored_names
  • load_json: unmarshals JSON into a dict & performs the above operation
  • save: saves the document
  • delete: removes the document from the collection
  • validate_fields_extra: validates your fields based on the dict passed in. The dict uses the same format as fields above. This method can be used to make certain fields required only in specific situations.

and the following @classmethods:

  • from_json: returns a new instance of class constructed with the input JSON
  • find: calls pymongo.collection's find
  • find_one: calls pymongo.collection's find_one

In addition, the following methods are passed on to the pymongo.collection instance:

  • aggregate
  • count
  • create_index
  • ensure_index
  • drop_index
  • drop_indexes
  • index_information
  • reindex
  • group
  • distinct
  • write_concern
  • find_and_modify

Any arguments are passed verbatim to the pymongo.collection instance, so please refer to pymongos documentation.


All development happens on GitHub. Feel free to report any issues there.

If you wish to contribute code, please note the following:

  • The project is BSD-licensed, and is not copyleft
  • Please work off the master branch, and not any other published branches that might exist
  • Make sure you're following conventions
  • Github pull requests are fine, as are patches emailed to