Skip to main content

Simple PyMongo "schema-less" object wrapper

Project description

This library is a simple "schema-less" object wrapper around the
pymongo library (
Mogo provides helpers to use PyMongo in an MVC environment
(things like dot-attribute syntax, model methods,
reference fields, etc.)

While pymongo is straightforward to use and really flexible, it
doesn't help with MVC because you are working with plain dicts
and can't attach model logic anywhere.

Mogo is licensed under the Apache License, Version 2.0

* Put classes / structure around pymongo results
* Models are dicts, so dot-attribute or key access is valid. Dot attribute
gives "smart" values, key access gives "raw" pymongo values.
* Support for specifiying Field() attributes without requiring
them or enforcing types.
* Simple ReferenceField implementation.

* PyMongo -

You can install it from PyPI with:

pip install mogo

Alternatively you should be able to grab it via git and run the following

python install

To run the tests, make sure you have a MongoDB instance running
on your local machine. It will write and delete entries to
mogotest db, so if by some bizarre coincidence you have / need that,
you might want to alter the DBNAME constant in the mogo/

After installation, or from the root project directory, run:

nosettest tests/

If you don't have nose, it's available with:

pip install nose


All the major classes and functions are available under the top level
mogo module:

import mogo
# or
from mogo import Model, Field, connect, ReferenceField


Mogo uses a single global connection, so that once you connect, you can
just start accessing your model class methods. Connecting looks like:

from mogo import connect

connect("my_database") # connects to a local mongodb server with default port
connect("foobar", host="", port=28088)
connect(uri="mongodb://user:pass@") # for heroku, etc.

If you need to use an alternate connection for a chunk of code, without
losing your main connection, you can use the following style:

from mogo import connect, session

# do normal stuff
with mogo.session("my_alternate_database"):
# do stuff with other database

Models are subclasses of dicts with some predefined class and instance
methods. They are designed so that you should be able to use them
with an existing MongoDB project (DATA BE WARNED: THIS IS ALPHA!)
All you need is a class with the proper collection name, and connect
to the DB before you access it.

Yes, this means the most basic example is just a class with nothing:

class Hero(Model):

You can now do things like:

hero = Hero.find({"name": "Malcolm Reynolds"}).first()

By default, it will use the lowercase name of the model as the
collection name. So, in the above example, the equivalent pymongo
call would be:

db.hero.find({"name": "Malcolm Reynolds"})[0]

Of course, Models are much more useful with methods:

class Hero(Model):
def swashbuckle(self):
print "%s is swashbuckling!" % self["name"]

mal = Hero.find({"name": "Mal"}).first()
# prints "Mal is swashbuckling!"

Since Models just subclass dictionaries, you can use (most) of the
normal dictionary methods (see `update` later on):

hero = Hero.find_one({"name": "Book"})
hero.get("powers", ["big darn hero"]) # returns ["big darn hero"]
hero_dict = hero.copy()
for key, value in hero.iteritems():
print key, value

To save or update values in the database, you use either `save` or
`update`. (Imagine that.) If it is a new object, you have to `save`
it first:

mal = Hero(name="Malcom Reynolds")

`save` will always overwrite the entire entry in the database.
This is the same behavior that PyMongo uses, and it is helpful for
simpler list and dictionary usage:

zoe = Hero(name="Zoe", powers=["warrior woman"])
zoe["powers"].append("big darn hero")

...however, this can ultimately be inefficient, not to
mention produce race conditions and have people saving over each
other's changes.

This is where `update` comes in. Note that the `update` method does
NOT function like the dictionary method. It has two roles,
depending on whether it is called from a class or from an instance.

If it is called from a class, it just passes everything on to PyMongo
like you might expect:

Hero.update({"name": "Malcolm Reynolds"},
{"$set":{"name": "Capt. Tightpants"}}, safe=True)
# equals the following in PyMongo
db.hero.update({"name": "Malcolm Reynolds"},
{"$set":{"name": "Capt. Tightpants"}}, safe=True)

If it is called from an instance, it uses keyword arguments to set
attributes, and then sends off a PyMongo "$set" update:

hero = Hero.find_one({"name": "River Tam"})
hero.update(powers=["telepathy", "mystic weirdness"])
# equals the following in PyMongo
hero = db.hero.find_one({"name": "River Tam"})
db.hero.update({"_id": hero["_id"]},
{"$set": {"powers": ["telepathy", "mystic weirdness"]}})

(BETA) If you call it from a cursor, it will use the query you
originally provided to the cursor. This does not currently respect
additional filtering like `where()`, does not check types when
setting values, and has not been exhaustively tested. (So beware.)

hero_cursor = Hero.find({"name": {"$in": ["River", "Simon"]}})
hero_cursor.update({"$push": {"powers": "siblingness"}})
# or, for you keyword-liking people...

Using a Field is (usually) necessary for a number of reasons. While
you can remain completely schemaless in Mongo, you will probably go
a little nutty if you don't document the standard top level fields
you are using.

Fields just go on the model like so:

class Hero(Model):
name = Field()

...and enable dot-attribute access, as well as some other goodies.
Fields take several optional arguments -- the first argument is a
type, and if used the field will validate any value passed as an
instance of that (sub)class. For example:

class Hero(Model):
name = Field(unicode)

# the following will raise a ValueError exception...
wash = Hero(name="Wash")
# but this is fine
wash = Hero(name=u"Wash")

If you don't want this validation, just don't pass in any type. If you
want to customize getting and setting, you can pass in `set\_callback`
and `get\_callback` functions to the Field constructor:

class Ship(Model):
type = Field(set_callback=lambda x: "Firefly")

ship = Ship(type="firefly")
print ship.type #prints "Firefly"
ship.type = "NCC 1701"
print ship.type #prints "Firefly"
# overwriting the "real" stored value
ship["type"] = "Millenium Falcon"
print ship.type # prints "Millenium Falcon"

You can also pass an optional default=VALUE, where VALUE is either a
static value like "foo" or 42, or it is a callable that returns a static
value like time.time() or (Thanks @nod!)

class Ship(Model):
name = Field(unicode, default=u"Dormunder")

The ReferenceField class allows (simple) model references to be used.
The "search" class method lets you pass in model instances and compare.

So most real world models will look more this:

class Ship(Model):
name = Field(unicode, required=True)
age = Field(int, default=10)
type = Field(unicode, default="Firefly")

def new(cls, name):
""" Creating a strict interface for new models """

def crew(self):

class Crew(Model):
name = Field(unicode, required=True)
joined = Field(float,, required=True)
ship = ReferenceField(Ship)

...and simple usage would look like this:

serenity ="Serenity")
mal = Crew(name=u"Malcom Reynolds", ship=None)
mal.ship = serenity

print [ for person in serenity.crew]
# results in [u"Malcom Reynolds",]
print mal.joined
# prints out the datetime that the instance was created

Note -- only use a ReferenceField with legacy data if you have been
storing DBRef's as the values. If you've just been storing ObjectIds or
something, it may be easier for existing data to just use a Field() with
a `(set|get)\_callback` do the retrieval logic yourself.

MongoDB lets you store any fields in any collection -- this means it is
particularly well suited for storing and querying across inheritance
relationships. I've recently added a new model type of `PolyModel` that
lets you define this in a (hopefully) simple way.

class Person(PolyModel):
""" The 'base' person model """
name = Field(unicode, required=True)
role = Field(unicode, default=u"person")

# custom method
def is_good(self):
""" All people are innately good. :) """
return True

# required to determine what `type` something is
def get_model_key(self):
return "role"

As you can see, we use the "role" field to determine what type a person
is -- by default, they are all just "person" and therefore should return
a Person instance. We need to register some new people types:

class Villain(Person):
role = Field(unicode, default=u"villain")

# Overwriting method
def is_good(self):
""" All villains are not good """
return False

class FlipFlopper(Person):
role = Field(unicode, default=u"questionable")
alliance = Field(unicode, default=u"good")

def is_good(self):
return self.alliance == "good"

def trade_alliance(self):
if self.alliance == "good":
self.alliance = "bad"
self.alliance = "good"

The PolyModel.register decorator takes an optional value argument, which
is what is used to compare to the field specified by `get_model_key` in
the base model. It works with the following pseudo-logic:

* Create a new Person instance (either from the DB or __init__)
* key = Person.get_model_key() # in this case, it's "role"
* Get current value of "role" (or use the default)
* Check the registered models, find one that matches the role value
* If a registered model class is found, use that.
* Otherwise, use the base class (Person in this case)

Using the above classes that we created / registered, here's a usage example:

simon = Person(name="Simon Tam")
simon.is_good() # True
badger = Villain(name="Badger")
badger.is_good() # False
jayne = FlipFlopper(name="Jayne")

Person.find().count() # should be 3
jayne = Person.find(name="Jayne")
isinstance(jayne, FlipFlopper) # True
jayne.is_good() # True
jayne.is_good() # False

Villain.find().count() # should be 1

* Mailing List Web:
* Mailing List Address:

If you play with this in any way, I'd love to hear about it.

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Files for mogo, version 0.4.0
Filename, size File type Python version Upload date Hashes
Filename, size mogo-0.4.0.tar.gz (13.6 kB) File type Source Python version None Upload date Hashes View

Supported by

Pingdom Pingdom Monitoring Google Google Object Storage and Download Analytics Sentry Sentry Error logging AWS AWS Cloud computing DataDog DataDog Monitoring Fastly Fastly CDN DigiCert DigiCert EV certificate StatusPage StatusPage Status page