Skip to main content

Create, query and manage graphs of relationships between your Django models

Project description

connections is a small app for Django that allows you to model any kind of relationship between instances of any model. It’s primary use is for building a social graph that you can query and manage easily.


connections requires Python 2.6/3.2 or newer and Django 1.5 or newer.

How to install

Using pip:

$ pip install django-connections


$ git clone
$ cd django-connections
$ python install


Add connections to your settings:

    # ...

Then, run migrate:

$ python migrate

Note: If you’re running Django 1.6 or lower, you should run syncdb instead.

Using connections

With connections you essentially build directed graphs, where each node is a model instance and each edge is a Connection instance. Which two models a connection can connect, is determined by a Relationship instance that you predefine.

Defining relationships

Assume you’re LitHub, a social coding site in its infancy, and you need to let your users star repositories they find interesting. With connections, you would first define a relationship:

>>> from django.contrib.auth.models import User
>>> from connections import define_relationship
>>> from lithub.models import Repo
>>> repo_stars = define_relationship('star_repo', from_model=User, to_model=Repo)

define_relationship creates and registers a new Relationship instance between the given models, with the name 'star_repo'. Names of relationships must be unique across your project. You may alternatively specify the models of the relationship as strings, e.g. 'auth.User' or 'lithub.Repo'.

Any time you need to reference a relationship, you can either import the module variable (as defined above), or use connections.get_relationship(name).

Managing connections

Let’s say that milo found a nice Python project on LitHub that he’d like to star, for future reference. In connections this can be modelled by creating a connection between milo and the repository instance:

>>> milo = User.objects.get(pk=104)
>>> foopy = Repo.objects.get(pk=47)
>>> star_repo.create_connection(milo, foopy)
'star_repo (auth.User:104 -> lithub.Repo:47)'

Connections are unidirectional, meaning that if foo is connected with bar, the reverse – that bar is connected to foo – is not implied. If you’d like to model a symmetrical relationship, that is, one that only makes sense if both sides have agreed in the relationship (e.g. friendship or even marriage), you’d have to create two opposite connections, one for each side of the relationship.

Let’s see what repositories milo has starred:

>>> repo_stars.connected_objects(milo)
[<Repo: foopy>]

We can also query for the reverse, that is, what users have starred foopy:

>>> repo_stars.connected_to_objects(foopy)
[<User: milo>]

There are several other methods you may use to query or manage connections, that you may read about in API Reference.

Best practices

The preferred idiom is to define relationships in app/ files, keeping a module-global reference to each relationship instance, through which you manage connections between your model instances.

If you’re using Django 1.7 or later you may have any modules automatically imported at start-up:

    # ...

API Reference

Class Relationship

Represents a predefined type of connection between two nodes in a (directed) graph. You may imagine relationships as the kind of an edge in the graph.

>>> from connections.models import Relationship
>>> rel = Relationship('rel_name', from_content_type, to_content_type)

Instance properties


Returns a Connection query set matching all connections of this relationship.

Instance methods

create_connection(from_obj, to_obj)

Creates and returns a new Connection instance between the given objects. If a connection already exists, the existing connection will be returned instead of creating a new one.

get_connection(from_obj, to_obj)

Returns a Connection instance for the given objects or None if there’s no connection.

connection_exists(from_obj, to_obj)

Returns True if a connection between the given objects exists, else False.


Returns a Connection query set matching all connections with the given object as a source.


Returns a Connection query set matching all connections with the given object as a destination.


Returns a query set matching all connected objects with the given object as a source.


Returns an iterable of the IDs of all objects connected with the given object as a source (i.e. the Connection.to_pk values).


Returns a query set matching all connected objects with the given object as a destination.


Returns an iterable of the IDs of all objects connected with the given object as a destination (i.e. the Connection.from_pk values).

distance_between(from_obj, to_obj, limit=2)

Calculates and returns an integer for the distance between two objects. A distance of 0 means from_obj and to_obj are the same objects, 1 means from_obj has a direct connection to to_obj, 2 means that one or more of from_obj’s connected objects are directly connected to to_obj, and so on. limit limits the depth of connections traversal. Returns None if the two objects are not connected within limit distance.

Class Connection

Represents a connection between two nodes in the graph. Connections must be treated as unidirectional, i.e. creating a connection from one node to another should not imply the reverse.

Model attributes


The name of the relationship. To access the relationship instance, use the Connection.relationship property.


The primary key of the instance acting as source.


The primary key of the instance acting as destination.


A datetime instance of the time the connection was created.

Instance properties


Returns the Relationship instance the connection is about.


The source instance.


The destination instance.

Project details

Release history Release notifications | RSS feed

This version


Download files

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

Source Distribution

django-connections-0.1.tar.gz (11.2 kB view hashes)

Uploaded source

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page