pyFaceGraph

pyFaceGraph is a Python client library for the Facebook Graph API. It is being developed and maintained by iPlatform.

Installation

Via pip or easy_install:

pip install pyfacegraph
easy_install pyfacegraph

You can install an 'edge' version via git:

git clone 'git://github.com/iplatform/pyFaceGraph.git'
cd pyFaceGraph
python setup.py install

Graph API Tutorial

To begin using the API, create a new Graph with an access token:

>>> from facegraph import Graph
>>> g = Graph(ACCESS_TOKEN) #  Access token is optional.
>>> g
<Graph('https://graph.facebook.com/') at 0x...>

Addressing Nodes

Each Graph contains an access token and a URL. The graph you just created will have a URL of 'https://graph.facebook.com/' by default (this is defined as the class attribute Graph.API_ROOT). The URL is represented as a URLObject; see http://github.com/zacharyvoase/urlobject for more information. Remember that you can treat it exactly as you would a unicode string; it just supports a few more methods to enable easy URL manipulation.

>>> g.url
<URLObject(u'https://graph.facebook.com/') at 0x...>
>>> print g.url
https://graph.facebook.com/
>>> unicode(g.url)
u'https://graph.facebook.com/'

To address child nodes, Graph supports dynamic attribute and item lookups:

>>> g.me
<Graph('https://graph.facebook.com/me') at 0x...>
>>> g.me.home
<Graph('https://graph.facebook.com/me/home') at 0x...>
>>> g['me']['home']
<Graph('https://graph.facebook.com/me/home') at 0x...>
>>> g[123456789]
<Graph('https://graph.facebook.com/123456789') at 0x...>

Note that a Graph instance is rarely modified; these methods all return copies of the original object. In addition, the API is lazy: HTTP requests will never be made unless you explicitly make them.

Retrieving Nodes

You can fetch data by calling a Graph instance:

>>> about_me = g.me()
>>> about_me
Node({'about': '...', 'id': '1503223370'})

This returns a Node object, which contains the retrieved data. Node is a subclass of bunch.Bunch [pypi], so you can access keys using either attribute or item syntax:

>>> about_me.id
'1503223370'
>>> about_me.first_name
'Zachary'
>>> about_me.hometown.name
'London, United Kingdom'
>>> about_me['hometown']['name']
'London, United Kingdom'

Accessing non-existent attributes or items will return a Graph instance corresponding to a child node. This Graph can then be called normally, to retrieve the child node it represents:

>>> 'home' in about_me  # Not present in the data itself
False
>>> about_me.home
<Graph('https://graph.facebook.com/me/home') at 0x...>
>>> about_me.home()
Node({'data': [...]})

Creating, Updating and Deleting Nodes

With the Graph API, node manipulation is done via HTTP POST requests. Graph.post() will POST to the current URL, with varying semantics for each endpoint:

>>> post = g.me.feed.post(message="Test.")  # Status update
>>> post
Node({'id': '...'})
>>> g[post.id].comments.post(message="A comment.") # Comment on the post
Node({'id': '...'})
>>> g[post.id].likes.post()  # Like the post
True

>>> event = g[121481007877204]()
>>> event.name
'Facebook Developer Garage London May 2010'
>>> event.rsvp_status is None
True
>>> event.attending.post()  # Attend the given event
True

Any keyword arguments passed to post() will be added as form data. Consult the Facebook API docs for a complete reference on URLs and options.

Nodes can be deleted by adding ?method=delete to the URL; the delete() method is a helpful shortcut:

>>> g[post.id].delete()
True

Canvas Applications

Facebook is rolling out support for OAuth 2.0 in canvas applications with the new signed_request parameter; you can read about it here.

pyFaceGraph comes with a simple function for verifying and decoding this parameter:

>>> from facegraph import decode_signed_request, InvalidSignature

>>> decode_signed_request(APP_SECRET, GET['signed_request'])
{'0': 'payload', 'algorithm': 'HMAC-SHA256'}

>>> decode_signed_request('wrong-secret', GET['signed_request'])
Traceback (most recent call last):
    ...
InvalidSignature

Django

pyFaceGraph comes with basic support for building client applications in Django. At the moment, this consists of:

• Abstract class-based views for OAuth 2.0 authorization
• Abstract middleware to attach Graph instances to each request

OAuth

pyFaceGraph defines two abstract class-based views (using django-clsview) and a mixin; these help your app obtain OAuth 2.0 acess tokens to access the Graph API on behalf of a Facebook user.

• FacebookOAuthView: Defines some common methods useful to all Facebook-related CBVs. Should be used as an inherited ‘mixin’, i.e.:

  class Callback(CallbackView, FacebookOAuthView):
      pass
  

  Examples of methods implemented here are redirect_uri() (which must be the same for both the authorize and callback views), client_id() and client_secret() and fetch_url()

• AuthorizeView: Acts solely to redirect users to Facebook for authorization. You can override authorize_url() to change the URL which the user is redirected to. scope() and display() are two shortcuts for the Facebook-specific parameters, which you might need to alter for your requirements.

• CallbackView: The view the user is redirected back to from Facebook, upon successful authorization. It’s up to you to write the body of this (by defining __call__()), but you can call get_access_token() to fetch the access token. Suggestions include saving the access token to the database, or storing it in request.session.

See the source of facegraph.django.views for the ultimate reference guide to these views.

Example

Take a look at the example project in test/graphdevtools, which shows how to construct an app that interfaces with Facebook (and provides a dashboard for generating access tokens with varying permissions). You’ll need to create local_settings.py in that directory and define SECRET_KEY, FACEBOOK_CLIENT_ID and FACEBOOK_CLIENT_SECRET (for the latter two you need to register a Facebook application).

Middleware

The provided FacebookGraphMiddleware will attach a facegraph.Graph instance to each request, accessible as request.graph. You will need to subclass this middleware to define your own method of fetching the access token. For example:

## myapp/middleware.py:

from facegraph.django.middleware import FacebookGraphMiddleware

class GraphMiddleware(FacebookGraphMiddleware):
    def access_token(self, request):
        return request.session.get('access_token')

## settings.py:

MIDDLEWARE_CLASSES = (
  # ... 
  'myapp.middleware.GraphMiddleware',
  # ...
)

Note that this will still attach a Graph even if the access token is None. To check for authentication, just use if request.graph.access_token: in your view code.

There is also a piece of middleware for canvas applications: FacebookCanvasMiddleware will verify and decode Facebook’s new signed_request parameter for each request, and attach the value as request.fbrequest. By default, settings.FACEBOOK_APP_SECRET will be used as the application secret for signature verification, but you can subclass the middleware and override the app_secret() method to modify this behavior:

## myapp/middleware.py:

from django.conf import settings
from facegraph.django.middleware import FacebookCanvasMiddleware

class CanvasMiddleware(FacebookCanvasMiddleware):
    def app_secret(self, request):
        return settings.MY_FACEBOOK_APP_SECRET

## settings.py:

MIDDLEWARE_CLASSES = (
  # ... 
  'myapp.middleware.CanvasMiddleware',
  # ...
)

Finally, FacebookCanvasGraphMiddleware is a subclass of FacebookGraphMiddleware which will use the signed request as the source of the access token; no subclassing is necessary, just place it after CanvasMiddleware in MIDDLEWARE_CLASSES:

MIDDLEWARE_CLASSES = (
  # ... 
  'myapp.middleware.CanvasMiddleware',
  'facegraph.django.middleware.FacebookCanvasGraphMiddleware',
  # ...
)

(Un)license

This is free and unencumbered software released into the public domain.

Anyone is free to copy, modify, publish, use, compile, sell, or distribute this software, either in source code form or as a compiled binary, for any purpose, commercial or non-commercial, and by any means.

In jurisdictions that recognize copyright laws, the author or authors of this software dedicate any and all copyright interest in the software to the public domain. We make this dedication for the benefit of the public at large and to the detriment of our heirs and successors. We intend this dedication to be an overt act of relinquishment in perpetuity of all present and future rights to this software under copyright law.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

For more information, please refer to http://unlicense.org/