Niall's Weblog

Recently I have been working on a pretty simple Facebook application. I’ve found that the tough thing about writing a Facebook app is not the app per se, but figuring out what a Facebook app actually is, and how it is supposed to work! Anyway, I’m hoping to shed some light on the subject as I figure it out. This first post is mostly some background and description of the various Python implementations of the Facebook API.

Unfortunately, Facebook’s documentation is pretty bad, with a poorly maintained wiki full of out-of-date and plain misleading information. Also, it is all heavily biased toward PHP. On the one hand, this makes a certain amount of sense, since PHP is a very popular Web scripting language, and Facebook itself is implemented in PHP. However, Facebook apps are fundamentally a HTTP-level construct – that is, Facebook makes HTTP requests of various sorts to your application. HTTP is obviously language independent – any language capable of speaking HTTP can be use to implement a Facebook app. I feel that Facebook should focus more on the system at the HTTP level rather than having a load of PHP code and other examples. Anyway, I digress – what you really care about is writing Facebook apps in Python!

There really two distinct things going on in a Facebook app. One is answering HTTP requests from their servers in a specific way. As I wrote above, any language capable of speaking HTTP can handle this. The second is sending requests back to Facebook, in the form that they expect, with data correctly marshaled and so on. While you could implement this interface yourself, there are existing libraries which do all this marshaling and de-marshaling for you. The two I found for Python were minifb and PyFacebook.

I chose to go with PyFacebook, mostly because it seems actively maintained and has a little more documentation than minifb. The biggest gripe I have with PyFacebook is that it is very Django-centric. The second biggest gripe I have is that it is not very well documented. There is a tutorial on the Facebook wiki which is sadly lacking polish, but its still better than nothing.

The PyFacebook source and documentation talk about some Pylons middleware stuff. I tried to use this but found it to be more hassle than its worth – you can just use their provided Facebook class yourself and avoid some dicking about with middleware.py and so on. Actually all you need if you go this route is their __init__.py file. I dropped this in a subdir named ‘facebook’ in my Pylons lib directory. Then you can just do this to use it in your controller:

1

from intothebin.lib.facebook import Facebook

Then from your callback URL handler, you can do stuff like:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19

def callback(self):
    def redirect(fb, url):
        if fb.in_canvas:
            return '<fb:redirect url="%s" />' %(url, )
        else:
            response.status_int = 302
            response.headers['location'] = url
            return 'Moved temporarily'
    api_key = config['pyfacebook.apikey']
    secret_key = config['pyfacebook.secret']
    appid = config['pyfacebook.appid']
    auth_token = request.params.get('auth_token', None)
    fb = Facebook(api_key, secret_key, app_name='myapp',
            callback_path='/myapp/callback',
            auth_token=auth_token)
    if not fb.check_session(request) or not auth_token:
        log.info("got an unauthenticated session request")
        return redirect(fb, fb.get_login_url())
    fb.auth.getSession()

I’ll talk more about what all this stuff means, and what you actually need to do, in the next post. Stay tuned for more!