How we built a Twitter Application

Twitter, just like many other Web2.0 sites, has an excellent API. So much so, writing a twitter app is considered to be “Hello World” of Web2.0 Programming.

I wrote an application recently. It is DM++. By default, this application allows you to receive direct messages from twitter users you have @replied even if you currently don’t follow them. You can even configure application to be able to receive direct messages from people you have @replied in only last 2 days, or receive Dm from all other twitter users.

Look how many people are currently complaining Cant DM U as U dont follow me or read our Faq

Now, let me illustrate some steps in building this application.

One of the modules is where I need to find out whether the userx follows usery. This request needs authentication headers added for twitter to respond, else U receive a “Not authorized” response

import urllib2,base64,simplejson

def is_follows(follower, following):

    theurl = 'http://twitter.com/friendships/exists.json?user_a=%s&user_b=%s'%(follower, following)

    username = 'uname'
    password = 'pwd'

    handle = urllib2.Request(theurl)

    authheader =  "Basic %s" % base64.encodestring('%s:%s' % (username, password))

    handle.add_header("Authorization", authheader)

    try:
        return simplejson.load(urllib2.urlopen(handle))
    except IOError, e:
        # This is reached when allocated API requests to IP are completed.
        print "parsing the is_follows json from twitter, failed"
    return False


if __name__=='__main__':
    print is_follows(u'scorpion032',u'scobleizer')

This is an independent module, intended so, as this functionality is best that way, simplest without the bloat of any other functions and it is portable.

There is a very good Python-twitter wrapper by Google, many functions on twitter are now very simple, for anything more than finding out if userx follows usery as above, this can be very useful.

import twitter
api=twitter.Api('uname','pwd')

Thats it and you have an authenticated api object from which U can do nearly all twitter functions. Infact it is working with this wrapper, that tempted me to write this application. There are a host of things you can do with this wrapper including what is shown below, that I have used.

This code retrieves all Direct Messages that are received since it was last retrieved. We store the fetchinfo table pk=1 field with the last date time it was retrieved

try:
    f=FetchInfo.objects.get(pk=1).last_datetime
except:
    f=None
dm_list=api.GetDirectMessages(since=f)

Following snippet sends the Direct Messages to the users that can receive it, else it is marked filtered. ‘canhesend()’ is a function where we find verify if the user is able to send direct messages based on his access_level setting

    if canhesend(adm.sent_by.username,user):
            try:
                api.PostDirectMessage(user,"via @"+adm.sent_by.username+": "+adm.text)
                api.PostDirectMessage(adm.sent_by.username,"Your Dm has been sent to @"+user)
                adm.is_sent=True
                adm.save()
            except HTTPError:
                adm.is_filter=True
                adm.save()
        else:
            adm.is_filter=True
            adm.save()

adm is the parameter that loops on direct messages.

Among other important things see how we seperate the usernames and text which are sent to us, there is a very nice input from stackoverflow, which we incorporated.

We have a text of the format '@abc [@def ..] This part is text' which we want seperated to user names and text part. We want retrieved in the following format:

l=['abc','def']
s='This part is text'

This is how we went about doing it:

import re
rx = re.compile("((?:@\w+ +)+)(.*)")
t='@abc   @def  @xyz Hello this part is text and my email is [email protected]'
a,s = rx.match(t).groups()
l = re.split('[@ ]+',a)[1:-1]
print l
print s

Would you like to download 10+ free Django and Python books? Get them here