A twitter bot

August 21, 2009

One of the first experiments that I did with python and twitter was a BOT. I was interested in testing how would an interactive application work using twitter, and a reasonable model for me is that it would answer back when a user sent a message, after performing a given action.

To test it, I built this bot, and it is kinda running at http://twitter.com/interweb. Send it a message and it answers back with the “feeling lucky” result in a google query. You may need to follow it to see the answer, according to your security options. Try it with @interweb test.

The code is based around the latest version of python twitter library, and it pretty straight forward to use and adapt for other bots. It runs on the crontab, polling the api for new messages, and save the answered ones in a pickled db file.

For a production grade bot, it should use a queue and start as a daemon, with a slower time to check the api for new statuses. Don’t forget to ask for a bigger rate limit to twitter, so your application dont get blacklisted. Also, one can take advantage of geocoding data to narrow the answer, along with other truly web 3.14 ideas !

As usual, here goes the code.  Cheers !


import twitter
from urllib import urlencode
import simplejson
import urllib2
import os
import pickle

GOOGLE_API_URL = 'http://ajax.googleapis.com/ajax/services/search/web?v=1.0&%s'

my_user_name = ''  #here goes your twitter username
my_pass = '' #here goes the passwd
filename = "statuses.db"

if os.path.exists(filename):
 last_statuses = pickle.load(file(filename, 'r+b'))
else:
 last_statuses = {}

api = twitter.Api(username=my_user_name, password=my_pass)

def query(wot):
 params = {'q': wot, 'rsz': 'small'}
 url = GOOGLE_API_URL % urlencode(params)
 try:
 # Query Google
 resp = urllib2.urlopen(url)
 except URLError, e:
 print e.reason
 return "Ouch, got an error trying to search what you asked me."

 print resp.geturl()

 response = resp.read()
 # Parse response
 try:
 data = simplejson.loads(response)
 results = data['responseData']['results']

 if results:
 resp.close()
 return 'Feeling lucky: %s (+ %s results from Google)' % (results[0]['unescapedUrl'],
 data['responseData']['cursor'].get('estimatedResultCount'))

 except (ValueError, KeyError), e:
 print "Couldn't parse Google response due to %r: %s" % (e, response)
 return "Ouch, got some unknown error."

 resp.close();
 return "Google returned 0 results for: %s" % wot

reps = api.GetReplies()
updated = False

for s in reps:
 if s.user.screen_name == my_user_name:
 continue
 if s.id in last_statuses.keys():
 continue
 print "%s (status id: %s): %s" % (s.user.screen_name, s.id, s.text)
 q = s.text.replace("@%s"% s.in_reply_to_screen_name, '')
 r = query(q)
 api.PostUpdate('@%s %s' % (s.user.screen_name, r))
 last_statuses[s.id] = True
 updated = True
 print last_statuses

if updated == True:
 pickle.dump(last_statuses, file(filename, 'w+b'))

Advertisements

2 Responses to “A twitter bot”


  1. Nice work!

    By any chance do you have a similar one in php?

  2. gm Says:

    No, I never did that in PHP, alto I think it may not be difficult. Good luck.


Leave a Reply

Please log in using one of these methods to post your comment:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: