script/stream/recorder_tweetstream.py
author Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
Tue, 18 Jan 2011 10:08:03 +0100
changeset 11 54d7f1486ac4
parent 9 bb44692e09ee
child 15 5d552b6a0e55
permissions -rw-r--r--
implement get_oauth_token

from getpass import getpass
from iri_tweet import models, utils
from optparse import OptionParser
from sqlalchemy.orm import sessionmaker
from sqlite3 import *
import logging
import os
import socket
import sys
import tweetstream
socket._fileobject.default_bufsize = 0



#columns_tweet = [u'favorited', u'truncated', u'text', u'created_at', u'source', u'in_reply_to_status_id', u'in_reply_to_screen_name', u'in_reply_to_user_id', u'geo', u'id', u'user']
columns_tweet = [u'user', u'favorited', u'contributors', u'truncated', u'text', u'created_at', u'retweeted', u'in_reply_to_status_id_str', u'coordinates', u'in_reply_to_user_id_str', u'entities', u'in_reply_to_status_id', u'place', u'in_reply_to_user_id', u'id', u'in_reply_to_screen_name', u'retweet_count', u'geo', u'id_str', u'source']
#columns_user = [u'id', u'verified', u'profile_sidebar_fill_color', u'profile_text_color', u'followers_count', u'protected', u'location', u'profile_background_color', u'utc_offset', u'statuses_count', u'description', u'friends_count', u'profile_link_color', u'profile_image_url', u'notifications', u'geo_enabled', u'profile_background_image_url', u'screen_name', u'profile_background_tile', u'favourites_count', u'name', u'url', u'created_at', u'time_zone', u'profile_sidebar_border_color', u'following']
columns_user = [u'follow_request_sent', u'profile_use_background_image', u'id', u'verified', u'profile_sidebar_fill_color', u'profile_text_color', u'followers_count', u'protected', u'location', u'profile_background_color', u'id_str', u'utc_offset', u'statuses_count', u'description', u'friends_count', u'profile_link_color', u'profile_image_url', u'notifications', u'show_all_inline_media', u'geo_enabled', u'profile_background_image_url', u'name', u'lang', u'following', u'profile_background_tile', u'favourites_count', u'screen_name', u'url', u'created_at', u'contributors_enabled', u'time_zone', u'profile_sidebar_border_color', u'is_translator', u'listed_count']
#just put it in a sqlite3 tqble


class ReconnectingTweetStream(tweetstream.TrackStream):
    """TweetStream class that automatically tries to reconnect if the
    connecting goes down. Reconnecting, and waiting for reconnecting, is
    blocking.

    :param username: See :TweetStream:

    :param password: See :TweetStream:

    :keyword url: See :TweetStream:

    :keyword reconnects: Number of reconnects before a ConnectionError is
        raised. Default is 3

    :error_cb: Optional callable that will be called just before trying to
        reconnect. The callback will be called with a single argument, the
        exception that caused the reconnect attempt. Default is None

    :retry_wait: Time to wait before reconnecting in seconds. Default is 5

    """

    def __init__(self, user, password, keywords, url="track", reconnects=3, error_cb=None, retry_wait=5, **kwargs):
        self.max_reconnects = reconnects
        self.retry_wait = retry_wait
        self._reconnects = 0
        self._error_cb = error_cb
        super(ReconnectingTweetStream,self).__init__(user, password, keywords, url, **kwargs)

    def next(self):
        while True:
            try:
                return super(ReconnectingTweetStream,self).next()
            except tweetstream.ConnectionError, e:
                self._reconnects += 1
                if self._reconnects > self.max_reconnects:
                    raise ConnectionError("Too many retries")

                # Note: error_cb is not called on the last error since we
                # raise a ConnectionError instead
                if  callable(self._error_cb):
                    self._error_cb(e)

                time.sleep(self.retry_wait)
        # Don't listen to auth error, since we can't reasonably reconnect
        # when we get one.



def process_tweet(tweet, session, debug):
    
    logging.debug("Process_tweet :" + repr(tweet))
    processor = utils.TwitterProcessor(tweet, None, session)
    processor.process()

def main(username, password, track, session, debug, reconnects):

    username = username or raw_input('Twitter username: ')
    password = password or getpass('Twitter password: ')

    track_list = track or raw_input('Keywords to track (comma seperated): ').strip()
    track_list = [k for k in track_list.split(',')]

    stream = ReconnectingTweetStream(username, password, track_list, reconnects=reconnects)
    try:
        for tweet in stream:
            process_tweet(tweet, session, debug)
            session.commit()
    finally:
        stream.close()
        
def get_options():
    parser = OptionParser()
    parser.add_option("-f", "--file", dest="filename",  
                      help="write tweet to FILE", metavar="FILE", default="enmi2010_twitter.db")
    parser.add_option("-u", "--user", dest="username",
                      help="Twitter user", metavar="USER", default=None)
    parser.add_option("-w", "--password", dest="password",
                      help="Twitter password", metavar="PASSWORD", default=None)
    parser.add_option("-t", "--track", dest="track",
                      help="Twitter track", metavar="TRACK")
    parser.add_option("-n", "--new", dest="new", action="store_true",
                      help="new database", default=False)
    parser.add_option("-r", "--reconnects", dest="reconnects",
                      help="Reconnects", metavar="RECONNECTS", default=10, type='int')
    
    utils.set_logging_options(parser)

    return parser.parse_args()
    

if __name__ == '__main__':
    

    (options, args) = get_options()
    
    utils.set_logging(options)
        
    if options.debug:
        print "OPTIONS : "
        print repr(options)
    
    if options.new and os.path.exists(options.filename):
        os.remove(options.filename)
    
    engine, metadata = models.setup_database('sqlite:///'+options.filename, echo=(options.debug))
    Session = sessionmaker(bind=engine)
    session = Session()

    try:
        try:
            main(options.username, options.password, options.track, session, options.debug, options.reconnects)
        except KeyboardInterrupt:
            print '\nGoodbye!'
        session.commit()
    finally:
        session.close()