def backfill_tweets(self):
""" Get all tweets from the author of the source tweet since the last
time this method was called, adding them to the database. """
name = self.source.user.screen_name
new_tweets = search(['@'+name, 'from:'+name], self.last_tweet_id)
if not new_tweets:
return
logger.info("Backfilling %s tweets for %s, last id %s"
% (len(new_tweets), name, self.last_tweet_id))
new_tweets = [db.merge(tweet_to_Tweet(t)) for t in new_tweets]
new_ids = [t.id for t in new_tweets]
stray = [db.merge(t) for t in self.stray_tweets if t.id not in new_ids]
logger.info("%s other stray tweets" % len(stray))
# doing the db.merge already added the tweets
self.tweets.extend(stray)
self.tweets.extend(new_tweets)
self.update_last_tweet_id()
self.stray_tweets = []
logger.info("%s total tweets in %s" % (len(self.tweets), self))
def query_twitter(how_long=0, interval=5):
""" Interface function """
reset_location_cache()
# can send 180 requests per 15 min = 5 sec
start = time()
# make sure we don't create duplicates.
# keeping track of this ourselves saves many db hits
# if we don't specify go indefinitely
last_tweet_id = 0
while time() - start < how_long:
tweets = search(search_terms, last_tweet_id)
if not tweets: # if we dont get anything back, sleep and try again
sleep(interval)
continue
# if a retrieved tweet has a loc/user with a matching ID already in the
# db, that loc/user is updated instead of a new one added, bc of merge
try:
db.add_all([db.merge(tweet_to_Tweet(t)) for t in tweets])
db.commit()
last_tweet_id = tweets[0]['id_str']
except OperationalError:
pass
sleep(interval)
def test_tweets_to_Tweets():
""" Sample tweets should be converted properly """
fields = ['text', 'id_str', 'user']
tweets = [util.tweet_to_Tweet(sample.tweets[0], fields),
util.tweet_to_Tweet(sample.tweets[1], fields)]
sample.assert_matching_tweets(tweets)
def rec(tweet):
receive_tweet(incidents, search_queue, tweet_to_Tweet(tweet))
