def post(self):
try:
self.open()
self.tweets = []
self.lshj = LshJaccard(num_bands=20, rows_per_band=10)
except:
self.session['tw_auth'] = None
self.redirect('/')
return
while True:
try:
for shingles_list, original_document in shingle_generator(self.get_next()):
# get minhash signatures for each shingle list
min_hash_signatures = minhash.run(shingles_list)
#create document and run LSH for Jaccard Distance
doc_obj = JaccardDocument(original_document, shingles_list, min_hash_signatures)
logging.info('Running Jaccard LSH Current Tweet: %s', original_document)
results = self.lshj.run(doc_obj)
if results:
logging.info('.....RESULTS.....')
logging.info('.....score: %s', str(results['score']))
logging.info('.....match_found: %s', str(results['match_found']))
logging.info(results['document_1'])
logging.info(results['document_2'])
logging.info('---------------------------------------------------')
logging.info('Results: %s', str(results['score']))
#TODO update the code the read this and prints out score, docs and match boolean flag
#self.tweets.append(str(results['score']))
except NotFound as nf:
logging.error('TwitterReadNode.GetNext completed, %s', nf.value)
break
self.close(save=True)
class TwitterReadNode(TwitterGetTweets, PipeNode):
def open(self):
if not ('tw_auth' in self.session):
logging.error("Not logged in into twitter, tw_auth key not found in session dict")
raise NotLoggedIn("Not logged in into twitter...auth not in session dict")
auth = self.session['tw_auth']
api = tweepy.API(auth)
if not api:
logging.error("Not logged in into twitter, no tweepy api")
raise NotLoggedIn("Not logged in into twitter...no tweepy api")
# Read tweets from the stream
self.twitter_listener = super(TwitterReadNode, self).get()
logging.info('TwitterReadNode.Open completed')
def get_next(self):
while not self.twitter_listener.queue.empty():
yield self.twitter_listener.queue.dequeue()
raise NotFound('Tweets exhausted')
def close(self, save=False):
tweets = '<br/>\n— '.join(self.tweets)
if save:
all_the_tweets = TwitterStreamDump(content=tweets)
all_the_tweets.put()
logging.info('TwitterReadNode.Close completed')
num_tweets_status = "Number of tweets received - %s" % str(self.twitter_listener.tweet_counter)
banner = 'Done getting tweets at %s. %s' % (time.strftime("%a, %d %b %Y %H:%M:%S +0000", time.gmtime()), num_tweets_status)
self.session['tw_status'] = banner
self.session['tweets'] = tweets
self.redirect('/')
def post(self):
try:
self.open()
self.tweets = []
self.lshj = LshJaccard(num_bands=20, rows_per_band=10)
except:
self.session['tw_auth'] = None
self.redirect('/')
return
while True:
try:
for shingles_list, original_document in shingle_generator(self.get_next()):
# get minhash signatures for each shingle list
min_hash_signatures = minhash.run(shingles_list)
#create document and run LSH for Jaccard Distance
doc_obj = JaccardDocument(original_document, shingles_list, min_hash_signatures)
logging.info('Running Jaccard LSH Current Tweet: %s', original_document)
results = self.lshj.run(doc_obj)
if results:
logging.info('.....RESULTS.....')
logging.info('.....score: %s', str(results['score']))
logging.info('.....match_found: %s', str(results['match_found']))
logging.info(results['document_1'])
logging.info(results['document_2'])
logging.info('---------------------------------------------------')
logging.info('Results: %s', str(results['score']))
#TODO update the code the read this and prints out score, docs and match boolean flag
#self.tweets.append(str(results['score']))
except NotFound as nf:
logging.error('TwitterReadNode.GetNext completed, %s', nf.value)
break
self.close(save=True)