def enrich_tweet(self, scrap):
"""Function that calls all tweet fetchers """
self._fetch_counters(scrap)
self._remove_punctuations()
self._fetch_text(scrap)
self._fetch_hashtags(scrap)
self._fetch_date(scrap, False)
self._fetch_username(scrap)
self._fetch_tweet_id(scrap)
self._sentiment = TwitterClient().get_sentiment(self.text)
if args.train:
print("Forcing the model to retrain")
model = train(args.eval)
else:
try:
model = joblib.load(model_filename)
except:
model = train(args.eval)
cv = model['vec']
clf = model['clf']
if args.tweets_file is not None:
with open(args.tweets_file, 'r') as f:
tweets = f.read().split('\n')
df = pd.DataFrame(tweets, columns=['text'])
else:
client = TwitterClient()
tweets = client.get_tweets(query=args.query, count=200)
df = pd.DataFrame(tweets)
tc = TextCleaner()
cleaned_text = tc.fit_transform(df.text)
counts = cv.transform(cleaned_text)
preds = clf.predict(counts)
for text, pred in zip(df.text, preds):
print("\nSentiment: %s. Tweet: %s" % (pred, text))
def __init__(self, cfg):
TwitterClient.__init__(self, cfg)
psycopg2.extensions.register_type(psycopg2.extensions.UNICODE)
self.db = psycopg2.connect(c.get('planet','db'))
from testserver import Server # se server.py
import logging
LOG_FILE = 'kwitter.log'
logging.basicConfig(filename=LOG_FILE, level=logging.DEBUG)
if __name__ == '__main__':
# Ladda in tweets att testa, från fil
with open("test_tweets.txt", "r") as source_tweets:
tweets = source_tweets.readlines()
# Samt felmeddelande
with open("test_errormsg.txt", "r") as error_source:
error_msg = error_source.readlines()
title = "Kwitter" # Fönstertitel
client = TwitterClient() # Testverktyget
try:
# Försök starta Firefox + ladda sidan
server = Server()
client.get_url(server.port)
client.assert_connection(title)
try:
msg = "Startar testning"
logging.info(msg)
print(msg)
client.test_tweets(tweets, error_msg[0])
client.test_checkboxes()
client.test_refresh()
logging.basicConfig(filename=LOG_FILENAME,
level=logging.INFO,
format='%(asctime)s:%(levelname)s:%(message)s')
logging.info("Starting tweetStream %s" % __version__)
if len(sys.argv) < 2:
logging.fatal("No username given! Stopping...")
print "Usage: %s [username] [-c]" % sys.argv[0]
sys.exit(1)
uname = sys.argv[1]
logging.info("Given username %s" % uname)
client = TwitterClient(uname)
if len(sys.argv) == 3 and sys.argv[2] == '-c':
client.start()
shown = []
while True:
try:
time.sleep(3)
c = client.storage.get_all_sorted()
for t in c:
if t.guid not in shown:
print t
shown.append(t.guid)
except KeyboardInterrupt:
client.stop()
sys.exit(0)
else:
help=
"Number of tweets to consider - *Warning* This may trigger multiple search requests. Each twitter search requests has a limitation 100 tweets. Default will be best effort",
default=1000)
parser.add_argument("-st",
"--search_type",
help="recent(default)/popular/mixed",
default="recent")
parser.add_argument("-lang",
"--language",
help="language tweets",
default="en")
### End of arguments ###
args = parser.parse_args()
tClient = TwitterClient()
if args.tweet_count <= 100:
search_results = tClient.search_until(k=args.keyword,
tc=args.tweet_count,
st=args.search_type,
ud=args.since_date,
lan=args.language)
opin_dict = tClient.opinion_mining(search_results['statuses'])
'''
te_search_results = twitter.search(q=keyword, count=tweet_count, result_type='mixed', since_id=since, lang="te")
trans = Translator()
for result in te_search_results['statuses']:
print result['text']
trans.translate(result['text'])
'''
from twitterclient import TwitterClient
import argparse
from collections import OrderedDict
import pprint
### Start of arguments ###
parser = argparse.ArgumentParser()
parser.add_argument("Movie", help = "keyword to search for")
parser.add_argument("since_date", help = "The date from which tweets should be analyzed")
parser.add_argument("-v", "--verbose", help = "Print traces for debugging", type=bool, default=False)
### End of arguments ###
####main####
args = parser.parse_args()
tClient = TwitterClient()
ReviewRules = [('VeryGood' , 90),
('Good' , 80),
('Watchable', 70),
('Average' , 60),
('Bad' , 30),
('VeryBad' , 0)]
multi_search_results = tClient.multi_search_until(k=args.Movie, ud=args.since_date, verbose=args.verbose)
total_tweets, opinion_dictionary = tClient.review_mining_multi(multi_search_results, ReviewRules)
'''
opinion dictionary is a dictionary of dictionaries has the following structure
{Sentiment : {tweet_id: tweet_text}}
def __init__(self, cparser, user, listname, token, secret, members): self.confwrap = TwitterConfWrapper(cparser, user, listname, token, secret) self.members = set([self._normalize_handle(r[0]) for r in members]) TwitterClient.__init__(self, self.confwrap)
