from iri_tweet import models, processor, utils
from optparse import OptionParser
import anyjson
import re
import twitter
def get_option():
parser = OptionParser()
parser.add_option("-l", "--log", dest="logfile",
help="log to file", metavar="LOG", default="stderr")
parser.add_option("-v", dest="verbose", action="count",
help="verbose", metavar="VERBOSE", default=0)
parser.add_option("-q", dest="quiet", action="count",
help="quiet", metavar="QUIET", default=0)
parser.add_option("-Q", dest="query",
help="query", metavar="QUERY")
parser.add_option("-P", dest="rpp", metavar="RPP", default="50",
help="Result per page")
parser.add_option("-t", dest="token_filename", metavar="TOKEN_FILENAME", default=".oauth_token",
help="Token file name")
parser.add_argument("-k", "--key", dest="consumer_key",
help="Twitter consumer key", metavar="CONSUMER_KEY", required=True)
parser.add_argument("-s", "--secret", dest="consumer_secret",
help="Twitter consumer secret", metavar="CONSUMER_SECRET", required=True)
return parser.parse_args()
def get_auth(options, access_token):
consumer_key = options.consumer_key
consumer_secret = options.consumer_secret
auth = twitter.OAuth(token=access_token[0], token_secret=access_token[1], consumer_key=consumer_key, consumer_secret=consumer_secret)
return auth
if __name__ == "__main__":
(options, args) = get_option()
access_token = utils.get_oauth_token(consumer_key=options.consumer_key, consumer_secret=options.consumer_secret, token_file_path=options.token_filename)
auth = get_auth(options, access_token)
t = twitter.Twitter(domain="api.twitter.com",api_version="1.1",secure=True, auth=auth)
conn_str = args[0].strip()
if not re.match("^\w+://.+", conn_str):
conn_str = 'sqlite:///' + conn_str
engine, metadata, Session = models.setup_database(conn_str, echo=((options.verbose-options.quiet)>0), create_all=True)
session = None
try:
session = Session()
#conn.row_factory = sqlite3.Row
#curs = conn.cursor()
#curs.execute("create table if not exists tweet_tweet (json);")
#conn.commit()
results = None
page = 1
print options.query
while page <= int(1500/int(options.rpp)) and ( results is None or len(results) > 0):
results = t.search.tweets(q=options.query, rpp=options.rpp, page=page, include_entities=True)
for tweet in results["results"]:
print tweet
tweet_str = anyjson.serialize(tweet)
#invalidate user id
processor = processor.TwitterProcessorStatus(json_dict=tweet, json_txt=tweet_str, source_id=None, session=session, consumer_token=None, access_token=None, token_filename=options.token_filename, user_query_twitter=False, logger=None)
processor.process()
session.flush()
session.commit()
page += 1
#session.commit()
finally:
if session:
session.close()