import twython
from sqlite3 import *
import datetime, time
import email.utils
from optparse import OptionParser
import os.path
import os
import sys
import simplejson
#options filename rpp page total_pages start_date end_date
def adapt_datetime(ts):
return time.mktime(ts.timetuple())
def adapt_geo(geo):
return simplejson.dumps(geo)
def convert_geo(s):
return simplejson.loads(s)
register_adapter(datetime.datetime, adapt_datetime)
register_converter("geo", convert_geo)
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_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']
def processDate(entry):
ts = email.utils.parsedate(entry["created_at"])
entry["created_at_ts"] = datetime.datetime.fromtimestamp(time.mktime(ts))
def processPage(page, cursor, debug):
for entry in page:
if debug:
print "ENTRY : " + repr(entry)
curs.execute("select id from tweet_tweet where id = ?", (entry["id"],))
res = curs.fetchone()
if res:
continue
entry_user = entry["user"]
processDate(entry_user)
cursor.execute("insert into tweet_user ("+",".join(entry_user.keys())+") values (:"+",:".join(entry_user.keys())+");", entry_user);
new_id = cursor.lastrowid
processDate(entry)
entry["user"] = new_id
if entry["geo"]:
entry["geo"] = adapt_geo(entry["geo"])
new_id = cursor.execute("insert into tweet_tweet ("+",".join(entry.keys())+") values (:"+",:".join(entry.keys())+");", entry);
if __name__ == "__main__" :
parser = OptionParser()
parser.add_option("-f", "--file", dest="filename",
help="write tweet to FILE", metavar="FILE", default="enmi2010_twitter_rest.db")
parser.add_option("-r", "--rpp", dest="rpp",
help="Results per page", metavar="RESULT_PER_PAGE", default=200, type='int')
parser.add_option("-p", "--page", dest="page",
help="page result", metavar="PAGE", default=1, type='int')
parser.add_option("-t", "--total-page", dest="total_page",
help="Total page number", metavar="TOTAL_PAGE", default=16, type='int')
parser.add_option("-s", "--screenname", dest="screen_name",
help="Twitter screen name", metavar="SCREEN_NAME")
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("-n", "--new", dest="new", action="store_true",
help="new database", default=False)
parser.add_option("-d", "--debug", dest="debug", action="store_true",
help="debug", default=False)
(options, args) = parser.parse_args()
if options.debug:
print "OPTIONS : "
print repr(options)
if options.screen_name is None:
print "No Screen name. Exiting"
sys.exit()
if options.new and os.path.exists(options.filename):
os.remove(options.filename)
conn = connect(options.filename)
conn.row_factory = Row
curs = conn.cursor()
curs.execute("create table if not exists tweet_user ("+ ",".join(columns_user) +", created_at_ts integer);")
curs.execute("create table if not exists tweet_tweet ("+ ",".join(columns_tweet) +", created_at_ts integer);")
curs.execute("create index if not exists id_index on tweet_tweet (id asc);");
curs.execute("select count(*) from tweet_tweet;")
res = curs.fetchone()
old_total = res[0]
twitter = twython.setup(username=options.username, password=options.password, headers="IRI enmi (python urllib)")
twitter = twython.Twython(twitter_token = "54ThDZhpEjokcMgHJOMnQA", twitter_secret = "wUoL9UL2T87tfc97R0Dff2EaqRzpJ5XGdmaN2XK3udA")
search_results = None
page = options.page-1
while (page < options.total_page and ( search_results is None or len(search_results) > 0)):
page += 1
try:
search_results = twitter.getUserTimeline(screen_name=options.screen_name, count=options.rpp, page=page)
except twython.TwythonError, (e):
print "NAME : "+ options.screen_name + " ERROR : " + repr(e.msg)
break
print "NAME : "+ options.screen_name +" PAGE : " + repr(page) + " tweet: " + repr(len(search_results)) + " (total page : " + unicode(options.total_page) + " : rpp : "+unicode(options.rpp)+")"
processPage(search_results, curs, options.debug)
conn.commit()
curs.execute("select count(*) from tweet_tweet;")
res = curs.fetchone()
total = res[0]
print "Tweet for " + options.screen_name + " : " + unicode(total - old_total) +", Tweet total : " + repr(total)
conn.close()