weckbot/getter.py

#!/usr/bin/env python
# -*- coding: utf-8 -*-

import HTMLParser, tweepy, os, sqlite3, datetime, dateutil.parser, time

user = "Twitter Handle (without @)"
cke = "Consumer Key"
cse = "Consumer Secret"
ato = "Access Token"
ase = "Access Secret"

database_filename = "Database.db"

search = "@" + user

auth = tweepy.OAuthHandler(cke, cse)
auth.set_access_token(ato, ase)
api = tweepy.API(auth)

sql_conn = sqlite3.connect(database_filename)
cur = sql_conn.cursor()

try:
 cur.execute("SELECT max(tweet_id) FROM tweets")
 savepoint = int(cur.fetchone()[0])
except:
 savepoint = ""
 print "No savepoint found. Trying to get as many results as possible."

while True:
 timeline = tweepy.Cursor(api.search, q=search, since_id=savepoint).items()

 for status in timeline:
  if status.text.find("@" + user) > 1:
   continue
  text = status.text.encode("UTF-8")[status.text.find(" ")+1:].strip()
  sender = status.user.screen_name.encode("UTF-8")
  twid = int(status.id)
  try:
   if "-" in text:
    loc = text.find("-")
    date = dateutil.parser.parse(text[:loc-1],dayfirst=True,fuzzy=True)
    comment = HTMLParser.HTMLParser().unescape(text[loc+1:]).strip()
   else:
    date = dateutil.parser.parse(text,dayfirst=True,fuzzy=True) 
    comment = ""
  except ValueError:
   api.update_status("@%s Sorry, ich verstehe deinen Tweet nicht... :(" % sender, twid)
   continue

  cur.execute("INSERT INTO tweets VALUES(%i,'%s','%s','%s',0)" % (twid,date.strftime("%Y-%m-%dT%H:%M:%S"),sender,comment))
  sql_conn.commit()

  savepoint = twid

 time.sleep(60)