Пример #1
0
def snapshot_twitteraccounts(verbose=False):
    """Create today's SnapshotOfTwitterAccounts"""
    api = open_api()
    friends = api.GetFriends()

    for friend in friends:
        if verbose: print 'Scraping %s...' % friend.screen_name
        screen_name = friend.screen_name.lower()
        if screen_name=='theannotator':
            # legacy reasons
            screen_name = 'TheAnnotator'
        followers = friend.followers_count
        following = friend.friends_count
        tweets = friend.statuses_count
        today = datetime.now().date()
        # How long since we scraped this account?
        latest = Session.query(SnapshotOfTwitterAccount)\
                .filter(SnapshotOfTwitterAccount.screen_name==screen_name)\
                .order_by(SnapshotOfTwitterAccount.timestamp.desc())\
                .first()
        if latest and latest.timestamp>=today:
            if verbose: print ' -> most recent snapshot for %s has already been processed.' % screen_name
            continue
        # Create a snapshot 
        sn = SnapshotOfTwitterAccount(\
                timestamp=today,\
                screen_name=screen_name,\
                followers=followers,\
                following=following,\
                tweets=tweets)
        Session.add(sn)
        if verbose: print '  -> ',sn.toJson()
    Session.commit()
Пример #2
0
def snapshot_twitteraccounts(verbose=False):
    """Create today's SnapshotOfTwitterAccounts"""
    api = tweepy.API()
    for screen_name in TRACKED_ACCOUNTS:
        if verbose: print 'Scraping %s...' % screen_name
        u = api.get_user(screen_name)
        followers = u.followers_count
        following = u.friends_count
        tweets = u.statuses_count
        today = datetime.now().date()
        # How long since we scraped this account?
        latest = Session.query(SnapshotOfTwitterAccount)\
                .filter(SnapshotOfTwitterAccount.screen_name==screen_name)\
                .order_by(SnapshotOfTwitterAccount.timestamp.desc())\
                .first()
        if latest and latest.timestamp>=today:
            if verbose: print ' -> most recent snapshot for %s has already been processed.' % screen_name
            continue
        # Create a snapshot 
        sn = SnapshotOfTwitterAccount(\
                timestamp=today,\
                screen_name=screen_name,\
                followers=followers,\
                following=following,\
                tweets=tweets)
        Session.add(sn)
        if verbose: print '  -> ',sn.toJson()
    Session.commit()