diff -r 2bf0fd3432bf -r 8124cde38141 script/lib/iri_tweet/export_twitter_alchemy.py --- a/script/lib/iri_tweet/export_twitter_alchemy.py Tue Jun 28 22:23:56 2011 +0200 +++ b/script/lib/iri_tweet/export_twitter_alchemy.py Thu Jun 30 02:10:21 2011 +0200 @@ -85,6 +85,8 @@ help="list of file to process", metavar="LIST_CONF", default=None) parser.add_option("-E", "--extended", dest="extended_mode", action="store_true", help="Trigger polemic extended mode", metavar="EXTENDED", default=False) + parser.add_option("--user-whitelist", dest="user_whitelist", action="store", + help="A list of user screen name", metavar="USER_WHITELIST",default=None) set_logging_options(parser) @@ -123,7 +125,9 @@ tei = tweet_exclude_table.insert() for line in f: conn.execute(tei.values(id=long(line.strip()))) - + user_whitelist_file = options.user_whitelist + user_whitelist = None + if options.listconf: parameters = [] @@ -162,7 +166,10 @@ duration = params.get("duration", None) content_file = params.get("content_file", None) hashtags = params.get('hashtags', []) - + + if user_whitelist_file: + with open(user_whitelist_file, 'r+') as f: + user_whitelist = list(set([s.strip() for s in f])) start_date = parse_date(start_date_str) ts = time.mktime(start_date.timetuple()) @@ -174,7 +181,7 @@ te = ts + duration end_date = start_date + datetime.timedelta(seconds=duration) - query = get_filter_query(session, start_date, end_date, hashtags, tweet_exclude_table) + query = get_filter_query(session, start_date, end_date, hashtags, tweet_exclude_table, user_whitelist) query_res = query.all()