def old_counts(): results = store.get_db()[STREAM_NAME].map_reduce(MAP_FN, REDUCE_FN, 'xyz') tag_2_count = [(r['_id'], int(r['value'])) for r in results.find()] dico = dict(tag_2_count) v = val_or_0 new_dico = {} for t in HASHTAGS: new_dico['#' + t] = v(dico, t.lower()) return new_dico
def get_imgs_from_tweets(stream_name, hashtag=None): store.drop_coll(stream_name) gather.only_new_tweets(stream_name) query = {'entities.media.media_url': {'$exists': True}} if hashtag is not None: hashtag_re = hashtags.make_re(hashtag) query['entities.hashtags.text'] = hashtag_re coll = store.get_db()[stream_name] pics = [get_pic(t) for t in coll.find(query).sort('id_str', pymongo.DESCENDING)] return [p for p in pics if not None]
def get_coll(): return store.get_db()[MSG_COLL]