def test(): db = connect_db() data = db['topics'].find_one() data = data['words'] data = sorted(data, key = lambda x: x['contribution'], reverse = True) make_demo(data) return
def __init__(self, s_time, e_time, source_type, num=-1): ''' s_time: start time e_time: end time source_type: list of sources ''' db = connect_db() self.cursors = [ db[t].find({'timestamp': {'$gte': s_time, '$lte': e_time}}) for t in source_type] self.limit = num return
def __init__(self, s_time, e_time, source_type, keywords, num=-1): ''' s_time: start time e_time: end time source_type: list of sources ''' db = connect_db() if keywords == []: self.cursors = [ db[t].find({'timestamp': {'$gte': s_time, '$lte': e_time}}, timeout=True) for t in source_type] else: self.cursors = [ db[t].find({'timestamp': {'$gte': s_time, '$lte': e_time}, 'tokens': { '$all': keywords}}, timeout=True) for t in source_type] self.limit = num return
def __init__(self, s_time, e_time, source_type, num=-1): ''' s_time: start time e_time: end time source_type: list of sources ''' db = connect_db() self.cursors = [ db[t].find({'timestamp': { '$gte': s_time, '$lte': e_time }}) for t in source_type ] self.limit = num return
def save_model(model): ''' Given a model, save it to the database ''' db = connect_db() for t in model.show_topics(model.num_topics, num_words=100, formatted=False): topic = {} topic['words'] = [] topic['period'] ={ 'start_time':START_TIME, 'end_time': END_TIME } topic['keys'] = KEYS topic['sources'] = SOURCES for i in t: topic['words'].append({'contribution': i[0], 'token': i[1]}) db['topics'].insert(topic) logging.info('have saved the topics into the database') return
def __init__(self): self.new_operation = list() timer = 1 while True: try: self.conn = connect_db() break except Exception: time.sleep(5) # TODO: may need to restart mysqld if timer >= 10: break timer += 1 # initialize paths # LOG.info('Initializing paths ...') for item in NEW_PATHS: if not os.path.exists(item): os.makedirs(item) # get driver # LOG.info('Initializing drivers ...') self.driver = get_webdriver()