Пример #1
0
def update_attribute_day():
    bulk_action = []
    count = 0
    user_info_list = {}
    start_ts = time.time()
    while True:
        r_user_info = update_day_redis.rpop(UPDATE_DAY_REDIS_KEY)
        if r_user_info:
            r_user_info = json.loads(r_user_info)
            uid = r_user_info.keys()[0]
            user_info_list[uid] = r_user_info[uid]
            count += 1
        else:
            break

        if count % 1000==0:
            uid_list = user_info_list.keys()
            #get user_list hashtag_results
            hashtag_results = update_day_hashtag(uid_list)
            #get user geo today
            geo_results = update_day_geo(uid_list, user_info_list)
            #get user activeness evaluate
            activeness_results = update_day_activeness(geo_results, user_info_list)
            #get user influence
            influence_results = update_day_influence(uid_list, user_info_list)
            #get user sensitive
            sensitive_results, sensitive_string_results, sensitive_dict_results = get_sensitive(uid_list)
            #update to es by bulk action
            save_bulk_action(uid_list, hashtag_results, geo_results, activeness_results, influence_results, sensitive_results, sensitive_string_results, sensitive_dict_results)
            user_info_list = {}
            end_ts = time.time()
            #log_should_delete
            #print '%s sec count 1000' % (end_ts - start_ts)
            #log_should_delete
            start_ts = end_ts
    
    if user_info_list != {}:
        uid_list = user_info_list.keys()
        print 'uid_list:', uid_list
        #get user_list hashtag_results
        print 'get hashtag result'
        hashtag_results = update_day_hashtag(uid_list)
        #get user geo today
        print 'get geo'
        geo_results = update_day_geo(uid_list, user_info_list)
        #get user activeness evaluate
        print 'get activeness'
        activeness_results = update_day_activeness(geo_results, user_info_list)
        #get user influence
        print 'get influence'
        influence_results = update_day_influence(uid_list, user_info_list)
        #get user sensitive
        print 'get sensitive'
        sensitive_results, sensitive_string_results, sensitive_dict_results = get_sensitive(uid_list)

        #update to es by bulk action
        save_bulk_action(uid_list, hashtag_results, geo_results, activeness_results, influence_results, sensitive_results, sensitive_string_results, sensitive_dict_results)
Пример #2
0
def update_attribute_day():
    bulk_action = []
    count = 0
    user_info_list = {}
    start_ts = time.time()
    while True:
        r_user_info = update_day_redis.rpop(UPDATE_DAY_REDIS_KEY)
        if r_user_info:
            r_user_info = json.loads(r_user_info)
            uid = r_user_info.keys()[0]
            user_info_list[uid] = r_user_info[uid]
            count += 1
        else:
            break

        if count % 100==0:
            uid_list = user_info_list.keys()
            #get user_list hashtag_results
            hashtag_results = update_day_hashtag(uid_list)
            sensitive_hashtag_results = update_day_sensitive_hashtag(uid_list)
            #get user geo today
            geo_results = update_day_geo(uid_list, user_info_list)
            #get user activeness evaluate
            activeness_results = update_day_activeness(geo_results, user_info_list)
            #get user influence
            influence_results = update_day_influence(uid_list, user_info_list)
            #get user sensitive
            sensitive_results = update_day_sensitive(uid_list)
            #print 'sensitive_results:', sensitive_results
            profile_results = update_day_profile(uid_list)
            #update to es by bulk action
            save_bulk_action(uid_list, hashtag_results, sensitive_hashtag_results, geo_results, activeness_results, influence_results, sensitive_results, profile_results)
            user_info_list = {}
            end_ts = time.time()
            print '%s sec count 1000' % (end_ts - start_ts)
            start_ts = end_ts

    if user_info_list != {}:
        uid_list = user_info_list.keys()
        #get user_list hashtag_results
        hashtag_results = update_day_hashtag(uid_list)
        sensitive_hashtag_results = update_day_sensitive_hashtag(uid_list)
        #get user geo today
        geo_results = update_day_geo(uid_list, user_info_list)
        #get user activeness evaluate
        activeness_results = update_day_activeness(geo_results, user_info_list)
        #get user influence
        influence_results = update_day_influence(uid_list, user_info_list)
        #get user sensitive
        sensitive_results = update_day_sensitive(uid_list)
        #print 'sensitive_results:', sensitive_results
        profile_results = update_day_profile(uid_list)
        #update to es by bulk action
        save_bulk_action(uid_list, hashtag_results,sensitive_hashtag_results, geo_results, activeness_results, influence_results, sensitive_results, profile_results)
Пример #3
0
def update_attribute_day():
    bulk_action = []
    count = 0
    user_info_list = {}
    start_ts = time.time()
    while True:
        r_user_info = update_day_redis.rpop(UPDATE_DAY_REDIS_KEY)
        if r_user_info:
            r_user_info = json.loads(r_user_info)
            uid = r_user_info.keys()[0]
            user_info_list[uid] = r_user_info[uid]
            count += 1
        else:
            break

        if count % 1000 == 0:
            uid_list = user_info_list.keys()
            #get user_list hashtag_results
            hashtag_results = update_day_hashtag(uid_list)
            #get user geo today
            geo_results = update_day_geo(uid_list, user_info_list)
            #get user activeness evaluate
            activeness_results = update_day_activeness(geo_results,
                                                       user_info_list)
            #get user influence
            influence_results = update_day_influence(uid_list, user_info_list)
            #get user sensitive
            sensitive_results, sensitive_string_results, sensitive_dict_results = get_sensitive(
                uid_list)
            #update to es by bulk action
            save_bulk_action(uid_list, hashtag_results, geo_results,
                             activeness_results, influence_results,
                             sensitive_results, sensitive_string_results,
                             sensitive_dict_results)
            user_info_list = {}
            end_ts = time.time()
            #log_should_delete
            #print '%s sec count 1000' % (end_ts - start_ts)
            #log_should_delete
            start_ts = end_ts

    if user_info_list != {}:
        uid_list = user_info_list.keys()
        print 'uid_list:', uid_list
        #get user_list hashtag_results
        print 'get hashtag result'
        hashtag_results = update_day_hashtag(uid_list)
        #get user geo today
        print 'get geo'
        geo_results = update_day_geo(uid_list, user_info_list)
        #get user activeness evaluate
        print 'get activeness'
        activeness_results = update_day_activeness(geo_results, user_info_list)
        #get user influence
        print 'get influence'
        influence_results = update_day_influence(uid_list, user_info_list)
        #get user sensitive
        print 'get sensitive'
        sensitive_results, sensitive_string_results, sensitive_dict_results = get_sensitive(
            uid_list)

        #update to es by bulk action
        save_bulk_action(uid_list, hashtag_results, geo_results,
                         activeness_results, influence_results,
                         sensitive_results, sensitive_string_results,
                         sensitive_dict_results)