revise

libs/es.py

@@ -417,8 +417,12 @@ class ESPerform(object):
                     {"_score": {"order": "desc"}},
                     {"create_time_val": {"order": "desc"}},
                     {"language_type": {"order": "asc"}},
-                ]
+                ],
+                "collapse": {
+                    "field": "user_id"
+              }
             }
+
             if len(have_read_topic_id_list) > 0:
                 q["query"]["function_score"]["query"]["bool"]["must_not"] = {
                     "terms": {

linucb/views/collect_data.py

@@ -219,7 +219,9 @@ class CollectData(object):
 
                                 topic_tag_id_dict = dict()
                                 tag_list = list()
-                                exposure_sql_query_results = TopicTag.objects.using(settings.SLAVE_DB_NAME).filter(topic_id__in=exposure_topic_id_list).values_list("topic_id","tag_id","is_online","is_collection")
+                                exposure_sql_query_results = TopicTag.objects.using(settings.SLAVE_DB_NAME).\
+                                    filter(topic_id__in=exposure_topic_id_list).\
+                                    values_list("topic_id","tag_id","is_online","is_collection")
                                 # if len(exposure_sql_query_results)>0:
                                 for topic_id,tag_id,is_online,is_collection in exposure_sql_query_results:
                                     if is_online and is_collection == 1: