Merge branch 'master' into hera

linucb/views/collect_data.py

@@ -163,46 +163,49 @@ class CollectData(object):
 
                             raw_val_dict = json.loads(ori_msg.value)
 
-                            if "type" in raw_val_dict and "on_click_feed_topic_card" == raw_val_dict["type"]:
-                                topic_id = raw_val_dict["params"]["business_id"] or raw_val_dict["params"]["topic_id"]
-                                device_id = raw_val_dict["device"]["device_id"]
-                                user_id = raw_val_dict["user_id"] if "user_id" in raw_val_dict else None
+                            if "type" in raw_val_dict and \
+                                    (raw_val_dict["type"] in ("on_click_feed_topic_card","tag_zone_click_focus")):
 
-                                logging.info("consume topic_id:%s,device_id:%s" % (str(topic_id), str(device_id)))
-
-                                tag_list = list()
                                 click_topic_tag_list = list()
-                                collection_tag_sql_query_results = TopicTag.objects.using(settings.SLAVE_DB_NAME).filter(topic_id=topic_id).values_list("tag_id","is_online","is_collection")
-                                # if len(collection_tag_sql_query_results)>0:
-                                for tag_id,is_online,is_collection in collection_tag_sql_query_results:
-                                    if is_online and is_collection == 1:
-                                        click_topic_tag_list.append(tag_id)
+                                if "on_click_feed_topic_card" == raw_val_dict["type"]:
+                                    topic_id = raw_val_dict["params"]["business_id"] or raw_val_dict["params"]["topic_id"]
+                                    device_id = raw_val_dict["device"]["device_id"]
+                                    user_id = raw_val_dict["user_id"] if "user_id" in raw_val_dict else None
+
+                                    logging.info("consume topic_id:%s,device_id:%s" % (str(topic_id), str(device_id)))
+                                    topic_tag_list = list(TopicTag.objects.using(settings.SLAVE_DB_NAME).filter(topic_id=topic_id,is_online=True).values_list("tag_id",flat=True))
+                                    tag_query_results = Tag.objects.using(settings.SLAVE_DB_NAME).filter(id__in=topic_tag_list,is_online=True,is_deleted=False).values_list("id","collection","is_ai")
+                                    for id,collection,is_ai in tag_query_results:
+                                        if collection and is_ai:
+                                            click_topic_tag_list.append(id)
+
+                                    logging.info("positive tag_list,device_id:%s,topic_id:%s,tag_list:%s" % (
+                                    str(device_id), str(topic_id), str(click_topic_tag_list)))
+                                else:
+                                    tag_name = raw_val_dict["params"]["query"]
+                                    query_type = raw_val_dict["params"]["type"]
+                                    device_id = raw_val_dict["device"]["device_id"]
+                                    user_id = raw_val_dict["user_id"] if "user_id" in raw_val_dict else None
+                                    if query_type=="do":
+                                        tag_list = list(Tag.objects.using(settings.SLAVE_DB_NAME).filter(name=tag_name,is_online=True,is_deleted=False).values_list("id",flat=True))
+                                        click_topic_tag_list.extend(tag_list)
+                                    logging.info("query tag attention,positive tag_list,device_id:%s,query_name:%s,tag_list:%s" % (
+                                    str(device_id), tag_name, str(click_topic_tag_list)))
 
-                                logging.info("click_topic_tag_list:%s"%(str(click_topic_tag_list)))
-                                click_sql_query_results = TopicTag.objects.using(settings.SLAVE_DB_NAME).filter(topic_id=topic_id).values_list("tag_id","is_online")
-                                for tag_id,is_online in click_sql_query_results:
-                                    if is_online:
-                                        tag_list.append(tag_id)
-                                        tag_sql_query_results = Tag.objects.using(settings.SLAVE_DB_NAME).filter(
-                                            id=tag_id).values_list("id", "collection", "is_ai")
-                                        for id, collection, is_ai in tag_sql_query_results:
-                                            if (is_ai == 1) and id not in click_topic_tag_list:
-                                                click_topic_tag_list.append(id)
-                                logging.info("click_topic_tag_list:%s"%(str(click_topic_tag_list)))
 
+                                logging.info("click_topic_tag_list:%s"%(str(click_topic_tag_list)))
 
                                 is_click = 1
                                 is_vote = 0
 
                                 reward = 1 if is_click or is_vote else 0
 
-                                logging.info("positive tag_list,device_id:%s,topic_id:%s,tag_list:%s" % (
-                                str(device_id), str(topic_id), str(click_topic_tag_list)))
                                 for tag_id in click_topic_tag_list:
                                     self.update_user_linucb_tag_info(reward, device_id, tag_id, user_feature)
 
                                 # 更新该用户的推荐tag数据，放在 更新完成user tag行为信息之后
-                                self.update_recommend_tag_list(device_id, user_feature, user_id,click_topic_tag_list=click_topic_tag_list)
+                                if len(click_topic_tag_list)>0:
+                                    self.update_recommend_tag_list(device_id, user_feature, user_id,click_topic_tag_list=click_topic_tag_list)
                             elif "type" in raw_val_dict and "page_precise_exposure" == raw_val_dict["type"]:
                                 if isinstance(raw_val_dict["params"]["exposure_cards"],str):
                                     exposure_cards_list = json.loads(raw_val_dict["params"]["exposure_cards"])

search/utils/topic.py

@@ -10,7 +10,7 @@ from alpha_types.venus import TOPIC_SEARCH_SORT
 from libs.es import ESPerform
 from .common import TopicDocumentField
 from search.utils.common import *
-
+from trans2es.models.pictorial import PictorialTopics
 
 class TopicUtils(object):
 
@@ -876,38 +876,41 @@ class TopicUtils(object):
 
     @classmethod
     def list_topic_ids(cls, filters, nfilters, sorts_by, offset=0, size=10, index_name="topic", filter_online=True):
-
-        must = cls.process_filters(filters, filter_online=filter_online)
-        q = {
-            "query": {
-                "bool": {
-                    "must": must,
-                    "must_not": cls.process_nfilters(nfilters),
+        try:
+            must = cls.process_filters(filters, filter_online=filter_online)
+            q = {
+                "query": {
+                    "bool": {
+                        "must": must,
+                        "must_not": cls.process_nfilters(nfilters),
+                    }
                 }
             }
-        }
 
-        if sorts_by:
-            sorts = cls.process_sort(sorts_by)
-            if sorts:
-                q["sort"] = sorts
+            if sorts_by:
+                sorts = cls.process_sort(sorts_by)
+                if sorts:
+                    q["sort"] = sorts
 
-        try:
             result_dict = ESPerform.get_search_results(
                 ESPerform.get_cli(), sub_index_name=index_name,
                 query_body=q, offset=offset, size=size
             )
 
-            return {
-                "hits": result_dict["hits"],
-                "total_count": result_dict["total_count"]
-            }
+            if len(result_dict["hits"])>0:
+                topic_id_list = [item["_source"]["id"] for item in result_dict["hits"]]
+
+                return (topic_id_list,result_dict["total_count"])
+            elif offset==0 and "pictorial_id" in filters: # 防止帖子同步延迟，画报详情页为空
+                pictorial_id = int(filters["pictorial_id"])
+                topic_id_list = list(PictorialTopics.objects.filter(pictorial_id=pictorial_id,is_online=True,is_deleted=False).values_list("topic_id", flat=True)[offset:size])
+
+                return (topic_id_list,len(topic_id_list))
+            else:
+                return ([], 0)
         except:
             logging.error("catch exception,err_msg:%s" % traceback.format_exc())
-            return {
-                "hits": [],
-                "total_count": 0
-            }
+            return ([],0)
 
     @classmethod
     def business_topic_id(cls, filters, nfilters, sorts_by, offset=0, size=10, index_name="topic", filter_online=True):

search/views/topic.py

@@ -146,8 +146,8 @@ def get_home_recommend_topic_ids(user_id, device_id, tag_id, offset, size, query
             "have_read_topic_list": json.dumps(have_read_topic_id_list),
         }
         redis_client.hmset(redis_key, redis_dict)
-        # 每个session key保存30天
-        redis_client.expire(redis_key, 60 * 60 * 24 * 30)
+        # 每个session key保存60天
+        redis_client.expire(redis_key, 60 * 60 * 24 * 60)
 
         return recommend_topic_list,rank_topic_id_list
     except:
@@ -208,7 +208,7 @@ def home_recommend(device_id="", user_id=-1, offset=0, size=10, query_type=Topic
         return {"linucb_topic_ids": recommend_topic_ids,"rank_topic_ids":rank_topic_ids}
     except:
         logging.error("catch exception,err_msg:%s" % traceback.format_exc())
-        return {"recommend_topic_ids": []}
+        return {"linucb_topic_ids": [],"rank_topic_ids":[]}
 
 
 @bind("physical/search/discover_page")
@@ -398,13 +398,12 @@ def topic_search(filters, nfilters=None, sorts_by=None, offset=0, size=10):
     """帖子搜索。"""
 
     try:
-        result_list = TopicUtils.list_topic_ids(filters=filters, nfilters=nfilters,
+        (topic_id_list,total_count) = TopicUtils.list_topic_ids(filters=filters, nfilters=nfilters,
                                                 sorts_by=sorts_by, offset=offset, size=size)
-        logging.info("get result_list:%s"%result_list)
-        topic_ids = [item["_source"]["id"] for item in result_list["hits"]]
+
         return {
-            "topic_ids": topic_ids,
-            "total_count": result_list["total_count"]
+            "topic_ids": topic_id_list,
+            "total_count": total_count
         }
     except:
         logging.error("catch exception, err_msg:%s" % traceback.format_exc())