Merge branch 'master' of git.wanmeizhensuo.com:alpha/physical

injection/data_sync/tasks.py

@@ -20,6 +20,8 @@ def write_to_es(es_type, pk_list, use_batch_query_set=False):
 
         if es_type == "register_user_tag":
             RegisterUserTag.get_register_user_tag(pk_list)
+        elif es_type == "attention_user_tag":
+            RegisterUserTag.get_user_attention_tag(pk_list)
         else:
             type_info_map = get_type_info_map()
             type_info = type_info_map[es_type]

linucb/utils/register_user_tag.py

@@ -9,7 +9,7 @@ import traceback
 import json
 import pickle
 from django.conf import settings
-from trans2es.models.tag import AccountUserTag
+from trans2es.models.tag import AccountUserTag,CommunityTagFollow
 from libs.es import ESPerform
 import libs.tools as Tools
 from search.utils.common import *
@@ -30,8 +30,40 @@ class RegisterUserTag(object):
     linucb_user_id_register_tag_topic_id_prefix = "physical:linucb:register_tag_topic_recommend:user_id:"
 
     linucb_register_user_tag_key = "physical:linucb:register_user_tag_info"
+
+    @classmethod
+    def get_user_attention_tag(cls, pk_list):
+        """
+        :remark 获取用户关注标签
+        :param pk_list:
+        :return:
+        """
+        try:
+            user_id_dict = dict()
+            query_results = CommunityTagFollow.objects.filter(pk__in=pk_list,is_deleted=False,is_online=True)
+            for item in query_results:
+                tag_id = item.tag_id
+                user_id = item.user_id
+
+                user_tag_list = CommunityTagFollow.objects.filter(user=user_id,is_deleted=False,is_online=True).values_list("tag_id", flat=True)
+                user_id_dict[user_id] = user_tag_list
+
+            for user_id in user_id_dict:
+                redis_user_tag_id_data = redis_client.hget(cls.linucb_register_user_tag_key, user_id)
+                redis_user_tag_id_list = json.loads(redis_user_tag_id_data) if redis_user_tag_id_data else []
+                redis_user_tag_id_list.extend(user_id_dict[user_id])
+                redis_client.hset(cls.linucb_register_user_tag_key, user_id, json.dumps(list(set(redis_user_tag_id_list))))
+        except:
+            logging.error("catch exception,err_msg:%s" % traceback.format_exc())
+
+
     @classmethod
     def get_register_user_tag(cls,pk_list):
+        """
+        :remark 用户注册时选的标签
+        :param pk_list:
+        :return:
+        """
         try:
             # user_id_set = set()
             user_id_dict = dict()
@@ -43,7 +75,10 @@ class RegisterUserTag(object):
                 user_id_dict[user_id] = user_tag_list
 
             for user_id in user_id_dict:
-                redis_client.hset(cls.linucb_register_user_tag_key, user_id, json.dumps(list(user_id_dict[user_id])))
+                redis_user_tag_id_data = redis_client.hget(cls.linucb_register_user_tag_key, user_id)
+                redis_user_tag_id_list = json.loads(redis_user_tag_id_data) if redis_user_tag_id_data else []
+                redis_user_tag_id_list.extend(user_id_dict[user_id])
+                redis_client.hset(cls.linucb_register_user_tag_key, user_id, json.dumps(list(set(redis_user_tag_id_list))))
 
         #         if user_id not in user_id_set:
         #             user_id_set.add(user_id)

linucb/views/collect_data.py

@@ -188,8 +188,8 @@ class CollectData(object):
                                 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(tag_list)))
-                                for tag_id in tag_list:
+                                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
@@ -216,14 +216,17 @@ 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")
-                                for topic_id,tag_id,is_online in exposure_sql_query_results:
-                                    if is_online:
-                                        # 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 collection == 1 or is_ai == 1:
+                                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:
                                         tag_list.append(tag_id)
+                                    if is_online:
+                                        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 tag_list:
+                                                tag_list.append(id)
 
                                     if topic_id not in topic_tag_id_dict:
                                         topic_tag_id_dict[topic_id] = list()

search/views/topic.py

@@ -63,27 +63,31 @@ def get_home_recommend_topic_ids(user_id, device_id, tag_id, offset, size, query
         redis_field_list = [b'have_read_topic_list']
         redis_field_val_list = redis_client.hmget(redis_key, redis_field_list)
 
+        # 获取已读帖子
+        have_read_topic_id_list = list()
+
+        if redis_field_val_list[0]:
+            if query is None:
+                have_read_topic_id_list = list(json.loads(redis_field_val_list[0]))
+            else:
+                if offset>0: # 首次搜索时不需要过滤已读
+                    have_read_topic_id_list = list(json.loads(redis_field_val_list[0]))
+
+        # linucb 推荐帖子
         topic_recommend_redis_key = "physical:linucb:topic_recommend:device_id:" + str(device_id)
 
         recommend_topic_list=list()
         recommend_topic_dict = redis_client.hgetall(topic_recommend_redis_key)
         if b"data" in recommend_topic_dict:
             recommend_topic_id_list = json.loads(recommend_topic_dict[b"data"])
+            # 推荐帖子是强插的，要保证推荐帖子不在已读里
+            recommend_topic_id_list = list(set(recommend_topic_id_list)-set(have_read_topic_id_list))
             cursor = int(str(recommend_topic_dict[b"cursor"], encoding="utf-8"))
             newcursor = cursor + 6
             if len(recommend_topic_id_list) > newcursor:
                 recommend_topic_list = recommend_topic_id_list[cursor:newcursor]
                 redis_client.hset(topic_recommend_redis_key,"cursor",newcursor)
 
-        have_read_topic_id_list = list()
-
-        if redis_field_val_list[0]:
-            if query is None:
-                have_read_topic_id_list = list(json.loads(redis_field_val_list[0]))
-            else:
-                if offset>0:
-                    have_read_topic_id_list = list(json.loads(redis_field_val_list[0]))
-
         user_similar_score_redis_key = "physical:user_similar_score:user_id:" + str(user_id)
         redis_user_similar_score_redis_val = redis_client.get(user_similar_score_redis_key)
         user_similar_score_redis_list = json.loads(
@@ -92,7 +96,6 @@ def get_home_recommend_topic_ids(user_id, device_id, tag_id, offset, size, query
 
         redis_tag_data = redis_client.hget("physical:linucb:register_user_tag_info", user_id)
         attention_tag_list = json.loads(redis_tag_data) if redis_tag_data else []
-        logging.info("attention_tag_list:%s"%(str(attention_tag_list)))
         if len(recommend_topic_list)>0:
             size = size-len(recommend_topic_list)
             have_read_topic_id_list.extend(recommend_topic_list)