Merge branch 'similar_sort' into 'master'

Similar sort

See merge request !136

injection/data_sync/tasks.py

@@ -65,7 +65,7 @@ def sync_face_similar_data_to_redis():
 @shared_task
 def sync_user_similar_score():
     try:
-        results_items = UserSimilarScore.objects.filter(is_deleted=False).distinct().values("user_id").values_list(flat=True)
+        results_items = UserSimilarScore.objects.filter(is_deleted=False).distinct().values("user_id").values_list("user_id",flat=True)
         redis_key_prefix = "physical:user_similar_score:user_id:"
 
         logging.info("duan add,begin sync user_similar_score!")

linucb/views/collect_data.py

@@ -79,6 +79,8 @@ class CollectData(object):
 
     def consume_data_from_kafka(self,topic_name=None):
         try:
+            user_feature = [1,1]
+
             kafka_consumer_obj = KafkaManager.get_kafka_consumer_ins(topic_name)
             for ori_msg in kafka_consumer_obj:
                 logging.info(ori_msg)
@@ -90,7 +92,7 @@ class CollectData(object):
                         topic_id = raw_val_dict["params"]["business_id"]
                         device_id = raw_val_dict["device"]["device_id"]
 
-                        logging.info("consume topic_id:%s,device_id:%s" % (str(topic_id),str()))
+                        logging.info("consume topic_id:%s,device_id:%s" % (str(topic_id),str(device_id)))
 
                         tag_list = list()
                         sql_query_results = TopicTag.objects.filter(is_online=True,topic_id=topic_id)
@@ -100,7 +102,28 @@ class CollectData(object):
                         is_click = 1
                         is_vote = 0
 
-                        user_feature = [is_click, is_vote]
+                        reward = 1 if is_click or is_vote else 0
+
+                        for tag_id in 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)
+                    elif "type" in raw_val_dict and "page_precise_exposure"==raw_val_dict["type"]:
+                        exposure_cards_list = raw_val_dict["params"]["exposure_cards"]
+                        device_id = raw_val_dict["device"]["device_id"]
+                        for item in exposure_cards_list:
+                            exposure_topic_id = item["card_id"]
+                            logging.info("consume exposure topic_id:%s,device_id:%s" % (str(exposure_topic_id),str(device_id)))
+
+                        tag_list = list()
+                        sql_query_results = TopicTag.objects.filter(is_online=True,topic_id=topic_id)
+                        for sql_item in sql_query_results:
+                            tag_list.append(sql_item.tag_id)
+
+                            is_click = 0
+                            is_vote = 0
+
                             reward = 1 if is_click or is_vote else 0
 
                             for tag_id in tag_list:

trans2es/management/commands/trans2es_data2es_parallel.py

@@ -27,6 +27,7 @@ from trans2es.models.group import Group
 from trans2es.models.topic import Topic,ActionSumAboutTopic
 from search.utils.common import *
 from linucb.views.collect_data import CollectData
+from injection.data_sync.tasks import sync_user_similar_score
 
 class Job(object):
     __es = None
@@ -207,5 +208,8 @@ class Command(BaseCommand):
             if len(options["sync_type"]) and options["sync_type"]=="linucb":
                 collect_obj = CollectData()
                 collect_obj.consume_data_from_kafka()
+
+            if len(options["sync_type"]) and options["sync_type"]=="similar":
+                sync_user_similar_score()
         except:
             logging.error("catch exception,err_msg:%s" % traceback.format_exc())

trans2es/models/face_user_contrast_similar.py

@@ -32,7 +32,7 @@ class UserSimilarScore(models.Model):
         verbose_name=u"首页推荐用"
         db_table="user_similar_score"
 
-    id = models.IntegerField(verbose_name=u"主键ID",db_index=True)
+    id = models.IntegerField(verbose_name=u"主键ID",primary_key=True)
     is_deleted = models.BooleanField(verbose_name=u"是否删除")
     user_id = models.IntegerField(verbose_name=u"用户ID")
     contrast_user_id = models.BigIntegerField(verbose_name="参数对比的用户id", db_index=True)