Merge branch 'similar_sort' into 'master'

modify linucb

See merge request alpha/physical!143

linucb/views/collect_data.py

@@ -83,57 +83,65 @@ class CollectData(object):
 
             kafka_consumer_obj = KafkaManager.get_kafka_consumer_ins(topic_name)
             for ori_msg in kafka_consumer_obj:
-                logging.info(ori_msg)
+                try:
+                    logging.info(ori_msg)
 
-                raw_val_dict = json.loads(ori_msg.value)
+                    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"]
-                    device_id = raw_val_dict["device"]["device_id"]
+                    if "type" in raw_val_dict and "on_click_feed_topic_card" == raw_val_dict["type"]:
+                        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(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 = 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(tag_list)))
-                    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 = json.loads(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)))
+                        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=exposure_topic_id)
+                        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_click = 1
                         is_vote = 0
 
                         reward = 1 if is_click or is_vote else 0
 
-                        logging.info("negative tag_list,device_id:%s,topic_id:%s,tag_list:%s" % (str(device_id),str(exposure_topic_id),str(tag_list)))
+                        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:
-                            self.update_user_linucb_tag_info(reward,device_id,tag_id,user_feature)
+                            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)
-                else:
-                    logging.warning("unknown type msg:%s" % raw_val_dict.get("type","missing type"))
+                    elif "type" in raw_val_dict and "page_precise_exposure" == raw_val_dict["type"]:
+                        exposure_cards_list = json.loads(raw_val_dict["params"]["exposure_cards"])
+                        device_id = raw_val_dict["device"]["device_id"]
+                        for item in exposure_cards_list:
+                            if "card_id" not in item:
+                                continue
+                            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=exposure_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
+
+                            logging.info("negative tag_list,device_id:%s,topic_id:%s,tag_list:%s" % (
+                            str(device_id), str(exposure_topic_id), str(tag_list)))
+                            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)
+                    else:
+                        logging.warning("unknown type msg:%s" % raw_val_dict.get("type", "missing type"))
+                except:
+                    logging.error("catch exception,err_msg:%s" % traceback.format_exc())
 
             return True
         except: