update

task/tractate_analysis_in_7000.py

@@ -77,7 +77,7 @@ spark.sql("CREATE TEMPORARY FUNCTION arrayMerge AS 'com.gmei.hive.common.udf.UDF
 sql = """
 select card_id from strategy_content_exposure_index where card_content_type="user_post" and preciseexposure_num>=50 and ctr>=0.05 and avg_page_stay>=20 and create_day="2020-09-17"
 """
-second_demands_count_dict, tags_v3_count_dict,card_id_list = get_user_post_from_mysql(sql)
+second_demands_count_dict, tags_v3_count_dict, second_demands_card_id_list,tags_v3_card_id_list = get_user_post_from_mysql(sql)
 print(card_id_list)
 t = 2
 day_num = 0 - t
@@ -271,8 +271,7 @@ for res in sql_res:
     # print(res)
     cl_id = res.cl_id
     card_id = res.card_id
-    if int(card_id) not in card_id_list:
-        continue
+
     if cl_id in device_id_dict:
         # print("has device")
         # print(type(card_id),card_id)
@@ -280,6 +279,8 @@ for res in sql_res:
         session_pv = res.session_pv
         # card_id_set.update(card_id)
         if cl_id in second_demands_tag_dict:
+            if int(card_id) not in second_demands_card_id_list:
+                continue
             for tag_id in second_demands_tag_dict[cl_id]:
                 if tag_id in second_demands_id_count:
                     second_demands_id_count[tag_id][int(card_id)] = 1
@@ -293,6 +294,8 @@ for res in sql_res:
                 #     baoguang_dict[tag_id] = session_pv
 
         if cl_id in projects_demands_tag_dict:
+            if int(card_id) not in tags_v3_card_id_list:
+                continue
             for tag_id in projects_demands_tag_dict[cl_id]:
                 if tag_id in projects_demands_id_count:
                     projects_demands_id_count[tag_id][int(card_id)] = 1
@@ -330,7 +333,7 @@ for tag_id in projects_demands_tag_count:
         "tag_name": tag_id,
         "device_count": projects_demands_tag_count[tag_id],
         "tractate_count": tags_v3_count_dict.get(tag_id),
-        "exporsure_count": len(second_demands_id_count[tag_id]) if second_demands_id_count.get(tag_id) else 0,
+        "exporsure_count": len(projects_demands_id_count[tag_id]) if projects_demands_id_count.get(tag_id) else 0,
 
     }
     print(temp_dict)