update

task/tractate_analysis_in_7000.py

@@ -277,7 +277,7 @@ baoguang_dict = {}
 # 找出card_id 对应帖子的标签 并分类汇总 得到 标签-计数字段
 for res in sql_res:
     # partition_date = res.partition_date
-    print(res)
+    # print(res)
     cl_id = res.cl_id
     if cl_id in device_id_dict:
         card_id = res.card_id
@@ -286,10 +286,10 @@ for res in sql_res:
         if card_id in second_demands_tag_dict:
             for tag_id in second_demands_tag_dict[card_id]:
                 if tag_id in second_demands_id_count:
-                    second_demands_id_count[tag_id].append(card_id)
+                    second_demands_id_count[tag_id].update(card_id)
                 else:
-                    second_demands_id_count[tag_id] = []
-                    second_demands_id_count[tag_id].append(card_id)
+                    second_demands_id_count[tag_id] = set()
+                    second_demands_id_count[tag_id].update(card_id)
 
                 if tag_id in baoguang_dict:
                     baoguang_dict[tag_id] += session_pv
@@ -299,10 +299,10 @@ for res in sql_res:
         if card_id in projects_demands_tag_dict:
             for tag_id in projects_demands_tag_dict[card_id]:
                 if tag_id in projects_demands_id_count:
-                    projects_demands_id_count[tag_id].append(card_id)
+                    projects_demands_id_count[tag_id].update(card_id)
                 else:
-                    projects_demands_id_count[tag_id] = []
-                    projects_demands_id_count[tag_id].append(card_id)
+                    projects_demands_id_count[tag_id] = set()
+                    projects_demands_id_count[tag_id].update(card_id)
 
 final_projects_list = []
 second_demands_list = []
@@ -311,7 +311,7 @@ for tag_id in second_demands_id_count:
         "tag_name": tag_id,
         "device_count": second_demands_tag_count[tag_id],
         "tractate_count": second_demands_count_dict.get(tag_id),
-        "exporsure_count": len(set(second_demands_id_count[tag_id])),
+        "exporsure_count": len(second_demands_id_count[tag_id]),
         # "session_pv_count": len(set(second_demands_id_count[tag_id]))
     }
     print(temp_dict)