update

task/tractate_analysis_in_7000.py

@@ -78,7 +78,7 @@ 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)
-print(card_id_list,tags_v3_count_dict)
+print(card_id_list)
 t = 1
 day_num = 0 - t
 now = (datetime.datetime.now() + datetime.timedelta(days=day_num))
@@ -181,54 +181,45 @@ total_tag_count = {}
 total_tag_count_pro = {}
 temp_null_count = 0
 
-# bulk_dict = {
-#     0: [0, 0, 0],
-#     10: [0, 0, 0],
-#     50: [0, 0, 0],
-#     100: [0, 0, 0],
-#     200: [0, 0, 0],
-#     500: [0, 0, 0],
-#     1000: [0, 0, 0],
-# }
-# print(device_id_dict)
-for redis_count, device_id in enumerate(device_id_dict):
-
-    # if redis_count >= 50:break
-    second_demands = []
-    projects = []
-    total_answer_content_num = 0
-    total_tractate_content_num = 0
-    total_diary_content_num = 0
-    # print(sql_res)
-    try:
-        res = get_user_portrait_tag3_from_redis(device_id)
-    except:
-        continue
-    if res.get("second_demands"):
-        second_demands = res.get("second_demands")
-        # print(count_res)
-        for tag in second_demands:
-            if tag in second_demands_tag_count:
-                second_demands_tag_count[tag] += 1
-            else:
-                second_demands_tag_count[tag] = 1
-            if tag in second_demands_count_dict:
-                total_tractate_content_num += second_demands_count_dict[tag]
-        second_demands_tag_dict[device_id] = second_demands
-    if res.get("projects"):
-        projects = res.get("projects")
-        # print(count_res)
-        for tag in projects:
-            if tag in projects_demands_tag_count:
-                projects_demands_tag_count[tag] += 1
-            else:
-                projects_demands_tag_count[tag] = 1
-
-            if tag in tags_v3_count_dict:
-                total_tractate_content_num += tags_v3_count_dict[tag]
-        projects_demands_tag_dict[device_id] = projects
-    # print(total_answer_content_num, total_tractate_content_num, total_diary_content_num)
-    tmp_count_num = 0
+
+# for redis_count, device_id in enumerate(device_id_dict):
+#
+#     # if redis_count >= 50:break
+#     second_demands = []
+#     projects = []
+#     total_answer_content_num = 0
+#     total_tractate_content_num = 0
+#     total_diary_content_num = 0
+#     # print(sql_res)
+#     try:
+#         res = get_user_portrait_tag3_from_redis(device_id)
+#     except:
+#         continue
+#     if res.get("second_demands"):
+#         second_demands = res.get("second_demands")
+#         # print(count_res)
+#         for tag in second_demands:
+#             if tag in second_demands_tag_count:
+#                 second_demands_tag_count[tag] += 1
+#             else:
+#                 second_demands_tag_count[tag] = 1
+#             if tag in second_demands_count_dict:
+#                 total_tractate_content_num += second_demands_count_dict[tag]
+#         second_demands_tag_dict[device_id] = second_demands
+#     if res.get("projects"):
+#         projects = res.get("projects")
+#         # print(count_res)
+#         for tag in projects:
+#             if tag in projects_demands_tag_count:
+#                 projects_demands_tag_count[tag] += 1
+#             else:
+#                 projects_demands_tag_count[tag] = 1
+#
+#             if tag in tags_v3_count_dict:
+#                 total_tractate_content_num += tags_v3_count_dict[tag]
+#         projects_demands_tag_dict[device_id] = projects
+#     # print(total_answer_content_num, total_tractate_content_num, total_diary_content_num)
+#     tmp_count_num = 0
 
 # 7000片内容的曝光
 
@@ -284,12 +275,14 @@ for res in sql_res:
         if card_id not in card_id_list:continue
 
         session_pv = res.session_pv
-        card_id_set.update(card_id)
+        # card_id_set.update(card_id)
         if cl_id in second_demands_tag_dict:
             for tag_id in second_demands_tag_dict[cl_id]:
                 if tag_id in second_demands_id_count:
+                    print('+1')
                     second_demands_id_count[tag_id][card_id] = 1
                 else:
+                    print('0')
                     second_demands_id_count[tag_id] = {}
                     second_demands_id_count[tag_id][card_id] = 1
 
@@ -301,8 +294,10 @@ for res in sql_res:
         if cl_id in projects_demands_tag_dict:
             for tag_id in projects_demands_tag_dict[cl_id]:
                 if tag_id in projects_demands_id_count:
+                    print('+1')
                     projects_demands_id_count[tag_id][card_id] = 1
                 else:
+                    print(0)
                     projects_demands_id_count[tag_id] = {}
                     projects_demands_id_count[tag_id][card_id] = 1
 
@@ -311,8 +306,8 @@ second_demands_list = []
 for tag_id in second_demands_tag_count:
     temp_dict = {
         "tag_name": tag_id,
-        "device_count": second_demands_tag_count[tag_id],
-        "tractate_count": second_demands_count_dict.get(tag_id),
+        # "device_count": second_demands_tag_count[tag_id],
+        # "tractate_count": second_demands_count_dict.get(tag_id),
 
         "exporsure_count": len(second_demands_id_count[tag_id]) if second_demands_id_count.get(tag_id) else 0,
         # "session_pv_count": len(set(second_demands_id_count[tag_id]))
@@ -323,8 +318,8 @@ print("----------------------------------------------")
 for tag_id in projects_demands_tag_count:
     temp_dict = {
         "tag_name": tag_id,
-        "device_count": projects_demands_tag_count[tag_id],
-        "tractate_count": tags_v3_count_dict.get(tag_id),
+        # "device_count": projects_demands_tag_count[tag_id],
+        # "tractate_count": tags_v3_count_dict.get(tag_id),
         "exporsure_count": len(projects_demands_id_count[tag_id]) if projects_demands_id_count.get(tag_id) else 0,
         # "session_pv_count": len(set(second_demands_id_count[tag_id]))
     }