update

output/out_put_user_post_each_strategy.py

@@ -197,7 +197,7 @@ select t2.device_id from
 WHERE spam_pv.device_id IS NULL
   and dev.device_id is null
 
-""".format(today_str=today_str, last_30_day_str=last_30_day_str)
+""".format(today_str=today_str, last_30_day_str=today_str)
 
 print(huidu_device_id_sql)
 huidu_device_id_df = spark.sql(huidu_device_id_sql)
@@ -331,7 +331,10 @@ for res in sql_res:
                     if transaction_type not in second_demands_id_count:
                         second_demands_id_count[transaction_type] = {}
                     if tag_id in second_demands_id_count[transaction_type]:
-                        second_demands_id_count[transaction_type][tag_id][int(card_id)] = 1
+                        try:
+                            second_demands_id_count[transaction_type][tag_id][int(card_id)] += 1
+                        except:
+                            second_demands_id_count[transaction_type][tag_id][int(card_id)] = 1
                     else:
                         second_demands_id_count[transaction_type][tag_id] = {}
                         second_demands_id_count[transaction_type][tag_id][int(card_id)] = 1
@@ -360,7 +363,10 @@ for res in sql_res:
                     if transaction_type not in projects_demands_id_count:
                         projects_demands_id_count[transaction_type] = {}
                     if tag_id in projects_demands_id_count[transaction_type]:
-                        projects_demands_id_count[transaction_type][tag_id][int(card_id)] = 1
+                        try:
+                            projects_demands_id_count[transaction_type][tag_id][int(card_id)] += 1
+                        except:
+                            projects_demands_id_count[transaction_type][tag_id][int(card_id)] = 1
                     else:
                         projects_demands_id_count[transaction_type][tag_id] = {}
                         projects_demands_id_count[transaction_type][tag_id][int(card_id)] = 1
@@ -399,8 +405,13 @@ for tag_id in second_demands_tag_count:
     for transaction_type in second_demands_id_count:
         if second_demands_id_count[transaction_type].get(tag_id):
             temp_dict[transaction_type] = len(second_demands_id_count[transaction_type][tag_id])
+            try:
+                temp_dict[transaction_type+"_pv"] = sum(second_demands_id_count[transaction_type][tag_id].values())
+            except:
+                temp_dict[transaction_type + "_pv"] = 0
         else:
             temp_dict[transaction_type] = 0
+            temp_dict[transaction_type+"_pv"] = 0
     second_demands_csv_list.append(temp_dict)
     print(temp_dict)
 second_demands_data = pandas.DataFrame(second_demands_csv_list)
@@ -433,8 +444,13 @@ for tag_id in projects_demands_tag_count:
     for transaction_type in projects_demands_id_count:
         if projects_demands_id_count[transaction_type].get(tag_id):
             temp_dict[transaction_type] = len(projects_demands_id_count[transaction_type][tag_id])
+            try:
+                temp_dict[transaction_type + "_pv"] = sum(projects_demands_id_count[transaction_type][tag_id].values())
+            except:
+                temp_dict[transaction_type + "_pv"] = 0
         else:
             temp_dict[transaction_type] = 0
+            temp_dict[transaction_type + "_pv"] = 0
     projects_csv_list.append(temp_dict)
     print(temp_dict)
 projects_data = pandas.DataFrame(projects_csv_list)