2025-06-10-视频内容池冷启动品类实验

tasks/data_tasks/fwh_data_recycle.py

@@ -19,7 +19,7 @@ class FwhDataRecycle:
 
     PUBLISH_SUCCESS_STATUS = 2
 
-    STAT_PERIOD = 3 * 24 * 3600
+    STAT_PERIOD = 2 * 24 * 3600
 
     def __init__(self):
         self.denet_client = DatabaseConnector(denet_config)
@@ -140,6 +140,7 @@ class SaveFwhDataToDatabase(FwhDataRecycle):
         """
         fetch_response = self.piaoquan_client.fetch(fetch_query, cursor_type=DictCursor)
         gh_id_list = [i['gzh_id'] for i in fetch_response]
+        gh_id_list = ['gh_5e543853d8f0']
         return gh_id_list
 
     def get_stat_published_articles(self, gh_id):
@@ -205,12 +206,12 @@ class FwhDataExportTemp(FwhDataRecycle):
 
     def get_publish_articles(self):
         sql = f"""
-            select accountName, title, article_group, ItemIndex, show_view_count, from_unixtime(createTime, '%Y-%m-%d'), root_source_id_list
+            -- select accountName, title, article_group, ItemIndex, show_view_count, from_unixtime(createTime, '%Y-%m-%d'), root_source_id_list
+            select accountName, title, ItemIndex, from_unixtime(createTime, '%Y-%m-%d'), sum(show_view_count), root_source_id_list
             from official_articles_v2
-            where ghId in (
-                select gzh_id from article_gzh_developer
-                )
-            and from_unixtime(publish_timestamp) between '2025-06-08' and '2025-06-09';
+            where accountName = '票圈精彩'
+            and from_unixtime(publish_timestamp) between '2025-06-07' and '2025-06-10'
+            group by accountName, title, ItemIndex;
         """
         return self.piaoquan_client.fetch(query=sql, cursor_type=DictCursor)
 
@@ -220,7 +221,7 @@ class FwhDataExportTemp(FwhDataRecycle):
         """
         root_source_id_tuple = tuple(json.loads(root_source_id_list))
         query = f"""
-            select sum(uv) as 'uv', sum(first_uv) as 'first_uv', sum(split_uv) as 'split_uv'
+            select sum(uv) as 'uv', sum(first_uv) as 'first_uv', sum(split_uv) as 'split_uv', sum(split0) as 'T+0_fission'
             from changwen_data_rootsourceid 
             where root_source_id
             in %s;
@@ -237,10 +238,11 @@ class FwhDataExportTemp(FwhDataRecycle):
             article['uv'] = fission_info[0]['uv']
             article['first_uv'] = fission_info[0]['first_uv']
             article['split_uv'] = fission_info[0]['split_uv']
+            article['T+0_fission'] = fission_info[0]['T+0_fission']
             L.append(article)
         df = pd.DataFrame(L)
         df.to_csv('temp.csv', index=False)
 
 
 if __name__ == '__main__':
-    FwhDataExportTemp().deal()
+    SaveFwhDataToDatabase().deal()