luojunhui 3 weeks ago
parent
commit
fc19988b0e

+ 2 - 0
applications/tasks/cold_start_tasks/article_pool_cold_start.py

@@ -1,6 +1,7 @@
 from __future__ import annotations
 
 import time
+import asyncio
 import datetime
 import traceback
 
@@ -305,6 +306,7 @@ class ArticlePoolColdStart(ArticlePoolColdStartStrategy, ArticlePoolFilterStrate
                             plan_id=plan_id,
                             category=category,
                         )
+                        await asyncio.sleep(120)
                     except Exception as e:
                         await feishu_robot.bot(
                             title="文章冷启动异常",

+ 6 - 7
applications/tasks/data_recycle_tasks/recycle_daily_publish_articles.py

@@ -98,7 +98,7 @@ class RecycleDailyPublishArticlesTask(Const):
 
     async def recycle_single_account(self, account):
         """recycle single account"""
-        query = f"""
+        query = """
             select max(publish_timestamp) as publish_timestamp from official_articles_v2 where ghId = %s;
         """
         response = await self.pool.async_fetch(
@@ -189,7 +189,7 @@ class CheckDailyPublishArticlesTask(RecycleDailyPublishArticlesTask):
 
     async def check_account(self, account: dict, date_string: str) -> bool:
         """check account data"""
-        query = f"""
+        query = """
             select accountName, count(1) as publish_count 
             from official_articles_v2 where ghId = %s and from_unixtime(publish_timestamp) > %s;
         """
@@ -280,7 +280,7 @@ class UpdateRootSourceIdAndUpdateTimeTask(Const):
         self.log_client = log_client
 
     async def get_article_list(self):
-        query = f"""select ContentUrl, wx_sn from official_articles_v2 where publish_timestamp in %s;"""
+        query = """select ContentUrl, wx_sn from official_articles_v2 where publish_timestamp in %s;"""
         article_list = await self.pool.async_fetch(
             query=query, db_name="piaoquan_crawler", params=(tuple([0, -1]),)
         )
@@ -306,9 +306,7 @@ class UpdateRootSourceIdAndUpdateTimeTask(Const):
                 mini_program = data.get("mini_program", [])
                 if mini_program:
                     root_source_id_list = [
-                        urllib.parse.parse_qs(urllib.parse.unquote(i["path"]))[
-                            "rootSourceId"
-                        ][0]
+                        urllib.parse.parse_qs(urllib.parse.unquote(i["path"])).get("root_source_id", [""])[0]
                         for i in mini_program
                     ]
                 else:
@@ -333,7 +331,7 @@ class UpdateRootSourceIdAndUpdateTimeTask(Const):
                     "status": "fail",
                 }
             )
-        query = f"""
+        query = """
             update official_articles_v2 set publish_timestamp = %s, root_source_id_list = %s
             where wx_sn = %s;
         """
@@ -347,6 +345,7 @@ class UpdateRootSourceIdAndUpdateTimeTask(Const):
             ),
         )
         if publish_timestamp_s == self.REQUEST_FAIL_STATUS:
+            article['wx_sn'] = wx_sn
             return article
         else:
             return None