Merge branch '2024-11-11-add-negative-category-for-accounts' of Server/title_with_video into 2024-09-23newDbTasks

server/api/response.py

@@ -270,6 +270,12 @@ class Response(object):
                     "traceId": self.trace_id,
                     "miniprogramList": card_list
                 }
+            case 96:
+                return {
+                    "traceId": self.trace_id,
+                    "code": 96,
+                    "error": "该文章品类不符合这个账号，不做冷启动处理"
+                }
             case 97:
                 return {
                     "traceId": self.trace_id,

tasks/history_task.py

@@ -17,6 +17,7 @@ class historyContentIdTask(object):
     """
     TASK_PROCESSING_STATUS = 101
     EXIT_STATUS = 97
+    MISMATCH_STATUS = 96
     TASK_INIT_STATUS = 0
     TASK_PUBLISHED_STATUS = 4
 
@@ -31,6 +32,7 @@ class historyContentIdTask(object):
         self.article_crawler_video_table = self.config.article_crawler_video_table
         self.gh_id_dict = json.loads(self.config.get_config_value("testAccountLevel2"))
         self.history_coroutines = self.config.get_config_value("historyArticleCoroutines")
+        self.account_negative_category = json.loads(self.config.get_config_value("account_negative_category"))
 
     async def get_tasks(self):
         """
@@ -257,6 +259,27 @@ class historyContentIdTask(object):
         else:
             return False
 
+    async def check_title_category(self, content_id, gh_id) -> bool:
+        """
+        判断该文章的品类是否属于该账号的品类
+        :param content_id:
+        :param gh_id:
+        :return:
+        """
+        bad_category_set = set(self.account_negative_category.get(gh_id, []))
+        if bad_category_set:
+            sql = f"""
+                SELECT category
+                FROM article_category
+                WHERE produce_content_id = '{content_id}';
+            """
+            result = await self.mysql_client.async_select(sql)
+            if result:
+                category = result[0][0]
+                if category in bad_category_set:
+                    return True
+        return False
+
     async def process_task(self, params):
         """
         异步执行
@@ -266,7 +289,21 @@ class historyContentIdTask(object):
         content_id = params['content_id']
         trace_id = params['trace_id']
         flow_pool_level = params['flow_pool_level']
+        gh_id = params['gh_id']
         if flow_pool_level == "autoArticlePoolLevel4":
+            # 校验文章是否属于该账号的negative 类型
+            negative_category_status = await self.check_title_category(content_id=content_id, gh_id=gh_id)
+            if negative_category_status:
+                # 修改状态为品类不匹配状态
+                affected_rows = await self.update_content_status(
+                    trace_id=trace_id,
+                    new_content_status=self.MISMATCH_STATUS,
+                    ori_content_status=self.TASK_INIT_STATUS
+                )
+                if affected_rows == 0:
+                    print("修改行数为 0，多个进程抢占同一个 task， 抢占失败，进程退出")
+                    return
+            # 校验文章是否晋升 or 退场
             exit_status = await self.check_title_whether_exit(content_id)
             if exit_status:
                 # 修改状态为退出状态