orbitkit 0.8.43__tar.gz → 0.8.45__tar.gz

{orbitkit-0.8.43/orbitkit.egg-info → orbitkit-0.8.45}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: orbitkit
-Version: 0.8.43
+Version: 0.8.45
 Summary: This project is only for Orbit Tech internal use.
 Home-page: https://github.com/clown-0726/orbitkit
 Author: Lilu Cao

orbitkit-0.8.45/orbitkit/VERSION

@@ -0,0 +1 @@
+0.8.45

{orbitkit-0.8.43 → orbitkit-0.8.45}/orbitkit/airflow_handler/file_flow_exit_process.py

@@ -28,8 +28,6 @@ class FlowUpdater:
         return getattr(self, collection_name)
 
     def _handle_convert(self, status, attachments, db_store_path_set, attachments_pdf):
-        if not attachments or not attachments_pdf:
-            raise ValueError("Missing attachments: neither 'attachments' nor 'attachments_pdf' was provided.")
         if not status:
             return {
                 'x_status_list.status_convert.status': 'convert_failed',
@@ -37,11 +35,16 @@ class FlowUpdater:
                 'x_status_list.status_convert.status_meta': 'meta_init'
             }
 
+        if not attachments or not attachments_pdf:
+            raise ValueError("Missing attachments: neither 'attachments' nor 'attachments_pdf' was provided.")
+
         store_path_set = set()
         parent_id_store_path_map = {i['parent_id']: i['store_path'] for i in attachments_pdf}
         x_attachments_pdf = []
         for item in attachments:
-            store_path = item['store_path']
+            # The video branch will generate store_path_pre, so check for the existence of store_path_pre first.
+            # For other branches, the value will be null.
+            store_path = item['store_path_pre'] or item['store_path']
             parent_id = item['id']
             if store_path not in db_store_path_set:
                 raise ValueError(f"store_path not found in db: {store_path}")
@@ -108,7 +111,6 @@ class FlowUpdater:
         attachments_pdf = op_meta_record['x_attachments_pdf']
         data_source = op_meta_record['data_source']
 
-        # 校验参数
         if not report_id or not status or not start_stage or not current_stage or not target_stage or (not attachments and start_stage == 'convert'):
             raise ValueError(f"Invalid op_meta_record: {op_meta_record}")
         if status == 'success' and target_stage != current_stage:
@@ -116,7 +118,6 @@ class FlowUpdater:
             return
         attachments = [i for i in attachments if i['category'] == 'x_attachments']
 
-        # 确定结束阶段
         end_stage = target_stage if status == 'success' else current_stage if status == 'failed' else None
         if end_stage is None:
             logger.info(f"Invalid status: {status}.")
@@ -126,11 +127,9 @@ class FlowUpdater:
         if start_index > end_index:
             raise ValueError(f"start_stage cannot be after end_stage: {start_stage} -> {end_stage}.")
 
-        # 开始执行回更逻辑
         logger.info(
             f"😊 _id: {report_id}-{status}, start_step: {self.step_tuple[start_index]}, end_step: {self.step_tuple[end_index]}")
 
-        # 查询这个报告是否存在于当前数据源
         db_doc = self._check_and_create_collection(data_source).find_one({'_id': report_id},
                                                                          {'_id': 1, 'x_attachments': 1,
                                                                           'x_status_list': 1})
@@ -142,10 +141,8 @@ class FlowUpdater:
             logger.warning(f"{db_doc['_id']} statxus is not 'crawl_downloaded'")
             return
 
-        # 构建数据库中存在store_path 防止出现数据库中x_attachments与x_attachments_pdf数据不一致问题
         db_store_path = {f"s3://{i['bucket']}/{i['store_path']}" for i in db_doc['x_attachments']}
 
-        # 构建更新参数
         update_params = {}
         step_status = True
         for index, step in enumerate(self.step_tuple[start_index:end_index + 1], 1):
@@ -155,7 +152,6 @@ class FlowUpdater:
             item = self._step_handle(step, step_status, attachments, db_store_path, attachments_pdf)
             update_params.update(item)
 
-        # 执行更新
         if update_params:
             # logger.info(json.dumps(update_params, ensure_ascii=False, indent=2))
             self.update_mongo_data(report_id, data_source, update_params, kafka_ignore)

{orbitkit-0.8.43 → orbitkit-0.8.45/orbitkit.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: orbitkit
-Version: 0.8.43
+Version: 0.8.45
 Summary: This project is only for Orbit Tech internal use.
 Home-page: https://github.com/clown-0726/orbitkit
 Author: Lilu Cao

orbitkit-0.8.43/orbitkit/VERSION

@@ -1 +0,0 @@
-0.8.43