音视频支持分段

apps/knowledge/serializers/document.py

@@ -880,6 +880,21 @@ class DocumentSerializers(serializers.Serializer):
         @post(post_function=post_embedding)
         @transaction.atomic
         def save(self, instance: Dict, with_valid=False, **kwargs):
+            # 添加调试日志
+            from common.utils.logger import maxkb_logger
+            maxkb_logger.info(f"Save called with instance keys: {list(instance.keys())}")
+            maxkb_logger.info(f"save method - limit: {instance.get('limit')}, split_patterns: {instance.get('patterns')}, with_filter: {instance.get('with_filter')}")
+
+            # 确保分块参数使用正确的键名
+            # 前端传递的是 limit, patterns, with_filter
+            # 我们需要将其转换为 split_limit, split_patterns, split_with_filter
+            if instance.get('limit') is not None:
+                instance['split_limit'] = instance.get('limit')
+            if instance.get('patterns') is not None:
+                instance['split_patterns'] = instance.get('patterns')
+            if instance.get('with_filter') is not None:
+                instance['split_with_filter'] = instance.get('with_filter')
+
             if with_valid:
                 DocumentInstanceSerializer(data=instance).is_valid(raise_exception=True)
                 self.is_valid(raise_exception=True)
@@ -947,6 +962,13 @@ class DocumentSerializers(serializers.Serializer):
             # 添加音视频STT模型参数到meta
             if instance.get('stt_model_id'):
                 source_meta['stt_model_id'] = instance.get('stt_model_id')
+            # 添加分块参数到meta
+            if instance.get('split_limit') is not None:
+                source_meta['split_limit'] = instance.get('split_limit')
+            if instance.get('split_patterns') is not None:
+                source_meta['split_patterns'] = instance.get('split_patterns')
+            if instance.get('with_filter') is not None:
+                source_meta['split_with_filter'] = instance.get('with_filter')
             meta = {**instance.get('meta'), **source_meta} if instance.get('meta') is not None else source_meta
             meta = convert_uuid_to_str(meta)
 
@@ -1274,7 +1296,18 @@ class DocumentSerializers(serializers.Serializer):
             
             # 插入文档
             for document in instance_list:
+                # 确保分块参数使用正确的键名
+                # 前端传递的是 limit, patterns, with_filter
+                # 我们需要将其转换为 split_limit, split_patterns, split_with_filter
+                if document.get('limit') is not None:
+                    document['split_limit'] = document.get('limit')
+                if document.get('patterns') is not None:
+                    document['split_patterns'] = document.get('patterns')
+                if document.get('with_filter') is not None:
+                    document['split_with_filter'] = document.get('with_filter')
+
                 maxkb_logger.info(f"Processing document: {document.get('name')}, paragraphs count: {len(document.get('paragraphs', []))}")
+                maxkb_logger.info(f"Document split params - split_limit: {document.get('split_limit')}, split_patterns: {document.get('split_patterns')}, split_with_filter: {document.get('split_with_filter')}")
                 document_paragraph_dict_model = DocumentSerializers.Create.get_document_paragraph_model(
                     knowledge_id,
                     document