音视频支持分段

apps/knowledge/serializers/document.py

@@ -880,9 +880,15 @@ class DocumentSerializers(serializers.Serializer):
         @post(post_function=post_embedding)
         @transaction.atomic
         def save(self, instance: Dict, with_valid=False, **kwargs):
-            # 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
+            # 从 self.data 中获取分块参数
+            if self.data.get('limit') is not None:
+                instance['split_limit'] = self.data.get('limit')
+            if self.data.get('patterns') is not None:
+                instance['split_patterns'] = self.data.get('patterns')
+            if self.data.get('with_filter') is not None:
+                instance['split_with_filter'] = self.data.get('with_filter')
+
+            # 同时也支持从 instance 中获取分块参数（向后兼容）
             if instance.get('limit') is not None:
                 instance['split_limit'] = instance.get('limit')
             if instance.get('patterns') is not None:
@@ -1256,30 +1262,22 @@ class DocumentSerializers(serializers.Serializer):
             from common.utils.logger import maxkb_logger
             import os
 
-            # 添加详细日志
-            maxkb_logger.info(f"batch_save called with workspace_id: {workspace_id}, knowledge_id: {knowledge_id}")
-            maxkb_logger.info(f"instance_list contains {len(instance_list)} documents")
+            # 获取分块参数
+            split_limit = self.data.get('limit')
+            split_patterns = self.data.get('patterns')
+            split_with_filter = self.data.get('with_filter')
 
             # 在处理文档之前，先确保所有分块参数都被正确转换
             for idx, doc in enumerate(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 doc.get('limit') is not None:
-                    doc['split_limit'] = doc.get('limit')
-                    maxkb_logger.info(f"Document {idx}: Converting limit={doc.get('limit')} to split_limit")
-                if doc.get('patterns') is not None:
-                    doc['split_patterns'] = doc.get('patterns')
-                    maxkb_logger.info(f"Document {idx}: Converting patterns={doc.get('patterns')} to split_patterns")
-                if doc.get('with_filter') is not None:
-                    doc['split_with_filter'] = doc.get('with_filter')
-                    maxkb_logger.info(f"Document {idx}: Converting with_filter={doc.get('with_filter')} to split_with_filter")
-
-                maxkb_logger.info(f"Document {idx}: {doc.keys()}")
-                if 'stt_model_id' in doc:
-                    maxkb_logger.info(f"  - stt_model_id present: {doc['stt_model_id']}")
-                if 'llm_model_id' in doc:
-                    maxkb_logger.info(f"  - llm_model_id present: {doc['llm_model_id']}")
+                if split_limit is not None:
+                    doc['split_limit'] = split_limit
+                if split_patterns is not None:
+                    doc['split_patterns'] = split_patterns
+                if split_with_filter is not None:
+                    doc['split_with_filter'] = split_with_filter
 
             for document in instance_list:
                 # 检查是否是MinerU类型的文档（需要同时有llm_model_id和vision_model_id）

apps/knowledge/views/document.py

@@ -45,10 +45,22 @@ class DocumentView(APIView):
              get_knowledge_operation_object(keywords.get('knowledge_id')),
              {'name': r.data.get('name')}), )
     def post(self, request: Request, workspace_id: str, knowledge_id: str):
+        # 准备分块参数
+        serializer_data = {
+            'workspace_id': workspace_id,
+            'knowledge_id': knowledge_id
+        }
+
+        # 添加分块参数到 serializer_data
+        if 'limit' in request.data:
+            serializer_data['limit'] = request.data.get('limit')
+        if 'patterns' in request.data:
+            serializer_data['patterns'] = request.data.get('patterns')
+        if 'with_filter' in request.data:
+            serializer_data['with_filter'] = request.data.get('with_filter')
+
         return result.success(
-            DocumentSerializers.Create(
-                data={'workspace_id': workspace_id, 'knowledge_id': knowledge_id},
-            ).save(request.data))
+            DocumentSerializers.Create(data=serializer_data).save(request.data))
 
     @extend_schema(
         methods=['GET'],