音视频支持分段

apps/knowledge/serializers/document.py

@@ -1297,15 +1297,25 @@ class DocumentSerializers(serializers.Serializer):
 
             # 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
+                # 如果文档对象中已经有分块参数（前端传递的），优先使用
-                # 前端传递的是 limit, patterns, with_filter
+                # 否则使用从 serializer_data 获取的参数
-                # 我们需要将其转换为 split_limit, split_patterns, split_with_filter
+                if doc.get('split_limit') is not None:
-                if split_limit is not None:
+                    maxkb_logger.info(f"Document {idx} already has split_limit: {doc.get('split_limit')}")
+                elif split_limit is not None:
                     doc['split_limit'] = split_limit
-                if split_patterns is not None:
+                    maxkb_logger.info(f"Document {idx}: Setting split_limit from serializer_data: {split_limit}")
+
+                if doc.get('split_patterns') is not None:
+                    maxkb_logger.info(f"Document {idx} already has split_patterns: {doc.get('split_patterns')}")
+                elif split_patterns is not None:
                     doc['split_patterns'] = split_patterns
-                if split_with_filter is not None:
+                    maxkb_logger.info(f"Document {idx}: Setting split_patterns from serializer_data: {split_patterns}")
+
+                if doc.get('split_with_filter') is not None:
+                    maxkb_logger.info(f"Document {idx} already has split_with_filter: {doc.get('split_with_filter')}")
+                elif split_with_filter is not None:
                     doc['split_with_filter'] = split_with_filter
+                    maxkb_logger.info(f"Document {idx}: Setting split_with_filter from serializer_data: {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

@@ -434,6 +434,9 @@ class DocumentView(APIView):
             from common.utils.logger import maxkb_logger
             maxkb_logger.info(f"=== BatchCreate View ===")
             maxkb_logger.info(f"request.data type: {type(request.data)}")
+            maxkb_logger.info(f"request.query_params: {dict(request.query_params)}")
+            maxkb_logger.info(f"request.POST: {dict(request.POST)}")
+
             if isinstance(request.data, list):
                 maxkb_logger.info(f"request.data is list with {len(request.data)} items")
                 # 检查第一个文档的参数
@@ -449,14 +452,34 @@ class DocumentView(APIView):
                 'workspace_id': workspace_id
             }
 
-            # 从第一个文档中提取分块参数（所有文档使用相同的分块设置）
+            # 尝试从多个地方获取分块参数
+            # 1. 从 query_params 获取（GET 参数）
+            if 'limit' in request.query_params:
+                serializer_data['limit'] = request.query_params.get('limit')
+                maxkb_logger.info(f"Got limit from query_params: {serializer_data['limit']}")
+            if 'patterns' in request.query_params:
+                serializer_data['patterns'] = request.query_params.get('patterns')
+            if 'with_filter' in request.query_params:
+                serializer_data['with_filter'] = request.query_params.get('with_filter')
+
+            # 2. 从 POST 数据获取（表单数据）
+            if 'limit' in request.POST:
+                serializer_data['limit'] = request.POST.get('limit')
+                maxkb_logger.info(f"Got limit from POST: {serializer_data['limit']}")
+            if 'patterns' in request.POST:
+                serializer_data['patterns'] = request.POST.getlist('patterns')
+            if 'with_filter' in request.POST:
+                serializer_data['with_filter'] = request.POST.get('with_filter')
+
+            # 3. 从文档对象中获取
             if isinstance(request.data, list) and len(request.data) > 0:
                 first_doc = request.data[0]
-                if 'limit' in first_doc:
+                if 'limit' in first_doc and 'limit' not in serializer_data:
                     serializer_data['limit'] = first_doc.get('limit')
-                if 'patterns' in first_doc:
+                    maxkb_logger.info(f"Got limit from first doc: {serializer_data['limit']}")
+                if 'patterns' in first_doc and 'patterns' not in serializer_data:
                     serializer_data['patterns'] = first_doc.get('patterns')
-                if 'with_filter' in first_doc:
+                if 'with_filter' in first_doc and 'with_filter' not in serializer_data:
                     serializer_data['with_filter'] = first_doc.get('with_filter')
 
             maxkb_logger.info(f"BatchCreate serializer_data: {serializer_data}")

ui/src/views/document/UploadDocument.vue

@@ -209,8 +209,24 @@ function submit() {
         }
       }
       // 传递分段规则（如果有）
-      if (SetRulesRef.value?.form?.patterns) {
+      if (SetRulesRef.value?.form) {
-        doc.split_patterns = SetRulesRef.value.form.patterns
+        // 传递patterns
+        if (SetRulesRef.value.form.patterns) {
+          doc.split_patterns = SetRulesRef.value.form.patterns
+        }
+        // 传递limit
+        if (SetRulesRef.value.form.limit !== undefined) {
+          doc.split_limit = SetRulesRef.value.form.limit
+        }
+        // 传递with_filter
+        if (SetRulesRef.value.form.with_filter !== undefined) {
+          doc.split_with_filter = SetRulesRef.value.form.with_filter
+        }
+        console.log('Media chunking params:', {
+          split_patterns: doc.split_patterns,
+          split_limit: doc.split_limit,
+          split_with_filter: doc.split_with_filter
+        })
       }
       console.log('Final doc object for media:', doc)
     }