Merge branch 'PingChuan' into shiy-login

app/api/dify-dataset/api/documentApi.ts

@@ -225,6 +225,44 @@ export async function fetchUploadFileInfo(
     return response.data;
 }
 
+/**
+ * 下载文档原始文件
+ * 通过代理路由下载 Dify 知识库中的原始文件
+ *
+ * @param uploadFileInfo - 上传文件信息（从 fetchUploadFileInfo 获取）
+ * @returns File 对象
+ */
+export async function downloadOriginalFile(
+    uploadFileInfo: UploadFileInfo
+): Promise<File> {
+    if (!uploadFileInfo.download_url) {
+        throw new Error('无法获取原始文件下载地址');
+    }
+
+    // download_url 格式: /files/xxx/file-preview?...
+    // 转换为代理路由: /api/dataset/dify-files/xxx/file-preview?...
+    const downloadPath = uploadFileInfo.download_url.replace(/^\/files\//, '');
+    const proxyUrl = `${API_URL}/dify-files/${downloadPath}`;
+
+    console.log('[Dataset Client] 下载原始文件:', {
+        originalUrl: uploadFileInfo.download_url,
+        proxyUrl,
+    });
+
+    const response = await axios.get(proxyUrl, {
+        responseType: 'blob',
+        withCredentials: true,
+    });
+
+    const file = new File(
+        [response.data],
+        uploadFileInfo.name || 'document',
+        { type: uploadFileInfo.mime_type || 'application/octet-stream' }
+    );
+
+    return file;
+}
+
 /**
  * 预处理规则 ID
  */
@@ -244,6 +282,7 @@ export interface PreProcessingRule {
 export interface SegmentationConfig {
     separator: string;
     max_tokens: number;
+    chunk_overlap?: number;
 }
 
 /**