process_file

Instructions

Implementation Reference

• src/index.ts:113-140 (registration)

  Registers the 'process_file' tool with the MCP server, including description, input schema using Zod, and an async wrapper handler that calls the processFile function and handles errors.

  mcpServer.registerTool("process_file", {
    description: "使用 GLM-4.5V 处理文件（上传并提取内容）。支持 PDF、DOCX、DOC、XLS、XLSX、PPT、PPTX、PNG、JPG、JPEG、CSV 等格式",
    inputSchema: {
      filePath: z.string().describe("文件路径（本地文件路径）"),
      extractPrompt: z.string().optional().describe("可选的内容提取提示词，用于指导如何提取文件内容"),
    },
  }, async ({ filePath, extractPrompt }) => {
    try {
      const result = await processFile(filePath, extractPrompt);
      return {
        content: [{
          type: "text" as const,
          text: JSON.stringify(result, null, 2)
        }]
      };
    } catch (error) {
      return {
        content: [{
          type: "text" as const,
          text: JSON.stringify({
            ok: false,
            error: error instanceof Error ? error.message : "Unknown error"
          }, null, 2)
        }],
        isError: true
      };
    }
  });

• src/index.ts:307-362 (handler)

  Core handler function for processing files: validates file path, size, and type; uploads to GLM API; retrieves extracted content; returns structured result.

  async function processFile(filePath: string, extractPrompt?: string): Promise<FileProcessingResult> {
    const startTime = Date.now();
  
    try {
      console.error(`[DEBUG] processFile called with path: ${filePath}`);
  
      // 检查文件是否存在
      const resolvedPath = path.resolve(filePath);
      const stats = await fs.stat(resolvedPath);
      const fileSize = stats.size;
      const filename = path.basename(filePath);
  
      console.error(`[DEBUG] File size: ${fileSize} bytes, filename: ${filename}`);
  
      // 检查文件大小限制
      const maxSize = isImageFile(filename) ? 5 * 1024 * 1024 : 50 * 1024 * 1024; // 图片5MB，其他50MB
      if (fileSize > maxSize) {
        throw new Error(`文件大小超过限制。图片文件最大5MB，其他文件最大50MB。当前文件大小：${(fileSize / 1024 / 1024).toFixed(2)}MB`);
      }
  
      // 检查文件格式
      if (!isSupportedFileType(filename)) {
        throw new Error(`不支持的文件格式。支持的格式：PDF、DOCX、DOC、XLS、XLSX、PPT、PPTX、PNG、JPG、JPEG、CSV`);
      }
  
      // 1. 上传文件
      console.error(`[DEBUG] Uploading file...`);
      const fileId = await uploadFileToGLM(resolvedPath, filename);
      console.error(`[DEBUG] File uploaded with ID: ${fileId}`);
  
      // 2. 获取文件内容
      console.error(`[DEBUG] Getting file content...`);
      const content = await getFileContentFromGLM(fileId);
      console.error(`[DEBUG] Content extracted, length: ${content.length}`);
  
      const processingTime = Date.now() - startTime;
  
      return {
        ok: true,
        fileId,
        content,
        fileType: getFileType(filename),
        filename,
        metadata: {
          uploadTime: startTime,
          fileSize,
          processingTime
        }
      };
    } catch (error) {
      return {
        ok: false,
        error: error instanceof Error ? error.message : "Unknown error"
      };
    }
  }

• src/index.ts:36-48 (schema)

  TypeScript interface defining the output structure of the file processing result.

  interface FileProcessingResult {
    ok: boolean;
    fileId?: string;
    content?: string;
    fileType?: string;
    filename?: string;
    error?: string;
    metadata?: {
      uploadTime: number;
      fileSize: number;
      processingTime: number;
    };
  }

• src/index.ts:485-523 (helper)

  Helper function that uploads the file to the GLM API using FormData and returns the assigned file ID.

  async function uploadFileToGLM(filePath: string, filename: string): Promise<string> {
    const glmApiKey = process.env.GLM_API_KEY;
    if (!glmApiKey) {
      throw new Error("GLM_API_KEY environment variable is required");
    }
  
    try {
      // 读取文件
      const fileBuffer = await fs.readFile(filePath);
  
      // 创建 FormData
      const formData = new FormData();
      const blob = new Blob([new Uint8Array(fileBuffer)]);
      formData.append('file', blob, filename);
      formData.append('purpose', 'file-extract');
  
      const response = await fetch('https://open.bigmodel.cn/api/paas/v4/files', {
        method: 'POST',
        headers: {
          'Authorization': `Bearer ${glmApiKey}`
        },
        body: formData
      });
  
      if (!response.ok) {
        const errorText = await response.text();
        throw new Error(`文件上传失败: ${response.status} ${response.statusText} - ${errorText}`);
      }
  
      const result = await response.json();
      if (!result.id) {
        throw new Error('上传响应中缺少文件ID');
      }
  
      return result.id;
    } catch (error) {
      throw new Error(`文件上传失败: ${error instanceof Error ? error.message : 'Unknown error'}`);
    }
  }

• src/index.ts:525-557 (helper)

  Helper function that retrieves the extracted content from the GLM API using the file ID, handling JSON or text responses.

  async function getFileContentFromGLM(fileId: string): Promise<string> {
    const glmApiKey = process.env.GLM_API_KEY;
    if (!glmApiKey) {
      throw new Error("GLM_API_KEY environment variable is required");
    }
  
    try {
      const response = await fetch(`https://open.bigmodel.cn/api/paas/v4/files/${fileId}/content`, {
        method: 'GET',
        headers: {
          'Authorization': `Bearer ${glmApiKey}`
        }
      });
  
      if (!response.ok) {
        const errorText = await response.text();
        throw new Error(`获取文件内容失败: ${response.status} ${response.statusText} - ${errorText}`);
      }
  
      // 根据响应类型处理内容
      const contentType = response.headers.get('content-type') || '';
  
      if (contentType.includes('application/json')) {
        const jsonResult = await response.json();
        return JSON.stringify(jsonResult, null, 2);
      } else {
        // 对于其他类型，尝试作为文本读取
        return await response.text();
      }
    } catch (error) {
      throw new Error(`获取文件内容失败: ${error instanceof Error ? error.message : 'Unknown error'}`);
    }
  }