GLM-4.5V MCP Server

import { McpServer } from "@modelcontextprotocol/sdk/server/mcp.js"; import { StdioServerTransport } from "@modelcontextprotocol/sdk/server/stdio.js"; import fs from "node:fs/promises"; import path from "node:path"; import { z } from "zod"; import sharp from "sharp"; const mcpServer = new McpServer({ name: "gusheng-toolbox", version: "0.1.0", }); interface ImageResult { ok: boolean; image?: { source: string; mime: string; dataUrl: string; width?: number; height?: number; }; error?: string; } interface VisionResult { ok: boolean; result?: string | object; error?: string; metadata?: { mode: string; returnJson: boolean; timestamp: number; }; } interface FileProcessingResult { ok: boolean; fileId?: string; content?: string; fileType?: string; filename?: string; error?: string; metadata?: { uploadTime: number; fileSize: number; processingTime: number; }; } // 注册图片读取工具 mcpServer.registerTool("read_image", { description: "读取本地/URL图片并返回 dataURL 与尺寸信息", inputSchema: { path: z.string().describe("图片路径或URL"), maxSide: z.number().optional().describe("最大边长，用于缩放"), }, }, async ({ path: imagePath, maxSide }) => { try { const result = await readImage(imagePath, maxSide); 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 }; } }); // 注册视觉查询工具 mcpServer.registerTool("vision_query", { description: "调用 GLM-4.5V 对图片进行 OCR/问答/检测", inputSchema: { path: z.string().describe("图片路径或URL"), prompt: z.string().describe("查询提示词"), mode: z.enum(["describe", "ocr", "qa", "detect"]).default("describe").describe("查询模式"), returnJson: z.boolean().default(false).describe("是否返回JSON格式结果"), }, }, async ({ path: imagePath, prompt, mode, returnJson }) => { try { const result = await visionQuery(imagePath, prompt, mode, returnJson); 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 }; } }); // 注册文件处理工具 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 }; } }); // 图片压缩函数 async function compressImage(buffer: Buffer, maxSide: number = 1024, quality: number = 80): Promise<Buffer> { try { const image = sharp(buffer); const metadata = await image.metadata(); let width = metadata.width || 1024; let height = metadata.height || 1024; // 计算缩放比例 if (width > maxSide || height > maxSide) { const ratio = Math.min(maxSide / width, maxSide / height); width = Math.round(width * ratio); height = Math.round(height * ratio); } return await image .resize(width, height, { fit: 'inside' }) .jpeg({ quality, progressive: true }) .toBuffer(); } catch (error) { console.error("Image compression failed:", error); return buffer; // 失败时返回原始buffer } } async function readImage(imagePath: string, maxSide?: number): Promise<ImageResult> { try { console.error(`[DEBUG] readImage called with path: ${imagePath.substring(0, 50)}...`); console.error(`[DEBUG] Path starts with data:? ${imagePath.startsWith("data:")}`); let buffer: Buffer; if (imagePath.startsWith("data:")) { // Data URL 格式 console.error(`[DEBUG] Processing data URL in readImage`); const commaIndex = imagePath.indexOf(','); if (commaIndex === -1) { throw new Error("Invalid data URL format: no comma found"); } const base64Data = imagePath.substring(commaIndex + 1); if (!base64Data) { throw new Error("Invalid data URL format: no base64 data"); } console.error(`[DEBUG] Base64 data length: ${base64Data.length}`); buffer = Buffer.from(base64Data, 'base64'); } else if (imagePath.startsWith("http://") || imagePath.startsWith("https://")) { // HTTP/HTTPS URL图片 console.error(`[DEBUG] Fetching HTTP/HTTPS image`); const response = await fetch(imagePath); if (!response.ok) { throw new Error(`Failed to fetch image: ${response.statusText}`); } buffer = Buffer.from(await response.arrayBuffer()); } else { // 本地文件 console.error(`[DEBUG] Reading local file: ${imagePath}`); const resolvedPath = path.resolve(imagePath); buffer = await fs.readFile(resolvedPath); } console.error(`[DEBUG] Original buffer size: ${buffer.length}`); // 压缩图片 const compressedBuffer = await compressImage(buffer, maxSide || 1024); console.error(`[DEBUG] Compressed buffer size: ${compressedBuffer.length}`); const mime = "image/jpeg"; // 压缩后统一为 JPEG 格式 const dataUrl = `data:${mime};base64,${compressedBuffer.toString("base64")}`; return { ok: true, image: { source: imagePath, mime, dataUrl, width: (await sharp(compressedBuffer).metadata()).width, height: (await sharp(compressedBuffer).metadata()).height } }; } catch (error) { return { ok: false, error: error instanceof Error ? error.message : "Unknown error" }; } } async function visionQuery(imagePath: string, prompt: string, mode: string, returnJson: boolean): Promise<VisionResult> { try { let imageBase64: string; let buffer: Buffer; if (imagePath.startsWith("data:")) { // Data URL 格式 const base64Data = imagePath.split(',')[1]; if (!base64Data) { throw new Error("Invalid data URL format"); } buffer = Buffer.from(base64Data, 'base64'); } else if (imagePath.startsWith("http://") || imagePath.startsWith("https://")) { // HTTP/HTTPS URL图片 const response = await fetch(imagePath); if (!response.ok) { throw new Error(`Failed to fetch image: ${response.statusText}`); } buffer = Buffer.from(await response.arrayBuffer()); } else { // 本地文件 const resolvedPath = path.resolve(imagePath); buffer = await fs.readFile(resolvedPath); } // 压缩图片以减少token使用量 const compressedBuffer = await compressImage(buffer, 800, 75); // 更小尺寸和质量 imageBase64 = compressedBuffer.toString("base64"); const payload = buildGlmPayload({ prompt, imageBase64, mode, returnJson }); const glmBaseUrl = process.env.GLM_BASE_URL || "https://open.bigmodel.cn/api/paas/v4/chat/completions"; const glmApiKey = process.env.GLM_API_KEY; if (!glmApiKey) { throw new Error("GLM_API_KEY environment variable is required"); } const response = await fetch(glmBaseUrl, { method: "POST", headers: { "Content-Type": "application/json", "Authorization": `Bearer ${glmApiKey}` }, body: JSON.stringify(payload) }); if (!response.ok) { throw new Error(`GLM API request failed: ${response.statusText}`); } const data = await response.json(); const result = normalizeGlmResult(data, { mode, returnJson }); return { ok: true, result, metadata: { mode, returnJson, timestamp: Date.now() } }; } catch (error) { return { ok: false, error: error instanceof Error ? error.message : "Unknown error" }; } } // 文件处理主函数 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: ${content.length}`); const processingTime = Date.now() - startTime; return { ok: true, fileId, content, fileType: getFileType(filename), filename, metadata: { uploadTime: startTime, fileSize, processingTime } }; } catch (error) { return { ok: false, error: error instanceof Error ? error.message : "Unknown error" }; } } function guessExt(imagePath: string): string { const ext = imagePath.split(".").pop()?.toLowerCase() || "png"; return ext === "jpg" ? "jpeg" : ext; } function truncatePrompt(prompt: string, maxLength: number = 500): string { if (prompt.length <= maxLength) return prompt; return prompt.substring(0, maxLength) + "..."; } function buildGlmPayload(opts: { prompt: string; imageBase64: string; mode: string; returnJson: boolean; }) { const { prompt, imageBase64, mode, returnJson } = opts; // 截断过长的 prompt const truncatedPrompt = truncatePrompt(prompt, 300); let systemPrompt = ""; switch (mode) { case "ocr": systemPrompt = "识别图片中的文字。"; break; case "qa": systemPrompt = "根据图片回答问题。"; break; case "detect": systemPrompt = "识别图片中的物体。"; break; default: systemPrompt = "描述图片内容。"; } if (returnJson) { systemPrompt += "用JSON格式回答。"; } return { model: "glm-4v-plus", messages: [ { role: "system", content: systemPrompt }, { role: "user", content: [ { type: "text", text: truncatedPrompt }, { type: "image_url", image_url: { url: `data:image/jpeg;base64,${imageBase64}` } } ] } ], temperature: 0.1, max_tokens: 1000 }; } function normalizeGlmResult(data: any, opts: { mode: string; returnJson: boolean }) { if (data.error) { throw new Error(data.error.message || "GLM API error"); } const content = data.choices?.[0]?.message?.content || ""; if (opts.returnJson) { try { return JSON.parse(content); } catch { return { text: content, parsed: false }; } } return content; } // 文件处理辅助函数 function isImageFile(filename: string): boolean { const imageExtensions = ['.png', '.jpg', '.jpeg', '.gif', '.bmp', '.webp']; const ext = path.extname(filename).toLowerCase(); return imageExtensions.includes(ext); } function isSupportedFileType(filename: string): boolean { const supportedExtensions = [ '.pdf', '.docx', '.doc', '.xls', '.xlsx', '.ppt', '.pptx', '.png', '.jpg', '.jpeg', '.csv', '.txt' ]; const ext = path.extname(filename).toLowerCase(); return supportedExtensions.includes(ext); } function getFileType(filename: string): string { const ext = path.extname(filename).toLowerCase(); const typeMap: { [key: string]: string } = { '.pdf': 'PDF文档', '.docx': 'Word文档', '.doc': 'Word文档', '.xls': 'Excel表格', '.xlsx': 'Excel表格', '.ppt': 'PowerPoint演示文稿', '.pptx': 'PowerPoint演示文稿', '.png': 'PNG图片', '.jpg': 'JPEG图片', '.jpeg': 'JPEG图片', '.csv': 'CSV数据文件', '.txt': '文本文件' }; return typeMap[ext] || '未知文件类型'; } 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: ${error instanceof Error ? error.message : 'Unknown error'}`); } } async function getFileContentFromGLM(fileId: string): Promise<string> { const glmApiKey = process.env.GLM_API_KEY; if (!glmApiKey) { throw new Error("GLM_API_KEY environment variable is required"); } try { const response = await fetch(`https://open.bigmodel.cn/api/paas/v4/files/${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'}`); } } async function main() { // 显示 MCP 服务器配置信息 console.error("=".repeat(60)); console.error("🚀 Bytenote MCP Server"); console.error("=".repeat(60)); console.error(`📦 Server Name: glv`); console.error(`🔖 Version: 0.1.0`); console.error(`📡 Transport: StdioServerTransport`); console.error(`🔧 Available Tools:`); console.error(` • read_image - 读取本地/URL图片并返回 dataURL 与尺寸信息`); console.error(` • vision_query - 调用 GLM-4.5V 对图片进行 OCR/问答/检测`); console.error(` • process_file - 使用 GLM-4.5V 处理文件（上传并提取内容）`); console.error(`🌐 GLM API Endpoint: ${process.env.GLM_BASE_URL || "https://open.bigmodel.cn/api/paas/v4/chat/completions"}`); console.error(`🔑 API Key Status: ${process.env.GLM_API_KEY ? "✅ Configured" : "❌ Missing"}`); if (!process.env.GLM_API_KEY) { console.error(`⚠️ Warning: GLM_API_KEY not found in environment variables`); console.error(` Please set your API key in .env file`); } console.error("=".repeat(60)); console.error("🔄 Starting MCP server..."); await mcpServer.connect(new StdioServerTransport()); console.error("✅ MCP server connected and ready!"); console.error("=".repeat(60)); } if (import.meta.url === `file://${process.argv[1]}`) { main().catch((error) => { console.error("Server error:", error); process.exit(1); }); }