longchain4j 学习系列(4)-mcp调用

继续学习langchain4j，以下是langchain4j 调用MCP的示例：

1、添加pom依赖

<dependency>
    <groupId>dev.langchain4j</groupId>
    <artifactId>langchain4j-mcp</artifactId>
    <version>1.1.0-beta7</version>
</dependency>

这里要注意版本号，不同的版本api细节略有不同。

2、Mcp Client初始化

/*** 初始化SSE客户端** @param sseUrl SSE服务器连接地址* @return McpClient实例*/
private static McpClient initSseClient(String sseUrl) {// 构建默认MCP客户端return new DefaultMcpClient.Builder().clientName("yjmyzz.cnblogs.com").protocolVersion("2024-11-05").toolExecutionTimeout(Duration.ofSeconds(10))// 配置HTTP传输层参数.transport(new HttpMcpTransport.Builder()// 设置SSE服务器连接URL.sseUrl(sseUrl)// 设置连接超时时间.timeout(Duration.ofSeconds(10))// 启用请求日志记录.logRequests(true)// 启用响应日志记录.logResponses(true).build()).build();
}

3、定义1个Assistant接口

    private interface Assistant {String chat(String userMessage);}

4、调用MCP

    /*** 直接获取订单状态信息** @param orderId 订单ID，用于查询指定订单的状态* @return ResponseEntity<String> 包含订单状态信息的响应实体，成功时返回订单状态JSON字符串，* 失败时返回包含错误信息的JSON字符串*/@GetMapping(value = "/order", produces = MediaType.APPLICATION_JSON_VALUE)public ResponseEntity<String> getOrderStatusDirect(@RequestParam String orderId) {McpClient mcpClient = null;try {// 初始化SSE客户端连接mcpClient = initSseClient("http://localhost:8070/sse");// 构建AI助手服务，配置聊天模型和工具提供者Assistant assistant = AiServices.builder(Assistant.class).chatModel(deepSeekChatModel).toolProvider(McpToolProvider.builder().mcpClients(mcpClient).build()).build();// 调用AI助手查询订单状态String response = assistant.chat("查询订单状态，订单号：" + orderId);return ResponseEntity.ok(response);} catch (Exception e) {log.error("查询订单状态时发生错误", e);return ResponseEntity.ok("{\"error\":\"查询订单状态失败: " + e.getMessage() + "\"}");} finally {// 确保MCP客户端连接被正确关闭if (mcpClient != null) {try {mcpClient.close();} catch (Exception e) {log.error("关闭MCP客户端时发生错误", e);}}}}

注：这里的mcp server 可参考 spring-ai 学习系列(5)-MCP(webflux sse)

日志：

2025-10-03T20:30:43.277+08:00  INFO 18312 --- [longchain4j-study] [nio-8080-exec-2] d.l.http.client.log.LoggingHttpClient    : HTTP request:
- method: POST
- url: https://api.deepseek.com/chat/completions
- headers: [Authorization: Beare...00], [User-Agent: langchain4j-openai], [Content-Type: application/json]
- body: {"model" : "deepseek-chat","messages" : [ {"role" : "user","content" : "查询订单状态，订单号：25070602"} ],"temperature" : 0.7,"stream" : false,"max_tokens" : 2048,"tools" : [ {"type" : "function","function" : {"name" : "queryOrderStatus","description" : "根据订单号查询订单状态","parameters" : {"type" : "object","properties" : {"orderNo" : {"type" : "string","description" : "订单号,格式为8位数字,比如：25070601"}},"required" : [ "orderNo" ]}}} ]
}...2025-10-03T20:30:45.770+08:00  INFO 18312 --- [longchain4j-study] [nio-8080-exec-2] d.l.http.client.log.LoggingHttpClient    : HTTP request:
- method: POST
- url: https://api.deepseek.com/chat/completions
- headers: [Authorization: Beare...00], [User-Agent: langchain4j-openai], [Content-Type: application/json]
- body: {"model" : "deepseek-chat","messages" : [ {"role" : "user","content" : "查询订单状态，订单号：25070602"}, {"role" : "assistant","content" : "我来帮您查询订单号为25070602的订单状态。","tool_calls" : [ {"id" : "call_00_QnwEmctjY5kaQf5MEQ5C2i9Y","type" : "function","function" : {"name" : "queryOrderStatus","arguments" : "{\"orderNo\": \"25070602\"}"}} ]}, {"role" : "tool","tool_call_id" : "call_00_QnwEmctjY5kaQf5MEQ5C2i9Y","content" : "\"订单号：25070602，订单状态：已完成\""} ],"temperature" : 0.7,"stream" : false,"max_tokens" : 2048,"tools" : [ {"type" : "function","function" : {"name" : "queryOrderStatus","description" : "根据订单号查询订单状态","parameters" : {"type" : "object","properties" : {"orderNo" : {"type" : "string","description" : "订单号,格式为8位数字,比如：25070601"}},"required" : [ "orderNo" ]}}} ]
}

　

文中代码：GitHub - yjmyzz/langchain4j-study at day04