上文《Netty挖掘机(三)结合spring搭建netty脚手架》,结合Spring+Netty实现API接口服务模块,本文继续码字,以Springboot的方式接入Netty实现API接口服务。
我们知道,spring-boot-starter-web 默认的web容器是Tomcat,同样也支持修改容器
如要把Tomcat 切换为Jetty,我们首先要做的是先移除包中的Tomcat依赖,再加入Jetty容器
1 2 3 4 5 6 7 8 9 10 11 12 13 14 | <dependency> <groupId>org.springframework.boot</groupId> <artifactId>spring-boot-starter-web</artifactId> <exclusions> <exclusion> <artifactId>spring-boot-starter-tomcat</artifactId> <groupId>org.springframework.boot</groupId> </exclusion> </exclusions> </dependency> <dependency> <groupId>org.springframework.boot</groupId> <artifactId>spring-boot-starter-jetty</artifactId> </dependency> |
以上操作,即可将默认的Tomcat容器切换为Jetty容器。
然而,当使用Netty替代web容器的时候,也是上面这种做法吗?
回顾下Netty和Servlet的区别
-
Servlet
Servlet是一种Java EE规范 ,是用Java编写的服务器端程序。其主要功能在于交互式地浏览和修改数据,生成动态Web内容。狭义的Servlet是指Java语言实现的一个接口,广义的Servlet是指任何实现了这个Servlet接口的类,一般情况下,人们将Servlet理解为后者。
Tomcat和Jetty 是一个servlet容器,而Jetty比大多数的servlet容器要更轻,即轻量级的servlet容器。
-
Netty
而Netty是一个异步事件驱动的基于NIO的网络应用程序框架,它支持扩展实现自己的servlet容器。
当需要面对处理大量网络协议的时候,建议使用Netty,当仅用于HTTP应用程序的时候,可以使用servlet容器。
所以当使用Netty替代tomcat等servlet容器时,可以实现一个自定义的servlet容器。
搭建思路
- 首先Springboot 中移除Tomcat 依赖;
- 引入Servlet依赖,自己实现一个Servlet上下文,并加入Spring上下文;
- Netty捕获到一个完整的Http 请求后,转换为ServletRequest,交给DispatcherServlet处理;
- DispatcherServlet内部进行解析, 调用HandlerMapping寻找处理器,找到对应的处理器Controller后执行并返回处理结果;

process.jpg
搭建
Maven pom配置
-
主要引入了Netty、Springboot的相关依赖,并且排除了内嵌的tomcat包依赖;
-
项目中使用了Servlet来接收http,而我们又排除了tomcat依赖,所以需手动加上servlet-api依赖;
全局返回实体
1 2 3 4 5 6 7 8 9 10 11 12 | @Data public class ResultVO { private Integer result = 0; private String msg; private String data; public static ResultVO create(String data){ ResultVO ret = new ResultVO(); ret.setMsg("success"); ret.setData(data); return ret; } } |
定义功能类
1 2 3 4 5 6 7 8 | @RestController @RequestMapping("/user") public class UserController { @GetMapping("/get") public ResultVO getUser() { return ResultVO.create("my name is jerry"); } } |
Netty 启动类
通过Spring上下文找到dispatcherServlet,并传入Channel初始化配置类,用于执行具体的业务逻辑。
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 | @Slf4j @Component public class HttpServer { @Value("${netty.port:8080}") private int port; @Autowired private DispatcherServlet dispatcherServlet; static EventLoopGroup boss = new NioEventLoopGroup(); static EventLoopGroup worker = new NioEventLoopGroup(); public void start() { log.info("############# start server at port: {}... #############", port); try { ServerBootstrap bootstrap = new ServerBootstrap(); bootstrap // 绑定两个组 .group(boss, worker) // 创建NioServerSocketChannel实例 .channel(NioServerSocketChannel.class) // 添加Channel初始化配置类 .childHandler(new HttpChannelInitializer(dispatcherServlet)) // 服务于boss线程(accept connect) // 设置TCP中的连接队列大小,如果队列满了,会发送一个ECONNREFUSED错误信息给C端,即“ Connection refused” .option(ChannelOption.SO_BACKLOG, HttpChannelOptionConstants.SO_BACKLOG) // 设置关闭tcp的Nagle算法(尽可能发送大块数据,避免网络中充斥着许多小数据块),要求高实时性 .childOption(ChannelOption.TCP_NODELAY, HttpChannelOptionConstants.TCP_NODELAY) // 设置启用心跳保活机制 .childOption(ChannelOption.SO_KEEPALIVE, HttpChannelOptionConstants.SO_KEEPALIVE); bootstrap.bind(port).sync(); } catch (InterruptedException e) { e.printStackTrace(); } } } |
管道初始化配置类
每次有请求进来时,调用初始化配置,将处理器添加到管道。
1 2 3 4 5 6 7 8 9 10 11 | @Override protected void initChannel(Channel ch) throws Exception { ChannelPipeline pipeline = ch.pipeline(); pipeline.addLast(new HttpRequestDecoder()); pipeline.addLast(new HttpResponseEncoder()); pipeline.addLast(new HttpObjectAggregator(64*1024)); pipeline.addLast(new HttpFirstServerHandler()); pipeline.addLast(new HttpLastServerHandler(dispatcherServlet)); } |
- 添加了一个支持最大消息为64*1024 kb大小的
HttpFirstServerHandler ,即聚合了多个Http片段。而一般不设置的话默认是有多个Http的片段组合成一次请求的; - 添加了两个处理器,一个用于将Netty接收到的Http请求包装到
HttpServletRequest 中,一个用于接收MockHttpServletRequest ,并使用dispatcherServlet 执行内部的业务逻辑。
包装HttpServletRequest
Netty捕获到一个完整的Http 请求后,转换为ServletRequest,转发到下一个处理器
假设管道配置了聚合http片段时,入站处理器需要继承
代码如下
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 | public void channelRead0(ChannelHandlerContext channelHandlerContext, FullHttpRequest fullHttpRequest) throws Exception { MockHttpServletRequest servletRequest = new MockHttpServletRequest(); String uri = fullHttpRequest.uri(); if(uri.indexOf(favicon) >= 0) { return; } uri = URLDecoder.decode(uri, "UTF-8"); String method = fullHttpRequest.method().name(); ByteBuf content = fullHttpRequest.content(); fullHttpRequest.headers().forEach((header) -> servletRequest.addHeader(header.getKey(), header.getValue())); UriComponents uriComponents = UriComponentsBuilder.fromUriString(uri).build(); String path = uriComponents.getPath(); servletRequest.setRequestURI(uri); servletRequest.setServletPath(path); servletRequest.setScheme(ofNullable(uriComponents.getScheme()).orElse(null)); servletRequest.setServerName(ofNullable(uriComponents.getHost()).orElse(null)); servletRequest.setServerPort(ofNullable(uriComponents.getPort()).orElse(null)); servletRequest.setMethod(method); byte[] byteArr = new byte[content.readableBytes()]; content.getBytes(content.readerIndex(), byteArr); String contentStr = new String(byteArr, "UTF-8"); servletRequest.setContent(byteArr); if(HttpMethod.GET.name().equalsIgnoreCase(method)) { QueryStringDecoder uriDecoder = new QueryStringDecoder(uri); uriDecoder.parameters().entrySet().stream().forEach(param-> servletRequest.addParameter(param.getKey(), param.getValue().get(0))); } else if(HttpMethod.POST.name().equalsIgnoreCase(method)) { Map<String, String> contentMap = new Gson().fromJson(contentStr, new TypeToken<Map<String, String>>(){}.getType()); servletRequest.addParameters(contentMap); } channelHandlerContext.fireChannelRead(servletRequest); } |
处理实际的业务逻辑
DispatcherServlet内部进行解析, 调用HandlerMapping寻找处理器,找到对应的处理器Controller后执行并返回处理结果;
假设管道配置了聚合http片段时,入站处理器需要继承
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 | protected void channelRead0(ChannelHandlerContext ctx, MockHttpServletRequest servletRequest) throws Exception { long startTime = System.currentTimeMillis(); MockHttpServletResponse servletResponse = new MockHttpServletResponse(); dispatcherServlet.service(servletRequest, servletResponse); String respContent = servletResponse.getContentAsString(); ByteBuf resultBuf = Unpooled.copiedBuffer(respContent, CharsetUtil.UTF_8); HttpResponseStatus status = HttpResponseStatus.OK; FullHttpResponse response = new DefaultFullHttpResponse(HttpVersion.HTTP_1_1, status, resultBuf); // 添加响应头信息 HttpHeaders headers = response.headers(); headers.add(HttpHeaderNames.CONTENT_TYPE, HttpHeaderValues.APPLICATION_JSON + "; charset=UTF-8"); headers.add(HttpHeaderNames.CONTENT_LENGTH, resultBuf.readableBytes()); ctx.write(response); log.info("================= REQUEST END, cost {} ms ================= ", System.currentTimeMillis() - startTime); } |
启动类
1 2 3 4 | public static void main(String[] args) { ApplicationContext applicationContext = SpringApplication.run(ServerApplication.class, args); applicationContext.getBean(HttpServer.class).start(); } |
好了,接下来开始启动程序,发现报错,很明显,描述是指缺少了
1 2 3 4 5 | Caused by: org.springframework.context.ApplicationContextException: Unable to start ServletWebServerApplicationContext due to missing ServletWebServerFactory bean. at org.springframework.boot.web.servlet.context.ServletWebServerApplicationContext.getWebServerFactory(ServletWebServerApplicationContext.java:206) at org.springframework.boot.web.servlet.context.ServletWebServerApplicationContext.createWebServer(ServletWebServerApplicationContext.java:180) at org.springframework.boot.web.servlet.context.ServletWebServerApplicationContext.onRefresh(ServletWebServerApplicationContext.java:154) ... 8 common frames omitted |
查看源码,可以发现逻辑走向是这样的,
先获取webServer和servletContext,如果两者都为null,则去获取ServletWebServer工厂,进而得到webServer,否则则开始启动上下文。
如果不忽略tomcat,按正常的逻辑跑下去,一般是获取到

servletWebServerFactory.jpg
现在我们排除了tomcat的依赖,想要结合Netty自己实现一个Servlet上下文,则需要走else if的逻辑,即可得:servletContext不能为null。
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 | public class ServletWebServerApplicationContext extends GenericWebApplicationContext implements ConfigurableWebServerApplicationContext { private void createWebServer() { WebServer webServer = this.webServer; ServletContext servletContext = this.getServletContext(); if (webServer == null && servletContext == null) { // 默认获得Tomcat工厂实现类,用于获取Tomcat作为WebServer ServletWebServerFactory factory = this.getWebServerFactory(); this.webServer = factory.getWebServer(new ServletContextInitializer[]{this.getSelfInitializer()}); } else if (servletContext != null) { try { this.getSelfInitializer().onStartup(servletContext); } catch (ServletException var4) { throw new ApplicationContextException("Cannot initialize servlet context", var4); } } this.initPropertySources(); } } public class ServletWebServerApplicationContext extends GenericWebApplicationContext implements ConfigurableWebServerApplicationContext { protected ServletWebServerFactory getWebServerFactory() { String[] beanNames = this.getBeanFactory().getBeanNamesForType(ServletWebServerFactory.class); if (beanNames.length == 0) { throw new ApplicationContextException("Unable to start ServletWebServerApplicationContext due to missing ServletWebServerFactory bean."); } // ...... } } |
按以上的分析,由于是在
这里可以使用
什么是
SpringApplicationRunListener ?在调用
SpringApplication.run() 的过程中,会进行初始化事件体系,如果发现有自定义广播,则会将其设置成自身的事件广播,否则使用默认的SimpleApplicationEventMulticaster ,代码在org.springframework.context.support.AbstractApplicationContext#initApplicationEventMulticaster 。
SpringApplicationRunListener 接口参与了Springboot的生命周期,开放了基于生命周期各个过程的接口,使得用户方便在指定的生命周期广播相应的事件
创建监听器
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 | public class ServletListener implements SpringApplicationRunListener { public ServletListener(SpringApplication application, String[] args) { super(); } @Override public void contextPrepared(ConfigurableApplicationContext context) { class SelfServletContext extends MockServletContext { @Override public ServletRegistration.Dynamic addServlet(String servletName, Servlet servlet) { return null; } @Override public FilterRegistration.Dynamic addFilter(String filterName, Filter filter) { return null; } } ServletContext servletContext = new SelfServletContext(); ((ServletWebServerApplicationContext) context).setServletContext(servletContext); } } |
spring.factories
1 | org.springframework.boot.SpringApplicationRunListener=cn.binary.jerry.netty.config.ServletListener |
接下来开始启动程序,启动正常,并且调试接口: http://localhost:8082/user/get/jy
发现在最后一个Handler中的
查看源码发现是在以下代码中,未找到ServletConfig这个配置,而
1 2 3 4 5 6 7 8 9 10 11 12 13 14 | public abstract class FrameworkServlet extends HttpServletBean implements ApplicationContextAware { private transient ServletConfig config; public ServletConfig getServletConfig() { return this.config; } private void publishRequestHandledEvent(HttpServletRequest request, HttpServletResponse response, long startTime, @Nullable Throwable failureCause) { if (this.publishEvents && this.webApplicationContext != null) { long processingTime = System.currentTimeMillis() - startTime; // 由于this.getServletConfig() 为null,获取不到servletName,导致报NPT this.webApplicationContext.publishEvent(new ServletRequestHandledEvent(this, request.getRequestURI(), request.getRemoteAddr(), request.getMethod(), this.getServletConfig().getServletName(), WebUtils.getSessionId(request), this.getUsernameForRequest(request), processingTime, failureCause, response.getStatus())); } } } |
既然没有配置,我们就手动帮它配,并且进行初始化
1 2 3 4 5 6 7 8 9 10 11 | public static void main(String[] args) { ApplicationContext applicationContext = SpringApplication.run(ServerApplication.class, args); // 获取DispatcherServlet DispatcherServlet dispatcherServlet = applicationContext.getBean(DispatcherServlet.class); CommonUtils.setDeclaredFieldValue(dispatcherServlet, "config", new MockServletConfig()); try { dispatcherServlet.init(); } catch (ServletException e) { } applicationContext.getBean(HttpServer.class).start(); } |
接下来开始启动程序,启动正常,并且调试接口: http://localhost:8082/user/get/jy ,一切正常了!
注意
在Netty In Action中提到,Server端的程序不建议使用

channelRead_code.jpg
到这里,基于Springboot+Netty实现的API框架已成功搭建好了哈。
具体源码请查看github:
https://github.com/qJerry/Netty-Analyze-Demo/tree/master/Chapter1-3
Ending......
阿黑在下一章节将继续源码剖析!
请关注微信公众号:进击的阿黑,谢谢。