为了支持异步处理,在Servlet 3.0中,在ServletRequest上提供了startAsync()方法: AsyncContext startAsync() throws Java.lang.IllegalStateException; AsyncContext startAsync(ServletRequest servletRequest, ServletResponse servletResponse) throws java.lang.IllegalStateException
这两个方法都会返回AsyncContext接口的实现对象,前者会直接利用原有的请求与响应对象来创建AsyncContext,后者可以传入自行创建的请求、响应封装对象。在调用了startAsync()方法取得AsyncContext对象之后,此次请求的响应会被延后,并释放容器分配的线程。
可以通过AsyncContext的getRequest()、getResponse()方法取得请求、响应对象,此次对客户端的响应将暂缓至调用AsyncContext的complete()或dispatch()方法为止,前者表示响应完成,后者表示将调派指定的URL进行响应。
若要能调用ServletRequest的startAsync()以取得AsyncContext,必须告知容器此Servlet支持异步处理,如果使用@WebServlet来标注,则可以设置其asyncSupported为true。例如: @WebServlet(urlPatterns = "/some.do", asyncSupported = true) public class AsyncServlet extends HttpServlet { ... 如果使用web.xml设置Servlet,则可以在<servlet>中设置<async-supported>标签为true: ... < servlet> <servlet-name>AsyncServlet</servlet-name> <servlet-class>cc.openhome.AsyncServlet</servlet-class> <async-supported>true</async-supported> < /servlet> ...
如果Servlet将会进行异步处理,若其前端有过滤器,则过滤器亦需标示其支持异步处理,如果使用@WebFilter,同样可以设置其asyncSupported为true。例如: @WebFilter(urlPatterns = "/some.do", asyncSupported = true) public class AsyncFilter implements Filter{ ...
如果使用web.xml设置过滤器,则可以设置<async-supported>标签为true: ... < filter> <filter-name>AsyncFilter</filter-name> <filter-class>cc.openhome.AsyncFilter</filter-class> <async-supported>true</async-supported> < /filter> ...
下面示范一个异步处理的简单例子: AsyncContextDemo AsyncServlet.java package cc.openhome; import java.io.*; import java.util.concurrent.*; import javax.servlet.*; import javax.servlet.annotation.*; import javax.servlet.http.*; @WebServlet(name="AsyncServlet", urlPatterns={"/async.do"}, asyncSupported = true) public class AsyncServlet extends HttpServlet { private ExecutorService executorService = Executors.newFixedThreadPool(10); @Override protected void doGet(HttpServletRequest request, HttpServletResponse response) throws ServletException, IOException { response.setContentType("text/html; charset=UTF8"); AsyncContext ctx = request.startAsync(); executorService.submit(new AsyncRequest(ctx)); } @Override public void destroy() { executorService.shutdown(); } }
首先告诉容器,这个Servlet支持异步处理?,对于每个请求,Servlet会取得其AsyncContext?,并释放容器所分配的线程,响应被延后。对于这些被延后响应的请求,创建一个实现Runnable接口的AsyncRequest对象,并将其调度一个线程池(Thread pool)?,线程池的线程数量是固定的,让这些必须长时间处理的请求,在这些有限数量的线程中完成,而不用每次请求都占用容器分配的线程。
AsyncRequest是个实现Runnable的类,其模拟了长时间处理: AsyncContextDemo AsyncRequest.java package cc.openhome; import java.io.PrintWriter; import javax.servlet.AsyncContext; public class AsyncRequest implements Runnable { private AsyncContext ctx; public AsyncRequest(AsyncContext ctx) { this.ctx = ctx; } @Override public void run() { try { Thread.sleep(10000); PrintWriter out = ctx.getResponse().getWriter(); out.println("久等了...XD"); ctx.complete(); } catch (Exception e) { throw new RuntimeException(e); } } }
请求与响应对象都封装在AsyncContext中,所以AsyncRequest建构时必须接受AsyncContext实例。范例中以暂停线程的方式来模拟长时间处理?,并输出简单的字符串作为响应文字?,最后调用AsyncContext的complete()对客户端完成响应?。