CompleteableFuture异步编程框架

一、CompleteableFuture

CompleteableFuture是Java 8中的一个类，它提供了一种异步编程的方式。使用CompleteableFuture可以提高程序的并发性和性能。

二、案例引出

2.1 案例代码

1. 目录结构
   

2. 有3个被调用的service(MemberService,OrderService,ProductService)，每个service中有个search方法，分别耗时5,4,3秒钟。
   

3. 测试代码，需要计算查询需求的总体耗时
   

2.2 引出CompleteableFuture异步调用

有一个查询需求，需要分别调用这3个service的search方法，获取对应会员、订单、产品列表。这3个调用的返回结果之间没有依赖关系。

1. 串行调用

    public void query1(){List<Member> members = memberService.search();List<Order> orders = orderService.search();List<Product> products = productService.search();}

耗时为3+4+5=12秒

2. 并行调用–也就是异步调用

    public void query2() throws ExecutionException, InterruptedException {CompletableFuture<List<Member>> memberFuture = CompletableFuture.supplyAsync(() -> memberService.search());CompletableFuture<List<Order>> orderFuture = CompletableFuture.supplyAsync(() -> orderService.search());CompletableFuture<List<Product>> productFuture = CompletableFuture.supplyAsync(() -> productService.search());CompletableFuture.allOf(memberFuture,orderFuture,productFuture).get();}

耗时为max{5,4,3}=5秒

三、CompletableFuture常用方法

1. supplyAsync( )

开启一个异步调用，并返回一个CompletableFuture对象

    public void query2(){CompletableFuture<List<Member>> memberFuture = CompletableFuture.supplyAsync(() -> memberService.search());CompletableFuture<List<Order>> orderFuture = CompletableFuture.supplyAsync(() -> orderService.search());CompletableFuture<List<Product>> productFuture = CompletableFuture.supplyAsync(() -> productService.search());}

2. runAsync( )

    public void query2(){CompletableFuture.runAsync(() -> memberService.search());CompletableFuture.runAsync(() -> orderService.search());CompletableFuture.runAsync(() -> productService.search());}

3. get( )

阻塞式地获取一个CompletableFuture对象的结果，抛出ExecutionException, InterruptedException异常

    public void query2() throws ExecutionException, InterruptedException {CompletableFuture<List<Member>> memberFuture = CompletableFuture.supplyAsync(() -> memberService.search());List<Member> members = memberFuture.get();}

4.链式调用–thenApply( )

返回的结果如果是CompletableFuture，可以继续调用相关api

thenApply( )方法：前面任务执行完后执行当前任务。消费上一次调用的结果，返回另一个结果

    public void query2() throws ExecutionException, InterruptedException {CompletableFuture<List<Order>> future = CompletableFuture.supplyAsync(() -> memberService.search()).thenApply(lastResult -> orderService.search());List<Order> orders = future.get();}

5. thenRun( )

前面任务执行完后执行当前任务。不管上一次调用的结果是什么，没有参数，没有返回值

    public void query2() throws ExecutionException, InterruptedException {CompletableFuture.supplyAsync(() -> memberService.search()).thenRun(() -> orderService.search());}

6. thenAccept( )

前面任务执行完后执行当前任务。消费上一次调用的结果，没有返回值

    public void query2() throws ExecutionException, InterruptedException {CompletableFuture.supplyAsync(() -> memberService.search()).thenAccept(lastResult -> orderService.search());}

7. anyOf( )

将多个future对象合并，返回最早运行结束的对象

    public void query2() throws ExecutionException, InterruptedException {CompletableFuture<List<Member>> memberFuture = CompletableFuture.supplyAsync(() -> memberService.search());CompletableFuture<List<Order>> orderFuture = CompletableFuture.supplyAsync(() -> orderService.search());CompletableFuture<List<Product>> productFuture = CompletableFuture.supplyAsync(() -> productService.search());CompletableFuture<Object> future = CompletableFuture.anyOf(memberFuture, orderFuture, productFuture);System.out.println(future.get());}

8. allOf( )

等待所有并行任务执行后，再执行后面的操作。
allOf( )：返回CompletableFuture<Void>对象，本身并不会阻塞

示例：等待memberService和orderService都结束后，再执行后面的异步操作。

    public void query2(){long a = System.currentTimeMillis();CompletableFuture<List<Member>> memberFuture = CompletableFuture.supplyAsync(() -> memberService.search());CompletableFuture<List<Order>> orderFuture = CompletableFuture.supplyAsync(() -> orderService.search());CompletableFuture<Void> allFutures = CompletableFuture.allOf(memberFuture, orderFuture);allFutures.thenRun(()->{try {Thread.sleep(3000);} catch (InterruptedException e) {throw new RuntimeException(e);}long b = System.currentTimeMillis();System.out.println("耗时111:"+(b-a));});}