[转] CompletableFuture教程

原文地址:https://mp.weixin.qq.com/s/dsANLERHskYhPWlnabnjhg  作者:指北君

1 CompletableFuture的静态方法使用

  CompleteableFuture的静态方法有如下

   之前的文章里面已经讲过suuplyAsync,以及runAsync。我们就直接看其他方法

 delayedExcutor

  delayedExcutor其作用是构建一个延迟执行任务的Excutor,默认使用ForkJoinPool. 也可以使用自定义的Excutor。

一个延迟5秒执行任务的Excutor,默认使用使用ForkJoinPool.commonPool()。
Executor executor = CompletableFuture.delayedExecutor(5l, TimeUnit.SECONDS);

allof和anyof

  allof和anyof 为等待多个CompletableFuture完成之后返回一个CompletableFuture。

  • allof返回无result,

  • anyof返回为最先完成的CompletableFuture。

  可以看如下示例。

CompletableFuture<String> supplyAsync1 = CompletableFuture.supplyAsync(()-> {
    try {Thread.sleep(4 * 1000);} catch (InterruptedException e) {e.printStackTrace();}
    return "supplyAsync1";
});
CompletableFuture<String> supplyAsync2 = CompletableFuture.supplyAsync(() -> {
    try {Thread.sleep(2 * 1000);} catch (InterruptedException e) {e.printStackTrace();}
    return "supplyAsync2";
});
CompletableFuture.anyOf(supplyAsync1,supplyAsync2).thenAccept((str)-> {
    System.out.println(LocalDateTime.now() + " anyOf complete : " + str);
});
CompletableFuture.allOf(supplyAsync1,supplyAsync2).thenAccept((str)-> {
    System.out.println(LocalDateTime.now() + " allOf complete "+ str );
});

  执行结果如下:

start second: 2021-10-24T12:39:40.562001600
2021-10-24T12:39:42.611118800 anyOf complete : supplyAsync2
2021-10-24T12:39:44.611233200 allOf complete null

  failedStage和failedFuture是返回一个已知异常的CompletableFuture。这个下面和其他异常一起举例。

2 CompletableFuture的其余方法使用

  CompletableFuture中方法可以大致分为run,apply,accept几个类别。其对应的参数分别为Runnable,Function,Consummer等几个函数式表达式。

  1. run代表当前CompletableFuture完成后执行的一些列操作,无输入参数,无返回结果,所以只是Runnable为参数。()-> { option }

  2. apply代表以当前CompletableFuture完成后的结果为参数进行的操作,并且会返回一个新的CompletableFuture,所以以Function为参数。(s)-> {return s;}

  3. accept代表以当前CompletableFuture完成后的结果为参数,执行的操作,无返回结果,直接消费。以Consumer为参数,(s)-> { option }。

2.1 Run方法

  Run方法相关参数为Runnable,为直接执行的操作。

  thenRun 完成之后直接执行。

  thenRunAsync 使用线程池异步执行,线程池默认为ForkJoinPool.commonPool

  runAfterBoth/ runAfterEither 两个CompletableFuture同时完成或者某一个完成就执行的操作。

  runAfterBothAsync/runAfterEitherAsync 同理为使用线程池异步执行的操作。

public class CompletableFutureThenRun {
    public static void main(String[] args) {
        System.out.println(" CompletableFutureThenRun main start : " + LocalDateTime.now());
        CompletableFuture<String> cf1 = CompletableFuture.supplyAsync(() -> {
            System.out.println(" CompletableFutureThenRun  cf1: " + LocalDateTime.now());
            try {
                Thread.sleep(5 * 1000);
            } catch (InterruptedException e) {
                e.printStackTrace();
            }
            return "supplyAsync";
        });

        CompletableFuture<Void> cf2 = CompletableFuture.runAsync(() -> {
            System.out.println(" CompletableFutureThenRun  cf2: " + LocalDateTime.now());
            try {
                Thread.sleep(2 * 1000);
            } catch (InterruptedException e) {
                e.printStackTrace();
            }
        });

        cf1.runAfterBoth(cf2,()-> {
            System.out.println(Thread.currentThread().getName()+" CompletableFutureThenRun runAfterBoth: " + LocalDateTime.now());
        });
        cf1.runAfterBothAsync(cf2,()-> {
            System.out.println( Thread.currentThread().getName()+" CompletableFutureThenRun runAfterBothAsync: " + LocalDateTime.now());
        });
        cf1.runAfterEither(cf2,()-> {
            System.out.println( Thread.currentThread().getName()+" CompletableFutureThenRun runAfterEither: " + LocalDateTime.now());
        });
        cf1.runAfterEitherAsync(cf2,()-> {
            System.out.println( Thread.currentThread().getName()+" CompletableFutureThenRun runAfterEitherAsync: " + LocalDateTime.now());
        });
        cf1.thenRunAsync(()-> {
            System.out.println(Thread.currentThread().getName()+" CompletableFutureThenRun thenRunAsync: " + LocalDateTime.now());
        });
        cf1.thenRun(()-> {
            System.out.println(Thread.currentThread().getName()+" CompletableFutureThenRun thenRun: " + LocalDateTime.now());
        });
        System.out.println(Thread.currentThread().getName() + " CompletableFutureThenRun  last: " + LocalDateTime.now());

        try {
            Thread.sleep(10*1000);
        } catch (InterruptedException e) {
            e.printStackTrace();
        }
    }
}

  上述执行结果:

CompletableFutureThenRun main start : 2021-10-25T01:48:52.416000900
CompletableFutureThenRun  cf1: 2021-10-25T01:48:52.492008500
CompletableFutureThenRun  cf2: 2021-10-25T01:48:52.493008600
main CompletableFutureThenRun  last: 2021-10-25T01:48:52.495008800
ForkJoinPool.commonPool-worker-7 CompletableFutureThenRun runAfterEitherAsync: 2021-10-25T01:48:54.495208800
ForkJoinPool.commonPool-worker-3 CompletableFutureThenRun runAfterEither: 2021-10-25T01:48:54.495208800
ForkJoinPool.commonPool-worker-5 CompletableFutureThenRun thenRun: 2021-10-25T01:48:57.493508600
ForkJoinPool.commonPool-worker-3 CompletableFutureThenRun thenRunAsync: 2021-10-25T01:48:57.494508700
ForkJoinPool.commonPool-worker-3 CompletableFutureThenRun runAfterBoth: 2021-10-25T01:48:57.494508700
ForkJoinPool.commonPool-worker-3 CompletableFutureThenRun runAfterBothAsync: 2021-10-25T01:48:57.495508800

  apply 与accept相关的方法类似,此处不一一举例了。

  下面我们根据一些情景举例来说明方法如何使用:

2.2 多个 CompletableFuture组合在一起执行

  情景一:先去取快递,然后再去买菜,然后回家做饭。

CompletableFuture<String> cf = CompletableFuture.supplyAsync(()-> {
    System.out.println(LocalDateTime.now() + " 正在取快递! ");
    try {Thread.sleep(2 * 1000);} catch (InterruptedException e) {e.printStackTrace();}
    return "快递1";
}).thenApply((str) -> {
    System.out.println(LocalDateTime.now() + " 拿到了: "+str);
    System.out.println(LocalDateTime.now() + " 买菜中。。。 ");
    try {Thread.sleep(2 * 1000);} catch (InterruptedException e) {e.printStackTrace();}
    return str + " 和 蔬菜";
}).thenApply((str2)-> {
    System.out.println(LocalDateTime.now() + " 现在有了: ["+str2+"]");
    try {Thread.sleep(2 * 1000);} catch (InterruptedException e) {e.printStackTrace();}
    return "带着 [" + str2 + " ]回家做饭" ;
});
System.out.println( LocalDateTime.now() + " 美好的一天: "+ cf.join());

  下面看一下上面的执行结果,

2021-10-25T01:10:16.831465600 正在取快递! 
2021-10-25T01:10:18.861668600 拿到了: 快递1
2021-10-25T01:10:18.911673600 买菜中。。。 
2021-10-25T01:10:20.911873600 现在有了: [快递1 和 蔬菜]
2021-10-25T01:10:16.831465600 美好的一天: 带着 [快递1 和 蔬菜 ]回家做饭

  可以看到最后一行输出的时间比较早,这是因为join会阻塞线程,直到此CompletableFuture执行完并获取到值。

  情景二:和女朋友一起出门,我去取快递,女朋友去买菜,然后一起回家做饭。

CompletableFuture<String> cf1 = CompletableFuture.supplyAsync(()->{
    System.out.println(LocalDateTime.now() + " 正在取快递! ");
    try {Thread.sleep(2 * 1000);} catch (InterruptedException e) {e.printStackTrace();}
    return "快递";
});
CompletableFuture<String> cf2 = CompletableFuture.supplyAsync(()->{
    System.out.println(LocalDateTime.now() + " 女朋友正在买菜! ");
    try {Thread.sleep(4 * 1000);} catch (InterruptedException e) {e.printStackTrace();}
    return "蔬菜";
});

cf1.thenAcceptBoth(cf2,(str1 ,str2 )->{
    System.out.println(LocalDateTime.now() +  " ["+ str1 + "]["+str2+"] 带回来了,开始做饭 ");
}).join();

  此处使用 thenAcceptBoth 需要在两个CompletableFuture都完成的情况下,才能执行,所以最后使用join()使其阻塞到可以执行当前的操作。

 

  情景三:和女朋友一起出门,我去取快递,女朋友去买菜,谁先弄完谁就先回去。

cf1.acceptEither(cf2,(str1 )-> {
    System.out.println(LocalDateTime.now() +  " ["+ str1 +"] 带回来了,先回家吧! ");
}).join();

  我先拿到了快递,就快快的回家了,然后就挨了一顿毒打。

2.3 在两个CompletableFuture运行后再次计算

  晚饭过后和女朋友讨论做什么事情,然而发生了分歧:

CompletableFuture<List<String>> cf1 = CompletableFuture.supplyAsync(()->{
    List<String> strings = Arrays.asList("看电影", "打扑克");
    System.out.println(LocalDateTime.now() + " 晚饭后女朋友说,想要: " + strings);
    try {Thread.sleep(2 * 1000);} catch (InterruptedException e) {e.printStackTrace();}
    return strings;
});
CompletableFuture<List<String>> cf2 = CompletableFuture.supplyAsync(()->{
    List<String> strings = Arrays.asList("看电影", "打游戏");
    System.out.println(LocalDateTime.now() + " 晚饭后,我想: " + strings);
    try {Thread.sleep(4 * 1000);} catch (InterruptedException e) {e.printStackTrace();}
    return strings;
});

cf1.thenCombine(cf2,(list1,list2) -> {
        System.out.println("遭受了一顿毒打之后。。。!!!");
    List<String> collect = list1.stream().filter(str -> list2.contains(str)).collect(Collectors.toList());
    System.out.println(LocalDateTime.now() + " 综合两个人的想法,最终决定: " + collect);
    return collect;
}).join();

  女朋友想看电影,或者打扑克,但是我想打游戏。最后遭受一顿毒打之后,还是说出了或者看电影。最终选择了看电影。



4 CompletableFuture的异常处理

  CompletableFuture和异常相关的方法有如下

4.1 whenComplete/whenCompleteAsync

CompletableFuture<T> whenComplete(BiConsumer<? super T, ? super Throwable> action)

  whenCompletable使用有BiConsumer里面会有两个参数,下边是一个示例。参数需要两个分别为str,exception, 如果有异常exception有值,str为null。如果stringCompletableFuture正常完成,则exception为null。但是不管是否有异常,表达式里面的方法均会执行。

  有点类似try finally{},有没有异常均可执行。

CompletableFuture<String> whenCompleteCF = stringCompletableFuture.whenComplete((str, exception) -> {
     if(exception != null){
         System.out.println("whenComplete : " + exception);
         exception.printStackTrace();
     }
     System.out.println("whenComplete execute whither error throw ");
});

4.2 exceptionally

  exceptionally方法中为一个Function参数,需要一个输入值,为当前CompletableFuture抛出的异常。

  其返回值有两个结果:

  1. 如果当前CompletableFuture无异常完成,则返回与原CompletableFuture的result相同的CompletableFuture,注意知识result相同,并不是同一个类。

  2. 如果当前CompletableFuture有异常抛出,那么返回新的CompletableFuture以及新处理后的result。

        CompletableFuture<String> stringCompletableFuture = CompletableFuture.supplyAsync(() -> {
            throw new RuntimeException(" CompletableFuture throw one exception");
//            return "cc";
        });

        CompletableFuture<String> exceptionally = stringCompletableFuture.exceptionally((exception) -> {
            System.out.println("exceptionally only execute  when error throw ");
            return "exception";
        });
         System.out.println("exceptionally  : " + exceptionally.join());

  上述示例无异常抛出时结果如下:

exceptionally  :cc

  有异常抛出时结果如下:

exceptionally only execute  when error throw 
exceptionally  :exception

4.3 handle/handleAsync

public <U> CompletableFuture<U> handle(BiFunction<? super T, Throwable, ? extends U> fn)

  handle 和 whenComplete 比较类似,无论有没有异常,里面的方法均会执行到。但是有有一些区别,handle参数为BiFunction,有返回值,whenComplet的参数为BiComsumer 无返回值。

  下面的实例中,如果有异常则参数中的str为null,如果没有异常exception为null。

CompletableFuture<String> handle = stringCompletableFuture.handle((str, exception) -> {
    System.out.println("handle : " + str);
    if(exception != null ){
        System.out.println("stringCompletableFuture1 have exception :" );
        exception.printStackTrace();
    }
    return "handle complete ";
});

  有异常的执行结果:

stringCompletableFuture1 have exception :
handle.join(); :handle complete

  无异常的执行结果

handle :cc
handle.join(); :handle complete

4.4 failedStage/failedFuture

  failedStage和failedFuture均为静态方法,会返回一个已完成的给定异常的CompletableFuture。

  failedStage返回的是CompletionStage,failedFuture返回为CompletableFuture对象

CompletionStage<Object> test_exception = CompletableFuture.failedStage(new RuntimeException("test exception"));
CompletableFuture<Object> test_exception1 = CompletableFuture.failedFuture(new RuntimeException("test exception"));

  最后给一个可以直接食用的示例,可以根据不同的需求进行改良哦!

public static Map<String, List<Integer>> testMap = new ConcurrentHashMap<>();

static {
    testMap.put("A", Arrays.asList(1,2,3,4,5));
    testMap.put("B", Arrays.asList(6,7,8,9,10));
    testMap.put("C", Arrays.asList(11,12,13,14,15));
    testMap.put("D", Arrays.asList(21,22,23,24,25));
    testMap.put("E", Arrays.asList(31,32,33,34,35));
}

public static void main(String[] args) {
    System.out.println(" CompletableFutureDemo5 main start : " + LocalDateTime.now());
    List<String> strings = Arrays.asList("A", "B", "C", "D", "E");

    ExecutorService testPool = new ForkJoinPool(4);
    List<CompletableFuture<List<Integer>>> collect = strings.stream().map(
            key -> CompletableFuture.supplyAsync(() -> {
                return obtainTheList(key);
            },testPool).exceptionally((exc)->{
                System.out.println(" hit  the exception " );
                throw new RuntimeException(exc);
            })
    ).collect(Collectors.toList());

    System.out.println(" CompletableFutureDemo5 supplyAsync end : " + LocalDateTime.now());
    try {
        List<List<Integer>> integerCollect = collect.stream().map(CompletableFuture::join).collect(Collectors.toList());
    }catch (Exception e){
        System.out.println(" catch  the exception " + e.getMessage());
        e.printStackTrace();
    }
    System.out.println(" CompletableFutureDemo5 main end : " + LocalDateTime.now());
    try {
        Thread.sleep(5*1000);
    } catch (InterruptedException e) {
        e.printStackTrace();
    }

}

private static List<Integer> obtainTheList(String key) {
    List<Integer> integers = testMap.get(key);
    if( key.equals("C") ){
        throw new RuntimeException("exception test !");
    }
    try {
        Thread.sleep(2*1000);
        System.out.println(" obtainTheList thread name : " + Thread.currentThread().getName() +" : "+ LocalDateTime.now());
    } catch (InterruptedException e) {
        e.printStackTrace();
    }
    return integers==null? new ArrayList() :integers;
}

总结

  本片用了一些示例来讲解CompletableFuture,我们可以在开发中的一些场景中使用起来了。特别是异步多线程去拿一些数据的时候,非常好用哦。

  详细的Demo,请看传送门:https://github.com/javatechnorth/java-study-note/tree/master/multiThread/src/main/java/org/javanorth/currency/completableFuture

  作者:指北君,操千曲而后晓声,观千剑而后识器。



原文地址:https://www.cnblogs.com/huanshilang/p/15573401.html