RxJava2 源码分析(二) : subscribeOn()线

2019-05-27  本文已影响0人  ShawZ

前言

上回讲到subscribe()订阅观察者,这回咱们来聊聊subscribeOn()方法,官方解释是:在指定的线程中订阅该被观察者,你肯定知道subscribeOn调用多次只有最上面的那一次是生效的,真的是这样吗?留着这个问题,start read the fucking code~~

1.示例

 Observable.create(...)
                .doOnNext(new Consumer<Integer>() {
                    @Override
                    public void accept(Integer integer) throws Exception {
                        Log.d(TAG, "accept: first Thread = " + Thread.currentThread().getName());
                    }
                })
                .subscribeOn(Schedulers.computation())
                .doOnNext(new Consumer<Integer>() {
                    @Override
                    public void accept(Integer integer) throws Exception {
                        Log.d(TAG, "accept: second Thread = " + Thread.currentThread().getName());
                    }
                })
                .subscribeOn(Schedulers.io())
                .subscribe(new Observer<Integer>() {...}
                });


//日志
D: onSubscribe: Thread = main
D: accept: first Thread = RxComputationThreadPool-2
D: accept: second Thread = RxComputationThreadPool-2
D: onNext: Thread = RxComputationThreadPool-2
D: onComplete: Thread = RxComputationThreadPool-2

我们先不关心just,doOnNext操作符,这里仅仅是用来打印日志,结合上回的分析开始本次源码分析~

public final Observable<T> subscribeOn(Scheduler scheduler) {
    ObjectHelper.requireNonNull(scheduler, "scheduler is null");
    //懂的自然懂~,这里直接返回ObservableSubscribeOn实例
    return RxJavaPlugins.onAssembly(new ObservableSubscribeOn<T>(this, scheduler));
}

public final class ObservableSubscribeOn<T> extends AbstractObservableWithUpstream<T, T> {
    final Scheduler scheduler;

    public ObservableSubscribeOn(ObservableSource<T> source, Scheduler scheduler) {
        //持有上游的Source对象,也就是ObservableCreate被观察者对象
        super(source);
        //持有我们传入的线程调度对象
        this.scheduler = scheduler;
    }
}

2.开始订阅

有了上篇的分析,我们知道这里直接调用的是ObservableSubscribeOn的subscribeActual()方法

    public void subscribeActual(final Observer<? super T> observer) {
        //实例化SubscribeOnObserver观察者对象,传入observer进行持有
        final SubscribeOnObserver<T> parent = new SubscribeOnObserver<T>(observer);
        //打印第一行日志 ---onSubscribe: Thread = main
        //因为当前线程是UI线程,所以是线程名为main
        observer.onSubscribe(parent);
        //这里没有直接订阅,而是先进行了线程切换,才开始继续’向上订阅‘
        parent.setDisposable(scheduler.scheduleDirect(new SubscribeTask(parent)));
    }

    //SubscribeTask是ObservableSubscribeOn的内部类,实现了Runnable接口,注定要在线程里完成使命
    final class SubscribeTask implements Runnable {
        private final SubscribeOnObserver<T> parent;

        SubscribeTask(SubscribeOnObserver<T> parent) {
            //持有我们生成的SubscribeOnObserver对象
            this.parent = parent;
        }

        @Override
        public void run() {
            //source是指‘上游的’ObservableSource对象
            source.subscribe(parent);
        }
    }

    //接着再看scheduler.scheduleDirect()
    public Disposable scheduleDirect(@NonNull Runnable run) {
        return scheduleDirect(run, 0L, TimeUnit.NANOSECONDS);
    }

    public Disposable scheduleDirect(@NonNull Runnable run, long delay, @NonNull TimeUnit unit) {
        //1.createWorker是Scheduler的抽象函数,我们外部传入的是Schedulers.io()对象,即IoScheduler实例
        //这里的Worker是EventLoopWorker类实例,持有线程池之类的,这里不究细节
        final Worker w = createWorker();
        //属性的钩子函数
        final Runnable decoratedRun = RxJavaPlugins.onSchedule(run);
        //DisposeTask实现Runnable,Disposable接口,进行一层封装
        DisposeTask task = new DisposeTask(decoratedRun, w);
        //将我们创建的task传进去
        w.schedule(task, delay, unit);

        return task;
    }

    public Disposable schedule(@NonNull Runnable action, long delayTime, @NonNull TimeUnit unit) {
        //首先判断是否已经被安排了(取消订阅了)
            if (tasks.isDisposed()) {
                // don't schedule, we are unsubscribed
                return EmptyDisposable.INSTANCE;
            }
        //这里有很多线程池,Worker缓存池的东西,我们先不纠这些细节,跳过往下看scheduleActual()
            return threadWorker.scheduleActual(action, delayTime, unit, tasks);
        }

public ScheduledRunnable scheduleActual(final Runnable run, long delayTime, @NonNull TimeUnit unit, @Nullable DisposableContainer parent) {
        Runnable decoratedRun = RxJavaPlugins.onSchedule(run);
        //我也不知道,为什么要将SubscribeTask对象包裹这么多次。。。。。。
        ScheduledRunnable sr = new ScheduledRunnable(decoratedRun, parent);

        if (parent != null) {
            if (!parent.add(sr)) {
                return sr;
            }
        }

        Future<?> f;
        try {
            //延迟默认是0
            if (delayTime <= 0) {
                //最终看到这里,我们的SubscribeTask对象在线程池中终于被安排上了,好累~~
                f = executor.submit((Callable<Object>)sr);
            } else {
                f = executor.schedule((Callable<Object>)sr, delayTime, unit);
            }
            sr.setFuture(f);
        } catch (RejectedExecutionException ex) {
            if (parent != null) {
                parent.remove(sr);
            }
            RxJavaPlugins.onError(ex);
        }

        return sr;
    }

3.总结

subscribeOn的职责

  1. 调用onSubscribe()开始订阅
  2. 切换指定线程线程
  3. 在指定线程中继续向上订阅

后续的向上订阅到发射数据,都是在指定的线程中执行,这样也解释了我们之前打印的日志,在UI线程中订阅,切换到io线程中继续向上订阅,再次切换到Computation线程向上订阅,后续的发射发射数据都是RxComputationThreadPool线程中,也证明了调用多次subscribeOn()只有最上面那个有效。

上一篇下一篇

猜你喜欢

热点阅读