rxjava 操作符详解

Observable的 combineLatest

只看上面的图片不知道操作符是什么意思？下面通过源码分析这个操作符的意思

Observable的 combineLatest最终都会返回ObservableCombineLatest对象。看一下他的subscribeActual

public void subscribeActual(Observer<? super R> observer) {
    ObservableSource<? extends T>[] sources = this.sources;
    int count = 0;
    if (sources == null) {
        //sourcesIterable转化为sources
        sources = new Observable[8];
        for (ObservableSource<? extends T> p : sourcesIterable) {
            if (count == sources.length) {
                ObservableSource<? extends T>[] b = new ObservableSource[count + (count >> 2)];
                System.arraycopy(sources, 0, b, 0, count);
                sources = b;
            }
            sources[count++] = p;
        }
    } else {
        count = sources.length;
    }

    if (count == 0) {
        EmptyDisposable.complete(observer);
        return;
    }

    LatestCoordinator<T, R> lc = new LatestCoordinator<T, R>(observer, combiner, count, bufferSize, delayError);
    lc.subscribe(sources);
}

上面只是简单获取了他的ObservableSource个数 然后调用LatestCoordinator的subscribe方法

static final class LatestCoordinator<T, R> extends AtomicInteger implements Disposable {
    private static final long serialVersionUID = 8567835998786448817L;
    final Observer<? super R> downstream;
    final Function<? super Object[], ? extends R> combiner;
    final CombinerObserver<T, R>[] observers;
    Object[] latest;
    final SpscLinkedArrayQueue<Object[]> queue;
    final boolean delayError;

    volatile boolean cancelled;

    volatile boolean done;

    final AtomicThrowable errors = new AtomicThrowable();

    int active;
    int complete;

    @SuppressWarnings("unchecked")
    LatestCoordinator(Observer<? super R> actual,
            Function<? super Object[], ? extends R> combiner,
            int count, int bufferSize, boolean delayError) {
        this.downstream = actual;
        this.combiner = combiner;
        this.delayError = delayError;
        this.latest = new Object[count];
        //根据ObservableSource的数量初始化相同数量的CombinerObserver
        CombinerObserver<T, R>[] as = new CombinerObserver[count];
        for (int i = 0; i < count; i++) {
            as[i] = new CombinerObserver<T, R>(this, i);
        }
        this.observers = as;
        this.queue = new SpscLinkedArrayQueue<Object[]>(bufferSize);
    }

    public void subscribe(ObservableSource<? extends T>[] sources) {
        Observer<T>[] as = observers;
        int len = as.length;
        //调用下游的onSubscribe
        downstream.onSubscribe(this);
        //开始订阅每个ObservableSource
        for (int i = 0; i < len; i++) {
            if (done || cancelled) {
                return;
            }
            sources[i].subscribe(as[i]);
        }
    }
}

上面的代码订阅了声明的多个ObservableSource，这里直接新建了CombinerObserver来处理订阅结果。注意每个CombinerObserver和ObservableSource是一一绑定的。
现在看一下CombinerObserver是怎么处理的

static final class CombinerObserver<T, R> extends AtomicReference<Disposable> implements Observer<T> {
        private static final long serialVersionUID = -4823716997131257941L;

        final LatestCoordinator<T, R> parent;

        final int index;

        CombinerObserver(LatestCoordinator<T, R> parent, int index) {
            this.parent = parent;
            this.index = index;
        }

        @Override
        public void onSubscribe(Disposable d) {
            DisposableHelper.setOnce(this, d);
        }

        //
        @Override
        public void onNext(T t) {
            parent.innerNext(index, t);
        }

        @Override
        public void onError(Throwable t) {
            parent.innerError(index, t);
        }

        @Override
        public void onComplete() {
            parent.innerComplete(index);
        }

        public void dispose() {
            DisposableHelper.dispose(this);
        }
    }

上面的代码直接回调了订阅结果，看一下LatestCoordinator的innerNext

void innerNext(int index, T item) {
    boolean shouldDrain = false;
    synchronized (this) {
        //这里存储每个ObservableSource数组onnext返回的结果，每个object对应一个ObservableSource
        Object[] latest = this.latest;
        if (latest == null) {
            return;
        }
        //active表示ObservableSource数组中有几个执行完了
        Object o = latest[index];
        int a = active;
        if (o == null) {
            active = ++a;
        }
        //每次有最新值都替换保存
        latest[index] = item;
        //数组中的ObservableSource都执行完了
        if (a == latest.length) {
            queue.offer(latest.clone());
            shouldDrain = true;
        }
    }
    if (shouldDrain) {
        drain();
    }
}

latest的长度和ObservableSource数组的长度一样 ，对应保存每个ObservableSource的返回值。上面的代码是当所有ObservableSource都有返回值了才会执行下一步

void drain() {
    if (getAndIncrement() != 0) {
        return;
    }

    final SpscLinkedArrayQueue<Object[]> q = queue;
    final Observer<? super R> a = downstream;
    final boolean delayError = this.delayError;

    int missed = 1;
    for (;;) {

        for (;;) {
            if (cancelled) {
                clear(q);
                return;
            }

            //delayError==false 的时候 只要出错直接返回
            if (!delayError && errors.get() != null) {
                cancelSources();
                clear(q);
                a.onError(errors.terminate());
                return;
            }

            boolean d = done;
            //结果数组 这里是都不为null
            Object[] s = q.poll();
            boolean empty = s == null;

            if (d && empty) {
                clear(q);
                Throwable ex = errors.terminate();
                if (ex == null) {
                    a.onComplete();
                } else {
                    a.onError(ex);
                }
                return;
            }

            if (empty) {
                break;
            }

            R v;

            try {
                //调用传进来的Function方法
                v = ObjectHelper.requireNonNull(combiner.apply(s), "The combiner returned a null value");
            } catch (Throwable ex) {
                Exceptions.throwIfFatal(ex);
                errors.addThrowable(ex);
                cancelSources();
                clear(q);
                ex = errors.terminate();
                a.onError(ex);
                return;
            }
            
            //返回结果给下游。
            a.onNext(v);
        }

        missed = addAndGet(-missed);
        if (missed == 0) {
            break;
        }
    }
}

从上面可知道 当拿到所有ObservableSource返回的值之后， 直接调用Function方法获取结果，再把结果传递给下游。

自此，就知道这个操作符的作用是，等待所有的ObservableSource返回值，当每个ObservableSource都有返回值之后，在拿这些所有值进行操作

Observable
    .combineLatest(Observable.just(1, 2, 3), Observable.intervalRange(1, 3, 2, 2, TimeUnit.SECONDS), new BiFunction<Integer, Long, Long>() {
        @Override
        public Long apply(Integer first, Long second) throws Exception {
            return first * second;
        }
    })
    .subscribe(new Consumer<Long>() {
        @Override
        public void accept(Long value) throws Exception {
            //pring value
        }
    });

因此 上面的结果就是 3 6 9 ，由于第二个ObservableSource延迟执行 第一个ObservableSource，一直都是3.

---