更新
我正在寻找一种方法来控制另一个观察者的流量。例如,让我们有2个单调增加(重要)整数的可观察量:
source : 1----2-2---2--3--3--4----4--5---6----8---9---10--------11------
control : -1----3----------------5-----------6-------9-----------12------
我需要生成一个新的observable,其元素与源完全匹配,但它们的时序由控制可观察的控制方式如下:源值应始终小于或等于控制值。这意味着只有所有源值(大于最近发布的控件)应该等到它们被控件“释放”
source : 1----2-2---2--3--3--4----4--5---6----8---9---10--------11------
control : -1----3----------------5-----------6-------9-----------12------
expected result: -1----2-2--2--3--3-----4-4--5------6-------8-9---------10-11---
请查看以下代码示例:
private static <T, C> Observable<T> combine(Observable<T> source, Observable<C> control, BiFunction<T, C, Boolean> predicate) {
// ???
}
@Test
public void testControl() throws InterruptedException {
Subject<Integer> control = PublishSubject.create();
Observable<Integer> source = Observable.fromArray(1, 2, 2, 2, 3, 3, 4, 4, 5, 6, 8, 10, 11);
Observable<Integer> combined = combine(source, control, (s, c) -> s <= c);
control.subscribe(val -> System.out.println("Control: " + val));
combined.observeOn(Schedulers.io()).subscribe(val -> System.out.println("Value: " + val));
control.onNext(3); // should release 1,2,2,2,3,3
Thread.sleep(1000);
control.onNext(6); // should release 4,4,5,6
Thread.sleep(1000);
control.onNext(11); // should release 8,10,11
Thread.sleep(1000);
}
答案 0 :(得分:1)
由于我没有找到任何优雅的解决方案,我最终自己实现了它。如果我有人可以提出更优雅的解决方案,我将很高兴(在这种情况下,我将不接受这个答案,并接受更好的解决方案)。以下是我的解决方案:
private static <T, C> Observable<T> combine(Observable<T> source, Observable<C> control, BiFunction<T, C, Boolean> predicate) {
return Observable.create(emitter -> {
Queue<T> buffer = new ArrayDeque<>();
AtomicReference<C> lastControl = new AtomicReference<>();
CompletableSubject sourceCompletable = CompletableSubject.create();
CompletableSubject controlCompletable = CompletableSubject.create();
Disposable disposable = new CompositeDisposable(
control.subscribe(
val -> {
lastControl.set(val);
synchronized (buffer) {
while (!buffer.isEmpty() && predicate.apply(buffer.peek(), val)) {
emitter.onNext(buffer.poll());
}
}
},
emitter::onError,
controlCompletable::onComplete),
source.subscribe(
val -> {
C lastControlVal = lastControl.get();
synchronized (buffer) {
if (lastControlVal != null && predicate.apply(val, lastControlVal)) {
emitter.onNext(val);
} else {
buffer.add(val);
}
}
},
emitter::onError,
sourceCompletable::onComplete),
controlCompletable.andThen(sourceCompletable).subscribe(emitter::onComplete));
emitter.setDisposable(disposable);
});
}
答案 1 :(得分:0)
您需要将buffer功能(将为您缓冲项目)与combineLatest功能组合在一起(这将确定何时可以释放缓冲区)。
Observable<Integer> source;
Observable<Integer> control;
Observable<Integer> canRelease = Observable.combineLatest(source, control, (s, c) -> s < c ? s : null).filter(val -> val != null);
Observable<Integer> result = source.buffer(canRelease).flatMap(Observable::from);
生成的Observable将缓冲源项,直到canRelease
中有值。每次源Observable最新项目变得小于控制Observable最新项目时,canRelease
将发出;