引言
上篇文章中，我们了解了RxJava基本的无背压数据流实现原理，本篇我们依然从案例着手，学习有背压下数据流响应实现。何为背压?大多数情况下，上游发射数据的速度大于下游处理数据的速度，背压策略就是控制数据流速，在RxJava中通过设置下游的处理能力实现“响应式拉取”解决背压问题。
样例
下面是同步订阅带背压的样例:
private void testBackPressure() {
//同步订阅事件，发送一个接收一个，不会出现被观察者发送事件速度 > 观察者接收事件速度的情况。
// 可是，却会出现被观察者发送事件数量 > 观察者接收事件数量的问题
Flowable upstream = Flowable.create(new FlowableOnSubscribe() {
@Override
public void subscribe(FlowableEmitter emitter) throws Exception {
emitter.onNext(1);
emitter.onNext(2);
emitter.onNext(3);
emitter.onComplete();
}
}, BackpressureStrategy.ERROR);//背压策略，下游接收不到数据的时候报MissingBackpressureException异常
Subscriber subscriber = new Subscriber() {
@Override
public void onSubscribe(Subscription s) {
Log.e(TAG, "Flowable:onSubscribe");
//设置下游接收的事件个数,如果不设置
//这里只接收两个事件
//同步订阅的情况下，如果不设置接收能力，也会报背压异常
s.request(2);
}
@Override
public void onNext(Integer integer) {
Log.e(TAG, "Flowable:onNext: " + integer);
}
@Override
public void onError(Throwable t) {
Log.e(TAG, "Flowable:onError: ", t);
}
@Override
public void onComplete() {
Log.e(TAG, "Flowable:onComplete: ");
}
};
upstream.subscribe(subscriber);
}
可见写法和无背压的模式基本一致，而里面出现的元素也和无背压一一对应:
1.Flowable:被观察者,对应Observable;
2.FlowableOnSubscribe:中间件，类似于“管道”，用于发射数据，对应ObservableOnSubscribe;
- FlowableEmitter:中间件，类似于“泵”，用于真正发射数据，对应ObservableEmitter;
- Subscription:数据流管道开关，类似于“阀门”,对应于Disposable，它可以控制下游的处理能力;
- Subscriber:被观察者，对应Observer。
好了找到对应关系，我们猜想数据流的流向也是和无背压基本一致，事实上也是如此，因此本篇不再重点说明流向，而侧重点放到“响应式拉取”的实现上。由于和无背压的流程相似,不详细阐述他们之间的关系，不熟悉的请看上篇博客。下面先大概看看各元素的构成:
Flowable
public abstract class Flowable implements Publisher {
....
}
public interface Publisher {
/**
* Request {@link Publisher} to start streaming data.
*
* This is a "factory method" and can be called multiple times, each time starting a new {@link Subscription}.
*
* Each {@link Subscription} will work for only a single {@link Subscriber}.
*
* A {@link Subscriber} should only subscribe once to a single {@link Publisher}.
*
* If the {@link Publisher} rejects the subscription attempt or otherwise fails it will
* signal the error via {@link Subscriber#onError}.
*
* @param s the {@link Subscriber} that will consume signals from this {@link Publisher}
*/
public void subscribe(Subscriber s);
}
实现了Publisher接口的subscribe方法:
@BackpressureSupport(BackpressureKind.SPECIAL)
@SchedulerSupport(SchedulerSupport.NONE)
@Override
public final void subscribe(Subscriber s) {
...
try {
s = RxJavaPlugins.onSubscribe(this, s);
....
subscribeActual(s);
} catch (NullPointerException e) { // NOPMD
throw e;
} catch (Throwable e) {
Exceptions.throwIfFatal(e);
// can't call onError because no way to know if a Subscription has been set or not
// can't call onSubscribe because the call might have set a Subscription already
RxJavaPlugins.onError(e);
NullPointerException npe = new NullPointerException("Actually not, but can't throw other exceptions due to RS");
npe.initCause(e);
throw npe;
}
}
依然调用的subscribeActual抽象方法，我们接着看create操作符返回的FlowableCreate,它实现了subscribeActual方法。
FlowableCreate
public final class FlowableCreate extends Flowable {
//用户传入的FlowableOnSubscribe
final FlowableOnSubscribe source;
//背压策略
final BackpressureStrategy backpressure;
public FlowableCreate(FlowableOnSubscribe source, BackpressureStrategy backpressure) {
this.source = source;
this.backpressure = backpressure;
}
@Override
public void subscribeActual(Subscriber t) {
BaseEmitter emitter;
//根据背压策略，构造对应的发射器
switch (backpressure) {
case MISSING: {
emitter = new MissingEmitter(t);
break;
}
case ERROR: {
emitter = new ErrorAsyncEmitter(t);
break;
}
case DROP: {
emitter = new DropAsyncEmitter(t);
break;
}
case LATEST: {
emitter = new LatestAsyncEmitter(t);
break;
}
default: {
emitter = new BufferAsyncEmitter(t, bufferSize());
break;
}
}
//订阅时的观察方法，在使用的时候，需要在这个方法里面设置下游的处理事件个数
t.onSubscribe(emitter);
try {
//通过emitter发射数据，用户实现
source.subscribe(emitter);
} catch (Throwable ex) {
Exceptions.throwIfFatal(ex);
emitter.onError(ex);
}
}
}
和ObservableCreate类似，封装了FlowableOnSubscribe和背压策略，在subscribeActual中根据背压策略和传入的观察者设置不同的发射器，然后通过source.subscribe(emitter)发射数据，在这方法里，emitter封装了观察者Subscriber，最终调用它的接收数据方法。
下面我们着重看看发射器的结构，它们既然和背压策略紧密联系，那么背压功能必然由它们实现。
FlowableEmitter族
FlowableEmitter接口
先看它们的顶层接口:
public interface FlowableEmitter extends Emitter {
...
long requested();
...
}
其他方法和ObservableEmitter方法基本一致，这里多了requested方法，它用于返回当前可以接受事件的个数。另外发射器族还实现了Subscription接口:
public interface Subscription {
//设置下游接收能力，n为事件个数，多次调用时，容量累加
public void request(long n);
//断开管道
public void cancel();
}
所以发射器拥有设置下游接收容量和断开管道的功能。
接下来我们再看看背压数据流发射器的基类BaseEmitter:
BaseEmitter
abstract static class BaseEmitter
extends AtomicLong
implements FlowableEmitter, Subscription {
private static final long serialVersionUID = 7326289992464377023L;
//真正的被观察者
final Subscriber actual;
final SequentialDisposable serial;
BaseEmitter(Subscriber actual) {
this.actual = actual;
this.serial = new SequentialDisposable();
}
@Override
public void onComplete() {
if (isCancelled()) {
return;
}
try {
//观察者onComplete
actual.onComplete();
} finally {
serial.dispose();
}
}
@Override
public void onError(Throwable e) {
if (e == null) {
e = new NullPointerException("onError called with null. Null values are generally not allowed in 2.x operators and sources.");
}
if (isCancelled()) {
RxJavaPlugins.onError(e);
return;
}
try {
//观察者onError
actual.onError(e);
} finally {
serial.dispose();
}
}
@Override
public final void cancel() {
serial.dispose();
onUnsubscribed();
}
void onUnsubscribed() {
// default is no-op
}
@Override
public final boolean isCancelled() {
return serial.isDisposed();
}
//重点：调用层调用这个方法设置处理能力
@Override
public final void request(long n) {
if (SubscriptionHelper.validate(n)) {
//计数器+n
BackpressureHelper.add(this, n);
onRequested();
}
}
void onRequested() {
// default is no-op
}
@Override
public final void setDisposable(Disposable s) {
serial.update(s);
}
@Override
public final void setCancellable(Cancellable c) {
setDisposable(new CancellableDisposable(c));
}
@Override
public final long requested() {
//返回当前可以处理的事件个数
return get();
}
@Override
public final FlowableEmitter serialize() {
return new SerializedEmitter(this);
}
}
继承AtomicLong类，说明发射器是原子操作的Long型数，本质是一个计数器。主要覆写了onError、onComplete等方法,其中request方法是设置Long型数，代表当前发射器的事件容量，调用者在接收数据的onSubscribe方法中必须调用这个方法，设置容量。至于其他方法的实现我们再看他的子类实现。
NoOverflowBaseAsyncEmitter
abstract static class NoOverflowBaseAsyncEmitter extends BaseEmitter {
private static final long serialVersionUID = 4127754106204442833L;
NoOverflowBaseAsyncEmitter(Subscriber actual) {
super(actual);
}
@Override
public final void onNext(T t) {
if (isCancelled()) {
return;
}
if (t == null) {
onError(new NullPointerException("onNext called with null. Null values are generally not allowed in 2.x operators and sources."));
return;
}
//取当前剩余容量
if (get() != 0) {
//被观察者onNext方法
actual.onNext(t);
//容量-1
BackpressureHelper.produced(this, 1);
} else {
//剩余容量为0，超出处理能力交给子类实现
onOverflow();
}
}
//溢出的数据再这个方法中处理，不同的背压策略有不同的实现
abstract void onOverflow();
}
到目前为止，我们知道了request方法是做+计数，onNext方法是做减计数，在同步订阅方法中通过这两点实现响应式拉取，即:我能吃多少就只能吃多少，超过的事件交给onOverflow处理。接下来我们看看具体的几个发射器实现
发射器的实现
MissingEmitter
MissingEmitter对应MISSING策略:
static final class MissingEmitter extends BaseEmitter {
private static final long serialVersionUID = 3776720187248809713L;
MissingEmitter(Subscriber actual) {
super(actual);
}
@Override
public void onNext(T t) {
if (isCancelled()) {
return;
}
if (t != null) {
actual.onNext(t);
} else {
onError(new NullPointerException("onNext called with null. Null values are generally not allowed in 2.x operators and sources."));
return;
}
for (;;) {
long r = get();
//当前计数器为0或者安全-1返回,则退出循环
if (r == 0L || compareAndSet(r, r - 1)) {
return;
}
}
}
}
发现它没有继承NoOverflowBaseAsyncEmitter，所以没有做溢出处理，在onNext方法中检查当前计数器的值，安全减1或者计数值为0时返回，可见它接收到事件能处理就处理，处理不了就睁一只鸭闭一只眼直接返回。
ErrorAsyncEmitter
static final class ErrorAsyncEmitter extends NoOverflowBaseAsyncEmitter {
private static final long serialVersionUID = 338953216916120960L;
ErrorAsyncEmitter(Subscriber actual) {
super(actual);
}
@Override
void onOverflow() {
//事件溢出处理
onError(new MissingBackpressureException("create: could not emit value due to lack of requests"));
}
}
ErrorAsyncEmitter在事件溢出时直接抛出背压异常。
DropAsyncEmitter
static final class DropAsyncEmitter extends NoOverflowBaseAsyncEmitter {
private static final long serialVersionUID = 8360058422307496563L;
DropAsyncEmitter(Subscriber actual) {
super(actual);
}
@Override
void onOverflow() {
//事件溢出，就当没看见
}
}
到目前为止，我们梳理了背压原理，其实本质就是在发射数据的时候设置了计数器，没接收一个事件计数器减一，背压策略就是处理数据溢出的情况。