聊聊reactive streams的Mono及Flux

  • Post author:
  • Post category:其他



本文主要讲一下reactive streams的Publisher接口的两个抽象类Mono与Flux

Publisher

reactive-streams-1.0.1-sources.jar!/org/reactivestreams/Publisher.java

/**
 * A {@link Publisher} is a provider of a potentially unbounded number of sequenced elements, publishing them according to
 * the demand received from its {@link Subscriber}(s).
 * <p>
 * A {@link Publisher} can serve multiple {@link Subscriber}s subscribed {@link #subscribe(Subscriber)} dynamically
 * at various points in time.
 *
 * @param <T> the type of element signaled.
 */
public interface Publisher<T> {

    /**
     * Request {@link Publisher} to start streaming data.
     * <p>
     * This is a "factory method" and can be called multiple times, each time starting a new {@link Subscription}.
     * <p>
     * Each {@link Subscription} will work for only a single {@link Subscriber}.
     * <p>
     * A {@link Subscriber} should only subscribe once to a single {@link Publisher}.
     * <p>
     * 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<? super T> s);
}

Mono

reactor-core-3.1.2.RELEASE-sources.jar!/reactor/core/publisher/Mono.java

public abstract class Mono<T> implements Publisher<T> {
    //...
    /**
     * Expose the specified {@link Publisher} with the {@link Mono} API, and ensure it will emit 0 or 1 item.
     * The source emitter will be cancelled on the first `onNext`.
     * <p>
     * <img class="marble" src="https://raw.githubusercontent.com/reactor/reactor-core/v3.1.1.RELEASE/src/docs/marble/from1.png" alt="">
     * <p>
     * @param source the {@link Publisher} source
     * @param <T> the source type
     *
     * @return the next item emitted as a {@link Mono}
     */
    public static <T> Mono<T> from(Publisher<? extends T> source) {
        if (source instanceof Mono) {
            @SuppressWarnings("unchecked")
            Mono<T> casted = (Mono<T>) source;
            return casted;
        }
        if (source instanceof Flux) {
            @SuppressWarnings("unchecked")
            Flux<T> casted = (Flux<T>) source;
            return casted.next();
        }
        return onAssembly(new MonoFromPublisher<>(source));
    }

    /**
     * Create a new {@link Mono} that emits the specified item, which is captured at
     * instantiation time.
     *
     * <p>
     * <img class="marble" src="https://raw.githubusercontent.com/reactor/reactor-core/v3.1.1.RELEASE/src/docs/marble/just.png" alt="">
     * <p>
     * @param data the only item to onNext
     * @param <T> the type of the produced item
     *
     * @return a {@link Mono}.
     */
    public static <T> Mono<T> just(T data) {
        return onAssembly(new MonoJust<>(data));
    }
    //...
}

Flux

reactor-core-3.1.2.RELEASE-sources.jar!/reactor/core/publisher/Flux.java

public abstract class Flux<T> implements Publisher<T> {
    //......
    /**
     * Programmatically create a {@link Flux} with the capability of emitting multiple
     * elements in a synchronous or asynchronous manner through the {@link FluxSink} API.
     * <p>
     * This Flux factory is useful if one wants to adapt some other multi-valued async API
     * and not worry about cancellation and backpressure (which is handled by buffering
     * all signals if the downstream can't keep up).
     * <p>
     * For example:
     *
     * <pre><code>
     * Flux.&lt;String&gt;create(emitter -&gt; {
     *
     *     ActionListener al = e -&gt; {
     *         emitter.next(textField.getText());
     *     };
     *     // without cleanup support:
     *
     *     button.addActionListener(al);
     *
     *     // with cleanup support:
     *
     *     button.addActionListener(al);
     *     emitter.onDispose(() -> {
     *         button.removeListener(al);
     *     });
     * }, FluxSink.OverflowStrategy.LATEST);
     * </code></pre>
     *
     * @param <T> The type of values in the sequence
     * @param backpressure the backpressure mode, see {@link OverflowStrategy} for the
     * available backpressure modes
     * @param emitter Consume the {@link FluxSink} provided per-subscriber by Reactor to generate signals.
     * @return a {@link Flux}
     */
    public static <T> Flux<T> create(Consumer<? super FluxSink<T>> emitter, OverflowStrategy backpressure) {
        return onAssembly(new FluxCreate<>(emitter, backpressure, FluxCreate.CreateMode.PUSH_PULL));
    }

    /**
     * Decorate the specified {@link Publisher} with the {@link Flux} API.
     * <p>
     * <img class="marble" src="https://raw.githubusercontent.com/reactor/reactor-core/v3.1.1.RELEASE/src/docs/marble/from.png" alt="">
     * <p>
     * @param source the source to decorate
     * @param <T> The type of values in both source and output sequences
     *
     * @return a new {@link Flux}
     */
    public static <T> Flux<T> from(Publisher<? extends T> source) {
        if (source instanceof Flux) {
            @SuppressWarnings("unchecked")
            Flux<T> casted = (Flux<T>) source;
            return casted;
        }

        if (source instanceof Fuseable.ScalarCallable) {
            try {
                @SuppressWarnings("unchecked") T t =
                        ((Fuseable.ScalarCallable<T>) source).call();
                if (t != null) {
                    return just(t);
                }
                return empty();
            }
            catch (Exception e) {
                return error(e);
            }
        }
        return wrap(source);
    }

    /**
     * Programmatically create a {@link Flux} by generating signals one-by-one via a
     * consumer callback and some state, with a final cleanup callback. The
     * {@code stateSupplier} may return {@literal null} but your cleanup {@code stateConsumer}
     * will need to handle the null case.
     * <p>
     * <img class="marble" src="https://raw.githubusercontent.com/reactor/reactor-core/v3.1.1.RELEASE/src/docs/marble/generate.png" alt="">
     * <p>
     *
     * @param <T> the value type emitted
     * @param <S> the per-subscriber custom state type
     * @param stateSupplier called for each incoming Subscriber to provide the initial state for the generator bifunction
     * @param generator Consume the {@link SynchronousSink} provided per-subscriber by Reactor
     * as well as the current state to generate a <strong>single</strong> signal on each pass
     * and return a (new) state.
     * @param stateConsumer called after the generator has terminated or the downstream cancelled, receiving the last
     * state to be handled (i.e., release resources or do other cleanup).
     *
     * @return a {@link Flux}
     */
    public static <T, S> Flux<T> generate(Callable<S> stateSupplier, BiFunction<S, SynchronousSink<T>, S> generator, Consumer<? super S> stateConsumer) {
        return onAssembly(new FluxGenerate<>(stateSupplier, generator, stateConsumer));
    }
}

实例

Mono

    @Test
    public void testMonoBasic(){
        Mono.fromSupplier(() -> "Hello").subscribe(System.out::println);
        Mono.justOrEmpty(Optional.of("Hello")).subscribe(System.out::println);
        Mono.create(sink -> sink.success("Hello")).subscribe(System.out::println);
    }

Mono ,是指最多只能触发(emit) (事件)一次。它对应于 RxJava 库的 Single 和 Maybe 类型或者是java的Optional。因此一个异步任务,如果只是想要在完成时给出完成信号,就可以使用 Mono<Void>。

调用 Flux<T>的single()将返回一个 Mono<T>,而连接两个 monos一起使用 concatWith 将产生一个 Flux。

Flux

    @Test
    public void testBasic(){
        Flux.just("Hello", "World").subscribe(System.out::println);
        Flux.fromArray(new Integer[] {1, 2, 3}).subscribe(System.out::println);
        Flux.empty().subscribe(System.out::println);
        Flux.range(1, 10).subscribe(System.out::println);
        Flux.interval(Duration.of(10, ChronoUnit.SECONDS)).subscribe(System.out::println);
    }

Flux 相当于一个 RxJava Observable,能够发出 0~N 个数据项,然后(可选地)completing 或 erroring。处理多个数据项作为stream。

小结

Mono和Flux都是实现Publisher接口的抽象类,一个相当于Optional,一个相当于有0..N的stream。两个都是spring 5 reactive编程的重要基础概念。

doc