Guava源码学习（五）EventBus

基于版本：Guava 22.0

Wiki：EventBus

0. EventBus简介

提供了发布-订阅模型，可以方便的在EventBus上注册订阅者，发布者可以简单的将事件传递给EventBus，EventBus会自动将事件传递给相关联的订阅者。

支持同步/异步模式。

只能用于线程间通信。

1. EventBus类图

EventBus是同步实现

AsyncEventBus是异步实现

2. 代码实例

        EventBus eventBus = new EventBus();

        eventBus.register(new Object() {
            @Subscribe
            public void listen(Object subReport) throws Exception {
                System.out.println("receive object event!");
            }

            @Subscribe
            public void listen(Integer subReport) throws Exception {
                System.out.println("receive integer event!");
            }
        });

        eventBus.post(Integer.valueOf(1));

这段代码的输出如下：

receive integer event!
receive object event!

可以看到我们声明了一个EventBus，然后向其注册了两个订阅者（含有Subscribe注解的方法），然后调用post方法向EventBus发布了一条消息。

消息类型是Integer，由于订阅者的关注对象是Integer与Object，都与这条消息有关，所以两个订阅者都收到了通知。

但是这个发布-订阅模式是如何实现的呢？我们下面来逐步分析一下EventBus的源码。

3. EventBus.register()

EventBus.register
  /**
   * Registers all subscriber methods on {@code object} to receive events.
   *
   * @param object object whose subscriber methods should be registered.
   */
  public void register(Object object) {
    subscribers.register(object);
  }

SubscriberRegistry.register
  /**
   * Registers all subscriber methods on the given listener object.
   */
  void register(Object listener) {
    Multimap<Class<?>, Subscriber> listenerMethods = findAllSubscribers(listener);//获取传入的listener中含有的所有的订阅者

    for (Map.Entry<Class<?>, Collection<Subscriber>> entry : listenerMethods.asMap().entrySet()) {
      Class<?> eventType = entry.getKey();//订阅的事件类型
      Collection<Subscriber> eventMethodsInListener = entry.getValue();//对应的订阅者本身

      CopyOnWriteArraySet<Subscriber> eventSubscribers = subscribers.get(eventType);

      if (eventSubscribers == null) {
        CopyOnWriteArraySet<Subscriber> newSet = new CopyOnWriteArraySet<Subscriber>();
        eventSubscribers =
            MoreObjects.firstNonNull(subscribers.putIfAbsent(eventType, newSet), newSet);//线程安全的修改subscribers对象
      }

      eventSubscribers.addAll(eventMethodsInListener);
    }
  }

  /**
   * Returns all subscribers for the given listener grouped by the type of event they subscribe to.
    分析传入的对象，遍历其中所有含有Subscribe注解而且只含有一个参数的方法，然后将其包装成订阅者并返回。
   */
  private Multimap<Class<?>, Subscriber> findAllSubscribers(Object listener) {
    Multimap<Class<?>, Subscriber> methodsInListener = HashMultimap.create();
    Class<?> clazz = listener.getClass();
    for (Method method : getAnnotatedMethods(clazz)) {//getAnnotatedMethods方法最终会调用到下面的SubscriberRegistry.getAnnotatedMethodsNotCached方法，这个方法会用反射处理传入的clazz及其所有的父类，提取出含有Subscribe注解并且有且只有一个参数的方法
      Class<?>[] parameterTypes = method.getParameterTypes();
      Class<?> eventType = parameterTypes[0];//获取这个方法的唯一参数的Class
      methodsInListener.put(eventType, Subscriber.create(bus, listener, method));//将EventBus，订阅者对象，订阅者方法包装一下并放入map，后续触发事件的时候会用到
    }
    return methodsInListener;
  }

SubscriberRegistry.getAnnotatedMethodsNotCached
  private static ImmutableList<Method> getAnnotatedMethodsNotCached(Class<?> clazz) {
    Set<? extends Class<?>> supertypes = TypeToken.of(clazz).getTypes().rawTypes();//连clazz的父类也会处理
    Map<MethodIdentifier, Method> identifiers = Maps.newHashMap();
    for (Class<?> supertype : supertypes) {
      for (Method method : supertype.getDeclaredMethods()) {
        if (method.isAnnotationPresent(Subscribe.class) && !method.isSynthetic()) {//含有Subscribe注解，而且不是合成的方法
          // TODO(cgdecker): Should check for a generic parameter type and error out
          Class<?>[] parameterTypes = method.getParameterTypes();
          checkArgument(
              parameterTypes.length == 1,
              "Method %s has @Subscribe annotation but has %s parameters."
                  + "Subscriber methods must have exactly 1 parameter.",
              method,
              parameterTypes.length);//方法必须有且只有一个参数

          MethodIdentifier ident = new MethodIdentifier(method);
          if (!identifiers.containsKey(ident)) {
            identifiers.put(ident, method);
          }
        }
      }
    }
    return ImmutableList.copyOf(identifiers.values());
  }

从SubscriberRegistry.getAnnotatedMethods到SubscriberRegistry.getAnnotatedMethodsNotCached的跳转会涉及到Guava的LoadingCache，其他的逻辑都非常直观。

register方法的大概流程如下：

a. 解析传入的Object对象，用反射分析对应的class及其所有父类，找到所有用Subscribe注解修饰，而且只有一个参数的方法。由这个方法可以生成订阅者对象。

b. 解析这个方法的参数（订阅者订阅的事件类型）

c. 每个EventBus在初始化时都会与一个SubscriberRegistry关联，这个SubscriberRegistry内部维护了一个名为subscribers的ConcurrentMap，这个ConcurrentMap的 key是EventBus关心的事件类型，value是订阅了这些事件的订阅者的集合，subscribers在后续发布事件的流程中非常有用（可以通过发布的事件类型，找到这个事件所关联的所有订阅者，并通知这些相关的订阅者）。现在用a,b中的信息来更新subscribers，由于subscribers可能被并发操作，所以用到了ConcurrentMap.putIfAbsent方法以保证线程安全。

总之，调用register方法会更新subscribers，subscribers中含有事件与订阅者的关联关系。

4. EventBus.post()

EventBus.post
  public void post(Object event) {
    Iterator<Subscriber> eventSubscribers = subscribers.getSubscribers(event);//获取发布事件所关联的所有订阅者（包括event的所有父类/接口所关联的订阅者）
    if (eventSubscribers.hasNext()) {
      dispatcher.dispatch(event, eventSubscribers);//用分派器触发所有订阅者的订阅方法，分派器有同步/异步的不同实现
    } else if (!(event instanceof DeadEvent)) {
      // the event had no subscribers and was not itself a DeadEvent
      post(new DeadEvent(this, event));//如果这个事件没有订阅者，则将这个事件包装为DeadEvent事件，然后重新触发post方法
    }
  }

SubscriberRegistry.getSubscribers
  /**
   * Gets an iterator representing an immutable snapshot of all subscribers to the given event at
   * the time this method is called.
   */
  Iterator<Subscriber> getSubscribers(Object event) {
    ImmutableSet<Class<?>> eventTypes = flattenHierarchy(event.getClass());//flattenHierarchy方法会解析出event的class，及其所有父类/接口的class

    List<Iterator<Subscriber>> subscriberIterators =
        Lists.newArrayListWithCapacity(eventTypes.size());

    for (Class<?> eventType : eventTypes) {//遍历event关联的所有class，找到这些class关联的所有订阅者，添加到subscriberIterators中
      CopyOnWriteArraySet<Subscriber> eventSubscribers = subscribers.get(eventType);
      if (eventSubscribers != null) {
        // eager no-copy snapshot
        subscriberIterators.add(eventSubscribers.iterator());
      }
    }

    return Iterators.concat(subscriberIterators.iterator());
  }

SubscriberRegistry.flattenHierarchy
  /**
   * Flattens a class's type hierarchy into a set of {@code Class} objects including all
   * superclasses (transitively) and all interfaces implemented by these superclasses.
   */
  @VisibleForTesting
  static ImmutableSet<Class<?>> flattenHierarchy(Class<?> concreteClass) {
    try {
      return flattenHierarchyCache.getUnchecked(concreteClass);//这里会用Guava的TypeToken工具解析出concreteClass的class，以及所有父类/接口的class并返回
    } catch (UncheckedExecutionException e) {
      throw Throwables.propagate(e.getCause());
    }
  }

post方法的大概流程如下：

a. 用Guava的TypeToken工具解析Event的class，以及Event的所有父类/接口的class，所有订阅了这些class的订阅者后续都会收到通知

b. 调用EventBus关联的Dispatcher的dispatcher方法，将Event发布给所有关联的订阅者。

下一步我们会解析Dispatcher，以及Dispatcher是如何实现EventBus的同步/异步语义的。

5. EventBus的默认Dispatcher

EventBus
  /**
   * Creates a new EventBus named "default".
   */
  public EventBus() {
    this("default");
  }

  /**
   * Creates a new EventBus with the given {@code identifier}.
   *
   * @param identifier a brief name for this bus, for logging purposes. Should be a valid Java
   *     identifier.
   */
  public EventBus(String identifier) {
    this(
        identifier,
        MoreExecutors.directExecutor(),//由当前线程直接运行提交任务的“线程池”
        Dispatcher.perThreadDispatchQueue(),
        LoggingHandler.INSTANCE);
  }

Dispatcher.perThreadDispatchQueue()
  static Dispatcher perThreadDispatchQueue() {
    return new PerThreadQueuedDispatcher();
  }

Dispatcher.PerThreadQueuedDispatcher
  /**
   * Implementation of a {@link #perThreadDispatchQueue()} dispatcher.
   */
  private static final class PerThreadQueuedDispatcher extends Dispatcher {

    // This dispatcher matches the original dispatch behavior of EventBus.

    /**
     * Per-thread queue of events to dispatch.
     */
    private final ThreadLocal<Queue<Event>> queue =
        new ThreadLocal<Queue<Event>>() {
          @Override
          protected Queue<Event> initialValue() {
            return Queues.newArrayDeque();
          }
        };//为每个调用post方法的工作线程维护一个发布队列，工作线程独立操作这个队列完成时间发布流程，所以是线程安全的

    /**
     * Per-thread dispatch state, used to avoid reentrant event dispatching.
     */
    private final ThreadLocal<Boolean> dispatching =
        new ThreadLocal<Boolean>() {
          @Override
          protected Boolean initialValue() {
            return false;
          }
        };

    @Override
    void dispatch(Object event, Iterator<Subscriber> subscribers) {
      checkNotNull(event);
      checkNotNull(subscribers);
      Queue<Event> queueForThread = queue.get();
      queueForThread.offer(new Event(event, subscribers));//将发布事件+事件所关联的所有订阅者包装成一个Event对象，并提交至发布队列

      if (!dispatching.get()) {
        dispatching.set(true);
        try {
          Event nextEvent;
          while ((nextEvent = queueForThread.poll()) != null) {//尝试从发布队列中poll元素
            while (nextEvent.subscribers.hasNext()) {//遍历发布事件的所有订阅者，向订阅者派发事件，由于EventBus默认使用的线程池是MoreExecutors.directExecutor()，所以实际上发布者会串行而且同步的向事件的所有订阅者派发事件，直到全部派发结束，post方法才会返回，所以EventBus在默认情况下是同步的。
              nextEvent.subscribers.next().dispatchEvent(nextEvent.event);
            }
          }
        } finally {
          dispatching.remove();
          queue.remove();
        }
      }
    }

    private static final class Event {
      private final Object event;
      private final Iterator<Subscriber> subscribers;

      private Event(Object event, Iterator<Subscriber> subscribers) {
        this.event = event;
        this.subscribers = subscribers;
      }
    }
  }

Subscriber.dispatchEvent
  /**
   * Dispatches {@code event} to this subscriber using the proper executor.
   */
  final void dispatchEvent(final Object event) {//向订阅者派发事件
    executor.execute(
        new Runnable() {
          @Override
          public void run() {
            try {
              invokeSubscriberMethod(event);//用反射调用订阅的方法
            } catch (InvocationTargetException e) {
              bus.handleSubscriberException(e.getCause(), context(event));
            }
          }
        });
  }

Subscriber.invokeSubscriberMethod
  /**
   * Invokes the subscriber method. This method can be overridden to make the invocation
   * synchronized.
   */
  @VisibleForTesting
  void invokeSubscriberMethod(Object event) throws InvocationTargetException {
    try {
      method.invoke(target, checkNotNull(event));
    } catch (IllegalArgumentException e) {
      throw new Error("Method rejected target/argument: " + event, e);
    } catch (IllegalAccessException e) {
      throw new Error("Method became inaccessible: " + event, e);
    } catch (InvocationTargetException e) {
      if (e.getCause() instanceof Error) {
        throw (Error) e.getCause();
      }
      throw e;
    }
  }

同步语义的实现关键在于

a. 使用的线程池为MoreExecutors.directExecutor()，这实际上不能算是一个真正的线程池，所有提交到这个池子里的任务，都是由提交任务的线程自身来执行的。

b. EventBus为每个调用post方法的线程维护了一个发布队列，工作线程会将事件提交到这个私有队列里，然后逐个通知事件所关联的订阅者，由于使用的线程池是MoreExecutors.directExecutor()，所以这个过程实际上是完全串行而且同步执行的，调用post方法的工作线程实际上会在将事件通知给所有订阅者后才会返回，从而实现了同步的EventBus语义。

6. AsyncEventBus的Dispatcher

AsyncEventBus
  public AsyncEventBus(Executor executor) {
    super("default", executor, Dispatcher.legacyAsync(), LoggingHandler.INSTANCE);

Dispatcher.legacyAsync()
  static Dispatcher legacyAsync() {
    return new LegacyAsyncDispatcher();
  }

Dispatcher.LegacyAsyncDispatcher
  private static final class LegacyAsyncDispatcher extends Dispatcher {

    // This dispatcher matches the original dispatch behavior of AsyncEventBus.
    //
    // We can't really make any guarantees about the overall dispatch order for this dispatcher in
    // a multithreaded environment for a couple reasons:
    //
    // 1. Subscribers to events posted on different threads can be interleaved with each other
    //    freely. (A event on one thread, B event on another could yield any of
    //    [a1, a2, a3, b1, b2], [a1, b2, a2, a3, b2], [a1, b2, b3, a2, a3], etc.)
    // 2. It's possible for subscribers to actually be dispatched to in a different order than they
    //    were added to the queue. It's easily possible for one thread to take the head of the
    //    queue, immediately followed by another thread taking the next element in the queue. That
    //    second thread can then dispatch to the subscriber it took before the first thread does.
    //
    // All this makes me really wonder if there's any value in queueing here at all. A dispatcher
    // that simply loops through the subscribers and dispatches the event to each would actually
    // probably provide a stronger order guarantee, though that order would obviously be different
    // in some cases.

    /**
     * Global event queue.
     */
    private final ConcurrentLinkedQueue<EventWithSubscriber> queue =
        Queues.newConcurrentLinkedQueue();//公用的，支持并发访问的发布队列

    @Override
    void dispatch(Object event, Iterator<Subscriber> subscribers) {
      checkNotNull(event);
      while (subscribers.hasNext()) {//将Event+订阅者包装成EventWithSubscriber，放入发布队列中
        queue.add(new EventWithSubscriber(event, subscribers.next()));
      }

      EventWithSubscriber e;
      while ((e = queue.poll()) != null) {
        e.subscriber.dispatchEvent(e.event);//从发布队列中提取EventWithSubscriber，然后将事件发布给订阅者，这一过程是发送到AsyncEventBus初始化时提供的线程池里执行的，所以是异步的。也就是说post方法返回的时候，可能发布过程尚未完成，还有订阅者没有收到消息。
      }
    }

    private static final class EventWithSubscriber {
      private final Object event;
      private final Subscriber subscriber;

      private EventWithSubscriber(Object event, Subscriber subscriber) {
        this.event = event;
        this.subscriber = subscriber;
      }
    }
  }

异步语义实现的关键在于：

a. 发布事件使用的线程池是创建AsyncEventBus时传入的线程池，一般来说都是正常的含有多个工作线程的线程池，提交到线程池里的任务可以被异步执行

b. LegacyAsyncDispatcher中维护了一个公用的ConcurrentLinkedQueue，这个AsyncEventBus收到的所有Event都会被put到这个队列中，put完Event的线程同时也会从这个队列中poll Event，然后submit这些Event到线程池中，如果这个线程池是普通的含有工作线程的线程池，那么submit完Event之后post方法即可立即返回，传递Event给订阅者的任务会由线程池里的工作线程完成。这样就实现了异步语义。

总的来说，EventBus的源码还是比较清晰易懂的，实现手法也非常优雅，值得我们学习。