akka-typed(2) - typed-actor交流方式和交流协议

   akka系统是一个分布式的消息驱动系统。akka应用由一群负责不同运算工作的actor组成，每个actor都是被动等待外界的某种消息来驱动自己的作业。所以，通俗点描述：akka应用就是一群actor相互之间发送消息的系统，每个actor接收到消息后开始自己负责的工作。对于akka-typed来说，typed-actor只能接收指定类型的消息，所以actor之间的消息交流需要按照消息类型来进行，即需要协议来规范消息交流机制。想想看，如果用户需要一个actor做某件事，他必须用这个actor明白的消息类型来发送消息，这就是一种交流协议。

所谓消息交流方式包括单向和双向两类。如果涉及两个actor之间的消息交换，消息发送方式可以是单向和双向的。但如果是从外界向一个actor发送消息，那么肯定只能是单向的发送方式了，因为消息发送两端只有一端是actor。

典型的单向消息发送fire-and-forget如下：

import akka.actor.typed._
import scaladsl._

object Printer {
  case class PrintMe(message: String)
// 只接收PrintMe类型message
  def apply(): Behavior[PrintMe] =
    Behaviors.receive {
      case (context, PrintMe(message)) =>
        context.log.info(message)
        Behaviors.same
    }
}

object FireAndGo extends App {
  // system就是一个root-actor
val system: ActorRef[Printer.PrintMe] = ActorSystem(Printer(), "fire-and-forget-sample")
val printer: ActorRef[Printer.PrintMe] = system
// 单向消息发送，printMe类型的消息
printer ! Printer.PrintMe("hello")
printer ! Printer.PrintMe("world!")

system.asInstanceOf[ActorSystem[Printer.PrintMe]].terminate()

}

当然，在现实中通常我们要求actor去进行某些运算然后返回运算结果。这就涉及到actor之间双向信息交换了。第一种情况：两个actor之间的消息是任意无序的，这是一种典型的无顺序request-response模式。就是说一个response不一定是按照request的接收顺序返回的，只是它们之间能够交流而已。不过，在akka-typed中这种模式最基本的要求就是发送的消息类型必须符合接收方actor的类型。

好了，我们先对这个模式做个示范。所有actor的定义可以先从它的消息类型开始。对每个参加双向交流的actor来说，可以从request和response两种消息来反映它的功能：

object FrontEnd {
  sealed trait FrontMessages
  case class SayHi(who: String) extends FrontMessages
}
object BackEnd {
  //先从这个actor的回应消息开始
     sealed trait Response
     case class  HowAreU(msg: String) extends Response
     case object Unknown extends Response

  //可接收消息类型
  sealed trait BackMessages
  //这个replyTo应该是一个能处理Reponse类型消息的actor
  case class MakeHello(who: String, replyTo: ActorRef[Response]) extends BackMessages

}

这个FrontEnd接收SayHi消息后开始工作，不过目前还没有定义返回的消息类型。BackEnd接到MakeHello类型消息后返回response类型消息。从这个角度来讲，返回的对方actor必须能够处理Response类型的消息。

我们试试实现这个FrontEnd actor：

object FrontEnd {
  sealed trait FrontMessages
  case class SayHi(who: String) extends FrontMessages
  
  def apply(backEnd: ActorRef[BackEnd.BackMessages]): Behavior[FrontMessages] =  {
     Behaviors.receive { (ctx,msg) => msg match {
       case SayHi(who) =>
         ctx.log.info("requested to say hi to {}", who)
         backEnd ! BackEnd.MakeHello(who, ???)
       
     }
  }
}

MakeHello需要一个replyTo，应该是什么呢？不过它一定是可以处理Response类型消息的actor。但我们知道这个replyTo就是FrontEnd，不过FrontEnd只能处理FrontMessages类型消息，应该怎么办呢？可不可以把replyTo直接写成FrontEnd呢？虽然可以这么做，但这个MakeHello消息就只能跟FrontEnd绑死了。如果其它的actor也需要用到这个MakeHello的话就需要另外定义一个了。所以，最好的解决方案就是用某种类型转换方式来实现。如下：

import akka.actor.typed._
import scaladsl._

object FrontEnd {
  sealed trait FrontMessages
  case class SayHi(who: String) extends FrontMessages

  case class WrappedBackEndResonse(res: BackEnd.Response) extends FrontMessages

  def apply(backEnd: ActorRef[BackEnd.BackMessages]): Behavior[FrontMessages] = {
    Behaviors.setup[FrontMessages] { ctx =>
                                                   //ctx.messageAdapter(ref => WrappedBackEndResonse(ref))
      val backEndRef: ActorRef[BackEnd.Response] = ctx.messageAdapter(WrappedBackEndResonse)
      Behaviors.receive { (ctx, msg) =>
        msg match {
          case SayHi(who) =>
            ctx.log.info("requested to say hi to {}", who)
            backEnd ! BackEnd.MakeHello(who, backEndRef)
            Behaviors.same
          //messageAdapter将BackEnd.Response转换成WrappedBackEndResponse
          case WrappedBackEndResonse(msg) => msg match {
            case BackEnd.HowAreU(msg) =>
              ctx.log.info(msg)
              Behaviors.same
            case BackEnd.Unknown =>
              ctx.log.info("Unable to say hello")
              Behaviors.same
          }
        }
      }
    }
  }
}

首先，我们用ctx.mesageAdapter产生了ActorRef[BackEnd.Response]，正是我们需要提供给MakeHello消息的replyTo。看看这个messageAdapter函数：

def messageAdapter[U: ClassTag](f: U => T): ActorRef[U]

如果我们进行类型替换U -> BackEnd.Response, T -> FrontMessage 那么：

      val backEndRef: ActorRef[BackEnd.Response] = 
            ctx.messageAdapter((response: BackEnd.Response) => WrappedBackEndResonse(response))

实际上这个messageAdapter函数在本地ActorContext范围内登记了一个从BackEnd.Response类型到FrontMessages的转换。把接收到的BackEnd.Response立即转换成WrappedBackEndResponse(response)。

还有一种两个actor之间的双向交流模式是 1:1 request-response，即一对一模式。一对一的意思是发送方发送消息后等待回应消息。这就意味着收信方需要在完成运算任务后立即向发信方发送回应，否则造成发信方的超时异常。无法避免的是，这种模式依然会涉及消息类型的转换，如下：

object FrontEnd {
  sealed trait FrontMessages
  case class SayHi(who: String) extends FrontMessages

  case class WrappedBackEndResonse(res: BackEnd.Response) extends FrontMessages
  case class ErrorResponse(errmsg: String) extends FrontMessages

  def apply(backEnd: ActorRef[BackEnd.BackMessages]): Behavior[FrontMessages] = {
    Behaviors.setup[FrontMessages] { ctx =>
      //ask需要超时上限
      import scala.concurrent.duration._
      import scala.util._
      implicit val timeOut: Timeout = 3.seconds
      Behaviors.receive[FrontMessages] { (ctx, msg) =>
        msg match {
          case SayHi(who) =>
            ctx.log.info("requested to say hi to {}", who)
            
            ctx.ask(backEnd,(backEndRef: ActorRef[BackEnd.Response]) => BackEnd.MakeHello(who,backEndRef) ){
              case Success(backResponse) => WrappedBackEndResonse(backResponse)
              case Failure(err) =>ErrorResponse(err.getLocalizedMessage)
            }
            Behaviors.same

          case WrappedBackEndResonse(msg) => msg match {
            case BackEnd.HowAreU(msg) =>
              ctx.log.info(msg)
              Behaviors.same
            case BackEnd.Unknown =>
              ctx.log.info("Unable to say hello")
              Behaviors.same
          }
          case ErrorResponse(errmsg) =>
            ctx.log.info("ask error: {}",errmsg)
            Behaviors.same
        }
      }
    }
  }
}

似乎类型转换是在ask里实现的，看看这个函数：

  def ask[Req, Res](target: RecipientRef[Req], createRequest: ActorRef[Res] => Req)(
      mapResponse: Try[Res] => T)(implicit responseTimeout: Timeout, classTag: ClassTag[Res]): Unit

req -> BackEnd.BackMessages, res -> BackEnd.Response, T -> FrontMessages。现在ask可以写成下面这样：

 

            ctx.ask[BackEnd.BackMessages,BackEnd.Response](backEnd,
               (backEndRef: ActorRef[BackEnd.Response]) => BackEnd.MakeHello(who,backEndRef) ){
              case Success(backResponse:BackEnd.Response) => WrappedBackEndResonse(backResponse)
              case Failure(err) =>ErrorResponse(err.getLocalizedMessage)
            }

 

这样看起来更明白点，也就是说ask把接收的BackEnd.Response转换成了FrontEnd处理的消息类型WrappedBackEndRespnse，也就是FrontMessages

还有一种ask模式是在actor之外进行的，如下：

object AskDemo extends App {
  import akka.actor.typed.scaladsl.AskPattern._
  import scala.concurrent._
  import scala.concurrent.duration._
  import akka.util._
  import scala.util._

  implicit val system: ActorSystem[BackEnd.BackMessages] = ActorSystem(BackEnd(), "front-app")
  // asking someone requires a timeout if the timeout hits without response
  // the ask is failed with a TimeoutException
  implicit val timeout: Timeout = 3.seconds

  val result: Future[BackEnd.Response] =
        system.asInstanceOf[ActorRef[BackEnd.BackMessages]]
            .ask[BackEnd.Response]((ref: ActorRef[BackEnd.Response]) =>
          BackEnd.MakeHello("John", ref))

  // the response callback will be executed on this execution context
  implicit val ec = system.executionContext

  result.onComplete {
    case Success(res)  => res match {
      case BackEnd.HowAreU(msg) =>
        println(msg)
      case BackEnd.Unknown =>
        println("Unable to say hello")
    }
    case Failure(ex)  => 
      println(s"error: ${ex.getMessage}")
  }

  system.terminate()

}

 这个ask是在akka.actor.typed.scaladsl.AskPattern包里。函数款式如下：

   def ask[Res](replyTo: ActorRef[Res] => Req)(implicit timeout: Timeout, scheduler: Scheduler): Future[Res]

 

向ask传入一个函数ActorRef[BackEnd.Response] => BackEnd.BackMessages，然后返回Future[BackEnd.Response]。这个模式中接收回复方是在ActorContext之外，不存在消息截获机制，所以不涉及消息类型的转换。

另一种单actor双向消息交换模式，即自己ask自己。在ActorContext内向自己发送消息并提供回应消息的接收，如pipeToSelf：

 

object PipeFutureTo {
  trait CustomerDataAccess {
    def update(value: Customer): Future[Done]
  }

  final case class Customer(id: String, version: Long, name: String, address: String)

  object CustomerRepository {

    sealed trait Command

    final case class Update(value: Customer, replyTo: ActorRef[UpdateResult]) extends Command

    sealed trait UpdateResult

    final case class UpdateSuccess(id: String) extends UpdateResult

    final case class UpdateFailure(id: String, reason: String) extends UpdateResult

    private final case class WrappedUpdateResult(result: UpdateResult, replyTo: ActorRef[UpdateResult])
      extends Command

    private val MaxOperationsInProgress = 10

    def apply(dataAccess: CustomerDataAccess): Behavior[Command] = {
      Behaviors.setup[Command] { ctx =>
          implicit val dispatcher =  ctx.system.dispatchers.lookup(DispatcherSelector.fromConfig("my-dispatcher"))
          next(dataAccess, operationsInProgress = 0)
      }
    }

    private def next(dataAccess: CustomerDataAccess, operationsInProgress: Int)(implicit ec: ExecutionContextExecutor): Behavior[Command] = {
      Behaviors.receive { (context, command) =>
        command match {
          case Update(value, replyTo) =>
            if (operationsInProgress == MaxOperationsInProgress) {
              replyTo ! UpdateFailure(value.id, s"Max $MaxOperationsInProgress concurrent operations supported")
              Behaviors.same
            } else {
              val futureResult = dataAccess.update(value)
              context.pipeToSelf(futureResult) {
                // map the Future value to a message, handled by this actor
                case Success(_) => WrappedUpdateResult(UpdateSuccess(value.id), replyTo)
                case Failure(e) => WrappedUpdateResult(UpdateFailure(value.id, e.getMessage), replyTo)
              }
              // increase operationsInProgress counter
              next(dataAccess, operationsInProgress + 1)
            }

          case WrappedUpdateResult(result, replyTo) =>
            // send result to original requestor
            replyTo ! result
            // decrease operationsInProgress counter
            next(dataAccess, operationsInProgress - 1)
        }
      }
    }
  }
}