How to create a Source that can receive elements later via a method call?

为君一笑 提交于 2019-11-26 23:59:19

There are three ways this can be achieved:

1. Post Materialization with SourceQueue

You can use Source.queue that materializes the Flow into a SourceQueue:

case class Weather(zipCode : String, temperature : Double, raining : Boolean)

val bufferSize = 100

//if the buffer fills up then this strategy drops the oldest elements
//upon the arrival of a new element.
val overflowStrategy = akka.stream.OverflowStrategy.dropHead

val queue = Source.queue(bufferSize, overflowStrategy)
                  .filter(!_.raining)
                  .to(Sink foreach println)
                  .run() // in order to "keep" the queue Materialized value instead of the Sink's

queue offer Weather("02139", 32.0, true)

2. Post Materialization with Actor

There is a similar question and answer here, the gist being that you materialize the stream as an ActorRef and send messages to that ref:

val ref = Source.actorRef[Weather](Int.MaxValue, fail)
                .filter(!_.raining)
                .to(Sink foreach println )
                .run() // in order to "keep" the ref Materialized value instead of the Sink's

ref ! Weather("02139", 32.0, true)

3. Pre Materialization with Actor

Similarly, you could explicitly create an Actor that contains a message buffer, use that Actor to create a Source, and then send that Actor messages as described in the answer here:

object WeatherForwarder {
  def props : Props = Props[WeatherForwarder]
}

//see provided link for example definition
class WeatherForwarder extends Actor {...}

val actorRef = actorSystem actorOf WeatherForwarder.props 

//note the stream has not been instatiated yet
actorRef ! Weather("02139", 32.0, true) 

//stream already has 1 Weather value to process which is sitting in the 
//ActorRef's internal buffer
val stream = Source(ActorPublisher[Weather](actorRef)).runWith{...}

After playing around and looking for a good solution to this I came across this solution which is clean, simple, and works both pre and post materialization. https://stackoverflow.com/a/32553913/6791842

  val (ref: ActorRef, publisher: Publisher[Int]) =
    Source.actorRef[Int](bufferSize = 1000, OverflowStrategy.fail)
      .toMat(Sink.asPublisher(true))(Keep.both).run()

  ref ! 1 //before

  val source = Source.fromPublisher(publisher)

  ref ! 2 //before
  Thread.sleep(1000)
  ref ! 3 //before

  source.runForeach(println)

  ref ! 4 //after
  Thread.sleep(1000)
  ref ! 5 //after

Output:

1
2
3
4
5
易学教程内所有资源均来自网络或用户发布的内容,如有违反法律规定的内容欢迎反馈
该文章没有解决你所遇到的问题?点击提问,说说你的问题,让更多的人一起探讨吧!