Question

我目前正在玩akka流，并尝试了以下示例。

在请求某个HTTP端点时从kafka获取第一个元素。这是我编写的代码及其工作原理。

get {
      path("ticket" / IntNumber) { ticketNr =>

        val future = Consumer.plainSource(consumerSettings, Subscriptions.topics("tickets"))
          .take(1)
          .completionTimeout(5 seconds)
          .runWith(Sink.head)

        onComplete(future) {
          case Success(record) => complete(HttpEntity(ContentTypes.`text/html(UTF-8)`, record.value()))
          case _ => complete(HttpResponse(StatusCodes.NotFound))
        }
      }
    }

我只是想知道这是否是使用（akka）流的思想方式。那么还有更多＆＃34;直接＆＃34;将kafka流连接到HTTP响应流的方法？

例如，POSTing时我这样做：

val kafkaTicketsSink = Flow[String]
    .map(new ProducerRecord[Array[Byte], String]("tickets", _))
    .toMat(Producer.plainSink(producerSettings))(Keep.right)    

post {
          path("ticket") {
            (entity(as[Ticket]) & extractMaterializer) { (ticket, mat) => {
                val f = Source.single(ticket).map(t => t.description).runWith(kafkaTicketsSink)(mat)
                onComplete(f) { _ =>
                  val locationHeader = headers.Location(s"/ticket/${ticket.id}")
                  complete(HttpResponse(StatusCodes.Created, headers = List(locationHeader)))
                }
              }
            }
          }
        }

也许这也可以改进??

Answer 1

您可以使用Sink.queue保留单个背压流。每次收到请求时，您都可以从物化队列中提取元素。如果可用，这应该给你一个元素，否则给你背压。

有些事情：

  val queue = Consumer.plainSource(consumerSettings, Subscriptions.topics("tickets"))
    .runWith(Sink.queue())

  get {
    path("ticket" / IntNumber) { ticketNr =>

      val future: Future[Option[ConsumerRecord[String, String]]] = queue.pull()

      onComplete(future) {
        case Success(Some(record)) => complete(HttpEntity(ContentTypes.`text/html(UTF-8)`, record.value()))
        case _ => complete(HttpResponse(StatusCodes.NotFound))
      }
    }
  }

Sink.queue的更多信息可以在docs找到。

连接akka溪流kafka和akka http

1 个答案: