Question

Akka-http documentation说：

除了将服务器端绑定的套接字视为源[IncomingConnection]和每个连接作为源[HttpRequest]带有一个接收器[HttpResponse]

假设我们从多个Source [IncomingConnection]获取包含传入连接的合并源。

然后，假设我们从Source [IncomingConnection]获得Source [HttpRequest]（参见下面的代码）。

然后，没问题，我们可以提供一个流程来将HttpRequest转换为HttpResponse。

这就是问题 - 我们如何才能正确地收集回复？我们如何加入对连接的响应？

用例背后的整个想法是可以优先处理来自不同连接的传入请求。在许多情况下应该是有用的我猜...

提前致谢！

修改解决方案基于@RamonJRomeroyVigil的答案：

服务器代码：

val in1 = Http().bind(interface = "localhost", port = 8200)
val in2 = Http().bind(interface = "localhost", port = 8201)

val connSrc = Source.fromGraph(FlowGraph.create() { implicit b =>
  import FlowGraph.Implicits._

  val merge = b.add(Merge[IncomingConnection](2))

  in1 ~> print("in1") ~> merge.in(0)
  in2 ~> print("in2") ~> merge.in(1)

  SourceShape(merge.out)
})

val reqSrc : Source[(HttpRequest, IncomingConnection), _] =
  connSrc.flatMapConcat { conn =>
    Source.empty[HttpResponse]
      .via(conn.flow)
      .map(request => (request, conn))
  }

val flow: Flow[(HttpRequest, IncomingConnection), (HttpResponse, IncomingConnection), _] =
  Flow[(HttpRequest, IncomingConnection)].map{
      case (HttpRequest(HttpMethods.GET, Uri.Path("/ping"), _, entity, _), conn: IncomingConnection) =>
        println(s"${System.currentTimeMillis()}: " +
          s"process request from ${conn.remoteAddress.getHostName}:${conn.remoteAddress.getPort}")
        (HttpResponse(entity = "pong"), conn)
    }

reqSrc.via(flow).to(Sink.foreach { case (resp, conn) =>
  Source.single(resp).via(conn.flow).runWith(Sink.ignore)
}).run()

def print(prefix: String) = Flow[IncomingConnection].map { s =>
  println(s"$prefix [ ${System.currentTimeMillis()} ]: ${s.remoteAddress}"); s
}

所以，我从控制台使用curl并看到以下内容：

% curl http://localhost:8200/ping
curl: (52) Empty reply from server

第二次卷曲请求失败：

% curl http://localhost:8200/ping
curl: (7) Failed to connect to localhost port 8200: Connection refused

在服务器控制台上，我发送第一个请求时会看到以下内容：

in1 [ 1450287301512 ]: /127.0.0.1:52461
1450287301626: process request from localhost:52461
[INFO] [12/16/2015 20:35:01.641] [default-akka.actor.default-dispatcher-6] [akka://default/system/IO-TCP-STREAM/server-1-localhost%2F127.0.0.1%3A8200] Message [akka.io.Tcp$Unbound$] from Actor[akka://default/system/IO-TCP/selectors/$a/0#119537130] to Actor[akka://default/system/IO-TCP-STREAM/server-1-localhost%2F127.0.0.1%3A8200#-1438663077] was not delivered. [1] dead letters encountered. This logging can be turned off or adjusted with configuration settings 'akka.log-dead-letters' and 'akka.log-dead-letters-during-shutdown'.
[INFO] [12/16/2015 20:35:01.641] [default-akka.actor.default-dispatcher-6] [akka://default/system/IO-TCP-STREAM/server-2-localhost%2F127.0.0.1%3A8201] Message [akka.io.Tcp$Unbound$] from Actor[akka://default/system/IO-TCP/selectors/$a/1#679898594] to Actor[akka://default/system/IO-TCP-STREAM/server-2-localhost%2F127.0.0.1%3A8201#1414174163] was not delivered. [2] dead letters encountered. This logging can be turned off or adjusted with configuration settings 'akka.log-dead-letters' and 'akka.log-dead-letters-during-shutdown'.

发送第二个请求时没有任何内容。

因此，内部连接流（如@RamonJRomeroyVigil所述）或其他内容似乎存在问题......

基本上代码不起作用。

仍在调查此问题。

Answer 1

以下解决方案基于问题评论中提供的进一步信息。

鉴于

val connSrc : Source[IncomingConnection,_] = ???

flatMapConcat方法解决了所述的具体问题：

val reqSrc : Source[(HttpRequest, IncomingConnection), _] =
  connSrc.flatMapConcat { conn =>
    Source.empty[HttpResponse]
          .via(conn.flow)
          .map(request => (request, conn))
  }

这提供了(HttpRequest, IncomingConnection)元组的来源。

假设您有一个将请求转换为响应的处理步骤

val flow : Flow[(HttpRequest, IncomingConnection), (HttpResponse, IncomingConnection), _] = ???

您可以将回复发送回客户：

reqSrc.via(flow).to(Sink.foreach { case (resp, conn) =>
  Source.single(resp).via(conn.flow).runWith(Sink.ignore)
})

警告：此解决方案调用conn.flow两次：一次创建生成请求的流，然后再创建一个流以发送响应。我不知道这种用例是否会破坏IncomingConnection逻辑中的某些内容。

Akka-http进程HttpRequests来自不同连接的一个流程

1 个答案: