Akka(29): Http:Server-Side-Api,Low-Level-Api

 Akka-http针对Connection的两头都提供了方便编程的Api,分别是Server-Side-Api和Client-Side-Api。通过这两个Api可以大大提高编程效率。当然,上期我们提到过,Http-Server是Akka-http的核心部分,所有系统集成功能都是在Server-Side实现的。Akka-http-Server-Side-Api可以说是最先进的Http-Server编程工具,支持:

  • Full support for HTTP persistent connections
  • Full support for HTTP pipelining
  • Full support for asynchronous HTTP streaming including “chunked” transfer encoding accessible through an idiomatic API
  • Optional SSL/TLS encryption
  • WebSocket support

Server-Side-Api又分两个层次:Low-level-Server-Side-Api和High-level-Server-Side-Api。Low-level-server-api支持HTTP/1.1Server所有功能,包括:

  • Connection management
  • Parsing and rendering of messages and headers
  • Timeout management (for requests and connections)
  • Response ordering (for transparent pipelining support)

其它Server功能如请求解析request routing,文件服务file serving,数据压缩compression等都放在了High-level-server-api里。Akka-http是基于Akka-stream编写的,所以我们需要从Akka-stream运算模式来理解Akka-http的类型表现形式。

一个Http-Server是绑定在一个Socket上来接收客户端上传的request进行相关的服务提供的。Server对Socket的绑定在Akka-http里的可以Stream形式来表现:

val serverSource: Source[Http.IncomingConnection, Future[Http.ServerBinding]] =
  Http().bind(interface = "localhost", port = 8080)

Server-Side Socket绑定实际上就是一个Akka-stream-source:Source[IncomingConnection]:

  /**
   * Creates a [[akka.stream.scaladsl.Source]] of [[akka.http.scaladsl.Http.IncomingConnection]] instances which represents a prospective HTTP server binding
   * on the given `endpoint`.
   *
   * If the given port is 0 the resulting source can be materialized several times. Each materialization will
   * then be assigned a new local port by the operating system, which can then be retrieved by the materialized
   * [[akka.http.scaladsl.Http.ServerBinding]].
   *
   * If the given port is non-zero subsequent materialization attempts of the produced source will immediately
   * fail, unless the first materialization has already been unbound. Unbinding can be triggered via the materialized
   * [[akka.http.scaladsl.Http.ServerBinding]].
   *
   * If an [[ConnectionContext]] is given it will be used for setting up TLS encryption on the binding.
   * Otherwise the binding will be unencrypted.
   *
   * If no `port` is explicitly given (or the port value is negative) the protocol's default port will be used,
   * which is 80 for HTTP and 443 for HTTPS.
   *
   * To configure additional settings for a server started using this method,
   * use the `akka.http.server` config section or pass in a [[akka.http.scaladsl.settings.ServerSettings]] explicitly.
   */
  def bind(interface: String, port: Int = DefaultPortForProtocol,
           connectionContext: ConnectionContext = defaultServerHttpContext,
           settings:          ServerSettings    = ServerSettings(system),
           log:               LoggingAdapter    = system.log)(implicit fm: Materializer): Source[Http.IncomingConnection, Future[ServerBinding]] = {
    val fullLayer = fuseServerBidiFlow(settings, connectionContext, log)

    tcpBind(interface, choosePort(port, connectionContext), settings)
      .map(incoming ⇒ {
        val serverFlow = fullLayer.addAttributes(prepareAttributes(settings, incoming)) join incoming.flow
        IncomingConnection(incoming.localAddress, incoming.remoteAddress, serverFlow)
      })
      .mapMaterializedValue(materializeTcpBind)
  }

run这个Source[IncomingConnection]产生一串连接Connection: 

  /**
   * Represents one accepted incoming HTTP connection.
   */
  final case class IncomingConnection(
    localAddress:  InetSocketAddress,
    remoteAddress: InetSocketAddress,
    flow:          Flow[HttpResponse, HttpRequest, NotUsed]) {

    /**
     * Handles the connection with the given flow, which is materialized exactly once
     * and the respective materialization result returned.
     */
    def handleWith[Mat](handler: Flow[HttpRequest, HttpResponse, Mat])(implicit fm: Materializer): Mat =
      flow.joinMat(handler)(Keep.right).run()

    /**
     * Handles the connection with the given handler function.
     */
    def handleWithSyncHandler(handler: HttpRequest ⇒ HttpResponse)(implicit fm: Materializer): Unit =
      handleWith(Flow[HttpRequest].map(handler))

    /**
     * Handles the connection with the given handler function.
     */
    def handleWithAsyncHandler(handler: HttpRequest ⇒ Future[HttpResponse], parallelism: Int = 1)(implicit fm: Materializer): Unit =
      handleWith(Flow[HttpRequest].mapAsync(parallelism)(handler))
  }

IncomingConnection类型提供了个handleWith这样的streaming函数进行request到response的转换。用户可以下面的方式提供自定义的转换方法:

调用handleWith传入Flow[HttpRequest,HttpResponse,_],如:

  def req2Resp: HttpRequest => HttpResponse = _ => HttpResponse(entity=
    HttpEntity(ContentTypes.`text/html(UTF-8)`,"<h> Hello World! </h>"))
  val flow = Flow.fromFunction(req2Resp)

提供HttpRequest=>HttpResponse函数传人handleWithSyncHandler:

def syncHandler: HttpRequest => HttpResponse = {
    case HttpRequest(HttpMethods.GET,Uri.Path("/"),_headers,_entiy,_protocol) =>
      HttpResponse(entity=
        HttpEntity(ContentTypes.`text/html(UTF-8)`,"<h> Hello World! </h>"))

    case req: HttpRequest =>
      req.discardEntityBytes() // important to drain incoming HTTP Entity stream
      HttpResponse(404, entity = "Unknown resource!")
  }

提供HttpRequest=>Future[HttpResponse]函数传人handleWithASyncHandler:

  def asyncHandler: HttpRequest => Future[HttpResponse] = {
    case HttpRequest(HttpMethods.GET,Uri.Path("/"),_headers,_entiy,_protocol) => Future {
      HttpResponse(entity=
        HttpEntity(ContentTypes.`text/html(UTF-8)`,"<h> Hello World! </h>")) }

    case req: HttpRequest => Future {
      req.discardEntityBytes() // important to drain incoming HTTP Entity stream
      HttpResponse(404, entity = "Unknown resource!")
    }
  }

run Source[IncomingConnection,Future[ServerBinding]]返回结果为Future[ServerBinding]:

  val futBinding: Future[Http.ServerBinding] =
    connSource.to { Sink.foreach{ connection =>
        println(s"client address ${connection.remoteAddress}")
  //      connection handleWith flow
  //    connection handleWithSyncHandler syncHandler
      connection handleWithAsyncHandler asyncHandler
    }}.run()

我们可以通过ServerBinding来释放绑定的Socket:

 /**
   * Represents a prospective HTTP server binding.
   *
   * @param localAddress  The local address of the endpoint bound by the materialization of the `connections` [[akka.stream.scaladsl.Source]]
   *
   */
  final case class ServerBinding(localAddress: InetSocketAddress)(private val unbindAction: () ⇒ Future[Unit]) {

    /**
     * Asynchronously triggers the unbinding of the port that was bound by the materialization of the `connections`
     * [[akka.stream.scaladsl.Source]]
     *
     * The produced [[scala.concurrent.Future]] is fulfilled when the unbinding has been completed.
     */
    def unbind(): Future[Unit] = unbindAction()
  }

我们可以调用这个unbind():

  futBinding.flatMap(_.unbind())

整个示范源代码如下:

import akka.actor._
import akka.stream._
import akka.stream.scaladsl._
import akka.http.scaladsl.Http
import akka.http.scaladsl.model._
import scala.concurrent._

object LowLevelServerApi extends App {
  implicit val httpSys = ActorSystem("actorSystem")
  implicit val httpMat = ActorMaterializer()
  implicit val httpEc = httpSys.dispatcher

  val (interface,port) = ("localhost",8088)
  val connSource: Source[Http.IncomingConnection,Future[Http.ServerBinding]] =
    Http().bind(interface,port)

  def req2Resp: HttpRequest => HttpResponse = _ => HttpResponse(entity=
    HttpEntity(ContentTypes.`text/html(UTF-8)`,"<h> Hello World! </h>"))
  val flow = Flow.fromFunction(req2Resp)

  def syncHandler: HttpRequest => HttpResponse = {
    case HttpRequest(HttpMethods.GET,Uri.Path("/"),_headers,_entiy,_protocol) =>
      HttpResponse(entity=
        HttpEntity(ContentTypes.`text/html(UTF-8)`,"<h> Hello World! </h>"))

    case req: HttpRequest =>
      req.discardEntityBytes() // important to drain incoming HTTP Entity stream
      HttpResponse(404, entity = "Unknown resource!")
  }

  def asyncHandler: HttpRequest => Future[HttpResponse] = {
    case HttpRequest(HttpMethods.GET,Uri.Path("/"),_headers,_entiy,_protocol) => Future {
      HttpResponse(entity=
        HttpEntity(ContentTypes.`text/html(UTF-8)`,"<h> Hello World! </h>")) }

    case req: HttpRequest => Future {
      req.discardEntityBytes() // important to drain incoming HTTP Entity stream
      HttpResponse(404, entity = "Unknown resource!")
    }
  }

  val futBinding: Future[Http.ServerBinding] =
    connSource.to { Sink.foreach{ connection =>
        println(s"client address ${connection.remoteAddress}")
  //      connection handleWith flow
  //    connection handleWithSyncHandler syncHandler
      connection handleWithAsyncHandler asyncHandler
    }}.run()

  println(s"Server running at $interface $port. Press any key to exit ...")

  scala.io.StdIn.readLine()

  futBinding.flatMap(_.unbind())
    .onComplete(_ => httpSys.terminate())
  
}

本文参与腾讯云自媒体分享计划,欢迎正在阅读的你也加入,一起分享。

发表于

我来说两句

0 条评论
登录 后参与评论

相关文章

来自专栏个人分享

MongoDB项目中常用方法

  前台通过DBCollection 根据名称获取相应的value 然后加入list中~

853
来自专栏JadePeng的技术博客

使用Mongodb 做对象缓存

mongodb高效的访问速度,用来快速存取数据再合适不过了,缓存神马的,可以用这个的 另外,有的时候,如果仅仅存储几条数据,单独去建立一张表代价太大,这个时候,...

3285
来自专栏码匠的流水账

聊聊storm的WindowedBoltExecutor

storm-2.0.0/storm-client/src/jvm/org/apache/storm/topology/WindowedBoltExecutor....

732
来自专栏码匠的流水账

聊聊storm的tickTuple

storm-2.0.0/storm-client/src/jvm/org/apache/storm/utils/TupleUtils.java

692
来自专栏码匠的流水账

聊聊storm的WindowedBolt

storm-2.0.0/storm-client/src/jvm/org/apache/storm/topology/IWindowedBolt.java

612
来自专栏智能大石头

MDK C++中对内联的极度优化

先来看看我们SmartIRQ的具体实现 // 智能IRQ,初始化时备份,销毁时还原 class SmartIRQ { public: force_inl...

2156
来自专栏移动端周边技术扩展

weex开发之push路径定义与处理

1909
来自专栏Java与Android技术栈

RxJava 之 ParallelFlowable

RxJava 2.0.5 版本新增了 ParallelFlowable API,它允许并行地执行一些运算符,譬如map、filter、concatMap、fla...

642
来自专栏函数式编程语言及工具

FunDA(16)- 示范:整合并行运算 - total parallelism solution

   在对上两篇讨论中我们介绍了并行运算的两种体现方式:并行构建数据源及并行运算用户自定义函数。我们分别对这两部分进行了示范。本篇我准备示范把这两种情况集成一体...

18110
来自专栏码匠的流水账

聊聊storm的LoggingMetricsConsumer

storm-2.0.0/storm-client/src/jvm/org/apache/storm/metric/LoggingMetricsConsumer....

933

扫码关注云+社区