Server WebSockets

Ktor supports the WebSocket protocol and allows you to create applications that require real-time data transfer from and to the server. For example, WebSockets can be used to create a chat application.

Ktor allows you to:

Configure basic WebSocket settings, such as frame size, a ping period, and so on.
Handle a WebSocket session for exchanging messages between the server and client.
Add WebSocket extensions. For example, you can use the Deflate extension or implement a custom extension.

Add dependencies

To enable WebSockets support, you need to include the ktor-websockets artifact in the build script:

implementation "io.ktor:ktor-websockets:$ktor_version"

implementation("io.ktor:ktor-websockets:$ktor_version")

        
                    <dependency>
                        <groupId>io.ktor</groupId>
                        <artifactId>ktor-websockets</artifactId>
                        <version>${ktor_version}</version>
                    </dependency>
                    
        

Install WebSockets

To install the WebSockets plugin, pass it to the install function in the application initialization code. Depending on the way used to create a server, this can be the embeddedServer function call ...

            import io.ktor.features.*
            // ...
            fun main() {
                embeddedServer(Netty, port = 8080) {
                    install(WebSockets)
                    // ...
                }.start(wait = true)
            }
        

... or a specified module.

            import io.ktor.features.*
            // ...
            fun Application.module() {
                install(WebSockets)
                // ...
            }
        

Configure WebSockets settings

Optionally, you can configure the plugin by passing WebSocketOptions to the install function:

    install(WebSockets) {
        pingPeriod = Duration.ofSeconds(15)
        timeout = Duration.ofSeconds(15)
        maxFrameSize = Long.MAX_VALUE
        masking = false
    }

Handle WebSockets sessions

API overview

After you installed and configured the WebSockets plugin, you are ready to handle a WebSocket session. First, you need to define a WebSocket endpoint on a server by calling the webSocket function inside the routing block:

routing { 
    webSocket("/echo") {
       // Handle a WebSocket session
    }
}

For such an endpoint, a server accepts WebSocket requests to ws://localhost:8080/echo when a default configuration is used.

Inside the webSocket block, you need to handle a WebSocket session, which is represented by the DefaultWebSocketServerSession class. Session configuration might look as follows:

Use the send function to send text content to the client.
Use the incoming and outgoing properties to access the channels for receiving and sending WebSocket frames. A frame is represented by the Frame class.
When handing a session, you can check a frame type, for example:
- Frame.Text is a text frame. For this frame type, you can read its content using Frame.Text.readText().
- Frame.Binary is a binary frame. For this type, you can read its content using Frame.Binary.readBytes().
- Frame.Close is a closing frame. You can call Frame.Close.readReason() to get a close reason for the current session.
Use the close function to send a close frame with the specified reason.

Below we'll take a look at the examples of using this API.

Example: Handle a single session

The example below shows how to create the echo WebSocket endpoint to handle a session with one client:

    routing {
        webSocket("/echo") {
            send("Please enter your name")
            for (frame in incoming) {
                when (frame) {
                    is Frame.Text -> {
                        val receivedText = frame.readText()
                        if (receivedText.equals("bye", ignoreCase = true)) {
                            close(CloseReason(CloseReason.Codes.NORMAL, "Client said BYE"))
                        } else {
                            send(Frame.Text("Hi, $receivedText!"))
                        }
                    }
                }
            }
        }
    }
}

You can find the full example here: server-websockets.

Example: Handle multiple sessions

To handle multiple WebSocket sessions (for example, for a chat application), you need to store each session on a server. For example, you can define a connection with a unique name and associate it with a specified session. A sample Connection class below shows how to do this:

class Connection(val session: DefaultWebSocketSession) {
    companion object {
        var lastId = AtomicInteger(0)
    }

    val name = "user${lastId.getAndIncrement()}"
}

Then, you can create a new connection inside the webSocket handler when a new client connects to the WebSocket endpoint:

    routing {
        val connections = Collections.synchronizedSet<Connection?>(LinkedHashSet())
        webSocket("/chat") {
            val thisConnection = Connection(this)
            connections += thisConnection
            send("You've logged in as [${thisConnection.name}]")

            for (frame in incoming) {
                when (frame) {
                    is Frame.Text -> {
                        val receivedText = frame.readText()
                        val textWithUsername = "[${thisConnection.name}]: $receivedText"
                        connections.forEach {
                            it.session.send(textWithUsername)
                        }
                    }
                }
            }
        }
    }
}

You can find the full example here: server-websockets.

Testing

You can test WebSocket conversations by using the handleWebSocketConversation function inside the withTestApplication block:

class ModuleTest {
    @Test
    fun testConversation() {
        withTestApplication(Application::module) {
            handleWebSocketConversation("/echo") { incoming, outgoing ->
                val greetingText = (incoming.receive() as Frame.Text).readText()
                assertEquals("Please enter your name", greetingText)

                outgoing.send(Frame.Text("JetBrains"))
                val responseText = (incoming.receive() as Frame.Text).readText()
                assertEquals("Hi, JetBrains!", responseText)

                outgoing.send(Frame.Text("bye"))
                val closeReason = (incoming.receive() as Frame.Close).readReason()?.message
                assertEquals("Client said BYE", closeReason)
            }
        }
    }
}

The WebSocket API and Ktor

The standard events from the WebSocket API map to Ktor in the following way:

onConnect happens at the start of the block.
onMessage happens after successfully reading a message (for example, with incoming.receive()) or using suspended iteration with for(frame in incoming).
onClose happens when the incoming channel is closed. That would complete the suspended iteration, or throw a ClosedReceiveChannelException when trying to receive a message`.
onError is equivalent to other exceptions.

In both onClose and onError, the closeReason property is set.

webSocket("/echo") {
    println("onConnect")
    try {
        for (frame in incoming){
            val text = (frame as Frame.Text).readText()
            println("onMessage")
            received += text
            outgoing.send(Frame.Text(text))
        }
    } catch (e: ClosedReceiveChannelException) {
        println("onClose ${closeReason.await()}")
    } catch (e: Throwable) {
        println("onError ${closeReason.await()}")
        e.printStackTrace()
    }
}

In this sample, the infinite loop is only exited with an exception is risen: either a ClosedReceiveChannelException or another exception.

Last modified: 01 September 2021