Using WebSocketStream to write a client
Non-standard: This feature is non-standard and is not on a standards track. Do not use it on production sites facing the Web: it will not work for every user. There may also be large incompatibilities between implementations and the behavior may change in the future.
The WebSocketStream
API is a Promise
-based alternative to WebSocket
for creating and using client-side WebSocket connections. WebSocketStream
uses the Streams API to handle receiving and sending messages, meaning that socket connections can take advantage of stream backpressure automatically (no additional action required by the developer), regulating the speed of reading or writing to avoid bottlenecks in the application.
This article explains how to use the WebSocketStream
API to create a WebSocket client.
Feature detection
To check whether the WebSocketStream
API is supported, you can use the following:
if ("WebSocketStream" in self) {
// WebSocketStream is supported
}
Creating a WebSocketStream object
To create a WebSocket client, you first need to create a new WebSocketStream
instance using the WebSocketStream()
constructor. In its simplest form, it takes the URL of the WebSocket server as an argument:
const wss = new WebSocketStream("wss://example.com/wss");
It can also take an options object containing custom protocols and/or an AbortSignal
(see Closing the connection):
const controller = new AbortController();
const queueWSS = new WebSocketStream("wss://example.com/queue", {
protocols: ["amqp", "mqtt"],
signal: controller.signal,
});
Sending and receiving data
The WebSocketStream
instance has an opened
property — this returns a promise that fulfills with an object containing a ReadableStream
and a WritableStream
instance once the WebSocket connection is opened successfully:
const { readable, writable } = await wss.opened;
Calling getReader()
and getWriter()
on these objects provides us with a ReadableStreamDefaultReader
and a WritableStreamDefaultWriter
respectively, which can be used to read from and write to the socket connection:
const reader = readable.getReader();
const writer = writable.getWriter();
To write data to the socket, you can use WritableStreamDefaultWriter.write()
:
writer.write("My message");
To read data from the socket, you can continuously call ReadableStreamDefaultReader.read()
until the stream has finished, which is indicated by done
being true
:
while (true) {
const { value, done } = await reader.read();
if (done) {
break;
}
// Process value in some way
}
The browser automatically controls the rate at which the client receives and sends data by applying backpressure when needed. If data is arriving faster than the client can read()
it, the underlying Streams API exerts backpressure on the server. In addition, write()
operations will only proceed if it is safe to do so.
Closing the connection
With WebSocketStream
, the information previously available via the WebSocket
close
and error
events is now available via the closed
property — this returns a promise that fulfills with an object containing the closing code (see the full list of CloseEvent
status codes) and reason indicating why the server closed the connection:
const { code, reason } = await wss.closed;
As mentioned earlier, the WebSocket connection can be closed using an AbortController
. The necessary AbortSignal
is passed to the WebSocketStream
constructor during creation, and AbortController.abort()
can then be called when required:
const controller = new AbortController();
const wss = new WebSocketStream("wss://example.com/wss", {
signal: controller.signal,
});
// some time later
controller.abort();
Alternatively you can use the WebSocketStream.close()
method to close a connection. This is mainly used if you wish to specify a custom code and/or reason:
wss.close({
code: 4000,
reason: "Night draws to a close",
});
Note: Depending on the server setup and status code you use, the server may choose to ignore a custom code in favor of a valid code that is correct for the closing reason.
A complete sample client
To demonstrate basic usage of WebSocketStream
, we've created a sample client. You can see the full listing at the bottom of the article, and follow along with the explanation below.
Note: To get the example working, you'll also need a server component. We wrote our client to work along with the Deno server explained in Writing a WebSocket server in JavaScript (Deno), but any compatible server will do.
The HTML for the demo is as follows. It includes informational <h2>
and <p>
elements, a <button>
to close the WebSocket connection that is initially disabled, and a <div>
for us to write output messages into.
<h2>WebSocketStream Test</h2>
<p>Sends a ping every five seconds</p>
<button id="close" disabled>Close socket connection</button>
<div id="output"></div>
Now on to the JavaScript. First we grab references to the output <div>
and the close <button>
, and define a utility function that writes messages to the <div>
:
const output = document.querySelector("#output");
const closeBtn = document.querySelector("#close");
function writeToScreen(message) {
const pElem = document.createElement("p");
pElem.textContent = message;
output.appendChild(pElem);
}
Next, we create an if ... else
structure to feature detect WebSocketStream
and output an informative message on non-supporting browsers:
if (!("WebSocketStream" in self)) {
writeToScreen("Your browser does not support WebSocketStream");
} else {
// supporting code path
}
In the supporting code path, we begin by defining a variable containing the WebSocket server URL, and constructing a new WebSocketServer
instance:
const wsURL = "ws://127.0.0.1/";
const wss = new WebSocketStream(wsURL);
Note:
Best practice is to use secure WebSockets (wss://
) in production apps. However, in this demo we are connecting to localhost, therefore we need to use the non-secure WebSocket protocol (ws://
) for the example to work.
The main bulk of our code is contained within the start()
function, which we define and then immediately invoke. We await the opened
promise, then after it fulfills write a message to let the reader know the connection is successful and create ReadableStreamDefaultReader
and WritableStreamDefaultWriter
instances from the returned readable
and writable
properties.
Next, we create a start()
function that sends "ping" messages to the server and receives "pong" messages back, and invoke it. In the function body we await the wss.opened
promise and create a reader and writer from its fulfillment values. Once the socket is open, we communicate that to the user and enable the close button. Next, we write()
a "ping"
value to the socket and communicate that to the user. At this point, the server will respond with a "pong"
message. We await the read()
of the response, communicate it to the user, then write another "ping"
to the server after a timeout of 5 seconds. This continues the "ping"
/"pong"
loop indefinitely.
async function start() {
const { readable, writable } = await wss.opened;
writeToScreen("CONNECTED");
closeBtn.disabled = false;
const reader = readable.getReader();
const writer = writable.getWriter();
writer.write("ping");
writeToScreen("SENT: ping");
while (true) {
const { value, done } = await reader.read();
writeToScreen(`RECEIVED: ${value}`);
if (done) {
break;
}
setTimeout(async () => {
try {
await writer.write("ping");
writeToScreen("SENT: ping");
} catch (e) {
writeToScreen(`Error writing to socket: ${e.message}`);
}
}, 5000);
}
}
start();
Note:
The setTimeout()
function wraps the write()
call in a try...catch
block to handle any errors that can arise if the application tries to write to the stream after it has been closed.
We now include a promise-style code section to inform the user of the code and reason if the WebSocket connection is closed, as signalled by the closed
promise fulfilling:
wss.closed.then((result) => {
writeToScreen(
`DISCONNECTED: code ${result.closeCode}, message "${result.reason}"`,
);
console.log("Socket closed", result.closeCode, result.reason);
});
Finally, we add an event listener to the close button that closes the connection using the close()
method, with a code and custom reason. The function also disables the close button — we don't want users to press it once the connection is already closed.
closeBtn.addEventListener("click", () => {
wss.close({
code: 1000,
reason: "That's all folks",
});
closeBtn.disabled = true;
});
Full listing
<!doctype html>
<html lang="en">
<head>
<meta charset="utf-8" />
<title>WebSocketStream Test</title>
</head>
<body>
<h2>WebSocketStream Test</h2>
<p>Sends a ping every five seconds</p>
<button id="close" disabled>Close socket connection</button>
<div id="output"></div>
<script>
const output = document.querySelector("#output");
const closeBtn = document.querySelector("#close");
function writeToScreen(message) {
const pElem = document.createElement("p");
pElem.textContent = message;
output.appendChild(pElem);
}
if (!("WebSocketStream" in self)) {
writeToScreen("Your browser does not support WebSocketStream");
} else {
const wsURL = "ws://127.0.0.1/";
const wss = new WebSocketStream(wsURL);
console.log(wss.url);
async function start() {
const { readable, writable, extensions, protocol } = await wss.opened;
writeToScreen("CONNECTED");
closeBtn.disabled = false;
const reader = readable.getReader();
const writer = writable.getWriter();
writer.write("ping");
writeToScreen("SENT: ping");
while (true) {
const { value, done } = await reader.read();
writeToScreen(`RECEIVED: ${value}`);
if (done) {
break;
}
setTimeout(() => {
writer.write("ping");
writeToScreen("SENT: ping");
}, 5000);
}
}
start();
wss.closed.then((result) => {
writeToScreen(
`DISCONNECTED: code ${result.closeCode}, message "${result.reason}"`,
);
console.log("Socket closed", result.closeCode, result.reason);
});
closeBtn.addEventListener("click", () => {
wss.close({
code: 1000,
reason: "That's all folks",
});
closeBtn.disabled = true;
});
}
</script>
</body>
</html>