WebRTC codelab

##Overview

WebRTC enables real-time communication in the browser.

This tutorial explains how to build a simple video and text chat application.

For more information about WebRTC, see Getting started with WebRTC on HTML5 Rocks.

You can submit feedback about the WebRTC APIs at bit.ly/webrtcfeedback.

Prerequisites

Basic knowledge:

1. HTML, CSS and JavaScript
2. git
3. Chrome DevTools

Experience of Node.js and socket.io would also be useful.

Installed on your development machine:

1. Google Chrome or Firefox.
2. Code editor.
3. Web cam.
4. git, in order to get the source code.
5. The source code.
6. Node.js with socket.io and node-static. (Node.js hosting would also be an advantage -- see below for some options.)

The instructions in this codelab assume you are using Mac OS, Linux or Windows. Unless you know what you're doing, it's probably easier not to attempt this codelab from a Chromebook!

It would also be useful to have an Android device with Google Chrome installed in order to try out the examples on mobile. To run WebRTC APIs on Chrome for Android, you must enable WebRTC from the chrome://flags page.

Running The Examples

To run any of the examples provided in the complete directory of this repo simply cd into the specific step directory you'd like to see locally and run

node server.js

That directory's demo will then be live here.

Step 0: Get the code

Using git, clone the codelab repository onto your development computer. If you haven't used git before, there are several tutorials and reference guides available from the git website.

Step 1: Create a blank HTML5 document

Complete example: complete/step1.

1. Create a bare-bones HTML document.
2. Test it out locally (see instructions above on running demos).
   • Why is there nothing on the screen?

Step 2: Get video from your webcam

Complete example: complete/step2.

1. Add a video element to your page.

2. Add the following JavaScript to the script element on your page, to enable getUserMedia() to set the source of the video from the web cam:

    navigator.getUserMedia = navigator.getUserMedia ||
      navigator.webkitGetUserMedia || navigator.mozGetUserMedia;
   
    var constraints = {video: true};
   
    function successCallback(localMediaStream) {
      window.stream = localMediaStream; // stream available to console
      var video = document.querySelector("video");
      video.src = window.URL.createObjectURL(localMediaStream);
      video.play();
    }
   
    function errorCallback(error){
      console.log("navigator.getUserMedia error: ", error);
    }
   
    navigator.getUserMedia(constraints, successCallback, errorCallback);
   

3. Test it out locally (see instructions above on running demos).

Explanation

getUserMedia is called like this:

navigator.getUserMedia(constraints, successCallback, errorCallback);

The constraints argument allows us to specify the media to get, in this case video only:

var constraints = {"video": true}

If successful, the video stream from the webcam is set as the source of the video element:

function successCallback(localMediaStream) {
  window.stream = localMediaStream; // stream available to console
  var video = document.querySelector("video");
  video.src = window.URL.createObjectURL(localMediaStream);
  video.play();
}

Bonus points

1. Inspect the stream object from the console.
2. Try calling stream.stop().
3. What does stream.getVideoTracks() return?
4. Look at the constraints object: what happens when you change it to {audio: true, video: true}?
5. What size is the video element? How can you get the video's natural size from JavaScript? Use the Chrome Dev Tools to check. Use CSS to make the video full width. How would you ensure the video is no higher than the viewport?
6. Try adding CSS filters to the video element (more ideas here).
7. Try changing constraints: see the sample at simpl.info/getusermedia/constraints.

For example:

video {
  filter: hue-rotate(180deg) saturate(200%);
  -moz-filter: hue-rotate(180deg) saturate(200%);
  -webkit-filter: hue-rotate(180deg) saturate(200%);
}

Step 3: Stream video with RTCPeerConnection

Complete example: complete/step3.

RTCPeerConnection is the WebRTC API for video and audio calling.

This example sets up a connection between two peers on the same page. Not much use, but good for understanding how RTCPeerConnection works!

1. Get rid of the JavaScript you've entered so far -- we're going to do something different!

2. Edit the HTML so there are two video elements and three buttons: Start, Call and Hang Up:

    <video id="localVideo" autoplay></video>
    <video id="remoteVideo" autoplay></video>
   
    <div>
      <button id="startButton">Start</button>
      <button id="callButton">Call</button>
      <button id="hangupButton">Hang Up</button>
    </div>
   

3. Add the JavaScript from complete/step3/index.html.

4. Test it out locally (see instructions above on running demos).

Explanation

This code does a lot!

• Get and share local and remote descriptions: metadata about local media in SDP¹ format.
• Get and share ICE² candidates: network information.
• Pass the local stream to the remote RTCPeerConnection.

Bonus points

1. Take a look at chrome://webrtc-internals. (There is a full list of Chrome URLs at chrome://about.)
2. Style the page with CSS:
   • Put the videos side by side.
   • Make the buttons the same width, with bigger text.
   • Make sure it works on mobile.
3. From the Chrome Dev Tools console, inspect localStream, localPeerConnection and remotePeerConnection.
4. Take a look at localPeerConnection.localDescription. What does SDP format look like?

Step 4: Stream arbitrary data with RTCDataChannel

Complete example: complete/step4.

For this step, we'll use RTCDataChannel to send text between two textareas on the same page. Not very useful, except to demonstrate how the API works.

1. Create a new document and add the following HTML:

    <textarea id="dataChannelSend" disabled></textarea>
    <textarea id="dataChannelReceive" disabled></textarea>
   
    <div id="buttons">
      <button id="startButton">Start</button>
      <button id="sendButton">Send</button>
      <button id="closeButton">Stop</button>
    </div>
   

2. Add the JavaScript from complete/step4/index.html.

3. Test it out locally (see instructions above on running demos).

Explanation

This code uses RTCPeerConnection and RTCDataChannel to enable exchange of text messages.

Most of the code in this section is the same as for the RTCPeerConnection example. Additional code is as follows:

function sendData(){
  var data = document.getElementById("dataChannelSend").value;
  sendChannel.send(data);
}
...
localPeerConnection = new RTCPeerConnection(servers,
  {optional: [{RtpDataChannels: true}]});
sendChannel = localPeerConnection.createDataChannel("sendDataChannel",
  {reliable: false});
sendChannel.onopen = handleSendChannelStateChange;
sendChannel.onclose = handleSendChannelStateChange;
...
remotePeerConnection = new RTCPeerConnection(servers,
  {optional: [{RtpDataChannels: true}]});
function gotReceiveChannel(event) {
  receiveChannel = event.channel;
  receiveChannel.onmessage = gotMessage;
}
...
remotePeerConnection.ondatachannel = gotReceiveChannel;
function gotMessage(event) {
  document.getElementById("dataChannelReceive").value = event.data;
}

The syntax of RTCDataChannel is deliberately similar to WebSocket, with a send() method and a message event.

Notice the use of constraints.

Bonus points

1. Try out RTCDataChannel file sharing with Sharefest. When would RTCDataChannel need to provide reliable delivery of data, and when might performance be more important -- even if that means losing some data?
2. Use CSS to improve page layout, and add a placeholder attribute to the dataChannelReceive textarea.
3. Test the page on a mobile device.

Step 5: Set up a signaling server and exchange messages

Complete example: complete/step5.

RTCPeerConnection instances need to exchange metadata in order to set up and maintain a WebRTC 'call':

• Candidate (network) information.
• Offer and answer messages providing information about media such as resolution and codecs.

In other words, an exchange of metadata is required before peer-to-peer audio, video or data streaming can take place. This process is called signaling.

In the examples already completed, the 'sender' and 'receiver' RTCPeerConnection objects are on the same page, so signaling is simply a matter of passing objects between methods.

In a real world application, the sender and receiver RTCPeerConnections are not on the same page, and we need a way for them to communicate metadata.

For this, we use a signaling server: a server that can exchange messages between a WebRTC app (client) running in one browser and a client in another browser. The actual messages are stringified JavaScript objects.

To reiterate: metadata exchange between WebRTC clients (via a signaling server) is required for RTCPeerConnection to do audio, video and data streaming (peer to peer).

In this step we'll build a simple Node.js signaling server, using the socket.io Node module and JavaScript library for messaging. Experience of Node.js and socket.io will be useful, but not crucial -- the messaging components are very simple. In this example, the server (the Node app) is server.js and the client (the web app) is index.html.

The Node server application in this step has two tasks.

To act as a messaging intermediary:

socket.on('message', function (message) {
  log('Got message: ', message);
  socket.broadcast.emit('message', message);
});

To manage WebRTC video chat 'rooms':

if (numClients == 0){
  socket.join(room);
  socket.emit('created', room);
} else if (numClients == 1) {
  io.sockets.in(room).emit('join', room);
  socket.join(room);
  socket.emit('joined', room);
} else { // max two clients
  socket.emit('full', room);
}

Our simple WebRTC application will only permit a maximum of two peers to share a room.

1. Ensure you have Node, socket.io and node-static installed. Node can be downloaded from nodejs.org; installation is straightforward and quick. To install socket.io and node-static, run Node Package Manager from a terminal in your application directory:

    npm install socket.io
    npm install node-static
   

(You don't need to learn about node-static for this exercise: it just makes the server simpler.)

2. Using the code from the step 5 directory, run the server (server.js). To start the server, run the following command from a terminal in your application directory:

    node server.js
   

3. From your browser, open localhost:2013. Open a new tab page or window in any browser and open localhost:2013 again, then repeat.

4. To see what's happening, check the Chrome DevTools console (Command-Option-J, or Ctrl-Shift-J).

Bonus points

1. Try deploying your messaging server so you can access it via a public URL. (Free trials and easy deployment options for Node are available on several hosting sites including nodejitsu, heroku and nodester.)

2. What alternative messaging mechanisms are available? (Take a look at apprtc.appspot.com.) What problems might we encounter using 'pure' WebSocket? (Take a look at Arnout Kazemier's presentation, WebSuckets.)

3. What issues might be involved with scaling this application? Can you develop a method for testing thousands or millions of simultaneous room requests.

4. Try out Remy Sharp's tool nodemon. This monitors any changes in your Node.js application and automatically restarts the server when changes are saved.

5. This app uses a JavaScript prompt to get a room name. Work out a way to get the room name from the URL, for example localhost:2013/foo would give the room name foo.

Step 6: RTCPeerConnection with messaging

Complete example: complete/step6.

In this step, we build a video chat client, using the signaling server we created in Step 5 and the RTCPeerConnection code from Step 3.

This step users adapter.js. This is a JavaScript shim, maintained by Google, that abstracts away browser differences and spec changes.

1. Ensure you have Node, socket.io and node-static installed and working. If in doubt, try the code in Step 5.

2. Using the code from the step 6 directory, run the server (server.js). To start the server, run the following from your application directory:

    node server.js
   

3. From your browser, open localhost:2013. Open a new tab page or window and open localhost:2013 again.

4. View logging from the Chrome DevTools console and WebRTC debug information from chrome://webrtc-internals.

Bonus points

1. This application only supports one-to-one video chat. How might you change the design to enable more than one person to share the same video chat room? (Look at talky.io for an example of this in action.)

2. The example has the room name foo hard coded. What would be the best way to enable other room names?

3. Does the app work on mobile? Try it out on a phone, on a 7" and a 10" tablet. What layout, UI and UX changes would be required to ensure a good mobile experience?

4. Deploy your app at a public URL (see above for hosting options). Try different network configurations, for example with one user on wifi and another on 3G. Any problems?

5. How would users share the room name? Try to build an alternative to sharing room names.

Step 7: File sharing using RTCDataChannel

Complete example: complete/step7.

Step 5 showed how to exchange text messages using RTCDataChannel.

This step makes it possible to share entire files: in this example, photos captured via getUserMedia().

The core parts of this step are as follows:

1. Establish a data channel. Note that we don't add any media streams to the peer connection in this step.

2. Grab the user's webcam video stream with getUserMedia():

    var video = document.getElementById('video');
    getUserMedia({video: true}, function(stream) {
        video.src = window.URL.createObjectURL(stream);
    }, getMediaErrorCallback);
   

3. When the user clicks on the Snap button, get a snapshot (a video frame) from the video stream and display it:

    var canvas = document.querySelector('canvas);
    var context = photo.getContext('2d');
    context.drawImage(video, 0, 0, canvasWidth, canvasHeight);
   

4. When the user clicks on the Send button, convert the image to bytes and send them over a data channel:

    // Split data channel message in chunks
    var CHUNK_LEN = 64000;
    // Get the image bytes and calculate the number of chunks
    var img = canvas.getImageData(0, 0, canvasWidth, canvasHeight);
    var len = img.data.byteLength;
    var numChunks = len / CHUNK_LEN | 0;
    // Let the other peer know in advance how many bytes to expect in total
    dataChannel.send(len);
    // Split the photo in chunks and send it over the data channel
    for (var i = 0; i < n; i++) {
        var start = i * CHUNK_LEN;
        var end = (i+1) * CHUNK_LEN;
        dataChannel.send(img.data.subarray(start, end));
    }
    // Send the reminder, if any
    if (len % CHUNK_LEN) {
        dataChannel.send(img.data.subarray(n * CHUNK_LEN));
    }
   

5. The receiving side converts data channel message bytes back to an image and displays this to the user:

    var buf, count;
    // dc is an RTCDataChannel initialized somewhere else
    dc.onmessage = function(event) {
      if (typeof event.data === 'string') {
          buf = new Uint8ClampedArray(parseInt(event.data));
          count = 0;
          console.log('Expecting a total of ' + buf.byteLength + ' bytes');
          return;
      }
      var data = new Uint8ClampedArray(event.data);
      buf.set(data, count);
      count += data.byteLength;
      if (count == buf.byteLength) {
          // we're done: all data chunks have been received
          renderPhoto(buf);
      }
    }
    function renderPhoto(data) {
        var photo = document.createElement('canvas');
        trail.insertBefore(photo, trail.firstChild);
        var canvas = photo.getContext('2d');
        var img = canvas.createImageData(300, 150);
        img.data.set(data);
        canvas.putImageData(img, 0, 0);
    }
   

Playing with the sample code

1. As in previous steps, start a local server with node server.js and navigate to http://localhost:2013.
2. Allow the app to use your camera.
3. The app will create a random room. Copy and paste the URL into a new window.
4. Click the 'Snap & send' button and see what happens in the other window.

Bonus points

1. Try combinations of different browsers, e.g. Firefox or Opera.
2. Send the room URL to another person in your codelab. You should be able to connect via WiFi.
3. Deploy the app to a public URL and try 'snap & sending' photos with your friends.
4. How can you change the code to make it possible to share any file type with your peers?

Step 8: Use a WebRTC library: SimpleWebRTC

Complete example: complete/step8.

Abstraction libraries such as SimpleWebRTC make it simple to create WebRTC applications. (You can find other examples of WebRTC libraries at bit.ly/webrtcwebaudio.)

1. Create a new document using the code from complete/step8/index.html.
2. Open the document in multiple windows or tab.

Bonus points

1. Find a WebRTC library for RTCDataChannel. (Hint: there's one named PeerJS!)
2. Set up your own signaling server using the SimpleWebRTC server signalmaster.

Step 9: Support three or more users in the same room

This is a DIY step!

How would you implement an app combining video chat and file exchange for three or more users?

How many users can an app sustain in full mesh mode, i.e. with a peer connection between every user?

What are the alternatives? (Take a look at the HTML5 Rocks WebRTC infrastructure article for inspiration.)

Bonus points

1. So far, the app hasn't had any work done on layout. Sort it out! Make sure your app works well across different devices.

Footnotes

1. Session Description Protocol ↩

2. Interactive Connectivity Establishment Protocol ↩