Touch events

In order to provide quality support for touch-based user interfaces, touch events offer the ability to interpret finger (or stylus) activity on touch screens or trackpads.

The touch events interfaces are relatively low-level APIs that can be used to support application specific multi-touch interactions such as a two-finger gesture. A multi-touch interaction starts when a finger (or stylus) first touches the contact surface. Other fingers may subsequently touch the surface and optionally move across the touch surface. The interaction ends when the fingers are removed from the surface. During this interaction, an application receives touch events during the start, move and end phases.

Touch events are similar to mouse events except they support simultaneous touches and at different locations on the touch surface. The TouchEvent interface encapsulates all of the touch points that are currently active. The Touch interface, which represents a single touch point, includes information such as the position of the touch point relative to the browser viewport.

Definitions

Surface: The touch-sensitive surface. This may be a screen or trackpad.

Touch point: A point of contact with the surface. This may be a finger (or elbow, ear, nose, whatever, but typically a finger) or stylus.

Interfaces

TouchEvent: Represents an event that occurs when the state of touches on the surface changes.
Touch: Represents a single point of contact between the user and the touch surface.
TouchList: Represents a group of touches; this is used when the user has, for example, multiple fingers on the surface at the same time.

Example

This example tracks multiple touch points at a time, allowing the user to draw in a <canvas> with more than one finger at a time. It will only work on a browser that supports touch events.

Note: The text below uses the term "finger" when describing the contact with the surface, but it could, of course, also be a stylus or other contact method.

Create a canvas

<canvas id="canvas" width="600" height="600" style="border:solid black 1px;">
  Your browser does not support canvas element.
</canvas>
<br>
<button onclick="startup()">Initialize</button>
<br>
Log: <pre id="log" style="border: 1px solid #ccc;"></pre>

Setting up the event handlers

When the page loads, the startup() function shown below should be called by our <body> element's onload attribute (but in the example we use a button to trigger it, due to limitations of the MDN live example system).

function startup() {
  var el = document.getElementsByTagName("canvas")[0];
  el.addEventListener("touchstart", handleStart, false);
  el.addEventListener("touchend", handleEnd, false);
  el.addEventListener("touchcancel", handleCancel, false);
  el.addEventListener("touchmove", handleMove, false);
  log("initialized.");
}

This simply sets up all the event listeners for our <canvas> element so we can handle the touch events as they occur.

Tracking new touches

We'll keep track of the touches in-progress.

var ongoingTouches = [];

When a touchstart event occurs, indicating that a new touch on the surface has occurred, the handleStart() function below is called.

function handleStart(evt) {
  evt.preventDefault();
  log("touchstart.");
  var el = document.getElementsByTagName("canvas")[0];
  var ctx = el.getContext("2d");
  var touches = evt.changedTouches;
        
  for (var i = 0; i < touches.length; i++) {
    log("touchstart:" + i + "...");
    ongoingTouches.push(copyTouch(touches[i]));
    var color = colorForTouch(touches[i]);
    ctx.beginPath();
    ctx.arc(touches[i].pageX, touches[i].pageY, 4, 0, 2 * Math.PI, false);  // a circle at the start
    ctx.fillStyle = color;
    ctx.fill();
    log("touchstart:" + i + ".");
  }
}

This calls event.preventDefault() to keep the browser from continuing to process the touch event (this also prevents a mouse event from also being delivered). Then we get the context and pull the list of changed touch points out of the event's TouchEvent.changedTouches property.

After that, we iterate over all the Touch objects in the list, pushing them onto an array of active touch points and drawing the start point for the draw as a small circle; we're using a 4-pixel wide line, so a 4 pixel radius circle will show up neatly.

Drawing as the touches move

Each time one or more fingers moves, a touchmove event is delivered, resulting in our handleMove() function being called. Its responsibility in this example is to update the cached touch information and to draw a line from the previous position to the current position of each touch.

function handleMove(evt) {
  evt.preventDefault();
  var el = document.getElementsByTagName("canvas")[0];
  var ctx = el.getContext("2d");
  var touches = evt.changedTouches;

  for (var i = 0; i < touches.length; i++) {
    var color = colorForTouch(touches[i]);
    var idx = ongoingTouchIndexById(touches[i].identifier);

    if (idx >= 0) {
      log("continuing touch "+idx);
      ctx.beginPath();
      log("ctx.moveTo(" + ongoingTouches[idx].pageX + ", " + ongoingTouches[idx].pageY + ");");
      ctx.moveTo(ongoingTouches[idx].pageX, ongoingTouches[idx].pageY);
      log("ctx.lineTo(" + touches[i].pageX + ", " + touches[i].pageY + ");");
      ctx.lineTo(touches[i].pageX, touches[i].pageY);
      ctx.lineWidth = 4;
      ctx.strokeStyle = color;
      ctx.stroke();

      ongoingTouches.splice(idx, 1, copyTouch(touches[i]));  // swap in the new touch record
      log(".");
    } else {
      log("can't figure out which touch to continue");
    }
  }
}

This iterates over the changed touches as well, but it looks in our cached touch information array for the previous information about each touch in order to determine the starting point for each touch's new line segment to be drawn. This is done by looking at each touch's Touch.identifier property. This property is a unique integer for each touch, and remains consistent for each event during the duration of each finger's contact with the surface.

This lets us get the coordinates of the previous position of each touch and use the appropriate context methods to draw a line segment joining the two positions together.

After drawing the line, we call Array.splice() to replace the previous information about the touch point with the current information in the ongoingTouches array.

Handling the end of a touch

When the user lifts a finger off the surface, a touchend event is sent. We handle both of these the same way: by calling the handleEnd() function below. Its job is to draw the last line segment for each touch that ended and remove the touch point from the ongoing touch list.

function handleEnd(evt) {
  evt.preventDefault();
  log("touchend");
  var el = document.getElementsByTagName("canvas")[0];
  var ctx = el.getContext("2d");
  var touches = evt.changedTouches;

  for (var i = 0; i < touches.length; i++) {
    var color = colorForTouch(touches[i]);
    var idx = ongoingTouchIndexById(touches[i].identifier);

    if (idx >= 0) {
      ctx.lineWidth = 4;
      ctx.fillStyle = color;
      ctx.beginPath();
      ctx.moveTo(ongoingTouches[idx].pageX, ongoingTouches[idx].pageY);
      ctx.lineTo(touches[i].pageX, touches[i].pageY);
      ctx.fillRect(touches[i].pageX - 4, touches[i].pageY - 4, 8, 8);  // and a square at the end
      ongoingTouches.splice(idx, 1);  // remove it; we're done
    } else {
      log("can't figure out which touch to end");
    }
  }
}

This is very similar to the previous function; the only real differences are that we draw a small square to mark the end and that when we call Array.splice(), we simply remove the old entry from the ongoing touch list, without adding in the updated information. The result is that we stop tracking that touch point.

Handling canceled touches

If the user's finger wanders into browser UI, or the touch otherwise needs to be canceled, the touchcancel event is sent, and we call the handleCancel() function below.

function handleCancel(evt) {
  evt.preventDefault();
  log("touchcancel.");
  var touches = evt.changedTouches;
  
  for (var i = 0; i < touches.length; i++) {
    ongoingTouches.splice(i, 1);  // remove it; we're done
  }
}

Since the idea is to immediately abort the touch, we simply remove it from the ongoing touch list without drawing a final line segment.

Convenience functions

This example uses two convenience functions that should be looked at briefly to help make the rest of the code more clear.

Selecting a color for each touch

In order to make each touch's drawing look different, the colorForTouch() function is used to pick a color based on the touch's unique identifier. This identifier is an opaque number, but we can at least rely on it differing between the currently-active touches.

function colorForTouch(touch) {
  var r = touch.identifier % 16;
  var g = Math.floor(touch.identifier / 3) % 16;
  var b = Math.floor(touch.identifier / 7) % 16;
  r = r.toString(16); // make it a hex digit
  g = g.toString(16); // make it a hex digit
  b = b.toString(16); // make it a hex digit
  var color = "#" + r + g + b;
  log("color for touch with identifier " + touch.identifier + " = " + color);
  return color;
}

The result from this function is a string that can be used when calling <canvas> functions to set drawing colors. For example, for a Touch.identifier value of 10, the resulting string is "#aaa".

Copying a touch object

Some browsers (mobile Safari, for one) re-use touch objects between events, so it's best to copy the bits you care about, rather than referencing the entire object.

function copyTouch(touch) {
  return { identifier: touch.identifier, pageX: touch.pageX, pageY: touch.pageY };
}

Finding an ongoing touch

The ongoingTouchIndexById() function below scans through the ongoingTouches array to find the touch matching the given identifier, then returns that touch's index into the array.

function ongoingTouchIndexById(idToFind) {
  for (var i = 0; i < ongoingTouches.length; i++) {
    var id = ongoingTouches[i].identifier;
    
    if (id == idToFind) {
      return i;
    }
  }
  return -1;    // not found
}

Showing what's going on

function log(msg) {
  var p = document.getElementById('log');
  p.innerHTML = msg + "\n" + p.innerHTML;
}

If your browser supports it, you can see it live.

jsFiddle example

Additional tips

This section provides additional tips on how to handle touch events in your web application.

Handling clicks

Since calling preventDefault() on a touchstart or the first touchmove event of a series prevents the corresponding mouse events from firing, it's common to call preventDefault() on touchmove rather than touchstart. That way, mouse events can still fire and things like links will continue to work. Alternatively, some frameworks have taken to refiring touch events as mouse events for this same purpose. (This example is oversimplified and may result in strange behavior. It is only intended as a guide.)

function onTouch(evt) {
  evt.preventDefault();
  if (evt.touches.length > 1 || (evt.type == "touchend" && evt.touches.length > 0))
    return;

  var newEvt = document.createEvent("MouseEvents");
  var type = null;
  var touch = null;

  switch (evt.type) {
    case "touchstart": 
      type = "mousedown";
      touch = evt.changedTouches[0];
      break;
    case "touchmove":
      type = "mousemove";
      touch = evt.changedTouches[0];
      break;
    case "touchend":        
      type = "mouseup";
      touch = evt.changedTouches[0];
      break;
  }

  newEvt.initMouseEvent(type, true, true, evt.originalTarget.ownerDocument.defaultView, 0,
    touch.screenX, touch.screenY, touch.clientX, touch.clientY,
    evt.ctrlKey, evt.altKey, evt.shiftKey, evt.metaKey, 0, null);
  evt.originalTarget.dispatchEvent(newEvt);
}

Calling preventDefault() only on a second touch

One technique for preventing things like pinchZoom on a page is to call preventDefault() on the second touch in a series. This behavior is not well defined in the touch events spec, and results in different behavior for different browsers (i.e., iOS will prevent zooming but still allow panning with both fingers; Android will allow zooming but not panning; Opera and Firefox currently prevent all panning and zooming.) Currently, it's not recommended to depend on any particular behavior in this case, but rather to depend on meta viewport to prevent zooming.

Specifications

Specification	Status	Comment
Touch Events – Level 2 The definition of 'Touch' in that specification.	Editor's Draft	Added `radiusX`, `radiusY`, `rotationAngle`, `force` properties
Touch Events The definition of 'Touch' in that specification.	Recommendation	Initial definition.

Note that unfortunately touch events may not fire at all on laptops with touch functionality (test page).

Feature	Chrome	Firefox (Gecko)	Internet Explorer	Opera	Safari
Basic support	22.0	18.0 (18.0)^[1]	No support	No support	No support

Feature	Android	Android Webview	Chrome for Android	Firefox Mobile (Gecko)	Firefox OS	IE Mobile	Opera Mobile	Safari Mobile
Basic support	(Yes)	(Yes)	(Yes)	6.0 (6.0)	(Yes)	11	(Yes)	(Yes)

[1] The dom.w3c_touch_events.enabled tri-state preference can be used to disable (0), enable (1), and auto-detect (2) support for standard touch events; by default, they're on auto-detect (2).

Prior to Gecko 12 (Firefox 12.0 / Thunderbird 12.0 / SeaMonkey 2.9), Gecko did not support multi-touch; only one touch at a time was reported.

As of Gecko 24.0 (Firefox 24.0 / Thunderbird 24.0 / SeaMonkey 2.21), the touch events support introduced with Gecko 18.0 (Firefox 18.0 / Thunderbird 18.0 / SeaMonkey 2.15) has been disabled on the desktop version of Firefox (bug 888304), as some popular sites including Google and Twitter were not working properly. Once the bug is fixed, the API will be enabled again. To enable it anyway, open about:config and set the dom.w3c_touch_events.enabled preference to 2. The mobile versions including Firefox for Android and Firefox OS are not affected by this change. Also, the API has been enabled on the Metro-style version of Firefox for Windows 8.

Prior to Gecko 6.0 (Firefox 6.0 / Thunderbird 6.0 / SeaMonkey 2.3), Gecko offered a proprietary touch event API. That API is now deprecated; you should switch to this one.