build your own touch slider with hammer.js

There are countless pre-built libraries for sliders, carousels, and everything in-between, and it seems like a major one comes out every other year or so. While most of those work fine in a vacuum, your choices quickly narrow when you need to add part or a whole slider in the middle of an existing mobile app without invasive JavaScript, CSS, and bloated markup. In these scenarios, HammerJS comes to the rescue by making it easier and more time-effective to build one from scratch than you may have thought.

Real-world Example

Mobile interface for Real Thread: a deceptive amount of JavaScript

I ran into this problem developing Real Thread’s mobile interface. Within that box where the shirt image sits, you’ll find:

  • Angular code that swaps background images out with the shirt color you’ve selected
  • A file uploader handled by DropzoneJS
  • Touch events that allow you to move, pinch–scale, and delete the artwork you’ve uploaded

The designs called for a way to swipe left and right between the front and the back of the shirt — in other words: a slider. But with everything else going on in this markup, this isn’t a typical use case for the normal slider libraries meant for marketing pages. So for my own sanity — and to keep the app running lean — I made my own with only a little bit of vanilla JavaScript and HammerJS.

Is HammerJS right for me?

HammerJS is a 7kb multi-touch library that mimics vanilla JavaScript mouse and touch events. You still get the expected stuff like clientX and clientY to track current position, but you get access to new events like swipe, pinch, and rotate, along with new properties like deltaX, deltaY, distance, direction, rotation, and scale, to name a few.

Before I waste any more of your time, ask yourself: which sounds more like me?

  1. I want to copy & paste stuff and just have it work. Sorry; HammerJS is only a minimal event library and has nothing to do with the DOM. You’ll have to write some JavaScript to make it work, or find a slider that works out-of-box.
  2. I need to incorporate touch or gestures into my app, but it’s a weird interaction and I’m thinking about just writing it from scratch. HammerJS may be what you need — it not only provides a versatile set of events and event properties, but also allows you to register custom touch events as well. Anyone with basic JavaScript experience can pick it up because it behaves like a normal event callback.

Check out their examples to play around with it.

The Markup

It’s a basic requirement of sliders to actually slide (i.e., not be stacked layers that simply fade opacity). The easiest way to accomplish this is to have one container that is as wide as all the slides, side-by-side (.slider), within a container that crops everything to width: 100% (.has-slider) so no scrolling happens. The markup should be something similar to:

<div class="has-slider">
  <div class="slider">
    <div class="slider-panel"></div>
    <div class="slider-panel"></div>
    <div class="slider-panel"></div>

And the important CSS looks like (Sass notation):

  overflow: hidden
  width: 100%
  width: 300% // 100% * 3 slides

Since CSS transforms are our most performant way to move something around the page, we’ll take advantage of transform: translateX() to do the heavy lifting for us. Since .slider is the entire width of all 3 slides (300%), we’ll apply translateX() to that container with JavaScript to move the proper .slider-panel we want to show within the viewport.

Of course we’ll need more CSS than this to style it properly, but I’ll skip explaining the rest of it. Now we need is a little bit of JavaScript to get it working.

Working with Hammer

I prefer working with managers. Compared to the standalone new Hammer() syntax, managers give you the flexibility of tying multiple events together (such as pinch and pan) on the same element. For example, if you wanted to recreate the pinch + rotate gesture in the latest iOS Photos app, you could do that in HammerJS using a manager. Managers don’t have any advantages if you’re only listening for one event (like in our example), but I do this by default because it allows me to track additional events later if I need to without rearranging my code.

Step 1: Adding the listener

If we add the following code to our project:

We don’t have a working slider yet, but in your console you’ll get an object with lots of useful things like deltaX, deltaY, rotation, scale, etc. One way to visualize that output is like this:

Since we’re building a horizontal slider, deltaX will be the most useful. This tracks the net horizontal distance between where the gesture started and where the last registered touch event was. To illustrate how handy this is, let’s compare what we’d have to do without HammerJS:

Without HammerJS

  1. Set up different event listeners for touchstart, touchmove, touchend, and touchcancel
  2. Store the values for the initial clientX and clientY for the first pointer in touchstart in a variable outside the scope of your event listeners
  3. Every touchmove event, calculate the stored values minus the current clientX and clientY to get the net distance
  4. Find a workaround for IE & Edge (OSX Safari and Firefox don’t support it either, but effectively none of your users will need touch support for those browsers)

With HammerJS

  1. Make one pan event listener
  2. Net distance already done for you with deltaX and deltaY

It’s your choice, but I’m sticking with HammerJS.

Step 2: Applying the transform

Pressing forward, let’s apply a CSS transform to our element (and save it in a variable so we don’t repeat ourselves #DRYlife):

Now we have the slider moving 1:1 with our finger, but it’s not “snapping” into place yet, and it seems to always be stuck on the first slide. But we’re getting there, and it only took us 7 lines of JavaScript!

Step 3: Keeping track of which slide we’re on

Things will get a little more complicated here, but this is the last major addition. Let’s add an activeSlide variable to keep track of which slide we’re on, and then have it snap into place when we’re done with the event with a new function:

Now this changes slides on release, but it has 2 key problems: it jumps between slides rather than animating smoothly, and it’s also too sensitive — it changes slides if you drag it even 1 pixel in either direction.

Wrapping Up

The only things remaining for this to be a full-fledged slider are adding pagination, and applying a CSS transition at just the right time to animate smoothly when snapping to a slide. The reason you don’t want to have a transition applied all the time is this will really make your touch events feel sluggish when dragging:

Instead, we’ll stick the transition on an .is-animating state class in our CSS rather than .slider. And we’ll let JavaScript add and remove the class at the appropriate times (i.e., only between touch events). The final result, with pagination, looks like this:

If you inspect the JS for this last CodePen, it’s a little more object-oriented but is still the same code we’ve been working with all along. The only thing remaining would be to add a “flick” listener to the slider, in case a user swiped their finger quickly. velocityX helps us out here. This can be added by modifying the .isFinal conditional since we only want to register a flick at the end of a gesture:

This is all merely proof of concept, and needs refactoring to fit into your setup. But anyways, there you have it: a fairly decent slider that doesn’t get in the way of your application.