Note: keyCode is now deprecated.
Multiple keystroke detection is easy if you understand the concept
The way I do it is like this:
var map = {}; // You could also use an array
onkeydown = onkeyup = function(e){
e = e || event; // to deal with IE
map[e.keyCode] = e.type == 'keydown';
/* insert conditional here */
}
This code is very simple: Since the computer only passes one keystroke at a time, an array is created to keep track of multiple keys. The array can then be used to check for one or more keys at once.
Just to explain, let’s say you press A and B, each fires a keydown
event that sets map[e.keyCode]
to the value of e.type == keydown
, which evaluates to either true or false. Now both map[65]
and map[66]
are set to true
. When you let go of A
, the keyup
event fires, causing the same logic to determine the opposite result for map[65]
(A), which is now false, but since map[66]
(B) is still “down” (it hasn’t triggered a keyup event), it remains true.
The map
array, through both events, looks like this:
// keydown A
// keydown B
[
65:true,
66:true
]
// keyup A
// keydown B
[
65:false,
66:true
]
There are two things you can do now:
A) A Key logger (example) can be created as a reference for later when you want to quickly figure out one or more key codes. Assuming you have defined an html element and pointed to it with the variable element
.
element.innerHTML = '';
var i, l = map.length;
for(i = 0; i < l; i ++){
if(map[i]){
element.innerHTML += '<hr>' + i;
}
}
Note: You can easily grab an element by its id
attribute.
<div id="element"></div>
This creates an html element that can be easily referenced in javascript with element
alert(element); // [Object HTMLDivElement]
You don’t even have to use document.getElementById()
or $()
to grab it. But for the sake of compatibility, use of jQuery’s $()
is more widely recommended.
Just make sure the script tag comes after the body of the HTML. Optimization tip: Most big-name websites put the script tag after the body tag for optimization. This is because the script tag blocks further elements from loading until its script is finished downloading. Putting it ahead of the content allows the content to load beforehand.
B (which is where your interest lies) You can check for one or more keys at a time where /*insert conditional here*/
was, take this example:
if(map[17] && map[16] && map[65]){ // CTRL+SHIFT+A
alert('Control Shift A');
}else if(map[17] && map[16] && map[66]){ // CTRL+SHIFT+B
alert('Control Shift B');
}else if(map[17] && map[16] && map[67]){ // CTRL+SHIFT+C
alert('Control Shift C');
}
Edit: That isn’t the most readable snippet. Readability’s important, so you could try something like this to make it easier on the eyes:
function test_key(selkey){
var alias = {
"ctrl": 17,
"shift": 16,
"A": 65,
/* ... */
};
return key[selkey] || key[alias[selkey]];
}
function test_keys(){
var keylist = arguments;
for(var i = 0; i < keylist.length; i++)
if(!test_key(keylist[i]))
return false;
return true;
}
Usage:
test_keys(13, 16, 65)
test_keys('ctrl', 'shift', 'A')
test_key(65)
test_key('A')
Is this better?
if(test_keys('ctrl', 'shift')){
if(test_key('A')){
alert('Control Shift A');
} else if(test_key('B')){
alert('Control Shift B');
} else if(test_key('C')){
alert('Control Shift C');
}
}
(end of edit)
This example checks for CtrlShiftA, CtrlShiftB, and CtrlShiftC
It’s just as simple as that 🙂
Notes
Keeping Track of KeyCodes
As a general rule, it is good practice to document code, especially things like Key codes (like // CTRL+ENTER
) so you can remember what they were.
You should also put the key codes in the same order as the documentation (CTRL+ENTER => map[17] && map[13]
, NOT map[13] && map[17]
). This way you won’t ever get confused when you need to go back and edit the code.
A gotcha with if-else chains
If checking for combos of differing amounts (like CtrlShiftAltEnter and CtrlEnter), put smaller combos after larger combos, or else the smaller combos will override the larger combos if they are similar enough. Example:
// Correct:
if(map[17] && map[16] && map[13]){ // CTRL+SHIFT+ENTER
alert('Whoa, mr. power user');
}else if(map[17] && map[13]){ // CTRL+ENTER
alert('You found me');
}else if(map[13]){ // ENTER
alert('You pressed Enter. You win the prize!')
}
// Incorrect:
if(map[17] && map[13]){ // CTRL+ENTER
alert('You found me');
}else if(map[17] && map[16] && map[13]){ // CTRL+SHIFT+ENTER
alert('Whoa, mr. power user');
}else if(map[13]){ // ENTER
alert('You pressed Enter. You win the prize!');
}
// What will go wrong: When trying to do CTRL+SHIFT+ENTER, it will
// detect CTRL+ENTER first, and override CTRL+SHIFT+ENTER.
// Removing the else's is not a proper solution, either
// as it will cause it to alert BOTH "Mr. Power user" AND "You Found Me"
Gotcha: “This key combo keeps activating even though I’m not pressing the keys”
When dealing with alerts or anything that takes focus from the main window, you might want to include map = []
to reset the array after the condition is done. This is because some things, like alert()
, take the focus away from the main window and cause the ‘keyup’ event to not trigger. For example:
if(map[17] && map[13]){ // CTRL+ENTER
alert('Oh noes, a bug!');
}
// When you Press any key after executing this, it will alert again, even though you
// are clearly NOT pressing CTRL+ENTER
// The fix would look like this:
if(map[17] && map[13]){ // CTRL+ENTER
alert('Take that, bug!');
map = {};
}
// The bug no longer happens since the array is cleared
Gotcha: Browser Defaults
Here’s an annoying thing I found, with the solution included:
Problem: Since the browser usually has default actions on key combos (like CtrlD activates the bookmark window, or CtrlShiftC activates skynote on maxthon), you might also want to add return false
after map = []
, so users of your site won’t get frustrated when the “Duplicate File” function, being put on CtrlD, bookmarks the page instead.
if(map[17] && map[68]){ // CTRL+D
alert('The bookmark window didn\'t pop up!');
map = {};
return false;
}
Without return false
, the Bookmark window would pop up, to the dismay of the user.
The return statement (new)
Okay, so you don’t always want to exit the function at that point. That’s why the event.preventDefault()
function is there. What it does is set an internal flag that tells the interpreter to not allow the browser to run its default action. After that, execution of the function continues (whereas return
will immediately exit the function).
Understand this distinction before you decide whether to use return false
or e.preventDefault()
event.keyCode
is deprecated
User SeanVieira pointed out in the comments that event.keyCode
is deprecated.
There, he gave an excellent alternative: event.key
, which returns a string representation of the key being pressed, like "a"
for A, or "Shift"
for Shift.
I went ahead and cooked up a tool for examining said strings.
element.onevent
vs element.addEventListener
Handlers registered with addEventListener
can be stacked, and are called in the order of registration, while setting .onevent
directly is rather aggressive and overrides anything you previously had.
document.body.onkeydown = function(ev){
// do some stuff
ev.preventDefault(); // cancels default actions
return false; // cancels this function as well as default actions
}
document.body.addEventListener("keydown", function(ev){
// do some stuff
ev.preventDefault() // cancels default actions
return false; // cancels this function only
});
The .onevent
property seems to override everything and the behavior of ev.preventDefault()
and return false;
can be rather unpredictable.
In either case, handlers registered via addEventlistener
seem to be easier to write and reason about.
There is also attachEvent("onevent", callback)
from Internet Explorer’s non-standard implementation, but this is beyond deprecated and doesn’t even pertain to JavaScript (it pertains to an esoteric language called JScript). It would be in your best interest to avoid polyglot code as much as possible.
A helper class
To address confusion/complaints, I’ve written a “class” that does this abstraction (pastebin link):
function Input(el){
var parent = el,
map = {},
intervals = {};
function ev_kdown(ev)
{
map[ev.key] = true;
ev.preventDefault();
return;
}
function ev_kup(ev)
{
map[ev.key] = false;
ev.preventDefault();
return;
}
function key_down(key)
{
return map[key];
}
function keys_down_array(array)
{
for(var i = 0; i < array.length; i++)
if(!key_down(array[i]))
return false;
return true;
}
function keys_down_arguments()
{
return keys_down_array(Array.from(arguments));
}
function clear()
{
map = {};
}
function watch_loop(keylist, callback)
{
return function(){
if(keys_down_array(keylist))
callback();
}
}
function watch(name, callback)
{
var keylist = Array.from(arguments).splice(2);
intervals[name] = setInterval(watch_loop(keylist, callback), 1000/24);
}
function unwatch(name)
{
clearInterval(intervals[name]);
delete intervals[name];
}
function detach()
{
parent.removeEventListener("keydown", ev_kdown);
parent.removeEventListener("keyup", ev_kup);
}
function attach()
{
parent.addEventListener("keydown", ev_kdown);
parent.addEventListener("keyup", ev_kup);
}
function Input()
{
attach();
return {
key_down: key_down,
keys_down: keys_down_arguments,
watch: watch,
unwatch: unwatch,
clear: clear,
detach: detach
};
}
return Input();
}
This class doesn’t do everything and it won’t handle every conceivable use case. I’m not a library guy. But for general interactive use it should be fine.
To use this class, create an instance and point it to the element you want to associate keyboard input with:
var input_txt = Input(document.getElementById("txt"));
input_txt.watch("print_5", function(){
txt.value += "FIVE ";
}, "Control", "5");
What this will do is attach a new input listener to the element with #txt
(let’s assume it’s a textarea), and set a watchpoint for the key combo Ctrl+5
. When both Ctrl
and 5
are down, the callback function you passed in (in this case, a function that adds "FIVE "
to the textarea) will be called. The callback is associated with the name print_5
, so to remove it, you simply use:
input_txt.unwatch("print_5");
To detach input_txt
from the txt
element:
input_txt.detach();
This way, garbage collection can pick up the object (input_txt
), should it be thrown away, and you won’t have an old zombie event listener left over.
For thoroughness, here is a quick reference to the class’s API, presented in C/Java style so you know what they return and what arguments they expect.
Boolean key_down (String key);
Returns
true
ifkey
is down, false otherwise.Boolean keys_down (String key1, String key2, ...);
Returns
true
if all keyskey1 .. keyN
are down, false otherwise.void watch (String name, Function callback, String key1, String key2, ...);
Creates a “watchpoint” such that pressing all of
keyN
will trigger the callbackvoid unwatch (String name);
Removes said watchpoint via its name
void clear (void);
Wipes the “keys down” cache. Equivalent to
map = {}
abovevoid detach (void);
Detaches the
ev_kdown
andev_kup
listeners from the parent element, making it possible to safely get rid of the instance
Update 2017-12-02 In response to a request to publish this to github, I have created a gist.
Update 2018-07-21 I’ve been playing with declarative style programming for a while, and this way is now my personal favorite: fiddle, pastebin
Generally, it’ll work with the cases you would realistically want (ctrl, alt, shift), but if you need to hit, say, a+w
at the same time, it wouldn’t be too difficult to “combine” the approaches into a multi-key-lookup.
I hope this thoroughly explained answer mini-blog was helpful 🙂