I think this is the most interesting part of the grid. This feature required two different things: the ability to focus on a grid cell and the ability to capture navigation keys to move the selected cell.

Focusing on a grid cell

It turns out that all you need to do in HTML to get focus on a non-form element means that you apply a tabindex to it. Let’s take a look at this in practice. Look at this HTML.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
<style type="text/css">
.tabindex-example span {
height: 100px;
display: inline-block;
}
.tabindex-example .orange {
background-color: orange;
width: 100px;
}
.tabindex-example .purple {
background-color: purple;
width: 150px;
}
.tabindex-example .yellow {
background-color: yellow;
width: 200px;
}
</style>
<div class="tabindex-example">
<span tabindex="0" class="orange">&nbsp;</span>
<span tabindex="0" class="purple">&nbsp;</span>
<span tabindex="0" class="yellow">&nbsp;</span>
</div>

And, here it is in the browser. Click on the orange rectange. Once you see the highlight around the element, press the TAB key and watch that focus move to the purple. Then, after another TAB key press, the highlight will move to the yellow element.

     

That addressses the problem with how to get the TAB key to stop on one of the grid cells. The real problem comes with how to apply tabindexes on the cells of the grid. On big tables, applying the tabindex property on each of the cells or prerendering them leads to really slow grid render and response times. So, grijq takes the “apply a tabindex on demand” approach. For example, when a cell receives focus, then it applies a tabindex="0" to each of the cells in that row.

1
cell.parent().children().prop('tabindex', '0');

That happens throughout the code in different ways, when the user interacts with the different portions of the grid. This is the least amount of interaction that guarantees smooth movement horizontally with the TAB key presses. When the focus gets to the end or beginning of the row, then it also applies it to the next or previous row, respectively.

1
2
3
4
5
6
7
if(cell.next().length === 0) {
var nextRow = row.next();
nextRow.children().prop('tabindex', '0');
} else if(cell.prev().length === 0) {
var previousRow = row.prev();
previousRow.children().prop('tabindex', '0');
}

Key navigation

Key navigation is mundane, lower level programming that, while makes the user very happy, kind of sucks to implement. The jQuery UI library even defines constants that we can use in capturing the key presses.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
grijq.element.keydown(function(e) {
switch(e.keyCode) {
case $.ui.keyCode.LEFT:
target.prev().focus();
e.preventDefault();
break;
case $.ui.keyCode.RIGHT:
target.next().focus();
e.preventDefault();
break;
case $.ui.keyCode.UP:
var index = target.prevAll().length + 1;
var tr = target.closest('tr').prev();
tr.children().prop('tabindex', '0');
$(':nth-child(' + index + ')', tr).focus();
e.preventDefault();
break;
case $.ui.keyCode.DOWN:
var index = target.prevAll().length + 1;
var tr = target.closest('tr').next();
tr.children().prop('tabindex', '0');
$(':nth-child(' + index + ')', tr).focus();
e.preventDefault();
break;
}
});

Very simple, really, once grijq gets the tabindexes into place. If we stop here, we have a simple and relatively elegant solution. Unfortunately, as we find out tomorrow, making cells editable completely screws with this simple solution. When we allow users to type into an input element, we introduce another tabindex-ed element into the mix which messes up this code above. But, I’ll leave that until tomorrow.

Head on over to the project page and check out the examples.