You need to use window.getComputedStyle
for those browsers that support it, and element.currentStyle
for those that support that. Or you could use jQuery $(element).css('width')
which should abstract the difference (although I haven't tested the latter).
It seems the following does not do what I had thought it would, at least not for width and height. After searching around I found this other SO question where it is stated to be impossible (at least not without parsing the Stylesheet?!). Seems mad to me, I shall keep looking just in case.
get CSS rule's percentage value in jQuery
if( window.getComputedStyle ) {
value = window.getComputedStyle(element,null).width;
} else if( element.currentStyle ) {
value = element.currentStyle.width;
}
update
I've found that this works...! but only for firefox :( To me it would make sense that if the element has nothing to compute it's width against (i.e. it's not part of the document flow) it should return it's original value:
function isElementFluid(elm){
var clone = elm.cloneNode(false);
if( window.getComputedStyle ) {
value = window.getComputedStyle(clone,null).width;
} else if( clone.currentStyle ) {
value = clone.currentStyle.width;
}
return (value && String(value).indexOf('%') != -1 );
}
(have not tested for IE)
Yet again another instance of where I agree with FireFox's implementation and frown at Chrome or Safari.
update 2
Ok, not a fan of being defeated by computers ;) so have come up with this function -- totally over the top, but it does seem to work. Again I have yet to test this on IE as I don't have a Windows machine to hand at the moment. It's annoying when the original FF only version is quite succinct, but the logic here is sound - it falls back to what a normal human would do in testing if something is stretchy.
function isElementFluid(elm){
var wrapper, clone = elm.cloneNode(false), ow, p1, p2;
if( window.getComputedStyle ) {
value = window.getComputedStyle(clone,null).width;
} else if( clone.currentStyle ) {
value = clone.currentStyle.width;
}
/// the browsers that fail to work as Firefox does
/// return an empty width value, so here we fall back.
if ( !value ) {
/// remove styles that can get in the way
clone.style.margin = '0';
clone.style.padding = '0';
clone.style.maxWidth = 'none';
clone.style.minWidth = 'none';
/// create a wrapper that we can control, my reason for
/// using an unknown element is that it stands less chance
/// of being affected by stylesheets - this could be improved
/// to avoid possible erroneous results by overriding more css
/// attributes with inline styles.
wrapper = document.createElement('wrapper');
wrapper.style.display = 'block';
wrapper.style.width = '500px';
wrapper.style.padding = '0';
wrapper.style.margin = '0';
wrapper.appendChild(clone);
/// insert the element in the same location as our target
elm.parentNode.insertBefore(wrapper,elm);
/// store the clone's calculated width
ow = clone.offsetWidth;
/// change the wrapper size once more
wrapper.style.width = '600px';
/// if the new width is the same as before, most likely a fixed width
if( clone.offsetWidth == ow ){
/// tidy up
elm.parentNode.removeChild(wrapper);
return false;
}
/// otherwise, calculate the percentages each time - if they
/// match then it's likely this is a fluid element
else {
p1 = Math.floor(100/500*ow);
p2 = Math.floor(100/600*clone.offsetWidth);
/// tidy up
elm.parentNode.removeChild(wrapper);
return (p1 == p2) ? Math.round(p1)+'%' : false;
}
}
else {
p1 = (value && String(value).indexOf('%') != -1);
return p1 ? value : false;
}
}