What's the best way to make sure an element exists?


Before adding / removing a class (just an example from the top of my head) I always do the following:

if ( typeof(myelement) != "undefined" && myelement !=  null && typeof(myelement) == 'object' ) {
//add / remove class

I feel like this is not a good approach. What are the checks you have to do to make sure an element is in the DOM and can be used for various purposes? Would ==‘object’ be enough?


Generally i just check to ensure that the value is not equal to ‘undefined’.

Commonly duck typing is used to perform checks such as this, where we just check the important properties to ensure that it’s close enough to being an element.

For example:

function isElement(elem) {
    if (typeof HTMLElement === 'object') {
        return elem instanceof HTMLElement;
    return elem &&
        typeof elem === 'object' &&
        elem.nodeType === 1 &&
        typeof elem.nodeName === 'string';

If the browser has HTMLElement object support, a simple test against that is good enough.

If not though, we check first that the object exists and that it is an object before checking for a couple of obvious properties that an HTML element should have. If they’re there, it’s close enough to being an element for us to accept that it is one.

I find this interesting because I always thought just checking the nodeType was sufficient. Are there instances where that can trip you up?

Thanks :slight_smile:

Why don’t you declare the obj variable? Should it be elem?

Yes it should, and has been corrected. Thanks for catching that.

Just checking the nodeType property can trip you up if no object actually exists, for then the script stops executing with an error.

Checking that the object exists first does tend to be mandatory.
Checking that it’s an object is not strictly necessary, but it does help others coming across the code to know that you expect it to be an object from that point onwards.
Checking the nodeType also tends to be mandatory, if you want to avoid matching other nodes such as text nodes, or comment nodes.

From what I understand though, that’s not quite enough in some browsers when working with elements that belong to other windows or frames, which is why nodeName is also used.
It would be useful though if we had some tests to confirm that this is actually the case.

And while it is possible to pass in something that is not an element and will in some older browsers be mistaken for one, such as:

    nodeType: 1,
    nodeName: 'P'

We are not attempting to remove all doubt here. Instead, we are looking to see if it’s close enough to what an element should be, If it’s close enough then we’ll accept that it as it is. That’s duck typing for you :slight_smile:

To answer the original OP, if you are looking for an element node do not just check for typeof==‘object’,
because null is an object, and null is what you get if you look for something specific in the dom that isn’t there,
like trying getElementById on an element that hasn’t been appended to the document, or previous/nextSibling when you are at the first/last child.

if(node && node.nodeType==1) is good enough for most of us.