Develop Reference

How to get the content of all spans that have the email attribute with JavaScript

I want to find all spans in a document that have the "email" attribute and then for each email address i'll check with my server if the email is approved and inject to the span
content an img with a "yes" or a "no". I don't need the implementation of the PHP side, only JavaScript.
So say "newsletter#zend.com" is approved in my db and the HTML code is:
<span dir="ltr"><span class="yP" email="newsletter#zend.com">Zend Technologies</span></span>
Then the JavaScript will change the HTML to:
<span dir="ltr"><span class="yP" email="newsletter#zend.com"><img src="yes.gif" />Zend Technologies</span></span>
I need someone to guide me to the right direction on how to approach this.
Note: i don't want to use jQuery.

If you don't want to use a library, and you don't want to limit yourself to browsers that support querySelectorAll, you're probably best off with a simple recursive-descent function or getElementsByTagName. Here's an RD example:
The function (off-the-cuff, untested):
function findEmailSpans(container, callback) {
var node, emailAttr;
for (node = container.firstChild; node; node = node.nextSibling) {
if (node.nodeType === 1 && node.tagName === "SPAN") { // 1 = Element
emailAttr = node.getAttribute("email");
if (emailAttr) {
callback(node, emailAttr);
}
}
}
switch (node.nodeType) {
case 1: // Element
case 9: // Document
case 11: // DocumentFragment
findEmailSpans(node, callback);
break;
}
}
}
Calling it:
findEmailSpans(document.documentElement, function(span, emailAttr) {
// Do something with `span` and `emailAttr`
});
Alternately, if you want to rely on getElementsByTagName (which is quite widely supported) and don't mind building such a large NodeList in memory, that would be simpler and might be faster: It would let you get one flat NodeList of all span elements, so then you'd just have a simple loop rather than a recursive-descent function (not that the RD function is either difficult or slow, but still). Something like this:
var spans = document.getElementsByTagName("span"),
index, node, emailAttr;
for (index = 0; index < spans.length; ++index) {
node = spans.item(index);
emailAttr = node.getAttribute("email");
if (emailAttr) {
// Do something with `node` and `emailAttr`
}
}
You'll want to compare and decide which method suits you best, each probably has pros and cons.
References:
DOM3 Core Spec
However, for this sort of thing I really would recommend getting and using a good JavaScript library like jQuery, Prototype, YUI, Closure, or any of several others. With any good library, it can look something like this (jQuery):
$("span[email]").each(function() {
// Here, `this` refers to the span that has an email attribute
});
...or this (Prototype):
$$("span[email]").each(function() {
// Here, `this` refers to the span that has an email attribute
});
...and the others won't be massively more complex. Using a library to factor our common ops like searching for things in the DOM lets you concentrate on the actual problem you're trying to solve. Both jQuery and (recent versions of) Prototype will defer to querySelectorAll on browsers that support it (and I imagine most others will, too), and fall back to their own search functions on browsers that don't.

You would use document.getElementsByTagName() to get a list of all spans. Then, check each span for the email attribute using Element.hasAttribute. Then you would use the Node interface to create and insert newe tags accordingly.
EDIT:
window.addEventListener('load', callback, true);
var callback = function() {
var spanTags = document.getElementsByTagName('span');
for (var i = 0; i < spanTags.length; i += 1) {
if (spanTags[i].hasAttribute('email')) {
var imgElement = document.createElement('img');
imgElement.setAttribute('src', 'yes.gif');
spanTags[i].insertBefore(imgElement, spanTags[i].firstChild);
}
}
}

Remove all child elements of a DOM node in JavaScript

How would I go about removing all of the child elements of a DOM node in JavaScript?
Say I have the following (ugly) HTML:
<p id="foo">
<span>hello</span>
<div>world</div>
</p>
And I grab the node I want like so:
var myNode = document.getElementById("foo");
How could I remove the children of foo so that just <p id="foo"></p> is left?
Could I just do:
myNode.childNodes = new Array();
or should I be using some combination of removeElement?
I'd like the answer to be straight up DOM; though extra points if you also provide an answer in jQuery along with the DOM-only answer.

Option 1 A: Clearing innerHTML.
This approach is simple, but might not be suitable for high-performance applications because it invokes the browser's HTML parser (though browsers may optimize for the case where the value is an empty string).
doFoo.onclick = () => {
const myNode = document.getElementById("foo");
myNode.innerHTML = '';
}
<div id='foo' style="height: 100px; width: 100px; border: 1px solid black;">
<span>Hello</span>
</div>
<button id='doFoo'>Remove via innerHTML</button>
Option 1 B: Clearing textContent
As above, but use .textContent. According to MDN this will be faster than innerHTML as browsers won't invoke their HTML parsers and will instead immediately replace all children of the element with a single #text node.
doFoo.onclick = () => {
const myNode = document.getElementById("foo");
myNode.textContent = '';
}
<div id='foo' style="height: 100px; width: 100px; border: 1px solid black;">
<span>Hello</span>
</div>
<button id='doFoo'>Remove via textContent</button>
Option 2 A: Looping to remove every lastChild:
An earlier edit to this answer used firstChild, but this is updated to use lastChild as in computer-science, in general, it's significantly faster to remove the last element of a collection than it is to remove the first element (depending on how the collection is implemented).
The loop continues to check for firstChild just in case it's faster to check for firstChild than lastChild (e.g. if the element list is implemented as a directed linked-list by the UA).
doFoo.onclick = () => {
const myNode = document.getElementById("foo");
while (myNode.firstChild) {
myNode.removeChild(myNode.lastChild);
}
}
<div id='foo' style="height: 100px; width: 100px; border: 1px solid black;">
<span>Hello</span>
</div>
<button id='doFoo'>Remove via lastChild-loop</button>
Option 2 B: Looping to remove every lastElementChild:
This approach preserves all non-Element (namely #text nodes and <!-- comments --> ) children of the parent (but not their descendants) - and this may be desirable in your application (e.g. some templating systems that use inline HTML comments to store template instructions).
This approach wasn't used until recent years as Internet Explorer only added support for lastElementChild in IE9.
doFoo.onclick = () => {
const myNode = document.getElementById("foo");
while (myNode.lastElementChild) {
myNode.removeChild(myNode.lastElementChild);
}
}
<div id='foo' style="height: 100px; width: 100px; border: 1px solid black;">
<!-- This comment won't be removed -->
<span>Hello <!-- This comment WILL be removed --></span>
<!-- But this one won't. -->
</div>
<button id='doFoo'>Remove via lastElementChild-loop</button>
Bonus: Element.clearChildren monkey-patch:
We can add a new method-property to the Element prototype in JavaScript to simplify invoking it to just el.clearChildren() (where el is any HTML element object).
(Strictly speaking this is a monkey-patch, not a polyfill, as this is not a standard DOM feature or missing feature. Note that monkey-patching is rightfully discouraged in many situations.)
if( typeof Element.prototype.clearChildren === 'undefined' ) {
Object.defineProperty(Element.prototype, 'clearChildren', {
configurable: true,
enumerable: false,
value: function() {
while(this.firstChild) this.removeChild(this.lastChild);
}
});
}
<div id='foo' style="height: 100px; width: 100px; border: 1px solid black;">
<span>Hello <!-- This comment WILL be removed --></span>
</div>
<button onclick="this.previousElementSibling.clearChildren()">Remove via monkey-patch</button>

In 2022+, use the replaceChildren() API!
Replacing all children can now be done with the (cross-browser supported) replaceChildren API:
container.replaceChildren(...arrayOfNewChildren);
This will do both:
remove all existing children, and
append all of the given new children, in one operation.
You can also use this same API to just remove existing children, without replacing them:
container.replaceChildren();
This is fully supported in Chrome/Edge 86+, Firefox 78+, and Safari 14+. It is fully specified behavior. This is likely to be faster than any other proposed method here, since the removal of old children and addition of new children is done without requiring innerHTML, and in one step instead of multiple.

Use modern Javascript, with remove!
const parent = document.getElementById("foo")
while (parent.firstChild) {
parent.firstChild.remove()
}
This is a newer way to write node removal in ES5. It is vanilla JS and reads much nicer than relying on parent.
All modern browsers are supported.
Browser Support - 97% Jun '21

The currently accepted answer is wrong about innerHTML being slower (at least in IE and Chrome), as m93a correctly mentioned.
Chrome and FF are dramatically faster using this method (which will destroy attached jquery data):
var cNode = node.cloneNode(false);
node.parentNode.replaceChild(cNode, node);
in a distant second for FF and Chrome, and fastest in IE:
node.innerHTML = '';
InnerHTML won't destroy your event handlers or break jquery references, it's also recommended as a solution here:
https://developer.mozilla.org/en-US/docs/Web/API/Element.innerHTML.
The fastest DOM manipulation method (still slower than the previous two) is the Range removal, but ranges aren't supported until IE9.
var range = document.createRange();
range.selectNodeContents(node);
range.deleteContents();
The other methods mentioned seem to be comparable, but a lot slower than innerHTML, except for the outlier, jquery (1.1.1 and 3.1.1), which is considerably slower than anything else:
$(node).empty();
Evidence here:
http://jsperf.com/innerhtml-vs-removechild/167 http://jsperf.com/innerhtml-vs-removechild/300
https://jsperf.com/remove-all-child-elements-of-a-dom-node-in-javascript
(New url for jsperf reboot because editing the old url isn't working)
Jsperf's "per-test-loop" often gets understood as "per-iteration", and only the first iteration has nodes to remove so the results are meaningless, at time of posting there were tests in this thread set up incorrectly.

If you use jQuery:
$('#foo').empty();
If you don't:
var foo = document.getElementById('foo');
while (foo.firstChild) foo.removeChild(foo.firstChild);

var myNode = document.getElementById("foo");
var fc = myNode.firstChild;
while( fc ) {
myNode.removeChild( fc );
fc = myNode.firstChild;
}
If there's any chance that you have jQuery affected descendants, then you must use some method that will clean up jQuery data.
$('#foo').empty();
The jQuery .empty() method will ensure that any data that jQuery associated with elements being removed will be cleaned up.
If you simply use DOM methods of removing the children, that data will remain.

The fastest...
var removeChilds = function (node) {
var last;
while (last = node.lastChild) node.removeChild(last);
};
Thanks to Andrey Lushnikov for his link to jsperf.com (cool site!).
EDIT: to be clear, there is no performance difference in Chrome between firstChild and lastChild. The top answer shows a good solution for performance.

Use elm.replaceChildren().
It’s experimental without wide support, but when executed with no params will do what you’re asking for, and it’s more efficient than looping through each child and removing it. As mentioned already, replacing innerHTML with an empty string will require HTML parsing on the browser’s part.
MDN Documentation
Update It's widely supported now

If you only want to have the node without its children you could also make a copy of it like this:
var dupNode = document.getElementById("foo").cloneNode(false);
Depends on what you're trying to achieve.

Ecma6 makes it easy and compact
myNode.querySelectorAll('*').forEach( n => n.remove() );
This answers the question, and removes “all child nodes”.
If there are text nodes belonging to myNode they can’t be selected with CSS selectors, in this case we’ve to apply (also):
myNode.textContent = '';
Actually the last one could be the fastest and more effective/efficient solution.
.textContent is more efficient than .innerText and .innerHTML, see: MDN

Here's another approach:
function removeAllChildren(theParent){
// Create the Range object
var rangeObj = new Range();
// Select all of theParent's children
rangeObj.selectNodeContents(theParent);
// Delete everything that is selected
rangeObj.deleteContents();
}

element.textContent = '';
It's like innerText, except standard. It's a bit slower than removeChild(), but it's easier to use and won't make much of a performance difference if you don't have too much stuff to delete.

Here is what I usually do :
HTMLElement.prototype.empty = function() {
while (this.firstChild) {
this.removeChild(this.firstChild);
}
}
And voila, later on you can empty any dom element with :
anyDom.empty()

In response to DanMan, Maarten and Matt. Cloning a node, to set the text is indeed a viable way in my results.
// #param {node} node
// #return {node} empty node
function removeAllChildrenFromNode (node) {
var shell;
// do not copy the contents
shell = node.cloneNode(false);
if (node.parentNode) {
node.parentNode.replaceChild(shell, node);
}
return shell;
}
// use as such
var myNode = document.getElementById('foo');
myNode = removeAllChildrenFromNode( myNode );
Also this works for nodes not in the dom which return null when trying to access the parentNode. In addition, if you need to be safe a node is empty before adding content this is really helpful. Consider the use case underneath.
// #param {node} node
// #param {string|html} content
// #return {node} node with content only
function refreshContent (node, content) {
var shell;
// do not copy the contents
shell = node.cloneNode(false);
// use innerHTML or you preffered method
// depending on what you need
shell.innerHTML( content );
if (node.parentNode) {
node.parentNode.replaceChild(shell, node);
}
return shell;
}
// use as such
var myNode = document.getElementById('foo');
myNode = refreshContent( myNode );
I find this method very useful when replacing a string inside an element, if you are not sure what the node will contain, instead of worrying how to clean up the mess, start out fresh.

Using a range loop feels the most natural to me:
for (var child of node.childNodes) {
child.remove();
}
According to my measurements in Chrome and Firefox, it is about 1.3x slower. In normal circumstances, this will perhaps not matter.

There are couple of options to achieve that:
The fastest ():
while (node.lastChild) {
node.removeChild(node.lastChild);
}
Alternatives (slower):
while (node.firstChild) {
node.removeChild(node.firstChild);
}
while (node.hasChildNodes()) {
node.removeChild(node.lastChild);
}
Benchmark with the suggested options

var empty_element = function (element) {
var node = element;
while (element.hasChildNodes()) { // selected elem has children
if (node.hasChildNodes()) { // current node has children
node = node.lastChild; // set current node to child
}
else { // last child found
console.log(node.nodeName);
node = node.parentNode; // set node to parent
node.removeChild(node.lastChild); // remove last node
}
}
}
This will remove all nodes within the element.

A one-liner to iteratively remove all the children of a node from the DOM
Array.from(node.children).forEach(c => c.remove())
Or
[...node.children].forEach(c => c.remove())

innerText is the winner! http://jsperf.com/innerhtml-vs-removechild/133. At all previous tests inner dom of parent node were deleted at first iteration and then innerHTML or removeChild where applied to empty div.

Simplest way of removing the child nodes of a node via Javascript
var myNode = document.getElementById("foo");
while(myNode.hasChildNodes())
{
myNode.removeChild(myNode.lastChild);
}

let el = document.querySelector('#el');
if (el.hasChildNodes()) {
el.childNodes.forEach(child => el.removeChild(child));
}

i saw people doing:
while (el.firstNode) {
el.removeChild(el.firstNode);
}
then someone said using el.lastNode is faster
however I would think that this is the fastest:
var children = el.childNodes;
for (var i=children.length - 1; i>-1; i--) {
el.removeNode(children[i]);
}
what do you think?
ps:
this topic was a life saver for me. my firefox addon got rejected cuz i used innerHTML. Its been a habit for a long time. then i foudn this. and i actually noticed a speed difference. on load the innerhtml took awhile to update, however going by addElement its instant!

Why aren't we following the simplest method here "remove" looped inside while.
const foo = document.querySelector(".foo");
while (foo.firstChild) {
foo.firstChild.remove();
}
Selecting the parent div
Using "remove" Method inside a While loop for eliminating First child element , until there is none left.

Generally, JavaScript uses arrays to reference lists of DOM nodes. So, this will work nicely if you have an interest in doing it through the HTMLElements array. Also, worth noting, because I am using an array reference instead of JavaScript proto's this should work in any browser, including IE.
while(nodeArray.length !== 0) {
nodeArray[0].parentNode.removeChild(nodeArray[0]);
}

Just saw someone mention this question in another and thought I would add a method I didn't see yet:
function clear(el) {
el.parentNode.replaceChild(el.cloneNode(false), el);
}
var myNode = document.getElementById("foo");
clear(myNode);
The clear function is taking the element and using the parent node to replace itself with a copy without it's children. Not much performance gain if the element is sparse but when the element has a bunch of nodes the performance gains are realized.

Functional only approach:
const domChildren = (el) => Array.from(el.childNodes)
const domRemove = (el) => el.parentNode.removeChild(el)
const domEmpty = (el) => domChildren(el).map(domRemove)
"childNodes" in domChildren will give a nodeList of the immediate children elements, which is empty when none are found. In order to map over the nodeList, domChildren converts it to array. domEmpty maps a function domRemove over all elements which removes it.
Example usage:
domEmpty(document.body)
removes all children from the body element.

You can remove all child elements from a container like below:
function emptyDom(selector){
const elem = document.querySelector(selector);
if(elem) elem.innerHTML = "";
}
Now you can call the function and pass the selector like below:
If element has id = foo
emptyDom('#foo');
If element has class = foo
emptyDom('.foo');
if element has tag = <div>
emptyDom('div')

element.innerHTML = "" (or .textContent) is by far the fastest solution
Most of the answers here are based on flawed tests
For example:
https://jsperf.com/innerhtml-vs-removechild/15
This test does not add new children to the element between each iteration. The first iteration will remove the element's contents, and every other iteration will then do nothing.
In this case, while (box.lastChild) box.removeChild(box.lastChild) was faster because box.lastChild was null 99% of the time
Here is a proper test: https://jsperf.com/innerhtml-conspiracy
Finally, do not use node.parentNode.replaceChild(node.cloneNode(false), node). This will replace the node with a copy of itself without its children. However, this does not preserve event listeners and breaks any other references to the node.

Best Removal Method for ES6+ Browser (major browsers released after year 2016):
Perhaps there are lots of way to do it, such as Element.replaceChildren().
I would like to show you an effective solution with only one redraw & reflow supporting all ES6+ browsers.
function removeChildren(cssSelector, parentNode){
var elements = parentNode.querySelectorAll(cssSelector);
let fragment = document.createDocumentFragment();
fragment.textContent=' ';
fragment.firstChild.replaceWith(...elements);
}
Usage: removeChildren('.foo',document.body);: remove all elements with className foo in <body>

If you want to empty entire parent DOM then it's very simple...
Just use .empty()
function removeAll() {
$('#parent').empty();
}
<script src="https://cdnjs.cloudflare.com/ajax/libs/jquery/3.3.1/jquery.min.js"></script>
<button onclick="removeAll()">Remove all the element of parent</button>
<div id="parent">
<h3>Title</h3>
<p>Child 1</p>
<p>Child 2</p>
<p>Child 3</p>
</div>

What kind of JavaScript is this?

I have an application that has this format scattered around but I dont know what kind it is. It's not jQuery, so what is it?
$('some_edit').style.display = "block";
$('some_views').style.display = "none";
I get this in firebug and I know the element is present:
$("some_edit").style is undefined

It could be many things - examine the source code (or use Firebug) and see what JS libraries are being loaded.

A lot of people have defined the '$' symbol as a substitute for document.getElementById().
Basically:
function $(id) { return document.getElementById(id); }
$("ElementID").innerHTML = "Text"; //Usage
A more proper, "namespace" example:
var DOM = { // creating the namespace "DOM"
$: (function() {
if(document.getElementById)
return function(id){ return document.getElementById(id); }
else if(document.all)
return function(id) { return document.all[id]; }
else
return function(id) { /* I don't even want to get into document.layers */ }
})()
};
// Later in the code:
{
function ExampleFunction() {
// ...
DOM.$("ElementID").style.backgroundColor = "#96d0a0"; // a nice minty green color
// ...
}
}
I have used a self-invocation pattern (function(){ ... }()) in this example.

at first i thought the jquery selector would likely have been $("#some_edit") and then .css(). so I would have said, prototype or mootools or a home brew $.
you can certainly discount both mootools and prototype, because if the selector returns an object, then the style property will be available (ignoring best practices in both frameworks on setting styles).
this leaves, the site uses homebrew $ assignment or jquery, which is not being used correctly.
actually, $("foo").style.blah in jquery will produce this very exception (even if the selector was good) - here is jsfiddle to the rescue
case point jquery (triggers):
http://www.jsfiddle.net/dimitar/vmsZn/
case point prototype (works):
http://www.jsfiddle.net/dimitar/vmsZn/1/
case point mootools (works):
http://www.jsfiddle.net/dimitar/vmsZn/2/

It is setting the display style for the two page elements - the display property specifies the type of box an element should generate.
block = The element will generate a block box (a line break before and after the element)
none = The element will generate no box at all

Put a [0] in front of $('some_views') to return the Native DOM Element.
$('some_views')[0].style.display = "none";
or $('some_views').get(0).style.display = "none";
or $('some_views').css('display', 'none') to iterate through the collection of DOM elements.

It's JQuery -- uses $ as its key variable.
Added:
Could also be mootools. Also uses $
Added:
'some_edit' would be the id of an element.
ps. I agree $ could be anything. Odds are though that it is JQuery or Mootools. "When you hear hoof beats, think horses, not zebras."

Remove element created by JavaScript on load

How can I remove elements which are created by JavaScript, I tried using CSS by setting display:none; but that doesn't seem to solve my problem, also I can't remove them since I don't have them in HTML, any other ways? Thank you
UPDATE:
Can't use any JavaScript such as jQuery, MooTools, ExtJS etc, and actual elements I want to remove are divs, with a specified class so I can't use getElementById.
I found this script on Google, but it doesn't seem to work but this is what I need:
HERE

This is fairly simple to do this using jQuery.
$("#myId").remove();
will remove
<div id="myId"></div>
Edit: You can also do it with "old school" javascript.
The function you're looking for is removeChild()
Example:
function removeElement(divNum) {
var d = document.getElementById('myDiv');
var olddiv = document.getElementById(divNum);
d.removeChild(olddiv);
}

You will want something like this to take advantage of browser support where you can:
if(document.getElementsByClassName){
// Safari 3+ and Firefox 3+
var itms = document.getElementsByClassName('your_class');
for(var i = 0; i<itms.length; i++){
var it = itms[i];
if(it.tagName == "DIV"){
it.parentNode.removeChild(it);
i = i - 1; //It actually gets removed from the array, so we need to drop our count
}
}
} else {
// All other browsers
var itms = document.getElementsByTagName('div');
for(var i = 0; i<itms.length; i++){
var it = itms[i];
// Test that className contains your class
if(/your_class/.test(it.className)) it.parentNode.removeChild(it);
}
}

JavaScript handles all memory mangement for you using garbage collection so once all references to an object cease to exist that object will be handled by the browsers specific implementation.
If you have the dom element itself:
if(node && node.parentNode){
// return a ref to the removed child
node.parentNode.removeChild(node);
}

Since you say you can't use Javascript, you're pretty stuck. Have you tried this CSS:
.classname {
display: none !important;
}
It's possible that the created elements have inline styles set on them, in which case normal CSS is overridden. I believe the !important will override inline styles.
Of course, the best solution is not to add the element in the first place... but I'm guessing you're in one of those (unfathomably common) scenarios where you can't change or get rid of the JS?

Not all browsers have a
document.getElementsByClassName
method, but for your purpose you can
fake it well enough- This method does
not work like the native
HTMLElement.getElementsByClassName- it
returns an array, not a live nodelist.
You can specify a parent element and a
tag name to speed it up.
function getElementsByClass(css, pa, tag){
pa= pa || document.body;
tag= tag || '*';
css= RegExp('\\b'+css+'\\b');
var A= [], elements, L, i= 0, tem;
elements= pa.getElementsByTagName(tag);
L= elements.length;
while(i<L){
tem= elements[i++];
if(css.test(tem.className)) A[A.length]= tem;
}
return A;
}
// test
var A= getElementsByClass('removeClass'), who;
while(A.length){
who= A.pop(); // remove from the end first, in case nested items also have the class
if(who.parentNode) who.parentNode.removeChild(who);
who= null;
}
If you have assigned event handlers to
the elements being removed, you should
remove the handlers before the
elements.

You will probably have to be more specific.
The general answer is 'with JavaScript'. As long as you have a way of navigating to the element through the DOM, you can then remove it from the DOM.
It's much easier if you can use a library like jQuery or prototype, but anything you can do with these you can do with JavaScript.
marcgg has assumed that you know the ID of the element: if you don't but can trace it in the DOM structure, you can do something like this (in prototype - don't know jQuery)
var css_selector = 'table#fred tr td span.mydata';
$$(css).invoke('remove');
If you can't use a JS library, you'll have to do the navigation through the DOM yourself, using Element.getElementsByTagName() a lot.
Now you've specified your question a bit: use Element.getElementsByTagName, and loop through them looking at their className property.

Use:
document.removeChild('id_of_element');

Unique element ID, even if element doesn't have one

I'm writing a GreaseMonkey script where I'm iterating through a bunch of elements. For each element, I need a string ID that I can use to reference that element later. The element itself doesn't have an id attribute, and I can't modify the original document to give it one (although I can make DOM changes in my script). I can't store the references in my script because when I need them, the GreaseMonkey script itself will have gone out of scope. Is there some way to get at an "internal" ID that the browser uses, for example? A Firefox-only solution is fine; a cross-browser solution that could be applied in other scenarios would be awesome.
Edit:
If the GreaseMonkey script is out of scope, how are you referencing the elements later? They GreaseMonkey script is adding events to DOM objects. I can't store the references in an array or some other similar mechanism because when the event fires, the array will be gone because the GreaseMonkey script will have gone out of scope. So the event needs some way to know about the element reference that the script had when the event was attached. And the element in question is not the one to which it is attached.
Can't you just use a custom property on the element? Yes, but the problem is on the lookup. I'd have to resort to iterating through all the elements looking for the one that has that custom property set to the desired id. That would work, sure, but in large documents it could be very time consuming. I'm looking for something where the browser can do the lookup grunt work.
Wait, can you or can you not modify the document? I can't modify the source document, but I can make DOM changes in the script. I'll clarify in the question.
Can you not use closures? Closuses did turn out to work, although I initially thought they wouldn't. See my later post.
It sounds like the answer to the question: "Is there some internal browser ID I could use?" is "No."

The answer is no, there isn't an internal id you can access. Opera and IE (maybe Safari?) support .sourceIndex (which changes if DOM does) but Firefox has nothing of this sort.
You can simulate source-index by generating Xpath to a given node or finding the index of the node from document.getElementsByTagName('*') which will always return elements in source order.
All of this requires a completely static file of course. Changes to DOM will break the lookup.
What I don't understand is how you can loose references to nodes but not to (theoretical) internal id's? Either closures and assignments work or they don't. Or am I missing something?

Closure is the way to go. This way you'll have exact reference to the element that even will survive some shuffling of DOM.
Example for those who don't know closures:
var saved_element = findThatDOMNode();
document.body.onclick = function()
{
alert(saved_element); // it's still there!
}
If you had to store it in a cookie, then I recommend computing XPath for it (e.g. walk up the DOM counting previous siblings until you find element with an ID and you'll end up with something like [#id=foo]/div[4]/p[2]/a).
XPointer is W3C's solution to that problem.

A bit confused by the wording of your question - you say that you "need a string ID that [you] can use to reference that element later, " but that you "can't store the references in [your] script because when [you] need them, the GreaseMonkey script itself will have gone out of scope."
If the script will have gone out of scope, then how are you referencing them later?!
I am going to ignore the fact that I am confused by what you are getting at and tell you that I write Greasemonkey scripts quite often and can modify the DOM elements I access to give them an ID property. This is code you can use to get a pseudo-unique value for temporary use:
var PseudoGuid = new (function() {
this.empty = "00000000-0000-0000-0000-000000000000";
this.GetNew = function() {
var fourChars = function() {
return (((1 + Math.random()) * 0x10000)|0).toString(16).substring(1).toUpperCase();
}
return (fourChars() + fourChars() + "-" + fourChars() + "-" + fourChars() + "-" + fourChars() + "-" + fourChars() + fourChars() + fourChars());
};
})();
// usage example:
var tempId = PseudoGuid.GetNew();
someDomElement.id = tempId;
That works for me, I just tested it in a Greasemonkey script myself.
UPDATE: Closures are the way to go - personally, as a hard-core JavaScript developer, I don't know how you didn't think of those immediately. :)
myDomElement; // some DOM element we want later reference to
someOtherDomElement.addEventListener("click", function(e) {
// because of the closure, here we have a reference to myDomElement
doSomething(myDomElement);
}, false);
Now, myDomElement is one of the elements you apparently, from your description, already have around (since you were thinking of adding an ID to it, or whatever).
Maybe if you post an example of what you are trying to do, it would be easier to help you, assuming this doesn't.

UPDATE: Closures are indeed the answer. So after fiddling with it some more, I figured out why closures were initially problematic and how to fix it. The tricky thing with a closure is you have to be careful when iterating through the elements not to end up with all of your closures referencing the same element. For example, this doesn't work:
for (var i = 0; i < elements.length; i++) {
var element = elements[i];
var button = document.createElement("button");
button.addEventListener("click", function(ev) {
// do something with element here
}, false)
}
But this does:
var buildListener = function(element) {
return function(ev) {
// do something with event here
};
};
for (var i = 0; i < elements.length; i++) {
var element = elements[i];
var button = document.createElement("button");
button.addEventListener("click", buildListener(element), false)
}
Anyway, I decided not to select one answer because the question had two answers: 1) No, there are no internal IDs you can use; 2) you should use closures for this. So I simply upvoted the first people to say whether there were internal IDs or who recommended generating IDs, plus anyone who mentioned closures. Thanks for the help!

If you can write to the DOM (I'm sure you can). I would solve this like this:
Have a function return or generate an ID:
//(function () {
var idCounter = new Date().getTime();
function getId( node ) {
return (node.id) ? node.id : (node.id = 'tempIdPrefix_' + idCounter++ );
}
//})();
Use this to get ID's as needed:
var n = document.getElementById('someid');
getId(n); // returns "someid"
var n = document.getElementsByTagName('div')[1];
getId(n); // returns "tempIdPrefix_1224697942198"
This way you don't need to worry about what the HTML looks like when the server hands it to you.

If you're not modifying the DOM you can get them all by indexed order:
(Prototype example)
myNodes = document.body.descendants()
alert(document.body.descendants()[1].innerHTML)
You could loop through all of the nodes and give them a unique className that you could later select easily.

You can set the id attribute to a computed value. There is a function in the prototype library that can do this for you.
http://www.prototypejs.org/api/element/identify
My favorite javascript library is jQuery. Unfortunately jQuery does not have a function like identify. However, you can still set the id attribute to a value that you generate on your own.
http://docs.jquery.com/Attributes/attr#keyfn
Here is a partial snippet from jQuery docs that sets id for divs based on the position in the page:
$(document).ready(function(){
$("div").attr("id", function (arr) {
return "div-id" + arr;
});
});

You can generate a stable, unique identifier for any given node in a DOM with the following function:
function getUniqueKeyForNode (targetNode) {
const pieces = ['doc'];
let node = targetNode;
while (node && node.parentNode) {
pieces.push(Array.prototype.indexOf.call(node.parentNode.childNodes, node));
node = node.parentNode
}
return pieces.reverse().join('/');
}
This will create identifiers such as doc/0, doc/0/0, doc/0/1, doc/0/1/0, doc/0/1/1 for a structure like this one:
<div>
<div />
<div>
<div />
<div />
</div>
</div>
There are also a few optimisations and changes you can make, for example:
In the while loop, break when that node has an attribute you know to be unique, for example #id
Not reverse() the pieces, currently it is just there to look more like the DOM structure the ID's are generated from
Not include the first piece doc if you don't need an identifier for the document node
Save the identifier on the node in some way, and reuse that value for child nodes to avoid having to traverse all the way up the tree again.
If you're writing these identifiers back to XML, use another concatenation character if the attribute you're writing is restricted.

Use mouse and/or positional properties of the element to generate a unique ID.

In javascript, you could attach a custom ID field to the node
if(node.id) {
node.myId = node.id;
} else {
node.myId = createId();
}
// store myId
It's a bit of hack, but it'll give each and every node an id you can use. Of course, document.getElementById() won't pay attention to it.

You can also use pguid (page-unique identifier) for unique identifier generation:
pguid = b9j.pguid.next() // A unique id (suitable for a DOM element)
// is generated
// Something like "b9j-pguid-20a9ff-0"
...
pguid = b9j.pguid.next() // Another unique one... "b9j-pguid-20a9ff-1"
// Build a custom generator
var sequence = new b9j.pguid.Sequence({ namespace: "frobozz" })
pguid = sequence.next() "frobozz-c861e1-0"
http://appengine.bravo9.com/b9j/documentation/pguid.html

I 'think' I've just solved a problem similar to this. However, I'm using jQuery in a browser DOM environment.
var objA = $("selector to some dom element");
var objB = $("selector to some other dom element");
if( objA[0] === objB[0]) {
//GREAT! the two objects point to exactly the same dom node
}

OK, there is no ID associated to DOM element automatically.
DOM has a hierarchycal structure of elements which is the main information.
From this perspective, you can associate data to DOM elements with jQuery or jQLite. It can solve some issues when you have to bind custom data to elements.