We Keep Coding

custom querySelectorAll implemention

This was given to me as an interview question -- didn't get the job, but I still want to figure it out.
The objective is to write two querySelectorAll functions: one called qsa1 which works for selectors consisting of a single tag name (e.g. div or span) and another called qsa2 which accepts arbitrarily nested tag selectors (such as p span or ol li code).
I got the first one easily enough, but the second one is a bit trickier.
I suspect that, in order to handle a variable number of selectors, the proper solution might be recursive, but I figured I'd try to get something working that is iterative first. Here's what I've got so far:
qsa2 = function(node, selector) {
var selectors = selector.split(" ");
var matches;
var children;
var child;
var parents = node.getElementsByTagName(selectors[0]);
if (parents.length > 0) {
for (var i = 0; i < parents.length; i++) {
children = parents[i].getElementsByTagName(selectors[1]);
if (children.length > 0) {
for (var i = 0; i < parents.length; i++) {
child = children[i];
matches.push(child); // somehow store our result here
}
}
}
}
return matches;
}
The first problem with my code, aside from the fact that it doesn't work, is that it only handles two selectors (but it should be able to clear the first, second, and fourth cases).
The second problem is that I'm having trouble returning the correct result. I know that, just as in qsa1, I should be returning the same result as I'd get by calling the getElementsByTagName() function which "returns a live NodeList of elements with the given tag name". Creating an array and pushing or appending the Nodes to it isn't cutting it.
How do I compose the proper return result?
(For context, the full body of code can be found here)

Here's how I'd do it
function qsa2(selector) {
var next = document;
selector.split(/\s+/g).forEach(function(sel) {
var arr = [];
(Array.isArray(next) ? next : [next]).forEach(function(el) {
arr = arr.concat( [].slice.call(el.getElementsByTagName(sel) ));
});
next = arr;
});
return next;
}
Assume we always start with the document as context, then split the selector on spaces, like you're already doing, and iterate over the tagnames.
On each iteration, just overwrite the outer next variable, and run the loop again.
I've used an array and concat to store the results in the loop.
This is somewhat similar to the code in the question, but it should be noted that you never create an array, in fact the matches variable is undefined, and can't be pushed to.

You have syntax errors here:
if (parents.length > 0) {
for (var i = 0; i < parents.length; i++) {
children = parents[i].getElementsByTagName(selectors[1]);
if (children.length > 0) {
for (var i = 0; i < parents.length; i++) { // <-----------------------
Instead of going over the length of the children, you go over the length of the parent.
As well as the fact that you are reusing iteration variable names! This means the i that's mapped to the length of the parent is overwritten in the child loop!
On a side note, a for loop won't iterate over the elements if it's empty anyway, so your checks are redundant.
It should be the following:
for (var i = 0; i < parents.length; i++) {
children = parents[i].getElementsByTagName(selectors[1]);
for (var k = 0; k < children.length; i++) {
Instead of using an iterative solution, I would suggest using a recursive solution like the following:
var matches = [];
function recursivelySelectChildren(selectors, nodes){
if (selectors.length != 0){
for (var i = 0; i < nodes.length; i++){
recursivelySelectChildren(nodes[i].getElementsByTagName(selectors[0]), selectors.slice(1))
}
} else {
matches.push(nodes);
}
}
function qsa(selector, node){
node = node || document;
recursivelySelectChildren(selector.split(" "), [node]);
return matches;
}

Using querySelectorAll to get ALL elements with that class name, not only the first

I've ditched jquery about 9(ish) months ago and needed a selector engine (without all the hassle and don't mind ie<7 support) so i made a simplified version of document.querySelectorAll by creating this function:
// "qsa" stands for: "querySelectorAll"
window.qsa = function (el) {
var result = document.querySelectorAll(el)[0];
return result;
};
This works perfectly fine for 95% of the time but I've had this problem for a while now and i have researched mdn, w3c, SO and not to forget Google :) but have not yet found the answer as to why I only get the first element with the requested class.
And I know that only the first element being returned is caused by the "[0]" at the end, but the function won't work if I remove it so I've tried to make a for loop with an index variable that increases in value depending on the length of elements with that class like this:
window.qsa = function (el) {
var result, el = document.querySelectorAll(el);
for(var i = 0; i < el.length; ++i) {
result = el[i];
}
return result;
};
Again that did not work so I tried a while loop like this:
window.qsa = function (el) {
var result, i = 0, el = document.querySelectorAll(el);
while(i < el.length) {
i++;
}
result = el[i];
return result;
};
By now I'm starting to wonder if anything works? and I'm getting very frustrated with document.querySelectorAll...
But my stubborn inner-self keeps going and I keep on failing (tiering cycle) so I know that now is REALLY the time to ask these questions :
Why is it only returning the first element with that class and not all of them?
Why does my for loop fail?
Why does my while loop fail?
And thank you because any / all help is much appreciated.

Why is it only returning the first element with that class and not all of them?
Because you explicitly get the first element off the results and return that.
Why does my for loop fail?
Because you overwrite result with a new value each time you go around the end of loop. Then you return the last thing you get.
Why does my while loop fail?
The same reason.
If you want all the elements, then you just get the result of running the function:
return document.querySelectorAll(el)
That will give you a NodeList object containing all the elements.
Now that does what you say you want, I'm going to speculate about what your real problem is (i.e. why you think it doesn't work).
You haven't shown us what you do with the result of running that function, but my guess is that you are trying to treat it like an element.
It isn't an element. It is a NodeList, which is like an array.
If you wanted to, for instance, change the background colour of an element you could do this:
element.style.backgroundColor = "red";
If you want to change the background colour of every element in a NodeList, then you have to change the background colour of each one in turn: with a loop.
for (var i = 0; i < node_list.length; i++) {
var element = node_list[i];
element.style.backgroundColor = "red";
}

You are returning a single element. You can return the array. If you want to be able to act on all elements at once, jQuery style, you can pass a callback into your function;
window.qsa = function(query, callback) {
var els = document.querySelectorAll(query);
if (typeof callback == 'function') {
for (var i = 0; i < els.length; ++i) {
callback.call(els[i], els[i], i);
}
}
return els;
};
qsa('button.change-all', function(btn) {
// You can reference the element using the first parameter
btn.addEventListener('click', function(){
qsa('p', function(p, index){
// Or you can reference the element using `this`
this.innerHTML = 'Changed ' + index;
});
});
});
qsa('button.change-second', function(btn) {
btn.addEventListener('click', function(){
var second = qsa('p')[1];
second.innerHTML = 'Changed just the second one';
});
});
<p>One</p>
<p>Two</p>
<p>Three</p>
<button class='change-all'>Change Paragraphs</button>
<button class='change-second'>Change Second Paragraph</button>
Then you can call either use the callback
qsa('P', function(){
this.innerHTML = 'test';
});
Or you can use the array that is returned
var pList = qsa('p');
var p1 = pList[0];

This loop
for(var i = 0; i < el.length; ++i) {
result = el[i];
}
overwrites your result variable every time. That's why you always get only one element.
You can use the result outside though, and iterate through it. Kinda like
var result = window.qsa(el)
for(var i = 0; i < result.length; ++i) {
var workOn = result[i];
// Do something with workOn
}

Storing elements in memory to prevent updating the DOM too often?

Currently I have a loop that updates the DOM in each iteration; I have learned this is a bad practice & you should update the DOM as little as possible for better speed.
So I was wondering how I go about editing the below so I can store all the elements in one element or something & then do a single DOM addition once the loop ends.
Here is the loop..
for (var i = spot; i < spot + batchSize && i < cats.options.length; i++) {
// Check if the cat is selected
if (cats.options[i].selected == true) {
// Set this category's values to some variables
var cat_id = cats.options[i].getAttribute('value');
var cat_name = cats.options[i].text;
if (checkCatSICAdd(cat_id) === false) {
// Now we create the new element
var new_option = document.createElement('option');
// Add attribute
new_option.setAttribute('value',cat_id);
// Create text node
var new_text_node = document.createTextNode(cat_name);
// Append new text node to new option element we created
new_option.appendChild(new_text_node);
// Append new option tag to select list
sel_cats.appendChild(new_option);
} else {
failed++;
}
}
}

Working with DOM element in the loop is slow - no matter if you attach them to the document or not. Attaching them at the end is a bit faster since only only redraw is required but it's still slow.
The proper way is generating a plain old string containing HTML and attaching this string to the DOM using the innerHTML property of a DOM element.

Your code should be ok. The DOM won't actually redraw until the Javascript has finished executing. However, if you've encountered a problem browser that does perform badly, you could try creating a new select before your loop that is not yet attached to the DOM, populating it as you are now, then replacing sel_cats with that new select at the end. That way, the DOM is only updated once.

Your way is good enough unless you have great many items added to sel_cats - you add to the DOM only once.
The only way to improve efficiency might be to store the options as raw HTML then assign that after the loop:
var arrHTML = [];
for (var i = spot; i < spot + batchSize && i < cats.options.length; i++) {
// Check if the cat is selected
if (cats.options[i].selected == true) {
// Set this category's values to some variables
var cat_id = cats.options[i].value;
var cat_name = cats.options[i].text;
if (checkCatSICAdd(cat_id) === false) {
arrHTML.push("<option value=\"" + cat_id + "\">" + cat_name + "</option>";
}
else {
failed++;
}
}
}
sel_cats.innerHTML = arrHTML.join("");

Once you have the select list assigned to a variable, remove it from the dom using removeChild on its parent tag. You can then use appendChild in the loop before adding the select list back into the dom.

Your code is way bloated, DOM 0 methods will be much faster.
If speed really matters, store spot + batchSize && i < cats.options.length in variables so they aren't re-calcuated on each loop (modern browsers probably don't, but older ones did):
for (var i=spot, j=spot+batchSize, k=cats.options.length; i < j && i < k; i++) {
// Store reference to element
var opt = cats.options[i];
// The selected property is boolean, no need to compare
if (opt.selected) {
// if checkCatSICAdd() returns boolean, just use it
// but maybe you need the boolean comparison
if (checkCatSICAdd(opt.name) === false) {
// Wrapped for posting
sel_cats.options[sel_cats.options.length] =
new Option(opt.value, opt.name);
} else {
failed++;
}
}
}

How can I loop through ALL DOM elements on a page?

I'm trying to loop over ALL elements on a page, so I want to check every element that exists on this page for a special class.
So, how do I say that I want to check EVERY element?

You can pass a * to getElementsByTagName() so that it will return all elements in a page:
var all = document.getElementsByTagName("*");
for (var i=0, max=all.length; i < max; i++) {
// Do something with the element here
}
Note that you could use querySelectorAll(), if it's available (IE9+, CSS in IE8), to just find elements with a particular class.
if (document.querySelectorAll)
var clsElements = document.querySelectorAll(".mySpeshalClass");
else
// loop through all elements instead
This would certainly speed up matters for modern browsers.
Browsers now support foreach on NodeList. This means you can directly loop the elements instead of writing your own for loop.
document.querySelectorAll('*').forEach(function(node) {
// Do whatever you want with the node object.
});
Performance note - Do your best to scope what you're looking for by using a specific selector. A universal selector can return lots of nodes depending on the complexity of the page. Also, consider using document.body.querySelectorAll instead of document.querySelectorAll when you don’t care about <head> children.

Was looking for same. Well, not exactly. I only wanted to list all DOM Nodes.
var currentNode,
ni = document.createNodeIterator(document.documentElement, NodeFilter.SHOW_ELEMENT);
while(currentNode = ni.nextNode()) {
console.log(currentNode.nodeName);
}
To get elements with a specific class, we can use filter function.
var currentNode,
ni = document.createNodeIterator(
document.documentElement,
NodeFilter.SHOW_ELEMENT,
function(node){
return node.classList.contains('toggleable') ? NodeFilter.FILTER_ACCEPT : NodeFilter.FILTER_REJECT;
}
);
while(currentNode = ni.nextNode()) {
console.log(currentNode.nodeName);
}
Found solution on
MDN

As always the best solution is to use recursion:
loop(document);
function loop(node){
// do some thing with the node here
var nodes = node.childNodes;
for (var i = 0; i <nodes.length; i++){
if(!nodes[i]){
continue;
}
if(nodes[i].childNodes.length > 0){
loop(nodes[i]);
}
}
}
Unlike other suggestions, this solution does not require you to create an array for all the nodes, so its more light on the memory. More importantly, it finds more results. I am not sure what those results are, but when testing on chrome it finds about 50% more nodes compared to document.getElementsByTagName("*");

Here is another example on how you can loop through a document or an element:
function getNodeList(elem){
var l=new Array(elem),c=1,ret=new Array();
//This first loop will loop until the count var is stable//
for(var r=0;r<c;r++){
//This loop will loop thru the child element list//
for(var z=0;z<l[r].childNodes.length;z++){
//Push the element to the return array.
ret.push(l[r].childNodes[z]);
if(l[r].childNodes[z].childNodes[0]){
l.push(l[r].childNodes[z]);c++;
}//IF
}//FOR
}//FOR
return ret;
}

For those who are using Jquery
$("*").each(function(i,e){console.log(i+' '+e)});

Andy E. gave a good answer.
I would add, if you feel to select all the childs in some special selector (this need happened to me recently), you can apply the method "getElementsByTagName()" on any DOM object you want.
For an example, I needed to just parse "visual" part of the web page, so I just made this
var visualDomElts = document.body.getElementsByTagName('*');
This will never take in consideration the head part.

from this link
javascript reference
<html>
<head>
<title>A Simple Page</title>
<script language="JavaScript">
<!--
function findhead1()
{
var tag, tags;
// or you can use var allElem=document.all; and loop on it
tags = "The tags in the page are:"
for(i = 0; i < document.all.length; i++)
{
tag = document.all(i).tagName;
tags = tags + "\r" + tag;
}
document.write(tags);
}
// -->
</script>
</head>
<body onload="findhead1()">
<h1>Heading One</h1>
</body>
</html>
UPDATE:EDIT
since my last answer i found better simpler solution
function search(tableEvent)
{
clearResults()
document.getElementById('loading').style.display = 'block';
var params = 'formAction=SearchStocks';
var elemArray = document.mainForm.elements;
for (var i = 0; i < elemArray.length;i++)
{
var element = elemArray[i];
var elementName= element.name;
if(elementName=='formAction')
continue;
params += '&' + elementName+'='+ encodeURIComponent(element.value);
}
params += '&tableEvent=' + tableEvent;
createXmlHttpObject();
sendRequestPost(http_request,'Controller',false,params);
prepareUpdateTableContents();//function js to handle the response out of scope for this question
}

Getting all elements using var all = document.getElementsByTagName("*"); for (var i=0, max=all.length; i < max; i++); is ok if you need to check every element but will result in checking or looping repeating elements or text.
Below is a recursion implementation that checks or loop each element of all DOM elements only once and append:
(Credits to #George Reith for his recursion answer here: Map HTML to JSON)
function mapDOMCheck(html_string, json) {
treeObject = {}
dom = new jsdom.JSDOM(html_string) // use jsdom because DOMParser does not provide client-side Window for element access
document = dom.window.document
element = document.querySelector('html')
// Recurse and loop through DOM elements only once
function treeHTML(element, object) {
var nodeList = element.childNodes;
if (nodeList != null) {
if (nodeList.length) {
object[element.nodeName] = []; // IMPT: empty [] array for parent node to push non-text recursivable elements (see below)
for (var i = 0; i < nodeList.length; i++) {
console.log("nodeName", nodeList[i].nodeName);
if (nodeList[i].nodeType == 3) { // if child node is **final base-case** text node
console.log("nodeValue", nodeList[i].nodeValue);
} else { // else
object[element.nodeName].push({}); // push {} into empty [] array where {} for recursivable elements
treeHTML(nodeList[i], object[element.nodeName][object[element.nodeName].length - 1]);
}
}
}
}
}
treeHTML(element, treeObject);
}

Use *
var allElem = document.getElementsByTagName("*");
for (var i = 0; i < allElem.length; i++) {
// Do something with all element here
}

i think this is really quick
document.querySelectorAll('body,body *').forEach(function(e) {

You can try with
document.getElementsByClassName('special_class');

Javascript Closure and DOM Builder

I am making a DOM builder which I have working succesfully but now I am trying to assign some shorthand functions so that div() -> e("div")
Here is my code:
//assign these objects to a namespace, defaults to window
(function(parent) {
/**
* Creates string of element
* #param tag The element to add
* #param options An object of attributes to add
* #param child ... n Child elements to nest
* #return HTML string to use with innerHTML
*/
var e = function(tag, options) {
var html = '<'+tag;
var children = Array.prototype.slice.apply(arguments, [(typeof options === "string") ? 1 : 2]);
if(options && typeof options !== "string") {
for(var option in options) {
html += ' '+option+'="'+options[option]+'"';
}
}
html += '>';
for(var child in children) {
html += children[child];
}
html += '</'+tag+'>';
return html;
}
//array of tags as shorthand for e(<tag>) THIS PART NOT WORKING
var tags = "div span strong cite em li ul ol table th tr td input form textarea".split(" "), i=0;
for(; i < tags.length; i++) {
(function(el) { //create closure to keep EL in scope
parent[el] = function() {
var args = Array.prototype.slice.call(arguments);
console.log(args);
args[0] = el; //make the first argument the shorthand tag
return e.apply(e,args);
};
})(tags[i]);
}
//assign e to parent
parent.e = e;
})(window);
What's currently happening is the args array is getting modified each time I call one of the shorthand functions and I assume what needs to happen is a closure somewhere so the args array I created does not get affected each time it is called. Here is the output of the unit tests:
div( div( span("Content")), span()) expected: <div><div><span>Content</span></div><span></span></div> result: <div><span></span></div>
div( div( span( e("b", e("b", e("b")))), span())) expected: <div><div><span><b><b><b></b></b></b></span><span></span></div></div> result: <div></div>

Though this doesn't directly answer your question,
for(var el in tags) {
is not entirely correct. tags is an array, not an object, so its properties cannot be enumerated using for (... in ...). Try
for(var el = 0; el < tags.length; el++) {
This can make a huge difference towards the interpreter's understanding of your code... and the correct execution of your algorithm.

Blonde moment, I was overwriting the first element when I meant to use args.unshift(el);

#MvanGeest - doing a for..in on an array is technically allowed. Arrays are still objects in javascript. The index of the array will be the key if iterated using a for..in loop. Obviously not the point of using an array in that case, but thought I would clarify.
#Anurag - the forEach method is not supported in IE8 (not sure about 9) so that might not be a reliable method to use until later on in the future.