Ajay Poshak

September 16, 2019

Walking the DOM

Document Object Model (DOM) is the HTML structure of the page. Traversing the DOM comes handy in cases such as to access the last element of a carousel or to extract all text content from a page.

Let's understand DOM first

All operations in DOM start with document object. It is the entry point of DOM.

Every thing in DOM can be classified as node. And node can be of three types: element, text, comment.

  1. text - In general, text nodes are leaf nodes, which means that they do not contain any child nodes. They only contain a string, sometimes that string can be the newline character among other things.

  2. Element - It is any HTML Element.

  3. Comment is also a node in DOM.

To differentiate among different nodes whether it is an text, element or comment, we need to examine its nodeType property:

For example, consider following HTML snippet:

1<!-- A Paragraph tag -->
3<p>This is a text</p>

It has one element - the p tag, and one text node - "This is a text" and also one comment node, "<!-- A Paragraph tag -->"

Above concept has one exception, the document node, which is not an HTML Element.

DOM interface has methods accordingly:

document.body.children - list all children element

document.body.childNodes - lists all children nodes

Please refer to above section for difference between element and nodes.


This property gives us access to all child nodes to any given DOMElement. But it has one interesting detail. The list returned by it, is not an actual array. It is an iterable list. Which means that we can use for-of loop to iterate over it, but cannot use regular array methods like Array.map.

We can still convert that list to array using Array.from, and then can use all sorts of array methods.

DOM Walker

Now let's focus on a contrived example. Lets say we need to extract all text data from a web page.

To do that we need to apply all we have learned here. We would start from document's body then process all its children and store any text we find in an array. Then do the same to the children of document's children until there are no children left for processing.

We can use nodeType to differentiate between a text node and an element node. So the pseudoCode would like this

  1. Take the document.body as root element and push it to an array(traversalQ)
  2. Use that array as queue, and keep removing elements from the front until array is empty:
    • Remove the first element from array (current)
    • Check if this element is a textNode, if yes then push its textContent to result array
    • Or if this element is a elementNode, then check if it has any children using childNode.length, then push all its children to traversalQ array
    • Repeat step 2)

Its JS implementation would look like this.

1function getAllTextInDocument(root) {
2  const texts = [];
4  const traversalQ = [];
6  traversalQ.push(root);
8  while (traversalQ.length > 0) {
9    const current = traversalQ.shift();
11    // If text node then push in texts array
12    if (current && current.nodeType === 3 && current.textContent.trim()) {
13      // Here trim helps us to remove newline text nodes
14      texts.push(current.textContent);
15    } else if (current && current.childNodes.length > 0) {
16      // If it is an element then push it in traversalQ for further traversal
18      const currentChildren = current.childNodes;
20      for (const child of currentChildren) {
21        traversalQ.push(child);
22      }
23    }
24  }

There is one more detail in above example. Whenever we are traversing the DOM, then the newlines are also considered as textNode. But we don't want newlines in our result, so we trimmed the textContent value of textNode to remove any newlines.

A few more bits about DOM

There are common properties on all DOM elements. And they inherit those properties from their parents. Each DOM node belongs to a corresponsing built-in class.The root of hierarchy is EventTarget class.

DOM Classes hierarchy

These classes are:

EventTarget: As it is a root interface hence all DOM elements inherits behaviour defined in it. addEventListener, removeEventListener & disptachEvent are the methods provided in this class.

Node: It is an abstract class which serves as base class for all DOM elements. It provides methods like firstChild, lastChild, previousSibling, nextSibling and properties like textContent.

Text, Comment and Node: These classes inherits from Node interface. They represent text nodes, element nodes & comment nodes.

HTMLElement: It provides metadata attributes like title, lang, translate, and user interaction elements like draggable, innerText, accessKey attributes.

HTMLInputElement, HTMLBodyElement, HTMLFormElement: These elements inherit HTMLElement. HTMLInputElement deals with <input> tag, HTMLBodyElement relates to <body> tag, and HTMLFormElement corresponds to <form> tag.

Other interesting thing is that all these tags inherit from Object as well, so they get "plain object" methods like 'hasOwnProperty' available to them.

If we try doing console.log(document.body) on browser console, and observe the prototype chain then we can see the above hierarchy in it.

Happy DOM Walking !!!