Straightforward HTML parser for Node.js and browser

  • hyntax

Downloads in past


125151.1.93 years ago5 years agoMinified + gzip package size for hyntax in KB


Hyntax project logo — lego bricks in the shape of a capital letter H

Straightforward HTML parser for JavaScript. Live Demo.
  • Simple. API is straightforward, output is clear.
  • Forgiving. Just like a browser, normally parses invalid HTML.
  • Supports streaming. Can process HTML while it's still being loaded.
  • No dependencies.

Table Of Contents


npm install hyntax

const { tokenize, constructTree } = require('hyntax')
const util = require('util')

const inputHTML = `
      <input type="text" placeholder="Don't type">
      <button>Don't press</button>

const { tokens } = tokenize(inputHTML)
const { ast } = constructTree(tokens)

console.log(JSON.stringify(tokens, null, 2))
console.log(util.inspect(ast, { showHidden: false, depth: null }))

TypeScript Typings

Hyntax is written in JavaScript but has integrated TypeScript typings to help you navigate around its data structures. There is also Types Reference which covers most common types.


Use StreamTokenizer and StreamTreeConstructor classes to parse HTML chunk by chunk while it's still being loaded from the network or read from the disk.
const { StreamTokenizer, StreamTreeConstructor } = require('hyntax')
const http = require('http')
const util = require('util')

http.get('http://info.cern.ch', (res) => {
  const streamTokenizer = new StreamTokenizer()
  const streamTreeConstructor = new StreamTreeConstructor()

  let resultTokens = []
  let resultAst


    .on('data', (tokens) => {
      resultTokens = resultTokens.concat(tokens)
    .on('end', () => {
      console.log(JSON.stringify(resultTokens, null, 2))

    .on('data', (ast) => {
      resultAst = ast
    .on('end', () => {
      console.log(util.inspect(resultAst, { showHidden: false, depth: null }))
}).on('error', (err) => {
  throw err;


Here are all kinds of tokens which Hyntax will extract out of HTML string.
Overview of all possible tokens
Each token conforms to Tokenizer.Token interface.

AST Format

Resulting syntax tree will have at least one top-level Document Node with optional children nodes nested within.

  nodeType: TreeConstructor.NodeTypes.Document,
  content: {
    children: [
        nodeType: TreeConstructor.NodeTypes.AnyNodeType,
        content: {…}
        nodeType: TreeConstructor.NodeTypes.AnyNodeType,
        content: {…}

Content of each node is specific to node's type, all of them are described in AST Node Types reference.

API Reference


Hyntax has its tokenizer as a separate module. You can use generated tokens on their own or pass them further to a tree constructor to build an AST.


tokenize(html: String): Tokenizer.Result


  • html
HTML string to process
Type: string.

Returns Tokenizer.Result

Tree Constructor

After you've got an array of tokens, you can pass them into tree constructor to build an AST.


constructTree(tokens: Tokenizer.AnyToken[]): TreeConstructor.Result


  • tokens
Array of tokens received from the tokenizer.
Type: Tokenizer.AnyToken

Returns TreeConstructor.Result

Types Reference


interface Result {
  state: Tokenizer.State
  tokens: Tokenizer.AnyToken[]

  • state
The current state of tokenizer. It can be persisted and passed to the next tokenizer call if the input is coming in chunks.
  • tokens
Array of resulting tokens.
Type: Tokenizer.AnyToken


interface Result {
  state: State
  ast: AST

  • state
The current state of the tree constructor. Can be persisted and passed to the next tree constructor call in case when tokens are coming in chunks.
  • ast
Resulting AST.
Type: TreeConstructor.AST


Generic Token, other interfaces use it to create a specific Token type.
interface Token<T extends TokenTypes.AnyTokenType> {
  type: T
  content: string
  startPosition: number
  endPosition: number

  • type
One of the Token types.
  • content
Piece of original HTML string which was recognized as a token.
  • startPosition
Index of a character in the input HTML string where the token starts.
  • endPosition
Index of a character in the input HTML string where the token ends.


Shortcut type of all possible tokens.
type AnyTokenType =
  | Text
  | OpenTagStart
  | AttributeKey
  | AttributeAssigment
  | AttributeValueWrapperStart
  | AttributeValue
  | AttributeValueWrapperEnd
  | OpenTagEnd
  | CloseTag
  | OpenTagStartScript
  | ScriptTagContent
  | OpenTagEndScript
  | CloseTagScript
  | OpenTagStartStyle
  | StyleTagContent
  | OpenTagEndStyle
  | CloseTagStyle
  | DoctypeStart
  | DoctypeEnd
  | DoctypeAttributeWrapperStart
  | DoctypeAttribute
  | DoctypeAttributeWrapperEnd
  | CommentStart
  | CommentContent
  | CommentEnd


Shortcut to reference any possible token.
type AnyToken = Token<TokenTypes.AnyTokenType>


Just an alias to DocumentNode. AST always has one top-level DocumentNode. See AST Node Types
type AST = TreeConstructor.DocumentNode

AST Node Types

There are 7 possible types of Node. Each type has a specific content.
type DocumentNode = Node<NodeTypes.Document, NodeContents.Document>	

type DoctypeNode = Node<NodeTypes.Doctype, NodeContents.Doctype>

type TextNode = Node<NodeTypes.Text, NodeContents.Text>

type TagNode = Node<NodeTypes.Tag, NodeContents.Tag>

type CommentNode = Node<NodeTypes.Comment, NodeContents.Comment>

type ScriptNode = Node<NodeTypes.Script, NodeContents.Script>

type StyleNode = Node<NodeTypes.Style, NodeContents.Style>

Interfaces for each content type:


Generic Node, other interfaces use it to create specific Nodes by providing type of Node and type of the content inside the Node.
interface Node<T extends NodeTypes.AnyNodeType, C extends NodeContents.AnyNodeContent> {
  nodeType: T
  content: C


Shortcut type of all possible Node types.
type AnyNodeType =
  | Document
  | Doctype
  | Tag
  | Text
  | Comment
  | Script
  | Style

Node Content Types


Shortcut type of all possible types of content inside a Node.
type AnyNodeContent =
  | Document
  | Doctype
  | Text
  | Tag
  | Comment
  | Script
  | Style


interface Document {
  children: AnyNode[]


interface Doctype {
  start: Tokenizer.Token<Tokenizer.TokenTypes.DoctypeStart>
  attributes?: DoctypeAttribute[]
  end: Tokenizer.Token<Tokenizer.TokenTypes.DoctypeEnd>


interface Text {
  value: Tokenizer.Token<Tokenizer.TokenTypes.Text>


interface Tag {
  name: string
  selfClosing: boolean
  openStart: Tokenizer.Token<Tokenizer.TokenTypes.OpenTagStart>
  attributes?: TagAttribute[]
  openEnd: Tokenizer.Token<Tokenizer.TokenTypes.OpenTagEnd>
  children?: AnyNode[]
  close?: Tokenizer.Token<Tokenizer.TokenTypes.CloseTag>


interface Comment {
  start: Tokenizer.Token<Tokenizer.TokenTypes.CommentStart>
  value: Tokenizer.Token<Tokenizer.TokenTypes.CommentContent>
  end: Tokenizer.Token<Tokenizer.TokenTypes.CommentEnd>


interface Script {
  openStart: Tokenizer.Token<Tokenizer.TokenTypes.OpenTagStartScript>
  attributes?: TagAttribute[]
  openEnd: Tokenizer.Token<Tokenizer.TokenTypes.OpenTagEndScript>
  value: Tokenizer.Token<Tokenizer.TokenTypes.ScriptTagContent>
  close: Tokenizer.Token<Tokenizer.TokenTypes.CloseTagScript>


interface Style {
  openStart: Tokenizer.Token<Tokenizer.TokenTypes.OpenTagStartStyle>,
  attributes?: TagAttribute[],
  openEnd: Tokenizer.Token<Tokenizer.TokenTypes.OpenTagEndStyle>,
  value: Tokenizer.Token<Tokenizer.TokenTypes.StyleTagContent>,
  close: Tokenizer.Token<Tokenizer.TokenTypes.CloseTagStyle>


interface DoctypeAttribute {
  startWrapper?: Tokenizer.Token<Tokenizer.TokenTypes.DoctypeAttributeWrapperStart>,
  value: Tokenizer.Token<Tokenizer.TokenTypes.DoctypeAttribute>,
  endWrapper?: Tokenizer.Token<Tokenizer.TokenTypes.DoctypeAttributeWrapperEnd>


interface TagAttribute {
  key?: Tokenizer.Token<Tokenizer.TokenTypes.AttributeKey>,
  startWrapper?: Tokenizer.Token<Tokenizer.TokenTypes.AttributeValueWrapperStart>,
  value?: Tokenizer.Token<Tokenizer.TokenTypes.AttributeValue>,
  endWrapper?: Tokenizer.Token<Tokenizer.TokenTypes.AttributeValueWrapperEnd>