Parsing HTML using google gumbo parser

npm install gumbo-parser
4 downloads in the last week
72 downloads in the last month

Gumbo Parser

(npm install not fixed on windows yet)

Using google's gumbo parser to parse HTML in node.

var gumbo = require("gumbo-parser");
var tree = gumbo(htmlstring);


There's only one method:

You can also pass in the options
gumbo(htmlstring, {
  // The tab-stop size, for computing positions in source code that uses tabs.
  // default: 8
  tabStop: 8,
  // Whether or not to stop parsing when the first error is encountered.
  // default: false
  stopOnFirstError: true


  document: {
    // the document element (see below)

  root: {
    // the html element (se 'Element' below)

  nodeName (string) (same as tagname)
  nodeType (number) 1
  tagName (string)  (normalized to lowercase)
  originalTag (string) original text from tag
  originalTagEnd (string) original closing tag from original text, if there was one
  children (array) -> replicating childNodes rather than children,
                      ie all text / comment children are included
  tagNamespace (string) "HTML", "SVG" or "MATHML"
  attributes (array)

  nodeName (string) #text or #cdata-section
  nodeType (number) 3
  textContent (string)

  nodeName (string) #document
  nodeType (number) 9
  children (array)
  hasDoctype true/false
  name: (string)            -> see below
  publicIdentifier (string)       "
  systemIdentifier (string)       "

  nodeName (string) #comment
  nodeType (number) 8
  textContent (string) content comment
  nodeValue (string) same as textcontent

  name: attribute name
  value: attribute value (currently always string, doh)
  nodeType: (number) 2

About html doctypes

An html document will always have the "html". If the document has anything else in the type, for example this html4 doctype:

<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "">

the first part within quotation marks will end up in the document.publicIdentifier, and the second part will be in document.systemIdentifier. You can read more about this here:

Untrusted content

If you plan on using gumbo-parser to clean user input, please read this comment from the gumbo-parsers authors.

Build and test:

node-gyp configure
node-gyp build
npm test


0.1.7 Fixes for build on snow leopard

0.1.6 Adding originalTag, originalTagName and tagNamespace if the tag is unknown, parse originalTag and set in as tag

0.1.5 Updating the gumbo-parser to the latest version. This includes some security fixes, and if you use this for user content, please update.

0.1.4 Temporary workaround for the latest changes in node 0.11, thanks Daniel

0.1.3 Fixes utf-8 bug, thanks Yonatan

0.1.2 Taking the (optional) options argument providing publicIdentifier and systemIdentifer for the doctype

0.1.1 Fix build on node 0.8

0.1.0 Passing { document: document, root: root } instead of only root

npm loves you