Skip to content

Latest commit

 

History

History
195 lines (149 loc) · 6.3 KB

README.md

File metadata and controls

195 lines (149 loc) · 6.3 KB

Gumbo Parser

Build Status

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

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

Usage

There's only one method: gumbo(htmlstring).

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,

  // fragment parsing
  // Option 1: just plain HTML in a 'body' context
  fragment: true

  // Option 2:
  // gumbo-style fragment parsing:
  // can be a valid tag for the ns
  fragmentContext: "div",
  // optional can be 'html', 'svg', 'mathml', defaults to html
  fragmentNamespace: "html"
});

returns:

// if you use normal document mode:
{
  document: {
    // the document element (see below)
  },

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

// if you use fragment parsing:
{
  childNodes: [
    list
  ]
}

Element:
  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)
  startPos (position) -> if element is inserted by parser, this value is undefined
  endPos (position)

TextNode:
  nodeName (string) #text or #cdata-section
  nodeType (number) 3
  textContent (string)
  originalText (string)
  startPos (position)

  note: In DOM3, CDATA is marked as nodeType 4. However, after checking that neither
  firefox, chrome nor safari marks CDATA as 4 (they use 3 or 8), and that CDATA is
  gone in DOM4, i decided to stick with the futuristic alternative.

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

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

Attribute
  name: attribute name
  value: attribute value (currently always string, doh)
  nodeType: (number) 2
  nameStart: (position)
  nameEnd: (position)
  valueStart: (position)
  valueEnd: (position)

Position
  line:   number
  column: number
  offset: number

About html doctypes

An html document will always have the document.name "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"
        "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">

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: http://www.whatwg.org/specs/web-apps/current-work/multipage/syntax.html#syntax-doctype.

Untrusted content / XSS cleaning

If you plan on using gumbo-parser to clean user input, the gumbo parser is one of the most well-tested and audited parsers available. Please read this comment from the gumbo-parsers authors.. There's a node module for XSS cleaning with the gumbo parser. Check Gumbo-Sanitize out!

Node 0.8

Contrary to what i previously said, node-gumbo-parser does build under node 0.8. You might have to npm update -g npm though.

Build and test:

node-gyp configure
node-gyp build
npm test

Changes

0.2.2 Update to use the latest NaN api, so it works for node 4.0

0.2.1 Celebrating some new stuff with a MINOR version change * Fragment parsing supports fragmentContext and fragmentNamespace Uses version 0.10.1, Big changes from the gumbo-parser-team: * Fragment parsing (instead my homebrew fragment parsing, the gumbo c-lib now supports fragments) * Parses all html5lib tests including template * 30-40% speed improvement See all changes here

0.1.13 Upgrade C lib Uses version 0.9.3, CDATA handling (see note in docs) See all changes here

0.1.12 io.js support! Thanks a lot to MicroMike

0.1.11 Upgrade C lib Uses version 0.9.2, performance improvements, duplicate attributes, semicolon fix, See all changes here

0.1.10 Visual Studio bugfix Thanks takenspc

0.1.9 Experimental fragment parsing Expose node positions from the parser, which also enables the user to see if an element is inserted by the parser or was in the text Update gumbo parser to a more secure version Update statement about security

0.1.8 Fix for BSD build problem

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