getDom.js

//USAGE
// needs jsdom, node-htmlparser
// require('getDom.js')(stream_or_string,callback)


//exposes -> response.sax
//  sax.onDirective
//  sax.onElement
//  sax.onAttribute
//  sax.onText
var EventEmitter = require("events").EventEmitter
  , htmlparser = require( "node-htmlparser" )
  , handler = new htmlparser.DefaultHandler( function ( error, dom ) {
    if ( error ) {
      this.callback(error)
    }
    else {
      while ( dom[ 0 ].type == "directive" || dom[ 0 ].type == "text" ) {
        dom.shift()
      }
      var path = [ dom[ 0 ] ]
        , indices = [ 0 ]
        , depth = 0
      while ( path.length ) {
        node = path[ depth ]
        //console.log(sys.inspect(node))
        if(!indices[depth]) switch(node.type) {
          case "tag":
            this.sax.emit("element",node.name)
            var attrs = node.attribs
            for(var property in attrs) {
              this.sax.emit("attribute",property,attrs[property])
            }
            break
          case "script":
            this.sax.emit("element","script")
            var attrs = node.attribs
            for(var property in attrs) {
              this.sax.emit("attribute",property,attrs[property])
            }
            if(attrs && !attrs.src) {
              this.sax.emit("text",node.children && node.children.length ? node.children[0].data : "")
            }
            break
          case "style":
            var attrs = node.attribs
            this.sax.emit("element",attrs && attrs.href ? "link":"style")
            for(var property in attrs) {
              this.sax.emit("attribute",property,attrs[property])
            }
            break
          case "text":
            this.sax.emit("text",node.data)
        }
        //iteration
        if ( node.children ) {
          if ( indices[ depth ] < node.children.length ) {
            path.push( node.children[ indices[ depth ] ] )
            indices[ depth ] += 1
            depth++
            indices[ depth ] = 0
            continue
          }
          else {
            path.pop()
            indices.pop()
            depth--
            if(node.type in {"tag":1,"script":1,"style":1}) this.sax.emit("elementEnd")
          }
        }
        else {
          path.pop()
          indices.pop()
          depth--
          if(node.type in {"tag":1,"script":1,"style":1}) this.sax.emit("elementEnd")
        }
      }
      this.callback(false,this.window)
    }
  }, { enforceEmptyTags: false } )

var jsdom = require("jsdom")
  , dom = jsdom.dom.level1.core

module.exports = function GetDOM(stream_or_string,callback) {	
	var window = jsdom.windowAugmentation(dom);
	var document = browser.document
	document.defaultView = window
	var selected = null
	var sax = new EventEmitter
    var parser = new htmlparser.Parser( handler )
      , parse = function ( data ) {
        parser.parseChunk( data )
      }
      , end = function ( data ) {
        parser.parseChunk( data )
        parser.done()
      }
    parser._handler.sax = sax
    parser._handler.callback = callback
    parser._handler.window = window
  sax.addListener("element", function(elemName) {
	var node = document.createElement(elemName)
	;(selected || document).appendChild(node)
	selected = node
  } )
  sax.addListener("attribute", function(name,value) {
	selected.setAttribute(name,value)
  } )
  sax.addListener("text", function(text) {
	if(selected) selected.appendChild(document.createTextNode(text))
  } )
  sax.addListener("elementEnd", function() {
	selected = selected.parentNode
  } )
  if(typeof stream_or_string == 'string') {
	end(stream_or_string)
  }
  else {
  	steam_or_string.on('data',function(data){
  		parse(data)
  	})
  	steam_or_string.on('end',function(){
  		end()
  	})
  }
}