analytics/webcontent/dojo17/dijit/_editor/html.js

define("dijit/_editor/html", [
	"dojo/_base/array",
	"dojo/_base/lang", // lang.getObject
	"dojo/_base/sniff", // has("ie")
	".."		// for exporting symbols to dijit._editor (remove for 2.0)
], function(array, lang, has, dijit){


// module:
//		dijit/_editor/html
// summary:
//		Utility functions used by editor

// Tests for DOMNode.attributes[] behavior:
//	 - dom-attributes-explicit - attributes[] only lists explicitly user specified attributes
//	 - dom-attributes-specified-flag (IE8) - need to check attr.specified flag to skip attributes user didn't specify
//	 - Otherwise, in IE6-7. attributes[] will list hundreds of values, so need to do outerHTML to get attrs instead.
var form = document.createElement("form");
has.add("dom-attributes-explicit", form.attributes.length == 0); // W3C
has.add("dom-attributes-specified-flag", form.attributes.length > 0 && form.attributes.length < 40);	// IE8

lang.getObject("_editor", true, dijit);

dijit._editor.escapeXml=function(/*String*/str, /*Boolean?*/noSingleQuotes){
	// summary:
	//		Adds escape sequences for special characters in XML: &<>"'
	//		Optionally skips escapes for single quotes
	str = str.replace(/&/gm, "&amp;").replace(/</gm, "&lt;").replace(/>/gm, "&gt;").replace(/"/gm, "&quot;");
	if(!noSingleQuotes){
		str = str.replace(/'/gm, "&#39;");
	}
	return str; // string
};

dijit._editor.getNodeHtml=function(/* DomNode */node){
	var output;
	switch(node.nodeType){
		case 1: //element node
			var lName = node.nodeName.toLowerCase();
			if(!lName || lName.charAt(0) == "/"){
				// IE does some strange things with malformed HTML input, like
				// treating a close tag </span> without an open tag <span>, as
				// a new tag with tagName of /span.  Corrupts output HTML, remove
				// them.  Other browsers don't prefix tags that way, so will
				// never show up.
				return "";
			}
			output = '<' + lName;

			//store the list of attributes and sort it to have the
			//attributes appear in the dictionary order
			var attrarray = [], attrhash = {};
			var attr;
			if(has("dom-attributes-explicit") || has("dom-attributes-specified-flag")){
				// IE8+ and all other browsers.
				var i = 0;
				while((attr = node.attributes[i++])){
					// ignore all attributes starting with _dj which are
					// internal temporary attributes used by the editor
					var n = attr.name;
					if(n.substr(0,3) !== '_dj' &&
						(!has("dom-attributes-specified-flag") || attr.specified) &&
						!(n in attrhash)){	// workaround repeated attributes bug in IE8 (LinkDialog test)
						var v = attr.value;
						if(n == 'src' || n == 'href'){
							if(node.getAttribute('_djrealurl')){
								v = node.getAttribute('_djrealurl');
							}
						}
						if(has("ie") === 8 && n === "style"){
							v = v.replace("HEIGHT:", "height:").replace("WIDTH:", "width:");
						}
						attrarray.push([n,v]);
						attrhash[n] = v;
					}
				}
			}else{
				// IE6-7 code path
				var clone = /^input$|^img$/i.test(node.nodeName) ? node : node.cloneNode(false);
				var s = clone.outerHTML;
				// Split up and manage the attrs via regexp
				// similar to prettyPrint attr logic.
				var rgxp_attrsMatch = /[\w-]+=("[^"]*"|'[^']*'|\S*)/gi
				var attrSplit = s.match(rgxp_attrsMatch);
				s = s.substr(0, s.indexOf('>'));
				array.forEach(attrSplit, function(attr){
					if(attr){
						var idx = attr.indexOf("=");
						if(idx > 0){
							var key = attr.substring(0,idx);
							if(key.substr(0,3) != '_dj'){
								if(key == 'src' || key == 'href'){
									if(node.getAttribute('_djrealurl')){
										attrarray.push([key,node.getAttribute('_djrealurl')]);
										return;
									}
								}
								var val, match;
								switch(key){
									case 'style':
										val = node.style.cssText.toLowerCase();
										break;
									case 'class':
										val = node.className;
										break;
									case 'width':
										if(lName === "img"){
											// This somehow gets lost on IE for IMG tags and the like
											// and we have to find it in outerHTML, known IE oddity.
											match=/width=(\S+)/i.exec(s);
											if(match){
												val = match[1];
											}
											break;
										}
									case 'height':
										if(lName === "img"){
											// This somehow gets lost on IE for IMG tags and the like
											// and we have to find it in outerHTML, known IE oddity.
											match=/height=(\S+)/i.exec(s);
											if(match){
												val = match[1];
											}
											break;
										}
									default:
										val = node.getAttribute(key);
								}
								if(val != null){
									attrarray.push([key, val.toString()]);
								}
							}
						}
					}
				}, this);
			}
			attrarray.sort(function(a,b){
				return a[0] < b[0] ? -1 : (a[0] == b[0] ? 0 : 1);
			});
			var j = 0;
			while((attr = attrarray[j++])){
				output += ' ' + attr[0] + '="' +
					(lang.isString(attr[1]) ? dijit._editor.escapeXml(attr[1], true) : attr[1]) + '"';
			}
			if(lName === "script"){
				// Browsers handle script tags differently in how you get content,
				// but innerHTML always seems to work, so insert its content that way
				// Yes, it's bad to allow script tags in the editor code, but some people
				// seem to want to do it, so we need to at least return them right.
				// other plugins/filters can strip them.
				output += '>' + node.innerHTML +'</' + lName + '>';
			}else{
				if(node.childNodes.length){
					output += '>' + dijit._editor.getChildrenHtml(node)+'</' + lName +'>';
				}else{
					switch(lName){
						case 'br':
						case 'hr':
						case 'img':
						case 'input':
						case 'base':
						case 'meta':
						case 'area':
						case 'basefont':
							// These should all be singly closed
							output += ' />';
							break;
						default:
							// Assume XML style separate closure for everything else.
							output += '></' + lName + '>';
					}
				}
			}
			break;
		case 4: // cdata
		case 3: // text
			// FIXME:
			output = dijit._editor.escapeXml(node.nodeValue, true);
			break;
		case 8: //comment
			// FIXME:
			output = '<!--' + dijit._editor.escapeXml(node.nodeValue, true) + '-->';
			break;
		default:
			output = "<!-- Element not recognized - Type: " + node.nodeType + " Name: " + node.nodeName + "-->";
	}
	return output;
};

dijit._editor.getChildrenHtml = function(/* DomNode */dom){
	// summary:
	//		Returns the html content of a DomNode and children
	var out = "";
	if(!dom){ return out; }
	var nodes = dom["childNodes"] || dom;

	//IE issue.
	//If we have an actual node we can check parent relationships on for IE,
	//We should check, as IE sometimes builds invalid DOMS.  If no parent, we can't check
	//And should just process it and hope for the best.
	var checkParent = !has("ie") || nodes !== dom;

	var node, i = 0;
	while((node = nodes[i++])){
		//IE is broken.  DOMs are supposed to be a tree.  But in the case of malformed HTML, IE generates a graph
		//meaning one node ends up with multiple references (multiple parents).  This is totally wrong and invalid, but
		//such is what it is.  We have to keep track and check for this because otherise the source output HTML will have dups.
		//No other browser generates a graph.  Leave it to IE to break a fundamental DOM rule.  So, we check the parent if we can
		//If we can't, nothing more we can do other than walk it.
		if(!checkParent || node.parentNode == dom){
			out += dijit._editor.getNodeHtml(node);
		}
	}
	return out; // String
};

return dijit._editor;
});