John Cappiello - Dojo.common-0.4.1

Documentation | Source
dojo.provide("dojo.xml.Parse");
dojo.require("dojo.dom");

//TODO: determine dependencies
// currently has dependency on dojo.xml.DomUtil nodeTypes constants...

// using documentFragment nomenclature to generalize in case we don't want to require passing a collection of nodes with a single parent

dojo.xml.Parse = function(){
	// summary:
	//		generic class for taking a DOM node and parsing it into an object
	//		based on the "dojo tag name" of that node.
	// 
	//		supported dojoTagName's:
	//			<prefix:tag> => prefix:tag
	//			<dojo:tag> => dojo:tag
	//			<dojoTag> => dojo:tag
	//			<tag dojoType="type"> => dojo:type
	//			<tag dojoType="prefix:type"> => prefix:type
	//			<tag dojo:type="type"> => dojo:type
	//			<tag class="classa dojo-type classb"> => dojo:type	

	var isIE = ((dojo.render.html.capable)&&(dojo.render.html.ie));

	// get normalized (lowercase) tagName
	// some browsers report tagNames in lowercase no matter what
	function getTagName(node){
		/*
		return ((node)&&(node["tagName"]) ? node.tagName.toLowerCase() : '');
		*/
		try{
			return node.tagName.toLowerCase();
		}catch(e){
			return "";
		}
	}

	// locate dojo qualified tag name
	function getDojoTagName(node){
		var tagName = getTagName(node);
		if (!tagName){
				return '';
		}
		// any registered tag
		if((dojo.widget)&&(dojo.widget.tags[tagName])){
			return tagName;
		}
		// <prefix:tag> => prefix:tag
		var p = tagName.indexOf(":");
		if(p>=0){
			return tagName;
		}
		// <dojo:tag> => dojo:tag
		if(tagName.substr(0,5) == "dojo:"){
			return tagName;
		}
		if(dojo.render.html.capable && dojo.render.html.ie && node.scopeName != 'HTML'){
			return node.scopeName.toLowerCase() + ':' + tagName;
		}
		// <dojoTag> => dojo:tag
		if(tagName.substr(0,4) == "dojo"){
			// FIXME: this assumes tag names are always lower case
			return "dojo:" + tagName.substring(4);
		}
		// <tag dojoType="prefix:type"> => prefix:type
		// <tag dojoType="type"> => dojo:type
		var djt = node.getAttribute("dojoType") || node.getAttribute("dojotype");
		if(djt){
			if (djt.indexOf(":")<0){
				djt = "dojo:"+djt;
			}
			return djt.toLowerCase();
		}
		// <tag dojo:type="type"> => dojo:type
		djt = node.getAttributeNS && node.getAttributeNS(dojo.dom.dojoml,"type");
		if(djt){
			return "dojo:" + djt.toLowerCase();
		}
		// <tag dojo:type="type"> => dojo:type
		try{
			// FIXME: IE really really doesn't like this, so we squelch errors for it
			djt = node.getAttribute("dojo:type");
		}catch(e){ 
			// FIXME: log?  
		}
		if(djt){ return "dojo:"+djt.toLowerCase(); }
		// <tag class="classa dojo-type classb"> => dojo:type	
		if((dj_global["djConfig"])&&(!djConfig["ignoreClassNames"])){ 
			// FIXME: should we make this optionally enabled via djConfig?
			var classes = node.className||node.getAttribute("class");
			// FIXME: following line, without check for existence of classes.indexOf
			// breaks firefox 1.5's svg widgets
			if((classes )&&(classes.indexOf)&&(classes.indexOf("dojo-")!=-1)){
				var aclasses = classes.split(" ");
				for(var x=0, c=aclasses.length; x<c; x++){
					if(aclasses[x].slice(0, 5) == "dojo-"){
						return "dojo:"+aclasses[x].substr(5).toLowerCase(); 
					}
				}
			}
		}
		// no dojo-qualified name
		return '';
	}


	this.parseElement = function(	/*DomNode*/node,
									/*Boolean*/hasParentNodeSet, 
									/*Boolean*/optimizeForDojoML, 
									/*Integer*/thisIdx	){
		// summary:
		//		recursively parse the passed node, returning a normalized data
		//		structure that represents the "attributes of interest" of said
		//		elements. If optimizeForDojoML is true, only nodes that contain
		//		a "dojo tag name" will be inspected for attributes.
		// node: the DomNode to be treated as the root of inspection
		// hasParentNodeSet: no-op, please pass "null"
		// optimizeForDojoML: should we ignore non-Dojo nodes? Defaults to false.
		// thisIdx:
		//		a way to specify a synthetic "index" property in the resulting
		//		data structure. Otherwise the index property of the top-level
		//		return element is always "0".

		// TODOC: document return structure of a non-trivial element set

		// run shortcuts to bail out of processing up front to save time and
		// object alloc if possible.
		var tagName = getTagName(node);
		//There's a weird bug in IE where it counts end tags, e.g. </dojo:button> as nodes that should be parsed.  Ignore these
		if(isIE && tagName.indexOf("/")==0){ return null; }

		try{
			var attr = node.getAttribute("parseWidgets");
			if(attr && attr.toLowerCase() == "false"){
				return {};
			}
		}catch(e){/*continue*/}

		
		// look for a dojoml qualified name
		// process dojoml only when optimizeForDojoML is true
		var process = true;
		if(optimizeForDojoML){
			var dojoTagName = getDojoTagName(node);
			tagName = dojoTagName || tagName;
			process = Boolean(dojoTagName);
		}

		var parsedNodeSet = {};
		parsedNodeSet[tagName] = [];
		var pos = tagName.indexOf(":");
		if(pos>0){
			var ns = tagName.substring(0,pos);
			parsedNodeSet["ns"] = ns;
			// honor user namespace filters
			if((dojo.ns)&&(!dojo.ns.allow(ns))){process=false;}
		}

		if(process){
			var attributeSet = this.parseAttributes(node);
			for(var attr in attributeSet){
				if((!parsedNodeSet[tagName][attr])||(typeof parsedNodeSet[tagName][attr] != "array")){
					parsedNodeSet[tagName][attr] = [];
				}
				parsedNodeSet[tagName][attr].push(attributeSet[attr]);
			}	
			// FIXME: we might want to make this optional or provide cloning instead of
			// referencing, but for now, we include a node reference to allow
			// instantiated components to figure out their "roots"
			parsedNodeSet[tagName].nodeRef = node;
			parsedNodeSet.tagName = tagName;
			parsedNodeSet.index = thisIdx||0;
		}

		var count = 0;
		for(var i = 0; i < node.childNodes.length; i++){
			var tcn = node.childNodes.item(i);
			switch(tcn.nodeType){
				case  dojo.dom.ELEMENT_NODE: // element nodes, call this function recursively
					var ctn = getDojoTagName(tcn) || getTagName(tcn);
					if(!parsedNodeSet[ctn]){
						parsedNodeSet[ctn] = [];
					}
					parsedNodeSet[ctn].push(this.parseElement(tcn, true, optimizeForDojoML, count));
					if(	(tcn.childNodes.length == 1)&&
						(tcn.childNodes.item(0).nodeType == dojo.dom.TEXT_NODE)){
						parsedNodeSet[ctn][parsedNodeSet[ctn].length-1].value = tcn.childNodes.item(0).nodeValue;
					}
					count++;
					break;
				case  dojo.dom.TEXT_NODE: // if a single text node is the child, treat it as an attribute
					if(node.childNodes.length == 1){
						parsedNodeSet[tagName].push({ value: node.childNodes.item(0).nodeValue });
					}
					break;
				default: break;
				/*
				case  dojo.dom.ATTRIBUTE_NODE: // attribute node... not meaningful here
					break;
				case  dojo.dom.CDATA_SECTION_NODE: // cdata section... not sure if this would ever be meaningful... might be...
					break;
				case  dojo.dom.ENTITY_REFERENCE_NODE: // entity reference node... not meaningful here
					break;
				case  dojo.dom.ENTITY_NODE: // entity node... not sure if this would ever be meaningful
					break;
				case  dojo.dom.PROCESSING_INSTRUCTION_NODE: // processing instruction node... not meaningful here
					break;
				case  dojo.dom.COMMENT_NODE: // comment node... not not sure if this would ever be meaningful 
					break;
				case  dojo.dom.DOCUMENT_NODE: // document node... not sure if this would ever be meaningful
					break;
				case  dojo.dom.DOCUMENT_TYPE_NODE: // document type node... not meaningful here
					break;
				case  dojo.dom.DOCUMENT_FRAGMENT_NODE: // document fragment node... not meaningful here
					break;
				case  dojo.dom.NOTATION_NODE:// notation node... not meaningful here
					break;
				*/
			}
		}
		//return (hasParentNodeSet) ? parsedNodeSet[node.tagName] : parsedNodeSet;
		//if(parsedNodeSet.tagName)dojo.debug("parseElement: RETURNING NODE WITH TAGNAME "+parsedNodeSet.tagName);
		return parsedNodeSet;
	};


	/* parses a set of attributes on a node into an object tree */
	this.parseAttributes = function(/*DomNode*/node){
		// summary:
		// 		creates an attribute object that maps attribute values for the
		// 		passed node. Note that this is similar to creating a JSON
		// 		representation of a DOM node.
		// usage:
		//		a node with the following serialization:
		//			<div foo="bar" baz="thud">...</div>	
		//		would yeild the following return structure when passed into this
		//		function:
		//			{
		//				"foo": {
		//					"value": "bar"
		//				},
		//				"baz": {
		//					"value": "thud"
		//				}
		//			}
		//
		var parsedAttributeSet = {};
		var atts = node.attributes;
		// TODO: should we allow for duplicate attributes at this point...
		// would any of the relevant dom implementations even allow this?
		var attnode, i=0;
		while((attnode=atts[i++])){
			if(isIE){
				if(!attnode){ continue; }
				if((typeof attnode == "object")&&
					(typeof attnode.nodeValue == 'undefined')||
					(attnode.nodeValue == null)||
					(attnode.nodeValue == '')){ 
					continue; 
				}
			}

			var nn = attnode.nodeName.split(":");
			nn = (nn.length == 2) ? nn[1] : attnode.nodeName;
						
			parsedAttributeSet[nn] = { 
				value: attnode.nodeValue 
			};
		}
		return parsedAttributeSet;
	};
};