@xmldom/xmldom 0.9.0-beta.1 → 0.9.0-beta.10

package/lib/sax.js

@@ -1,246 +1,224 @@
-'use strict'
+'use strict';
+
+var conventions = require('./conventions');
+var g = require('./grammar');
 
-var conventions = require("./conventions");
-var isHTMLRawTextElement = conventions.isHTMLRawTextElement;
 var isHTMLEscapableRawTextElement = conventions.isHTMLEscapableRawTextElement;
+var isHTMLMimeType = conventions.isHTMLMimeType;
+var isHTMLRawTextElement = conventions.isHTMLRawTextElement;
 var NAMESPACE = conventions.NAMESPACE;
-var MIME_TYPE = conventions.MIME_TYPE;
-
-//[4]   	NameStartChar	   ::=   	":" | [A-Z] | "_" | [a-z] | [#xC0-#xD6] | [#xD8-#xF6] | [#xF8-#x2FF] | [#x370-#x37D] | [#x37F-#x1FFF] | [#x200C-#x200D] | [#x2070-#x218F] | [#x2C00-#x2FEF] | [#x3001-#xD7FF] | [#xF900-#xFDCF] | [#xFDF0-#xFFFD] | [#x10000-#xEFFFF]
-//[4a]   	NameChar	   ::=   	NameStartChar | "-" | "." | [0-9] | #xB7 | [#x0300-#x036F] | [#x203F-#x2040]
-//[5]   	Name	   ::=   	NameStartChar (NameChar)*
-var nameStartChar = /[A-Z_a-z\xC0-\xD6\xD8-\xF6\u00F8-\u02FF\u0370-\u037D\u037F-\u1FFF\u200C-\u200D\u2070-\u218F\u2C00-\u2FEF\u3001-\uD7FF\uF900-\uFDCF\uFDF0-\uFFFD]///\u10000-\uEFFFF
-var nameChar = new RegExp("[\\-\\.0-9"+nameStartChar.source.slice(1,-1)+"\\u00B7\\u0300-\\u036F\\u203F-\\u2040]");
-var tagNamePattern = new RegExp('^'+nameStartChar.source+nameChar.source+'*(?:\:'+nameStartChar.source+nameChar.source+'*)?$');
-//var tagNamePattern = /^[a-zA-Z_][\w\-\.]*(?:\:[a-zA-Z_][\w\-\.]*)?$/
+var ParseError = conventions.ParseError;
+
 //var handlers = 'resolveEntity,getExternalSubset,characters,endDocument,endElement,endPrefixMapping,ignorableWhitespace,processingInstruction,setDocumentLocator,skippedEntity,startDocument,startElement,startPrefixMapping,notationDecl,unparsedEntityDecl,error,fatalError,warning,attributeDecl,elementDecl,externalEntityDecl,internalEntityDecl,comment,endCDATA,endDTD,endEntity,startCDATA,startDTD,startEntity'.split(',')
 
 //S_TAG,	S_ATTR,	S_EQ,	S_ATTR_NOQUOT_VALUE
 //S_ATTR_SPACE,	S_ATTR_END,	S_TAG_SPACE, S_TAG_CLOSE
-var S_TAG = 0;//tag name offerring
-var S_ATTR = 1;//attr name offerring
-var S_ATTR_SPACE=2;//attr name end and space offer
-var S_EQ = 3;//=space?
-var S_ATTR_NOQUOT_VALUE = 4;//attr value(no quot value only)
-var S_ATTR_END = 5;//attr value end and no space(quot end)
-var S_TAG_SPACE = 6;//(attr value end || tag end ) && (space offer)
-var S_TAG_CLOSE = 7;//closed el<el />
-
-/**
- * Creates an error that will not be caught by XMLReader aka the SAX parser.
- *
- * @param {string} message
- * @param {any?} locator Optional, can provide details about the location in the source
- * @constructor
- */
-function ParseError(message, locator) {
-	this.message = message
-	this.locator = locator
-	if(Error.captureStackTrace) Error.captureStackTrace(this, ParseError);
-}
-ParseError.prototype = new Error();
-ParseError.prototype.name = ParseError.name
-
-function XMLReader(){
+var S_TAG = 0; //tag name offerring
+var S_ATTR = 1; //attr name offerring
+var S_ATTR_SPACE = 2; //attr name end and space offer
+var S_EQ = 3; //=space?
+var S_ATTR_NOQUOT_VALUE = 4; //attr value(no quot value only)
+var S_ATTR_END = 5; //attr value end and no space(quot end)
+var S_TAG_SPACE = 6; //(attr value end || tag end ) && (space offer)
+var S_TAG_CLOSE = 7; //closed el<el />
 
-}
+function XMLReader() {}
 
 XMLReader.prototype = {
-	parse:function(source,defaultNSMap,entityMap){
+	parse: function (source, defaultNSMap, entityMap) {
 		var domBuilder = this.domBuilder;
 		domBuilder.startDocument();
-		_copy(defaultNSMap ,defaultNSMap = {})
-		parse(source,defaultNSMap,entityMap,
-				domBuilder,this.errorHandler);
+		_copy(defaultNSMap, (defaultNSMap = {}));
+		parse(source, defaultNSMap, entityMap, domBuilder, this.errorHandler);
 		domBuilder.endDocument();
-	}
-}
-function parse(source,defaultNSMapCopy,entityMap,domBuilder,errorHandler){
-	var isHTML = MIME_TYPE.isHTML(domBuilder.mimeType);
+	},
+};
+
+function parse(source, defaultNSMapCopy, entityMap, domBuilder, errorHandler) {
+	var isHTML = isHTMLMimeType(domBuilder.mimeType);
+
 	function fixedFromCharCode(code) {
 		// String.prototype.fromCharCode does not supports
 		// > 2 bytes unicode chars directly
 		if (code > 0xffff) {
 			code -= 0x10000;
-			var surrogate1 = 0xd800 + (code >> 10)
-				, surrogate2 = 0xdc00 + (code & 0x3ff);
+			var surrogate1 = 0xd800 + (code >> 10),
+				surrogate2 = 0xdc00 + (code & 0x3ff);
 
 			return String.fromCharCode(surrogate1, surrogate2);
 		} else {
 			return String.fromCharCode(code);
 		}
 	}
-	function entityReplacer(a){
-		var k = a.slice(1,-1);
+
+	function entityReplacer(a) {
+		var k = a.slice(1, -1);
 		if (Object.hasOwnProperty.call(entityMap, k)) {
 			return entityMap[k];
-		}else if(k.charAt(0) === '#'){
-			return fixedFromCharCode(parseInt(k.substr(1).replace('x','0x')))
-		}else{
-			errorHandler.error('entity not found:'+a);
+		} else if (k.charAt(0) === '#') {
+			return fixedFromCharCode(parseInt(k.substr(1).replace('x', '0x')));
+		} else {
+			errorHandler.error('entity not found:' + a);
 			return a;
 		}
 	}
-	function appendText(end){//has some bugs
-		if(end>start){
-			var xt = source.substring(start,end).replace(/&#?\w+;/g,entityReplacer);
-			locator&&position(start);
-			domBuilder.characters(xt,0,end-start);
-			start = end
+
+	function appendText(end) {
+		//has some bugs
+		if (end > start) {
+			var xt = source.substring(start, end).replace(/&#?\w+;/g, entityReplacer);
+			locator && position(start);
+			domBuilder.characters(xt, 0, end - start);
+			start = end;
 		}
 	}
-	function position(p,m){
-		while(p>=lineEnd && (m = linePattern.exec(source))){
+
+	function position(p, m) {
+		while (p >= lineEnd && (m = linePattern.exec(source))) {
 			lineStart = m.index;
 			lineEnd = lineStart + m[0].length;
 			locator.lineNumber++;
-			//console.log('line++:',locator,startPos,endPos)
 		}
-		locator.columnNumber = p-lineStart+1;
+		locator.columnNumber = p - lineStart + 1;
 	}
+
 	var lineStart = 0;
 	var lineEnd = 0;
-	var linePattern = /.*(?:\r\n?|\n)|.*$/g
+	var linePattern = /.*(?:\r\n?|\n)|.*$/g;
 	var locator = domBuilder.locator;
 
-	var parseStack = [{currentNSMap:defaultNSMapCopy}]
+	var parseStack = [{ currentNSMap: defaultNSMapCopy }];
 	var closeMap = {};
 	var start = 0;
-	while(true){
-		try{
-			var tagStart = source.indexOf('<',start);
-			if(tagStart<0){
-				if(!source.substr(start).match(/^\s*$/)){
+	while (true) {
+		try {
+			var tagStart = source.indexOf('<', start);
+			if (tagStart < 0) {
+				if (!source.substr(start).match(/^\s*$/)) {
 					var doc = domBuilder.doc;
-	    			var text = doc.createTextNode(source.substr(start));
-	    			doc.appendChild(text);
-	    			domBuilder.currentElement = text;
+					var text = doc.createTextNode(source.substr(start));
+					doc.appendChild(text);
+					domBuilder.currentElement = text;
 				}
 				return;
 			}
-			if(tagStart>start){
+			if (tagStart > start) {
 				appendText(tagStart);
 			}
-			switch(source.charAt(tagStart+1)){
-			case '/':
-				var end = source.indexOf('>',tagStart+3);
-				var tagName = source.substring(tagStart + 2, end).replace(/[ \t\n\r]+$/g, '');
-				var config = parseStack.pop();
-				if(end<0){
-
-	        		tagName = source.substring(tagStart+2).replace(/[\s<].*/,'');
-	        		errorHandler.error("end tag name: "+tagName+' is not complete:'+config.tagName);
-	        		end = tagStart+1+tagName.length;
-	        	}else if(tagName.match(/\s</)){
-	        		tagName = tagName.replace(/[\s<].*/,'');
-	        		errorHandler.error("end tag name: "+tagName+' maybe not complete');
-	        		end = tagStart+1+tagName.length;
-				}
-				var localNSMap = config.localNSMap;
-				var endMatch = config.tagName == tagName;
-				var endIgnoreCaseMach = endMatch || config.tagName&&config.tagName.toLowerCase() == tagName.toLowerCase()
-		        if(endIgnoreCaseMach){
-		        	domBuilder.endElement(config.uri,config.localName,tagName);
-					if(localNSMap){
-						for(var prefix in localNSMap){
-							domBuilder.endPrefixMapping(prefix) ;
-						}
+			switch (source.charAt(tagStart + 1)) {
+				case '/':
+					var config = parseStack.pop();
+					var end = source.indexOf('>', tagStart + 3);
+					var tagNameRaw = source.substring(tagStart + 2, end > 0 ? end : undefined);
+					var tagNameMatch = g.QName_group.exec(tagNameRaw);
+					// for the root level the config does not contain the tagName
+					var tagName =
+						tagNameMatch && tagNameMatch[1] ? tagNameMatch[1] : config.tagName || domBuilder.doc.documentElement.tagName;
+					if (end < 0) {
+						errorHandler.error('end tag name: ' + tagName + ' is not complete');
+						end = tagStart + 1 + tagName.length;
+					} else if (tagNameRaw.match(/</) && !isHTML) {
+						errorHandler.error('end tag name: ' + tagName + ' maybe not complete');
 					}
-					if(!endMatch){
-		            	errorHandler.fatalError("end tag name: "+tagName+' is not match the current start tagName:'+config.tagName ); // No known test case
+					var localNSMap = config.localNSMap;
+					var endMatch = config.tagName == tagName;
+					var endIgnoreCaseMach = endMatch || (config.tagName && config.tagName.toLowerCase() == tagName.toLowerCase());
+					if (endIgnoreCaseMach) {
+						domBuilder.endElement(config.uri, config.localName, tagName);
+						if (localNSMap) {
+							for (var prefix in localNSMap) {
+								if (Object.prototype.hasOwnProperty.call(localNSMap, prefix)) {
+									domBuilder.endPrefixMapping(prefix);
+								}
+							}
+						}
+						if (!endMatch) {
+							// No known test case
+							return errorHandler.fatalError(
+								'end tag name: ' + tagName + ' is not match the current start tagName:' + config.tagName
+							);
+						}
+					} else {
+						parseStack.push(config);
 					}
-		        }else{
-		        	parseStack.push(config)
-		        }
 
-				end++;
-				break;
+					end++;
+					break;
 				// end elment
-			case '?':// <?...?>
-				locator&&position(tagStart);
-				end = parseInstruction(source,tagStart,domBuilder);
-				break;
-			case '!':// <!doctype,<![CDATA,<!--
-				locator&&position(tagStart);
-				end = parseDCC(source,tagStart,domBuilder,errorHandler);
-				break;
-			default:
-				locator&&position(tagStart);
-				var el = new ElementAttributes();
-				var currentNSMap = parseStack[parseStack.length-1].currentNSMap;
-				//elStartEnd
-				var end = parseElementStartPart(
-					source,
-					tagStart,
-					el,
-					currentNSMap,
-					entityReplacer,
-					errorHandler,
-					isHTML
-				)
-				var len = el.length;
-
-
-				if(!el.closed && fixSelfClosed(source,end,el.tagName,closeMap)){
-					el.closed = true;
-					if(!isHTML){
-						errorHandler.warning('unclosed xml attribute');
-					}
-				}
-				if(locator && len){
-					var locator2 = copyLocator(locator,{});
-					//try{//attribute position fixed
-					for(var i = 0;i<len;i++){
-						var a = el[i];
-						position(a.offset);
-						a.locator = copyLocator(locator,{});
-					}
-					domBuilder.locator = locator2
-					if(appendElement(el,domBuilder,currentNSMap)){
-						parseStack.push(el)
+				case '?': // <?...?>
+					locator && position(tagStart);
+					end = parseProcessingInstruction(source, tagStart, domBuilder, errorHandler);
+					break;
+				case '!': // <!doctype,<![CDATA,<!--
+					locator && position(tagStart);
+					end = parseDoctypeCommentOrCData(source, tagStart, domBuilder, errorHandler);
+					break;
+				default:
+					locator && position(tagStart);
+					var el = new ElementAttributes();
+					var currentNSMap = parseStack[parseStack.length - 1].currentNSMap;
+					//elStartEnd
+					var end = parseElementStartPart(source, tagStart, el, currentNSMap, entityReplacer, errorHandler, isHTML);
+					var len = el.length;
+
+					if (!el.closed && fixSelfClosed(source, end, el.tagName, closeMap)) {
+						el.closed = true;
+						if (!isHTML) {
+							errorHandler.warning('unclosed xml attribute');
+						}
 					}
-					domBuilder.locator = locator;
-				}else{
-					if(appendElement(el,domBuilder,currentNSMap)){
-						parseStack.push(el)
+					if (locator && len) {
+						var locator2 = copyLocator(locator, {});
+						//try{//attribute position fixed
+						for (var i = 0; i < len; i++) {
+							var a = el[i];
+							position(a.offset);
+							a.locator = copyLocator(locator, {});
+						}
+						domBuilder.locator = locator2;
+						if (appendElement(el, domBuilder, currentNSMap)) {
+							parseStack.push(el);
+						}
+						domBuilder.locator = locator;
+					} else {
+						if (appendElement(el, domBuilder, currentNSMap)) {
+							parseStack.push(el);
+						}
 					}
-				}
 
-				if (isHTML && !el.closed) {
-					end = parseHtmlSpecialContent(source,end,el.tagName,entityReplacer,domBuilder)
-				} else {
-					end++;
-				}
+					if (isHTML && !el.closed) {
+						end = parseHtmlSpecialContent(source, end, el.tagName, entityReplacer, domBuilder);
+					} else {
+						end++;
+					}
 			}
-		}catch(e){
+		} catch (e) {
 			if (e instanceof ParseError) {
 				throw e;
 			}
-			errorHandler.error('element parse error: '+e)
+			errorHandler.error('element parse error: ' + e);
 			end = -1;
 		}
-		if(end>start){
+		if (end > start) {
 			start = end;
-		}else{
+		} else {
 			//TODO: 这里有可能sax回退，有位置错误风险
-			appendText(Math.max(tagStart,start)+1);
+			appendText(Math.max(tagStart, start) + 1);
 		}
 	}
 }
-function copyLocator(f,t){
+
+function copyLocator(f, t) {
 	t.lineNumber = f.lineNumber;
 	t.columnNumber = f.columnNumber;
 	return t;
 }
 
 /**
- * @see #appendElement(source,elStartEnd,el,selfClosed,entityReplacer,domBuilder,parseStack);
- * @return end of the elementStartPart(end of elementEndPart for selfClosed el)
+ * @returns end of the elementStartPart(end of elementEndPart for selfClosed el)
+ * @see {@link #appendElement}
  */
-function parseElementStartPart(
-	source,start,el,currentNSMap,entityReplacer,errorHandler, isHTML
-){
-
+function parseElementStartPart(source, start, el, currentNSMap, entityReplacer, errorHandler, isHTML) {
 	/**
 	 * @param {string} qname
 	 * @param {string} value
@@ -248,7 +226,7 @@ function parseElementStartPart(
 	 */
 	function addAttribute(qname, value, startIndex) {
 		if (el.attributeNames.hasOwnProperty(qname)) {
-			errorHandler.fatalError('Attribute ' + qname + ' redefined')
+			return errorHandler.fatalError('Attribute ' + qname + ' redefined');
 		}
 		el.addValue(
 			qname,
@@ -258,404 +236,635 @@ function parseElementStartPart(
 			// - trimming and collapsing multiple spaces into a single one for attributes that are not of type CDATA
 			value.replace(/[\t\n\r]/g, ' ').replace(/&#?\w+;/g, entityReplacer),
 			startIndex
-		)
+		);
 	}
+
 	var attrName;
 	var value;
 	var p = ++start;
-	var s = S_TAG;//status
-	while(true){
+	var s = S_TAG; //status
+	while (true) {
 		var c = source.charAt(p);
-		switch(c){
-		case '=':
-			if(s === S_ATTR){//attrName
-				attrName = source.slice(start,p);
-				s = S_EQ;
-			}else if(s === S_ATTR_SPACE){
-				s = S_EQ;
-			}else{
-				//fatalError: equal must after attrName or space after attrName
-				throw new Error('attribute equal must after attrName'); // No known test case
-			}
-			break;
-		case '\'':
-		case '"':
-			if(s === S_EQ || s === S_ATTR //|| s == S_ATTR_SPACE
-				){//equal
-				if(s === S_ATTR){
-					errorHandler.warning('attribute value must after "="')
-					attrName = source.slice(start,p)
+		switch (c) {
+			case '=':
+				if (s === S_ATTR) {
+					//attrName
+					attrName = source.slice(start, p);
+					s = S_EQ;
+				} else if (s === S_ATTR_SPACE) {
+					s = S_EQ;
+				} else {
+					//fatalError: equal must after attrName or space after attrName
+					throw new Error('attribute equal must after attrName'); // No known test case
 				}
-				start = p+1;
-				p = source.indexOf(c,start)
-				if(p>0){
+				break;
+			case "'":
+			case '"':
+				if (
+					s === S_EQ ||
+					s === S_ATTR //|| s == S_ATTR_SPACE
+				) {
+					//equal
+					if (s === S_ATTR) {
+						errorHandler.warning('attribute value must after "="');
+						attrName = source.slice(start, p);
+					}
+					start = p + 1;
+					p = source.indexOf(c, start);
+					if (p > 0) {
+						value = source.slice(start, p);
+						addAttribute(attrName, value, start - 1);
+						s = S_ATTR_END;
+					} else {
+						//fatalError: no end quot match
+						throw new Error("attribute value no end '" + c + "' match");
+					}
+				} else if (s == S_ATTR_NOQUOT_VALUE) {
 					value = source.slice(start, p);
-					addAttribute(attrName, value, start-1);
+					addAttribute(attrName, value, start);
+					errorHandler.warning('attribute "' + attrName + '" missed start quot(' + c + ')!!');
+					start = p + 1;
 					s = S_ATTR_END;
-				}else{
-					//fatalError: no end quot match
-					throw new Error('attribute value no end \''+c+'\' match');
+				} else {
+					//fatalError: no equal before
+					throw new Error('attribute value must after "="'); // No known test case
 				}
-			}else if(s == S_ATTR_NOQUOT_VALUE){
-				value = source.slice(start, p);
-				addAttribute(attrName, value, start);
-				errorHandler.warning('attribute "'+attrName+'" missed start quot('+c+')!!');
-				start = p+1;
-				s = S_ATTR_END
-			}else{
-				//fatalError: no equal before
-				throw new Error('attribute value must after "="'); // No known test case
-			}
-			break;
-		case '/':
-			switch(s){
-			case S_TAG:
-				el.setTagName(source.slice(start,p));
-			case S_ATTR_END:
-			case S_TAG_SPACE:
-			case S_TAG_CLOSE:
-				s =S_TAG_CLOSE;
-				el.closed = true;
-			case S_ATTR_NOQUOT_VALUE:
-			case S_ATTR:
-			case S_ATTR_SPACE:
 				break;
-			//case S_EQ:
-			default:
-				throw new Error("attribute invalid close char('/')") // No known test case
-			}
-			break;
-		case ''://end document
-			errorHandler.error('unexpected end of input');
-			if(s == S_TAG){
-				el.setTagName(source.slice(start,p));
-			}
-			return p;
-		case '>':
-			switch(s){
-			case S_TAG:
-				el.setTagName(source.slice(start,p));
-			case S_ATTR_END:
-			case S_TAG_SPACE:
-			case S_TAG_CLOSE:
-				break;//normal
-			case S_ATTR_NOQUOT_VALUE://Compatible state
-			case S_ATTR:
-				value = source.slice(start,p);
-				if(value.slice(-1) === '/'){
-					el.closed  = true;
-					value = value.slice(0,-1)
-				}
-			case S_ATTR_SPACE:
-				if(s === S_ATTR_SPACE){
-					value = attrName;
-				}
-				if(s == S_ATTR_NOQUOT_VALUE){
-					errorHandler.warning('attribute "'+value+'" missed quot(")!');
-					addAttribute(attrName, value, start)
-				}else{
-					if(!isHTML){
-						errorHandler.warning('attribute "'+value+'" missed value!! "'+value+'" instead!!')
-					}
-					addAttribute(value, value, start)
+			case '/':
+				switch (s) {
+					case S_TAG:
+						el.setTagName(source.slice(start, p));
+					case S_ATTR_END:
+					case S_TAG_SPACE:
+					case S_TAG_CLOSE:
+						s = S_TAG_CLOSE;
+						el.closed = true;
+					case S_ATTR_NOQUOT_VALUE:
+					case S_ATTR:
+						break;
+					case S_ATTR_SPACE:
+						el.closed = true;
+						break;
+					//case S_EQ:
+					default:
+						throw new Error("attribute invalid close char('/')"); // No known test case
 				}
 				break;
-			case S_EQ:
-				throw new Error('attribute value missed!!');
-			}
-//			console.log(tagName,tagNamePattern,tagNamePattern.test(tagName))
-			return p;
-		/*xml space '\x20' | #x9 | #xD | #xA; */
-		case '\u0080':
-			c = ' ';
-		default:
-			if(c<= ' '){//space
-				switch(s){
-				case S_TAG:
-					el.setTagName(source.slice(start,p));//tagName
-					s = S_TAG_SPACE;
-					break;
-				case S_ATTR:
-					attrName = source.slice(start,p)
-					s = S_ATTR_SPACE;
-					break;
-				case S_ATTR_NOQUOT_VALUE:
-					var value = source.slice(start, p);
-					errorHandler.warning('attribute "'+value+'" missed quot(")!!');
-					addAttribute(attrName, value, start)
-				case S_ATTR_END:
-					s = S_TAG_SPACE;
-					break;
-				//case S_TAG_SPACE:
-				//case S_EQ:
-				//case S_ATTR_SPACE:
-				//	void();break;
-				//case S_TAG_CLOSE:
-					//ignore warning
+			case '': //end document
+				errorHandler.error('unexpected end of input');
+				if (s == S_TAG) {
+					el.setTagName(source.slice(start, p));
 				}
-			}else{//not space
-//S_TAG,	S_ATTR,	S_EQ,	S_ATTR_NOQUOT_VALUE
-//S_ATTR_SPACE,	S_ATTR_END,	S_TAG_SPACE, S_TAG_CLOSE
-				switch(s){
-				//case S_TAG:void();break;
-				//case S_ATTR:void();break;
-				//case S_ATTR_NOQUOT_VALUE:void();break;
-				case S_ATTR_SPACE:
-					var tagName =  el.tagName;
-					if (!isHTML) {
-						errorHandler.warning('attribute "'+attrName+'" missed value!! "'+attrName+'" instead2!!')
+				return p;
+			case '>':
+				switch (s) {
+					case S_TAG:
+						el.setTagName(source.slice(start, p));
+					case S_ATTR_END:
+					case S_TAG_SPACE:
+					case S_TAG_CLOSE:
+						break; //normal
+					case S_ATTR_NOQUOT_VALUE: //Compatible state
+					case S_ATTR:
+						value = source.slice(start, p);
+						if (value.slice(-1) === '/') {
+							el.closed = true;
+							value = value.slice(0, -1);
+						}
+					case S_ATTR_SPACE:
+						if (s === S_ATTR_SPACE) {
+							value = attrName;
+						}
+						if (s == S_ATTR_NOQUOT_VALUE) {
+							errorHandler.warning('attribute "' + value + '" missed quot(")!');
+							addAttribute(attrName, value, start);
+						} else {
+							if (!isHTML) {
+								errorHandler.warning('attribute "' + value + '" missed value!! "' + value + '" instead!!');
+							}
+							addAttribute(value, value, start);
+						}
+						break;
+					case S_EQ:
+						throw new Error('attribute value missed!!');
+				}
+				return p;
+			/*xml space '\x20' | #x9 | #xD | #xA; */
+			case '\u0080':
+				c = ' ';
+			default:
+				if (c <= ' ') {
+					//space
+					switch (s) {
+						case S_TAG:
+							el.setTagName(source.slice(start, p)); //tagName
+							s = S_TAG_SPACE;
+							break;
+						case S_ATTR:
+							attrName = source.slice(start, p);
+							s = S_ATTR_SPACE;
+							break;
+						case S_ATTR_NOQUOT_VALUE:
+							var value = source.slice(start, p);
+							errorHandler.warning('attribute "' + value + '" missed quot(")!!');
+							addAttribute(attrName, value, start);
+						case S_ATTR_END:
+							s = S_TAG_SPACE;
+							break;
+						//case S_TAG_SPACE:
+						//case S_EQ:
+						//case S_ATTR_SPACE:
+						//	void();break;
+						//case S_TAG_CLOSE:
+						//ignore warning
+					}
+				} else {
+					//not space
+					//S_TAG,	S_ATTR,	S_EQ,	S_ATTR_NOQUOT_VALUE
+					//S_ATTR_SPACE,	S_ATTR_END,	S_TAG_SPACE, S_TAG_CLOSE
+					switch (s) {
+						//case S_TAG:void();break;
+						//case S_ATTR:void();break;
+						//case S_ATTR_NOQUOT_VALUE:void();break;
+						case S_ATTR_SPACE:
+							if (!isHTML) {
+								errorHandler.warning('attribute "' + attrName + '" missed value!! "' + attrName + '" instead2!!');
+							}
+							addAttribute(attrName, attrName, start);
+							start = p;
+							s = S_ATTR;
+							break;
+						case S_ATTR_END:
+							errorHandler.warning('attribute space is required"' + attrName + '"!!');
+						case S_TAG_SPACE:
+							s = S_ATTR;
+							start = p;
+							break;
+						case S_EQ:
+							s = S_ATTR_NOQUOT_VALUE;
+							start = p;
+							break;
+						case S_TAG_CLOSE:
+							throw new Error("elements closed character '/' and '>' must be connected to");
 					}
-					addAttribute(attrName, attrName, start);
-					start = p;
-					s = S_ATTR;
-					break;
-				case S_ATTR_END:
-					errorHandler.warning('attribute space is required"'+attrName+'"!!')
-				case S_TAG_SPACE:
-					s = S_ATTR;
-					start = p;
-					break;
-				case S_EQ:
-					s = S_ATTR_NOQUOT_VALUE;
-					start = p;
-					break;
-				case S_TAG_CLOSE:
-					throw new Error("elements closed character '/' and '>' must be connected to");
 				}
-			}
-		}//end outer switch
+		} //end outer switch
 		//console.log('p++',p)
 		p++;
 	}
 }
+
 /**
- * @return true if has new namespace define
+ * @returns `true` if a new namespace has been defined.
  */
-function appendElement(el,domBuilder,currentNSMap){
+function appendElement(el, domBuilder, currentNSMap) {
 	var tagName = el.tagName;
 	var localNSMap = null;
 	//var currentNSMap = parseStack[parseStack.length-1].currentNSMap;
 	var i = el.length;
-	while(i--){
+	while (i--) {
 		var a = el[i];
 		var qName = a.qName;
 		var value = a.value;
 		var nsp = qName.indexOf(':');
-		if(nsp>0){
-			var prefix = a.prefix = qName.slice(0,nsp);
-			var localName = qName.slice(nsp+1);
-			var nsPrefix = prefix === 'xmlns' && localName
-		}else{
+		if (nsp > 0) {
+			var prefix = (a.prefix = qName.slice(0, nsp));
+			var localName = qName.slice(nsp + 1);
+			var nsPrefix = prefix === 'xmlns' && localName;
+		} else {
 			localName = qName;
-			prefix = null
-			nsPrefix = qName === 'xmlns' && ''
+			prefix = null;
+			nsPrefix = qName === 'xmlns' && '';
 		}
 		//can not set prefix,because prefix !== ''
-		a.localName = localName ;
+		a.localName = localName;
 		//prefix == null for no ns prefix attribute
-		if(nsPrefix !== false){//hack!!
-			if(localNSMap == null){
-				localNSMap = {}
+		if (nsPrefix !== false) {
+			//hack!!
+			if (localNSMap == null) {
+				localNSMap = {};
 				//console.log(currentNSMap,0)
-				_copy(currentNSMap,currentNSMap={})
+				_copy(currentNSMap, (currentNSMap = {}));
 				//console.log(currentNSMap,1)
 			}
 			currentNSMap[nsPrefix] = localNSMap[nsPrefix] = value;
-			a.uri = NAMESPACE.XMLNS
-			domBuilder.startPrefixMapping(nsPrefix, value)
+			a.uri = NAMESPACE.XMLNS;
+			domBuilder.startPrefixMapping(nsPrefix, value);
 		}
 	}
 	var i = el.length;
-	while(i--){
+	while (i--) {
 		a = el[i];
-		var prefix = a.prefix;
-		if(prefix){//no prefix attribute has no namespace
-			if(prefix === 'xml'){
+		if (a.prefix) {
+			//no prefix attribute has no namespace
+			if (a.prefix === 'xml') {
 				a.uri = NAMESPACE.XML;
-			}if(prefix !== 'xmlns'){
-				a.uri = currentNSMap[prefix || '']
+			}
+			if (a.prefix !== 'xmlns') {
+				a.uri = currentNSMap[a.prefix];
 			}
 		}
 	}
 	var nsp = tagName.indexOf(':');
-	if(nsp>0){
-		prefix = el.prefix = tagName.slice(0,nsp);
-		localName = el.localName = tagName.slice(nsp+1);
-	}else{
-		prefix = null;//important!!
+	if (nsp > 0) {
+		prefix = el.prefix = tagName.slice(0, nsp);
+		localName = el.localName = tagName.slice(nsp + 1);
+	} else {
+		prefix = null; //important!!
 		localName = el.localName = tagName;
 	}
 	//no prefix element has default namespace
-	var ns = el.uri = currentNSMap[prefix || ''];
-	domBuilder.startElement(ns,localName,tagName,el);
+	var ns = (el.uri = currentNSMap[prefix || '']);
+	domBuilder.startElement(ns, localName, tagName, el);
 	//endPrefixMapping and startPrefixMapping have not any help for dom builder
 	//localNSMap = null
-	if(el.closed){
-		domBuilder.endElement(ns,localName,tagName);
-		if(localNSMap){
-			for(prefix in localNSMap){
-				domBuilder.endPrefixMapping(prefix)
+	if (el.closed) {
+		domBuilder.endElement(ns, localName, tagName);
+		if (localNSMap) {
+			for (prefix in localNSMap) {
+				if (Object.prototype.hasOwnProperty.call(localNSMap, prefix)) {
+					domBuilder.endPrefixMapping(prefix);
+				}
 			}
 		}
-	}else{
+	} else {
 		el.currentNSMap = currentNSMap;
 		el.localNSMap = localNSMap;
 		//parseStack.push(el);
 		return true;
 	}
 }
-function parseHtmlSpecialContent(source,elStartEnd,tagName,entityReplacer,domBuilder){
+
+function parseHtmlSpecialContent(source, elStartEnd, tagName, entityReplacer, domBuilder) {
 	// https://html.spec.whatwg.org/#raw-text-elements
 	// https://html.spec.whatwg.org/#escapable-raw-text-elements
 	// https://html.spec.whatwg.org/#cdata-rcdata-restrictions:raw-text-elements
 	// TODO: https://html.spec.whatwg.org/#cdata-rcdata-restrictions
 	var isEscapableRaw = isHTMLEscapableRawTextElement(tagName);
-	if(isEscapableRaw || isHTMLRawTextElement(tagName)){
-		var elEndStart =  source.indexOf('</'+tagName+'>',elStartEnd);
-		var text = source.substring(elStartEnd+1,elEndStart);
+	if (isEscapableRaw || isHTMLRawTextElement(tagName)) {
+		var elEndStart = source.indexOf('</' + tagName + '>', elStartEnd);
+		var text = source.substring(elStartEnd + 1, elEndStart);
 
-		if(isEscapableRaw){
-				text = text.replace(/&#?\w+;/g,entityReplacer);
+		if (isEscapableRaw) {
+			text = text.replace(/&#?\w+;/g, entityReplacer);
 		}
-				domBuilder.characters(text,0,text.length);
-				return elEndStart;
+		domBuilder.characters(text, 0, text.length);
+		return elEndStart;
 	}
-	return elStartEnd+1;
+	return elStartEnd + 1;
 }
-function fixSelfClosed(source,elStartEnd,tagName,closeMap){
+
+function fixSelfClosed(source, elStartEnd, tagName, closeMap) {
 	//if(tagName in closeMap){
 	var pos = closeMap[tagName];
-	if(pos == null){
+	if (pos == null) {
 		//console.log(tagName)
-		pos =  source.lastIndexOf('</'+tagName+'>')
-		if(pos<elStartEnd){//忘记闭合
-			pos = source.lastIndexOf('</'+tagName)
+		pos = source.lastIndexOf('</' + tagName + '>');
+		if (pos < elStartEnd) {
+			//忘记闭合
+			pos = source.lastIndexOf('</' + tagName);
 		}
-		closeMap[tagName] =pos
+		closeMap[tagName] = pos;
 	}
-	return pos<elStartEnd;
+	return pos < elStartEnd;
 	//}
 }
-function _copy(source,target){
-	for(var n in source){target[n] = source[n]}
+
+function _copy(source, target) {
+	for (var n in source) {
+		if (Object.prototype.hasOwnProperty.call(source, n)) {
+			target[n] = source[n];
+		}
+	}
 }
-function parseDCC(source,start,domBuilder,errorHandler){//sure start with '<!'
-	var next= source.charAt(start+2)
-	switch(next){
-	case '-':
-		if(source.charAt(start + 3) === '-'){
-			var end = source.indexOf('-->',start+4);
-			//append comment source.substring(4,end)//<!--
-			if(end>start){
-				domBuilder.comment(source,start+4,end-start-4);
-				return end+3;
-			}else{
-				errorHandler.error("Unclosed comment");
-				return -1;
+
+/**
+ * @typedef ParseUtils
+ * @property {function(relativeIndex: number?): string | undefined} char
+ * Provides look ahead access to a singe character relative to the current index.
+ * @property {function(): number} getIndex
+ * Provides read-only access to the current index.
+ * @property {function(reg: RegExp): string | null} getMatch
+ * Applies the provided regular expression enforcing that it starts at the current index and
+ * returns the complete matching string,
+ * and moves the current index by the length of the matching string.
+ * @property {function(): string} getSource
+ * Provides read-only access to the complete source.
+ * @property {function(places: number?): void} skip
+ * moves the current index by places (defaults to 1)
+ * @property {function(): number} skipBlanks
+ * Moves the current index by the amount of white space that directly follows the current index
+ * and returns the amount of whitespace chars skipped (0..n),
+ * or -1 if the end of the source was reached.
+ * @property {function(): string} substringFromIndex
+ * creates a substring from the current index to the end of `source`
+ * @property {function(compareWith: string): boolean} substringStartsWith
+ * Checks if source contains `compareWith`,
+ * starting from the current index.
+ * @see {@link parseUtils}
+ */
+
+/**
+ * A temporary scope for parsing and look ahead operations in `source`,
+ * starting from index `start`.
+ *
+ * Some operations move the current index by a number of positions,
+ * after which `getIndex` returns the new index.
+ *
+ * @param {string} source
+ * @param {number} start
+ * @returns {ParseUtils}
+ */
+function parseUtils(source, start) {
+	var index = start;
+
+	function char(n) {
+		n = n || 0;
+		return source.charAt(index + n);
+	}
+
+	function skip(n) {
+		n = n || 1;
+		index += n;
+	}
+
+	function skipBlanks() {
+		var blanks = 0;
+		while (index < source.length) {
+			var c = char();
+			if (c !== ' ' && c !== '\n' && c !== '\t' && c !== '\r') {
+				return blanks;
 			}
-		}else{
-			//error
-			return -1;
+			blanks++;
+			skip();
+		}
+		return -1;
+	}
+	function substringFromIndex() {
+		return source.substring(index);
+	}
+	function substringStartsWith(text) {
+		return source.substring(index, index + text.length) === text;
+	}
+
+	function getMatch(args) {
+		var expr = g.reg('^', args);
+		var match = expr.exec(substringFromIndex());
+		if (match) {
+			skip(match[0].length);
+			return match[0];
+		}
+		return null;
+	}
+	return {
+		char: char,
+		getIndex: function () {
+			return index;
+		},
+		getMatch: getMatch,
+		getSource: function () {
+			return source;
+		},
+		skip: skip,
+		skipBlanks: skipBlanks,
+		substringFromIndex: substringFromIndex,
+		substringStartsWith: substringStartsWith,
+	};
+}
+
+/**
+ * @param {ParseUtils} p
+ * @param {DOMHandler} errorHandler
+ * @returns {string}
+ */
+function parseDoctypeInternalSubset(p, errorHandler) {
+	/**
+	 * @param {ParseUtils} p
+	 * @param {DOMHandler} errorHandler
+	 * @returns {string}
+	 */
+	function parsePI(p, errorHandler) {
+		var match = g.PI.exec(p.substringFromIndex());
+		if (!match) {
+			return errorHandler.fatalError('processing instruction is not well-formed at position ' + p.getIndex());
 		}
-	default:
-		if(source.substr(start+3,6) == 'CDATA['){
-			var end = source.indexOf(']]>',start+9);
-			domBuilder.startCDATA();
-			domBuilder.characters(source,start+9,end-start-9);
-			domBuilder.endCDATA()
-			return end+3;
+		if (match[1].toLowerCase() === 'xml') {
+			return errorHandler.fatalError(
+				'xml declaration is only allowed at the start of the document, but found at position ' + p.getIndex()
+			);
 		}
-		//<!DOCTYPE
-		//startDTD(java.lang.String name, java.lang.String publicId, java.lang.String systemId)
-		var matchs = split(source,start);
-		var len = matchs.length;
-		if(len>1 && /!doctype/i.test(matchs[0][0])){
-			var name = matchs[1][0];
-			var pubid = false;
-			var sysid = false;
-			if(len>3){
-				if(/^public$/i.test(matchs[2][0])){
-					pubid = matchs[3][0];
-					sysid = len>4 && matchs[4][0];
-				}else if(/^system$/i.test(matchs[2][0])){
-					sysid = matchs[3][0];
+		p.skip(match[0].length);
+		return match[0];
+	}
+	// Parse internal subset
+	var source = p.getSource();
+	if (p.char() === '[') {
+		p.skip(1);
+		var intSubsetStart = p.getIndex();
+		p.skipBlanks();
+		while (p.getIndex() < source.length) {
+			var current = null;
+			// Only in external subset
+			// if (char() === '<' && char(1) === '!' && char(2) === '[') {
+			// 	parseConditionalSections(p, errorHandler);
+			// } else
+			if (p.char() === '<' && p.char(1) === '!') {
+				switch (p.char(2)) {
+					case 'E':
+						if (p.char(3) === 'L') {
+							current = p.getMatch(g.elementdecl);
+						} else if (p.char(3) === 'N') {
+							current = p.getMatch(g.EntityDecl);
+						}
+						break;
+					case 'A':
+						current = p.getMatch(g.AttlistDecl);
+						break;
+					case 'N':
+						current = p.getMatch(g.NotationDecl);
+						break;
+					case '-':
+						current = p.getMatch(g.Comment);
+						break;
 				}
+			} else if (p.char() === '<' && p.char(1) === '?') {
+				current = parsePI(p, errorHandler);
+			} else if (p.char() === '%') {
+				current = p.getMatch(g.PEReference);
+			} else {
+				return errorHandler.fatalError('Error detected in Markup declaration');
 			}
-			var lastMatch = matchs[len-1]
-			domBuilder.startDTD(name, pubid, sysid);
-			domBuilder.endDTD();
-
-			return lastMatch.index+lastMatch[0].length
+			if (!current) {
+				return errorHandler.fatalError('Error in internal subset at position ' + p.getIndex());
+			}
+			p.skipBlanks();
+			if (p.char() === ']') {
+				var internalSubset = source.substring(intSubsetStart, p.getIndex());
+				p.skip(1);
+				return internalSubset;
+			}
+			p.skipBlanks();
 		}
+		return errorHandler.fatalError('doctype internal subset is not well-formed, missing ]');
 	}
-	return -1;
 }
 
+/**
+ * Called when the parser encounters an element starting with '<!'.
+ *
+ * @param {string} source
+ * The xml.
+ * @param {number} start
+ * the start index of the '<!'
+ * @param {DOMHandler} domBuilder
+ * @param {DOMHandler} errorHandler
+ * @returns {number | never} The end index of the element.
+ * @throws {ParseError}
+ * In case the element is not well-formed.
+ */
+function parseDoctypeCommentOrCData(source, start, domBuilder, errorHandler) {
+	var p = parseUtils(source, start);
+
+	switch (p.char(2)) {
+		case '-':
+			// should be a comment
+			var comment = p.getMatch(g.Comment);
+			if (comment) {
+				domBuilder.comment(comment, g.COMMENT_START.length, comment.length - g.COMMENT_START.length - g.COMMENT_END.length);
+				return p.getIndex();
+			} else {
+				return errorHandler.fatalError('comment is not well-formed at position ' + p.getIndex());
+			}
+		case '[':
+			// should be CDATA
+			var cdata = p.getMatch(g.CDSect);
+			if (cdata) {
+				domBuilder.startCDATA();
+				domBuilder.characters(cdata, g.CDATA_START.length, cdata.length - g.CDATA_START.length - g.CDATA_END.length);
+				domBuilder.endCDATA();
+				return p.getIndex();
+			} else {
+				return errorHandler.fatalError('Invalid CDATA starting at position ' + start);
+			}
+		case 'D': {
+			// should be DOCTYPE
+			var doctype = {
+				name: undefined,
+				publicId: undefined,
+				systemId: undefined,
+				internalSubset: undefined,
+			};
+
+			if (!p.substringStartsWith(g.DOCTYPE_DECL_START)) {
+				return errorHandler.fatalError('Expected ' + g.DOCTYPE_DECL_START + ' at position ' + p.getIndex());
+			}
+			p.skip(g.DOCTYPE_DECL_START.length);
+			if (p.skipBlanks() < 1) {
+				return errorHandler.fatalError('Expected whitespace after ' + g.DOCTYPE_DECL_START + ' at position ' + p.getIndex());
+			}
+
+			// Parse the DOCTYPE name
+			doctype.name = p.getMatch(g.Name);
+			if (!doctype.name)
+				return errorHandler.fatalError('doctype name missing or contains unexpected characters at position ' + p.getIndex());
+			p.skipBlanks();
+
+			// Check for ExternalID
+			if (p.substringStartsWith(g.PUBLIC) || p.substringStartsWith(g.SYSTEM)) {
+				var match = g.ExternalID_match.exec(p.substringFromIndex());
+				if (!match) {
+					return errorHandler.fatalError('doctype external id is not well-formed at position ' + p.getIndex());
+				}
+				if (match.groups.SystemLiteralOnly !== undefined) {
+					doctype.systemId = match.groups.SystemLiteralOnly;
+				} else {
+					doctype.systemId = match.groups.SystemLiteral;
+					doctype.publicId = match.groups.PubidLiteral;
+				}
+				p.skip(match[0].length);
+			}
 
+			p.skipBlanks();
+			doctype.internalSubset = parseDoctypeInternalSubset(p, errorHandler);
+			p.skipBlanks();
+			if (p.char() !== '>') {
+				return errorHandler.fatalError('doctype not terminated with > at position ' + p.getIndex());
+			}
+			p.skip(1);
+			domBuilder.startDTD(doctype.name, doctype.publicId, doctype.systemId, doctype.internalSubset);
+			domBuilder.endDTD();
+			return p.getIndex();
+		}
+		default:
+			return errorHandler.fatalError('Not well-formed XML starting with "<!" at position ' + start);
+	}
+}
 
-function parseInstruction(source,start,domBuilder){
-	var end = source.indexOf('?>',start);
-	if(end){
-		var match = source.substring(start,end).match(/^<\?(\S*)\s*([\s\S]*?)\s*$/);
-		if(match){
-			var len = match[0].length;
-			domBuilder.processingInstruction(match[1], match[2]) ;
-			return end+2;
-		}else{//error
-			return -1;
+function parseProcessingInstruction(source, start, domBuilder, errorHandler) {
+	var match = source.substring(start).match(g.PI);
+	if (!match) {
+		return errorHandler.fatalError('Invalid processing instruction starting at position ' + start);
+	}
+	if (match[1].toLowerCase() === 'xml') {
+		if (start > 0) {
+			return errorHandler.fatalError(
+				'processing instruction at position ' + start + ' is an xml declaration which is only at the start of the document'
+			);
+		}
+		if (!g.XMLDecl.test(source.substring(start))) {
+			return errorHandler.fatalError('xml declaration is not well-formed');
 		}
 	}
-	return -1;
+	domBuilder.processingInstruction(match[1], match[2]);
+	return start + match[0].length;
 }
 
-function ElementAttributes(){
-	this.attributeNames = {}
+function ElementAttributes() {
+	this.attributeNames = {};
 }
+
 ElementAttributes.prototype = {
-	setTagName:function(tagName){
-		if(!tagNamePattern.test(tagName)){
-			throw new Error('invalid tagName:'+tagName)
+	setTagName: function (tagName) {
+		if (!g.QName_exact.test(tagName)) {
+			throw new Error('invalid tagName:' + tagName);
 		}
-		this.tagName = tagName
+		this.tagName = tagName;
 	},
-	addValue:function(qName, value, offset) {
-		if(!tagNamePattern.test(qName)){
-			throw new Error('invalid attribute:'+qName)
+	addValue: function (qName, value, offset) {
+		if (!g.QName_exact.test(qName)) {
+			throw new Error('invalid attribute:' + qName);
 		}
 		this.attributeNames[qName] = this.length;
-		this[this.length++] = {qName:qName,value:value,offset:offset}
+		this[this.length++] = { qName: qName, value: value, offset: offset };
 	},
-	length:0,
-	getLocalName:function(i){return this[i].localName},
-	getLocator:function(i){return this[i].locator},
-	getQName:function(i){return this[i].qName},
-	getURI:function(i){return this[i].uri},
-	getValue:function(i){return this[i].value}
-//	,getIndex:function(uri, localName)){
-//		if(localName){
-//
-//		}else{
-//			var qName = uri
-//		}
-//	},
-//	getValue:function(){return this.getValue(this.getIndex.apply(this,arguments))},
-//	getType:function(uri,localName){}
-//	getType:function(i){},
-}
-
-
-
-function split(source,start){
-	var match;
-	var buf = [];
-	var reg = /'[^']+'|"[^"]+"|[^\s<>\/=]+=?|(\/?\s*>|<)/g;
-	reg.lastIndex = start;
-	reg.exec(source);//skip <
-	while(match = reg.exec(source)){
-		buf.push(match);
-		if(match[1])return buf;
-	}
-}
+	length: 0,
+	getLocalName: function (i) {
+		return this[i].localName;
+	},
+	getLocator: function (i) {
+		return this[i].locator;
+	},
+	getQName: function (i) {
+		return this[i].qName;
+	},
+	getURI: function (i) {
+		return this[i].uri;
+	},
+	getValue: function (i) {
+		return this[i].value;
+	},
+	//	,getIndex:function(uri, localName)){
+	//		if(localName){
+	//
+	//		}else{
+	//			var qName = uri
+	//		}
+	//	},
+	//	getValue:function(){return this.getValue(this.getIndex.apply(this,arguments))},
+	//	getType:function(uri,localName){}
+	//	getType:function(i){},
+};
 
 exports.XMLReader = XMLReader;
-exports.ParseError = ParseError;
+exports.parseUtils = parseUtils;
+exports.parseDoctypeCommentOrCData = parseDoctypeCommentOrCData;