CB-12546: emulator specs.

2026-04-23 00:00:09 +08:00 · 2017-03-14 13:12:57 -07:00
parent eb6ada8091
commit 6395eda0c8
24 changed files with 655 additions and 446 deletions
@@ -2,21 +2,21 @@
 //[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 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 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_V
-//S_ATTR_S,	S_E,	S_S,	S_C
+//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_S=2;//attr name end and space offer
+var S_ATTR_SPACE=2;//attr name end and space offer
 var S_EQ = 3;//=space?
-var S_V = 4;//attr value(no quot value only)
-var S_E = 5;//attr value end and no space(quot end)
-var S_S = 6;//(attr value end || tag end ) && (space offer)
-var S_C = 7;//closed el<el />
+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(){
 	
@@ -33,7 +33,7 @@ XMLReader.prototype = {
 	}
 }
 function parse(source,defaultNSMapCopy,entityMap,domBuilder,errorHandler){
-  function fixedFromCharCode(code) {
+	function fixedFromCharCode(code) {
 		// String.prototype.fromCharCode does not supports
 		// > 2 bytes unicode chars directly
 		if (code > 0xffff) {
@@ -76,7 +76,7 @@ function parse(source,defaultNSMapCopy,entityMap,domBuilder,errorHandler){
 	}
 	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}]
@@ -87,7 +87,7 @@ function parse(source,defaultNSMapCopy,entityMap,domBuilder,errorHandler){
 			var tagStart = source.indexOf('<',start);
 			if(tagStart<0){
 				if(!source.substr(start).match(/^\s*$/)){
-					var doc = domBuilder.document;
+					var doc = domBuilder.doc;
 	    			var text = doc.createTextNode(source.substr(start));
 	    			doc.appendChild(text);
 	    			domBuilder.currentElement = text;
@@ -102,16 +102,36 @@ function parse(source,defaultNSMapCopy,entityMap,domBuilder,errorHandler){
 				var end = source.indexOf('>',tagStart+3);
 				var tagName = source.substring(tagStart+2,end);
 				var config = parseStack.pop();
-				var localNSMap = config.localNSMap;
-		        if(config.tagName != tagName){
-		            errorHandler.fatalError("end tag name: "+tagName+' is not match the current start tagName:'+config.tagName );
-		        }
-				domBuilder.endElement(config.uri,config.localName,tagName);
-				if(localNSMap){
-					for(var prefix in localNSMap){
-						domBuilder.endPrefixMapping(prefix) ;
-					}
+				if(end<0){
+					
+	        		tagName = source.substring(tagStart+2).replace(/[\s<].*/,'');
+	        		//console.error('#@@@@@@'+tagName)
+	        		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;
 				}
+				//console.error(parseStack.length,parseStack)
+				//console.error(config);
+				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) ;
+						}
+					}
+					if(!endMatch){
+		            	errorHandler.fatalError("end tag name: "+tagName+' is not match the current start tagName:'+config.tagName );
+					}
+		        }else{
+		        	parseStack.push(config)
+		        }
+				
 				end++;
 				break;
 				// end elment
@@ -124,33 +144,40 @@ function parse(source,defaultNSMapCopy,entityMap,domBuilder,errorHandler){
 				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,entityReplacer,errorHandler);
+				var end = parseElementStartPart(source,tagStart,el,currentNSMap,entityReplacer,errorHandler);
 				var len = el.length;
 				
-				if(locator){
-					if(len){
-						//attribute position fixed
-						for(var i = 0;i<len;i++){
-							var a = el[i];
-							position(a.offset);
-							a.offset = copyLocator(locator,{});
-						}
-					}
-					position(end);
-				}
+				
 				if(!el.closed && fixSelfClosed(source,end,el.tagName,closeMap)){
 					el.closed = true;
 					if(!entityMap.nbsp){
 						errorHandler.warning('unclosed xml attribute');
 					}
 				}
-				appendElement(el,domBuilder,parseStack);
+				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,{});
+					}
+					//}catch(e){console.error('@@@@@'+e)}
+					domBuilder.locator = locator2
+					if(appendElement(el,domBuilder,currentNSMap)){
+						parseStack.push(el)
+					}
+					domBuilder.locator = locator;
+				}else{
+					if(appendElement(el,domBuilder,currentNSMap)){
+						parseStack.push(el)
+					}
+				}
+				
 				
 				
 				if(el.uri === 'http://www.w3.org/1999/xhtml' && !el.closed){
@@ -160,8 +187,10 @@ function parse(source,defaultNSMapCopy,entityMap,domBuilder,errorHandler){
 				}
 			}
 		}catch(e){
-			errorHandler.error('element parse error: '+e);
+			errorHandler.error('element parse error: '+e)
+			//errorHandler.error('element parse error: '+e);
 			end = -1;
+			//throw e;
 		}
 		if(end>start){
 			start = end;
@@ -181,7 +210,7 @@ function copyLocator(f,t){
 * @see #appendElement(source,elStartEnd,el,selfClosed,entityReplacer,domBuilder,parseStack);
 * @return end of the elementStartPart(end of elementEndPart for selfClosed el)
 */
-function parseElementStartPart(source,start,el,entityReplacer,errorHandler){
+function parseElementStartPart(source,start,el,currentNSMap,entityReplacer,errorHandler){
 	var attrName;
 	var value;
 	var p = ++start;
@@ -193,7 +222,7 @@ function parseElementStartPart(source,start,el,entityReplacer,errorHandler){
 			if(s === S_ATTR){//attrName
 				attrName = source.slice(start,p);
 				s = S_EQ;
-			}else if(s === S_ATTR_S){
+			}else if(s === S_ATTR_SPACE){
 				s = S_EQ;
 			}else{
 				//fatalError: equal must after attrName or space after attrName
@@ -202,25 +231,30 @@ function parseElementStartPart(source,start,el,entityReplacer,errorHandler){
 			break;
 		case '\'':
 		case '"':
-			if(s === S_EQ){//equal
+			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).replace(/&#?\w+;/g,entityReplacer);
 					el.add(attrName,value,start-1);
-					s = S_E;
+					s = S_ATTR_END;
 				}else{
 					//fatalError: no end quot match
 					throw new Error('attribute value no end \''+c+'\' match');
 				}
-			}else if(s == S_V){
+			}else if(s == S_ATTR_NOQUOT_VALUE){
 				value = source.slice(start,p).replace(/&#?\w+;/g,entityReplacer);
 				//console.log(attrName,value,start,p)
 				el.add(attrName,value,start);
 				//console.dir(el)
 				errorHandler.warning('attribute "'+attrName+'" missed start quot('+c+')!!');
 				start = p+1;
-				s = S_E
+				s = S_ATTR_END
 			}else{
 				//fatalError: no equal before
 				throw new Error('attribute value must after "="');
@@ -230,14 +264,14 @@ function parseElementStartPart(source,start,el,entityReplacer,errorHandler){
 			switch(s){
 			case S_TAG:
 				el.setTagName(source.slice(start,p));
-			case S_E:
-			case S_S:
-			case S_C:
-				s = S_C;
+			case S_ATTR_END:
+			case S_TAG_SPACE:
+			case S_TAG_CLOSE:
+				s =S_TAG_CLOSE;
 				el.closed = true;
-			case S_V:
+			case S_ATTR_NOQUOT_VALUE:
 			case S_ATTR:
-			case S_ATTR_S:
+			case S_ATTR_SPACE:
 				break;
 			//case S_EQ:
 			default:
@@ -247,30 +281,36 @@ function parseElementStartPart(source,start,el,entityReplacer,errorHandler){
 		case ''://end document
 			//throw new Error('unexpected end of input')
 			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_E:
-			case S_S:
-			case S_C:
+			case S_ATTR_END:
+			case S_TAG_SPACE:
+			case S_TAG_CLOSE:
 				break;//normal
-			case S_V://Compatible state
+			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_S:
-				if(s === S_ATTR_S){
+			case S_ATTR_SPACE:
+				if(s === S_ATTR_SPACE){
 					value = attrName;
 				}
-				if(s == S_V){
+				if(s == S_ATTR_NOQUOT_VALUE){
 					errorHandler.warning('attribute "'+value+'" missed quot(")!!');
 					el.add(attrName,value.replace(/&#?\w+;/g,entityReplacer),start)
 				}else{
-					errorHandler.warning('attribute "'+value+'" missed value!! "'+value+'" instead!!')
+					if(currentNSMap[''] !== 'http://www.w3.org/1999/xhtml' || !value.match(/^(?:disabled|checked|selected)$/i)){
+						errorHandler.warning('attribute "'+value+'" missed value!! "'+value+'" instead!!')
+					}
 					el.add(value,value,start)
 				}
 				break;
@@ -287,64 +327,68 @@ function parseElementStartPart(source,start,el,entityReplacer,errorHandler){
 				switch(s){
 				case S_TAG:
 					el.setTagName(source.slice(start,p));//tagName
-					s = S_S;
+					s = S_TAG_SPACE;
 					break;
 				case S_ATTR:
 					attrName = source.slice(start,p)
-					s = S_ATTR_S;
+					s = S_ATTR_SPACE;
 					break;
-				case S_V:
+				case S_ATTR_NOQUOT_VALUE:
 					var value = source.slice(start,p).replace(/&#?\w+;/g,entityReplacer);
 					errorHandler.warning('attribute "'+value+'" missed quot(")!!');
 					el.add(attrName,value,start)
-				case S_E:
-					s = S_S;
+				case S_ATTR_END:
+					s = S_TAG_SPACE;
 					break;
-				//case S_S:
+				//case S_TAG_SPACE:
 				//case S_EQ:
-				//case S_ATTR_S:
+				//case S_ATTR_SPACE:
 				//	void();break;
-				//case S_C:
+				//case S_TAG_CLOSE:
 					//ignore warning
 				}
 			}else{//not space
-//S_TAG,	S_ATTR,	S_EQ,	S_V
-//S_ATTR_S,	S_E,	S_S,	S_C
+//S_TAG,	S_ATTR,	S_EQ,	S_ATTR_NOQUOT_VALUE
+//S_ATTR_SPACE,	S_ATTR_END,	S_TAG_SPACE, S_TAG_CLOSE
 				switch(s){
 				//case S_TAG:void();break;
 				//case S_ATTR:void();break;
-				//case S_V:void();break;
-				case S_ATTR_S:
-					errorHandler.warning('attribute "'+attrName+'" missed value!! "'+attrName+'" instead!!')
+				//case S_ATTR_NOQUOT_VALUE:void();break;
+				case S_ATTR_SPACE:
+					var tagName =  el.tagName;
+					if(currentNSMap[''] !== 'http://www.w3.org/1999/xhtml' || !attrName.match(/^(?:disabled|checked|selected)$/i)){
+						errorHandler.warning('attribute "'+attrName+'" missed value!! "'+attrName+'" instead2!!')
+					}
 					el.add(attrName,attrName,start);
 					start = p;
 					s = S_ATTR;
 					break;
-				case S_E:
+				case S_ATTR_END:
 					errorHandler.warning('attribute space is required"'+attrName+'"!!')
-				case S_S:
+				case S_TAG_SPACE:
 					s = S_ATTR;
 					start = p;
 					break;
 				case S_EQ:
-					s = S_V;
+					s = S_ATTR_NOQUOT_VALUE;
 					start = p;
 					break;
-				case S_C:
+				case S_TAG_CLOSE:
 					throw new Error("elements closed character '/' and '>' must be connected to");
 				}
 			}
-		}
+		}//end outer switch
+		//console.log('p++',p)
 		p++;
 	}
 }
 /**
- * @return end of the elementStartPart(end of elementEndPart for selfClosed el)
+ * @return true if has new namespace define
 */
-function appendElement(el,domBuilder,parseStack){
+function appendElement(el,domBuilder,currentNSMap){
 	var tagName = el.tagName;
 	var localNSMap = null;
-	var currentNSMap = parseStack[parseStack.length-1].currentNSMap;
+	//var currentNSMap = parseStack[parseStack.length-1].currentNSMap;
 	var i = el.length;
 	while(i--){
 		var a = el[i];
@@ -383,7 +427,7 @@ function appendElement(el,domBuilder,parseStack){
 			if(prefix === 'xml'){
 				a.uri = 'http://www.w3.org/XML/1998/namespace';
 			}if(prefix !== 'xmlns'){
-				a.uri = currentNSMap[prefix]
+				a.uri = currentNSMap[prefix || '']
 				
 				//{console.log('###'+a.qName,domBuilder.locator.systemId+'',currentNSMap,a.uri)}
 			}
@@ -412,7 +456,8 @@ function appendElement(el,domBuilder,parseStack){
 	}else{
 		el.currentNSMap = currentNSMap;
 		el.localNSMap = localNSMap;
-		parseStack.push(el);
+		//parseStack.push(el);
+		return true;
 	}
 }
 function parseHtmlSpecialContent(source,elStartEnd,tagName,entityReplacer,domBuilder){
@@ -442,7 +487,11 @@ function fixSelfClosed(source,elStartEnd,tagName,closeMap){
 	var pos = closeMap[tagName];
 	if(pos == null){
 		//console.log(tagName)
-		pos = closeMap[tagName] = source.lastIndexOf('</'+tagName+'>')
+		pos =  source.lastIndexOf('</'+tagName+'>')
+		if(pos<elStartEnd){//忘记闭合
+			pos = source.lastIndexOf('</'+tagName)
+		}
+		closeMap[tagName] =pos
 	}
 	return pos<elStartEnd;
 	//} 
@@ -533,7 +582,7 @@ ElementAttributes.prototype = {
 	},
 	length:0,
 	getLocalName:function(i){return this[i].localName},
-	getOffset:function(i){return this[i].offset},
+	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}
@@ -580,7 +629,5 @@ function split(source,start){
 	}
 }

-if(typeof require == 'function'){
-	exports.XMLReader = XMLReader;
-}
+exports.XMLReader = XMLReader;