fix(sax): Handle raw text elements in HTML

and drop warning for boolean attributes in HTML
xmldom · Mar 6, 2022 · ceff927 · ceff927
1 parent 5fd4e1b
commit ceff927
Show file tree

Hide file tree

Showing 8 changed files with 471 additions and 56 deletions.
diff --git a/lib/conventions.js b/lib/conventions.js
@@ -144,6 +144,54 @@ function isHTMLVoidElement(tagName) {
 	return HTML_VOID_ELEMENTS.hasOwnProperty(tagName.toLowerCase())
 }
 
+/**
+ * Tag names that are raw text elements according to HTML spec.
+ * The value denotes whether they are escapable or not.
+ *
+ * @see isHTMLEscapableRawTextElement
+ * @see isHTMLRawTextElement
+ * @see https://html.spec.whatwg.org/#raw-text-elements
+ * @see https://html.spec.whatwg.org/#escapable-raw-text-elements
+ */
+var HTML_RAW_TEXT_ELEMENTS = freeze({
+	script: false,
+	style: false,
+	textarea: true,
+	title: true,
+})
+
+/**
+ * Check if `tagName` is matching one of the HTML raw text element names.
+ * It includes escapable and not escapable raw text elements.
+ * This method doesn't check if such tags are allowed
+ * in the context of the current document/parsing.
+ *
+ * @param {string} tagName
+ * @return {boolean}
+ * @see HTML_RAW_TEXT_ELEMENTS
+ * @see https://html.spec.whatwg.org/#raw-text-elements
+ * @see https://html.spec.whatwg.org/#escapable-raw-text-elements
+ */
+function isHTMLRawTextElement(tagName) {
+	return HTML_RAW_TEXT_ELEMENTS.hasOwnProperty(tagName.toLowerCase())
+}
+/**
+ * Check if `tagName` is matching one of the HTML escapable raw text element names.
+ * This method doesn't check if such tags are allowed
+ * in the context of the current document/parsing.
+ *
+ * @param {string} tagName
+ * @return {boolean}
+ * @see isHTMLRawTextElement
+ * @see HTML_RAW_TEXT_ELEMENTS
+ * @see https://html.spec.whatwg.org/#raw-text-elements
+ * @see https://html.spec.whatwg.org/#escapable-raw-text-elements
+ */
+function isHTMLEscapableRawTextElement(tagName) {
+	const key = tagName.toLowerCase();
+	return HTML_RAW_TEXT_ELEMENTS.hasOwnProperty(key) && HTML_RAW_TEXT_ELEMENTS[key];
+}
+
 /**
  * All mime types that are allowed as input to `DOMParser.parseFromString`
  *
@@ -282,8 +330,11 @@ var NAMESPACE = freeze({
 exports.assign = assign
 exports.freeze = freeze
 exports.HTML_BOOLEAN_ATTRIBUTES = HTML_BOOLEAN_ATTRIBUTES
+exports.HTML_RAW_TEXT_ELEMENTS = HTML_RAW_TEXT_ELEMENTS
 exports.HTML_VOID_ELEMENTS = HTML_VOID_ELEMENTS
 exports.isHTMLBooleanAttribute = isHTMLBooleanAttribute
+exports.isHTMLRawTextElement = isHTMLRawTextElement
+exports.isHTMLEscapableRawTextElement = isHTMLEscapableRawTextElement
 exports.isHTMLVoidElement = isHTMLVoidElement
 exports.MIME_TYPE = MIME_TYPE
 exports.NAMESPACE = NAMESPACE
diff --git a/lib/sax.js b/lib/sax.js
@@ -1,6 +1,8 @@
 'use strict'
 
 var conventions = require("./conventions");
+var isHTMLRawTextElement = conventions.isHTMLRawTextElement;
+var isHTMLEscapableRawTextElement = conventions.isHTMLEscapableRawTextElement;
 var NAMESPACE = conventions.NAMESPACE;
 var MIME_TYPE = conventions.MIME_TYPE;
 
@@ -181,7 +183,7 @@ function parse(source,defaultNSMapCopy,entityMap,domBuilder,errorHandler){
 
 				if(!el.closed && fixSelfClosed(source,end,el.tagName,closeMap)){
 					el.closed = true;
-					if(!entityMap.nbsp){
+					if(!isHTML){
 						errorHandler.warning('unclosed xml attribute');
 					}
 				}
@@ -352,7 +354,7 @@ function parseElementStartPart(
 					errorHandler.warning('attribute "'+value+'" missed quot(")!');
 					addAttribute(attrName, value, start)
 				}else{
-					if(!isHTML || !value.match(/^(?:disabled|checked|selected)$/i)){
+					if(!isHTML){
 						errorHandler.warning('attribute "'+value+'" missed value!! "'+value+'" instead!!')
 					}
 					addAttribute(value, value, start)
@@ -400,7 +402,7 @@ function parseElementStartPart(
 				//case S_ATTR_NOQUOT_VALUE:void();break;
 				case S_ATTR_SPACE:
 					var tagName =  el.tagName;
-					if (!isHTML || !attrName.match(/^(?:disabled|checked|selected)$/i)) {
+					if (!isHTML) {
 						errorHandler.warning('attribute "'+attrName+'" missed value!! "'+attrName+'" instead2!!')
 					}
 					addAttribute(attrName, attrName, start);
@@ -503,24 +505,19 @@ function appendElement(el,domBuilder,currentNSMap){
 	}
 }
 function parseHtmlSpecialContent(source,elStartEnd,tagName,entityReplacer,domBuilder){
-	if(/^(?:script|textarea)$/i.test(tagName)){
+	// https://html.spec.whatwg.org/#raw-text-elements
+	// https://html.spec.whatwg.org/#escapable-raw-text-elements
+	// https://html.spec.whatwg.org/#cdata-rcdata-restrictions:raw-text-elements
+	// TODO: https://html.spec.whatwg.org/#cdata-rcdata-restrictions
+	if(isHTMLRawTextElement(tagName)){
 		var elEndStart =  source.indexOf('</'+tagName+'>',elStartEnd);
 		var text = source.substring(elStartEnd+1,elEndStart);
-		if(/[&<]/.test(text)){
-			if(/^script$/i.test(tagName)){
-				//if(!/\]\]>/.test(text)){
-					//lexHandler.startCDATA();
-					domBuilder.characters(text,0,text.length);
-					//lexHandler.endCDATA();
-					return elEndStart;
-				//}
-			}//}else{//text area
+
+		if(isHTMLEscapableRawTextElement(tagName)){
 				text = text.replace(/&#?\w+;/g,entityReplacer);
+		}
 				domBuilder.characters(text,0,text.length);
 				return elEndStart;
-			//}
-
-		}
 	}
 	return elStartEnd+1;
 }