Merge pull request #912 from capricorn86/task/908-domparserparsefroms…

…tring-doesnt-decode-html-entities #908@minor: Adds support for endoing HTML entities to the XML parser.…
capricorn86 · May 11, 2023 · a895911 · a895911
2 parents 2291dc3 + 5b3960a
commit a895911
Show file tree

Hide file tree

Showing 13 changed files with 115 additions and 99 deletions.
diff --git a/package-lock.json b/package-lock.json
diff --git a/packages/happy-dom/package.json b/packages/happy-dom/package.json
@@ -48,14 +48,13 @@
 	},
 	"dependencies": {
 		"css.escape": "^1.5.1",
-		"he": "^1.2.0",
+		"entities": "^4.5.0",
 		"webidl-conversions": "^7.0.0",
 		"whatwg-encoding": "^2.0.0",
 		"whatwg-mimetype": "^3.0.0",
 		"iconv-lite": "^0.6.3"
 	},
 	"devDependencies": {
-		"@types/he": "^1.1.2",
 		"@types/jest": "^29.4.0",
 		"@types/node": "^15.6.0",
 		"@types/node-fetch": "^2.6.1",

diff --git a/packages/happy-dom/src/nodes/character-data/CharacterDataUtility.ts b/packages/happy-dom/src/nodes/character-data/CharacterDataUtility.ts
@@ -1,26 +1,5 @@
 import ICharacterData from './ICharacterData';
 
-const HTML_ENTITIES = [
-	{ regex: /&quot;/g, value: '"' },
-	{ regex: /&#34;/g, value: '"' },
-	{ regex: /&#x22;/g, value: '"' },
-	{ regex: /&amp;/g, value: '&' },
-	{ regex: /&#38;/g, value: '&' },
-	{ regex: /&#x26;/g, value: '&' },
-	{ regex: /&apos;/g, value: "'" },
-	{ regex: /&#39;/g, value: "'" },
-	{ regex: /&#x27;/g, value: "'" },
-	{ regex: /&lt;/g, value: '<' },
-	{ regex: /&#60;/g, value: '<' },
-	{ regex: /&#x3C;/g, value: '<' },
-	{ regex: /&gt;/g, value: '>' },
-	{ regex: /&#62;/g, value: '>' },
-	{ regex: /&#x3E;/g, value: '>' },
-	{ regex: /&nbsp;/g, value: ' ' },
-	{ regex: /&#160;/g, value: ' ' },
-	{ regex: /&#xA0;/g, value: ' ' }
-];
-
 /**
  * Child node utility.
  */
@@ -90,20 +69,4 @@ export default class CharacterDataUtility {
 	): string {
 		return characterData.data.substring(offset, offset + count);
 	}
-
-	/**
-	 * Decodes unicode characters to text.
-	 *
-	 * @param html String.
-	 * @returns Decoded HTML string.
-	 */
-	public static decodeHTMLEntities(html: string): string {
-		if (!html) {
-			return '';
-		}
-		for (const entity of HTML_ENTITIES) {
-			html = html.replace(entity.regex, entity.value);
-		}
-		return html;
-	}
 }
diff --git a/packages/happy-dom/src/nodes/element/Element.ts b/packages/happy-dom/src/nodes/element/Element.ts
@@ -30,7 +30,6 @@ import INamedNodeMap from '../../named-node-map/INamedNodeMap';
 import Event from '../../event/Event';
 import ElementUtility from './ElementUtility';
 import HTMLCollection from './HTMLCollection';
-import CharacterDataUtility from '../character-data/CharacterDataUtility';
 import EventPhaseEnum from '../../event/EventPhaseEnum';
 
 /**
@@ -208,7 +207,7 @@ export default class Element extends Node implements IElement {
 				result += childNode.textContent;
 			}
 		}
-		return CharacterDataUtility.decodeHTMLEntities(result);
+		return result;
 	}
 
 	/**
@@ -253,7 +252,7 @@ export default class Element extends Node implements IElement {
 	 * @returns HTML.
 	 */
 	public get outerHTML(): string {
-		return new XMLSerializer().serializeToString(this);
+		return new XMLSerializer({ escapeEntities: false }).serializeToString(this);
 	}
 
 	/**
@@ -340,10 +339,13 @@ export default class Element extends Node implements IElement {
 	 * @returns HTML.
 	 */
 	public getInnerHTML(options?: { includeShadowRoots?: boolean }): string {
-		const xmlSerializer = new XMLSerializer();
+		const xmlSerializer = new XMLSerializer({
+			includeShadowRoots: options && options.includeShadowRoots,
+			escapeEntities: false
+		});
 		let xml = '';
 		for (const node of this.childNodes) {
-			xml += xmlSerializer.serializeToString(node, options);
+			xml += xmlSerializer.serializeToString(node);
 		}
 		return xml;
 	}

diff --git a/packages/happy-dom/src/nodes/html-template-element/HTMLTemplateElement.ts b/packages/happy-dom/src/nodes/html-template-element/HTMLTemplateElement.ts
@@ -50,10 +50,13 @@ export default class HTMLTemplateElement extends HTMLElement implements IHTMLTem
 	 * @override
 	 */
 	public getInnerHTML(options?: { includeShadowRoots?: boolean }): string {
-		const xmlSerializer = new XMLSerializer();
+		const xmlSerializer = new XMLSerializer({
+			includeShadowRoots: options && options.includeShadowRoots,
+			escapeEntities: false
+		});
 		let xml = '';
 		for (const node of this.content.childNodes) {
-			xml += xmlSerializer.serializeToString(node, options);
+			xml += xmlSerializer.serializeToString(node);
 		}
 		return xml;
 	}

diff --git a/packages/happy-dom/src/nodes/shadow-root/ShadowRoot.ts b/packages/happy-dom/src/nodes/shadow-root/ShadowRoot.ts
@@ -24,7 +24,9 @@ export default class ShadowRoot extends DocumentFragment implements IShadowRoot
 	 * @returns HTML.
 	 */
 	public get innerHTML(): string {
-		const xmlSerializer = new XMLSerializer();
+		const xmlSerializer = new XMLSerializer({
+			escapeEntities: false
+		});
 		let xml = '';
 		for (const node of this.childNodes) {
 			xml += xmlSerializer.serializeToString(node);

diff --git a/packages/happy-dom/src/xml-parser/XMLParser.ts b/packages/happy-dom/src/xml-parser/XMLParser.ts
@@ -9,7 +9,7 @@ import PlainTextElements from '../config/PlainTextElements';
 import IDocumentType from '../nodes/document-type/IDocumentType';
 import INode from '../nodes/node/INode';
 import IDocumentFragment from '../nodes/document-fragment/IDocumentFragment';
-import { decode } from 'he';
+import * as Entities from 'entities';
 
 /**
  * Markup RegExp.
@@ -99,7 +99,7 @@ export default class XMLParser {
 							// Plain text between tags.
 
 							currentNode.appendChild(
-								document.createTextNode(xml.substring(lastIndex, match.index))
+								document.createTextNode(Entities.decodeHTML(xml.substring(lastIndex, match.index)))
 							);
 						}
 
@@ -161,13 +161,17 @@ export default class XMLParser {
 							// Comment.
 
 							currentNode.appendChild(
-								document.createComment((match[6] ? '?' : '') + (match[3] || match[4] || match[6]))
+								document.createComment(
+									Entities.decodeHTML((match[6] ? '?' : '') + (match[3] || match[4] || match[6]))
+								)
 							);
 						} else if (match[5]) {
 							// Exclamation mark comment (usually <!DOCTYPE>).
 
+							const exclamationComment = Entities.decodeHTML(match[5]);
 							currentNode.appendChild(
-								this.getDocumentTypeNode(document, match[5]) || document.createComment(match[5])
+								this.getDocumentTypeNode(document, exclamationComment) ||
+									document.createComment(exclamationComment)
 							);
 						} else if (match[6]) {
 							// Processing instruction (not supported by HTML).
@@ -176,7 +180,9 @@ export default class XMLParser {
 							// Plain text between tags, including the match as it is not a valid start or end tag.
 
 							currentNode.appendChild(
-								document.createTextNode(xml.substring(lastIndex, markupRegexp.lastIndex))
+								document.createTextNode(
+									Entities.decodeHTML(xml.substring(lastIndex, markupRegexp.lastIndex))
+								)
 							);
 						}
 
@@ -205,7 +211,7 @@ export default class XMLParser {
 
 										const name = attributeMatch[1] || attributeMatch[5] || attributeMatch[9] || '';
 										const rawValue = attributeMatch[3] || attributeMatch[7] || '';
-										const value = rawValue ? decode(rawValue) : '';
+										const value = rawValue ? Entities.decodeHTMLAttribute(rawValue) : '';
 										const namespaceURI =
 											(<IElement>currentNode).tagName === 'SVG' && name === 'xmlns' ? value : null;
 
@@ -271,7 +277,9 @@ export default class XMLParser {
 
 							// Plain text elements such as <script> and <style> should only contain text.
 							currentNode.appendChild(
-								document.createTextNode(xml.substring(startTagIndex, match.index))
+								document.createTextNode(
+									Entities.decodeHTML(xml.substring(startTagIndex, match.index))
+								)
 							);
 
 							stack.pop();
@@ -289,7 +297,9 @@ export default class XMLParser {
 			if (lastIndex !== xml.length) {
 				// Plain text after tags.
 
-				currentNode.appendChild(document.createTextNode(xml.substring(lastIndex)));
+				currentNode.appendChild(
+					document.createTextNode(Entities.decodeHTML(xml.substring(lastIndex)))
+				);
 			}
 		}