#875@trivial: Continues on implementation.

capricorn86 · May 9, 2023 · da6edaa · da6edaa
1 parent 2a78a72
commit da6edaa
Show file tree

Hide file tree

Showing 3 changed files with 105 additions and 13 deletions.
diff --git a/packages/happy-dom/src/xml-parser/XMLParser.ts b/packages/happy-dom/src/xml-parser/XMLParser.ts
@@ -38,10 +38,9 @@ const MARKUP_REGEXP =
  * Group 7: Attribute value when the attribute has a value using single apostrophe (e.g. "value" in "<div name='value'>").
  * Group 8: Attribute apostrophe when the attribute has a value using single apostrophe (e.g. "name" in "<div name='value'>").
  * Group 9: Attribute name when the attribute has no value (e.g. "disabled" in "<div disabled>").
- * Group 10: Invalid characters to trim.
  */
 const ATTRIBUTE_REGEXP =
-	/\s*([a-zA-Z0-9-_:]+) *= *("{0,1})([^"]*)("{0,1})|\s*([a-zA-Z0-9-_:]+) *= *('{0,1})([^']*)('{0,1})|\s*([a-zA-Z0-9-_:]+)|\s*([^a-zA-Z0-9-_:]+$)/gm;
+	/\s*([a-zA-Z0-9-_:]+) *= *("{0,1})([^"]*)("{0,1})|\s*([a-zA-Z0-9-_:]+) *= *('{0,1})([^']*)('{0,1})|\s*([a-zA-Z0-9-_:]+)/gm;
 
 enum MarkupReadStateEnum {
 	startOrEndTag = 'startOrEndTag',
@@ -190,10 +189,11 @@ export default class XMLParser {
 							// Attribute name and value.
 
 							const attributeString = data.substring(startTagIndex, match.index);
-							if (attributeString) {
+							let hasAttributeStringEnded = true;
+
+							if (!!attributeString) {
 								const attributeRegexp = new RegExp(ATTRIBUTE_REGEXP, 'gm');
 								let attributeMatch: RegExpExecArray;
-								let attributeLatestIndex = 0;
 
 								while ((attributeMatch = attributeRegexp.exec(attributeString))) {
 									if (
@@ -212,20 +212,21 @@ export default class XMLParser {
 										(<IElement>currentNode).setAttributeNS(namespaceURI, name, value);
 
 										startTagIndex += attributeMatch[0].length;
-										attributeLatestIndex = attributeRegexp.lastIndex;
-									} else if (attributeMatch[10] && attributeLatestIndex === attributeMatch.index) {
-										// Invalid characters that should be trimmed.
-
-										startTagIndex += attributeMatch[0].length;
+									} else {
+										hasAttributeStringEnded = false;
+										break;
 									}
 								}
 							}
 
 							// We need to check if the attribute string is read completely.
 							// The attribute string can potentially contain "/>" or ">".
-							if (startTagIndex === match.index) {
+							if (hasAttributeStringEnded) {
 								// Non-self-closing tag
-								if (match[7] && !VoidElements.includes((<IElement>currentNode).tagName)) {
+								if (
+									(match[7] || (<IElement>currentNode).namespaceURI === NamespaceURI.html) &&
+									!VoidElements.includes((<IElement>currentNode).tagName)
+								) {
 									// Plain text elements such as <script> and <style> should only contain text.
 									plainTextTagName = PlainTextElements.includes((<IElement>currentNode).tagName)
 										? (<IElement>currentNode).tagName

diff --git a/packages/happy-dom/test/xml-parser/XMLParser.test.ts b/packages/happy-dom/test/xml-parser/XMLParser.test.ts
@@ -15,7 +15,7 @@ const GET_EXPECTED_HTML = (html: string): string =>
 		.replace('<?processing instruction?>', '<!--?processing instruction?-->')
 		.replace('<!Exclamation mark comment>', '<!--Exclamation mark comment-->')
 		.replace('<!DOCTYPE HTML', '<!DOCTYPE html')
-		.replace('<self-closing-custom-tag />', '<self-closing-custom-tag></self-closing-custom-tag>');
+		.replace('<img />', '<img>');
 
 describe('XMLParser', () => {
 	let window: IWindow;
@@ -535,5 +535,96 @@ describe('XMLParser', () => {
 				'index > 1'
 			);
 		});
+
+		it('Doesn\'t close non-void elements when using "/>" when namespace is HTML.', () => {
+			const root = XMLParser.parse(
+				window.document,
+				`
+                <span key1="value1"/>
+                <span key1="value1" key2/>
+                <span key2/>
+                `
+			);
+
+			expect(new XMLSerializer().serializeToString(root).replace(/\s/gm, '')).toBe(
+				`
+                <span key1="value1">
+                    <span key1="value1" key2="">
+                        <span key2=""></span>
+                    </span>
+                </span>`.replace(/\s/gm, '')
+			);
+		});
+
+		it('Parses malformed attributes.', () => {
+			const root = XMLParser.parse(
+				window.document,
+				`
+                <span key1="value1""></span>
+                <span key1="value1"" key2></span>
+                <span key1 key2 key3="value3""></span>
+                <img key1="value1"" key2/>
+                <img key1="value1""
+                />
+                <img key1="value1""
+                key2/>
+                <span key1 key2 key3="value3"''" " "></span>
+                <span key1="value1 > value2"></span>
+                <img key1="value1 /> value2"/>
+                `
+			);
+
+			expect(root.children.length).toBe(8);
+
+			expect(root.children[0].attributes.length).toBe(1);
+			expect(root.children[0].attributes[0].name).toBe('key1');
+			expect(root.children[0].attributes[0].value).toBe('value1');
+
+			expect(root.children[1].attributes.length).toBe(2);
+			expect(root.children[1].attributes[0].name).toBe('key1');
+			expect(root.children[1].attributes[0].value).toBe('value1');
+			expect(root.children[1].attributes[1].name).toBe('key2');
+			expect(root.children[1].attributes[1].value).toBe('');
+
+			expect(root.children[2].attributes.length).toBe(3);
+			expect(root.children[2].attributes[0].name).toBe('key1');
+			expect(root.children[2].attributes[0].value).toBe('');
+			expect(root.children[2].attributes[1].name).toBe('key2');
+			expect(root.children[2].attributes[1].value).toBe('');
+			expect(root.children[2].attributes[2].name).toBe('key3');
+			expect(root.children[2].attributes[2].value).toBe('value3');
+
+			expect(root.children[3].attributes.length).toBe(2);
+			expect(root.children[3].attributes[0].name).toBe('key1');
+			expect(root.children[3].attributes[0].value).toBe('value1');
+			expect(root.children[3].attributes[1].name).toBe('key2');
+			expect(root.children[3].attributes[1].value).toBe('');
+
+			expect(root.children[4].attributes.length).toBe(1);
+			expect(root.children[4].attributes[0].name).toBe('key1');
+			expect(root.children[4].attributes[0].value).toBe('value1');
+
+			expect(root.children[5].attributes.length).toBe(2);
+			expect(root.children[5].attributes[0].name).toBe('key1');
+			expect(root.children[5].attributes[0].value).toBe('value1');
+			expect(root.children[5].attributes[1].name).toBe('key2');
+			expect(root.children[5].attributes[1].value).toBe('');
+
+			expect(root.children[6].attributes.length).toBe(3);
+			expect(root.children[6].attributes[0].name).toBe('key1');
+			expect(root.children[6].attributes[0].value).toBe('');
+			expect(root.children[6].attributes[1].name).toBe('key2');
+			expect(root.children[6].attributes[1].value).toBe('');
+			expect(root.children[6].attributes[2].name).toBe('key3');
+			expect(root.children[6].attributes[2].value).toBe('value3');
+
+			expect(root.children[7].attributes.length).toBe(1);
+			expect(root.children[7].attributes[0].name).toBe('key1');
+			expect(root.children[7].attributes[0].value).toBe('value1 > value2');
+
+			expect(root.children[8].attributes.length).toBe(1);
+			expect(root.children[8].attributes[0].name).toBe('key1');
+			expect(root.children[8].attributes[0].value).toBe('value1 /> value2');
+		});
 	});
 });
diff --git a/packages/happy-dom/test/xml-parser/data/XMLParserHTML.ts b/packages/happy-dom/test/xml-parser/data/XMLParserHTML.ts
@@ -19,7 +19,7 @@ export default `
 				<!-- Comment 2 !-->
 			</article>
 			<img>
-			<self-closing-custom-tag />
+			<img />
 		</body>
 	</html>
 `.trim();