Merge pull request #123 from nonara/trim-fix

fix issue #115
taoqf · May 25, 2021 · 0632085 · 0632085
2 parents 416fc9f + d914efa
commit 0632085
Show file tree

Hide file tree

Showing 3 changed files with 58 additions and 7 deletions.
diff --git a/src/nodes/html.ts b/src/nodes/html.ts
@@ -133,6 +133,7 @@ export default class HTMLElement extends Node {
 
 		return JSON.stringify(attr.replace(/"/g, '&quot;'));
 	}
+
 	/**
 	 * Creates an instance of HTMLElement.
 	 * @param keyAttrs	id and class attribute
@@ -260,7 +261,7 @@ export default class HTMLElement extends Node {
 					// Whitespace node, postponed output
 					currentBlock.prependWhitespace = true;
 				} else {
-					let text = node.text;
+					let text = (<TextNode>node).trimmedText;
 					if (currentBlock.prependWhitespace) {
 						text = ` ${text}`;
 						currentBlock.prependWhitespace = false;
@@ -401,7 +402,7 @@ export default class HTMLElement extends Node {
 				if ((node as TextNode).isWhitespace) {
 					return;
 				}
-				node.rawText = node.rawText.trim();
+				node.rawText = (<TextNode>node).trimmedText;
 			} else if (node.nodeType === NodeType.ELEMENT_NODE) {
 				(node as HTMLElement).removeWhitespace();
 			}

diff --git a/src/nodes/text.ts b/src/nodes/text.ts
@@ -17,6 +17,45 @@ export default class TextNode extends Node {
 	 */
 	public nodeType = NodeType.TEXT_NODE;
 
+	private _trimmedText?: string;
+
+	/**
+	 * Returns text with all whitespace trimmed except single leading/trailing non-breaking space
+	 */
+	public get trimmedText() {
+		if (this._trimmedText !== undefined) return this._trimmedText;
+
+		const text = this.rawText;
+		let i = 0;
+		let startPos;
+		let endPos;
+
+		while (i >= 0 && i < text.length) {
+			if (/\S/.test(text[i])) {
+				if (startPos === undefined) {
+					startPos = i;
+					i = text.length;
+				} else {
+					endPos = i;
+					i = void 0;
+				}
+			}
+
+			if (startPos === undefined) i++;
+			else i--;
+		}
+
+		if (startPos === undefined) startPos = 0;
+		if (endPos === undefined) endPos = text.length - 1;
+
+		const hasLeadingSpace = startPos > 0 && /[^\S\r\n]/.test(text[startPos-1]);
+		const hasTrailingSpace = endPos < (text.length - 1) && /[^\S\r\n]/.test(text[endPos+1]);
+
+		this._trimmedText = (hasLeadingSpace ? ' ' : '') + text.slice(startPos, endPos + 1) + (hasTrailingSpace ? ' ' : '');
+
+		return this._trimmedText;
+	}
+
 	/**
 	 * Get unescaped text value of current node and its children.
 	 * @return {string} text content

diff --git a/test/html.js b/test/html.js
@@ -126,10 +126,10 @@ describe('HTML Parser', function () {
 			const script = root.firstChild;
 			const style = root.lastChild;
 			script.childNodes.should.not.be.empty;
-			script.childNodes.should.eql([new TextNode('1', script)]);
+			script.childNodes.should.eql([ new TextNode('1', script) ]);
 			script.text.should.eql('1');
 			style.childNodes.should.not.be.empty;
-			style.childNodes.should.eql([new TextNode('2&amp;', style)]);
+			style.childNodes.should.eql([ new TextNode('2&amp;', style) ]);
 			style.text.should.eql('2&');
 			style.rawText.should.eql('2&amp;');
 		});
@@ -198,14 +198,25 @@ describe('HTML Parser', function () {
 
 		describe('#removeWhitespace()', function () {
 			it('should remove whitespaces while preserving nodes with content', function () {
-				const root = parseHTML('<p> \r \n  \t <h5> 123 </h5></p>');
+				const root = parseHTML('<p> \r \n  \t <h5>  123  </h5></p>');
+
+				const textNode = new TextNode('  123  ');
+				textNode.rawText = textNode.trimmedText;
+				textNode.rawText.should.eql(' 123 ');
 
 				const p = new HTMLElement('p', {}, '', root);
-				p.appendChild(new HTMLElement('h5', {}, ''))
-					.appendChild(new TextNode('123'));
+				p
+					.appendChild(new HTMLElement('h5', {}, ''))
+					.appendChild(textNode);
 
 				root.firstChild.removeWhitespace().should.eql(p);
 			});
+
+			it('should preserve legitimate leading/trailing whitespace in TextNode', function () {
+				parseHTML('<p>Hello  <em>World</em>!</p>').removeWhitespace().firstChild.text.should.eql('Hello World!');
+				parseHTML('<p>\t\nHello\n\t<em>World</em>!</p>').removeWhitespace().firstChild.text.should.eql('HelloWorld!');
+				parseHTML('<p>\t\n  Hello \n\t<em>World</em>!</p>').removeWhitespace().firstChild.text.should.eql(' Hello World!');
+			});
 		});
 
 		describe('#rawAttributes', function () {