Desktop: Security: Fixes #6004: Prevent XSS in Goto Anything

.eslintignore

@@ -1927,6 +1927,9 @@ packages/renderer/headerAnchor.js.map
 packages/renderer/htmlUtils.d.ts
 packages/renderer/htmlUtils.js
 packages/renderer/htmlUtils.js.map
+packages/renderer/htmlUtils.test.d.ts
+packages/renderer/htmlUtils.test.js
+packages/renderer/htmlUtils.test.js.map
 packages/renderer/index.d.ts
 packages/renderer/index.js
 packages/renderer/index.js.map

.gitignore

@@ -1917,6 +1917,9 @@ packages/renderer/headerAnchor.js.map
 packages/renderer/htmlUtils.d.ts
 packages/renderer/htmlUtils.js
 packages/renderer/htmlUtils.js.map
+packages/renderer/htmlUtils.test.d.ts
+packages/renderer/htmlUtils.test.js
+packages/renderer/htmlUtils.test.js.map
 packages/renderer/index.d.ts
 packages/renderer/index.js
 packages/renderer/index.js.map

packages/app-desktop/plugins/GotoAnything.tsx

@@ -19,6 +19,7 @@ const { mergeOverlappingIntervals } = require('@joplin/lib/ArrayUtils.js');
 import markupLanguageUtils from '../utils/markupLanguageUtils';
 import focusEditorIfEditorCommand from '@joplin/lib/services/commands/focusEditorIfEditorCommand';
 import Logger from '@joplin/lib/Logger';
+import { MarkupToHtml } from '@joplin/renderer';
 
 const logger = Logger.create('GotoAnything');
 
@@ -81,7 +82,7 @@ class Dialog extends React.PureComponent<Props, State> {
 	private inputRef: any;
 	private itemListRef: any;
 	private listUpdateIID_: any;
-	private markupToHtml_: any;
+	private markupToHtml_: MarkupToHtml;
 	private userCallback_: any = null;
 
 	constructor(props: Props) {

packages/lib/htmlUtils.ts

@@ -1,7 +1,6 @@
 const urlUtils = require('./urlUtils.js');
 const Entities = require('html-entities').AllHtmlEntities;
 const htmlentities = new Entities().encode;
-const htmlparser2 = require('@joplin/fork-htmlparser2');
 const { escapeHtml } = require('./string-utils.js');
 
 // [\s\S] instead of . for multiline matching
@@ -138,40 +137,6 @@ class HtmlUtils {
 		return output.join(' ');
 	}
 
-	public stripHtml(html: string) {
-		const output: string[] = [];
-
-		const tagStack: any[] = [];
-
-		const currentTag = () => {
-			if (!tagStack.length) return '';
-			return tagStack[tagStack.length - 1];
-		};
-
-		const disallowedTags = ['script', 'style', 'head', 'iframe', 'frameset', 'frame', 'object', 'base'];
-
-		const parser = new htmlparser2.Parser({
-
-			onopentag: (name: string) => {
-				tagStack.push(name.toLowerCase());
-			},
-
-			ontext: (decodedText: string) => {
-				if (disallowedTags.includes(currentTag())) return;
-				output.push(decodedText);
-			},
-
-			onclosetag: (name: string) => {
-				if (currentTag() === name.toLowerCase()) tagStack.pop();
-			},
-
-		}, { decodeEntities: true });
-
-		parser.write(html);
-		parser.end();
-
-		return output.join('').replace(/\s+/g, ' ');
-	}
 }
 
 export default new HtmlUtils();

packages/renderer/htmlUtils.test.ts

@@ -0,0 +1,32 @@
+import htmlUtils from './htmlUtils';
+
+describe('htmlUtils', () => {
+
+	test('should strip off HTML', () => {
+		const testCases = [
+			[
+				'',
+				'',
+			],
+			[
+				'<b>test</b>',
+				'test',
+			],
+			[
+				'Joplin&circledR;',
+				'Joplin®',
+			],
+			[
+				'&lt;b&gttest&lt;/b&gt',
+				'&lt;b>test&lt;/b>',
+			],
+		];
+
+		for (const t of testCases) {
+			const [input, expected] = t;
+			const actual = htmlUtils.stripHtml(input);
+			expect(actual).toBe(expected);
+		}
+	});
+
+});

packages/renderer/htmlUtils.ts

@@ -97,8 +97,7 @@ class HtmlUtils {
 		return selfClosingElements.includes(tagName.toLowerCase());
 	}
 
-	// TODO: copied from @joplin/lib
+	public stripHtml(html: string) {
-	stripHtml(html: string) {
 		const output: string[] = [];
 
 		const tagStack: string[] = [];
@@ -130,7 +129,14 @@ class HtmlUtils {
 		parser.write(html);
 		parser.end();
 
-		return output.join('').replace(/\s+/g, ' ');
+		// In general, we want to get back plain text from this function, so all
+		// HTML entities are decoded. Howver, to prevent XSS attacks, we
+		// re-encode all the "<" characters, which should break any attempt to
+		// inject HTML tags.
+
+		return output.join('')
+			.replace(/\s+/g, ' ')
+			.replace(/</g, '&lt;');
 	}
 
 	public sanitizeHtml(html: string, options: any = null) {