fix: Filtering non ascii bad words (#35418)

.changeset/witty-ads-visit.md

@@ -0,0 +1,5 @@
+---
+'@rocket.chat/meteor': patch
+---
+
+Adds support for filtering bad words from messages for languages other than English

apps/meteor/server/services/messages/hooks/BeforeSaveBadWords.ts

@@ -1,18 +1,28 @@
 import { type IMessage } from '@rocket.chat/core-typings';
+import { Logger } from '@rocket.chat/logger';
 import type BadWordsFilter from 'bad-words';
 
 export class BeforeSaveBadWords {
 	badWords: BadWordsFilter | null = null;
 
+	badWordsRegex: RegExp | null = null;
+
+	protected logger: Logger;
+
+	constructor() {
+		this.logger = new Logger('BadWordsFilter');
+	}
+
 	async configure(badWordsList?: string, goodWordsList?: string) {
 		const { default: Filter } = await import('bad-words');
+		const badWords =
+			badWordsList
+				?.split(',')
+				.map((word) => word.trim())
+				.filter(Boolean) || [];
 
 		const options = {
-			list:
-				badWordsList
-					?.split(',')
-					.map((word) => word.trim())
-					.filter(Boolean) || [],
+			list: badWords,
 			// library definition does not allow optional definition
 			exclude: undefined,
 			splitRegex: undefined,
@@ -24,6 +34,13 @@ export class BeforeSaveBadWords {
 
 		this.badWords = new Filter(options);
 
+		try {
+			this.badWordsRegex = new RegExp(`(?<=^|[\\p{P}\\p{Z}])(${badWords.join('|')})(?=$|[\\p{P}\\p{Z}])`, 'gmiu');
+		} catch (error) {
+			this.badWordsRegex = null;
+			this.logger.error('Erorr when initializing bad words filter', error);
+		}
+
 		if (goodWordsList?.length) {
 			this.badWords.removeWords(...goodWordsList.split(',').map((word) => word.trim()));
 		}
@@ -31,6 +48,7 @@ export class BeforeSaveBadWords {
 
 	disable() {
 		this.badWords = null;
+		this.badWordsRegex = null;
 	}
 
 	async filterBadWords({ message }: { message: IMessage }): Promise<IMessage> {
@@ -42,6 +60,10 @@ export class BeforeSaveBadWords {
 			message.msg = this.badWords.clean(message.msg);
 		} catch (error) {
 			// ignore
+		} finally {
+			if (this.badWordsRegex) {
+				message.msg = message.msg.replace(this.badWordsRegex, (match) => '*'.repeat(match.length));
+			}
 		}
 
 		return message;

apps/meteor/tests/unit/server/services/messages/hooks/BeforeSaveBadWords.tests.ts

@@ -80,4 +80,48 @@ describe('Filter bad words before saving message', () => {
 
 		return expect(result.msg).to.be.equal('hell');
 	});
+
+	it('should filter non ascii bad words', async () => {
+		const badWords = new BeforeSaveBadWords();
+		await badWords.configure('バカ');
+
+		const message = createMessage('hell is バカ');
+
+		const result = await badWords.filterBadWords({ message });
+
+		return expect(result.msg).to.be.equal('**** is **');
+	});
+
+	it('should filter just the non ascii bad words', async () => {
+		const badWords = new BeforeSaveBadWords();
+		await badWords.configure('バカ');
+
+		const message = createMessage('バカ');
+
+		const result = await badWords.filterBadWords({ message });
+
+		return expect(result.msg).to.be.equal('**');
+	});
+
+	it('should filter non ascii bad words with punctuation', async () => {
+		const badWords = new BeforeSaveBadWords();
+		await badWords.configure('バカ');
+
+		const message = createMessage('バカ.');
+
+		const result = await badWords.filterBadWords({ message });
+
+		return expect(result.msg).to.be.equal('**.');
+	});
+
+	it('should not filter non ascii bad words, if part of another word ', async () => {
+		const badWords = new BeforeSaveBadWords();
+		await badWords.configure('バカ');
+
+		const message = createMessage('TESTバカTEST');
+
+		const result = await badWords.filterBadWords({ message });
+
+		return expect(result.msg).to.be.equal('TESTバカTEST');
+	});
 });