@rikatech/text-profanity

A fast, lightweight, and highly advanced profanity filter with comprehensive Anti-Bypass capabilities. Protects your text inputs across multiple languages (English, Vietnamese, and Japanese).

Installation

npm install @rikatech/text-profanity

Anti-Bypass Features

This package doesn't just match words; it automatically untangles complex attempts by users to evade filters:

• Noise & Punctuation Ignored: Dots, spaces, symbols are skipped over (e.g., f.u.c.k -> caught).
• Leetspeak (Teencode) Conversion: Numbers & symbols are reverted to letters (e.g., @->a, 3->e, 0->o, !->i).
• Repeating Characters: Extended vowels and characters are collapsed (e.g., shiiiiiiitttt -> shit).
• Unaccented/Extended Latin Support: Strips extended Unicode and intelligently matches unaccented variations to native dictionary items (e.g., d m catches đ m).
• Advanced Japanese Script Armor: Automatically folds Katakana to Hiragana (e.g. シネ -> しね), normalizes half-width characters, merges detached Dakuon (ﾊﾞｶ -> バカ -> ばか), and bypasses long vowels ー.

Usage

import { WordFilter } from "@rikatech/text-profanity";

// Initialize with target languages (built-in options: 'en', 'vi', 'ja')
const filter = new WordFilter({ 
    languages: ["en", "vi", "ja"], 
    placeholder: "*" // Default mask string
});

// 1. Check if text contains bad words
console.log(filter.hasBadWord("shiiiiiiit")); // Returns: true
console.log(filter.hasBadWord("hello world")); // Returns: false

// 2. Intelligent Masking (preserves text formatting & lengths)
console.log(filter.mask("you are a b.i.t.c.h")); 
// Output: "you are a *********"

// 3. Add custom bad words programmatically
filter.addBadWords(["scam", "spam"]);
console.log(filter.mask("This is a scam website!"));
// Output: "This is a **** website!"

Contributing and Tests

Written safely with TypeScript and rigorously tested with Jest parsing complex filter evasion scenarios.

npm run test