@fldx/sopan

Small TypeScript-first profanity filter for Indonesian text.

@fldx/sopan is dependency-free at runtime and exposes a compact API for JavaScript and TypeScript projects.

Install

npm install @fldx/sopan

pnpm add @fldx/sopan

yarn add @fldx/sopan

bun add @fldx/sopan

Usage

import { addWords, clean, containsProfanity, findProfanity } from "@fldx/sopan";

addWords(["kasar"]);

containsProfanity("dasar t41"); // true
findProfanity("A*N*J*I*N*G");
clean("dasar t41"); // "dasar ***"
clean("ka$aar!"); // "***!"

CommonJS is supported too:

const { clean, containsProfanity } = require("@fldx/sopan");

containsProfanity("ta1"); // true
clean("ta1"); // "***"

API

addWords(words)

Adds words to the shared default filter once, so future calls to containsProfanity, findProfanity, and clean detect them automatically.

addWords(["kasar", "kata-baru"]);

containsProfanity("ka$aar"); // true
clean("kata-baru"); // "***"

clearWords()

Removes words previously registered through addWords.

clearWords();

containsProfanity(input, options)

Returns true when the input contains a profane word.

containsProfanity("ini santai"); // false
containsProfanity("ini tai"); // true
containsProfanity("ka$aar", { additionalWords: ["kasar"] }); // true

findProfanity(input, options)

Returns match details with the configured dictionary word, raw token, normalized token, and index.

findProfanity("halo t41");
// [{ word: "tai", raw: "t41", normalized: "tai", index: 5 }]

findProfanity("ka$aar", { additionalWords: ["kasar"] });
// [{ word: "kasar", raw: "ka$aar", normalized: "kasar", index: 0 }]

clean(input, options)

Replaces profane words with "***" by default.

clean("dasar t41"); // "dasar ***"
clean("dasar t41", { replacement: "[redacted]" });
clean("dasar t41", { replacement: (match) => `[${match.word}]` });
clean("ka$aar", { additionalWords: ["kasar"], replacement: "[custom]" });

createFilter(options)

Creates a custom filter with your own word list.

const filter = createFilter({
  words: ["kasar", "contoh"] as const
});

filter.containsProfanity("ka$aar");

Matching Behavior

The default filter focuses on Indonesian profanity and normalizes:

• mixed casing: TAI
• simple leetspeak: t41, ta1
• symbol separators: a*n*j*i*n*g
• repeated letters: annjiiinggg

Matching is token-based, so tai is detected while santai is not.

The default Indonesian word list was expanded from public Indonesian rude-word references including Wiktionary's Indonesian "kata kasar" category, a translator-maintained Indonesian profanity list, a public GitHub gist, and selected regional Indonesian references for Sundanese, Javanese, Batak, and Medan usage. It intentionally keeps only single-token words in the core package.

Roadmap

• Expand Indonesian dictionary coverage.
• Add language packs without increasing the default bundle size.
• Add optional phrase matching for multi-word profanity.