Nepali profanity detection and censoring (Devanagari + romanized variants).
npm install nepali-profanity-checkerpowershell
npm ci
npm run build
npm test
`
Run interactive console to try phrases manually:
`powershell
npm run console
`
API
Import the library and use the helpers:
`js
const { containsProfanity, findProfanities, censorProfanity } = require('./dist/index.js');
console.log(containsProfanity('तँ हरामी हो')); // true
console.log(findProfanities('तँ हरामी हो')); // details about matches
console.log(censorProfanity('तँ हरामी हो')); // 'तँ * हो'
// Fuzzy (approximate) matching example
console.log(containsProfanity('you are h4rami', { fuzzy: true }));
`
Features
- Loads a curated JSON wordlist (src/data/badword.json) with variants and categories.
- Unicode normalization (NFKC) and diacritic removal for robust matching.
- Optional fuzzy matching (Levenshtein) to catch obfuscated/typo variants.
- Simple censoring that preserves Unicode codepoint lengths.
Publishing
Before publishing run:
`powershell
npm run build
npm pack # creates a tarball to inspect what will be published
then when ready:
npm publish
`
Contributing
Add/modify words in src/data/badword.json. Add unit tests under test/ and submit a PR.
If you would like to add some profane words, please make sure it is unique than the Current List down below
to avoid duplication. Please make sure to follow the json schema exactly [shown below] while making json data.
Current Profane Words List
`bash
[
'harami', 'murkha', 'badmas', 'saley',
'sali', 'kukur', 'randi', 'rande',
'bhate', 'bhalu', 'machikni', 'chhakka',
'lado', 'lauro', 'muji', 'puti',
'chhuchi', 'chhura', 'turi', 'geda',
'gu', 'gukhane', 'pakhe', 'boka',
'bokni', 'jantha', 'condo', 'chaak',
'chaak ko pwal', 'bhosdika', 'madarchod', 'bhenchod',
'lund', 'lodo', 'chod', 'chodna',
'rakhnu', 'pesa', 'gand', 'gandako',
'gandu', 'pesa garne', 'sungur', 'budi',
'budo', 'chikney', 'lundra', 'lundri',
'gadha', 'kutta', 'kichkich', 'thukk',
'nalayak', 'beijjat', 'ghrina', 'ninda',
'kachara', 'nikamma', 'sasto manche', 'gali',
'fohor', 'ghinlagdo', 'nindaniya', 'gand',
'gandako', 'gandu', 'chutiya', 'bekar',
'jhola', 'chor', 'hijada', 'pagal',
'kutne', 'latti', 'beshya', 'bokshi',
'chaak', 'Chickne', 'Gula', 'moot',
'murda', 'bajiya', 'khate', 'torpe',
'daka', 'paji',
'dhoti', 'radi',
'paad', 'mukhulla',
'gobre', 'bhusya',
'dhurt', 'bhaisi, 'kamina', 'kutti', 'haramzada'
]
`
Json Data Format
Each profane word must follow the structure below.
`bash
{
"base": "kutti", # Data type is string. The canonical form of the profane word written in Roman script.
"nepali": ["कुत्ती"], # Data type is string. The word written in Devanagari (Nepali/Hindi script).
"variants": [
"kutti",
"kuttiii",
"kutt1",
"k@tti"
], # Data type for variants is string[]. Common alternative spellings, misspellings, or obfuscated forms used to bypass profanity filters.
"category": ["sexual", "insult"] # Data type for category is string[]. Labels describing the type of profanity. Used for filtering and severity control.
}
``