seiyria

censor-sensor

  "description": "a better profanity filter",

    "build": "webpack --config webpack.config.js",

    "url": "git+https://github.com/seiyria/censor-sensor.git"

  "author": "Kyle Kemp <kyle@seiyria.com>",

    "url": "https://github.com/seiyria/censor-sensor/issues"

  "homepage": "https://github.com/seiyria/censor-sensor#readme",

    "awesome-typescript-loader": "^3.4.1",

Every other profanity filter just seems to be an array of words, checked against every word in a phrase. This is ridiculous - it's an O(N^2) check when it only needs to be O(N). Of course, this library, by default, only requires an O(N) check (and that's against your phrase; not my words). You can opt in to a "like" check which will check every word and try to match it to a substring.

Every other profanity filter also seems to neglect "tiers" of words. You might want to filter out slurs, but not common profanity. You can do that here.

If you want, you can pass a function that lets you censor words how you want. By default, it will be replaced with the `*` character in quantity based on the word (so "fuck" translates to "****").

That said, this library only has a list for english (`en`), but could be extended to add more fairly easily.

You can remove or add words at run time. By default, they'll be assigned the lowest tier (5).

Words are shamelessly based on [profanity-cleanser](https://github.com/LDNOOBW/profanity-cleanser), because I'm not creative enough to think of some of these.

import { CensorSensor } from 'censor-sensor';

// check for profanity (using string.includes)

censor.isProfaneIsh('bollockshead') // true

censor.isProfaneIsh('hello')        // true

// get the words that are profane from a phrase

censor.profaneIshWords('hello')     // ['hell']

censor.cleanProfanity('bollocks')   // '****' (by default)

censor.cleanProfanityIsh('hello')   // '****o' (by default)

censor.setCleanFunction((str) => Array.from(str, x => '%').join('')); // replace all bad characters with '%'

censor.cleanProfanity('bollocks')   // '%%%%%%%%' (by default)

censor.isProfane('bollocks')    // false 

censor.isProfane('uwotm8')                // false

censor.addLocale('custom', { uwotm8: 1 }) // the highest form of insult

censor.isProfane('uwotm8')                // true

As stated before, every word has a tier associated with it. Here is what each tier number means:

Feel free to contribute words, locales, or features.

import * as includes from 'lodash.includes';

  private locale: CensorLocale|string = 'en';

  private blackList: { [word: string]: boolean } = {};

  private customDictionary: { [word: string]: CensorTier } = {};

  private enabledTiers: { [tier: string]: boolean } = {

  private defaultCleanFunction: (str: string) => string = (str: string) => '****';

  private customCleanFunction: (str: string) => string;

  private get currentDictionary(): { [word: string]: CensorTier } {

    return merge({}, AllLocales[this.locale], this.customDictionary);

    return Object.assign({}, AllLocales[this.locale], this.customDictionary);

  private get cleanFunction(): (str: string) => string {

    return this.customCleanFunction || this.defaultCleanFunction;

  public addLocale(newLocale: string, dict: { [word: string]: CensorTier }): void {

  public setLocale(locale: CensorLocale|string): void {

  public disableTier(tier: CensorTier): void {

  public enableTier(tier: CensorTier): void {

  public addWord(word: string, tier: CensorTier = CensorTier.UserAdded): void {

  public removeWord(word: string): void {

  private prepareForParsing(phrase: string): string {

  private _isProfane(word: string, dict: { [word: string]: CensorTier }): boolean {

    if(!this.enabledTiers[wordTier]) return false;

  public isProfane(phrase: string): boolean {

    return some(this.prepareForParsing(phrase).split(' '), word => this._isProfane(word, dict));

    return this.prepareForParsing(phrase).split(' ').some(word => this._isProfane(word, dict));

  private _isProfaneIsh(phrase: string, dict: { [word: string]: CensorTier }): boolean {

    Object.keys(dict).forEach(dictWord => {

      if(includes(phrase, dictWord) && this.enabledTiers[tier]) isAnyMatch = true;

      if (phrase.indexOf(dictWord) !== -1 && this.enabledTiers[tier]) isAnyMatch = true;

  public isProfaneIsh(phrase: string): boolean {

    return this._isProfaneIsh(this.prepareForParsing(phrase), dict);

  private _profaneIshWords(phrase: string, dict: { [word: string]: CensorTier }): string[] {

      if(includes(phrase, dictWord) && this.enabledTiers[tier]) foundProfanity.push(dictWord);

      if(phrase.indexOf(dictWord) !== -1 && this.enabledTiers[tier]) foundProfanity.push(dictWord);

  public profaneIshWords(phrase: string) {

    return this._profaneIshWords(phrase, dict);

  public setCleanFunction(func: (str: string) => string): void {

  public cleanProfanity(phrase: string): string {

    const cleanFunc = this.cleanFunction;

    const joinPhrase = phrase.split(' ').map(word => {

      if(!this._isProfane(word.toLowerCase(), dict)) return word;

  public cleanProfanityIsh(phrase: string): string {

    const comparePhrase = this.prepareForParsing(phrase);

    const allProfanity = this.profaneIshWords(comparePhrase);

      const regex = new RegExp(word, 'gi');

      phrase = phrase.replace(regex, cleanFunc(word));

		{
		"name": "censor-sensor",
		"version": "1.0.3",
		"version": "1.0.4",
		"description": "a better profanity filter",
		@@ -35,7 +35,3 @@ "main": "./dist/index.js",
		},
		"dependencies": {
		"lodash.includes": "^4.3.0",
		"lodash.merge": "^4.6.1",
		"lodash.some": "^4.6.0"
		}
		"dependencies": {}
		}

		@@ -1,8 +0,3 @@

		import { AllLocales } from '../locales';

		import * as merge from 'lodash.merge';
		import * as includes from 'lodash.includes';
		import * as some from 'lodash.some';

		export type CensorLocale = 'en';
		@@ -38,3 +33,3 @@
		private get currentDictionary(): { [word: string]: CensorTier } {
		return merge({}, AllLocales[this.locale], this.customDictionary);
		return Object.assign({}, AllLocales[this.locale], this.customDictionary);
		}
		@@ -93,3 +88,3 @@

		return some(this.prepareForParsing(phrase).split(' '), word => this._isProfane(word, dict));
		return this.prepareForParsing(phrase).split(' ').some(word => this._isProfane(word, dict));
		}
		@@ -105,3 +100,3 @@

		if(includes(phrase, dictWord) && this.enabledTiers[tier]) isAnyMatch = true;
		if (phrase.indexOf(dictWord) !== -1 && this.enabledTiers[tier]) isAnyMatch = true;
		});
		@@ -124,3 +119,3 @@

		if(includes(phrase, dictWord) && this.enabledTiers[tier]) foundProfanity.push(dictWord);
		if(phrase.indexOf(dictWord) !== -1 && this.enabledTiers[tier]) foundProfanity.push(dictWord);
		});
		@@ -127,0 +122,0 @@

New alerts

Improved metrics

Worsened metrics

Dependency changes