bad-word-detector

Detect if Japanese/English input string contains blacklisted words, while allowing specified whitelisted words that contain blacklisted words.

Normalizes input string in the following ways:

• Removes accents from latin characters (ë→e)
• Fixes full-width latin characters to half-width (ｆｒｕｉｔ→fruit)
• Fixes half-width katakana to full-width (ｼﾞｭｰｽ→ジュース)
• Converts all hiragana to katakana (じゅーす→ジュース)
• Removes spaces & special characters

Optionally, you may specify a mode for each word in the bad word list to further control what gets marked as 'bad':

enum DetectionMode {
	Default = 1,
	ExactMatch = 2,
	UnNormalizedOnly = 3,
	UnNormalizedOnlyExactMatch = 4,
};

Example isBad outputs for the word apples with each detection mode are as follows:

input	Default	ExactMatch	UnNormalizedOnly	UnNormalizedOnlyExactMatch
apples	true	true	true	true
APPLES	true	true	false	false
iloveapples	true	false	true	false
iloveAPPLES	true	false	false	false

Example isBad outputs for the word モモ with each detection mode are as follows:

input	Default	ExactMatch	UnNormalizedOnly	UnNormalizedOnlyExactMatch
モモ	true	true	true	true
もも	true	true	false	false
モモが食べたい	true	false	true	false
ももが食べたい	true	false	false	false

Usage

const badWordList = {
	苺: {
		whitelist: ["苺大福"],
	},
	ジュース: {
		whitelist: [],
	},
	watermelon: {
		mode: 2,
		whitelist: [],
	}
};

const badWordDetector = new BadWordDetector(badWordList);

badWordDetector.isBad("苺が食べたい"); // true
badWordDetector.isBad("ｼﾞｭｰｽ"); // true
badWordDetector.isBad("じゅーす"); // true
badWordDetector.isBad("ｗaｔ🍉ｅｒmëlon"); // true

badWordDetector.isBad("苺大福が食べたい"); // false
badWordDetector.isBad("シューズ"); // false
badWordDetector.isBad("シュース"); // false
badWordDetector.isBad("melon"); // false
badWordDetector.isBad("water"); // false
badWordDetector.isBad("containswatermelonbutok"); // false
badWordDetector.isBad("i loveｗaｔ🍉ｅｒmëlon"); // false