[python_utils.git] / profanity_filter.py

#!/usr/bin/env python3

import logging
import random
import re
import string
import sys

import nltk
from nltk.stem import PorterStemmer

import decorator_utils
import string_utils


logger = logging.getLogger(__name__)


@decorator_utils.singleton
class ProfanityFilter(object):
    def __init__(self):
        self.bad_words = set(
            [
                'acrotomophilia',
                'anal',
                'analingus',
                'anally',
                'anilingus',
                'anus',
                'arsehol',
                'arsehole',
                'ass',
                'asses',
                'asshol',
                'asshole',
                'assmunch',
                'auto erot',
                'auto erotic',
                'autoerotic',
                'babeland',
                'babi batter',
                'baby batter',
                'ball gag',
                'ball gravi',
                'ball gravy',
                'ball kick',
                'ball kicking',
                'ball lick',
                'ball licking',
                'ball sack',
                'ball suck',
                'ball sucking',
                'ball zack',
                'bangbro',
                'bangbros',
                'bare legal',
                'bareback',
                'barely legal',
                'barenak',
                'barenaked',
                'bastardo',
                'bastinado',
                'bbc',
                'bbw',
                'bdsm',
                'beaver cleaver',
                'beaver lip',
                'beaver lips',
                'bestial',
                'bestiality',
                'bi curiou',
                'bi curious',
                'big black',
                'big breasts',
                'big knocker',
                'big knockers',
                'big tit',
                'big tits',
                'bimbo',
                'birdlock',
                'bitch',
                'bitches',
                'black cock',
                'blond action',
                'blond on blond',
                'blonde action',
                'blow j',
                'blow job',
                'blowjob',
                'blow my',
                'blow me',
                'blow ourselv',
                'blow ourselves',
                'blow your load',
                'blue waffl',
                'blue waffle',
                'blumpkin',
                'bollock',
                'bollocks',
                'bondag',
                'bondage',
                'boner',
                'boob',
                'boobs',
                'booti call',
                'booty call',
                'breast',
                'breasts',
                'brown shower',
                'brown showers',
                'brunett action',
                'brunette action',
                'bukkak',
                'bukkake',
                'bulldyk',
                'bulldyke',
                'bullet vibe',
                'bullshit',
                'bung hole',
                'bunghol',
                'bunghole',
                'busti',
                'busty',
                'butt',
                'buttcheek',
                'buttcheeks',
                'butthol',
                'butthole',
                'camel toe',
                'camgirl',
                'camslut',
                'camwhore',
                'carpet muncher',
                'carpetmuncher',
                'chocol rosebud',
                'chocolate rosebuds',
                'circlejerk',
                'chink',
                'cleveland steamer',
                'clit',
                'clitori',
                'clitoris',
                'clover clamp',
                'clover clamps',
                'clusterfuck',
                'cock',
                'cocks',
                'coprolagnia',
                'coprophilia',
                'cornhol',
                'cornhole',
                'cream pie',
                'creampi',
                'creampie',
                'cum',
                'cumming',
                'cunnilingu',
                'cunnilingus',
                'cunt',
                'damn',
                'darki',
                'darkie',
                'date rape',
                'daterap',
                'daterape',
                'deep throat',
                'deepthroat',
                'dick',
                'dildo',
                'dirti pillow',
                'dirti sanchez',
                'dirty pillow',
                'dirty sanchez',
                'dog style',
                'doggi style',
                'doggie style',
                'doggiestyl',
                'doggiestyle',
                'doggystyle',
                'dolcett',
                'domination',
                'dominatrix',
                'domm',
                'dommes',
                'donkey punch',
                'doubl dick',
                'doubl dong',
                'doubl penetr',
                'double dick',
                'double dong',
                'double penetration',
                'dp action',
                'dtf',
                'eat my ass',
                'ecchi',
                'ejacul',
                'erection',
                'erotic',
                'erotism',
                'escort',
                'ethical slut',
                'eunuch',
                'faggot',
                'fecal',
                'felch',
                'fellatio',
                'feltch',
                'female squirting',
                'femdom',
                'figging',
                'fingered',
                'fingering',
                'fingers',
                'fisted',
                'fisting',
                'fists',
                'foot fetish',
                'footjob',
                'frotting',
                'fuck button',
                'fuck',
                'fucked',
                'fucker',
                'fuckhead',
                'fuckin',
                'fucking',
                'fudge packer',
                'fudgepack',
                'fudgepacker',
                'futanari',
                'g spot',
                'g-spot',
                'gang bang',
                'gay sex',
                'gee spot',
                'genital',
                'giant cock',
                'girl gone wild',
                'girl on top',
                'girl on',
                'goatcx',
                'goatse',
                'goddamn',
                'gokkun',
                'golden shower',
                'goo girl',
                'goodpoop',
                'goregasm',
                'grope',
                'group sex',
                'gspot',
                'guro',
                'hand job',
                'handjob',
                'hard core',
                'hardcore',
                'hentai',
                'homoerotic',
                'honkey',
                'hooker',
                'horni',
                'horny',
                'hot chick',
                'how to kill',
                'how to murder',
                'huge fat',
                'humped',
                'humping',
                'humps',
                'incest',
                'intercourse',
                'jack off',
                'jail bait',
                'jailbait',
                'jerk off',
                'jigaboo',
                'jiggaboo',
                'jiggerboo',
                'jizz',
                'jugg',
                'kike',
                'kinbaku',
                'kinkster',
                'kinky',
                'knobbing',
                'leather restraint',
                'lemon party',
                'lolita',
                'lovemaking',
                'make me come',
                'male squirting',
                'masturb',
                'menage a trois',
                'milf',
                'missionary position',
                'motherfuck',
                'mound of venus',
                'mr hand',
                'muff diver',
                'muffdiv',
                'muffdiving',
                'nambla',
                'nawashi',
                'negro',
                'neonazi',
                'nig nog',
                'nigga',
                'nigger',
                'nimphomania',
                'nipple',
                'nip',
                'not safe for',
                'nsfl',
                'nsfw',
                'nude',
                'nudes',
                'nudity',
                'nut sack',
                'nutsack',
                'nympho',
                'nymphomania',
                'octopussy',
                'omorashi',
                'one night stand',
                'orgasm',
                'orgy',
                'paedophil',
                'paedophile',
                'panties',
                'panty',
                'pedobear',
                'pedophil',
                'pedophile',
                'pee',
                'pegging',
                'peni',
                'penis',
                'phone sex',
                'pigfucker',
                'piss pig',
                'piss',
                'pissing',
                'pisspig',
                'playboy',
                'pleasure chest',
                'pole smoker',
                'ponyplay',
                'poof',
                'poop chute',
                'poopchute',
                'porn',
                'pron',
                'pornhub',
                'porno',
                'pornographi',
                'pornography',
                'prince albert',
                'pthc',
                'pube',
                'pussi',
                'pussies',
                'pussy',
                'queaf',
                'queer',
                'raghead',
                'raging boner',
                'rape',
                'raping',
                'rapist',
                'rectum',
                'reverse cowgirl',
                'rimjob',
                'rimming',
                'rosy palm',
                'rusty trombone',
                's & m',
                's&m',
                's+m',
                'sadism',
                'scat',
                'schlong',
                'scissoring',
                'semen',
                'sex',
                'sexi',
                'sexo',
                'sexy',
                'shaved beaver',
                'shaved pussy',
                'shemale',
                'shibari',
                'shit',
                'shota',
                'shrimping',
                'slanteye',
                'slut',
                'smut',
                'snatch',
                'snm',
                'snowballing',
                'sodomi',
                'sodomize',
                'sodomy',
                'spic',
                'spooge',
                'spread legs',
                'squirting',
                'strap on',
                'strapon',
                'strappado',
                'strip club',
                'style doggy',
                'suck',
                'suicide girls',
                'sultry women',
                'swastika',
                'swinger',
                'taint',
                'tainted love',
                'taste my',
                'tea bagging',
                'threesome',
                'throating',
                'tied up',
                'tight white',
                'tit',
                'tits',
                'titti',
                'titties',
                'titty',
                'tongue in',
                'topless',
                'tosser',
                'towelhead',
                'tranny',
                'tribadism',
                'tub girl',
                'tubgirl',
                'tushy',
                'twat',
                'twink',
                'twinki',
                'twinkie',
                'undress',
                'upskirt',
                'urethra play',
                'urophilia',
                'vag',
                'vagina',
                'venus mound',
                'vibrator',
                'violet blue',
                'violet wand',
                'vorarephilia',
                'voyeur',
                'vulva',
                'wank',
                'wet dream',
                'wetback',
                'white power',
                'whore',
                'women rapping',
                'wrapping men',
                'wrinkled starfish',
                'xx',
                'xxx',
                'yaoi',
                'yellow shower',
                'yiffy',
                'zoophilia',
            ]
        )
        self.stemmer = PorterStemmer()

    def _normalize(self, text: str) -> str:
        """Normalize text.

        >>> _normalize('Tittie5')
        'titties'

        >>> _normalize('Suck a Dick!')
        'suck a dick'

        >>> _normalize('fucking a whore')
        'fuck a whore'

        """
        result = text.lower()
        result = result.replace("_", " ")
        result = result.replace('0', 'o')
        result = result.replace('1', 'l')
        result = result.replace('4', 'a')
        result = result.replace('5', 's')
        result = result.replace('3', 'e')
        for x in string.punctuation:
            result = result.replace(x, "")
        chunks = [self.stemmer.stem(word) for word in nltk.word_tokenize(result)]
        return ' '.join(chunks)

    def tokenize(self, text: str):
        for x in nltk.word_tokenize(text):
            for y in re.split('\W+', x):
                yield y

    def contains_bad_word(self, text: str) -> bool:
        """Returns True if text contains a bad word (or more than one)
        and False if no bad words were detected.

        >>> contains_bad_word('fuck you')
        True

        >>> contains_bad_word('FucK u')
        True

        >>> contains_bad_word('FuK U')
        False

        """
        words = [word for word in self.tokenize(text)]
        for word in words:
            if self.is_bad_word(word):
                logger.debug(f'"{word}" is profanity')
                return True

        if len(words) > 1:
            for bigram in string_utils.ngrams_presplit(words, 2):
                bigram = ' '.join(bigram)
                if self.is_bad_word(bigram):
                    logger.debug(f'"{bigram}" is profanity')
                    return True

        if len(words) > 2:
            for trigram in string_utils.ngrams_presplit(words, 3):
                trigram = ' '.join(trigram)
                if self.is_bad_word(trigram):
                    logger.debug(f'"{trigram}" is profanity')
                    return True
        return False

    def is_bad_word(self, word: str) -> bool:
        return word in self.bad_words or self._normalize(word) in self.bad_words

    def obscure_bad_words(self, text: str) -> str:
        """Obscure bad words that are detected by inserting random punctuation
        characters.

        """

        def obscure(word: str):
            out = ''
            last = ''
            for letter in word:
                if letter.isspace():
                    out += letter
                else:
                    while True:
                        char = random.choice(['#', '%', '!', '@', '&', '*'])
                        if last != char:
                            last = char
                            out += char
                            break
            return out

        words = self.tokenize(text)
        words.append('')
        words.append('')
        words.append('')
        out = ''

        cursor = 0
        while cursor < len(words) - 3:
            word = words[cursor]
            bigram = word + ' ' + words[cursor + 1]
            trigram = bigram + ' ' + words[cursor + 2]
            if self.is_bad_word(trigram):
                out += obscure(trigram) + ' '
                cursor += 3
            elif self.is_bad_word(bigram):
                out += obscure(bigram) + ' '
                cursor += 2
            elif self.is_bad_word(word):
                out += obscure(word) + ' '
                cursor += 1
            else:
                out += word + ' '
                cursor += 1
        return out.strip()


def main() -> None:
    import doctest

    doctest.testmod()
    pf = ProfanityFilter()
    phrase = ' '.join(sys.argv[1:])
    print(pf.contains_bad_word(phrase))
    print(pf.obscure_bad_words(phrase))
    sys.exit(0)


if __name__ == '__main__':
    main()