badwords-py 2.1.0__py3-none-any.whl

badwords/__init__.py

@@ -0,0 +1,5 @@
+"""A library for effective moderation of content."""
+
+from .check import ProfanityFilter
+
+__all__ = ["ProfanityFilter"]

badwords/check.py

@@ -0,0 +1,148 @@
+"""Module for checking text for badwords."""
+
+from __future__ import annotations
+
+from difflib import SequenceMatcher
+from pathlib import Path
+from typing import TYPE_CHECKING
+
+from .exceptions import NotSupportedLanguage
+from .text_processor import TextProcessor
+
+if TYPE_CHECKING:
+    from typing import Self
+else:
+    Self = "NotSupportedLanguage"
+
+
+class ProfanityFilter:
+    """A class for filtering profanity from text."""
+
+    def init(
+        self: Self,
+        languages: list[str] | None = None,
+        processing_normalize_text: bool = True,
+        processing_aggressive_normalize: bool = True,
+        processing_transliterate: bool = True,
+        processing_replace_homoglyphs: bool = True,
+    ) -> None:
+        """Initialize the profanity filter.
+
+        :param languages: List of languages to load profanity words for.
+        :param processing_normalize_text: Enable text normalization for all loaded words and word to filter
+        :param processing_aggressive_normalize: Enable aggressive text normalization for all loaded words and word to filter
+        :param processing_transliterate: Enable transliteration of text for all loaded words and word to filter
+        :param processing_replace_homoglyphs: Enable replacing of homoglyphs in the text for all loaded words and word to filter
+        """
+        self.resource_dir = Path(__file__).parent / "resource"
+
+        self.text_processor = TextProcessor(
+            processing_normalize_text=processing_normalize_text,
+            processing_aggressive_normalize=processing_aggressive_normalize,
+            processing_transliterate=processing_transliterate,
+            processing_replace_homoglyphs=processing_replace_homoglyphs
+        )
+
+        self.language_files = self.initialize_language_files()
+
+        if languages:
+            if all(i in self.language_files for i in languages):
+                self.language_files = languages
+            else:
+                raise NotSupportedLanguage
+
+        self.bad_words = self.initialize_bad_words()
+
+    def initialize_language_files(self: Self) -> list[str]:
+        """Initialize language files.
+
+        :return: Dictionary mapping language names to file paths.
+        """
+        return [str(path)[-6:-4] for path in (self.resource_dir).iterdir()]
+
+    def initialize_bad_words(self: Self) -> set[str]:
+        """Initialize the set of bad words from language files."""
+        bad_words: set[str] = set()
+        
+        for lang in self.language_files:
+            try:
+                # Sanitize the language code to prevent path traversal
+                lang = lang.lower().strip()
+                if not lang.isalpha():
+                    continue
+                    
+                file_path = self.resource_dir / f"{lang}.bdw"
+                if not file_path.exists():
+                    continue
+                    
+                with file_path.open(encoding="utf-8") as f:
+                    words = f.read().splitlines()
+                    processed_words = [self.text_processor.process_text(word) for word in words]
+                    bad_words.update(processed_words)
+            except Exception as e:
+                print(f"Error loading language file for {lang}: {e}")
+                continue
+                
+        return bad_words
+
+    def add_words(self: Self, words: list[str]) -> None:
+        """Add custom profanity words to the filter.
+
+        :param words: List of custom profanity words.
+        """
+        processed_words = [self.text_processor.process_text(word) for word in words]
+        self.bad_words.update(processed_words)
+
+    def similar(self: Self, a: str, b: str) -> float:
+        """Compute similarity ratio between two strings.
+
+        :param a: First string.
+        :param b: Second string.
+        :return: Similarity ratio.
+        """
+        return SequenceMatcher(None, a, b).ratio()
+
+    def filter_text(
+            self: Self, text: str,
+            match_threshold: float | None = None,
+            replace_character: str | None = None,
+        ) -> bool | str:
+        """Check if the given text contains profanity.
+
+        :param text: Input text to check.
+        :param match_threshold: Threshold for similarity match.
+        :param replace_character: Character to replace profane words with. If None,
+            return True/False.
+        :return: True if profanity found, False otherwise. If replace_character is
+            specified, return filtered text.
+        """
+        if not match_threshold:
+            match_threshold = 1
+
+        # Process the input text through all transformations
+        processed_text = self.text_processor.process_text(text)
+        words = processed_text.split()
+
+        for word in words:
+            # Check exact match
+            if word in self.bad_words:
+                if replace_character:
+                    return text.replace(word, replace_character * len(word))
+                return True
+
+            # Check similar matches if threshold is less than 1
+            if 0 < match_threshold < 1:
+                for bad_word in self.bad_words:
+                    if self.similar(word, bad_word) > match_threshold:
+                        if replace_character:
+                            return text.replace(word, replace_character * len(word))
+                        return True
+
+        return False
+
+    def get_all_languages(self: Self) -> list[str]:
+        """Get a list of all available languages.
+
+        :return: List of all language names.
+        """
+        return self.language_files

badwords/exceptions.py

@@ -0,0 +1,16 @@
+"""Exceptions module."""
+
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from typing import Self
+else:
+    Self = "NotSupportedLanguage"
+
+
+class NotSupportedLanguage(BaseException):
+    """Unsupport language check."""
+
+    def __str__(self: Self) -> str:
+        """String-like representation of exception."""
+        return "This language is not supported"

badwords/resource/br.bdw

@@ -0,0 +1,119 @@
+foderíamos
+fodêssemos
+foderíeis
+fodêramos
+fodêsseis
+merdimbuca
+putariinha
+chibundas
+chibundos
+cácété
+foderemos
+fodêreis
+fodíamos
+fudeção
+putariona
+boasudas
+boazudas
+cassetas
+chibunda
+chibundo
+culhões
+cunhões
+foderdes
+fodereis
+foderiam
+foderias
+fodermos
+foderás
+foderão
+fodessem
+fodesses
+fodestes
+fodíeis
+fudecão
+fudeçao
+fudidas?
+fudidos?
+peithola
+putarias
+xibundas
+xibundos
+babacas
+boasuda
+boazuda
+cacetas
+cacetes
+casseta
+cassete
+culhoes
+culhão
+cunhoes
+cunhão
+curalho
+fodamos
+fodasse
+fodemos
+fodendo
+foderam
+foderas
+foderei
+foderem
+foderes
+foderia
+foderum
+foderá
+fodesse
+fodeste
+fodinha
+fudecao
+fudedor
+fudendo
+furonas
+putaria
+putães
+putãos
+putões
+xibunda
+xibundo
+babaca
+bostas
+caceta
+cacete
+culhao
+cunhao
+fodais
+fodeis
+fodera
+fodete
+fodiam
+fodias
+fodão
+furona
+furão
+merdas
+putão
+sefoda
+bosta
+fodam
+fodao
+fodas
+fodei
+fodem
+foder
+fodes
+fodeu
+fodia
+fuder
+fudeu
+furao
+merda
+porra
+putos
+foda
+fode
+fodi
+fodo
+puto
+cus
+cu

badwords/resource/character_frequency.json

@@ -0,0 +1,12 @@
+{
+    "a": ["4", "@", "а"],
+    "b": ["8", "6", "в"],
+    "e": ["3", "ё", "е"],
+    "g": ["9", "6"],
+    "i": ["1", "!", "і"],
+    "l": ["1", "|", "!"],
+    "o": ["0", "о"],
+    "s": ["5", "$", "ѕ"],
+    "t": ["7", "+"],
+    "z": ["2", "z"]
+} 

badwords/resource/cz.bdw

@@ -0,0 +1,51 @@
+pokurvená
+pokurvení
+pokurvený
+pomočená
+pomočený
+zkurvysyni
+pochcaná
+pochcaný
+pochčije
+prasárna
+prasárny
+zkurvená
+zkurvení
+zkurvený
+zkurvysyn
+chcánky
+hovadina
+hovadiny
+kadění
+kokotina
+kokotiny
+pokurvit
+pokurví
+prdění
+chcaní
+chčije
+pochcat
+sračka
+sračky
+zkurvil
+zkurvit
+zkurví
+chcát
+mamrdi
+prdele
+prdět
+řitě
+hovna
+hovno
+kadit
+kadí
+kruci
+mamrd
+prdel
+sakra
+satan
+srát
+řiť
+prdy
+sere
+prd

badwords/resource/da.bdw

@@ -0,0 +1,60 @@
+åndsforsnottet
+skidespræller
+skvadderhoved
+pestspreder
+skvatpisser
+spytslikker
+torskepande
+undermåler
+åndsamøbe
+danglebær
+dinglebær
+lortefjæs
+lusepuster
+pattebørn
+pladderabe
+ringlebær
+satanedeme
+sjatpisser
+tøsedreng
+åndsbolle
+ærkefjols
+øgleyngel
+agurketud
+forpulede
+kvabodder
+kvajhoved
+kvajpande
+lorteøre
+pattebarn
+slapsvans
+forpulet
+nakkeost
+narrehat
+pikfjæs
+pikhoved
+skiderik
+abelort
+fandeme
+fåking
+fåkker
+møgdyr
+narrøv
+urinere
+bovlam
+fanden
+fandme
+focker
+narhat
+satan
+satme
+skide
+skvat
+fock
+fåk
+lort
+pjok
+skid
+svin
+urin
+sgu

badwords/resource/de.bdw

@@ -0,0 +1,62 @@
+leckmich[ai]m[aä]rsche?
+huren?s[oö]hne?
+schlappschwänze
+geht?sterben
+schwanzlutscher
+muschileckerin
+pferdescheiße
+schei(ss|ß)e?
+schlappschwanz
+drecksbälger
+mutterfick\w+
+sackgesichter
+schwanzköpfe
+wichsvorlagen
+angeschissen
+dreckstücke
+fickteuch
+gottverdammt
+kackbratzen?
+muschilecker
+wichsvorlage
+dreckskerle
+dreckstück
+hackfressen
+ihraffen
+sackgesicht
+saftärsche
+scheißkerl
+schwanzkopf
+verfickt
+drecksbalg
+dreckskerl
+drecksäue
+geschissen
+hackfresse
+kackfresse
+schlampen?
+arschloch
+drecksack
+huren?
+saftarsch
+schnepfen
+arsch
+drecksau
+fickdich
+sauhunde
+saukerle
+schnepfe
+brunzen
+fettsau
+sauhund
+saukerl
+seichen
+wichser
+kacke?
+poppen
+schiss
+husos
+pisse
+pisst
+huso
+nutte

badwords/resource/du.bdw

@@ -0,0 +1,39 @@
+opsodemieteren
+hoerententen
+godverdomme
+hoeretenten
+volkomenkut
+hoerentent
+opdonderen
+schijterts
+hoeretent
+rotzakken
+schijterd
+schijters
+schijtert
+slappelul
+verkloten
+goddomme
+godsamme
+kankeren
+oprotten
+schijten
+schijter
+verdomme
+verkloot
+pleuris
+rukkers
+schurft
+verdomd
+godver
+rotzak
+rukker
+scheit
+schijt
+stront
+tering
+zeiken
+klote
+tyfus
+gvd
+kut

badwords/resource/en.bdw

@@ -0,0 +1,66 @@
+(ape|bat|bull?|butt|dip|dog|dumb|ebo|holy|horse|jack|pedo|pig|ubi)sh(it|ti)s?
+d[il]+(ck|kc)(ass|bag|breath|eat|face|flip|head|hole|less|suck|weed)
+f[chj]?(a|au|aw|e|o+|u|uy)[ch]*k(e?d|e?rs?|[ei]?n+g?|t)
+m[aou](d+|t+|th|ht|z)(a|e|er|ir|ur)(c|f)c?[aou]c?k
+m[ou]th(a|er)(f|ph)[vue]+c?[gkqx]+(e?d|e?r|[ei]?n+g?)
+f[vu]+h*c+[hjvk]*(e?d|e?r|[ei]?n+g?|t)
+b+i+a*t+c+h+(e+d|e*r?[sz]+|[ei]?n+g?|y)
+f[vu]+c?[xkq]+(e?d|e?r|[ei]?n+g?|t)
+p(oo+|u+)s+(a+y+|e+|eh|ey|i+|ie|y+)s*
+m[ou]th(a|er)(f|ph)[vue]+c?[gkqx]+
+f[vu]+h*q(e?d|e?r|[ei]?n+g?|t)
+f[chj]?(a|au|aw|e|o+|u|uy)[ch]*k
+f[ck]+(e?d|e?r|[ei]?n+g?|t)
+kill your self
+fu󠀡󠀡cki󠀡󠀡ng
+dumb(f|ph)[vu]+c?[xkq]
+c(oc?|aw)ksuc?k
+d[il]+(ck|kc)(ed|ing)
+ph[vu]+h*c+[hjvk]*
+sonofabitch
+who+re(ed|ing|s)
+c(o|ah|aw)c?k(s|ed)
+s+h+[ily]+t+[ersy]*
+a+ss+w?hole?
+ph[vu]+c?[xkq]+
+f[vu]+c?[xkq]+
+bull+sh+[ily]+t+
+goddamn?it
+goddamn?
+a+ss+fu
+c(o|ah|aw)c?k
+d[il]+(ck|kc)
+b+i+a*t+c+h+
+buttfu
+cumm(er|ing)
+f[vu]+h*c
+f[vu]+h*q
+sh+i+[ae]+t+
+c+u+n+t+
+d+i+c+k+s*
+a+sholes?
+sh+e+i+t+
+F\.U\.C\.K
+chodes?
+cumshot
+fucking
+shit
+shtpost
+who+re?
+cumbag
+twats?
+cums?
+dafuq
+diсk
+coq+
+hore
+kock
+shat
+dik
+fck
+negr[ao]+e?s?
+f+u+c+k+
+fцск
+shіtty
+f_uc_k
+pussy

badwords/resource/fi.bdw

@@ -0,0 +1,57 @@
+perkeleeseen
+perkeleessä
+perkeleestä
+perkeleelle
+perkelettä
+vittupään
+vittupäät
+paskanaama
+vittupää
+jumalauta
+paskattaa
+paskoihin
+perkeleen
+perkeleet
+mulkeron
+mulkerot
+paskalla
+paskalle
+paskalta
+paskassa
+paskasta
+paskojen
+saatanaa
+saatanan
+saatanat
+aashole
+hitolla
+hitossa
+hitosta
+hittoon
+kyrpiä
+mulkero
+paskaan
+paskoja
+perkele
+perkule
+perseen
+pissata
+saatana
+kyrpii
+paskaa
+paskan
+paskat
+paskoi
+paskoo
+paskot
+pissaa
+hiton
+hitot
+hitto
+paska
+perse
+pissa
+prkle
+viddu
+prkl
+prk