Add formating before check for bad word, sort bad words.

antimat.py

@@ -11,9 +11,9 @@ global db
 ####### CREATE DB IF NOT EXIST
 
 if not os.path.exists('db.json'):
-	db = {"token": "None"}
+	db = {'token': 'None'}
 	js = json.dumps(db, indent=2)
-	with open("db.json", "w") as outfile:
+	with open('db.json', 'w') as outfile:
 		outfile.write(js)
 
 	print('Input token in "None" (db.json)')
@@ -28,7 +28,7 @@ def read_db():
 def write_db():
         global db
         js = json.dumps(db, indent=2)
-        with open("db.json", "w") as outfile:
+        with open('db.json', 'w') as outfile:
                 outfile.write(js)
 
 #############TOKEN INIT#####
@@ -39,10 +39,15 @@ bot = telebot.TeleBot(db['token'])
 
 ########## ALANYS WORD #####
 
-from re import search, match
+from re import search, match, sub, compile
 def is_bad(word, bad):
 	# Agressive mode, causes false positives
 	#res = search(bad, word)
+
+	# -М#ат$ => Мат
+	regex = compile('[^a-zA-Zа-яА-Я]')
+	word = regex.sub('', word)
+
 	res = match(bad, word)
 	if res is not None:
 		return True
@@ -53,8 +58,14 @@ def is_bad(word, bad):
 
 
 def catch(message):
-	bad_list = ["бл.+ть","еб.+ть","пиз.+ец","х.+й","бл+.дь","п.+зда","бля","f.+ck","блят","s.+ck","ебу", "ебал",
-	"трах","eby"]
+	bad_list = ['х.+й', 'xуи', 'xyи', 'хyи', 'xyй', 'xуй', 'ху.', '.+хуё', 'xyu', 'xui', 'хyй', 'хй', 'поху',
+	'бл.+ть', 'бля', 'блят',
+	'бл+.дь',
+	'трах', 'еб.+ть', 'ебу', 'ебал', '..ебен', 'ебан', 'ёбан', 'ебть', 'eby', '..ебись', 'уеб', 'уёб', 'ебей',
+	'п.+зда', 'пиз.+ец', 'пизд', 'пизец', 'пздец', 'п.+здец', 'пизд',
+	'пид.+р', 'пидр',
+	'f.+ck', 's.+ck', 'fck', 'sck']
+
 	words = message.text.split()
 
 	bad_found = False
@@ -69,7 +80,7 @@ def catch(message):
 
 	if bad_found:
 		bot.delete_message(message.chat.id, message.id)
-		bot.send_message(message.chat.id, f"Пользователь {telebot.util.user_link(message.from_user)} использовал непечатное выражение.", parse_mode="HTML")
+		bot.send_message(message.chat.id, f'Пользователь {telebot.util.user_link(message.from_user)} использовал непечатное выражение.', parse_mode='HTML')
 
 
 @bot.message_handler()
@@ -80,7 +91,7 @@ def catch_edited_messages(message):
 	catch(message)
 
 
-'''
+#'''
 while True:
 	try:
 		bot.polling()
@@ -90,3 +101,4 @@ while True:
 		pass
 '''
 bot.polling()
+'''