def test_confusable_regex__basic_ascii_regex_with_padding(self):
regex = confusable_regex('bore', include_character_padding=True)
reg = re.compile(regex)
self.assertTrue(
reg.search(
'Sometimes people say that life can be a ь.𝞂.ř.ɜ, but I don\'t agree'
))
def check_if_message_infringes(self, message):
word_boundary_chars = '(?:^|$|\\s|\\b)'
pattern = word_boundary_chars + confusable_regex(
self.banned_word, True) + word_boundary_chars
if re.search(pattern, message) is not None:
return True
return False
def test_confusable_regex__dont_treat_pipe_as_wildcard(self):
regex = confusable_regex('bore')
reg = re.compile(regex)
self.assertFalse(
reg.search(
'Sometimes people say that life can be a ||||, but I don\'t agree'
))
def test_confusable_regex__match_subwords(self):
regex = confusable_regex('bore', match_subword=True)
reg = re.compile(regex)
self.assertTrue(
reg.search(
'Sometimes people say that life can be a ь𝞂řɜd, but I don\'t agree'
))
self.assertTrue(
reg.search(
'Sometimes people say that life can be a ь𝞂řɜ, but I don\'t agree'
))
def test_confusable_regex__basic_ascii_regex_without_padding(self):
regex = confusable_regex('bore')
reg = re.compile(regex)
self.assertFalse(
reg.search(
'Sometimes people say that life can be a ь.𝞂.ř.ɜ, but I don\'t agree'
))
self.assertTrue(
reg.search(
'Sometimes people say that life can be a ь𝞂řɜ, but I don\'t agree'
))
def test_confusable_regex__match_multi_character_confusion(self):
regex = confusable_regex('‷')
reg = re.compile(regex)
self.assertFalse(
reg.search(
'Sometimes people say that life can be \' , but I don\'t agree'
))
self.assertTrue(
reg.search(
'Sometimes people say that life can be \'\'\' , but I don\'t agree'
))
def msg_contains_forbidden(msg: str, forbidden_words: List[str]) -> bool:
"""Returns True if msg contains a word in the forbidden_words list or
is confusable with any words in forbidden_words
"""
contains_forbidden = False
for word in forbidden_words:
regex_string = confusable_regex(word, include_character_padding=True)
regex = re.compile(regex_string)
contains_forbidden = regex.search(msg)
if contains_forbidden:
return contains_forbidden
return contains_forbidden
def confusables(data, line):
buf_ptr = line['buffer']
message = line['message']
tags = line['tags']
prefix = line['prefix']
if OPTIONS['list_of_words'] == "": # no words given, to look at
return weechat.WEECHAT_RC_OK
search_strings = OPTIONS['list_of_words'].split(',')
for i in search_strings:
regex_string = confusable_regex(i, include_character_padding=True)
regex = re.compile(regex_string)
conf_result = regex.search(
message) # get match to test with original string later
if regex.search(message) and conf_result.group().lower() != i.lower(
): # matching string is original string?
return {"tags": tags + ',' + OPTIONS['tags']}
# weechat.prnt_date_tags(buf_ptr,0,tags + ',' + OPTIONS['tags'],message)
return weechat.WEECHAT_RC_OK
def detect_banned_word(message, banned_word):
word_boundary_chars = '\\b'
pattern = word_boundary_chars + confusable_regex(banned_word, True) + word_boundary_chars
if re.search(pattern, message) is not None:
return True
return False
def test_confusable_regex__regex_does_not_match_if_only_subset_of_word(
self):
regex = confusable_regex('bore')
reg = re.compile(regex)
self.assertFalse(reg.search('Hopefully you don\'t get bored easily'))
def test_confusable_regex__regex_special_characters_are_escaped(self):
regex = confusable_regex('e|mo')
reg = re.compile(regex)
self.assertTrue(reg.search('elmo'))
self.assertFalse(reg.search('emo'))
