def test_get_table(self):
alphabet = Categories.get_alphabet(['LATIN'])
table = Homoglyphs.get_table(alphabet)
self.assertIn('s', table)
self.assertNotIn(CIRILLIC_ES, table)
alphabet = Categories.get_alphabet(['HANGUL', 'COMMON'])
table = Homoglyphs.get_table(alphabet)
self.assertGreater(len(table[u'ㅡ']), 0)
def test_to_ascii(self):
ss = Homoglyphs(strategy=STRATEGY_LOAD).to_ascii(CIRILLIC_HE)
self.assertEqual(ss, ['x'])
ss = Homoglyphs(strategy=STRATEGY_LOAD).to_ascii(CIRILLIC_HE)
self.assertEqual(ss, ['x'])
ss = Homoglyphs(strategy=STRATEGY_LOAD,
ascii_range=range(256)).to_ascii(CIRILLIC_HE + u'23.')
self.assertEqual(ss, ['x23.', u'×23.'])
def test_get_char_variants(self):
variants = Homoglyphs(['LATIN'])._get_char_variants('s')
self.assertIn('s', variants)
self.assertIn(u'ｓ', variants)
self.assertNotIn('S', variants)
self.assertNotIn(CIRILLIC_ES, variants)
variants = Homoglyphs(['LATIN'])._get_char_variants('c')
self.assertIn('c', variants)
self.assertNotIn(CIRILLIC_ES, variants)
variants = Homoglyphs(['LATIN', 'CYRILLIC'])._get_char_variants('c')
self.assertIn('c', variants)
self.assertIn(CIRILLIC_ES, variants)
def test_ascii_strategy(self):
ss = Homoglyphs(
categories=('LATIN', 'COMMON', 'CYRILLIC'),
ascii_strategy=STRATEGY_IGNORE,
).to_ascii(u'xхч2')
self.assertEqual(ss, [])
ss = Homoglyphs(
categories=('LATIN', 'COMMON', 'CYRILLIC'),
ascii_strategy=STRATEGY_REMOVE,
ascii_range=range(256),
).to_ascii(u'xхч2')
self.assertEqual(ss, ['xx2', u'x×2', u'×x2', u'××2'])
ss = Homoglyphs(
strategy=STRATEGY_LOAD,
ascii_strategy=STRATEGY_REMOVE,
).to_ascii(u'ч')
self.assertEqual(ss, [])
def test_strategy(self):
variants = Homoglyphs(
['LATIN'], strategy=STRATEGY_IGNORE)._get_char_variants(u'ё')
self.assertEqual(variants, [u'ё'])
variants = Homoglyphs(
['LATIN'], strategy=STRATEGY_LOAD)._get_char_variants(CIRILLIC_HE)
self.assertGreater(len(variants), 1)
self.assertIn('x', variants)
self.assertIn(CIRILLIC_HE, variants)
variants = Homoglyphs(
['LATIN'],
strategy=STRATEGY_REMOVE)._get_char_variants(CIRILLIC_HE)
self.assertEqual(variants, [])
for s in (STRATEGY_REMOVE, STRATEGY_IGNORE, STRATEGY_LOAD):
variants = Homoglyphs(['LATIN'],
strategy=s)._get_char_variants('d')
self.assertIn('d', variants)
def test_to_ascii(self):
ss = Homoglyphs(strategy=STRATEGY_LOAD).to_ascii(CIRILLIC_HE)
self.assertEqual(ss, ['x'])
ss = Homoglyphs(strategy=STRATEGY_LOAD).to_ascii(CIRILLIC_HE)
self.assertEqual(ss, ['x'])
ss = Homoglyphs(strategy=STRATEGY_LOAD).to_ascii(CIRILLIC_HE + u'23.')
self.assertEqual(ss, ['x23.'])
ss = Homoglyphs(
categories=('LATIN', 'COMMON', 'CYRILLIC'),
ascii_strategy=STRATEGY_IGNORE,
).to_ascii(u'xхч2')
self.assertEqual(ss, [])
ss = Homoglyphs(
categories=('LATIN', 'COMMON', 'CYRILLIC'),
ascii_strategy=STRATEGY_REMOVE,
).to_ascii(u'xхч2')
self.assertEqual(ss, ['xx2'])
def test_to_ascii(self):
ss = Homoglyphs(strategy=STRATEGY_LOAD).to_ascii(CIRILLIC_HE)
self.assertEqual(ss, ['x'])
ss = Homoglyphs(strategy=STRATEGY_LOAD).to_ascii(CIRILLIC_HE)
self.assertEqual(ss, ['x'])
ss = Homoglyphs(strategy=STRATEGY_LOAD,
ascii_range=range(256)).to_ascii(CIRILLIC_HE + u'23.')
self.assertEqual(ss, ['x23.', u'×23.'])
homoglyphs = Homoglyphs(languages={'en'}, strategy=STRATEGY_LOAD)
self.assertEqual(homoglyphs.to_ascii('ТЕСТ'), ['TECT'])
def test_get_table(self):
alphabet = Categories.get_alphabet(['LATIN'])
table = Homoglyphs.get_table(alphabet)
self.assertIn('s', table)
self.assertNotIn(CIRILLIC_ES, table)