def setUp(self):
self.ucd = UCD()
class UCDTests(unittest.TestCase):
def setUp(self):
self.ucd = UCD()
def tearDown(self):
pass
def ignore_findit(self):
from icu import Char, UProperty
maxchar = 0x10ffff
maxchar = 0xffff
for usv in range(maxchar):
char = chr(usv)
# if ((not self.ucd.is_specific_script(char)) and
# (not self.ucd.is_exemplar_wordbreak(char)) and
# (not Char.isUAlphabetic(char))):
if self.ucd.isformat(char) and not Char.hasBinaryProperty(
char, UProperty.DEFAULT_IGNORABLE_CODE_POINT):
print('%04X' % usv)
self.assertTrue(False)
# marks
def test_mark_true(self):
self.assertTrue(self.ucd.ismark(u'\u0301'))
def test_mark_false(self):
self.assertFalse(self.ucd.ismark(u'e'))
def test_nukta_true(self):
self.assertTrue(self.ucd.isnukta(u'\u093c'))
def test_nukta_false(self):
self.assertFalse(self.ucd.isnukta(u'\u0915'))
# always_combine
def test_nukta_always_combine(self):
self.assertTrue(self.ucd.is_always_combine(u'\u093c'))
def test_diacritic_always_combine(self):
self.assertFalse(self.ucd.is_always_combine(u'\u0300'))
def test_virama_always_combine(self):
self.assertFalse(self.ucd.is_always_combine(u'\u0ccd'))
def test_matra_always_combine(self):
self.assertFalse(self.ucd.is_always_combine(u'\u093e'))
# sometimes_combine
def test_nukta_sometimes_combine(self):
self.assertFalse(self.ucd.is_sometimes_combine(u'\u093c'))
def test_diacritic_sometimes_combine(self):
self.assertTrue(self.ucd.is_sometimes_combine(u'\u0300'))
def test_virama_sometimes_combine(self):
self.assertFalse(self.ucd.is_sometimes_combine(u'\u0ccd'))
def test_matra_sometimes_combine(self):
self.assertFalse(self.ucd.is_sometimes_combine(u'\u093e'))
# never_combine
def test_nukta_never_combine(self):
self.assertFalse(self.ucd.is_never_combine(u'\u093c'))
def test_diacritic_never_combine(self):
self.assertFalse(self.ucd.is_never_combine(u'\u0300'))
def test_virama_never_combine(self):
self.assertTrue(self.ucd.is_never_combine(u'\u0ccd'))
def test_matra_never_combine(self):
self.assertTrue(self.ucd.is_never_combine(u'\u093e'))
# other tests
def test_zwnj_true(self):
self.assertTrue(self.ucd.is_zwnj(u'\u200c'))
def test_zwnj_false(self):
self.assertFalse(self.ucd.is_zwnj(u'\u200d'))
def test_zwj_true(self):
self.assertTrue(self.ucd.is_zwj(u'\u200d'))
def test_zwj_false(self):
self.assertFalse(self.ucd.is_zwj(u'\u200c'))
def test_vs_true(self):
self.assertTrue(self.ucd.is_vs(u'\ufe00'))
def test_vs_false(self):
self.assertFalse(self.ucd.is_vs(u'\u1000'))
def test_number_true(self):
self.assertTrue(self.ucd.isnumber(u'1'))
def test_number_false(self):
self.assertFalse(self.ucd.isnumber(u'a'))
def test_format_true(self):
self.assertTrue(self.ucd.isformat(u'\u2060'))
def test_format_false(self):
self.assertFalse(self.ucd.isformat(u'a'))
def test_space_separator_true(self):
self.assertTrue(self.ucd.is_space_separator(u'\u200a'))
def test_space_separator_false(self):
self.assertFalse(self.ucd.is_space_separator(u'a'))
def test_pua_false_bmp(self):
self.assertFalse(self.ucd.is_pua(u'a'))
def test_pua_true_bmp(self):
self.assertTrue(self.ucd.is_pua(u'\ue000'))
def test_pua_false_nonbmp(self):
self.assertFalse(self.ucd.is_pua(u'\U0001D510'))
def test_pua_true_nonbmp_a(self):
self.assertTrue(self.ucd.is_pua(u'\U000fff80'))
def test_pua_true_nonbmp_b(self):
self.assertTrue(self.ucd.is_pua(u'\U000fff80'))
def test_script_specific_true_latin(self):
self.assertTrue(self.ucd.is_specific_script(u'\ua78c'))
def test_script_specific_false_latin(self):
self.assertFalse(self.ucd.is_specific_script(u'\u02bc'))
def test_script_specific_false_chinese(self):
self.assertFalse(self.ucd.is_specific_script(u'\ua700'))
def test_script_specific_false_vedic(self):
self.assertFalse(self.ucd.is_specific_script(u'\u1CD1'))
def test_wordbreak_katakana(self):
self.assertTrue(self.ucd.is_exemplar_wordbreak(u'\u309b'))
def test_wordbreak_aletter(self):
self.assertTrue(self.ucd.is_exemplar_wordbreak(u'\u05f3'))
def test_wordbreak_midletter(self):
self.assertFalse(self.ucd.is_exemplar_wordbreak(u'\u05f4'))
def test_wordbreak_chinese(self):
self.assertFalse(self.ucd.is_exemplar_wordbreak(u'\ua700'))
def test_nfc(self):
text = u'e\u0301'
self.assertEqual(u'\u00e9', self.ucd.normalize('NFC', text))
def test_nfd(self):
text = u'\u00e9'
self.assertEqual(u'e\u0301', self.ucd.normalize('NFD', text))
def test_nfc_tus10(self):
text = u'\u0061\u035C\u0315\u0300\u1DF6\u0062'
self.assertEqual(u'\u00E0\u0315\u1DF6\u035C\u0062',
self.ucd.normalize('NFC', text))
def test_nfd_tus10(self):
text = u'\u0061\u035C\u0315\u0300\u1DF6\u0062'
self.assertEqual(u'\u0061\u0300\u0315\u1DF6\u035C\u0062',
self.ucd.normalize('NFD', text))
def ignore_nfc_tus11(self):
text = u'\u0061\u0315\u0300\u05AE\u09FE\u0062'
self.assertEqual(u'\u00E0\u05AE\u09FE\u0315\u0062',
self.ucd.normalize('NFC', text))
def ignore_nfd_tus11(self):
text = u'\u0061\u0315\u0300\u05AE\u09FE\u0062'
self.assertEqual(u'\u0061\u05AE\u0300\u09FE\u0315\u0062',
self.ucd.normalize('NFD', text))
def ignore_nfc_tus12(self):
text = u'\u0061\u0315\u0300\u05AE\U0001E136\u0062'
self.assertEqual(u'\u00E0\u05AE\U0001E136\u0315\u0062',
self.ucd.normalize('NFC', text))
def ignore_nfd_tus12(self):
text = u'\u0061\u0315\u0300\u05AE\U0001E136\u0062'
self.assertEqual(u'\u0061\u05AE\u0300\U0001E136\u0315\u0062',
self.ucd.normalize('NFD', text))
def ignore_nfc_tus13(self):
text = u'\u0061\u3099\u093C\U00016FF0\u09BC\u0062'
self.assertEqual(u'\u0061\U00016FF0\u093C\u09BC\u3099\u0062',
tf.normalize('NFC', text))
def ignore_nfd_tus13(self):
text = u'\u0061\u3099\u093C\U00016FF0\u09BC\u0062'
self.assertEqual(u'\u0061\U00016FF0\u093C\u09BC\u3099\u0062',
tf.normalize('NFD', text))