def test_parse_begedkefet(self):
"""Expected to parse BeGeDKeFeT letters."""
test = hebrew.parse(u'בְּ/תוֹךְ')
expected = ['bet', 'dagesh-qal', 'sheva-na', 'sav', 'holam-male',
'khaf-sofit', 'sheva-nah']
self.assertEqual(test, expected, 'BGDKFT: bet, sav')
test = hebrew.parse(u'הַ/גְּדֹלִים')
expected = ['he', 'patah', 'gimel', 'dagesh-hazaq', 'sheva-na',
'dalet', 'holam-haser', 'lamed', 'hiriq-male', 'yod',
'mem-sofit']
self.assertEqual(test, expected, 'BGDKFT: gimel')
test = hebrew.parse(u'כִּי')
expected = ['kaf', 'dagesh-qal', 'hiriq-male', 'yod']
self.assertEqual(test, expected, 'BGDKFT: kaf')
test = hebrew.parse(u'נֶפֶשׁ')
expected = ['nun', 'segol', 'fe', 'segol', 'shin']
self.assertEqual(test, expected, 'BGDKFT: fe')
test = hebrew.parse(u'נֶעְתַּם')
expected = ['nun', 'segol', 'ayin', 'sheva', 'tav', 'dagesh-qal',
'patah', 'mem-sofit']
self.assertEqual(test, expected, 'BGDKFT: tav')
def test_parse_mapiq(self):
"""Expected to parse mapiq letters."""
test = hebrew.parse(u'אֶֽעֱשֶׂהּ־')
expected = ['alef', 'segol', 'ayin', 'hataf-segol', 'sin', 'segol',
'mapiq-he', 'mapiq']
self.assertEqual(test, expected, 'when mapiq-he')
test = hebrew.parse(u'וַ/יָּבִ֥יאּוּ')
expected = ['vav', 'patah', 'yod', 'dagesh-hazaq', 'patah', 'vet',
'hiriq-male', 'yod', 'mapiq-alef', 'mapiq', 'shuruq']
def test_parse_rare(self):
"""Expected to parse rare strings."""
test = hebrew.parse(u'תּוֹסְףְּ')
expected = ['tav', 'dagesh-qal', 'holam-male', 'samekh', 'sheva-na',
'pe-sofit', 'dagesh-qal', 'sheva-nah']
self.assertEqual(test, expected, 'rare: pe-sofit')
test = hebrew.parse(u'לְיִשָּׂשכָר')
expected = ['lamed', 'sheva-na', 'yod', 'hiriq', 'sin',
'dagesh-hazaq', 'qamats', 'sin', 'khaf', 'qamats', 'resh']
self.assertEqual(test, expected, 'rare: sin, dagesh, sin; Yissacher')
def test_parse_basic(self):
"""Expected to parse basic strings."""
test = hebrew.parse(u'בְּ/רֵאשִׁית')
expected = ['bet', 'dagesh-qal', 'sheva-na', 'resh', 'tsere', 'alef',
'shin', 'hiriq-male', 'yod', 'sav']
self.assertEqual(test, expected)
test = hebrew.parse(u'הַ/שָּׁמַיִם')
expected = ['he', 'patah', 'shin', 'dagesh-hazaq', 'qamats', 'mem',
'patah', 'yod', 'hiriq', 'mem-sofit']
self.assertEqual(test, expected)
def test_parse_qamats_qatan(self):
"""Expected to parse qamats-qatan."""
test = hebrew.parse(u'אָהֳלֹ/ה')
expected = ['alef', 'qamats-qatan', 'he', 'hataf-qamats', 'lamed',
'holam-haser', 'he']
self.assertEqual(test, expected, 'when precedes hataf-qamats')
test = hebrew.parse(u'וּ/בְ/כָל־')
expected = ['shuruq', 'vet', 'sheva-na', 'khaf', 'qamats-qatan',
'lamed']
self.assertEqual(test, expected, 'when ends in maqaf')
def test_parse_shuruq(self):
"""Expected to parse shuruq."""
test = hebrew.parse(u'וּ/בַ/לַּיְלָה')
expected = ['shuruq', 'vet', 'patah', 'lamed', 'dagesh-hazaq', 'patah',
'yod', 'sheva', 'lamed', 'qamats', 'he']
self.assertEqual(test, expected, 'when start of word')
test = hebrew.parse(u'וַיּוֹרֵהוּ')
expected = ['vav', 'patah', 'yod', 'dagesh-hazaq', 'holam-male',
'resh', 'tsere', 'he', 'shuruq']
self.assertEqual(test, expected, 'when end of word')
def test_parse_vav_dagesh_vowel(self):
"""Expected to parse vav, dagesh, {vowel}."""
test = hebrew.parse(u'אִוָּשֵׁעַ')
expected = ['alef', 'hiriq', 'vav', 'dagesh-hazaq', 'qamats', 'shin',
'tsere', 'ayin', 'patah-genuvah']
self.assertEqual(test, expected, 'when vowel is qamats')
test = hebrew.parse(u'קַוֹּה')
expected = ['qof', 'patah', 'vav', 'dagesh-hazaq', 'holam-haser', 'he']
self.assertEqual(test, expected, 'when vowel is holam-haser')
test = hebrew.parse(u'וּלְצַוּוֹת')
expected = ['shuruq', 'lamed', 'sheva-na', 'tsadi', 'patah', 'vav',
'dagesh-hazaq', 'holam-male', 'sav']
self.assertEqual(test, expected, 'when vowel is holam-male')
def test_parse_double_sheva(self):
"""Expected to parse double-sheva."""
test = hebrew.parse(u'וַ/יִּתְפְּרוּ')
expected = ['vav', 'patah', 'yod', 'dagesh-hazaq', 'hiriq', 'sav',
'sheva-nah', 'pe', 'dagesh-qal', 'sheva-na', 'resh',
'shuruq']
self.assertEqual(test, expected)
def test_parse_holam_haser_for_vav(self):
"""Expected to parse holam-haser for vav (unicode code point)."""
test = hebrew.parse(u'עֲוֺנִ/י')
expected = ['ayin', 'hataf-patah', 'vav', 'holam-haser', 'nun',
'hiriq-male', 'yod']
self.assertEqual(test, expected)
