def i2a(ipastring):
"""Convert an IPA string to an arpabet string.
Args:
ipastring: A string representing ipa using utf-8 encoding.
Returns:
arpabetstring: A list representing the corresponding arpabet.
"""
amapper = ARPABETMapper()
arpastring = amapper.map_unicode_string(ipastring,
ignore=True,
return_as_list=True)
return arpastring
def test_map_unicode_string_single(self):
mapper = ARPABETMapper()
values = [
(u"", u""),
(u"p", u"P"),
(u"p\u03B8", u"PTH"),
(u"\u027E", u"DX"),
(u"p\u0258\u026A", u"PEY"),
(u"p\u0258\u026Aw", u"PEYW"),
(u"p\u0258\u026A\u0258\u026Aw", u"PEYEYW"),
(u"p\u0258\u026A\u0251w", u"PEYAAW"),
(u"\u006A\u0075", u"YUW"),
]
for v, e in values:
self.assertEqual(mapper.map_unicode_string(v, single_char_parsing=True), e)
def test_map_unicode_string(self):
mapper = ARPABETMapper()
values = [
(None, None),
(u"", u""),
(u"p", u"P"),
(u"p\u03B8", u"PTH"),
(u"\u027E", u"DX"),
(u"p\u0258\u026A", u"PEY"),
(u"p\u0258\u026Aw", u"PEYW"),
(u"p\u0258\u026A\u0258\u026Aw", u"PEYEYW"),
(u"p\u0258\u026A\u0251w", u"PEYAAW"),
(u"\u006A\u0075", u"YUW"),
]
for v, e in values:
self.assertEqual(mapper.map_unicode_string(v), e)
def test_map_unicode_string_ignore(self):
mapper = ARPABETMapper()
values = [
(None, None),
(u"", u""),
(u"p", u"P"),
(u"p\u03B8", u"PTH"),
(u"\u027E", u"DX"),
(u"p\u0258\u026A", u"PEY"),
(u"p\u0258\u026Aw", u"PEYW"),
(u"p\u0258\u026A\u0258\u026Aw", u"PEYEYW"),
(u"p\u0258\u026A\u0251w", u"PEYAAW"),
(u"\u006A\u0075", u"YUW"),
(u"a\u006A\u0075", u"AEYUW"),
(u"o\u006A\u0075", u"OHYUW"),
(u"S\u006A\u0075", u"YUW"),
(u"S\u006A\u0075\u02C8be", u"YUWBEH"),
]
for v, e in values:
self.assertEqual(mapper.map_unicode_string(v, ignore=True), e)
def test_map_unicode_string_ignore(self):
mapper = ARPABETMapper()
values = [
(None, None),
(u"", u""),
(u"p", u"P"),
(u"p\u03B8", u"PTH"),
(u"\u027E", u"DX"),
(u"p\u0258\u026A", u"PEY"),
(u"p\u0258\u026Aw", u"PEYW"),
(u"p\u0258\u026A\u0258\u026Aw", u"PEYEYW"),
(u"p\u0258\u026A\u0251w", u"PEYAAW"),
(u"\u006A\u0075", u"YUW"),
(u"a\u006A\u0075", u"AEYUW"),
(u"o\u006A\u0075", u"OHYUW"),
(u"S\u006A\u0075", u"YUW"),
(u"S\u006A\u0075\u02C8be", u"YUWBEH"),
]
for v, e in values:
self.assertEqual(mapper.map_unicode_string(v, ignore=True), e)