def j2h(text):
i = 0
result = []
while i < len(text):
if i + 2 < len(text) and text[i] in LEADS and text[
i + 1] in VOWELS and text[i + 2] in TAILS:
result.append(jamo.j2h(text[i], text[i + 1], text[i + 2]))
i += 3
elif i + 1 < len(text) and text[i] in LEADS and text[i + 1] in VOWELS:
result.append(jamo.j2h(text[i], text[i + 1]))
i += 2
else:
result.append(text[i])
i += 1
return "".join(result)
def _get_text_from_candidates(candidates):
if len(candidates) == 0:
return ""
elif len(candidates) == 1:
return _jamo_char_to_hcj(candidates[0])
else:
return j2h(**dict(zip(["lead", "vowel", "tail"], candidates)))
def test_j2h(self):
"""j2h hardcoded tests.
Arguments may be integers corresponding to the U+11xx codepoints, the
actual U+11xx jamo characters, or HCJ.
Outputs a one-character Hangul string.
This function is defined solely for naming conisistency with
jamo_to_hangul.
"""
assert jamo.j2h('ㅎ', 'ㅏ', 'ㄴ') == "한",\
"j2h doesn't work. Hint: it's the same as jamo_to_hangul."
assert jamo.j2h('ㅎ', 'ㅏ') == "하",\
"j2h doesn't work. Hint: it's the same as jamo_to_hangul."
def test_j2h(self):
"""j2h hardcoded tests.
Arguments may be integers corresponding to the U+11xx codepoints, the
actual U+11xx jamo characters, or HCJ.
Outputs a one-character Hangul string.
This function is defined solely for naming conisistency with
jamo_to_hangul.
"""
assert jamo.j2h('ㅎ', 'ㅏ', 'ㄴ') == "한",\
"j2h doesn't work. Hint: it's the same as jamo_to_hangul."
assert jamo.j2h('ㅎ', 'ㅏ') == "하",\
"j2h doesn't work. Hint: it's the same as jamo_to_hangul."
def encode(self, text):
text = self.preprocess(text, vowel_type=self.vowel_type)
words = []
for word in TreebankWordTokenizer().tokenize(text):
from reo_toolkit import is_maori
if not is_maori(word):
words.append(word)
continue
encoded_text = []
for syllable in self.tokenize(word):
if not all(ch in alphabet for ch in syllable):
encoded_text.append(syllable)
continue
if syllable in vowels:
syllable = 'x' + syllable
try:
consonant, vowel = ''.join(
[self.encoder_dict[ch] for ch in syllable])
except KeyError:
logging.error(
"KeyError: phoneme {} not in encoder_dict".format(
syllable))
raise KeyError
try:
encoded = jamo.j2h(consonant, vowel)
except jamo.InvalidJamoError:
logging.error(
'InvalidJamoError - Consonant={} Vowel={} Syllable={}'.
format(consonant, vowel, syllable))
encoded_text.append(encoded)
words.append(''.join(encoded_text))
return TreebankWordDetokenizer().detokenize(words)
def get_text_from_candidates(candidates):
if len(candidates) == 0:
return ""
elif len(candidates) == 1:
return _jamo_char_to_hcj(candidates[0])
else:
return j2h(**dict(zip(["chosung", "jungsung", "jongsung"], candidates)))
def j2syl(string):
choseong = "[\u1100-\u1112]"
jungseong = "[\u1161-\u1175]"
jongseong = "[\u11A8-\u11C2]"
# CVC
matches = re.findall(f"{choseong}{jungseong}{jongseong}", string)
for match in matches:
syl = j2h(*match)
string = string.replace(match, syl)
# CV
matches = re.findall(f"{choseong}{jungseong}", string)
for match in matches:
syl = j2h(*match)
string = string.replace(match, syl)
return string
def compose(letters):
# insert placeholder
letters = re.sub("(^|[^\u1100-\u1112])([\u1161-\u1175])", r"\1ᄋ\2",
letters)
string = letters # assembled characters
# c+v+c
syls = set(
re.findall("[\u1100-\u1112][\u1161-\u1175][\u11A8-\u11C2]", string))
for syl in syls:
string = string.replace(syl, j2h(*syl))
# c+v
syls = set(re.findall("[\u1100-\u1112][\u1161-\u1175]", string))
for syl in syls:
string = string.replace(syl, j2h(*syl))
return string
def compose(self, string):
string = self.contract(string)
choseong = "[\u1100-\u1112]"
jungseong = "[\u1161-\u1175]"
jongseong = "[\u11A8-\u11C2]"
# CVC first
matches = re.findall(f"{choseong}{jungseong}{jongseong}", string)
for match in matches:
syl = j2h(*match)
string = string.replace(match, syl)
# CV
matches = re.findall(f"{choseong}{jungseong}", string)
for match in matches:
syl = j2h(*match)
string = string.replace(match, syl)
return string
