def romanize(self, string):
"""Convert a string from Hangul to Latin Script
:param string: string; any sentence, romanize if hangul else return the same text"""
if not self.has_hangul(string):
return string
sentence_latin = ''
syllables = list(string)
for i, syllable in enumerate(syllables):
if kr.is_hangul(syllable):
sentence_latin += self.jaeum(
syllable, self.__get_final_prior(syllables, i))
sentence_latin += self.moeum(syllable)
sentence_latin += self.batchim(
syllable, self.__get_initial_next(syllables, i))
else:
sentence_latin += syllable
sentence_latin = sentence_latin.strip()
if sentence_latin and not sentence_latin[0].isupper() \
and sentence_latin[0].isalpha():
sentence_latin = sentence_latin.capitalize()
return sentence_latin
def has_ssangbatchim(self, syllable):
"""True if param contains a double batchim
:param syllable: char; a syllable"""
if kr.is_hangul(syllable):
return self.is_ssangbatchim(kr.get_final(syllable))
else:
return False
def has_hangul(self, string):
"""Return True if there is any hangul char in the string"""
syllables = list(string)
for syllable in syllables:
if kr.is_hangul(syllable):
return True
return False
def __get_final_prior(self, syllables, i):
"""Get the batchim of the prior syllable of a word
:param syllabes: list; syllables in Hangul
:param i: int; the current index of the syllable in the sentence"""
if i > 0:
if syllables[i - 1] != '' and kr.is_hangul(syllables[i - 1]):
return kr.get_final(syllables[i - 1])
return ''
def __get_initial_next(self, syllables, i):
"""Get the initial letter of the next syllable of a word
:param syllabes: list; syllables in Hangul
:param i: int; the current index of the syllable in the sentence"""
if i < (len(syllables) - 1):
if syllables[i + 1] != '' and kr.is_hangul(syllables[i + 1]):
return kr.get_initial(syllables[i + 1])
return ''
def pull(text):
exploded = [list(hangul.split_char(c)) if hangul.is_hangul(c) else c for c in text]
for i in range(len(exploded)):
if i == 0:
continue
a = exploded[i - 1]
b = exploded[i]
if type(a) is list and type(b) is list:
if not a[2] and b[0] != "ㅇ":
a[2] = simplify_rule.get(b[0], b[0])
b[0] = "ㅇ"
elif a[2] in compound_rule and b[0] in compound_rule[a[2]]:
a[2] = compound_rule[a[2]][b[0]]
b[0] = "ㅇ"
return "".join([hangul.join_char(c) if type(c) is list else c for c in exploded])
def pull(text):
exploded = [
list(hangul.split_char(c)) if hangul.is_hangul(c) else c for c in text
]
for i in range(len(exploded)):
if i == 0:
continue
a = exploded[i - 1]
b = exploded[i]
if type(a) is list and type(b) is list:
if not a[2] and b[0] != "ㅇ":
next = b[0]
if next == "ㅉ":
next = "ㅈ"
elif next == "ㅃ":
next = "ㅂ"
elif next == "ㄸ":
next = "ㄷ"
a[2] = next
b[0] = "ㅇ"
elif a[2] == "ㄹ":
if b[0] == "ㄱ":
a[2] = "ㄺ"
elif b[0] == "ㅁ":
a[2] = "ㄻ"
elif b[0] == "ㅍ":
a[2] = "ㄿ"
b[0] = "ㅇ"
elif a[2] == "ㄱ" and b[0] == "ㅅ":
a[2] = "ㄳ"
b[0] = "ㅇ"
elif a[2] == "ㄴ" and b[0] == "ㅈ":
a[2] = "ㄵ"
b[0] = "ㅇ"
elif a[2] == "ㅂ" and b[0] == "ㅅ":
a[2] = "ㅄ"
b[0] = "ㅇ"
elif a[2] == "ㄹ":
if b[0] == "ㅂ":
a[2] = "ㄼ"
elif b[0] == "ㅅ":
a[2] = "ㄽ"
elif b[0] == "ㅌ":
a[2] = "ㄾ"
b[0] = "ㅇ"
return "".join(
[hangul.join_char(c) if type(c) is list else c for c in exploded])
def pull(text):
exploded = [list(hangul.split_char(c)) if hangul.is_hangul(c) else c for c in text]
for i in range(len(exploded)):
if i == 0:
continue
a = exploded[i - 1]
b = exploded[i]
if type(a) is list and type(b) is list:
if not a[2] and b[0] != "ㅇ":
next = b[0]
if next == "ㅉ":
next = "ㅈ"
elif next == "ㅃ":
next = "ㅂ"
elif next == "ㄸ":
next = "ㄷ"
a[2] = next
b[0] = "ㅇ"
elif a[2] == "ㄹ":
if b[0] == "ㄱ":
a[2] = "ㄺ"
elif b[0] == "ㅁ":
a[2] = "ㄻ"
elif b[0] == "ㅍ":
a[2] = "ㄿ"
b[0] = "ㅇ"
elif a[2] == "ㄱ" and b[0] == "ㅅ":
a[2] = "ㄳ"
b[0] = "ㅇ"
elif a[2] == "ㄴ" and b[0] == "ㅈ":
a[2] = "ㄵ"
b[0] = "ㅇ"
elif a[2] == "ㅂ" and b[0] == "ㅅ":
a[2] = "ㅄ"
b[0] = "ㅇ"
elif a[2] == "ㄹ":
if b[0] == "ㅂ":
a[2] = "ㄼ"
elif b[0] == "ㅅ":
a[2] = "ㄽ"
elif b[0] == "ㅌ":
a[2] = "ㄾ"
b[0] = "ㅇ"
return "".join([hangul.join_char(c) if type(c) is list else c for c in exploded])