def trans_symbol(symbol,prev_term,next_term):
if symbol in count_symbols:
return count_sym_han[count_symbols.index(symbol)]
elif prev_term not in puncs:
if hgtk.checker.is_hangul(prev_term) or hgtk.checker.is_hangul(next_term):
return sym_han[symbols.index(symbol)]
elif prev_term.isdigit() or next_term.isdigit():
return sym_han[symbols.index(symbol)]
elif real_latin(prev_term) or real_latin(next_term):
return sym_pro[symbols.index(symbol)]
else:
return ''
else:
return ''
def trans_number(n,prev_term,next_term): ## Context-given number reading
if hgtk.checker.is_hangul(prev_term) and hgtk.checker.is_hangul(next_term):
return readNumberKor(n,next_term)
elif real_latin(prev_term) or real_latin(next_term):
if hgtk.checker.is_hangul(next_term) and n>10:
return readNumberKor(n,next_term)
else:
return readNumberEng(n)
else: ## Maybe hanja
if prev_term in symbols or next_term in symbols:
return readOnlyNum(n)
elif n > 99999:
return readBigNum(n)
else:
return readNumber(n)
def leftword(chunks):
for i in range(len(chunks)):
eojeol = chunks[i]
for j in range(len(eojeol)):
term = chunks[i][j]
if real_latin(term):
chunks[i][j] = read_acronym(term)
elif not hgtk.checker.is_hangul(term) and term not in puncs:
chunks[i][j] = ''
return chunks
def trans_eojeol(chunks,
chunks_4num,
metadata,
if_num=True,
if_sym=True,
if_han=True,
if_eng=True,
if_puncs=True,
if_else=True):
for i in range(len(chunks)):
eojeol = chunks[i]
for j in range(len(eojeol)):
term = eojeol[j]
if term.isdigit():
if if_num:
term = int(term)
x, y = decide_context(term, chunks_4num, eojeol, i, j)
chunks[i][j] = trans_number(term, x,
y) ## Reflects context
else:
chunks[i][j] = term
elif term in symbols + count_symbols and i + j > 0: ## Symbols not sentence-first
if if_sym:
x, y = decide_context(term, chunks_4num, eojeol, i, j)
chunks[i][j] = trans_symbol(term, x,
y) ## Currently bypassing
else:
chunks[i][j] = term
elif hgtk.checker.is_hanja(term):
if if_han:
chunks[i][j] = trans_hanja(term) ## Double check
else:
chunks[i][j] = term
elif real_latin(term):
if if_eng:
chunks[i][j] = trans_latin(
term) ## Transliteration (or bypassing)
else:
chunks[i][j] = term
elif term in puncs:
if if_puncs:
chunks[i][j] = term ## Bypassing by default
else:
chunks[i][j] = ''
elif hgtk.checker.is_hangul(term):
chunks[i][j] = term ## Bypassing by default
else:
if if_else:
chunks[i][
j] = term # '' ## Currently bypassing but able to delete
else:
chunks[i][j] = ''
return chunks