def get_attachment(self, word: TextPunctuation) -> dict:
text = word.get_text()
sign = word.get_punctuation()
buffer_attach = dict()
attachment = self.get_data().zero_syll_words[text]
buffer_attach['attachment'] = attachment
buffer_attach['text'] = text
buffer_attach['sign'] = sign
return buffer_attach
def clean(self, words: list) -> list:
curr, foll = words[0], words[1]
buffer_text, buffer_signs = [], []
if len(curr.get_text()) != 1 and curr.get_text().isupper():
return [None, foll]
for i in range(len(curr.get_text())):
sym = curr.get_text()[i]
sign = curr.get_punctuation()[i]
if sym == '.' and isinstance(
foll, TextPunctuation) and not foll.get_text().istitle():
return [None, foll]
if not len(buffer_text) and sign == constants.PUNCT:
if sym in punctuation_to_erase:
continue
elif sym in dashes:
return [None, foll]
if sign == constants.HYPHEN:
if len(curr.get_text()) == i + 1:
return [None, foll]
else:
continue
if sign == constants.PUNCT and sym not in hyphen_dashes:
for j in range(i + 1, len(curr.get_text())):
next_sign = curr.get_punctuation()[j]
if next_sign != constants.PUNCT:
return [None, foll]
return [
TextPunctuation(''.join(buffer_text), buffer_signs), foll
]
sym_low = sym.lower()
if sym_low not in self.get_data().letters:
return [None, foll]
buffer_text.append(sym_low)
buffer_signs.append(sign)
curr = TextPunctuation(''.join(buffer_text), buffer_signs)
if isinstance(foll, TextPunctuation) and len(foll.get_text()) == 1:
foll_low = TextPunctuation(foll.get_text().lower(),
foll.get_punctuation())
if self.is_zero_syll(foll_low.get_text()):
buffer_attach = self.get_attachment(foll_low)
if buffer_attach['attachment'] == 'to_preceding':
buffer_text.append(foll_low.get_text())
buffer_signs.append(foll_low.get_punctuation())
foll = None
elif isinstance(foll, TextPunctuation) and len(
curr.get_text()) == 1 and self.is_zero_syll(curr.get_text()):
buffer_attach = self.get_attachment(curr)
if buffer_attach['attachment'] == 'to_following':
foll.set_text(curr.get_text() + foll.get_text())
foll.set_punctuation(curr.get_punctuation() +
foll.get_punctuation())
return [None, foll]
return [TextPunctuation(''.join(buffer_text), buffer_signs), foll]