def normalize(cls, word: Word) -> str:
res = word.source.strip().upper()
# Remove yer before linebreak unless tagged otherwise
if word.tagset.note is not None and not ("+ъ" in word.tagset.note
or "+ь" in word.tagset.note):
res = cls._replace_yer_before_linebreak(res)
# Remove milestones
res = re.sub(Milestone.REGEX, "", res)
if word.is_cardinal_number():
return word.tagset.pos # Non-spelled out numerals
if word.is_ordinal_number():
return str(
Number(res.replace("(", "").replace(")", "").replace(
" ", ""))) # Spelled-out numerals
res = replace_chars(
res,
characters.latin_special_characters,
characters.cyrillic_special_characters,
)
for idx in [idx for idx, char in enumerate(res)
if char == "V"]: # Izhitsa positional replacement
res = res[:idx] + cls._replace_izhitsa(res, idx) + res[idx + 1:]
# Orthography normalization
res = modif(res, word.tagset.pos if word.tagset is not None else "")
return res.replace("#", "").replace("(", "").replace(")", "")
def __get_upos(word: Word) -> str:
if word.pos == "сущ":
return "PROPN" if word.is_proper else "NOUN"
if word.pos in ("прил", "прил/ср", "прил/н", "числ/п"):
return "ADJ"
if word.pos == "числ" or word.is_cardinal_number():
return "NUM"
if word.pos == "мест":
return "PRON" # TODO Distinguish DET
if word.pos in ("гл", "гл/в", "прич", "прич/в", "инф", "инф/в", "суп"):
return "AUX" if word.tagset.role == "св" else "VERB"
if word.pos == "нар":
return "ADV"
if word.pos in ("пред", "посл"):
return "ADP"
if word.pos == "союз":
return "CCONJ" # TODO Distinguish SCONJ
if word.pos == "част":
return "PART"
if word.pos == "межд":
return "INTJ"
return "X"
