conex / espnet2 /text /sentencepiece_tokenizer.py
tobiasc's picture
Initial commit
ad16788
from pathlib import Path
from typing import Iterable
from typing import List
from typing import Union
import sentencepiece as spm
from typeguard import check_argument_types
from espnet2.text.abs_tokenizer import AbsTokenizer
class SentencepiecesTokenizer(AbsTokenizer):
def __init__(self, model: Union[Path, str]):
assert check_argument_types()
self.model = str(model)
# NOTE(kamo):
# Don't build SentencePieceProcessor in __init__()
# because it's not picklable and it may cause following error,
# "TypeError: can't pickle SwigPyObject objects",
# when giving it as argument of "multiprocessing.Process()".
self.sp = None
def __repr__(self):
return f'{self.__class__.__name__}(model="{self.model}")'
def _build_sentence_piece_processor(self):
# Build SentencePieceProcessor lazily.
if self.sp is None:
self.sp = spm.SentencePieceProcessor()
self.sp.load(self.model)
def text2tokens(self, line: str) -> List[str]:
self._build_sentence_piece_processor()
return self.sp.EncodeAsPieces(line)
def tokens2text(self, tokens: Iterable[str]) -> str:
self._build_sentence_piece_processor()
return self.sp.DecodePieces(list(tokens))