FunASR/funasr/tokenizer/build_tokenizer.py

from pathlib import Path
from typing import Iterable
from typing import Union
from abc import ABC
from abc import abstractmethod
from typing import Iterable
from typing import List
from pathlib import Path
from typing import Dict
from typing import Iterable
from typing import List
from typing import Union

import numpy as np

from funasr.tokenizer.abs_tokenizer import AbsTokenizer
from funasr.tokenizer.char_tokenizer import CharTokenizer
from funasr.tokenizer.phoneme_tokenizer import PhonemeTokenizer
from funasr.tokenizer.sentencepiece_tokenizer import SentencepiecesTokenizer
from funasr.tokenizer.word_tokenizer import WordTokenizer


def build_tokenizer(
    token_type: str,
    bpemodel: Union[Path, str, Iterable[str]] = None,
    non_linguistic_symbols: Union[Path, str, Iterable[str]] = None,
    remove_non_linguistic_symbols: bool = False,
    space_symbol: str = "<space>",
    delimiter: str = None,
    g2p_type: str = None,
    **kwargs,
):
    """A helper function to instantiate Tokenizer"""
    if token_type == "bpe":
        if bpemodel is None:
            raise ValueError('bpemodel is required if token_type = "bpe"')

        if remove_non_linguistic_symbols:
            raise RuntimeError(
                "remove_non_linguistic_symbols is not implemented for token_type=bpe"
            )
        return SentencepiecesTokenizer(bpemodel, **kwargs)

    elif token_type == "word":
        if remove_non_linguistic_symbols and non_linguistic_symbols is not None:
            return WordTokenizer(
                delimiter=delimiter,
                non_linguistic_symbols=non_linguistic_symbols,
                remove_non_linguistic_symbols=True,
            )
        else:
            return WordTokenizer(delimiter=delimiter, **kwargs)

    elif token_type == "char":
        return CharTokenizer(
            non_linguistic_symbols=non_linguistic_symbols,
            space_symbol=space_symbol,
            remove_non_linguistic_symbols=remove_non_linguistic_symbols,
            **kwargs
        )

    elif token_type == "phn":
        return PhonemeTokenizer(
            g2p_type=g2p_type,
            non_linguistic_symbols=non_linguistic_symbols,
            space_symbol=space_symbol,
            remove_non_linguistic_symbols=remove_non_linguistic_symbols,
            **kwargs
        )

    else:
        raise ValueError(
            f"token_mode must be one of bpe, word, char or phn: " f"{token_type}"
        )