from spacy.util import registry
from spacy.tokenizer import Tokenizer
import pathlib

@registry.tokenizers("customize_tokenizer")
def make_customize_tokenizer():
    def customize_tokenizer(nlp):
        tokenizer = Tokenizer(nlp.vocab)
        script_dir = pathlib.Path(__file__).parent.resolve()
        return tokenizer.from_disk(script_dir / "tokenizer")

    return customize_tokenizer