tokenization_t5.py

# coding=utf-8
# Copyright 2018 T5 Authors and The HuggingFace Inc. team.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

"""Tokenization class for Google T5 (sentence piece)."""

import logging
import os
import warnings
from shutil import copyfile
from typing import List, Optional

import regex as re
import sentencepiece as spm

from .tokenization_base import PreTrainedTokenizer

logger = logging.getLogger(__name__)

VOCAB_FILES_NAMES = {"vocab_file": "spiece.model"}

PRETRAINED_VOCAB_FILES_MAP = {
    "vocab_file": {"t5-base": "https://huggingface.co/t5-base/resolve/main/spiece.model"}
}

PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES = {
    "t5-base": 512,
}


class T5Tokenizer(PreTrainedTokenizer):
    """
    Construct a T5 tokenizer. Based on `SentencePiece <https://github.com/google/sentencepiece>`.

    Args:
        vocab_file (:obj:`str`):
            Path to the vocabulary file.
        eos_token (:obj:`str`, `optional`, defaults to :obj:`"</s>"`):
            The end of sequence token.
        unk_token (:obj:`str`, `optional`, defaults to :obj:`"<unk>"`):
            The unknown token. A token that is not in the vocabulary cannot
            be converted to an ID and is set to be this token instead.
        pad_token (:obj:`str`, `optional`, defaults to :obj:`"<pad>"`):
            The token used for padding, for example when batching sequences of different lengths.
        extra_ids (:obj:`int`, `optional`, defaults to 100):
            Add a number of extra ids added to the end of the vocabulary for use
            as sentinels. These tokens are accessible as "<extra_id_{%d}>" where
            "{%d}" is a number between 0 and extra_ids-1. Extra tokens are indexed
            from the end of the vocabulary up to beginning ("<extra_id_0>" is the
            last token in the vocabulary like in T5 preprocessing see `here
            <https://github.com/google-research/text-to-text-transfer-transformer/blob/9fd7b14a769417be33bc6c850f9598764913c833/t5/data/preprocessors.py#L2117>`__).
        additional_special_tokens (:obj:`List[str]`, `optional`):
            Additional special tokens used by the tokenizer.

    """

    vocab_files_names = VOCAB_FILES_NAMES
    pretrained_vocab_files_map = PRETRAINED_VOCAB_FILES_MAP
    max_model_input_sizes = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES

    def __init__(
        self,
        vocab_file,
        eos_token="</s>",
        unk_token="<unk>",
        pad_token="<pad>",
        extra_ids=100,
        additional_special_tokens=None,
        add_bos_token=False,
        **kwargs,
    ):
        # Add extra_ids to the special token list
        if extra_ids > 0 and additional_special_tokens is None:
            additional_special_tokens = [f"<extra_id_{i}>" for i in range(extra_ids)]
        elif extra_ids > 0 and additional_special_tokens is not None:
            extra_tokens = len(
                set(filter(lambda x: bool("extra_id" in str(x)), additional_special_tokens))
            )
            if extra_tokens != extra_ids:
                raise ValueError(
                    f"Both extra_ids ({extra_ids}) and additional_special_tokens "
                    f"({additional_special_tokens}) are privided to T5Tokenizer. "
                    "In this case the additional_special_tokens must include the extra_ids tokens"
                )

        super().__init__(
            eos_token=eos_token,
            unk_token=unk_token,
            pad_token=pad_token,
            additional_special_tokens=additional_special_tokens,
            **kwargs,
        )

        self.vocab_file = vocab_file
        self._extra_ids = extra_ids

        self.sp_model = spm.SentencePieceProcessor()
        self.sp_model.Load(vocab_file)
        self.add_bos_token = add_bos_token

    @property
    def vocab_size(self):
        return self.sp_model.get_piece_size() + self._extra_ids

    def get_vocab(self):
        vocab = {self.convert_ids_to_tokens(i): i for i in range(self.vocab_size)}
        vocab.update(self.added_tokens_encoder)
        return vocab

    def _tokenize(self, text):
        """Tokenize a string."""
        pieces = self.sp_model.encode(text, out_type=str)
        return pieces

    def _convert_token_to_id(self, token):
        """Converts a token (str) in an id using the vocab."""
        if token.startswith("<extra_id_"):
            match = re.match(r"<extra_id_(\d+)>", token)
            num = int(match.group(1))
            return self.vocab_size - num - 1
        return self.sp_model.piece_to_id(token)

    def _convert_id_to_token(self, index):
        """Converts an index (integer) in a token (str) using the vocab."""
        if index < self.sp_model.get_piece_size():
            token = self.sp_model.IdToPiece(index)
        else:
            token = f"<extra_id_{self.vocab_size - 1 - index}>"
        return token

    def convert_tokens_to_string(self, tokens):
        """Converts a sequence of tokens (string) to a single string."""
        current_sub_tokens = []
        out_string = ""
        for token in tokens:
            # make sure that special tokens are not decoded using sentencepiece model
            if token in self.all_special_tokens:
                out_string += self.sp_model.decode_pieces(current_sub_tokens) + token + " "
                current_sub_tokens = []
            else:
                current_sub_tokens.append(token)
        out_string += self.sp_model.decode_pieces(current_sub_tokens)
        return out_string.strip()

    def _add_eos_if_not_present(self, token_ids):
        if not self.add_bos_token:
            return token_ids
        if len(token_ids) > 0 and token_ids[-1] == self.eos_token_id:
            warnings.warn("This sequence already has {self.eos_token}.")
            return token_ids
        else:
            return token_ids + [self.eos_token_id]

    def build_inputs_with_special_tokens(
        self, token_ids_0: List[int], token_ids_1: Optional[List[int]] = None
    ) -> List[int]:
        """Add special tokens to a sequence or a pair of sequence.
        T5 format sentence input:

        - single sequence: tokens_a </s>
        - pair of sequences: tokens_a </s> tokens_b </s>

        Args:
            token_ids_0 (List[int]): The token ids of sentence 0.
            token_ids_1 (List[int], optional): The token ids of sentence 1. Defaults to None.

        Returns:
            :obj:`List[str]`: The sequence after adding special toekens.
        """
        token_ids_0 = self._add_eos_if_not_present(token_ids_0)
        if token_ids_1 is None:
            return token_ids_0
        else:
            token_ids_1 = self._add_eos_if_not_present(token_ids_1)
            return token_ids_0 + token_ids_1

    def save_vocabulary(self, save_directory, filename_prefix=None):
        """Save the tokenizer vocabulary to a directory or file."""
        if not os.path.isdir(save_directory):
            logger.error(f"Vocabulary path ({save_directory}) should be a directory")
            return
        out_vocab_file = os.path.join(
            save_directory,
            (filename_prefix + "-" if filename_prefix else "") + VOCAB_FILES_NAMES["vocab_file"],
        )

        if os.path.abspath(self.vocab_file) != os.path.abspath(out_vocab_file):
            copyfile(self.vocab_file, out_vocab_file)
            logger.info(f"Copy vocab file to {out_vocab_file}")

        return (out_vocab_file,)