tokenization_bart_japanese.py

# coding=utf-8
# Copyright 2020 The Facebook AI Research Team 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.

import os
from contextlib import contextmanager
from shutil import copyfile
from typing import Any, Dict, List, Optional, Tuple

import sentencepiece as spm

from transformers import AddedToken, PreTrainedTokenizer
from transformers import logging


logger = logging.get_logger(__name__)

SPIECE_UNDERLINE = "▁"

VOCAB_FILES_NAMES = {"vocab_file": "sentencepiece.bpe.model"}


PRETRAINED_VOCAB_FILES_MAP = {
    "vocab_file": {
        "Formzu/bart-base-japanese": (
            "https://huggingface.co/Formzu/bart-base-japanese/resolve/main/sentencepiece.bpe.model"
        ),
        "Formzu/bart-large-japanese": (
            "https://huggingface.co/Formzu/bart-large-japanese/resolve/main/sentencepiece.bpe.model"
        ),
    }
}

PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES = {
    "Formzu/bart-base-japanese": 1024,
    "Formzu/bart-large-japanese": 1024,
}


class BartJapaneseTokenizer(PreTrainedTokenizer):
    """
    Construct a BART tokenizer for Japanese text.

    Adapted from [`RobertaTokenizer`], [`XLNetTokenizer`] and [`MBartTokenizer`]. Based on
    [SentencePiece](https://github.com/google/sentencepiece).

    The tokenization method is `<bos> <tokens> <eos>`.

    Examples:

    ```python
    >>> from tokenization_bart_japanese import BartJapaneseTokenizer

    >>> tokenizer = BartJapaneseTokenizer.from_pretrained("Formzu/bart-base-japanese")
    >>> example_japanese_phrase = "今日は晴れています。"
    >>> expected_label = "天気"
    >>> inputs = tokenizer(example_japanese_phrase, return_tensors="pt")
    >>> labels = tokenizer(expected_label, return_tensors="pt")
    >>> inputs["labels"] = labels["input_ids"]
    ```"""

    vocab_files_names = VOCAB_FILES_NAMES
    max_model_input_sizes = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES
    pretrained_vocab_files_map = PRETRAINED_VOCAB_FILES_MAP
    model_input_names = ["input_ids", "attention_mask"]

    prefix_tokens: List[int] = []
    suffix_tokens: List[int] = []

    def __init__(
        self,
        vocab_file,
        bos_token="<s>",
        eos_token="</s>",
        sep_token="</s>",
        cls_token="<s>",
        unk_token="<unk>",
        pad_token="<pad>",
        mask_token="<mask>",
        tokenizer_file=None,
        src_lang=None,
        tgt_lang=None,
        sp_model_kwargs: Optional[Dict[str, Any]] = None,
        additional_special_tokens=None,
        **kwargs
    ):
        # Mask token behave like a normal word, i.e. include the space before it
        mask_token = AddedToken(mask_token, lstrip=True, rstrip=False) if isinstance(mask_token, str) else mask_token

        self.sp_model_kwargs = {} if sp_model_kwargs is None else sp_model_kwargs

        super().__init__(
            bos_token=bos_token,
            eos_token=eos_token,
            unk_token=unk_token,
            sep_token=sep_token,
            cls_token=cls_token,
            pad_token=pad_token,
            mask_token=mask_token,
            tokenizer_file=None,
            src_lang=src_lang,
            tgt_lang=tgt_lang,
            additional_special_tokens=additional_special_tokens,
            sp_model_kwargs=self.sp_model_kwargs,
            **kwargs,
        )


        self.sp_model = spm.SentencePieceProcessor(**self.sp_model_kwargs)
        self.sp_model.Load(str(vocab_file))
        self.vocab_file = vocab_file
        try:
            from zenhan import h2z
        except ModuleNotFoundError as error:
            raise error.__class__(
                "You need to install zenhan to use BartJapaneseTokenizer."
                "See https://pypi.org/project/zenhan/ for installation."
            )   
        try:
            from pyknp import Juman
        except ModuleNotFoundError as error:
            raise error.__class__(
                "You need to install pyknp to use BartJapaneseTokenizer."
                "See https://pypi.org/project/pyknp/ for installation."
            )        

        self.h2z = h2z
        self.jumanpp = Juman()

        # Original fairseq vocab and spm vocab must be "aligned":
        # Vocab    |    0    |    1    |   2    |    3    |    4   |   5    |   6    |   7    |   8    |   9
        # -------- | ------- | ------- | ------ | ------- | ------ | ------ | ------ | ------ | ------ | ------
        # fairseq  | '<s>'   | '<pad>' | '</s>' | '<unk>' |　'▁の'　|　'▁、'　|　'▁。'　|　'▁に'　|　'▁は'　|　'▁を'
        # spm      | '<unk>' | '<s>'   | '</s>' |　'▁の'　 |　'▁、'　|　'▁。'　|　'▁に'　|　'▁は'　|　'▁を'　|　'▁と'

        # Mimic fairseq token-to-id alignment for the first 4 token
        self.fairseq_tokens_to_ids = {"<s>": 0, "<pad>": 1, "</s>": 2, "<unk>": 3}

        # The first "real" token "▁の" has position 4 in the original fairseq vocab and position 3 in the spm vocab
        self.fairseq_offset = 1

        self.sp_model_size = len(self.sp_model)

        self.fairseq_tokens_to_ids["<mask>"] = len(self.sp_model) + self.fairseq_offset
        self.fairseq_ids_to_tokens = {v: k for k, v in self.fairseq_tokens_to_ids.items()}

        self.set_special_tokens()

    def __getstate__(self):
        state = self.__dict__.copy()
        state["sp_model"] = None
        state["sp_model_proto"] = self.sp_model.serialized_model_proto()
        return state

    def __setstate__(self, d):
        self.__dict__ = d

        # for backward compatibility
        if not hasattr(self, "sp_model_kwargs"):
            self.sp_model_kwargs = {}

        self.sp_model = spm.SentencePieceProcessor(**self.sp_model_kwargs)
        self.sp_model.LoadFromSerializedProto(self.sp_model_proto)

    @property
    def vocab_size(self):
        return len(self.sp_model) + self.fairseq_offset + 1  # Plus 1 for the mask token

    def get_special_tokens_mask(
        self, token_ids_0: List[int], token_ids_1: Optional[List[int]] = None, already_has_special_tokens: bool = False
    ) -> List[int]:
        """
        Retrieve sequence ids from a token list that has no special tokens added. This method is called when adding
        special tokens using the tokenizer `prepare_for_model` method.

        Args:
            token_ids_0 (`List[int]`):
                List of IDs.
            token_ids_1 (`List[int]`, *optional*):
                Optional second list of IDs for sequence pairs.
            already_has_special_tokens (`bool`, *optional*, defaults to `False`):
                Whether or not the token list is already formatted with special tokens for the model.

        Returns:
            `List[int]`: A list of integers in the range [0, 1]: 1 for a special token, 0 for a sequence token.
        """

        if already_has_special_tokens:
            return super().get_special_tokens_mask(
                token_ids_0=token_ids_0, token_ids_1=token_ids_1, already_has_special_tokens=True
            )

        prefix_ones = [1] * len(self.prefix_tokens)
        suffix_ones = [1] * len(self.suffix_tokens)
        if token_ids_1 is None:
            return prefix_ones + ([0] * len(token_ids_0)) + suffix_ones
        return prefix_ones + ([0] * len(token_ids_0)) + ([0] * len(token_ids_1)) + suffix_ones

    def build_inputs_with_special_tokens(
        self, token_ids_0: List[int], token_ids_1: Optional[List[int]] = None
    ) -> List[int]:
        """
        Build model inputs from a sequence or a pair of sequence for sequence classification tasks by concatenating and
        adding special tokens. A Japanese BART sequence has the following format, where `X` represents the sequence:

        - `input_ids` (for encoder) `[bos] X [eos]`
        - `decoder_input_ids`: (for decoder) `[bos] X [eos]`

        Pairs of sequences are not the expected use case, but they will be handled without a separator.

        Args:
            token_ids_0 (`List[int]`):
                List of IDs to which the special tokens will be added.
            token_ids_1 (`List[int]`, *optional*):
                Optional second list of IDs for sequence pairs.

        Returns:
            `List[int]`: List of [input IDs](../glossary#input-ids) with the appropriate special tokens.
        """
        if token_ids_1 is None:
            return self.prefix_tokens + token_ids_0 + self.suffix_tokens
        # We don't expect to process pairs, but leave the pair logic for API consistency
        return self.prefix_tokens + token_ids_0 + token_ids_1 + self.suffix_tokens

    def create_token_type_ids_from_sequences(
        self, token_ids_0: List[int], token_ids_1: Optional[List[int]] = None
    ) -> List[int]:
        """
        Create a mask from the two sequences passed to be used in a sequence-pair classification task. Japanese BART does not
        make use of token type ids, therefore a list of zeros is returned.

        Args:
            token_ids_0 (`List[int]`):
                List of IDs.
            token_ids_1 (`List[int]`, *optional*):
                Optional second list of IDs for sequence pairs.

        Returns:
            `List[int]`: List of zeros.

        """

        sep = [self.sep_token_id]
        cls = [self.cls_token_id]

        if token_ids_1 is None:
            return len(cls + token_ids_0 + sep) * [0]
        return len(cls + token_ids_0 + sep + sep + token_ids_1 + sep) * [0]

    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: str) -> List[str]:
        text = text
        text = self.h2z(text)
        text = self.jumanpp.analysis(text)
        text = ' '.join([mrph.midasi for mrph in text.mrph_list()])
        return self.sp_model.encode(text, out_type=str)

    def _convert_token_to_id(self, token):
        """Converts a token (str) in an id using the vocab."""
        if token in self.fairseq_tokens_to_ids:
            return self.fairseq_tokens_to_ids[token]
        spm_id = self.sp_model.PieceToId(token)

        # Need to return unknown token if the SP model returned 0
        return spm_id + self.fairseq_offset if spm_id else self.unk_token_id

    def _convert_id_to_token(self, index):
        """Converts an index (integer) in a token (str) using the vocab."""
        if index in self.fairseq_ids_to_tokens:
            return self.fairseq_ids_to_tokens[index]
        return self.sp_model.IdToPiece(index - self.fairseq_offset)

    def convert_tokens_to_string(self, tokens):
        """Converts a sequence of tokens (strings for sub-words) in a single string."""
        out_string = "".join(tokens).replace(SPIECE_UNDERLINE, " ").strip()
        return out_string

    def save_vocabulary(self, save_directory: str, filename_prefix: Optional[str] = None) -> Tuple[str]:
        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) and os.path.isfile(self.vocab_file):
            copyfile(self.vocab_file, out_vocab_file)
        elif not os.path.isfile(self.vocab_file):
            with open(out_vocab_file, "wb") as fi:
                content_spiece_model = self.sp_model.serialized_model_proto()
                fi.write(content_spiece_model)

        return (out_vocab_file,)

    def set_special_tokens(self) -> None:
        """Set prefix=[bos], suffix=[eos]."""
        self.prefix_tokens = [self.bos_token_id]
        self.suffix_tokens = [self.eos_token_id]
        self.add_tokens(self.all_special_tokens_extended, special_tokens=True)