# coding=utf-8 # Copyright 2021 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 shutil import copyfile from typing import List, Optional, Tuple from tokenizers import processors from ...tokenization_utils import AddedToken, BatchEncoding from ...tokenization_utils_fast import PreTrainedTokenizerFast from ...utils import is_sentencepiece_available, logging if is_sentencepiece_available(): from .tokenization_mbart50 import MBart50Tokenizer else: MBart50Tokenizer = None logger = logging.get_logger(__name__) VOCAB_FILES_NAMES = {"vocab_file": "sentencepiece.bpe.model", "tokenizer_file": "tokenizer.json"} FAIRSEQ_LANGUAGE_CODES = ["ar_AR", "cs_CZ", "de_DE", "en_XX", "es_XX", "et_EE", "fi_FI", "fr_XX", "gu_IN", "hi_IN", "it_IT", "ja_XX", "kk_KZ", "ko_KR", "lt_LT", "lv_LV", "my_MM", "ne_NP", "nl_XX", "ro_RO", "ru_RU", "si_LK", "tr_TR", "vi_VN", "zh_CN", "af_ZA", "az_AZ", "bn_IN", "fa_IR", "he_IL", "hr_HR", "id_ID", "ka_GE", "km_KH", "mk_MK", "ml_IN", "mn_MN", "mr_IN", "pl_PL", "ps_AF", "pt_XX", "sv_SE", "sw_KE", "ta_IN", "te_IN", "th_TH", "tl_XX", "uk_UA", "ur_PK", "xh_ZA", "gl_ES", "sl_SI"] # fmt: skip class MBart50TokenizerFast(PreTrainedTokenizerFast): """ Construct a "fast" MBART tokenizer for mBART-50 (backed by HuggingFace's *tokenizers* library). Based on [BPE](https://huggingface.co/docs/tokenizers/python/latest/components.html?highlight=BPE#models). This tokenizer inherits from [`PreTrainedTokenizerFast`] which contains most of the main methods. Users should refer to this superclass for more information regarding those methods. Args: vocab_file (`str`): Path to the vocabulary file. src_lang (`str`, *optional*): A string representing the source language. tgt_lang (`str`, *optional*): A string representing the target language. eos_token (`str`, *optional*, defaults to `""`): The end of sequence token. sep_token (`str`, *optional*, defaults to `""`): The separator token, which is used when building a sequence from multiple sequences, e.g. two sequences for sequence classification or for a text and a question for question answering. It is also used as the last token of a sequence built with special tokens. cls_token (`str`, *optional*, defaults to `""`): The classifier token which is used when doing sequence classification (classification of the whole sequence instead of per-token classification). It is the first token of the sequence when built with special tokens. unk_token (`str`, *optional*, defaults to `""`): 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 (`str`, *optional*, defaults to `""`): The token used for padding, for example when batching sequences of different lengths. mask_token (`str`, *optional*, defaults to `""`): The token used for masking values. This is the token used when training this model with masked language modeling. This is the token which the model will try to predict. Examples: ```python >>> from transformers import MBart50TokenizerFast >>> tokenizer = MBart50TokenizerFast.from_pretrained("facebook/mbart-large-50", src_lang="en_XX", tgt_lang="ro_RO") >>> src_text = " UN Chief Says There Is No Military Solution in Syria" >>> tgt_text = "Şeful ONU declară că nu există o soluţie militară în Siria" >>> model_inputs = tokenizer(src_text, text_target=tgt_text, return_tensors="pt") >>> # model(**model_inputs) should work ```""" vocab_files_names = VOCAB_FILES_NAMES model_input_names = ["input_ids", "attention_mask"] slow_tokenizer_class = MBart50Tokenizer prefix_tokens: List[int] = [] suffix_tokens: List[int] = [] def __init__( self, vocab_file=None, src_lang=None, tgt_lang=None, tokenizer_file=None, eos_token="", sep_token="", cls_token="", unk_token="", pad_token="", mask_token="", **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 kwargs["additional_special_tokens"] = kwargs.get("additional_special_tokens", []) or [] kwargs["additional_special_tokens"] += [ code for code in FAIRSEQ_LANGUAGE_CODES if code not in kwargs["additional_special_tokens"] ] super().__init__( vocab_file, src_lang=src_lang, tgt_lang=tgt_lang, tokenizer_file=tokenizer_file, eos_token=eos_token, sep_token=sep_token, cls_token=cls_token, unk_token=unk_token, pad_token=pad_token, mask_token=mask_token, **kwargs, ) self.vocab_file = vocab_file self.lang_code_to_id = { lang_code: self.convert_tokens_to_ids(lang_code) for lang_code in FAIRSEQ_LANGUAGE_CODES } self._src_lang = src_lang if src_lang is not None else "en_XX" self.tgt_lang = tgt_lang self.cur_lang_code_id = self.lang_code_to_id[self._src_lang] self.set_src_lang_special_tokens(self._src_lang) @property def can_save_slow_tokenizer(self) -> bool: return os.path.isfile(self.vocab_file) if self.vocab_file else False @property def src_lang(self) -> str: return self._src_lang @src_lang.setter def src_lang(self, new_src_lang: str) -> None: self._src_lang = new_src_lang self.set_src_lang_special_tokens(self._src_lang) 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. The special tokens depend on calling set_lang. An MBART-50 sequence has the following format, where `X` represents the sequence: - `input_ids` (for encoder) `[src_lang_code] X [eos]` - `labels`: (for decoder) `[tgt_lang_code] X [eos]` BOS is never used. 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 prepare_seq2seq_batch( self, src_texts: List[str], src_lang: str = "en_XX", tgt_texts: Optional[List[str]] = None, tgt_lang: str = "ro_RO", **kwargs, ) -> BatchEncoding: self.src_lang = src_lang self.tgt_lang = tgt_lang return super().prepare_seq2seq_batch(src_texts, tgt_texts, **kwargs) def _switch_to_input_mode(self): return self.set_src_lang_special_tokens(self.src_lang) def _switch_to_target_mode(self): return self.set_tgt_lang_special_tokens(self.tgt_lang) def set_src_lang_special_tokens(self, src_lang: str) -> None: """Reset the special tokens to the source lang setting. prefix=[src_lang_code] and suffix=[eos].""" self.cur_lang_code_id = self.convert_tokens_to_ids(src_lang) self.prefix_tokens = [self.cur_lang_code_id] self.suffix_tokens = [self.eos_token_id] prefix_tokens_str = self.convert_ids_to_tokens(self.prefix_tokens) suffix_tokens_str = self.convert_ids_to_tokens(self.suffix_tokens) self._tokenizer.post_processor = processors.TemplateProcessing( single=prefix_tokens_str + ["$A"] + suffix_tokens_str, pair=prefix_tokens_str + ["$A", "$B"] + suffix_tokens_str, special_tokens=list(zip(prefix_tokens_str + suffix_tokens_str, self.prefix_tokens + self.suffix_tokens)), ) def set_tgt_lang_special_tokens(self, tgt_lang: str) -> None: """Reset the special tokens to the target language setting. prefix=[src_lang_code] and suffix=[eos].""" self.cur_lang_code_id = self.convert_tokens_to_ids(tgt_lang) self.prefix_tokens = [self.cur_lang_code_id] self.suffix_tokens = [self.eos_token_id] prefix_tokens_str = self.convert_ids_to_tokens(self.prefix_tokens) suffix_tokens_str = self.convert_ids_to_tokens(self.suffix_tokens) self._tokenizer.post_processor = processors.TemplateProcessing( single=prefix_tokens_str + ["$A"] + suffix_tokens_str, pair=prefix_tokens_str + ["$A", "$B"] + suffix_tokens_str, special_tokens=list(zip(prefix_tokens_str + suffix_tokens_str, self.prefix_tokens + self.suffix_tokens)), ) def _build_translation_inputs( self, raw_inputs, return_tensors: str, src_lang: Optional[str], tgt_lang: Optional[str], **extra_kwargs ): """Used by translation pipeline, to prepare inputs for the generate function""" if src_lang is None or tgt_lang is None: raise ValueError("Translation requires a `src_lang` and a `tgt_lang` for this model") self.src_lang = src_lang inputs = self(raw_inputs, add_special_tokens=True, return_tensors=return_tensors, **extra_kwargs) tgt_lang_id = self.convert_tokens_to_ids(tgt_lang) inputs["forced_bos_token_id"] = tgt_lang_id return inputs def save_vocabulary(self, save_directory: str, filename_prefix: Optional[str] = None) -> Tuple[str]: if not self.can_save_slow_tokenizer: raise ValueError( "Your fast tokenizer does not have the necessary information to save the vocabulary for a slow " "tokenizer." ) 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) return (out_vocab_file,) __all__ = ["MBart50TokenizerFast"]