# coding=utf-8 # Copyright Microsoft Research and The HuggingFace Inc. team. All rights reserved. # # 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 LayoutLMv2.""" import collections import os import sys import unicodedata from typing import Dict, List, Optional, Tuple, Union from ...tokenization_utils import AddedToken, PreTrainedTokenizer, _is_control, _is_punctuation, _is_whitespace from ...tokenization_utils_base import ( BatchEncoding, EncodedInput, PreTokenizedInput, TextInput, TextInputPair, TruncationStrategy, ) from ...utils import PaddingStrategy, TensorType, add_end_docstrings, logging logger = logging.get_logger(__name__) VOCAB_FILES_NAMES = {"vocab_file": "vocab.txt"} LAYOUTLMV2_ENCODE_KWARGS_DOCSTRING = r""" add_special_tokens (`bool`, *optional*, defaults to `True`): Whether or not to encode the sequences with the special tokens relative to their model. padding (`bool`, `str` or [`~file_utils.PaddingStrategy`], *optional*, defaults to `False`): Activates and controls padding. Accepts the following values: - `True` or `'longest'`: Pad to the longest sequence in the batch (or no padding if only a single sequence if provided). - `'max_length'`: Pad to a maximum length specified with the argument `max_length` or to the maximum acceptable input length for the model if that argument is not provided. - `False` or `'do_not_pad'` (default): No padding (i.e., can output a batch with sequences of different lengths). truncation (`bool`, `str` or [`~tokenization_utils_base.TruncationStrategy`], *optional*, defaults to `False`): Activates and controls truncation. Accepts the following values: - `True` or `'longest_first'`: Truncate to a maximum length specified with the argument `max_length` or to the maximum acceptable input length for the model if that argument is not provided. This will truncate token by token, removing a token from the longest sequence in the pair if a pair of sequences (or a batch of pairs) is provided. - `'only_first'`: Truncate to a maximum length specified with the argument `max_length` or to the maximum acceptable input length for the model if that argument is not provided. This will only truncate the first sequence of a pair if a pair of sequences (or a batch of pairs) is provided. - `'only_second'`: Truncate to a maximum length specified with the argument `max_length` or to the maximum acceptable input length for the model if that argument is not provided. This will only truncate the second sequence of a pair if a pair of sequences (or a batch of pairs) is provided. - `False` or `'do_not_truncate'` (default): No truncation (i.e., can output batch with sequence lengths greater than the model maximum admissible input size). max_length (`int`, *optional*): Controls the maximum length to use by one of the truncation/padding parameters. If left unset or set to `None`, this will use the predefined model maximum length if a maximum length is required by one of the truncation/padding parameters. If the model has no specific maximum input length (like XLNet) truncation/padding to a maximum length will be deactivated. stride (`int`, *optional*, defaults to 0): If set to a number along with `max_length`, the overflowing tokens returned when `return_overflowing_tokens=True` will contain some tokens from the end of the truncated sequence returned to provide some overlap between truncated and overflowing sequences. The value of this argument defines the number of overlapping tokens. pad_to_multiple_of (`int`, *optional*): If set will pad the sequence to a multiple of the provided value. This is especially useful to enable the use of Tensor Cores on NVIDIA hardware with compute capability `>= 7.5` (Volta). return_tensors (`str` or [`~file_utils.TensorType`], *optional*): If set, will return tensors instead of list of python integers. Acceptable values are: - `'tf'`: Return TensorFlow `tf.constant` objects. - `'pt'`: Return PyTorch `torch.Tensor` objects. - `'np'`: Return Numpy `np.ndarray` objects. """ LAYOUTLMV2_ENCODE_PLUS_ADDITIONAL_KWARGS_DOCSTRING = r""" return_token_type_ids (`bool`, *optional*): Whether to return token type IDs. If left to the default, will return the token type IDs according to the specific tokenizer's default, defined by the `return_outputs` attribute. [What are token type IDs?](../glossary#token-type-ids) return_attention_mask (`bool`, *optional*): Whether to return the attention mask. If left to the default, will return the attention mask according to the specific tokenizer's default, defined by the `return_outputs` attribute. [What are attention masks?](../glossary#attention-mask) return_overflowing_tokens (`bool`, *optional*, defaults to `False`): Whether or not to return overflowing token sequences. If a pair of sequences of input ids (or a batch of pairs) is provided with `truncation_strategy = longest_first` or `True`, an error is raised instead of returning overflowing tokens. return_special_tokens_mask (`bool`, *optional*, defaults to `False`): Whether or not to return special tokens mask information. return_offsets_mapping (`bool`, *optional*, defaults to `False`): Whether or not to return `(char_start, char_end)` for each token. This is only available on fast tokenizers inheriting from [`PreTrainedTokenizerFast`], if using Python's tokenizer, this method will raise `NotImplementedError`. return_length (`bool`, *optional*, defaults to `False`): Whether or not to return the lengths of the encoded inputs. verbose (`bool`, *optional*, defaults to `True`): Whether or not to print more information and warnings. **kwargs: passed to the `self.tokenize()` method Return: [`BatchEncoding`]: A [`BatchEncoding`] with the following fields: - **input_ids** -- List of token ids to be fed to a model. [What are input IDs?](../glossary#input-ids) - **bbox** -- List of bounding boxes to be fed to a model. - **token_type_ids** -- List of token type ids to be fed to a model (when `return_token_type_ids=True` or if *"token_type_ids"* is in `self.model_input_names`). [What are token type IDs?](../glossary#token-type-ids) - **attention_mask** -- List of indices specifying which tokens should be attended to by the model (when `return_attention_mask=True` or if *"attention_mask"* is in `self.model_input_names`). [What are attention masks?](../glossary#attention-mask) - **labels** -- List of labels to be fed to a model. (when `word_labels` is specified). - **overflowing_tokens** -- List of overflowing tokens sequences (when a `max_length` is specified and `return_overflowing_tokens=True`). - **num_truncated_tokens** -- Number of tokens truncated (when a `max_length` is specified and `return_overflowing_tokens=True`). - **special_tokens_mask** -- List of 0s and 1s, with 1 specifying added special tokens and 0 specifying regular sequence tokens (when `add_special_tokens=True` and `return_special_tokens_mask=True`). - **length** -- The length of the inputs (when `return_length=True`). """ def load_vocab(vocab_file): """Loads a vocabulary file into a dictionary.""" vocab = collections.OrderedDict() with open(vocab_file, "r", encoding="utf-8") as reader: tokens = reader.readlines() for index, token in enumerate(tokens): token = token.rstrip("\n") vocab[token] = index return vocab def whitespace_tokenize(text): """Runs basic whitespace cleaning and splitting on a piece of text.""" text = text.strip() if not text: return [] tokens = text.split() return tokens table = dict.fromkeys(i for i in range(sys.maxunicode) if unicodedata.category(chr(i)).startswith("P")) def subfinder(mylist, pattern): matches = [] indices = [] for idx, i in enumerate(range(len(mylist))): if mylist[i] == pattern[0] and mylist[i : i + len(pattern)] == pattern: matches.append(pattern) indices.append(idx) if matches: return matches[0], indices[0] else: return None, 0 class LayoutLMv2Tokenizer(PreTrainedTokenizer): r""" Construct a LayoutLMv2 tokenizer. Based on WordPiece. [`LayoutLMv2Tokenizer`] can be used to turn words, word-level bounding boxes and optional word labels to token-level `input_ids`, `attention_mask`, `token_type_ids`, `bbox`, and optional `labels` (for token classification). This tokenizer inherits from [`PreTrainedTokenizer`] which contains most of the main methods. Users should refer to this superclass for more information regarding those methods. [`LayoutLMv2Tokenizer`] runs end-to-end tokenization: punctuation splitting and wordpiece. It also turns the word-level bounding boxes into token-level bounding boxes. """ vocab_files_names = VOCAB_FILES_NAMES def __init__( self, vocab_file, do_lower_case=True, do_basic_tokenize=True, never_split=None, unk_token="[UNK]", sep_token="[SEP]", pad_token="[PAD]", cls_token="[CLS]", mask_token="[MASK]", cls_token_box=[0, 0, 0, 0], sep_token_box=[1000, 1000, 1000, 1000], pad_token_box=[0, 0, 0, 0], pad_token_label=-100, only_label_first_subword=True, tokenize_chinese_chars=True, strip_accents=None, model_max_length: int = 512, additional_special_tokens: Optional[List[str]] = None, **kwargs, ): sep_token = AddedToken(sep_token, special=True) if isinstance(sep_token, str) else sep_token unk_token = AddedToken(unk_token, special=True) if isinstance(unk_token, str) else unk_token pad_token = AddedToken(pad_token, special=True) if isinstance(pad_token, str) else pad_token cls_token = AddedToken(cls_token, special=True) if isinstance(cls_token, str) else cls_token mask_token = AddedToken(mask_token, special=True) if isinstance(mask_token, str) else mask_token if not os.path.isfile(vocab_file): raise ValueError( f"Can't find a vocabulary file at path '{vocab_file}'. To load the vocabulary from a Google pretrained" " model use `tokenizer = BertTokenizer.from_pretrained(PRETRAINED_MODEL_NAME)`" ) self.vocab = load_vocab(vocab_file) self.ids_to_tokens = collections.OrderedDict([(ids, tok) for tok, ids in self.vocab.items()]) self.do_basic_tokenize = do_basic_tokenize if do_basic_tokenize: self.basic_tokenizer = BasicTokenizer( do_lower_case=do_lower_case, never_split=never_split, tokenize_chinese_chars=tokenize_chinese_chars, strip_accents=strip_accents, ) self.wordpiece_tokenizer = WordpieceTokenizer(vocab=self.vocab, unk_token=str(unk_token)) # additional properties self.cls_token_box = cls_token_box self.sep_token_box = sep_token_box self.pad_token_box = pad_token_box self.pad_token_label = pad_token_label self.only_label_first_subword = only_label_first_subword super().__init__( do_lower_case=do_lower_case, do_basic_tokenize=do_basic_tokenize, never_split=never_split, unk_token=unk_token, sep_token=sep_token, pad_token=pad_token, cls_token=cls_token, mask_token=mask_token, cls_token_box=cls_token_box, sep_token_box=sep_token_box, pad_token_box=pad_token_box, pad_token_label=pad_token_label, only_label_first_subword=only_label_first_subword, tokenize_chinese_chars=tokenize_chinese_chars, strip_accents=strip_accents, model_max_length=model_max_length, additional_special_tokens=additional_special_tokens, **kwargs, ) @property def do_lower_case(self): return self.basic_tokenizer.do_lower_case @property def vocab_size(self): return len(self.vocab) def get_vocab(self): return dict(self.vocab, **self.added_tokens_encoder) def _tokenize(self, text): split_tokens = [] if self.do_basic_tokenize: for token in self.basic_tokenizer.tokenize(text, never_split=self.all_special_tokens): # If the token is part of the never_split set if token in self.basic_tokenizer.never_split: split_tokens.append(token) else: split_tokens += self.wordpiece_tokenizer.tokenize(token) else: split_tokens = self.wordpiece_tokenizer.tokenize(text) return split_tokens def _convert_token_to_id(self, token): """Converts a token (str) in an id using the vocab.""" return self.vocab.get(token, self.vocab.get(self.unk_token)) def _convert_id_to_token(self, index): """Converts an index (integer) in a token (str) using the vocab.""" return self.ids_to_tokens.get(index, self.unk_token) def convert_tokens_to_string(self, tokens): """Converts a sequence of tokens (string) in a single string.""" out_string = " ".join(tokens).replace(" ##", "").strip() return out_string 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 BERT sequence has the following format: - single sequence: `[CLS] X [SEP]` - pair of sequences: `[CLS] A [SEP] B [SEP]` 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.cls_token_id] + token_ids_0 + [self.sep_token_id] cls = [self.cls_token_id] sep = [self.sep_token_id] return cls + token_ids_0 + sep + token_ids_1 + sep 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 ) if token_ids_1 is not None: return [1] + ([0] * len(token_ids_0)) + [1] + ([0] * len(token_ids_1)) + [1] return [1] + ([0] * len(token_ids_0)) + [1] 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. A BERT sequence pair mask has the following format: :: 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 | first sequence | second sequence | If `token_ids_1` is `None`, this method only returns the first portion of the mask (0s). 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 [token type IDs](../glossary#token-type-ids) according to the given sequence(s). """ 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) * [0] + len(token_ids_1 + sep) * [1] def save_vocabulary(self, save_directory: str, filename_prefix: Optional[str] = None) -> Tuple[str]: index = 0 if os.path.isdir(save_directory): vocab_file = os.path.join( save_directory, (filename_prefix + "-" if filename_prefix else "") + VOCAB_FILES_NAMES["vocab_file"] ) else: vocab_file = (filename_prefix + "-" if filename_prefix else "") + save_directory with open(vocab_file, "w", encoding="utf-8") as writer: for token, token_index in sorted(self.vocab.items(), key=lambda kv: kv[1]): if index != token_index: logger.warning( f"Saving vocabulary to {vocab_file}: vocabulary indices are not consecutive." " Please check that the vocabulary is not corrupted!" ) index = token_index writer.write(token + "\n") index += 1 return (vocab_file,) @add_end_docstrings(LAYOUTLMV2_ENCODE_KWARGS_DOCSTRING, LAYOUTLMV2_ENCODE_PLUS_ADDITIONAL_KWARGS_DOCSTRING) def __call__( self, text: Union[TextInput, PreTokenizedInput, List[TextInput], List[PreTokenizedInput]], text_pair: Optional[Union[PreTokenizedInput, List[PreTokenizedInput]]] = None, boxes: Union[List[List[int]], List[List[List[int]]]] = None, word_labels: Optional[Union[List[int], List[List[int]]]] = None, add_special_tokens: bool = True, padding: Union[bool, str, PaddingStrategy] = False, truncation: Union[bool, str, TruncationStrategy] = None, max_length: Optional[int] = None, stride: int = 0, pad_to_multiple_of: Optional[int] = None, return_tensors: Optional[Union[str, TensorType]] = None, return_token_type_ids: Optional[bool] = None, return_attention_mask: Optional[bool] = None, return_overflowing_tokens: bool = False, return_special_tokens_mask: bool = False, return_offsets_mapping: bool = False, return_length: bool = False, verbose: bool = True, **kwargs, ) -> BatchEncoding: """ Main method to tokenize and prepare for the model one or several sequence(s) or one or several pair(s) of sequences with word-level normalized bounding boxes and optional labels. Args: text (`str`, `List[str]`, `List[List[str]]`): The sequence or batch of sequences to be encoded. Each sequence can be a string, a list of strings (words of a single example or questions of a batch of examples) or a list of list of strings (batch of words). text_pair (`List[str]`, `List[List[str]]`): The sequence or batch of sequences to be encoded. Each sequence should be a list of strings (pretokenized string). boxes (`List[List[int]]`, `List[List[List[int]]]`): Word-level bounding boxes. Each bounding box should be normalized to be on a 0-1000 scale. word_labels (`List[int]`, `List[List[int]]`, *optional*): Word-level integer labels (for token classification tasks such as FUNSD, CORD). """ # Input type checking for clearer error def _is_valid_text_input(t): if isinstance(t, str): # Strings are fine return True elif isinstance(t, (list, tuple)): # List are fine as long as they are... if len(t) == 0: # ... empty return True elif isinstance(t[0], str): # ... list of strings return True elif isinstance(t[0], (list, tuple)): # ... list with an empty list or with a list of strings return len(t[0]) == 0 or isinstance(t[0][0], str) else: return False else: return False if text_pair is not None: # in case text + text_pair are provided, text = questions, text_pair = words if not _is_valid_text_input(text): raise ValueError("text input must of type `str` (single example) or `List[str]` (batch of examples). ") if not isinstance(text_pair, (list, tuple)): raise ValueError( "Words must be of type `List[str]` (single pretokenized example), " "or `List[List[str]]` (batch of pretokenized examples)." ) else: # in case only text is provided => must be words if not isinstance(text, (list, tuple)): raise ValueError( "Words must be of type `List[str]` (single pretokenized example), " "or `List[List[str]]` (batch of pretokenized examples)." ) if text_pair is not None: is_batched = isinstance(text, (list, tuple)) else: is_batched = isinstance(text, (list, tuple)) and text and isinstance(text[0], (list, tuple)) words = text if text_pair is None else text_pair if boxes is None: raise ValueError("You must provide corresponding bounding boxes") if is_batched: if len(words) != len(boxes): raise ValueError("You must provide words and boxes for an equal amount of examples") for words_example, boxes_example in zip(words, boxes): if len(words_example) != len(boxes_example): raise ValueError("You must provide as many words as there are bounding boxes") else: if len(words) != len(boxes): raise ValueError("You must provide as many words as there are bounding boxes") if is_batched: if text_pair is not None and len(text) != len(text_pair): raise ValueError( f"batch length of `text`: {len(text)} does not match batch length of `text_pair`:" f" {len(text_pair)}." ) batch_text_or_text_pairs = list(zip(text, text_pair)) if text_pair is not None else text is_pair = bool(text_pair is not None) return self.batch_encode_plus( batch_text_or_text_pairs=batch_text_or_text_pairs, is_pair=is_pair, boxes=boxes, word_labels=word_labels, add_special_tokens=add_special_tokens, padding=padding, truncation=truncation, max_length=max_length, stride=stride, pad_to_multiple_of=pad_to_multiple_of, return_tensors=return_tensors, return_token_type_ids=return_token_type_ids, return_attention_mask=return_attention_mask, return_overflowing_tokens=return_overflowing_tokens, return_special_tokens_mask=return_special_tokens_mask, return_offsets_mapping=return_offsets_mapping, return_length=return_length, verbose=verbose, **kwargs, ) else: return self.encode_plus( text=text, text_pair=text_pair, boxes=boxes, word_labels=word_labels, add_special_tokens=add_special_tokens, padding=padding, truncation=truncation, max_length=max_length, stride=stride, pad_to_multiple_of=pad_to_multiple_of, return_tensors=return_tensors, return_token_type_ids=return_token_type_ids, return_attention_mask=return_attention_mask, return_overflowing_tokens=return_overflowing_tokens, return_special_tokens_mask=return_special_tokens_mask, return_offsets_mapping=return_offsets_mapping, return_length=return_length, verbose=verbose, **kwargs, ) @add_end_docstrings(LAYOUTLMV2_ENCODE_KWARGS_DOCSTRING, LAYOUTLMV2_ENCODE_PLUS_ADDITIONAL_KWARGS_DOCSTRING) def batch_encode_plus( self, batch_text_or_text_pairs: Union[ List[TextInput], List[TextInputPair], List[PreTokenizedInput], ], is_pair: bool = None, boxes: Optional[List[List[List[int]]]] = None, word_labels: Optional[Union[List[int], List[List[int]]]] = None, add_special_tokens: bool = True, padding: Union[bool, str, PaddingStrategy] = False, truncation: Union[bool, str, TruncationStrategy] = None, max_length: Optional[int] = None, stride: int = 0, pad_to_multiple_of: Optional[int] = None, return_tensors: Optional[Union[str, TensorType]] = None, return_token_type_ids: Optional[bool] = None, return_attention_mask: Optional[bool] = None, return_overflowing_tokens: bool = False, return_special_tokens_mask: bool = False, return_offsets_mapping: bool = False, return_length: bool = False, verbose: bool = True, **kwargs, ) -> BatchEncoding: # Backward compatibility for 'truncation_strategy', 'pad_to_max_length' padding_strategy, truncation_strategy, max_length, kwargs = self._get_padding_truncation_strategies( padding=padding, truncation=truncation, max_length=max_length, pad_to_multiple_of=pad_to_multiple_of, verbose=verbose, **kwargs, ) return self._batch_encode_plus( batch_text_or_text_pairs=batch_text_or_text_pairs, is_pair=is_pair, boxes=boxes, word_labels=word_labels, add_special_tokens=add_special_tokens, padding_strategy=padding_strategy, truncation_strategy=truncation_strategy, max_length=max_length, stride=stride, pad_to_multiple_of=pad_to_multiple_of, return_tensors=return_tensors, return_token_type_ids=return_token_type_ids, return_attention_mask=return_attention_mask, return_overflowing_tokens=return_overflowing_tokens, return_special_tokens_mask=return_special_tokens_mask, return_offsets_mapping=return_offsets_mapping, return_length=return_length, verbose=verbose, **kwargs, ) def _batch_encode_plus( self, batch_text_or_text_pairs: Union[ List[TextInput], List[TextInputPair], List[PreTokenizedInput], ], is_pair: bool = None, boxes: Optional[List[List[List[int]]]] = None, word_labels: Optional[List[List[int]]] = None, add_special_tokens: bool = True, padding_strategy: PaddingStrategy = PaddingStrategy.DO_NOT_PAD, truncation_strategy: TruncationStrategy = TruncationStrategy.DO_NOT_TRUNCATE, max_length: Optional[int] = None, stride: int = 0, pad_to_multiple_of: Optional[int] = None, return_tensors: Optional[Union[str, TensorType]] = None, return_token_type_ids: Optional[bool] = None, return_attention_mask: Optional[bool] = None, return_overflowing_tokens: bool = False, return_special_tokens_mask: bool = False, return_offsets_mapping: bool = False, return_length: bool = False, verbose: bool = True, **kwargs, ) -> BatchEncoding: if return_offsets_mapping: raise NotImplementedError( "return_offset_mapping is not available when using Python tokenizers. " "To use this feature, change your tokenizer to one deriving from " "transformers.PreTrainedTokenizerFast." ) batch_outputs = self._batch_prepare_for_model( batch_text_or_text_pairs=batch_text_or_text_pairs, is_pair=is_pair, boxes=boxes, word_labels=word_labels, add_special_tokens=add_special_tokens, padding_strategy=padding_strategy, truncation_strategy=truncation_strategy, max_length=max_length, stride=stride, pad_to_multiple_of=pad_to_multiple_of, return_attention_mask=return_attention_mask, return_token_type_ids=return_token_type_ids, return_overflowing_tokens=return_overflowing_tokens, return_special_tokens_mask=return_special_tokens_mask, return_length=return_length, return_tensors=return_tensors, verbose=verbose, ) return BatchEncoding(batch_outputs) @add_end_docstrings(LAYOUTLMV2_ENCODE_KWARGS_DOCSTRING, LAYOUTLMV2_ENCODE_PLUS_ADDITIONAL_KWARGS_DOCSTRING) def _batch_prepare_for_model( self, batch_text_or_text_pairs, is_pair: bool = None, boxes: Optional[List[List[int]]] = None, word_labels: Optional[List[List[int]]] = None, add_special_tokens: bool = True, padding_strategy: PaddingStrategy = PaddingStrategy.DO_NOT_PAD, truncation_strategy: TruncationStrategy = TruncationStrategy.DO_NOT_TRUNCATE, max_length: Optional[int] = None, stride: int = 0, pad_to_multiple_of: Optional[int] = None, return_tensors: Optional[str] = None, return_token_type_ids: Optional[bool] = None, return_attention_mask: Optional[bool] = None, return_overflowing_tokens: bool = False, return_special_tokens_mask: bool = False, return_length: bool = False, verbose: bool = True, ) -> BatchEncoding: """ Prepares a sequence of input id, or a pair of sequences of inputs ids so that it can be used by the model. It adds special tokens, truncates sequences if overflowing while taking into account the special tokens and manages a moving window (with user defined stride) for overflowing tokens. Args: batch_ids_pairs: list of tokenized input ids or input ids pairs """ batch_outputs = {} for idx, example in enumerate(zip(batch_text_or_text_pairs, boxes)): batch_text_or_text_pair, boxes_example = example outputs = self.prepare_for_model( batch_text_or_text_pair[0] if is_pair else batch_text_or_text_pair, batch_text_or_text_pair[1] if is_pair else None, boxes_example, word_labels=word_labels[idx] if word_labels is not None else None, add_special_tokens=add_special_tokens, padding=PaddingStrategy.DO_NOT_PAD.value, # we pad in batch afterward truncation=truncation_strategy.value, max_length=max_length, stride=stride, pad_to_multiple_of=None, # we pad in batch afterward return_attention_mask=False, # we pad in batch afterward return_token_type_ids=return_token_type_ids, return_overflowing_tokens=return_overflowing_tokens, return_special_tokens_mask=return_special_tokens_mask, return_length=return_length, return_tensors=None, # We convert the whole batch to tensors at the end prepend_batch_axis=False, verbose=verbose, ) for key, value in outputs.items(): if key not in batch_outputs: batch_outputs[key] = [] batch_outputs[key].append(value) batch_outputs = self.pad( batch_outputs, padding=padding_strategy.value, max_length=max_length, pad_to_multiple_of=pad_to_multiple_of, return_attention_mask=return_attention_mask, ) batch_outputs = BatchEncoding(batch_outputs, tensor_type=return_tensors) return batch_outputs @add_end_docstrings(LAYOUTLMV2_ENCODE_KWARGS_DOCSTRING) def encode( self, text: Union[TextInput, PreTokenizedInput], text_pair: Optional[PreTokenizedInput] = None, boxes: Optional[List[List[int]]] = None, word_labels: Optional[List[int]] = None, add_special_tokens: bool = True, padding: Union[bool, str, PaddingStrategy] = False, truncation: Union[bool, str, TruncationStrategy] = None, max_length: Optional[int] = None, stride: int = 0, pad_to_multiple_of: Optional[int] = None, return_tensors: Optional[Union[str, TensorType]] = None, return_token_type_ids: Optional[bool] = None, return_attention_mask: Optional[bool] = None, return_overflowing_tokens: bool = False, return_special_tokens_mask: bool = False, return_offsets_mapping: bool = False, return_length: bool = False, verbose: bool = True, **kwargs, ) -> List[int]: encoded_inputs = self.encode_plus( text=text, text_pair=text_pair, boxes=boxes, word_labels=word_labels, add_special_tokens=add_special_tokens, padding=padding, truncation=truncation, max_length=max_length, stride=stride, pad_to_multiple_of=pad_to_multiple_of, return_tensors=return_tensors, return_token_type_ids=return_token_type_ids, return_attention_mask=return_attention_mask, return_overflowing_tokens=return_overflowing_tokens, return_special_tokens_mask=return_special_tokens_mask, return_offsets_mapping=return_offsets_mapping, return_length=return_length, verbose=verbose, **kwargs, ) return encoded_inputs["input_ids"] @add_end_docstrings(LAYOUTLMV2_ENCODE_KWARGS_DOCSTRING, LAYOUTLMV2_ENCODE_PLUS_ADDITIONAL_KWARGS_DOCSTRING) def encode_plus( self, text: Union[TextInput, PreTokenizedInput], text_pair: Optional[PreTokenizedInput] = None, boxes: Optional[List[List[int]]] = None, word_labels: Optional[List[int]] = None, add_special_tokens: bool = True, padding: Union[bool, str, PaddingStrategy] = False, truncation: Union[bool, str, TruncationStrategy] = None, max_length: Optional[int] = None, stride: int = 0, pad_to_multiple_of: Optional[int] = None, return_tensors: Optional[Union[str, TensorType]] = None, return_token_type_ids: Optional[bool] = None, return_attention_mask: Optional[bool] = None, return_overflowing_tokens: bool = False, return_special_tokens_mask: bool = False, return_offsets_mapping: bool = False, return_length: bool = False, verbose: bool = True, **kwargs, ) -> BatchEncoding: """ Tokenize and prepare for the model a sequence or a pair of sequences. .. warning:: This method is deprecated, `__call__` should be used instead. Args: text (`str`, `List[str]`, `List[List[str]]`): The first sequence to be encoded. This can be a string, a list of strings or a list of list of strings. text_pair (`List[str]` or `List[int]`, *optional*): Optional second sequence to be encoded. This can be a list of strings (words of a single example) or a list of list of strings (words of a batch of examples). """ # Backward compatibility for 'truncation_strategy', 'pad_to_max_length' padding_strategy, truncation_strategy, max_length, kwargs = self._get_padding_truncation_strategies( padding=padding, truncation=truncation, max_length=max_length, pad_to_multiple_of=pad_to_multiple_of, verbose=verbose, **kwargs, ) return self._encode_plus( text=text, boxes=boxes, text_pair=text_pair, word_labels=word_labels, add_special_tokens=add_special_tokens, padding_strategy=padding_strategy, truncation_strategy=truncation_strategy, max_length=max_length, stride=stride, pad_to_multiple_of=pad_to_multiple_of, return_tensors=return_tensors, return_token_type_ids=return_token_type_ids, return_attention_mask=return_attention_mask, return_overflowing_tokens=return_overflowing_tokens, return_special_tokens_mask=return_special_tokens_mask, return_offsets_mapping=return_offsets_mapping, return_length=return_length, verbose=verbose, **kwargs, ) def _encode_plus( self, text: Union[TextInput, PreTokenizedInput], text_pair: Optional[PreTokenizedInput] = None, boxes: Optional[List[List[int]]] = None, word_labels: Optional[List[int]] = None, add_special_tokens: bool = True, padding_strategy: PaddingStrategy = PaddingStrategy.DO_NOT_PAD, truncation_strategy: TruncationStrategy = TruncationStrategy.DO_NOT_TRUNCATE, max_length: Optional[int] = None, stride: int = 0, pad_to_multiple_of: Optional[int] = None, return_tensors: Optional[Union[str, TensorType]] = None, return_token_type_ids: Optional[bool] = None, return_attention_mask: Optional[bool] = None, return_overflowing_tokens: bool = False, return_special_tokens_mask: bool = False, return_offsets_mapping: bool = False, return_length: bool = False, verbose: bool = True, **kwargs, ) -> BatchEncoding: if return_offsets_mapping: raise NotImplementedError( "return_offset_mapping is not available when using Python tokenizers. " "To use this feature, change your tokenizer to one deriving from " "transformers.PreTrainedTokenizerFast. " "More information on available tokenizers at " "https://github.com/huggingface/transformers/pull/2674" ) return self.prepare_for_model( text=text, text_pair=text_pair, boxes=boxes, word_labels=word_labels, add_special_tokens=add_special_tokens, padding=padding_strategy.value, truncation=truncation_strategy.value, max_length=max_length, stride=stride, pad_to_multiple_of=pad_to_multiple_of, return_tensors=return_tensors, prepend_batch_axis=True, return_attention_mask=return_attention_mask, return_token_type_ids=return_token_type_ids, return_overflowing_tokens=return_overflowing_tokens, return_special_tokens_mask=return_special_tokens_mask, return_length=return_length, verbose=verbose, ) @add_end_docstrings(LAYOUTLMV2_ENCODE_KWARGS_DOCSTRING, LAYOUTLMV2_ENCODE_PLUS_ADDITIONAL_KWARGS_DOCSTRING) def prepare_for_model( self, text: Union[TextInput, PreTokenizedInput], text_pair: Optional[PreTokenizedInput] = None, boxes: Optional[List[List[int]]] = None, word_labels: Optional[List[int]] = None, add_special_tokens: bool = True, padding: Union[bool, str, PaddingStrategy] = False, truncation: Union[bool, str, TruncationStrategy] = None, max_length: Optional[int] = None, stride: int = 0, pad_to_multiple_of: Optional[int] = None, return_tensors: Optional[Union[str, TensorType]] = None, return_token_type_ids: Optional[bool] = None, return_attention_mask: Optional[bool] = None, return_overflowing_tokens: bool = False, return_special_tokens_mask: bool = False, return_offsets_mapping: bool = False, return_length: bool = False, verbose: bool = True, prepend_batch_axis: bool = False, **kwargs, ) -> BatchEncoding: """ Prepares a sequence or a pair of sequences so that it can be used by the model. It adds special tokens, truncates sequences if overflowing while taking into account the special tokens and manages a moving window (with user defined stride) for overflowing tokens. Please Note, for *text_pair* different than `None` and *truncation_strategy = longest_first* or `True`, it is not possible to return overflowing tokens. Such a combination of arguments will raise an error. Word-level `boxes` are turned into token-level `bbox`. If provided, word-level `word_labels` are turned into token-level `labels`. The word label is used for the first token of the word, while remaining tokens are labeled with -100, such that they will be ignored by the loss function. Args: text (`str`, `List[str]`, `List[List[str]]`): The first sequence to be encoded. This can be a string, a list of strings or a list of list of strings. text_pair (`List[str]` or `List[int]`, *optional*): Optional second sequence to be encoded. This can be a list of strings (words of a single example) or a list of list of strings (words of a batch of examples). """ # Backward compatibility for 'truncation_strategy', 'pad_to_max_length' padding_strategy, truncation_strategy, max_length, kwargs = self._get_padding_truncation_strategies( padding=padding, truncation=truncation, max_length=max_length, pad_to_multiple_of=pad_to_multiple_of, verbose=verbose, **kwargs, ) tokens = [] pair_tokens = [] token_boxes = [] pair_token_boxes = [] labels = [] if text_pair is None: if word_labels is None: # CASE 1: document image classification (training + inference) + CASE 2: token classification (inference) for word, box in zip(text, boxes): if len(word) < 1: # skip empty words continue word_tokens = self.tokenize(word) tokens.extend(word_tokens) token_boxes.extend([box] * len(word_tokens)) else: # CASE 2: token classification (training) for word, box, label in zip(text, boxes, word_labels): if len(word) < 1: # skip empty words continue word_tokens = self.tokenize(word) tokens.extend(word_tokens) token_boxes.extend([box] * len(word_tokens)) if self.only_label_first_subword: # Use the real label id for the first token of the word, and padding ids for the remaining tokens labels.extend([label] + [self.pad_token_label] * (len(word_tokens) - 1)) else: labels.extend([label] * len(word_tokens)) else: # CASE 3: document visual question answering (inference) # text = question # text_pair = words tokens = self.tokenize(text) token_boxes = [self.pad_token_box for _ in range(len(tokens))] for word, box in zip(text_pair, boxes): if len(word) < 1: # skip empty words continue word_tokens = self.tokenize(word) pair_tokens.extend(word_tokens) pair_token_boxes.extend([box] * len(word_tokens)) # Create ids + pair_ids ids = self.convert_tokens_to_ids(tokens) pair_ids = self.convert_tokens_to_ids(pair_tokens) if pair_tokens else None if ( return_overflowing_tokens and truncation_strategy == TruncationStrategy.LONGEST_FIRST and pair_ids is not None ): raise ValueError( "Not possible to return overflowing tokens for pair of sequences with the " "`longest_first`. Please select another truncation strategy than `longest_first`, " "for instance `only_second` or `only_first`." ) # Compute the total size of the returned encodings pair = bool(pair_ids is not None) len_ids = len(ids) len_pair_ids = len(pair_ids) if pair else 0 total_len = len_ids + len_pair_ids + (self.num_special_tokens_to_add(pair=pair) if add_special_tokens else 0) # Truncation: Handle max sequence length overflowing_tokens = [] overflowing_token_boxes = [] overflowing_labels = [] if truncation_strategy != TruncationStrategy.DO_NOT_TRUNCATE and max_length and total_len > max_length: ( ids, token_boxes, pair_ids, pair_token_boxes, labels, overflowing_tokens, overflowing_token_boxes, overflowing_labels, ) = self.truncate_sequences( ids, token_boxes, pair_ids=pair_ids, pair_token_boxes=pair_token_boxes, labels=labels, num_tokens_to_remove=total_len - max_length, truncation_strategy=truncation_strategy, stride=stride, ) if return_token_type_ids and not add_special_tokens: raise ValueError( "Asking to return token_type_ids while setting add_special_tokens to False " "results in an undefined behavior. Please set add_special_tokens to True or " "set return_token_type_ids to None." ) # Load from model defaults if return_token_type_ids is None: return_token_type_ids = "token_type_ids" in self.model_input_names if return_attention_mask is None: return_attention_mask = "attention_mask" in self.model_input_names encoded_inputs = {} if return_overflowing_tokens: encoded_inputs["overflowing_tokens"] = overflowing_tokens encoded_inputs["overflowing_token_boxes"] = overflowing_token_boxes encoded_inputs["overflowing_labels"] = overflowing_labels encoded_inputs["num_truncated_tokens"] = total_len - max_length # Add special tokens if add_special_tokens: sequence = self.build_inputs_with_special_tokens(ids, pair_ids) token_type_ids = self.create_token_type_ids_from_sequences(ids, pair_ids) token_boxes = [self.cls_token_box] + token_boxes + [self.sep_token_box] if pair_token_boxes: pair_token_boxes = pair_token_boxes + [self.sep_token_box] if labels: labels = [self.pad_token_label] + labels + [self.pad_token_label] else: sequence = ids + pair_ids if pair else ids token_type_ids = [0] * len(ids) + ([0] * len(pair_ids) if pair else []) # Build output dictionary encoded_inputs["input_ids"] = sequence encoded_inputs["bbox"] = token_boxes + pair_token_boxes if return_token_type_ids: encoded_inputs["token_type_ids"] = token_type_ids if return_special_tokens_mask: if add_special_tokens: encoded_inputs["special_tokens_mask"] = self.get_special_tokens_mask(ids, pair_ids) else: encoded_inputs["special_tokens_mask"] = [0] * len(sequence) if labels: encoded_inputs["labels"] = labels # Check lengths self._eventual_warn_about_too_long_sequence(encoded_inputs["input_ids"], max_length, verbose) # Padding if padding_strategy != PaddingStrategy.DO_NOT_PAD or return_attention_mask: encoded_inputs = self.pad( encoded_inputs, max_length=max_length, padding=padding_strategy.value, pad_to_multiple_of=pad_to_multiple_of, return_attention_mask=return_attention_mask, ) if return_length: encoded_inputs["length"] = len(encoded_inputs["input_ids"]) batch_outputs = BatchEncoding( encoded_inputs, tensor_type=return_tensors, prepend_batch_axis=prepend_batch_axis ) return batch_outputs def truncate_sequences( self, ids: List[int], token_boxes: List[List[int]], pair_ids: Optional[List[int]] = None, pair_token_boxes: Optional[List[List[int]]] = None, labels: Optional[List[int]] = None, num_tokens_to_remove: int = 0, truncation_strategy: Union[str, TruncationStrategy] = "longest_first", stride: int = 0, ) -> Tuple[List[int], List[int], List[int]]: """ Truncates a sequence pair in-place following the strategy. Args: ids (`List[int]`): Tokenized input ids of the first sequence. Can be obtained from a string by chaining the `tokenize` and `convert_tokens_to_ids` methods. token_boxes (`List[List[int]]`): Bounding boxes of the first sequence. pair_ids (`List[int]`, *optional*): Tokenized input ids of the second sequence. Can be obtained from a string by chaining the `tokenize` and `convert_tokens_to_ids` methods. pair_token_boxes (`List[List[int]]`, *optional*): Bounding boxes of the second sequence. labels (`List[int]`, *optional*): Labels of the first sequence (for token classification tasks). num_tokens_to_remove (`int`, *optional*, defaults to 0): Number of tokens to remove using the truncation strategy. truncation_strategy (`str` or [`~tokenization_utils_base.TruncationStrategy`], *optional*, defaults to `False`): The strategy to follow for truncation. Can be: - `'longest_first'`: Truncate to a maximum length specified with the argument `max_length` or to the maximum acceptable input length for the model if that argument is not provided. This will truncate token by token, removing a token from the longest sequence in the pair if a pair of sequences (or a batch of pairs) is provided. - `'only_first'`: Truncate to a maximum length specified with the argument `max_length` or to the maximum acceptable input length for the model if that argument is not provided. This will only truncate the first sequence of a pair if a pair of sequences (or a batch of pairs) is provided. - `'only_second'`: Truncate to a maximum length specified with the argument `max_length` or to the maximum acceptable input length for the model if that argument is not provided. This will only truncate the second sequence of a pair if a pair of sequences (or a batch of pairs) is provided. - `'do_not_truncate'` (default): No truncation (i.e., can output batch with sequence lengths greater than the model maximum admissible input size). stride (`int`, *optional*, defaults to 0): If set to a positive number, the overflowing tokens returned will contain some tokens from the main sequence returned. The value of this argument defines the number of additional tokens. Returns: `Tuple[List[int], List[int], List[int]]`: The truncated `ids`, the truncated `pair_ids` and the list of overflowing tokens. Note: The *longest_first* strategy returns empty list of overflowing tokens if a pair of sequences (or a batch of pairs) is provided. """ if num_tokens_to_remove <= 0: return ids, token_boxes, pair_ids, pair_token_boxes, labels, [], [], [] if not isinstance(truncation_strategy, TruncationStrategy): truncation_strategy = TruncationStrategy(truncation_strategy) overflowing_tokens = [] overflowing_token_boxes = [] overflowing_labels = [] if truncation_strategy == TruncationStrategy.ONLY_FIRST or ( truncation_strategy == TruncationStrategy.LONGEST_FIRST and pair_ids is None ): if len(ids) > num_tokens_to_remove: window_len = min(len(ids), stride + num_tokens_to_remove) overflowing_tokens = ids[-window_len:] overflowing_token_boxes = token_boxes[-window_len:] overflowing_labels = labels[-window_len:] ids = ids[:-num_tokens_to_remove] token_boxes = token_boxes[:-num_tokens_to_remove] labels = labels[:-num_tokens_to_remove] else: error_msg = ( f"We need to remove {num_tokens_to_remove} to truncate the input " f"but the first sequence has a length {len(ids)}. " ) if truncation_strategy == TruncationStrategy.ONLY_FIRST: error_msg = ( error_msg + "Please select another truncation strategy than " f"{truncation_strategy}, for instance 'longest_first' or 'only_second'." ) logger.error(error_msg) elif truncation_strategy == TruncationStrategy.LONGEST_FIRST: logger.warning( "Be aware, overflowing tokens are not returned for the setting you have chosen," f" i.e. sequence pairs with the '{TruncationStrategy.LONGEST_FIRST.value}' " "truncation strategy. So the returned list will always be empty even if some " "tokens have been removed." ) for _ in range(num_tokens_to_remove): if pair_ids is None or len(ids) > len(pair_ids): ids = ids[:-1] token_boxes = token_boxes[:-1] labels = labels[:-1] else: pair_ids = pair_ids[:-1] pair_token_boxes = pair_token_boxes[:-1] elif truncation_strategy == TruncationStrategy.ONLY_SECOND and pair_ids is not None: if len(pair_ids) > num_tokens_to_remove: window_len = min(len(pair_ids), stride + num_tokens_to_remove) overflowing_tokens = pair_ids[-window_len:] overflowing_token_boxes = pair_token_boxes[-window_len:] pair_ids = pair_ids[:-num_tokens_to_remove] pair_token_boxes = pair_token_boxes[:-num_tokens_to_remove] else: logger.error( f"We need to remove {num_tokens_to_remove} to truncate the input " f"but the second sequence has a length {len(pair_ids)}. " f"Please select another truncation strategy than {truncation_strategy}, " "for instance 'longest_first' or 'only_first'." ) return ( ids, token_boxes, pair_ids, pair_token_boxes, labels, overflowing_tokens, overflowing_token_boxes, overflowing_labels, ) def _pad( self, encoded_inputs: Union[Dict[str, EncodedInput], BatchEncoding], max_length: Optional[int] = None, padding_strategy: PaddingStrategy = PaddingStrategy.DO_NOT_PAD, pad_to_multiple_of: Optional[int] = None, return_attention_mask: Optional[bool] = None, ) -> dict: """ Pad encoded inputs (on left/right and up to predefined length or max length in the batch) Args: encoded_inputs: Dictionary of tokenized inputs (`List[int]`) or batch of tokenized inputs (`List[List[int]]`). max_length: maximum length of the returned list and optionally padding length (see below). Will truncate by taking into account the special tokens. padding_strategy: PaddingStrategy to use for padding. - PaddingStrategy.LONGEST Pad to the longest sequence in the batch - PaddingStrategy.MAX_LENGTH: Pad to the max length (default) - PaddingStrategy.DO_NOT_PAD: Do not pad The tokenizer padding sides are defined in self.padding_side: - 'left': pads on the left of the sequences - 'right': pads on the right of the sequences pad_to_multiple_of: (optional) Integer if set will pad the sequence to a multiple of the provided value. This is especially useful to enable the use of Tensor Core on NVIDIA hardware with compute capability `>= 7.5` (Volta). return_attention_mask: (optional) Set to False to avoid returning attention mask (default: set to model specifics) """ # Load from model defaults if return_attention_mask is None: return_attention_mask = "attention_mask" in self.model_input_names required_input = encoded_inputs[self.model_input_names[0]] if padding_strategy == PaddingStrategy.LONGEST: max_length = len(required_input) if max_length is not None and pad_to_multiple_of is not None and (max_length % pad_to_multiple_of != 0): max_length = ((max_length // pad_to_multiple_of) + 1) * pad_to_multiple_of needs_to_be_padded = padding_strategy != PaddingStrategy.DO_NOT_PAD and len(required_input) != max_length # Initialize attention mask if not present. if return_attention_mask and "attention_mask" not in encoded_inputs: encoded_inputs["attention_mask"] = [1] * len(required_input) if needs_to_be_padded: difference = max_length - len(required_input) if self.padding_side == "right": if return_attention_mask: encoded_inputs["attention_mask"] = encoded_inputs["attention_mask"] + [0] * difference if "token_type_ids" in encoded_inputs: encoded_inputs["token_type_ids"] = ( encoded_inputs["token_type_ids"] + [self.pad_token_type_id] * difference ) if "bbox" in encoded_inputs: encoded_inputs["bbox"] = encoded_inputs["bbox"] + [self.pad_token_box] * difference if "labels" in encoded_inputs: encoded_inputs["labels"] = encoded_inputs["labels"] + [self.pad_token_label] * difference if "special_tokens_mask" in encoded_inputs: encoded_inputs["special_tokens_mask"] = encoded_inputs["special_tokens_mask"] + [1] * difference encoded_inputs[self.model_input_names[0]] = required_input + [self.pad_token_id] * difference elif self.padding_side == "left": if return_attention_mask: encoded_inputs["attention_mask"] = [0] * difference + encoded_inputs["attention_mask"] if "token_type_ids" in encoded_inputs: encoded_inputs["token_type_ids"] = [self.pad_token_type_id] * difference + encoded_inputs[ "token_type_ids" ] if "bbox" in encoded_inputs: encoded_inputs["bbox"] = [self.pad_token_box] * difference + encoded_inputs["bbox"] if "labels" in encoded_inputs: encoded_inputs["labels"] = [self.pad_token_label] * difference + encoded_inputs["labels"] if "special_tokens_mask" in encoded_inputs: encoded_inputs["special_tokens_mask"] = [1] * difference + encoded_inputs["special_tokens_mask"] encoded_inputs[self.model_input_names[0]] = [self.pad_token_id] * difference + required_input else: raise ValueError("Invalid padding strategy:" + str(self.padding_side)) return encoded_inputs # Copied from transformers.models.bert.tokenization_bert.BasicTokenizer class BasicTokenizer(object): """ Constructs a BasicTokenizer that will run basic tokenization (punctuation splitting, lower casing, etc.). Args: do_lower_case (`bool`, *optional*, defaults to `True`): Whether or not to lowercase the input when tokenizing. never_split (`Iterable`, *optional*): Collection of tokens which will never be split during tokenization. Only has an effect when `do_basic_tokenize=True` tokenize_chinese_chars (`bool`, *optional*, defaults to `True`): Whether or not to tokenize Chinese characters. This should likely be deactivated for Japanese (see this [issue](https://github.com/huggingface/transformers/issues/328)). strip_accents (`bool`, *optional*): Whether or not to strip all accents. If this option is not specified, then it will be determined by the value for `lowercase` (as in the original BERT). do_split_on_punc (`bool`, *optional*, defaults to `True`): In some instances we want to skip the basic punctuation splitting so that later tokenization can capture the full context of the words, such as contractions. """ def __init__( self, do_lower_case=True, never_split=None, tokenize_chinese_chars=True, strip_accents=None, do_split_on_punc=True, ): if never_split is None: never_split = [] self.do_lower_case = do_lower_case self.never_split = set(never_split) self.tokenize_chinese_chars = tokenize_chinese_chars self.strip_accents = strip_accents self.do_split_on_punc = do_split_on_punc def tokenize(self, text, never_split=None): """ Basic Tokenization of a piece of text. For sub-word tokenization, see WordPieceTokenizer. Args: never_split (`List[str]`, *optional*) Kept for backward compatibility purposes. Now implemented directly at the base class level (see [`PreTrainedTokenizer.tokenize`]) List of token not to split. """ # union() returns a new set by concatenating the two sets. never_split = self.never_split.union(set(never_split)) if never_split else self.never_split text = self._clean_text(text) # This was added on November 1st, 2018 for the multilingual and Chinese # models. This is also applied to the English models now, but it doesn't # matter since the English models were not trained on any Chinese data # and generally don't have any Chinese data in them (there are Chinese # characters in the vocabulary because Wikipedia does have some Chinese # words in the English Wikipedia.). if self.tokenize_chinese_chars: text = self._tokenize_chinese_chars(text) # prevents treating the same character with different unicode codepoints as different characters unicode_normalized_text = unicodedata.normalize("NFC", text) orig_tokens = whitespace_tokenize(unicode_normalized_text) split_tokens = [] for token in orig_tokens: if token not in never_split: if self.do_lower_case: token = token.lower() if self.strip_accents is not False: token = self._run_strip_accents(token) elif self.strip_accents: token = self._run_strip_accents(token) split_tokens.extend(self._run_split_on_punc(token, never_split)) output_tokens = whitespace_tokenize(" ".join(split_tokens)) return output_tokens def _run_strip_accents(self, text): """Strips accents from a piece of text.""" text = unicodedata.normalize("NFD", text) output = [] for char in text: cat = unicodedata.category(char) if cat == "Mn": continue output.append(char) return "".join(output) def _run_split_on_punc(self, text, never_split=None): """Splits punctuation on a piece of text.""" if not self.do_split_on_punc or (never_split is not None and text in never_split): return [text] chars = list(text) i = 0 start_new_word = True output = [] while i < len(chars): char = chars[i] if _is_punctuation(char): output.append([char]) start_new_word = True else: if start_new_word: output.append([]) start_new_word = False output[-1].append(char) i += 1 return ["".join(x) for x in output] def _tokenize_chinese_chars(self, text): """Adds whitespace around any CJK character.""" output = [] for char in text: cp = ord(char) if self._is_chinese_char(cp): output.append(" ") output.append(char) output.append(" ") else: output.append(char) return "".join(output) def _is_chinese_char(self, cp): """Checks whether CP is the codepoint of a CJK character.""" # This defines a "chinese character" as anything in the CJK Unicode block: # https://en.wikipedia.org/wiki/CJK_Unified_Ideographs_(Unicode_block) # # Note that the CJK Unicode block is NOT all Japanese and Korean characters, # despite its name. The modern Korean Hangul alphabet is a different block, # as is Japanese Hiragana and Katakana. Those alphabets are used to write # space-separated words, so they are not treated specially and handled # like the all of the other languages. if ( (cp >= 0x4E00 and cp <= 0x9FFF) or (cp >= 0x3400 and cp <= 0x4DBF) # or (cp >= 0x20000 and cp <= 0x2A6DF) # or (cp >= 0x2A700 and cp <= 0x2B73F) # or (cp >= 0x2B740 and cp <= 0x2B81F) # or (cp >= 0x2B820 and cp <= 0x2CEAF) # or (cp >= 0xF900 and cp <= 0xFAFF) or (cp >= 0x2F800 and cp <= 0x2FA1F) # ): # return True return False def _clean_text(self, text): """Performs invalid character removal and whitespace cleanup on text.""" output = [] for char in text: cp = ord(char) if cp == 0 or cp == 0xFFFD or _is_control(char): continue if _is_whitespace(char): output.append(" ") else: output.append(char) return "".join(output) # Copied from transformers.models.bert.tokenization_bert.WordpieceTokenizer class WordpieceTokenizer(object): """Runs WordPiece tokenization.""" def __init__(self, vocab, unk_token, max_input_chars_per_word=100): self.vocab = vocab self.unk_token = unk_token self.max_input_chars_per_word = max_input_chars_per_word def tokenize(self, text): """ Tokenizes a piece of text into its word pieces. This uses a greedy longest-match-first algorithm to perform tokenization using the given vocabulary. For example, `input = "unaffable"` wil return as output `["un", "##aff", "##able"]`. Args: text: A single token or whitespace separated tokens. This should have already been passed through *BasicTokenizer*. Returns: A list of wordpiece tokens. """ output_tokens = [] for token in whitespace_tokenize(text): chars = list(token) if len(chars) > self.max_input_chars_per_word: output_tokens.append(self.unk_token) continue is_bad = False start = 0 sub_tokens = [] while start < len(chars): end = len(chars) cur_substr = None while start < end: substr = "".join(chars[start:end]) if start > 0: substr = "##" + substr if substr in self.vocab: cur_substr = substr break end -= 1 if cur_substr is None: is_bad = True break sub_tokens.append(cur_substr) start = end if is_bad: output_tokens.append(self.unk_token) else: output_tokens.extend(sub_tokens) return output_tokens