566 lines
22 KiB
Python
566 lines
22 KiB
Python
# coding=utf-8
|
||
# Copyright 2019-present CNRS, Facebook Inc. 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 classes for Flaubert."""
|
||
|
||
|
||
import json
|
||
import os
|
||
import re
|
||
import unicodedata
|
||
from typing import List, Optional, Tuple
|
||
|
||
from ...tokenization_utils import PreTrainedTokenizer
|
||
from ...utils import logging
|
||
|
||
|
||
logger = logging.get_logger(__name__)
|
||
|
||
VOCAB_FILES_NAMES = {
|
||
"vocab_file": "vocab.json",
|
||
"merges_file": "merges.txt",
|
||
}
|
||
|
||
|
||
def convert_to_unicode(text):
|
||
"""
|
||
Converts `text` to Unicode (if it's not already), assuming UTF-8 input.
|
||
"""
|
||
|
||
def ensure_text(s, encoding="utf-8", errors="strict"):
|
||
if isinstance(s, bytes):
|
||
return s.decode(encoding, errors)
|
||
elif isinstance(s, str):
|
||
return s
|
||
else:
|
||
raise TypeError(f"not expecting type '{type(s)}'")
|
||
|
||
return ensure_text(text, encoding="utf-8", errors="ignore")
|
||
|
||
|
||
# Copied from transformers.models.xlm.tokenization_xlm.get_pairs
|
||
def get_pairs(word):
|
||
"""
|
||
Return set of symbol pairs in a word. word is represented as tuple of symbols (symbols being variable-length
|
||
strings)
|
||
"""
|
||
pairs = set()
|
||
prev_char = word[0]
|
||
for char in word[1:]:
|
||
pairs.add((prev_char, char))
|
||
prev_char = char
|
||
return pairs
|
||
|
||
|
||
# Copied from transformers.models.xlm.tokenization_xlm.replace_unicode_punct
|
||
def replace_unicode_punct(text):
|
||
"""
|
||
Port of https://github.com/moses-smt/mosesdecoder/blob/master/scripts/tokenizer/replace-unicode-punctuation.perl
|
||
"""
|
||
text = text.replace("，", ",")
|
||
text = re.sub(r"。\s*", ". ", text)
|
||
text = text.replace("、", ",")
|
||
text = text.replace("”", '"')
|
||
text = text.replace("“", '"')
|
||
text = text.replace("∶", ":")
|
||
text = text.replace("：", ":")
|
||
text = text.replace("？", "?")
|
||
text = text.replace("《", '"')
|
||
text = text.replace("》", '"')
|
||
text = text.replace("）", ")")
|
||
text = text.replace("！", "!")
|
||
text = text.replace("（", "(")
|
||
text = text.replace("；", ";")
|
||
text = text.replace("１", "1")
|
||
text = text.replace("」", '"')
|
||
text = text.replace("「", '"')
|
||
text = text.replace("０", "0")
|
||
text = text.replace("３", "3")
|
||
text = text.replace("２", "2")
|
||
text = text.replace("５", "5")
|
||
text = text.replace("６", "6")
|
||
text = text.replace("９", "9")
|
||
text = text.replace("７", "7")
|
||
text = text.replace("８", "8")
|
||
text = text.replace("４", "4")
|
||
text = re.sub(r"．\s*", ". ", text)
|
||
text = text.replace("～", "~")
|
||
text = text.replace("’", "'")
|
||
text = text.replace("…", "...")
|
||
text = text.replace("━", "-")
|
||
text = text.replace("〈", "<")
|
||
text = text.replace("〉", ">")
|
||
text = text.replace("【", "[")
|
||
text = text.replace("】", "]")
|
||
text = text.replace("％", "%")
|
||
return text
|
||
|
||
|
||
# Copied from transformers.models.xlm.tokenization_xlm.remove_non_printing_char
|
||
def remove_non_printing_char(text):
|
||
"""
|
||
Port of https://github.com/moses-smt/mosesdecoder/blob/master/scripts/tokenizer/remove-non-printing-char.perl
|
||
"""
|
||
output = []
|
||
for char in text:
|
||
cat = unicodedata.category(char)
|
||
if cat.startswith("C"):
|
||
continue
|
||
output.append(char)
|
||
return "".join(output)
|
||
|
||
|
||
class FlaubertTokenizer(PreTrainedTokenizer):
|
||
"""
|
||
Construct a Flaubert tokenizer. Based on Byte-Pair Encoding. The tokenization process is the following:
|
||
|
||
- Moses preprocessing and tokenization.
|
||
- Normalizing all inputs text.
|
||
- The arguments `special_tokens` and the function `set_special_tokens`, can be used to add additional symbols (like
|
||
"__classify__") to a vocabulary.
|
||
- The argument `do_lowercase` controls lower casing (automatically set for pretrained vocabularies).
|
||
|
||
This tokenizer inherits from [`PreTrainedTokenizer`] which contains most of the main methods. Users should refer to
|
||
this superclass for more information regarding those methods.
|
||
|
||
Args:
|
||
vocab_file (`str`):
|
||
Vocabulary file.
|
||
merges_file (`str`):
|
||
Merges file.
|
||
do_lowercase (`bool`, *optional*, defaults to `False`):
|
||
Controls lower casing.
|
||
unk_token (`str`, *optional*, defaults to `"<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.
|
||
bos_token (`str`, *optional*, defaults to `"<s>"`):
|
||
The beginning of sequence token that was used during pretraining. Can be used a sequence classifier token.
|
||
|
||
<Tip>
|
||
|
||
When building a sequence using special tokens, this is not the token that is used for the beginning of
|
||
sequence. The token used is the `cls_token`.
|
||
|
||
</Tip>
|
||
|
||
sep_token (`str`, *optional*, defaults to `"</s>"`):
|
||
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.
|
||
pad_token (`str`, *optional*, defaults to `"<pad>"`):
|
||
The token used for padding, for example when batching sequences of different lengths.
|
||
cls_token (`str`, *optional*, defaults to `"</s>"`):
|
||
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.
|
||
mask_token (`str`, *optional*, defaults to `"<special1>"`):
|
||
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.
|
||
additional_special_tokens (`List[str]`, *optional*, defaults to `['<special0>', '<special1>', '<special2>', '<special3>', '<special4>', '<special5>', '<special6>', '<special7>', '<special8>', '<special9>']`):
|
||
List of additional special tokens.
|
||
lang2id (`Dict[str, int]`, *optional*):
|
||
Dictionary mapping languages string identifiers to their IDs.
|
||
id2lang (`Dict[int, str]`, *optional*):
|
||
Dictionary mapping language IDs to their string identifiers.
|
||
"""
|
||
|
||
vocab_files_names = VOCAB_FILES_NAMES
|
||
|
||
def __init__(
|
||
self,
|
||
vocab_file,
|
||
merges_file,
|
||
do_lowercase=False,
|
||
unk_token="<unk>",
|
||
bos_token="<s>",
|
||
sep_token="</s>",
|
||
pad_token="<pad>",
|
||
cls_token="</s>",
|
||
mask_token="<special1>",
|
||
additional_special_tokens=[
|
||
"<special0>",
|
||
"<special1>",
|
||
"<special2>",
|
||
"<special3>",
|
||
"<special4>",
|
||
"<special5>",
|
||
"<special6>",
|
||
"<special7>",
|
||
"<special8>",
|
||
"<special9>",
|
||
],
|
||
lang2id=None,
|
||
id2lang=None,
|
||
**kwargs,
|
||
):
|
||
do_lowercase_and_remove_accent = kwargs.pop("do_lowercase_and_remove_accent", None)
|
||
if do_lowercase_and_remove_accent is not None:
|
||
logger.warning(
|
||
"`do_lowercase_and_remove_accent` is passed as a keyword argument, but this won't do anything."
|
||
" `FlaubertTokenizer` will always set it to `False`."
|
||
)
|
||
# always `False`
|
||
self.do_lowercase_and_remove_accent = False
|
||
|
||
self.do_lowercase = do_lowercase
|
||
|
||
try:
|
||
import sacremoses
|
||
except ImportError:
|
||
raise ImportError(
|
||
"You need to install sacremoses to use FlaubertTokenizer. "
|
||
"See https://pypi.org/project/sacremoses/ for installation."
|
||
)
|
||
|
||
self.sm = sacremoses
|
||
|
||
# cache of sm.MosesPunctNormalizer instance
|
||
self.cache_moses_punct_normalizer = {}
|
||
# cache of sm.MosesTokenizer instance
|
||
self.cache_moses_tokenizer = {}
|
||
self.lang_with_custom_tokenizer = {"zh", "th", "ja"}
|
||
self.lang2id = lang2id
|
||
self.id2lang = id2lang
|
||
if lang2id is not None and id2lang is not None:
|
||
assert len(lang2id) == len(id2lang)
|
||
|
||
self.ja_word_tokenizer = None
|
||
self.zh_word_tokenizer = None
|
||
|
||
with open(vocab_file, encoding="utf-8") as vocab_handle:
|
||
self.encoder = json.load(vocab_handle)
|
||
self.decoder = {v: k for k, v in self.encoder.items()}
|
||
with open(merges_file, encoding="utf-8") as merges_handle:
|
||
merges = merges_handle.read().split("\n")[:-1]
|
||
merges = [tuple(merge.split()[:2]) for merge in merges]
|
||
self.bpe_ranks = dict(zip(merges, range(len(merges))))
|
||
self.cache = {}
|
||
|
||
super().__init__(
|
||
unk_token=unk_token,
|
||
bos_token=bos_token,
|
||
sep_token=sep_token,
|
||
pad_token=pad_token,
|
||
cls_token=cls_token,
|
||
mask_token=mask_token,
|
||
additional_special_tokens=additional_special_tokens,
|
||
lang2id=lang2id,
|
||
id2lang=id2lang,
|
||
**kwargs,
|
||
)
|
||
|
||
@property
|
||
# Copied from transformers.models.xlm.tokenization_xlm.XLMTokenizer.do_lower_case
|
||
def do_lower_case(self):
|
||
return self.do_lowercase_and_remove_accent
|
||
|
||
# Copied from transformers.models.xlm.tokenization_xlm.XLMTokenizer.moses_punct_norm
|
||
def moses_punct_norm(self, text, lang):
|
||
if lang not in self.cache_moses_punct_normalizer:
|
||
punct_normalizer = self.sm.MosesPunctNormalizer(lang=lang)
|
||
self.cache_moses_punct_normalizer[lang] = punct_normalizer
|
||
else:
|
||
punct_normalizer = self.cache_moses_punct_normalizer[lang]
|
||
return punct_normalizer.normalize(text)
|
||
|
||
# Copied from transformers.models.xlm.tokenization_xlm.XLMTokenizer.moses_tokenize
|
||
def moses_tokenize(self, text, lang):
|
||
if lang not in self.cache_moses_tokenizer:
|
||
moses_tokenizer = self.sm.MosesTokenizer(lang=lang)
|
||
self.cache_moses_tokenizer[lang] = moses_tokenizer
|
||
else:
|
||
moses_tokenizer = self.cache_moses_tokenizer[lang]
|
||
return moses_tokenizer.tokenize(text, return_str=False, escape=False)
|
||
|
||
# Copied from transformers.models.xlm.tokenization_xlm.XLMTokenizer.moses_pipeline
|
||
def moses_pipeline(self, text, lang):
|
||
text = replace_unicode_punct(text)
|
||
text = self.moses_punct_norm(text, lang)
|
||
text = remove_non_printing_char(text)
|
||
return text
|
||
|
||
# Copied from transformers.models.xlm.tokenization_xlm.XLMTokenizer.ja_tokenize
|
||
def ja_tokenize(self, text):
|
||
if self.ja_word_tokenizer is None:
|
||
try:
|
||
import Mykytea
|
||
|
||
self.ja_word_tokenizer = Mykytea.Mykytea(
|
||
f"-model {os.path.expanduser('~')}/local/share/kytea/model.bin"
|
||
)
|
||
except (AttributeError, ImportError):
|
||
logger.error(
|
||
"Make sure you install KyTea (https://github.com/neubig/kytea) and it's python wrapper"
|
||
" (https://github.com/chezou/Mykytea-python) with the following steps"
|
||
)
|
||
logger.error("1. git clone git@github.com:neubig/kytea.git && cd kytea")
|
||
logger.error("2. autoreconf -i")
|
||
logger.error("3. ./configure --prefix=$HOME/local")
|
||
logger.error("4. make && make install")
|
||
logger.error("5. pip install kytea")
|
||
raise
|
||
return list(self.ja_word_tokenizer.getWS(text))
|
||
|
||
@property
|
||
# Copied from transformers.models.xlm.tokenization_xlm.XLMTokenizer.vocab_size
|
||
def vocab_size(self):
|
||
return len(self.encoder)
|
||
|
||
# Copied from transformers.models.xlm.tokenization_xlm.XLMTokenizer.get_vocab
|
||
def get_vocab(self):
|
||
return dict(self.encoder, **self.added_tokens_encoder)
|
||
|
||
# Copied from transformers.models.xlm.tokenization_xlm.XLMTokenizer.bpe
|
||
def bpe(self, token):
|
||
word = tuple(token[:-1]) + (token[-1] + "</w>",)
|
||
if token in self.cache:
|
||
return self.cache[token]
|
||
pairs = get_pairs(word)
|
||
|
||
if not pairs:
|
||
return token + "</w>"
|
||
|
||
while True:
|
||
bigram = min(pairs, key=lambda pair: self.bpe_ranks.get(pair, float("inf")))
|
||
if bigram not in self.bpe_ranks:
|
||
break
|
||
first, second = bigram
|
||
new_word = []
|
||
i = 0
|
||
while i < len(word):
|
||
try:
|
||
j = word.index(first, i)
|
||
except ValueError:
|
||
new_word.extend(word[i:])
|
||
break
|
||
else:
|
||
new_word.extend(word[i:j])
|
||
i = j
|
||
|
||
if word[i] == first and i < len(word) - 1 and word[i + 1] == second:
|
||
new_word.append(first + second)
|
||
i += 2
|
||
else:
|
||
new_word.append(word[i])
|
||
i += 1
|
||
new_word = tuple(new_word)
|
||
word = new_word
|
||
if len(word) == 1:
|
||
break
|
||
else:
|
||
pairs = get_pairs(word)
|
||
word = " ".join(word)
|
||
if word == "\n </w>":
|
||
word = "\n</w>"
|
||
self.cache[token] = word
|
||
return word
|
||
|
||
def preprocess_text(self, text):
|
||
text = text.replace("``", '"').replace("''", '"')
|
||
text = convert_to_unicode(text)
|
||
text = unicodedata.normalize("NFC", text)
|
||
|
||
if self.do_lowercase:
|
||
text = text.lower()
|
||
|
||
return text
|
||
|
||
def _tokenize(self, text, bypass_tokenizer=False):
|
||
"""
|
||
Tokenize a string given language code using Moses.
|
||
|
||
Details of tokenization:
|
||
|
||
- [sacremoses](https://github.com/alvations/sacremoses): port of Moses
|
||
- Install with `pip install sacremoses`
|
||
|
||
Args:
|
||
- bypass_tokenizer: Allow users to preprocess and tokenize the sentences externally (default = False)
|
||
(bool). If True, we only apply BPE.
|
||
|
||
Returns:
|
||
List of tokens.
|
||
"""
|
||
lang = "fr"
|
||
if lang and self.lang2id and lang not in self.lang2id:
|
||
logger.error(
|
||
"Supplied language code not found in lang2id mapping. Please check that your language is supported by"
|
||
" the loaded pretrained model."
|
||
)
|
||
|
||
if bypass_tokenizer:
|
||
text = text.split()
|
||
else:
|
||
text = self.preprocess_text(text)
|
||
text = self.moses_pipeline(text, lang=lang)
|
||
text = self.moses_tokenize(text, lang=lang)
|
||
|
||
split_tokens = []
|
||
for token in text:
|
||
if token:
|
||
split_tokens.extend(list(self.bpe(token).split(" ")))
|
||
|
||
return split_tokens
|
||
|
||
# Copied from transformers.models.xlm.tokenization_xlm.XLMTokenizer._convert_token_to_id
|
||
def _convert_token_to_id(self, token):
|
||
"""Converts a token (str) in an id using the vocab."""
|
||
return self.encoder.get(token, self.encoder.get(self.unk_token))
|
||
|
||
# Copied from transformers.models.xlm.tokenization_xlm.XLMTokenizer._convert_id_to_token
|
||
def _convert_id_to_token(self, index):
|
||
"""Converts an index (integer) in a token (str) using the vocab."""
|
||
return self.decoder.get(index, self.unk_token)
|
||
|
||
# Copied from transformers.models.xlm.tokenization_xlm.XLMTokenizer.convert_tokens_to_string
|
||
def convert_tokens_to_string(self, tokens):
|
||
"""Converts a sequence of tokens (string) in a single string."""
|
||
out_string = "".join(tokens).replace("</w>", " ").strip()
|
||
return out_string
|
||
|
||
# Copied from transformers.models.xlm.tokenization_xlm.XLMTokenizer.build_inputs_with_special_tokens
|
||
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. An XLM sequence has the following format:
|
||
|
||
- single sequence: `<s> X </s>`
|
||
- pair of sequences: `<s> A </s> B </s>`
|
||
|
||
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.
|
||
|
||
"""
|
||
bos = [self.bos_token_id]
|
||
sep = [self.sep_token_id]
|
||
|
||
if token_ids_1 is None:
|
||
return bos + token_ids_0 + sep
|
||
return bos + token_ids_0 + sep + token_ids_1 + sep
|
||
|
||
# Copied from transformers.models.xlm.tokenization_xlm.XLMTokenizer.get_special_tokens_mask
|
||
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]
|
||
|
||
# Copied from transformers.models.xlm.tokenization_xlm.XLMTokenizer.create_token_type_ids_from_sequences
|
||
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. An XLM 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]
|
||
|
||
# Copied from transformers.models.xlm.tokenization_xlm.XLMTokenizer.save_vocabulary
|
||
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
|
||
vocab_file = os.path.join(
|
||
save_directory, (filename_prefix + "-" if filename_prefix else "") + VOCAB_FILES_NAMES["vocab_file"]
|
||
)
|
||
merge_file = os.path.join(
|
||
save_directory, (filename_prefix + "-" if filename_prefix else "") + VOCAB_FILES_NAMES["merges_file"]
|
||
)
|
||
|
||
with open(vocab_file, "w", encoding="utf-8") as f:
|
||
f.write(json.dumps(self.encoder, indent=2, sort_keys=True, ensure_ascii=False) + "\n")
|
||
|
||
index = 0
|
||
with open(merge_file, "w", encoding="utf-8") as writer:
|
||
for bpe_tokens, token_index in sorted(self.bpe_ranks.items(), key=lambda kv: kv[1]):
|
||
if index != token_index:
|
||
logger.warning(
|
||
f"Saving vocabulary to {merge_file}: BPE merge indices are not consecutive."
|
||
" Please check that the tokenizer is not corrupted!"
|
||
)
|
||
index = token_index
|
||
writer.write(" ".join(bpe_tokens) + "\n")
|
||
index += 1
|
||
|
||
return vocab_file, merge_file
|
||
|
||
# Copied from transformers.models.xlm.tokenization_xlm.XLMTokenizer.__getstate__
|
||
def __getstate__(self):
|
||
state = self.__dict__.copy()
|
||
state["sm"] = None
|
||
return state
|
||
|
||
# Copied from transformers.models.xlm.tokenization_xlm.XLMTokenizer.__setstate__
|
||
def __setstate__(self, d):
|
||
self.__dict__ = d
|
||
|
||
try:
|
||
import sacremoses
|
||
except ImportError:
|
||
raise ImportError(
|
||
"You need to install sacremoses to use XLMTokenizer. "
|
||
"See https://pypi.org/project/sacremoses/ for installation."
|
||
)
|
||
|
||
self.sm = sacremoses
|