File size: 7,664 Bytes
c81207b | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 | """MolParser Mobile tokenizer for Hugging Face Hub remote loading."""
from __future__ import annotations
import json
import os
import re
from pathlib import Path
from typing import Dict, Iterable, List, Optional, Sequence, Union
from huggingface_hub import hf_hub_download
from transformers import PreTrainedTokenizer
TOKENIZER_CONFIG_NAME = "tokenizer_config.json"
VOCAB_NAME = "vocab.txt"
class MolParserTokenizer(PreTrainedTokenizer):
model_input_names = ["input_ids", "attention_mask"]
padding_side = "right"
def __init__(
self,
vocab_list: Optional[List[str]] = None,
special_tokens: Optional[Dict[str, str]] = None,
additional_special_tokens: Optional[Sequence[str]] = None,
**kwargs,
):
if vocab_list is None:
vocab_list = []
if special_tokens is None:
special_tokens = {}
if additional_special_tokens is None:
additional_special_tokens = []
self.special_tokens = {
"cls_token": special_tokens.get("cls_token", "[CLS]"),
"pad_token": special_tokens.get("pad_token", "[PAD]"),
"sep_token": special_tokens.get("sep_token", "[SEP]"),
"unk_token": special_tokens.get("unk_token", "[UNK]"),
}
self.additional_special_tokens = list(dict.fromkeys(additional_special_tokens))
self.vocab_list = list(vocab_list)
all_tokens = self._build_full_vocab(self.vocab_list)
self.vocab = {token: idx for idx, token in enumerate(all_tokens)}
self.ids_to_tokens = {idx: token for token, idx in self.vocab.items()}
self._decode_skip_tokens = set(self.special_tokens.values())
self._compile_pattern()
super().__init__(
cls_token=self.special_tokens["cls_token"],
pad_token=self.special_tokens["pad_token"],
sep_token=self.special_tokens["sep_token"],
unk_token=self.special_tokens["unk_token"],
bos_token=self.special_tokens["cls_token"],
eos_token=self.special_tokens["sep_token"],
additional_special_tokens=self.additional_special_tokens,
**kwargs,
)
def _build_full_vocab(self, vocab_list: Sequence[str]) -> List[str]:
ordered_tokens: List[str] = []
for token in list(self.special_tokens.values()) + list(vocab_list) + list(self.additional_special_tokens):
if token not in ordered_tokens:
ordered_tokens.append(token)
return ordered_tokens
def _compile_pattern(self) -> None:
multi_char_tokens = sorted(self.vocab.keys(), key=len, reverse=True)
pattern = "(" + "|".join(re.escape(token) for token in multi_char_tokens) + "|.)"
self.pattern = re.compile(pattern)
@property
def vocab_size(self) -> int:
return len(self.vocab)
def __len__(self) -> int:
return len(self.vocab)
def get_vocab(self) -> Dict[str, int]:
return dict(self.vocab)
def _tokenize(self, text: str) -> List[str]:
return [token for token in self.pattern.findall(str(text)) if token]
def tokenize(self, text: str, **kwargs) -> List[str]:
return self._tokenize(text)
def _convert_token_to_id(self, token: str) -> int:
return self.vocab.get(token, self.unk_token_id)
def _convert_id_to_token(self, index: int) -> str:
return self.ids_to_tokens.get(int(index), self.unk_token)
def convert_tokens_to_string(self, tokens: Sequence[str]) -> str:
return "".join(tokens)
def build_inputs_with_special_tokens(self, token_ids_0, token_ids_1=None):
if token_ids_1 is None:
return list(token_ids_0)
return list(token_ids_0) + list(token_ids_1)
def encode(self, text: str, add_special_tokens: bool = False, **kwargs) -> List[int]:
token_ids = [self._convert_token_to_id(token) for token in self._tokenize(text)]
if add_special_tokens:
return [self.bos_token_id] + token_ids + [self.eos_token_id]
return token_ids
def decode(self, token_ids: Iterable[int], skip_special_tokens: bool = False, **kwargs) -> str:
tokens = [self._convert_id_to_token(idx) for idx in token_ids]
if skip_special_tokens:
# Match deploy/tokenizer.py: keep MolParser business tokens such as
# <sep>, <a>, </a>, <r>, </r>, <c>, </c>, and |Sg:n|.
tokens = [token for token in tokens if token not in self._decode_skip_tokens]
return "".join(tokens)
def batch_encode(self, texts: Sequence[str], add_special_tokens: bool = False) -> List[List[int]]:
return [self.encode(text, add_special_tokens=add_special_tokens) for text in texts]
def batch_decode(
self,
sequences: Sequence[Sequence[int]],
skip_special_tokens: bool = False,
**kwargs,
) -> List[str]:
return [self.decode(ids, skip_special_tokens=skip_special_tokens, **kwargs) for ids in sequences]
def to_dict(self) -> Dict[str, object]:
return {
"vocab_list": self.vocab_list,
"special_tokens": self.special_tokens,
"additional_special_tokens": self.additional_special_tokens,
"tokenizer_class": self.__class__.__name__,
"auto_map": {
"AutoTokenizer": [
"tokenization_molparser_mobile.MolParserTokenizer",
None,
]
},
}
@classmethod
def from_dict(cls, config: Dict[str, object]) -> "MolParserTokenizer":
return cls(
vocab_list=list(config["vocab_list"]),
special_tokens=dict(config["special_tokens"]),
additional_special_tokens=list(config.get("additional_special_tokens", [])),
)
def save_vocabulary(self, save_directory: str, filename_prefix: Optional[str] = None):
path = Path(save_directory)
path.mkdir(parents=True, exist_ok=True)
name = f"{filename_prefix}-{VOCAB_NAME}" if filename_prefix else VOCAB_NAME
vocab_path = path / name
vocab_path.write_text("\n".join(self.vocab_list) + "\n", encoding="utf-8")
return (str(vocab_path),)
def save_pretrained(self, save_directory: str, **kwargs):
os.makedirs(save_directory, exist_ok=True)
config_path = os.path.join(save_directory, TOKENIZER_CONFIG_NAME)
with open(config_path, "w", encoding="utf-8") as f:
json.dump(self.to_dict(), f, ensure_ascii=False, indent=2)
vocab_files = self.save_vocabulary(save_directory)
return (config_path, *vocab_files)
@classmethod
def from_pretrained(cls, pretrained_model_name_or_path: str, *args, **kwargs) -> "MolParserTokenizer":
config_path = Path(pretrained_model_name_or_path)
if config_path.is_dir():
config_path = config_path / TOKENIZER_CONFIG_NAME
elif config_path.is_file():
pass
else:
config_path = Path(
hf_hub_download(
repo_id=str(pretrained_model_name_or_path),
filename=TOKENIZER_CONFIG_NAME,
repo_type=kwargs.get("repo_type"),
revision=kwargs.get("revision"),
cache_dir=kwargs.get("cache_dir"),
token=kwargs.get("token"),
local_files_only=kwargs.get("local_files_only", False),
)
)
with open(config_path, "r", encoding="utf-8") as f:
config = json.load(f)
return cls.from_dict(config)
__all__ = ["MolParserTokenizer"]
|