File size: 4,959 Bytes
3984c38
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
547e194
3984c38
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1da69df
547e194
3984c38
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
547e194
1da69df
 
 
 
 
547e194
 
1da69df
 
 
 
3984c38
 
 
 
1da69df
 
3984c38
1da69df
 
 
 
 
547e194
3984c38
 
 
1da69df
3984c38
 
 
 
 
 
1da69df
 
 
547e194
 
1da69df
 
 
 
 
547e194
 
1da69df
 
3984c38
1da69df
3984c38
 
 
1da69df
3984c38
 
 
 
 
1da69df
 
7d4510b
1da69df
 
 
7d4510b
1da69df
 
 
 
3984c38
1da69df
 
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
"""

Custom Chess Tokenizer for the Chess Challenge.



This tokenizer treats each move as a single token using the extended UCI notation

from the Lichess dataset (e.g., WPe2e4, BNg8f6).



The dataset format uses:

- W/B prefix for White/Black

- Piece letter: P=Pawn, N=Knight, B=Bishop, R=Rook, Q=Queen, K=King

- Source and destination squares (e.g., e2e4)

- Special suffixes: (x)=capture, (+)=check, (+*)=checkmate, (o)/(O)=castling

"""

from __future__ import annotations

import json
import os
from pathlib import Path
from typing import Dict, List, Optional
import re

from transformers import PreTrainedTokenizer


class ChessTokenizer(PreTrainedTokenizer):
    model_input_names = ["input_ids", "attention_mask"]
    
    PAD_TOKEN = "[PAD]"
    BOS_TOKEN = "[BOS]"
    EOS_TOKEN = "[EOS]"
    UNK_TOKEN = "[UNK]"
    
    def __init__(

        self,

        vocab_file: Optional[str] = None,

        vocab: Optional[Dict[str, int]] = None,

        **kwargs,

    ):
        self._pad_token = self.PAD_TOKEN
        self._bos_token = self.BOS_TOKEN
        self._eos_token = self.EOS_TOKEN
        self._unk_token = self.UNK_TOKEN

        kwargs.pop("pad_token", None)
        kwargs.pop("bos_token", None)
        kwargs.pop("eos_token", None)
        kwargs.pop("unk_token", None)
        
        self.token_pattern = re.compile(r'[a-h][1-8]|[qrbn]')

        if vocab is not None:
            self._vocab = vocab
        elif vocab_file is not None and os.path.exists(vocab_file):
            with open(vocab_file, "r", encoding="utf-8") as f:
                self._vocab = json.load(f)
        else:
            self._vocab = self._create_default_vocab()
        
        self._ids_to_tokens = {v: k for k, v in self._vocab.items()}
        
        super().__init__(
            pad_token=self._pad_token,
            bos_token=self._bos_token,
            eos_token=self._eos_token,
            unk_token=self._unk_token,
            **kwargs,
        )
    
    def _create_default_vocab(self) -> Dict[str, int]:
        special_tokens = [self.PAD_TOKEN, self.BOS_TOKEN, self.EOS_TOKEN, self.UNK_TOKEN]
        vocab = {token: idx for idx, token in enumerate(special_tokens)}
        idx = len(vocab)
        
        # Squares (4-67)
        for f in 'abcdefgh':
            for r in '12345678':
                vocab[f"{f}{r}"] = idx
                idx += 1
                
        # Promotions (68-71)
        for p in ['q', 'r', 'b', 'n']:
            vocab[p] = idx
            idx += 1
        return vocab
    
    def _tokenize(self, text: str) -> List[str]:
        """

        Tokenizes text by first normalizing specific chess patterns 

        and then extracting squares/promotions.

        """
        text = (text.replace("(Q)", "q")
                    .replace("(R)", "r")
                    .replace("(B)", "b")
                    .replace("(N)", "n"))
        
        return self.token_pattern.findall(text)
    
    def _convert_token_to_id(self, token: str) -> int:
        """Convert a token to its ID."""
        return self._vocab.get(token, self._vocab.get(self.UNK_TOKEN, 0))
    
    def _convert_id_to_token(self, index: int) -> str:
        """Convert an ID to its token."""
        return self._ids_to_tokens.get(index, self.UNK_TOKEN)
    
    def convert_tokens_to_string(self, tokens: List[str]) -> str:
        """Reconstructs standard UCI string (e.g. "e2e4 a7a8q")"""
        special = {self.PAD_TOKEN, self.BOS_TOKEN, self.EOS_TOKEN, self.UNK_TOKEN}
        clean_tokens = [t for t in tokens if t not in special]
        
        output = []
        for token in clean_tokens:
            if token in ['q', 'r', 'b', 'n'] and output:
                output[-1] += token
            elif output and len(output[-1]) == 2 and output[-1][0] in 'abcdefgh':
                output[-1] += token
            else:
                output.append(token)
                
        return " ".join(output)
    
    def save_vocabulary(self, save_directory: str, filename_prefix: Optional[str] = None) -> tuple:
        if not os.path.isdir(save_directory):
            os.makedirs(save_directory, exist_ok=True)
        vocab_file = os.path.join(
            save_directory, (filename_prefix + "-" if filename_prefix else "") + "vocab.json"
        )
        with open(vocab_file, "w", encoding="utf-8") as f:
            json.dump(self._vocab, f, ensure_ascii=False, indent=2)
        return (vocab_file,)
    
    @classmethod
    def build_vocab_from_iterator(cls, iterator, min_frequency=1):
        return cls() 
        
    @classmethod
    def build_vocab_from_dataset(cls, **kwargs):
        return cls() 

    @property
    def vocab_size(self) -> int:
        return len(self._vocab)
    
    def get_vocab(self) -> Dict[str, int]:
        return dict(self._vocab)