File size: 5,475 Bytes
2e57899
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
"""
Custom Chess Tokenizer for the Chess Challenge.

This tokenizer splits moves into 3 parts:
1. Piece (e.g., WP)
2. From Square (e.g., e2)
3. To Square + Suffix (e.g., e4 or e4(x))
"""

from __future__ import annotations

import json
import os
from typing import Dict, List, Optional

from transformers import PreTrainedTokenizer


class ChessTokenizer(PreTrainedTokenizer):
    """
    A custom tokenizer for chess moves using a 3-part split.
    
    Splits "WPe2e4(x)" into ["WP", "e2", "e4(x)"].
    """
    
    model_input_names = ["input_ids", "attention_mask"]
    vocab_files_names = {"vocab_file": "vocab.json"}
    
    # Special tokens
    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,
    ):
        # Clean kwargs to avoid conflicts
        kwargs.pop("pad_token", None)
        kwargs.pop("bos_token", None)
        kwargs.pop("eos_token", None)
        kwargs.pop("unk_token", None)

        self.vocab_file = vocab_file
        
        # Load vocab
        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]:
        """Create a minimal default vocabulary with just special tokens."""
        special_tokens = [self.PAD_TOKEN, self.BOS_TOKEN, self.EOS_TOKEN, self.UNK_TOKEN]
        vocab = {token: idx for idx, token in enumerate(special_tokens)}
        return vocab
    
    @property
    def vocab_size(self) -> int:
        return len(self._vocab)
    
    def get_vocab(self) -> Dict[str, int]:
        return dict(self._vocab)
    
    def _tokenize(self, text: str) -> List[str]:
        """
        Tokenize a string of moves into 3 components per move.
        """
        tokens = []
        raw_moves = text.strip().split()
        
        for move in raw_moves:
            if len(move) >= 6:
                # 1. Piece (WP)
                tokens.append(move[:2])
                # 2. From (e2)
                tokens.append(move[2:4])
                # 3. To (e4 or e4(x)) - grab the rest
                tokens.append(move[4:])
            else:
                tokens.append(self.UNK_TOKEN)
        return tokens
    
    def _convert_token_to_id(self, token: str) -> int:
        return self._vocab.get(token, self._vocab.get(self.UNK_TOKEN))
    
    def _convert_id_to_token(self, index: int) -> str:
        return self._ids_to_tokens.get(index, self.UNK_TOKEN)
    
    def convert_tokens_to_string(self, tokens: List[str]) -> str:
        # Filter specials
        filtered = [t for t in tokens if t not in [self.PAD_TOKEN, self.BOS_TOKEN, self.EOS_TOKEN, self.UNK_TOKEN]]
        # Join with space. Result: "WP e2 e4 BN g8 f6"
        return " ".join(filtered)
    
    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_dataset(
        cls,
        dataset_name: str = "dlouapre/lichess_2025-01_1M",
        split: str = "train",
        column: str = "text",
        min_frequency: int = 100,
        max_samples: Optional[int] = 100000,
    ) -> "ChessTokenizer":
        from datasets import load_dataset
        
        print(f"Loading dataset {dataset_name} to build vocabulary...")
        dataset = load_dataset(dataset_name, split=split, streaming=True)
        
        unique_tokens = set()
        
        print("Building vocabulary...")
        count = 0
        for example in dataset:
            moves = example[column].strip().split()
            for move in moves:
                if len(move) >= 6:
                    unique_tokens.add(move[:2])    # Piece
                    unique_tokens.add(move[2:4])   # From
                    unique_tokens.add(move[4:])    # To (includes suffix like (x))
            count += 1
            if max_samples is not None and count >= max_samples:
                break

        special = [cls.PAD_TOKEN, cls.BOS_TOKEN, cls.EOS_TOKEN, cls.UNK_TOKEN]
        # Sort tokens to ensure deterministic IDs
        all_tokens = special + sorted(list(unique_tokens))
        
        vocab = {token: idx for idx, token in enumerate(all_tokens)}
        print(f"Built vocabulary with {len(vocab)} tokens")
        return cls(vocab=vocab)


# Kept for compatibility if other scripts import it
def count_vocab_from_dataset(*args, **kwargs):
    return {}