tokenizer.py

from __future__ import annotations

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

from transformers import PreTrainedTokenizer


class ChessTokenizer(PreTrainedTokenizer):
    """
    Chess tokenizer with structured move tokens:
    Each move is split into: [side][piece][from][to][suffixes].
    Example:
        "WPe2e4 BNg8xf6+" -> [W][P][e2][e4] [B][N][g8][f6][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]"
    
    MOVE_RE = re.compile(
        r"^(?P<side>[WB])"
        r"(?P<piece>[PNBRQK])"
        r"(?P<src>[a-h][1-8])"
        r"(?P<dst>[a-h][1-8])"
        r"(?P<suffix>.*)$"
    )
    
    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

        # Remove duplicates from kwargs
        kwargs.pop("pad_token", None)
        kwargs.pop("bos_token", None)
        kwargs.pop("eos_token", None)
        kwargs.pop("unk_token", None)
        
        # Load or create 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()
        
        # Reverse mapping
        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 = [self.PAD_TOKEN, self.BOS_TOKEN, self.EOS_TOKEN, self.UNK_TOKEN]

        sides = ["[W]", "[B]"]
        pieces = ["[P]", "[N]", "[B]", "[R]", "[Q]", "[K]"]
        squares = [f"[{f}{r}]" for f in "abcdefgh" for r in "12345678"]
        suffixes = ["[x]", "[+]", "[#]", "[O-O]", "[O-O-O]",
                    "[prom_Q]", "[prom_R]", "[prom_B]", "[prom_N]"]

        vocab_list = special + sides + pieces + squares + suffixes
        return {tok: i for i, tok in enumerate(vocab_list)}
    
    @classmethod
    def build_vocab_from_iterator(cls, iterator, min_frequency: int = 1) -> "ChessTokenizer":
        from collections import Counter

        token_counts = Counter()
        tokenizer = cls()

        for game in iterator:
            tokens = tokenizer._tokenize(game)
            token_counts.update(tokens)

        # Keep tokens meeting frequency threshold
        tokens = [t for t, c in token_counts.items() if c >= min_frequency]
        tokens = sorted(tokens)

        special = [cls.PAD_TOKEN, cls.BOS_TOKEN, cls.EOS_TOKEN, cls.UNK_TOKEN]
        vocab = {tok: i for i, tok in enumerate(special + tokens)}

        return cls(vocab=vocab)
    
    @classmethod
    def build_vocab_from_dataset(
        cls,
        dataset_name: str = "dlouapre/lichess_2025-01_1M",
        split: str = "train",
        column: str = "text",
        min_frequency: int = 500,
        max_samples: Optional[int] = 100000,
    ) -> "ChessTokenizer":
        from datasets import load_dataset

        dataset = load_dataset(dataset_name, split=split)
        if max_samples is not None:
            dataset = dataset.select(range(min(max_samples, len(dataset))))

        def game_iterator():
            for example in dataset:
                yield example[column]

        return cls.build_vocab_from_iterator(game_iterator(), min_frequency=min_frequency)
    
    @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]:
        tokens: List[str] = []

        moves = text.strip().split()
        for move in moves:
            # Castling
            if "O-O-O" in move:
                tokens.append("[W]" if move.startswith("W") else "[B]")
                tokens.append("[O-O-O]")
                continue
            if "O-O" in move:
                tokens.append("[W]" if move.startswith("W") else "[B]")
                tokens.append("[O-O]")
                continue

            m = self.MOVE_RE.match(move)
            if not m:
                tokens.append(self.UNK_TOKEN)
                continue

            tokens.append(f"[{m.group('side')}]")
            tokens.append(f"[{m.group('piece')}]")
            tokens.append(f"[{m.group('src')}]")
            tokens.append(f"[{m.group('dst')}]")

            suffix = m.group("suffix")
            if "x" in suffix:
                tokens.append("[x]")
            if "+" in suffix:
                tokens.append("[+]")
            if "*" in suffix:
                tokens.append("[#]")
            if "=" in suffix:
                promo = suffix.split("=")[-1].upper()
                tokens.append(f"[prom_{promo}]")

        return tokens
    
    def _convert_token_to_id(self, token: str) -> int:
        return self._vocab.get(token, self._vocab.get(self.UNK_TOKEN, 0))
    
    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:
        special = {self.PAD_TOKEN, self.BOS_TOKEN, self.EOS_TOKEN, self.UNK_TOKEN}
        return " ".join(t for t in tokens if t not in special)
    
    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,)