tokenizer.py

"""
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
- Promotion: (Q)=queen, (R)=rook, (B)=bishop, (N)=knight

New token strategy:
- we only retain the squares involved in the move and the promotion piece if any
- everything else (piece type, capture flag, check flag, etc.) is discarded
- the vocabulary size is thus minimal (72 tokens): 64 squares + 4 promotion pieces + 4 special tokens
"""

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)
        
        for f in 'abcdefgh':
            for r in '12345678':
                vocab[f"{f}{r}"] = idx
                idx += 1
                
        for p in ['q', 'r', 'b', 'n']:
            vocab[p] = idx
            idx += 1
        return vocab
    
    def _tokenize(self, text: str) -> List[str]:
        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:
        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}
        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)