Spaces:
Sleeping
Sleeping
File size: 11,602 Bytes
3e1f9da e5572a6 3e1f9da b5e858e 3e1f9da b5e858e 3e1f9da b5e858e 3e1f9da b5e858e 3e1f9da |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 |
"""Chess environment for OpenEnv using moonfish."""
import random
import uuid
from typing import Any, Dict, Optional, Tuple
import chess
from moonfish.psqt import board_evaluation, MG_PIECE_VALUES, count_pieces, get_phase
from moonfish.lib import search_move
from ..models import ChessAction, ChessObservation, ChessState, RewardConfig
class ChessEnvironment:
"""
Chess environment implementing the OpenEnv interface.
Uses python-chess for game logic and moonfish for position evaluation.
Designed for RL training where an agent plays as one color against
an opponent (which can be random, moonfish engine, or self-play).
"""
def __init__(
self,
reward_config: Optional[RewardConfig] = None,
max_moves: int = 500,
agent_color: Optional[
bool
] = None, # None = alternate, True = White, False = Black
opponent: Optional[
str
] = None, # None = self-play, "moonfish" = moonfish engine, "random" = random
opponent_depth: int = 2, # Search depth for moonfish opponent
):
"""
Initialize the chess environment.
Args:
reward_config: Configuration for reward shaping
max_moves: Maximum half-moves before draw (prevents infinite games)
agent_color: Which color the RL agent plays (None = alternates each episode)
opponent: Opponent type - None (self-play), "moonfish", or "random"
opponent_depth: Search depth when using moonfish as opponent
"""
self.reward_config = reward_config or RewardConfig()
self.max_moves = max_moves
self.agent_color_setting = agent_color
self.opponent = opponent
self.opponent_depth = opponent_depth
# Will be set on reset
self._board: Optional[chess.Board] = None
self._state: Optional[ChessState] = None
self._agent_color: bool = chess.WHITE
def reset(
self,
seed: Optional[int] = None,
episode_id: Optional[str] = None,
fen: Optional[str] = None,
**kwargs,
) -> ChessObservation:
"""
Initialize a new chess game episode.
Args:
seed: Random seed (unused for now, chess is deterministic)
episode_id: Unique identifier for this episode
fen: Optional starting position in FEN notation
Returns:
Initial observation of the board state
"""
# Create new board
if fen:
self._board = chess.Board(fen)
else:
self._board = chess.Board()
# Determine agent color
if self.agent_color_setting is None:
# Alternate each episode based on episode_id hash
if episode_id:
self._agent_color = hash(episode_id) % 2 == 0
else:
self._agent_color = chess.WHITE
else:
self._agent_color = self.agent_color_setting
# Initialize state
self._state = ChessState(
episode_id=episode_id or uuid.uuid4().hex,
step_count=0,
current_player="white" if self._board.turn else "black",
fen=self._board.fen(),
move_history=[],
)
# If agent plays Black and opponent is configured, opponent moves first
if self.opponent is not None and self._agent_color == chess.BLACK:
self._make_opponent_move()
return self._get_observation()
def step(
self, action: ChessAction, timeout_s: Optional[float] = None, **kwargs
) -> Tuple[ChessObservation, float, bool]:
"""
Execute a chess move and return the resulting state.
Args:
action: The move to make in UCI format (e.g., "e2e4")
timeout_s: Unused timeout parameter
Returns:
Tuple of (observation, reward, done)
"""
if self._board is None or self._state is None:
raise RuntimeError("Environment not initialized. Call reset() first.")
# Parse the move
try:
move = chess.Move.from_uci(action.move)
except ValueError:
# Invalid move format
return self._handle_illegal_move(f"Invalid move format: {action.move}")
# Check if move is legal
if move not in self._board.legal_moves:
return self._handle_illegal_move(f"Illegal move: {action.move}")
# Execute the move
self._board.push(move)
self._state.step_count += 1
self._state.move_history.append(action.move)
self._state.current_player = "white" if self._board.turn else "black"
self._state.fen = self._board.fen()
# Calculate reward and check for game end
reward, done = self._calculate_reward_and_done()
# If game not over and opponent is configured, make opponent move
if not done and self.opponent is not None:
self._make_opponent_move()
# Recalculate after opponent move
opp_reward, done = self._calculate_reward_and_done()
# Opponent's reward is negative of ours (zero-sum)
reward += -opp_reward if done else 0
observation = self._get_observation(done=done, reward=reward if done else None)
return observation, reward, done
@property
def state(self) -> ChessState:
"""Return the current episode state."""
if self._state is None:
raise RuntimeError("Environment not initialized. Call reset() first.")
return self._state
def close(self) -> None:
"""Clean up resources."""
self._board = None
self._state = None
def get_metadata(self) -> Dict[str, Any]:
"""Return environment metadata."""
return {
"name": "chess",
"version": "1.0.0",
"max_moves": self.max_moves,
"reward_config": {
"win": self.reward_config.win,
"loss": self.reward_config.loss,
"draw": self.reward_config.draw,
"illegal_move": self.reward_config.illegal_move,
"use_evaluation": self.reward_config.use_evaluation,
"evaluation_scale": self.reward_config.evaluation_scale,
},
}
def _get_observation(
self,
done: bool = False,
reward: Optional[float] = None,
result: Optional[str] = None,
error: Optional[str] = None,
) -> ChessObservation:
"""Build observation from current board state."""
assert self._board is not None
legal_moves = [move.uci() for move in self._board.legal_moves]
metadata: Dict[str, Any] = {}
# Add evaluation if configured
if self.reward_config.use_evaluation:
metadata["evaluation"] = board_evaluation(self._board)
# Add material count
metadata["material"] = self._get_material_count()
# Add game phase (0 = opening, 256 = endgame)
metadata["phase"] = get_phase(self._board)
metadata["fullmove_number"] = self._board.fullmove_number
metadata["halfmove_clock"] = self._board.halfmove_clock
if error:
metadata["error"] = error
# Determine result string if game is over
if done and result is None:
result = self._get_result_string()
return ChessObservation(
fen=self._board.fen(),
legal_moves=legal_moves,
is_check=self._board.is_check(),
done=done,
reward=reward,
result=result,
metadata=metadata,
)
def _calculate_reward_and_done(self) -> Tuple[float, bool]:
"""Calculate reward and check if episode is done."""
assert self._board is not None
# Check for game end
if self._board.is_checkmate():
# The side to move is checkmated, so the previous mover won
winner = not self._board.turn
if winner == self._agent_color:
return self.reward_config.win, True
else:
return self.reward_config.loss, True
if self._board.is_stalemate():
return self.reward_config.draw, True
if self._board.is_insufficient_material():
return self.reward_config.draw, True
if self._board.is_fifty_moves():
return self.reward_config.draw, True
if self._board.is_repetition(3):
return self.reward_config.draw, True
# Check move limit
if self._state and self._state.step_count >= self.max_moves:
return self.reward_config.draw, True
# Game continues
reward = 0.0
# Optional: Add evaluation-based intermediate rewards
if self.reward_config.use_evaluation:
eval_score = board_evaluation(self._board)
# Normalize evaluation to agent's perspective
if self._board.turn != self._agent_color:
eval_score = -eval_score
reward = eval_score * self.reward_config.evaluation_scale
return reward, False
def _handle_illegal_move(
self, error_msg: str
) -> Tuple[ChessObservation, float, bool]:
"""Handle an illegal move attempt."""
observation = self._get_observation(done=False, error=error_msg)
return observation, self.reward_config.illegal_move, False
def _get_result_string(self) -> str:
"""Get the game result as a string."""
assert self._board is not None
if self._board.is_checkmate():
return "1-0" if not self._board.turn else "0-1"
return "1/2-1/2"
def _get_material_count(self) -> Dict[str, int]:
"""Count material for both sides using moonfish piece values."""
assert self._board is not None
# count_pieces returns [wp, bp, wn, bn, wb, bb, wr, br, wq, bq]
pieces = count_pieces(self._board)
wp, bp, wn, bn, wb, bb, wr, br, wq, bq = pieces
white = (
wp * MG_PIECE_VALUES[chess.PAWN]
+ wn * MG_PIECE_VALUES[chess.KNIGHT]
+ wb * MG_PIECE_VALUES[chess.BISHOP]
+ wr * MG_PIECE_VALUES[chess.ROOK]
+ wq * MG_PIECE_VALUES[chess.QUEEN]
)
black = (
bp * MG_PIECE_VALUES[chess.PAWN]
+ bn * MG_PIECE_VALUES[chess.KNIGHT]
+ bb * MG_PIECE_VALUES[chess.BISHOP]
+ br * MG_PIECE_VALUES[chess.ROOK]
+ bq * MG_PIECE_VALUES[chess.QUEEN]
)
return {"white": white, "black": black}
def _make_opponent_move(self) -> None:
"""Make a move for the opponent using configured strategy."""
assert self._board is not None
assert self._state is not None
if not list(self._board.legal_moves):
return # No legal moves (game should be over)
if self.opponent == "moonfish":
# Use moonfish engine to find best move
move = search_move(self._board, depth=self.opponent_depth)
elif self.opponent == "random":
# Pick a random legal move
move = random.choice(list(self._board.legal_moves))
else:
return # No opponent configured
# Execute opponent's move
self._board.push(move)
self._state.step_count += 1
self._state.move_history.append(move.uci())
self._state.current_player = "white" if self._board.turn else "black"
self._state.fen = self._board.fen()
|