Initial commit from Openverse builder
This commit is contained in:
61
env.py
Normal file
61
env.py
Normal file
@@ -0,0 +1,61 @@
|
||||
import re
|
||||
from typing import Optional, Dict, Tuple, Any
|
||||
|
||||
import textarena as ta
|
||||
|
||||
class IteratedRockPaperScissorsEnv(ta.Env):
|
||||
def __init__(self, num_rounds: int = 5):
|
||||
self.num_rounds = num_rounds
|
||||
|
||||
def reset(self, num_players: int, seed: Optional[int] = None):
|
||||
self.state = ta.TwoPlayerState(num_players=num_players, seed=seed)
|
||||
game_state = {"round": 1, "points": {0:0,1:0}, "moves": {0:None,1:None}, "history": []}
|
||||
self.state.reset(game_state=game_state, player_prompt_function=self._generate_player_prompt)
|
||||
|
||||
def _generate_player_prompt(self, player_id: int, game_state: Dict[str, Any]) -> str:
|
||||
return (
|
||||
f"You are Player {player_id} in a {self.num_rounds}-round Rock-Paper-Scissors game.\nYour goal is to win as many rounds as possible.\n"
|
||||
"In each round, respond with one of: '[rock]', '[paper]', or '[scissors]'.\nYou may also use '[r]', '[p]', or '[s]' as shorthand.\n"
|
||||
)
|
||||
|
||||
def step(self, action: str) -> Tuple[bool, ta.Info]:
|
||||
player_id = self.state.current_player_id
|
||||
self.state.add_observation(from_id=player_id, to_id=player_id, message=action, observation_type=ta.ObservationType.PLAYER_ACTION)
|
||||
|
||||
move = self._parse_action(action)
|
||||
if move not in {"rock", "paper", "scissors"}:
|
||||
self.state.set_invalid_move(reason=f"Move not recognized. Use [rock], [paper], or [scissors].")
|
||||
else:
|
||||
self.state.game_state["moves"][player_id] = move
|
||||
self.state.add_observation(from_id=player_id, to_id=player_id, message=f"Player {player_id} selects move {move}.", observation_type=ta.ObservationType.GAME_ACTION_DESCRIPTION)
|
||||
|
||||
if self.state.game_state["moves"][1-player_id] != None: # Resolve the round
|
||||
p0_move = self.state.game_state["moves"][0]
|
||||
p1_move = self.state.game_state["moves"][1]
|
||||
result = self._resolve_round(p0_move, p1_move)
|
||||
self.state.game_state["history"].append({0:p0_move,1:p1_move})
|
||||
self.state.game_state["round"] += 1
|
||||
self.state.game_state["moves"] = {0:None, 1:None}
|
||||
|
||||
if result == 0:
|
||||
self.state.add_observation(message="Round result: Draw", observation_type=ta.ObservationType.GAME_MESSAGE)
|
||||
else:
|
||||
self.state.add_observation(message=f"Round result: Player {result-1} wins!", observation_type=ta.ObservationType.GAME_MESSAGE)
|
||||
self.state.game_state["points"][result-1] += 1
|
||||
|
||||
if self.state.game_state["round"] > self.num_rounds: # Check end condition
|
||||
wins = self.state.game_state.get("points", {0: 0, 1: 0})
|
||||
if wins[0] > wins[1]: self.state.set_winner(player_id=0, reason="Player 0 won the most rounds!")
|
||||
elif wins[1] > wins[0]: self.state.set_winner(player_id=1, reason="Player 1 won the most rounds!")
|
||||
else: self.state.set_draw("The match is a draw!")
|
||||
|
||||
return self.state.step()
|
||||
|
||||
def _parse_action(self, action: str) -> str:
|
||||
match = re.search(r"\[(rock|r|paper|p|scissors|s)\]", action.strip().lower())
|
||||
if not match: return ""
|
||||
return {"r": "rock", "p": "paper", "s": "scissors"}.get(match.group(1), match.group(1))
|
||||
|
||||
def _resolve_round(self, p0: str, p1: str) -> int:
|
||||
if p0 == p1: return 0
|
||||
return 1 if {"rock": "scissors", "paper": "rock", "scissors": "paper",}[p0] == p1 else 2
|
||||
Reference in New Issue
Block a user