gym-tic-tac-toe/gym_tic_tac_toe/envs/tic_tac_toe_env.py
Nicolai Czempin 73495ae7e4 cleanup
2017-04-04 23:07:45 +02:00

66 lines
2.1 KiB
Python

import gym
from gym import spaces
import numpy as np
class TicTacToeEnv(gym.Env):
metadata = {'render.modes': ['human']}
def __init__(self):
self.action_space = spaces.Discrete(9)
self.observation_space = spaces.Discrete(512*512*2) # flattened
def _step(self, action):
done = False
reward = 0
p, square = action
# p = p*2 - 1
# check move legality
board = self.state['board']
proposed = board[square]
om = self.state['on_move']
if (proposed != 0): # wrong player, not empty
print("illegal move ", action, ". (square occupied): ", square)
done = True
reward = -1 * om # player who did NOT make the illegal move
if (p != om): # wrong player, not empty
print("illegal move ", action, " not on move: ", p)
done = True
reward = -1 * om # player who did NOT make the illegal move
else:
board[square] = p
self.state['on_move'] = -p
# check game over
for i in range(3):
# horizontals and verticals
if ((board[i * 3] == p and board[i * 3 + 1] == p and board[i * 3 + 2 ] == p)
or (board[i + 0] == p and board[i + 3] == p and board[i + 6] == p)):
reward = p
done = True
break
return self.state, reward, done, {}
def _reset(self):
self.state = {}
self.state['board'] = [0, 0, 0, 0, 0, 0, 0, 0, 0]
self.state['on_move'] = 1
return self.state
def _render(self, mode='human', close=False):
if close:
return
print("on move: " , self.state['on_move'])
for i in range (9):
print (self.state['board'][i], end=" ")
print()
def move_generator(self):
moves = []
for i in range (9):
if (self.state['board'][i] == 0):
p = self.state['on_move']
m = [p, i]
moves.append(m)
return moves