def test_start(self):
ttt = ttt_env.TicTacToeEnv(strict=True)
self.assertFalse(ttt.done)
# At max one move may have been played by the env.
spaces = ttt_env.get_open_spaces(ttt.board_state)
num_open_spaces = len(spaces)
# i.e. either 8 or 9
self.assertGreater(num_open_spaces, 7)
# Play a move
observation, reward, done, unused_info = ttt.step(spaces[0])
# The environment should also have played a move.
spaces = ttt_env.get_open_spaces(observation)
self.assertEqual(num_open_spaces - 2, len(spaces))
# Since at-max 3 moves have been played, the game can't end.
self.assertEqual(reward, 0)
self.assertFalse(done)
def test_keep_playing(self):
ttt = ttt_env.TicTacToeEnv(strict=False)
done = False
while not done:
# sample an action from the action space.
action = ttt.action_space.sample()
# play it -- could be a no-op since we don't see if positions are empty.
unused_observation, reward, done, unused_info = ttt.step(action)
# done is True, so either:
# we won
# env won or
# no space left
we_won = reward == 1
env_won = reward == -1
space = bool(ttt_env.get_open_spaces(ttt.board_state))
self.assertTrue(we_won or env_won or not space)
