def test_max_rank(self):
children = {
1: tictactoe.Board(),
5: tictactoe.Board(),
3: tictactoe.Board()
}
children[1].rank = 0
children[5].rank = 1
children[3].rank = -1
maxKey = max(children.iterkeys(), key=(lambda key: children[key].rank))
self.assertEqual(5, maxKey)
def matching_diag(self):
board1 = tictactoe.Board()
board1.boardArr[0] = 'X'
board1.boardArr[4] = 'X'
board1.boardArr[8] = 'X'
self.assertTrue(board1.matching_diag('X'))
board2 = tictactoe.Board()
board2.boardArr[2] = 'X'
board2.boardArr[4] = 'X'
board2.boardArr[6] = 'X'
self.assertTrue(board2.matching_diag('X'))
board_no_match = tictactoe.Board()
self.assertFalse(board_no_match.matching_diag('X'))
self.assertFalse(board_no_match.matching_diag('O'))
def test_switch_player(self):
board = tictactoe.Board()
board.current_plyr = 'X'
board.switch_player()
self.assertEqual(board.current_plyr, 'O')
board.switch_player()
self.assertEqual(board.current_plyr, 'X')
def test_full_square(self):
board = tictactoe.Board()
column = row = 0
player = 'X'
board.mark_square(column, row, player)
self.assertFalse(board.mark_square(column, row, player))
def test_DiagonalWinner(self):
board = tictactoe.Board()
win = Win_Checker()
board.board[0][0] = 'X'
board.board[1][1] = 'X'
board.board[2][2] = 'X'
self.assertEqual(True, win.has_winner(board.board))
def test_move_on_taken_cell(self):
_ = tictactoe.EMPTY
X = tictactoe.CROSS
O = tictactoe.NOUGHT
board = tictactoe.Board()
newBoard = board.move(4).move(4) # should return None
self.assertEqual(None, newBoard)
def test_build_tree_pre_win(self):
_ = tictactoe.EMPTY
X = tictactoe.CROSS
O = tictactoe.NOUGHT
board = tictactoe.Board([_, O, O, _, O, X, X, _, _])
board.nextMove = O
tictactoe.buildGameTree(board)
self.assertEqual(-1, board.rank)
def test_goodInput(self):
#arrange
board = tictactoe.Board()
row = 0
column = 0
player = 'X'
marker = Square_Marker()
#act and assert
self.assertEqual(True, marker.mark_square(board.board, column, row, player))
def test_move_onEmpty(self):
_ = tictactoe.EMPTY
X = tictactoe.CROSS
O = tictactoe.NOUGHT
board = tictactoe.Board()
newBoard = board.move(4) # center cells
self.assertEqual([_] * 9, board.cells)
self.assertEqual([_, _, _, _, O, _, _, _, _], newBoard.cells)
self.assertEqual(X, newBoard.nextMove)
def test_badPlayer(self):
#arrange
board = tictactoe.Board()
row = 0
column = 0
player = 'b'
marker = Square_Marker()
#act and assert
self.assertEqual(False, marker.mark_square(board.board, column, row, player))
def test_outOfBounds(self):
#arrange
board = tictactoe.Board()
row = 4
column = 3
player = 'X'
marker = Square_Marker()
#act and assert
self.assertEqual(False, marker.mark_square(board.board, column, row, player))
def test_move_cross(self):
_ = tictactoe.EMPTY
X = tictactoe.CROSS
O = tictactoe.NOUGHT
board = tictactoe.Board()
newBoard = board.move(4).move(
0) # O in the center, X in the top left corner
self.assertEqual([_] * 9, board.cells)
self.assertEqual([X, _, _, _, O, _, _, _, _], newBoard.cells)
self.assertEqual(O, newBoard.nextMove)
def test_has_winner(self):
player = 'X'
winner_board = tictactoe.Board()
winner_board.boardArr[0] = 'X'
winner_board.boardArr[1] = 'X'
winner_board.boardArr[2] = 'X'
self.assertTrue(winner_board.has_winner(player))
col_board = tictactoe.Board()
col_board.boardArr[0] = 'X'
col_board.boardArr[3] = 'X'
col_board.boardArr[6] = 'X'
self.assertTrue(col_board.has_winner(player))
diag_board = tictactoe.Board()
diag_board.boardArr[0] = 'X'
diag_board.boardArr[4] = 'X'
diag_board.boardArr[8] = 'X'
self.assertTrue(diag_board.has_winner(player))
def random_v_random(n=1):
"""
Plays two random players against each other
"""
p1_strategy = strategies.RandomStrategy()
p2_strategy = strategies.RandomStrategy()
p1 = player.Player('X', p1_strategy)
p2 = player.Player('O', p2_strategy)
board = tictactoe.Board()
game = rl_game.Game(p1, p2, board)
game.play_one()
class ifCharacter(unittest.TestCase):
board = tt.Board()
def test_not_chara(self):
self.assertEqual(self.board.if_character('Y'), False)
def test_is_chara(self):
self.assertEqual(self.board.if_character('X'), True)
def test_isnot_chara(self):
self.assertEqual(self.board.if_character('NOT'), False)
def random_v_q(n=1000):
"""
Trains the RL agent against random
"""
for i in range(5):
p1_strategy = strategies.RandomStrategy()
p2_strategy = strategies.QStrategy('O')
p1 = player.Player('X', p1_strategy)
p2 = player.Player('O', p2_strategy)
board = tictactoe.Board()
game = rl_game.Game(p1, p2, board)
game.play_many(n)
class InBounds(unittest.TestCase):
board = tt.Board()
def test_Out_of_Bound(self):
self.assertEqual(self.board.in_bound(3, 3), False)
def test_in_Bound(self):
self.assertEqual(self.board.in_bound(2, 2), True)
def test_Out_of_Bound2(self):
self.assertEqual(self.board.in_bound(3, 12), False)
class Test_Has_Winner(unittest.TestCase):
board = BoardClass.Board()
def test_Horizontal_Win(self):
print("Testing horizontal win")
self.board.game_board = [['0', "X", '0'], ['0', "X", '0'],
['0', "X", '0']]
self.assertEqual(self.board.has_winner(), "0")
def q_v_q():
"""
Plays two Q-strategy players against each other
"""
p1_strategy = strategies.QStrategy('X')
p2_strategy = strategies.QStrategy('O')
p1 = player.Player('X', p1_strategy)
p2 = player.Player('O', p2_strategy)
board = tictactoe.Board()
game = rl_game.Game(p1, p2, board)
game.play_one()
p1.strategy.save_q()
p2.strategy.save_q()
def text_matching_col(self):
board1 = tictactoe.Board()
board1.boardArr[0] = 'X'
board1.boardArr[3] = 'X'
board1.boardArr[6] = 'X'
self.assertTrue(board1.matching_col('X'))
board2 = tictactoe.Board()
board2.boardArr[1] = 'X'
board2.boardArr[4] = 'X'
board2.boardArr[7] = 'X'
self.assertTrue(board2.matching_col('X'))
board3 = tictactoe.Board()
board3.boardArr[2] = 'X'
board3.boardArr[5] = 'X'
board3.boardArr[8] = 'X'
self.assertTrue(board3.matching_col('X'))
board_no_match = tictactoe.Board()
self.assertFalse(board_no_match.matching_col('X'))
self.assertFalse(board_no_match.matching_col('O'))
def test_NoWinnerTie(self):
board = tictactoe.Board()
win = Win_Checker()
board.board[0][0] = 'X'
board.board[0][1] = 'O'
board.board[0][2] = 'X'
board.board[1][0] = 'X'
board.board[1][1] = 'O'
board.board[1][2] = 'O'
board.board[2][0] = 'O'
board.board[2][1] = 'X'
board.board[2][2] = 'O'
self.assertEqual(False, win.has_winner(board.board))
def train_q(n=1000):
"""
Trains the RL agent
"""
for i in range(50):
p1_strategy = strategies.QStrategy('X')
p2_strategy = strategies.QStrategy('O')
p1 = player.Player('X', p1_strategy)
p2 = player.Player('O', p2_strategy)
board = tictactoe.Board()
game = rl_game.Game(p1, p2, board)
game.play_many(n)
p1.strategy.save_q()
p2.strategy.save_q()
class Test_check_unoccupied(unittest.TestCase):
board = BoardClass.Board()
def test_unoccupied(self):
print("Testing unoccupied")
self.board.game_board[2][2] = '0'
self.assertTrue(self.board.check_unoccupied(2, 2))
def test_occupied(self):
print("Testing occupied")
self.board.game_board[2][2] = 'X'
self.assertFalse(self.board.check_unoccupied(2, 2))
class Test_check_inbounds(unittest.TestCase):
game_board = BoardClass.Board()
def test_both_valid_equal(self):
print("Testing valid equal")
self.assertTrue(self.game_board.check_inbounds(2, 2))
def test_both_valid(self):
print("Testing both valid")
self.assertTrue(self.game_board.check_inbounds(2, 1))
def test_both_invalid(self):
print("Testing both invalid")
self.assertFalse(self.game_board.check_inbounds(4, 4))
def test_one_valid_one_invalid(self):
print("Testing valid and invalid")
self.assertFalse(self.game_board.check_inbounds(4, 1))
def test_space_marked(self):
board = tictactoe.Board()
board.mark_square(0, 'X')
self.assertEqual(board.boardArr[0], "X")
board.mark_square(1, 'X')
self.assertEqual(board.boardArr[1], "X")
board.mark_square(2, 'X')
self.assertEqual(board.boardArr[2], "X")
board.mark_square(3, 'X')
self.assertEqual(board.boardArr[3], "X")
board.mark_square(4, 'X')
self.assertEqual(board.boardArr[4], "X")
board.mark_square(5, 'X')
self.assertEqual(board.boardArr[5], "X")
board.mark_square(6, 'X')
self.assertEqual(board.boardArr[6], "X")
board.mark_square(7, 'X')
self.assertEqual(board.boardArr[7], "X")
board.mark_square(8, 'X')
self.assertEqual(board.boardArr[8], "X")
def human_v_q(human_player=1):
"""
Allows human to play versus q-learning agent
"""
if human_player == 1:
p1_strategy = strategies.Human()
p2_strategy = strategies.QStrategy('O')
else:
human_player = 2
p1_strategy = strategies.QStrategy('X')
p2_strategy = strategies.Human()
p1 = player.Player('X', p1_strategy)
p2 = player.Player('O', p2_strategy)
board = tictactoe.Board()
message = 'Welcome to tic tac toe!\n'+\
'You are playing against a random opponent and you are player '+str(human_player)+'.\n'+\
'To make a move, enter the number of the square which you would like to play, labelled as:\n'+\
'1, 2, 3\n4, 5, 6\n7, 8, 9\n\n'
print(message)
game = rl_game.Game(p1, p2, board)
game.play_one()
def play_training_game(network, c_puct, iterations, display=False):
initial_state = game.Board()
current_node = MCTS_node(initial_state, network)
draft_training_data = []
while current_node.state.is_end() == False:
for iter in range(iterations):
MCTS_tree_search(current_node, network, c_puct)
improved_policy = {}
for move in current_node.moves:
improved_policy[move] = float(
current_node.N[move]) / (current_node.visits - 1)
draft_training_data.append((current_node.state, improved_policy, 0))
move = get_move(improved_policy)
current_node = current_node.children[move]
if display:
print("\n\nMOVING FORWARD")
current_node.state.print_board()
print("Network value estimate:%s" % current_node.network_value)
game.print_policy(improved_policy)
draft_training_data.append((current_node.state, {(0, 0): 1}, 0))
reward = current_node.state.value()
training_data = []
for (state, policy, draft_reward) in draft_training_data:
training_data.append((state, policy, reward))
reward = -reward
#training_data[-1][0].print_board()
#print(training_data[-1][2])
#raw_input()
return training_data
class Test_check_valid_move(unittest.TestCase):
board = BoardClass.Board()
def test_valid_move(self):
print("Testing valid")
testMove = MoveClass.move(2, 2, 'X')
self.assertTrue(self.board.check_valid_move(testMove))
def test_valid_occupied(self):
print("Testing valid occupied")
self.board.game_board[2][2] = 'X'
testMove = MoveClass.move(2, 2, 'O')
self.assertFalse(self.board.check_valid_move(testMove))
def test_invalid(self):
print("Testing invalid")
testMove = MoveClass.move(3, 3, 'O')
self.assertFalse(self.board.check_valid_move(testMove))
def test_check_diag_left_diag_winner_player1(self):
board = tictactoe.Board()
board.board = [['X', 0, 0], [0, 'X', 0], [0, 0, 'X']]
winner = board.check_diag(board.board, 3)
self.assertEqual(winner, board.player1)
def test_column_no_winner(self):
board = tictactoe.Board()
board.board = [['X', 0, 0], ['O', 0, 0], ['X', 0, 0]]
winner = board.winner_in_column()
self.assertIsNone(winner)