def test_flop_check(self):
logic = Game_Helpers()
# 1 SB -> 2SB -> 6SB -> 4SB
blocking = np.zeros(52, dtype=int)
history = np.zeros(1300)
history = history.reshape(25, 13, 4)
stack = logic.chips_bitmask(500)
legal_moves_arr = logic.legal_moves(stack, history, blocking, 'FLOP')
legal_moves_expected = np.ones(52)
legal_moves_expected[0] = 0
self.assertEqual(legal_moves_expected.tolist(),
legal_moves_arr.tolist())
def test_blinds_and_call_covered(self):
logic = Game_Helpers()
blocking = np.zeros(52)
history = np.zeros(1300)
history = history.reshape(25, 13, 4)
history[0, 0, 0:1] = 1
history[1, 0, 0:2] = 1
history[2, 0, 0:2] = 1
stack = logic.chips_bitmask(500)
legal_moves_arr = logic.legal_moves(stack, history, blocking,
'PREFLOP')
legal_moves_expected = np.zeros(52)
legal_moves_expected[3:] = 1
self.assertEqual(legal_moves_expected.tolist(),
legal_moves_arr.tolist())
def test_normal_raise_uncovered_call_covered(self):
# 1 SB -> 2SB -> 16SB -> minraise = 28 SB but we only have a stack of 19 SB
logic = Game_Helpers()
blocking = np.zeros(52)
history = np.zeros(1300)
history = history.reshape(25, 13, 4)
history[0, 0, 0:1] = 1
history[1, 0, 0:2] = 1
history[2, 0:4, 0:4] = 1
stack = logic.chips_bitmask(190)
legal_moves_arr = logic.legal_moves(stack, history, blocking,
'PREFLOP')
legal_moves_expected = np.zeros(52)
legal_moves_expected[15] = 1
legal_moves_expected[18] = 1
legal_moves_expected[51] = 1
self.assertEqual(legal_moves_expected.tolist(),
legal_moves_arr.tolist())
def test_just_blinds_covered(self):
logic = Game_Helpers()
blocking = np.zeros(50, dtype=int)
history = np.zeros(1300)
history = history.reshape(25, 13, 4)
# Small Blind
history[0, 0, 0:1] = 1
# Big Blind
history[1, 0, 0:2] = 1
stack = logic.chips_bitmask(500)
legal_moves_arr = logic.legal_moves(stack, history, blocking,
'PREFLOP')
legal_moves_expected = np.zeros(52)
legal_moves_expected[1] = 1
legal_moves_expected[3:50] = 1
legal_moves_expected[51] = 1
self.assertEqual(legal_moves_expected.tolist(),
legal_moves_arr.tolist())
def test_37sb_raise(self):
logic = Game_Helpers()
blocking = np.zeros(52)
history = np.zeros(1300)
history = history.reshape(25, 13, 4)
history[0, 0, 0:1] = 1
history[1, 0, 0:2] = 1
history[2, 0:9, 0:4] = 1
history[2, 9, 0:3] = 1
stack = logic.chips_bitmask(500)
legal_moves_arr = logic.legal_moves(stack, history, blocking,
'PREFLOP')
legal_moves_expected = np.zeros(52)
legal_moves_expected[38] = 1
legal_moves_expected[49] = 1
legal_moves_expected[51] = 1
self.assertEqual(legal_moves_expected.tolist(),
legal_moves_arr.tolist())
def test_8sb_45sb_raises(self):
logic = Game_Helpers()
# 1 SB -> 2SB -> 6SB -> 4SB
blocking = np.zeros(52, dtype=int)
history = np.zeros(1300)
history = history.reshape(25, 13, 4)
history[0, 0, 0:1] = 1
history[1, 0, 0:2] = 1
history[2, :12, :] = 1
stack = logic.chips_bitmask(410)
legal_moves_arr = logic.legal_moves(stack, history, blocking,
'PREFLOP')
legal_moves_expected = np.zeros(52)
legal_moves_expected[40] = 1
legal_moves_expected[51] = 1
#legal_moves_expected[49] = 1
self.assertEqual(legal_moves_expected.tolist(),
legal_moves_arr.tolist())
def test_blinds_underraise(self):
# 1 SB -> 2SB -> 3 SB
logic = Game_Helpers()
blocking = np.zeros(50, dtype=int)
blocking[2] = 1
history = np.zeros(1300)
history = history.reshape(25, 13, 4)
history[0, 0, 0:1] = 1
history[1, 0, 0:2] = 1
history[2, 0, 0:3] = 1
stack = logic.chips_bitmask(500)
legal_moves_arr = logic.legal_moves(stack, history, blocking,
'PREFLOP')
legal_moves_expected = np.zeros(52)
legal_moves_expected[2] = 1
legal_moves_expected[51] = 1
self.assertEqual(legal_moves_expected.tolist(),
legal_moves_arr.tolist())
class Player():
def __init__(self, name, chips,agent):
self.agent = agent
self.name = name
self.cards = None
self.hand = np.zeros(shape = (1,13,4))
self.chips_preflop = np.zeros(shape = (1,13,4))
self.chips_flop = np.zeros(shape = (1,13,4))
self.chips_turn = np.zeros(shape = (1,13,4))
self.chips_river = np.zeros(shape = (1,13,4))
self.chips_dict = {'PREFLOP': self.chips_preflop,'FLOP':self.chips_flop,'TURN':self.chips_turn
,'RIVER':self.chips_river}
self.is_bb = np.zeros(shape = (1,13,4))
self.chips = chips
self.id = id
self.has_acted = False
self.active = True
self.is_allin = False
self.has_folded = False
self.is_blocking = False
self.next_player = None
self.position = None
self.current_stack = None
self.game_helpers = Game_Helpers()
self.game_state = None
self.chips_dict['PREFLOP'] = self.game_helpers.chips_bitmask_plane(chips)
self.state_memory = []
def reset(self,chips):
self.cards = None
self.hand = np.zeros(shape = (1,13,4))
self.chips_preflop = np.zeros(shape = (1,13,4))
self.chips_flop = np.zeros(shape = (1,13,4))
self.chips_turn = np.zeros(shape = (1,13,4))
self.chips_river = np.zeros(shape = (1,13,4))
self.chips_dict = {'PREFLOP': self.chips_preflop, 'FLOP': self.chips_flop, 'TURN': self.chips_turn
, 'RIVER': self.chips_river}
self.is_bb = np.zeros(shape = (1,13,4))
self.chips = chips
self.id = id
self.has_acted = False
self.active = True
self.is_allin = False
self.has_folded = False
self.is_blocking = False
self.next_player = None
self.position = None
self.current_stack = None
self.game_helpers = Game_Helpers()
self.game_state = None
self.chips_dict['PREFLOP'] = self.game_helpers.chips_bitmask_plane(chips)
self.agent.reset()
self.state_memory = []
def set_hand(self,cardA,cardB):
self.hand = self.hand.flatten()
self.hand[cardA] = 1
self.hand[cardB] = 1
self.hand = self.hand.reshape(1, 13, 4)
def build_player_game_state(self,public_state):
self.game_state = public_state
self.game_state = np.concatenate((self.game_state, self.is_bb))
self.game_state = np.concatenate((self.game_state, self.hand))
def set_street_stacks(self,history,street,player_number):
current_stack = self.game_helpers.get_stack(history, self.chips, self.position, player_number, street)
self.chips_dict[street] = self.game_helpers.chips_bitmask_plane(current_stack)
def set_reward(self,reward):
for i in range(0,len(self.state_memory)):
rl_state = None
if i <= len(self.state_memory) -2:
rl_state = {'state0': self.state_memory[i]['state'],'action0':self.state_memory[i]['action'],
'state1':self.state_memory[i+1]['state'],'reward':0}
elif i == len(self.state_memory) -1:
rl_state = {'state0': self.state_memory[i]['state'], 'action0': self.state_memory[i]['action'],
'state1': 'Terminal', 'reward': reward}
self.agent.save_rl_transition(rl_state)
def get_move(self,history,blocking,player_number,street,public_state):
self.set_street_stacks(history, street, player_number)
current_history = history[street]
current_stack = self.game_helpers.get_stack(history,self.chips,self.position,player_number,street)
bitmask_stack = self.game_helpers.chips_bitmask(current_stack)
legals = self.game_helpers.legal_moves(bitmask_stack, current_history , blocking, street)
legals = np.argwhere(legals == 1)
self.build_player_game_state(public_state)
moves,game_state_dict = self.agent.get_move(history,blocking,player_number,street,self.game_state,legals,None)
self.state_memory.append(game_state_dict)
return moves,bitmask_stack