def init_game(self):
''' Initialize players and state.
Returns:
dict: first state in one game
int: current player's id
'''
# initialize public variables
self.winner_id = None
self.history = []
# initialize players
self.players = [Player(num) for num in range(self.num_players)]
# initialize round to deal cards and determine landlord
self.round = Round()
self.round.initiate(self.players)
# initialize judger
self.judger = Judger(self.players)
# get state of first player
player_id = self.round.current_player
player = self.players[player_id]
others_hands = self._get_others_current_hand(player)
actions = list(self.judger.playable_cards[player_id])
state = player.get_state(self.round.public, others_hands, actions)
state['all_actions'] = state['actions']
self.state = state
return state, player_id
def test_playable_cards_from_hand(self):
# #solo
# in_cards, not_in_cards, hand = ('3', '4', '5', '6', '8', '9', 'J', 'Q', 'B', 'R'), (), '334445555689JQBR'
# playable_cards = Judger.playable_cards_from_hand(hand)
# for e in in_cards:
# self.assertIn(e, playable_cards)
# for e in not_in_cards:
# self.assertNotIn(e, playable_cards)
# #pair
# in_cards, not_in_cards, hand = ('33', '44', '55'), (), '334445555689JQBR'
# playable_cards = Judger.playable_cards_from_hand(hand)
# for e in in_cards:
# self.assertIn(e, playable_cards)
# for e in not_in_cards:
# self.assertNotIn(e, playable_cards)
# #trio
# in_cards, not_in_cards, hand = ('444', '555'), (), '334445555689JQBR'
# playable_cards = Judger.playable_cards_from_hand(hand)
# for e in in_cards:
# self.assertIn(e, playable_cards)
# for e in not_in_cards:
# self.assertNotIn(e, playable_cards)
# #bomb
# in_cards, not_in_cards, hand = ('5555', ), (), '334445555689JQBR'
# playable_cards = Judger.playable_cards_from_hand(hand)
# for e in in_cards:
# self.assertIn(e, playable_cards)
# for e in not_in_cards:
# self.assertNotIn(e, playable_cards)
# #rocket
# in_cards, not_in_cards, hand = ('BR', ), (), '334445555689JQBR'
# playable_cards = Judger.playable_cards_from_hand(hand)
# for e in in_cards:
# self.assertIn(e, playable_cards)
# for e in not_in_cards:
# self.assertNotIn(e, playable_cards)
# #solo_chain_5 -- #solo_chain_12
# in_cards = ('34567', '345678', '3456789', '3456789T', '3456789TJ',
# '3456789TJQ', '3456789TJQK', '3456789TJQKA', '45678', '456789',
# '3456789T', '3456789TJ', '456789TJQ', '456789TJQK', '456789TJQKA',
# '56789', '56789T', '56789TJ', '56789TJQ', '56789TJQK', '56789TJQKA',
# '6789T', '6789TJ', '6789TJQ', '6789TJQK', '6789TJQKA', '789TJ',
# '789TJQ', '789TJQK', '789TJQKA', '89TJQ', '89TJQK',
# '89TJQKA', '9TJQK', '9TJQKA', 'TJQKA')
# not_in_cards = ('JQKA2', )
# hand = '3344455556789TJQKA2BR'
# playable_cards = Judger.playable_cards_from_hand(hand)
# for e in in_cards:
# self.assertIn(e, playable_cards)
# for e in not_in_cards:
# self.assertNotIn(e, playable_cards)
# #pair_chain_3 -- #pair_chain_10
# in_cards = ('334455', '33445566', '3344556677', '334455667788', '33445566778899',
# '33445566778899TT', '33445566778899TTJJ', '33445566778899TTJJQQ', '445566', '44556677',
# '4455667788', '445566778899', '445566778899TT', '445566778899TTJJ', '445566778899TTJJQQ',
# '445566778899TTJJQQKK', '556677', '55667788', '5566778899', '5566778899TT',
# '5566778899TTJJ', '5566778899TTJJQQ', '5566778899TTJJQQKK', '5566778899TTJJQQKKAA',
# '667788', '66778899', '66778899TT', '66778899TTJJ', '66778899TTJJQQ',
# '66778899TTJJQQKK', '66778899TTJJQQKKAA', '778899', '778899TT', '778899TTJJ',
# '778899TTJJQQ', '778899TTJJQQKK', '778899TTJJQQKKAA', '8899TT', '8899TTJJ',
# '8899TTJJQQ', '8899TTJJQQKK', '8899TTJJQQKKAA', '99TTJJ', '99TTJJQQ',
# '99TTJJQQKK', '99TTJJQQKKAA', 'TTJJQQ', 'TTJJQQKK', 'TTJJQQKKAA',
# 'JJQQKK', 'JJQQKKAA', 'QQKKAA')
# not_in_cards = ('33445566778899TTJJQQKK', 'KKAA22')
# hand = '33444555566778899TTJJQQKKAA22'
# playable_cards = Judger.playable_cards_from_hand(hand)
# for e in in_cards:
# self.assertIn(e, playable_cards)
# for e in not_in_cards:
# self.assertNotIn(e, playable_cards)
# #trio_chain_2 -- #trio_chain_6
# in_cards = ('333444', '333444555', '333444555666', '333444555666777', '333444555666777888',
# '444555', '444555666', '444555666777', '444555666777888', '444555666777888999',
# '555666', '555666777', '555666777888', '555666777888999', '555666777888999TTT',
# '666777', '666777888', '666777888999', '666777888999TTT', '666777888999TTTJJJ',
# '777888', '777888999', '777888999TTT', '777888999TTTJJJ', '777888999TTTJJJQQQ',
# '888999', '888999TTT', '888999TTTJJJ', '888999TTTJJJQQQ', '888999TTTJJJQQQKKK',
# '999TTT', '999TTTJJJ', '999TTTJJJQQQ', '999TTTJJJQQQKKK', '999TTTJJJQQQKKKAAA',
# 'TTTJJJ', 'TTTJJJQQQ', 'TTTJJJQQQKKK', 'TTTJJJQQQKKKAAA',
# 'JJJQQQ', 'JJJQQQKKK', 'JJJQQQKKKAAA',
# 'QQQKKK', 'QQQKKKAAA',
# 'KKKAAA')
# not_in_cards = ('333444555666777888999', 'AAA222')
# hand = '333444455556667778889999TTTJJJQQQKKKAAA222BR'
# playable_cards = Judger.playable_cards_from_hand(hand)
# for e in in_cards:
# self.assertIn(e, playable_cards)
# for e in not_in_cards:
# self.assertNotIn(e, playable_cards)
# #trio_solo, trio_pair
# in_cards = ('3777', '4777', '5777', '6777', '7778',
# '7772', '44777', '55777', '77788', '77722')
# not_in_cards = ()
# hand = '344455677778889TJQKA2222BR'
# playable_cards = Judger.playable_cards_from_hand(hand)
# for e in in_cards:
# self.assertIn(e, playable_cards)
# for e in not_in_cards:
# self.assertNotIn(e, playable_cards)
# #trio_solo_chain_2 -- trio_solo_chain_5
# in_cards = ('34777888', '3777888T', '3456777888999TTTJJJQ', '66777888')
# not_in_cards = ('37777888', '777888999TTTJJJQK2BR', '777888999TTTJJJJ')
# hand = '34556677778888999TTTTJJJJQQQQKA2BR'
# playable_cards = Judger.playable_cards_from_hand(hand)
# for e in in_cards:
# self.assertIn(e, playable_cards)
# for e in not_in_cards:
# self.assertNotIn(e, playable_cards)
# #trio_pair_chain_2 -- #trio_pair_chain_4
# in_cards = ('5566777888', '55777888TT')
# not_in_cards = ('777888QQQQ', )
# hand = '34556677778888999TTTTJJJJQQQQKA2BR'
# playable_cards = Judger.playable_cards_from_hand(hand)
# for e in in_cards:
# self.assertIn(e, playable_cards)
# for e in not_in_cards:
# self.assertNotIn(e, playable_cards)
# #four_two_solo, #four_two_pair
# in_cards = ('357777', '557777', '567777', '777788', '55667777', '66777799', '557777TT', '55777788')
# not_in_cards = ('77778888', )
# hand = '34556677778888999TTTTJJJJQQQQKA2BR'
# playable_cards = Judger.playable_cards_from_hand(hand)
# for e in in_cards:
# self.assertIn(e, playable_cards)
# for e in not_in_cards:
# self.assertNotIn(e, playable_cards)
playable_cards = list(
Judger.playable_cards_from_hand(
'3333444455556666777788889999TTTTJJJJQQQQKKKKAAAA2222BR'))
all_cards_list = CARD_TYPE[1]
for c in playable_cards:
# if (c not in all_cards_list):
# print(c)
self.assertIn(c, all_cards_list)
for c in all_cards_list:
# if (c not in playable_cards):
# print('\t' + c)
self.assertIn(c, playable_cards)
self.assertEqual(len(playable_cards), len(all_cards_list))
class DoudizhuGame(object):
''' Provide game APIs for env to run doudizhu and get corresponding state
information.
An example of state during runtime:
{
'deck': '3333444455556666777788889999TTTTJJJJQQQQKKKKAAAA2222BR',
'seen_cards': 'TQA',
'landlord': 0,
'self': 2,
'initial_hand': '3456677799TJQKAAB',
'trace': [(0, '8222'), (1, 'pass'), (2, 'pass'), (0, '6KKK'),
(1, 'pass'), (2, 'pass'), (0, '8'), (1, 'Q')],
'played_cards': ['6', '8', '8', 'Q', 'K', 'K', 'K', '2', '2', '2'],
'others_hand': '333444555678899TTTJJJQQAA2R',
'current_hand': '3456677799TJQKAAB',
'actions': ['pass', 'K', 'A', 'B']
}
'''
def __init__(self, allow_step_back=False):
self.allow_step_back = allow_step_back
self.num_players = 3
def init_game(self):
''' Initialize players and state.
Returns:
dict: first state in one game
int: current player's id
'''
# initialize public variables
self.winner_id = None
self.history = []
# initialize players
self.players = [Player(num) for num in range(self.num_players)]
# initialize round to deal cards and determine landlord
self.round = Round()
self.round.initiate(self.players)
# initialize judger
self.judger = Judger(self.players)
# get state of first player
player_id = self.round.current_player
player = self.players[player_id]
others_hands = self._get_others_current_hand(player)
actions = list(self.judger.playable_cards[player_id])
state = player.get_state(self.round.public, others_hands, actions)
state['all_actions'] = state['actions']
self.state = state
return state, player_id
def step(self, action):
''' Perform one draw of the game
Args:
action (str): specific action of doudizhu. Eg: '33344'
Returns:
dict: next player's state
int: next player's id
'''
if self.allow_step_back:
# TODO: don't record game.round, game.players, game.judger if allow_step_back not set
pass
# perfrom action
player = self.players[self.round.current_player]
self.round.proceed_round(player, action)
if (action != 'pass'):
self.judger.calc_playable_cards(player)
if self.judger.judge_game(self.players, self.round.current_player):
self.winner_id = self.round.current_player
next_id = get_downstream_player_id(player, self.players)
self.round.current_player = next_id
# get next state
state = self.get_state(next_id)
self.state = state
#print("doudizhu状态:",state)
return state, next_id
def step_back(self):
''' Return to the previous state of the game
Returns:
(bool): True if the game steps back successfully
'''
if not self.round.trace:
return False
#winner_id will be always None no matter step_back from any case
self.winner_id = None
#reverse round
player_id, cards = self.round.trace.pop()
self.round.current_player = player_id
if (cards != 'pass'):
for card in cards:
self.round.played_cards.remove(card)
greater_player_id = self.round.find_last_greater_player_id_in_trace()
if (greater_player_id is not None):
self.round.greater_player = self.players[greater_player_id]
else:
self.round.greater_player = None
#reverse player
if (cards != 'pass'):
self.players[
player_id].played_cards = self.round.find_last_played_cards_in_trace(
player_id)
self.players[player_id].play_back()
#reverse judger.played_cards if needed
if (cards != 'pass'):
self.judger.restore_playable_cards(player_id)
self.state = self.get_state(self.round.current_player)
return True
def get_state(self, player_id):
''' Return player's state
Args:
player_id (int): player id
Returns:
(dict): The state of the player
'''
player = self.players[player_id]
others_hands = self._get_others_current_hand(player)
if self.is_over():
actions = None
all_actions = None
else:
actions, all_actions = player.available_actions(
self.round.greater_player, self.judger)
actions = list(actions)
all_actions = list(all_actions)
state = player.get_state(self.round.public, others_hands, actions)
state['all_actions'] = all_actions
return state
@staticmethod
def get_action_num():
''' Return the total number of abstract acitons
Returns:
int: the total number of abstract actions of doudizhu
'''
return 309
def get_player_id(self):
''' Return current player's id
Returns:
int: current player's id
'''
return self.round.current_player
def get_player_num(self):
''' Return the number of players in doudizhu
Returns:
int: the number of players in doudizhu
'''
return self.num_players
def is_over(self):
''' Judge whether a game is over
Returns:
Bool: True(over) / False(not over)
'''
if self.winner_id is None:
return False
return True
def _get_others_current_hand(self, player):
player_up = self.players[get_upstream_player_id(player, self.players)]
player_down = self.players[get_downstream_player_id(
player, self.players)]
others_hand = merge(player_up.current_hand,
player_down.current_hand,
key=functools.cmp_to_key(doudizhu_sort_card))
return cards2str(others_hand)
def test_judge_payoffs(self):
payoffs = Judger.judge_payoffs(0, 0)
self.assertEqual(payoffs[0], 1)
payoffs = Judger.judge_payoffs(2, 0)
self.assertEqual(payoffs[0], 1)
self.assertEqual(payoffs[1], 1)
class DoudizhuGame(object):
''' Provide game APIs for env to run doudizhu and get corresponding state
information.
An example of state during runtime:
{
'deck': '3333444455556666777788889999TTTTJJJJQQQQKKKKAAAA2222BR',
'seen_cards': 'TQA',
'landlord': 0,
'self': 2,
'initial_hand': '3456677799TJQKAAB',
'trace': [(0, '8222'), (1, 'pass'), (2, 'pass'), (0, '6KKK'),
(1, 'pass'), (2, 'pass'), (0, '8'), (1, 'Q')],
'played_cards': ['6', '8', '8', 'Q', 'K', 'K', 'K', '2', '2', '2'],
'others_hand': '333444555678899TTTJJJQQAA2R',
'current_hand': '3456677799TJQKAAB',
'actions': ['pass', 'K', 'A', 'B']
}
'''
def __init__(self, allow_step_back=False):
self.allow_step_back = allow_step_back
self.num_players = 3
def init_game(self):
''' Initialize players and state.
Returns:
dict: first state in one game
int: current player's id
'''
# initialize public variables
self.winner_id = None
self.history = []
# initialize players
self.players = [Player(num)
for num in range(self.num_players)]
# initialize round to deal cards and determine landlord
self.round = Round()
self.round.initiate(self.players)
# initialize judger
self.judger = Judger(self.players)
# get state of first player
player_id = self.round.current_player
player = self.players[player_id]
others_hands = self._get_others_current_hand(player)
actions = list(self.judger.playable_cards[player_id])
state = player.get_state(self.round.public, others_hands, actions)
self.state = state
return state, player_id
def step(self, action):
''' Perform one draw of the game
Args:
action (str): specific action of doudizhu. Eg: '33344'
Returns:
dict: next player's state
int: next player's id
'''
if self.allow_step_back:
# record game history
self._record_history()
# perfrom action
player = self.players[self.round.current_player]
self.round.proceed_round(player, action)
if self.judger.judge_game(self.players, self.round.current_player):
self.winner_id = self.round.current_player
next_id = get_downstream_player_id(player, self.players)
self.round.current_player = next_id
# get next state
state = self.get_state(next_id)
self.state = state
return state, next_id
def step_back(self):
''' Return to the previous state of the game
Returns:
(bool): True if the game steps back successfully
'''
if not self.history:
return False
self.winner_id, self.players, self.round, self.judger = self.history.pop()
return True
def get_state(self, player_id):
''' Return player's state
Args:
player_id (int): player id
Returns:
(dict): The state of the player
'''
player = self.players[player_id]
others_hands = self._get_others_current_hand(player)
if self.is_over():
actions = None
else:
actions = player.available_actions(self.round.greater_player, self.judger)
state = player.get_state(self.round.public, others_hands, actions)
return state
@staticmethod
def get_action_num():
''' Return the total number of abstract acitons
Returns:
int: the total number of abstract actions of doudizhu
'''
return 309
def get_player_id(self):
''' Return current player's id
Returns:
int: current player's id
'''
return self.round.current_player
def get_player_num(self):
''' Return the number of players in doudizhu
Returns:
int: the number of players in doudizhu
'''
return self.num_players
def is_over(self):
''' Judge whether a game is over
Returns:
Bool: True(over) / False(not over)
'''
if self.winner_id is None:
return False
return True
def _record_history(self):
''' Record history
'''
winner_id_cp = copy.deepcopy(self.winner_id)
players_cp = copy.deepcopy(self.players)
round_cp = copy.deepcopy(self.round)
judger_cp = copy.deepcopy(self.judger)
self.history.append((winner_id_cp, players_cp, round_cp, judger_cp))
def _get_others_current_hand(self, player):
player_up = self.players[get_upstream_player_id(player, self.players)]
player_down = self.players[get_downstream_player_id(
player, self.players)]
others_hand = (player_up.current_hand + player_down.current_hand)
others_hand.sort(key=functools.cmp_to_key(doudizhu_sort_card))
return cards2str(others_hand)
class SimpleDoudizhuGame(object):
''' Provide game APIs for env to run simple doudizhu and get corresponding state
information.
An example of state during runtime:
'''
def __init__(self, allow_step_back=False):
self.allow_step_back = allow_step_back
self.num_players = 3
def init_game(self):
''' Initialize players and state.
Returns:
dict: first state in one game
int: current player's id
'''
# initialize public variables
self.winner_id = None
self.history = []
# initialize players
self.players = [Player(num) for num in range(self.num_players)]
# initialize round to deal cards and determine landlord
self.round = Round()
self.round.initiate(self.players)
# initialize judger
self.judger = Judger(self.players)
# get state of first player
player_id = self.round.current_player
player = self.players[player_id]
others_hands = self._get_others_current_hand(player)
actions = list(self.judger.playable_cards[player_id])
state = player.get_state(self.round.public, others_hands, actions)
self.state = state
return state, player_id
def step(self, action):
''' Perform one draw of the game
Args:
action (str): specific action of doudizhu. Eg: '33344'
Returns:
dict: next player's state
int: next player's id
'''
if self.allow_step_back:
# TODO: don't record game.round, game.players, game.judger if allow_step_back not set
pass
# perfrom action
player = self.players[self.round.current_player]
self.round.proceed_round(player, action)
if (action != 'pass'):
self.judger.calc_playable_cards(player)
if self.judger.judge_game(self.players, self.round.current_player):
self.winner_id = self.round.current_player
next_id = get_downstream_player_id(player, self.players)
self.round.current_player = next_id
# get next state
state = self.get_state(next_id)
self.state = state
return state, next_id
def step_back(self):
''' Return to the previous state of the game
Returns:
(bool): True if the game steps back successfully
'''
if not self.round.trace:
return False
#winner_id will be always None no matter step_back from any case
self.winner_id = None
#reverse round
player_id, cards = self.round.step_back(self.players)
#reverse player
if (cards != 'pass'):
self.players[
player_id].played_cards = self.round.find_last_played_cards_in_trace(
player_id)
self.players[player_id].play_back()
#reverse judger.played_cards if needed
if (cards != 'pass'):
self.judger.restore_playable_cards(player_id)
self.state = self.get_state(self.round.current_player)
return True
def get_state(self, player_id):
''' Return player's state
Args:
player_id (int): player id
Returns:
(dict): The state of the player
'''
player = self.players[player_id]
others_hands = self._get_others_current_hand(player)
if self.is_over():
actions = None
else:
actions = list(
player.available_actions(self.round.greater_player,
self.judger))
state = player.get_state(self.round.public, others_hands, actions)
return state
@staticmethod
def get_action_num():
''' Return the total number of abstract acitons
Returns:
int: the total number of abstract actions of doudizhu
'''
return 131
def get_player_id(self):
''' Return current player's id
Returns:
int: current player's id
'''
return self.round.current_player
def get_player_num(self):
''' Return the number of players in doudizhu
Returns:
int: the number of players in doudizhu
'''
return self.num_players
def is_over(self):
''' Judge whether a game is over
Returns:
Bool: True(over) / False(not over)
'''
if self.winner_id is None:
return False
return True
def _get_others_current_hand(self, player):
player_up = self.players[get_upstream_player_id(player, self.players)]
player_down = self.players[get_downstream_player_id(
player, self.players)]
others_hand = merge(player_up.current_hand,
player_down.current_hand,
key=functools.cmp_to_key(doudizhu_sort_card))
return cards2str(others_hand)