datamllab / rlcard / 240

Build # 240

Build Type

push

travis-ci

Committed by Daochen

Commit Message

<a href="https://github.com/datamllab/rlcard/commit/<a class=hub.com/datamllab/rlcard/commit/<a class="double-link" href="https://git"><a class=hub.com/datamllab/rlcard/commit/<a class="double-link" href="https://git"><a class=hub.com/datamllab/rlcard/commit/<a class="double-link" href="https://git"><a class=hub.com/datamllab/rlcard/commit/<a class="double-link" href="https://git"><a class=hub.com/datamllab/rlcard/commit/02a537bd2753a1e5a6ca32c28a2eb3b827fb0024">02a537bd2<a href="https://github.com/datamllab/rlcard/commit/02a537bd2753a1e5a6ca32c28a2eb3b827fb0024">&lt;a href=&quot;https://github.com/datamllab/rlcard/commit/02a537bd2753a1e5a6ca32c28a2eb3b827fb0024">&lt;a href=&quot;https://github.com/datamllab/rlcard/commit/02a537bd2753a1e5a6ca32c28a2eb3b827fb0024&quot;&gt;&amp;lt;a href=&amp;quot;https://github.com/datamllab/rlcard/commit/02a537bd2753a1e5a6ca32c28a2eb3b827fb0024&amp;quot;&amp;gt;&amp;amp;lt;a href=&amp;amp;quot;https://github.com/datamllab/rlcard/commit/02a537bd2753a1e5a6ca32c28a2eb3b827fb0024&amp;amp;quot;&amp;amp;gt;&amp;amp;amp;quot;&amp;amp;amp;gt;Squashed commit of the following:

commit &amp;amp;lt;/a&amp;amp;gt;&amp;amp;lt;a class=&amp;amp;quot;double-link&amp;amp;quot; href=&amp;amp;quot;https://github.com/datamllab/rlcard/commit/&amp;amp;lt;a class=&amp;amp;quot;double-link&amp;amp;quot; href=&amp;amp;quot;https://github.com/datamllab/rlcard/commit/32abdb3705e677688190394f7766e7fcbdb7556a&amp;amp;quot;&amp;amp;gt;32abdb370&amp;amp;lt;/a&amp;amp;gt;&amp;amp;quot;&amp;amp;gt;32abdb370&amp;amp;lt;/a&amp;amp;gt;&amp;amp;lt;a href=&amp;amp;quot;https://github.com/datamllab/rlcard/commit/02a537bd2753a1e5a6ca32c28a2eb3b827fb0024&amp;quot;&amp;gt;
Merge: 18b65b1 d462bf3
Author: songyih &amp;amp;lt;songyih1996@gmail.com&amp;amp;gt;
Date:   Tue Oct 8 18:52:36 2019 -0700

    Merge branch &amp;amp;#39;dev&amp;amp;#39; of https://github.com/datamllab/rlcard into dev

commit &amp;lt;/a&amp;gt;&amp;lt;a class=&amp;quot;double-link&amp;quot; href=&amp;quot;https://github.com/datamllab/rlcard/commit/&amp;lt;a class=&amp;quot;double-link&amp;quot; href=&amp;quot;https://github.com/datamllab/rlcard/commit/d462bf35ba46d982f803bcefdeaf1476c8d96c75&amp;quot;&gt;d462bf35b&lt;/a&gt;&quot;&gt;d462bf35b&lt;/a&gt;&lt;a href=&quot;https://github.com/datamllab/rlcard/commit/02a537bd2753a1e5a6ca32c28a2eb3b827fb0024&quot;&gt;
Author: Kwei-Herng Lai &amp;lt;khlai037@datalab3.engr.tamu.edu&amp;gt;
Date:   Tue Oct 8 18:51:59 2019 -0700

    mahjong (fix table)

commit &lt;/a&gt;&lt;a class=&quot;double-link&quot; href=&quot;https://github.com/datamllab/rlcard/commit/&lt;a class=&quot;double-link&quot; href=&quot;https://github.com/datamllab/rlcard/commit/18b65b18334df5f8442aceff9123374505ddfcb1&quot;&gt;18b65b183&lt;/a&gt;&quot;&gt;18b65b183&lt;/a&gt;&lt;a href=&quot;https://github.com/datamllab/rlcard/commit/02a537bd2753a1e5a6ca32c28a2eb3b827fb0024">
Author: songyih &lt;songyih1996@gmail.com&gt;
Date:   Tue Oct 8 18:50:50 2019 -0700

    setup.py for pip hosting

commit </a><a class="double-link" href="https://github.com/datamllab/rlcard/commit/<a class="double-link" href="https://github.com/datamllab/rlcard/commit/f54c6e3812267646dcbbee1fe427319a107c6153">f54c6e381</a>">f54c6e381</a><a href="https://github.com/datamllab/rlcard/commit/02a537bd2753a1e5a6ca32c28a2eb3b827fb0024">
Author: Kwei-Herng Lai &lt;khlai037@datalab3.engr.tamu.edu&gt;
Date:   Tue Oct 8 18:50:38 2019 -0700

    mahjong (table)

commit </a><a class="double-link" href="https://github.com/datamllab/rlcard/commit/bfd6e149a">bfd6e149a
Author: songyih <songyih1996@gmail.com>
Date:   Tue Oct 8 18:49:38 2019 -0700

    setup.py for pip hosting

commit 
87.5

/rlcard/envs/mahjong.py

import numpy as np

from rlcard.utils.utils import *
from rlcard.envs.env import Env
from rlcard.games.mahjong.game import MahjongGame as Game
from rlcard.games.mahjong.card import MahjongCard as Card
from rlcard.games.mahjong.utils import *

class MahjongEnv(Env):
    ''' Mahjong Environment
    '''

    def __init__(self, allow_step_back=False):
        super().__init__(Game(allow_step_back), allow_step_back)
        self.action_id = card_encoding_dict
        self.de_action_id = {self.action_id[key]: key for key in self.action_id.keys()}
        self.state_shape = [6, 34, 4]

    def extract_state(self, state):
        ''' Encode state

        Args:
            state (dict): dict of original state

        Returns:
            numpy array: 6*5*15 array
                         6 : current hand
                             the union of the other two players' hand
                             the recent three actions
                             the union of all played cards
        '''
        players_pile = state['players_pile']
        hand_rep = encode_cards(state['current_hand'])
        piles_rep = []
        for p in players_pile.keys():
            piles_rep.append(encode_cards(pile2list(players_pile[p])))
        piles_rep = np.array(piles_rep)
        table_rep = encode_cards(state['table'])
        rep = [hand_rep, table_rep]
        rep.extend(piles_rep)
        obs = np.array(rep)

        extrated_state = {'obs': obs, 'legal_actions': self.get_legal_actions()}
        return extrated_state

    def get_payoffs(self):
        ''' Get the payoffs of players. Must be implemented in the child class.

        Returns:
            payoffs (list): a list of payoffs for each player
        '''
        _, player, _ = self.game.judger.judge_game(self.game)
        if player == -1:
            payoffs = [0, 0, 0, 0]
        else:
            payoffs = [-1, -1, -1, -1]
            payoffs[player] = 1
        return payoffs

    def decode_action(self, action_id):
        ''' Action id -> the action in the game. Must be implemented in the child class.

        Args:
            action_id (int): the id of the action

        Returns:
            action (string): the action that will be passed to the game engine.
        '''
        action = self.de_action_id[action_id]
        if action_id < 34:
            candidates = self.game.get_legal_actions(self.game.get_state(self.game.round.current_player))
            for card in candidates:
                if card.get_str() == action:
                    action = card
                    break
        return action

    def get_legal_actions(self):
        ''' Get all legal actions for current state

        Returns:
        if type(legal_actions[0]) == Card:
            print("GET:", [c.get_str() for c in legal_actions])
        else:
            print(legal_actions)
            legal_actions (list): a list of legal actions' id
        '''
        legal_action_id = []
        legal_actions = self.game.get_legal_actions(self.game.get_state(self.game.round.current_player))
        if legal_actions:
            for action in legal_actions:
                if isinstance(action, Card):
                    action = action.get_str()
                action_id = self.action_id[action]
                legal_action_id.append(action_id)
        else:
            print("##########################")
            print("No Legal Actions")
            print(self.game.judger.judge_game(self.game))
            print(self.game.is_over())
            print([len(p.pile) for p in self.game.players])
            #print(self.game.get_state(self.game.round.current_player))
            #exit()
        return legal_action_id

1	import numpy as np	1✔
2
3	from rlcard.utils.utils import *	1✔
4	from rlcard.envs.env import Env	1✔
5	from rlcard.games.mahjong.game import MahjongGame as Game	1✔
6	from rlcard.games.mahjong.card import MahjongCard as Card	1✔
7	from rlcard.games.mahjong.utils import *	1✔
8
9	class MahjongEnv(Env):	1✔
10	''' Mahjong Environment
11	'''
12
13	def __init__(self, allow_step_back=False):	1✔
14	super().__init__(Game(allow_step_back), allow_step_back)	1✔
15	self.action_id = card_encoding_dict	1✔
16	self.de_action_id = {self.action_id[key]: key for key in self.action_id.keys()}	1✔
17	self.state_shape = [6, 34, 4]	1✔
18
19	def extract_state(self, state):	1✔
20	''' Encode state
21
22	Args:
23	state (dict): dict of original state
24
25	Returns:
26	numpy array: 6515 array
27	6 : current hand
28	the union of the other two players' hand
29	the recent three actions
30	the union of all played cards
31	'''
32	players_pile = state['players_pile']	1✔
33	hand_rep = encode_cards(state['current_hand'])	1✔
34	piles_rep = []	1✔
35	for p in players_pile.keys():	1✔
36	piles_rep.append(encode_cards(pile2list(players_pile[p])))	1✔
37	piles_rep = np.array(piles_rep)	1✔
38	table_rep = encode_cards(state['table'])	1✔
39	rep = [hand_rep, table_rep]	1✔
40	rep.extend(piles_rep)	1✔
41	obs = np.array(rep)	1✔
42
43	extrated_state = {'obs': obs, 'legal_actions': self.get_legal_actions()}	1✔
44	return extrated_state	1✔
45
46	def get_payoffs(self):	1✔
47	''' Get the payoffs of players. Must be implemented in the child class.
48
49	Returns:
50	payoffs (list): a list of payoffs for each player
51	'''
52	_, player, _ = self.game.judger.judge_game(self.game)	1✔
53	if player == -1:	1✔
54	payoffs = [0, 0, 0, 0]	1✔
55	else:
56	payoffs = [-1, -1, -1, -1]	×
57	payoffs[player] = 1	×
58	return payoffs	1✔
59
60	def decode_action(self, action_id):	1✔
61	''' Action id -> the action in the game. Must be implemented in the child class.
62
63	Args:
64	action_id (int): the id of the action
65
66	Returns:
67	action (string): the action that will be passed to the game engine.
68	'''
69	action = self.de_action_id[action_id]	1✔
70	if action_id < 34:	1✔
71	candidates = self.game.get_legal_actions(self.game.get_state(self.game.round.current_player))	1✔
72	for card in candidates:	1✔
73	if card.get_str() == action:	1✔
74	action = card	1✔
75	break	1✔
76	return action	1✔
77
78	def get_legal_actions(self):	1✔
79	''' Get all legal actions for current state
80
81	Returns:
82	if type(legal_actions[0]) == Card:
83	print("GET:", [c.get_str() for c in legal_actions])
84	else:
85	print(legal_actions)
86	legal_actions (list): a list of legal actions' id
87	'''
88	legal_action_id = []	1✔
89	legal_actions = self.game.get_legal_actions(self.game.get_state(self.game.round.current_player))	1✔
90	if legal_actions:	1✔
91	for action in legal_actions:	1✔
92	if isinstance(action, Card):	1✔
93	action = action.get_str()	1✔
94	action_id = self.action_id[action]	1✔
95	legal_action_id.append(action_id)	1✔
96	else:
97	print("##########################")	×
98	print("No Legal Actions")	×
99	print(self.game.judger.judge_game(self.game))	×
100	print(self.game.is_over())	×
101	print([len(p.pile) for p in self.game.players])	×
102	#print(self.game.get_state(self.game.round.current_player))
103	#exit()
104	return legal_action_id	1✔