ApplesAndPigs/eval_table.py at main · Radzim/ApplesAndPigs · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
import itertools
import numpy as np
import pandas as pd


def check_valid(s):
    return not check_winner(str(3 - int(s[0])) + s[1:])

def check_winner(s):
    last_move = str(3 - int(s[0]))
    lines = [[2, 3, 4], [5, 6, 7], [8, 9, 10], [2, 5, 8], [3, 6, 9], [4, 7, 10], [2, 6, 10], [4, 6, 8]]
    for l in lines:
        if s[l[0]] == s[l[1]] == s[l[2]] == last_move:
            return True
    return False

def successors(s, game_mode='basic'):
    s = list(s)
    suc = []
    to_move = s[0]
    if s[2:].count(to_move) < 3:
        for i in range(9):
            if s[2+i] == '0':
                s2 = s.copy()
                s2[2+i] = to_move
                s2[0] = str(3-int(s2[0]))
                suc.append(''.join(s2))
    else:
        for i in range(9):
            if s[2+i] == to_move:
                for j in range(9):
                    if game_mode == 'basic':
                        if s[2+j] == '0':
                            s2 = s.copy()
                            s2[2 + j] = to_move
                            s2[2 + i] = '0'
                            s2[0] = str(3 - int(s2[0]))
                            suc.append(''.join(s2))
                    if game_mode == 'bogart':
                        if s[2 + 4] == s[0] and i != 4:
                            pass
                        elif s[2+j] == '0':
                            s2 = s.copy()
                            s2[2 + j] = to_move
                            s2[2 + i] = '0'
                            s2[0] = str(3 - int(s2[0]))
                            suc.append(''.join(s2))
    return suc

def get_eval_table(game_mode='basic'):
    try:
        return pd.read_pickle('eval_table_'+game_mode+'.pkl')
    except:
        if game_mode == 'basic' or game_mode == 'bogart':
            impossible = np.array(list(itertools.product([0, 1, 2], repeat=9)))
            possible1 = [i for i in impossible if 4 > sum(i == 1) == sum(i == 2)]
            possible2 = [i for i in impossible if 4 > sum(i == 1) == sum(i == 2) + 1] + [i for i in impossible if 3 == sum(i == 1) == sum(i == 2)]
            short = ['1_'+''.join(str(ii) for ii in i) for i in possible1] + ['2_'+''.join(str(ii) for ii in i) for i in possible2]
            valid = [i for i in short if check_valid(i)]
            eval_table = pd.DataFrame(valid, index=valid, columns=[''])
            eval_table['winner'] = eval_table.apply(lambda x: int(x.name[0])*2-3 if check_winner(x.name) else 0, axis=1)
            eval_table['successors'] = eval_table.apply(lambda x: successors(x.name, game_mode=game_mode) if x['winner']==0 else [], axis=1)
            eval_table = eval_table.drop(columns=[''])
            eval_table['eval'] = eval_table['winner']
            eval_table['successors_evals'] = eval_table.apply(lambda x: [eval_table.loc[i]['eval'] for i in x['successors']], axis=1)
            checker = 0
            while len(set(eval_table['eval'].values)) != checker:
                checker = len(set(eval_table['eval'].values))
                eval_table['eval'] = eval_table.apply(lambda x: (max(x['successors_evals'])/2 if x.name[0]=='1' else min(x['successors_evals'])/2) if -1 < x['eval'] < 1 else x['eval'], axis=1)
                eval_table['successors_evals'] = eval_table.apply(lambda x: [eval_table.loc[i]['eval'] for i in x['successors']], axis=1)
                print(set(eval_table['eval'].values))
            eval_table['best_move'] = eval_table.apply(lambda x: (x['successors'][np.argmax(x['successors_evals'])] if x.name[0]=='1' else x['successors'][np.argmin(x['successors_evals'])]) if -1 < x['eval'] < 1 else 'game_over', axis=1)
            eval_table['trickiness'] = eval_table.apply(lambda x: sum([-1/xx for xx in x['successors_evals'] if xx<0]) if x.name[0]=='1' else sum([1/xx for xx in x['successors_evals'] if xx>0]) if -1 < x['eval'] < 1 else x['eval'], axis=1)
            eval_table['successors_trickiness'] = eval_table.apply(lambda x: [eval_table.loc[i]['trickiness'] for i in x['successors']], axis=1)
            eval_table.to_pickle('eval_table_'+game_mode+'.pkl')
            return pd.read_pickle('eval_table_'+game_mode+'.pkl')