GoCNN/server.py at master · godmoves/GoCNN · 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
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
# -*- coding: utf-8 -*-
import xlrd
import time

import pandas
import gomill.sgf
import numpy as np

from visualization.go_driver import GoDriver
from board_evaluation.pachi_player import Pachi


# set default board size.
BOARD_SIZE = 9

pachi = Pachi(pachi_path="thirdparty/pachi-pachi-12.10-jowa/pachi")
godriver = GoDriver("data/working/cnn_5layer_64filter", board_size=BOARD_SIZE)


def parse_sgf_file(sgf_content):
    while 'AB' in sgf_content:
        pos_b = sgf_content.find('AB')
        sgf_content = to_sequence(sgf_content, pos_b, ';B')
    while 'AW' in sgf_content:
        pos_w = sgf_content.find('AW')
        sgf_content = to_sequence(sgf_content, pos_w, ';W')
    return sgf_content


def to_sequence(sgf_content, pos, prefix):
    pos += 1
    start_pos = pos + 1
    while pos + 4 < len(sgf_content) and sgf_content[pos + 4] == ']':
        pos += 4
    end_pos = pos + 1
    sgf_string = sgf_content[start_pos: end_pos]
    res = ""
    for i in range(0, len(sgf_string), 4):
        res += prefix + sgf_string[i: i + 4]
    res = sgf_content[: start_pos - 2] + res + sgf_content[end_pos:]
    return res


def board_eval(sgf_content):
    # reset go board
    # skip komi, we will handle this later
    godriver.reset_board()
    sgf_content = parse_sgf_file(sgf_content)
    try:
        sgf = gomill.sgf.Sgf_game.from_string(sgf_content)
    except ValueError:
        print('WARNING: no SGF data found')
        # if this is not a sgf file, we return blank command
        return np.zeros((BOARD_SIZE, BOARD_SIZE))
    sgf_iterator = sgf.main_sequence_iter()
    while True:
        try:
            it = sgf_iterator.next()
            color, move = it.get_move()
            if color is None:
                it = sgf_iterator.next()
                color, move = it.get_move()
        except StopIteration:
            break
        if move is not None:
            godriver.play(color, move)
    # scale the value range to [-1, 1]
    nn_matrix = godriver.evaluate_current_board() * 2 - 1

    # pachi player is used to reinforcement the neural network
    pachi_matrix, score = pachi.get_final_score_matrix(sgf_content)
    if pachi_matrix is None:
        pachi_matrix = nn_matrix
    else:
        assert len(pachi_matrix) == BOARD_SIZE ** 2
        pachi_matrix = np.array(pachi_matrix).reshape(BOARD_SIZE, BOARD_SIZE)

    final_matrix = 0.1 * nn_matrix + 0.9 * pachi_matrix

    # (0, 0) -> A1, (0, 8) ->I1, (8, 8)->I9
    return final_matrix


if __name__ == '__main__':
    # readbook = xlrd.open_workbook("./data/kifu_test/test9x9.xlsx")
    # sheet = readbook.sheet_by_index(0)
    # start_time = time.time()
    # for i in range(sheet.nrows):
    #     sgf_content = sheet.cell(i, 3).value.encode('utf-8')
    #     try:
    #         score = board_eval(sgf_content)
    #         print("sgf id {} score {}".format(i, np.sum(score)))
    #     except Exception as e:
    #         print("sgf id {} exception {}".format(i, e))
    #     if i % 100 == 0:
    #         print("time used: {}s".format(time.time() - start_time))

    sql = pandas.read_csv("./data/kifu_test/test_sql.csv")
    sgfs = sql["Sgf"]
    print("sgf nums {}".format(sgfs.size))
    start_time = time.time()
    for i, f in enumerate(sgfs):
        sgf_content = f.encode('utf-8')
        try:
            score = board_eval(sgf_content)
            print("sgf id {} score {}".format(i, np.sum(score)))
        except Exception as e:
            print("sgf id {} exception {}".format(i, e))
        if i % 100 == 0:
            print("time used: {}s".format(time.time() - start_time))