#!/usr/bin/python3
#-*-coding:utf-8-*-
#$File: tensor2fen.py
import numpy as np
import pickle
def tensor2fen(tensor_frd, tensor_emy):
'''
tensor_frd, tensor_emy: ndarray [9,10,16]
return fen string
'''
return state2fen(tensor2state(tensor_frd, tensor_emy))
def tensor2state(tensor_frd, tensor_emy):
'''
transform tensor 2 state
tensor_frd, tensor_emy ndarray [9,10,16]
return state ndarray [10,9]
'''
assert tensor_frd.shape == tensor_emy.shape
state = np.zeros((10,9), dtype=np.str)
chessfrdplayer = 'KAABBNNRRCCPPPPP'
chessemyplayer = 'kaabbnnrrccppppp'
for i in range(tensor_frd.shape[0]):
for j in range(tensor_frd.shape[1]):
if ~(tensor_frd[i][j] == 0).all():
layer = np.argmax(tensor_frd[i][j])
state[j][i] = chessfrdplayer[layer]
elif ~(tensor_emy[i][j] == 0).all():
layer = np.argmax(tensor_emy[i][j])
state[j][i] = chessemyplayer[layer]
else:
state[j][i] = ' '
return state
def state2fen(state):
'''
transfer the chessboard to fen string
state: state of the current chessboard
turn: which player to play
round: count of round
return: fen string
'''
fen = ''
[m,n] = state.shape
for i in range(m):
zcnt = 0
for j in range(n):
if state[i][j] != ' ':
if zcnt != 0:
fen += str(zcnt)
zcnt = 0
fen += state[i][j]
else:
zcnt += 1
if zcnt != 0:
fen += str(zcnt)
zcnt = 0
fen += '/'
fen = fen[:-1]
fen += ' w'
fen += ' - - 0 1'
return fen
def visualstate(state):
'''
visual the chessborad
'''
print(" -----------------------------------")
r = 9
for row in state:
print('%d | ' % r, end='')
for i in range(len(row)):
print("%s " % (row[i]), end='')
print('\b\b|\n', end='')
r -= 1
print(" -----------------------------------")
print(" -A---B---C---D---E---F---G---H---I-")
if __name__ == '__main__':
file = '../pred.tensor'
fh = open(file, 'rb')
data, label, pred = pickle.load(fh)
state = tensor2state(data[0][0],data[1][0])
fen = state2fen(state)
#from IPython import embed;embed()
print(state)
visualstate(state)
print(fen)
frdpos = data[0]
emypos = data[1]
fenlist = []
for k in range(frdpos.shape[0]):
fenlist.append(tensor2fen(frdpos[k],emypos[k]))
#print(fenlist)
label = tensor2fen(label[1],label[1])
#print(label)
