#coding:utf-8

import numpy as np
import chainer
from chainer import cuda, Variable, FunctionSet, optimizers
import chainer.functions  as F
import chainer.links as L
import os
import cv2
import logging
import datetime

# 確率的勾配降下法で学習させる際の１回分のバッチサイズ
batchsize = 100

# 学習の繰り返し回数
n_epoch   = 100

# 中間層の数
n_units   = 1000

IMGSIZE = 28

class TegakiNN:

    def __init__(self,epoch,patterns):
        self.epoch = epoch
        _path = 'result_{}'.format(patterns)
        print(_path)
        self.model = chainer.FunctionSet(l1=F.Linear(2352, n_units),
                    l2=F.Linear(n_units, n_units),
                    l3=F.Linear(n_units, 2))
        chainer.serializers.load_npz('{}/model_tegaki_{}'.format(
            _path,self.epoch),self.model)
        self.optimizer = chainer.optimizers.Adam()
        self.optimizer.setup(self.model)

        chainer.serializers.load_npz('{}/optimizer_tegaki_{}'.format(
            _path,self.epoch),self.optimizer)

    def predict(self,image):
        x = image.astype(np.float32)
        x = cv2.resize(x,(IMGSIZE*3,IMGSIZE))
        x = x.reshape(1,1,IMGSIZE*3,IMGSIZE)/255

        h1 = F.relu(self.model.l1(chainer.Variable(x)))
        h2 = F.relu(self.model.l2(h1))
        y = self.model.l3(h2)

        return np.argmax(y.data)