# coding: utf-8
#import cupy as cp
import numpy as cp
import numpy as np
from functions import *
from util import im2col, col2im, DW_im2col


class Relu:
    def __init__(self):
        self.mask = None

    def forward(self, x):
        self.mask = (x <= 0)
        out = x.copy()
        out[self.mask] = 0

        return out

    def backward(self, dout):
        dout[self.mask] = 0
        dx = dout

        return dx


class Sigmoid:
    def __init__(self):
        self.out = None

    def forward(self, x):
        out = sigmoid(x)
        self.out = out
        return out

    def backward(self, dout):
        dx = dout * (1.0 - self.out) * self.out

        return dx


class Affine:
    def __init__(self, W):
        self.W =W
#        self.b = b
        
        self.x = None
        self.original_x_shape = None
        # 重み・バイアスパラメータの微分
        self.dW = None
#        self.db = None

    def forward(self, x):
        # テンソル対応
        self.original_x_shape = x.shape
        x = x.reshape(x.shape[0], -1)
        self.x = x

        out = cp.dot(self.x, self.W) #+ self.b

        return out

    def backward(self, dout):
        dx = cp.dot(dout, self.W.T)
        self.dW = cp.dot(self.x.T, dout)
#        self.db = cp.sum(dout, axis=0)
        
        dx = dx.reshape(*self.original_x_shape)  # 入力データの形状に戻す（テンソル対応）
        return dx


class SoftmaxWithLoss:
    def __init__(self):
        self.loss = None
        self.y = None # softmaxの出力
        self.t = None # 教師データ

    def forward(self, x, t):
        self.t = t
        self.y = softmax(x)
        self.loss = cross_entropy_error(self.y, self.t)
        
        return self.loss

    def backward(self, dout=1):
        batch_size = self.t.shape[0]
        if self.t.size == self.y.size: # 教師データがone-hot-vectorの場合
            dx = (self.y - self.t) / batch_size
        else:
            dx = self.y.copy()
            dx[np.arange(batch_size), self.t] -= 1
            dx = dx / batch_size
        
        return dx


class Dropout:
    """
    http://arxiv.org/abs/1207.0580
    """
    def __init__(self, dropout_ratio=0.5):
        self.dropout_ratio = dropout_ratio
        self.mask = None

    def forward(self, x, train_flg=True):
        if train_flg:
            self.mask = np.random.rand(*x.shape) > self.dropout_ratio
            return x * self.mask
        else:
            return x * (1.0 - self.dropout_ratio)

    def backward(self, dout):
        return dout * self.mask


class LightNormalization:
    """
    """
    def __init__(self, momentum=0.9, running_mean=None, running_var=None):
        self.momentum = momentum
        self.input_shape = None # Conv層の場合は4次元、全結合層の場合は2次元

        # テスト時に使用する平均と分散
        self.running_mean = running_mean
        self.running_var = running_var

        # backward時に使用する中間データ
        self.batch_size = None
        self.xc = None
        self.std = None

    def forward(self, x, train_flg=True):
        self.input_shape = x.shape
        if x.ndim == 2:
            N, D = x.shape
            x = x.reshape(N, D, 1, 1)

        x = x.transpose(0, 2, 3, 1)
        out = self.__forward(x, train_flg)
        out = out.transpose(0, 3, 1, 2)

        return out.reshape(*self.input_shape)

    def __forward(self, x, train_flg):
        if self.running_mean is None:
            N, H, W, C = x.shape
            self.running_mean = cp.zeros(C, dtype=np.float32)
            self.running_var = cp.zeros(C, dtype=np.float32)

        if train_flg:
            mu = x.mean(axis=(0, 1, 2))
            xc = x - mu
            var = cp.mean(xc**2, axis=(0, 1, 2), dtype=np.float32)
            std = cp.sqrt(var + 10e-7, dtype=np.float32)
            xn = xc / std

            self.batch_size = x.shape[0]
            self.xc = xc
            self.xn = xn
            self.std = std
            self.running_mean = self.momentum * self.running_mean + (1-self.momentum) * mu
            self.running_var = self.momentum * self.running_var + (1-self.momentum) * var
        else:
            xc = x - self.running_mean
            xn = xc / ((cp.sqrt(self.running_var + 10e-7, dtype=np.float32)))

        out = xn
        return out

    def backward(self, dout):
        if dout.ndim == 2:
            N, D = dout.shape
            dout = dout.reshape(N, D, 1, 1)

        dout = dout.transpose(0, 2, 3, 1)
        dx = self.__backward(dout)
        dx = dx.transpose(0, 3, 1, 2)

        dx = dx.reshape(*self.input_shape)
        return dx

    def __backward(self, dout):
        dxn = dout
        dxc = dxn / self.std
        dstd = -cp.sum((dxn * self.xc) / (self.std * self.std), axis=0)
        dvar = 0.5 * dstd / self.std
        dxc += (2.0 / self.batch_size) * self.xc * dvar
        dmu = cp.sum(dxc, axis=0)
        dx = dxc - dmu / self.batch_size

        return dx

class BatchNormalization:
    """
    http://arxiv.org/abs/1502.03167
    """
    def __init__(self, gamma, beta, momentum=0.9, running_mean=None, running_var=None):
        self.gamma = gamma
        self.beta = beta
        self.momentum = momentum
        self.input_shape = None # Conv層の場合は4次元、全結合層の場合は2次元  

        # テスト時に使用する平均と分散
        self.running_mean = running_mean
        self.running_var = running_var  
        
        # backward時に使用する中間データ
        self.batch_size = None
        self.xc = None
        self.std = None
        self.dgamma = None
        self.dbeta = None

    def forward(self, x, train_flg=True):
        self.input_shape = x.shape
        if x.ndim != 2:
            N, C, H, W = x.shape
            x = x.reshape(N, -1)

        out = self.__forward(x, train_flg)
        
        return out.reshape(*self.input_shape)
            
    def __forward(self, x, train_flg):
        if self.running_mean is None:
            N, D = x.shape
            self.running_mean = cp.zeros(D, dtype=np.float32)
            self.running_var = cp.zeros(D, dtype=np.float32)
                        
        if train_flg:
            mu = x.mean(axis=0)
            xc = x - mu
            var = cp.mean(xc**2, axis=0, dtype=np.float32)
            std = cp.sqrt(var + 10e-7, dtype=np.float32)
            xn = xc / std
            
            self.batch_size = x.shape[0]
            self.xc = xc
            self.xn = xn
            self.std = std
            self.running_mean = self.momentum * self.running_mean + (1-self.momentum) * mu
            self.running_var = self.momentum * self.running_var + (1-self.momentum) * var            
        else:
            xc = x - self.running_mean
            xn = xc / ((cp.sqrt(self.running_var + 10e-7, dtype=np.float32)))
            
        out = self.gamma * xn + self.beta 
        return out

    def backward(self, dout):
        if dout.ndim != 2:
            N, C, H, W = dout.shape
            dout = dout.reshape(N, -1)

        dx = self.__backward(dout)

        dx = dx.reshape(*self.input_shape)
        return dx

    def __backward(self, dout):
        dbeta = dout.sum(axis=0)
        dgamma = cp.sum(self.xn * dout, axis=0)
        dxn = self.gamma * dout
        dxc = dxn / self.std
        dstd = -cp.sum((dxn * self.xc) / (self.std * self.std), axis=0)
        dvar = 0.5 * dstd / self.std
        dxc += (2.0 / self.batch_size) * self.xc * dvar
        dmu = cp.sum(dxc, axis=0)
        dx = dxc - dmu / self.batch_size
        
        self.dgamma = dgamma
        self.dbeta = dbeta
        
        return dx


class Convolution:
    def __init__(self, W, stride=1, pad=0):
        self.W = W
        self.stride = stride
        self.pad = pad
        
        self.x = None   
        self.col = None
        self.col_W = None
        
        self.dW = None

    def forward(self, x):
        FN, C, FH, FW = self.W.shape
        N, C, H, W = x.shape
        out_h = 1 + int((H + 2*self.pad - FH) / self.stride)
        out_w = 1 + int((W + 2*self.pad - FW) / self.stride)

        col = im2col(x, FH, FW, self.stride, self.pad)
        col_W = self.W.reshape(FN, -1).T

        out = cp.dot(col, col_W)
        out = out.reshape(N, out_h, out_w, -1).transpose(0, 3, 1, 2)

        self.x = x
        self.col = col
        self.col_W = col_W

        return out

    def backward(self, dout):
        FN, C, FH, FW = self.W.shape
        dout = dout.transpose(0,2,3,1).reshape(-1, FN)

        self.dW = cp.dot(self.col.T, dout)
        self.dW = self.dW.transpose(1, 0).reshape(FN, C, FH, FW)

        dcol = cp.dot(dout, self.col_W.T)
        dx = col2im(dcol, self.x.shape, FH, FW, self.stride, self.pad)

        return dx


class Pooling:
    def __init__(self, pool_h, pool_w, stride=1, pad=0):
        self.pool_h = pool_h
        self.pool_w = pool_w
        self.stride = stride
        self.pad = pad
        
        self.x = None
        self.arg_max = None

    def forward(self, x):
        N, C, H, W = x.shape
        out_h = int(1 + (H - self.pool_h) / self.stride)
        out_w = int(1 + (W - self.pool_w) / self.stride)

        col = im2col(x, self.pool_h, self.pool_w, self.stride, self.pad)
        col = col.reshape(-1, self.pool_h*self.pool_w)

        arg_max = cp.argmax(col, axis=1)
        out = cp.array(cp.max(col, axis=1), dtype=np.float32)
        out = out.reshape(N, out_h, out_w, C).transpose(0, 3, 1, 2)

        self.x = x
        self.arg_max = arg_max

        return out

    def backward(self, dout):
        dout = dout.transpose(0, 2, 3, 1)
        
        pool_size = self.pool_h * self.pool_w
        dmax = cp.zeros((dout.size, pool_size), dtype=np.float32)
        dmax[cp.arange(self.arg_max.size), self.arg_max.flatten()] = dout.flatten()
        dmax = dmax.reshape(dout.shape + (pool_size,)) 
        
        dcol = dmax.reshape(dmax.shape[0] * dmax.shape[1] * dmax.shape[2], -1)
        dx = col2im(dcol, self.x.shape, self.pool_h, self.pool_w, self.stride, self.pad)
        
        return dx

class DW_Convolution:
    def __init__(self, W, stride=1, pad=0):
        self.W = W
        self.stride = stride
        self.pad = pad

        self.x = None   
        self.col = None
        self.col_W = None
        
        self.dW = None
        self.db = None



    def forward(self, x):
        FN, C, FH, FW = self.W.shape
        N, C, H, W = x.shape
        out_h = 1 + int((H + 2*self.pad - FH) / self.stride)
        out_w = 1 + int((W + 2*self.pad - FW) / self.stride)

        col = DW_im2col(x, FH, FW, self.stride, self.pad)
        col_W = self.W.reshape(FN, -1).T

        outlist = []
        outlist = np.zeros((FN, N*H*W, 1))
        for count in range(FN):
            outlist[count] = np.dot(col[count, :, :], col_W[:, count]).reshape(-1,1)

        out = outlist.transpose(1,0,2)
        out = out.reshape(N, out_h, out_w, -1).transpose(0, 3, 1, 2)

        self.x = x
        self.col = col
        self.col_W = col_W
        return out


    def backward(self, dout):
        FN, C, FH, FW = self.W.shape
        N, XC, H, W = dout.shape
        dout = dout.transpose(0,2,3,1).reshape(-1, FN)


        dW_list = np.zeros((FN, FH*FW))
        dcol_list = np.zeros((N * H * W, FN, FH * FW))
        for count in range(FN):
            dW_list[count] = np.dot(self.col[count].transpose(1,0), dout[:, count])
            dcol_list[:,count,:] = np.dot(dout[:,count].reshape(-1,1), self.col_W.T[count,:].reshape(1,-1))
        self.dW = dW_list
        self.dW = self.dW.reshape(FN, C, FH, FW)


        dcol = dcol_list
        dx = col2im(dcol, self.x.shape, FH, FW, self.stride, self.pad)


        return dx