"""NAFNet-SIDD-w32 denoiser -> RRDBNet(23) 4x SR cascade (CascadeModel).
Requires rrdbnet.py alongside."""
import torch
import torch.nn as nn
import torch.nn.functional as F

from rrdbnet import RRDBNet


class LayerNormFunction(torch.autograd.Function):

    @staticmethod
    def forward(ctx, x, weight, bias, eps):
        ctx.eps = eps
        N, C, H, W = x.size()
        mu = x.mean(1, keepdim=True)
        var = (x - mu).pow(2).mean(1, keepdim=True)
        y = (x - mu) / (var + eps).sqrt()
        ctx.save_for_backward(y, var, weight)
        y = weight.view(1, C, 1, 1) * y + bias.view(1, C, 1, 1)
        return y

    @staticmethod
    def backward(ctx, grad_output):
        eps = ctx.eps

        N, C, H, W = grad_output.size()
        y, var, weight = ctx.saved_variables
        g = grad_output * weight.view(1, C, 1, 1)
        mean_g = g.mean(dim=1, keepdim=True)

        mean_gy = (g * y).mean(dim=1, keepdim=True)
        gx = 1. / torch.sqrt(var + eps) * (g - y * mean_gy - mean_g)
        return gx, (grad_output * y).sum(dim=3).sum(dim=2).sum(dim=0), grad_output.sum(dim=3).sum(dim=2).sum(
            dim=0), None

class LayerNorm2d(nn.Module):

    def __init__(self, channels, eps=1e-6):
        super(LayerNorm2d, self).__init__()
        self.register_parameter('weight', nn.Parameter(torch.ones(channels)))
        self.register_parameter('bias', nn.Parameter(torch.zeros(channels)))
        self.eps = eps

    def forward(self, x):
        return LayerNormFunction.apply(x, self.weight, self.bias, self.eps)



class SimpleGate(nn.Module):
    def forward(self, x):
        x1, x2 = x.chunk(2, dim=1)
        return x1 * x2

class NAFBlock(nn.Module):
    def __init__(self, c, DW_Expand=2, FFN_Expand=2, drop_out_rate=0.):
        super().__init__()
        dw_channel = c * DW_Expand
        self.conv1 = nn.Conv2d(in_channels=c, out_channels=dw_channel, kernel_size=1, padding=0, stride=1, groups=1, bias=True)
        self.conv2 = nn.Conv2d(in_channels=dw_channel, out_channels=dw_channel, kernel_size=3, padding=1, stride=1, groups=dw_channel,
                               bias=True)
        self.conv3 = nn.Conv2d(in_channels=dw_channel // 2, out_channels=c, kernel_size=1, padding=0, stride=1, groups=1, bias=True)
        
        # Simplified Channel Attention
        self.sca = nn.Sequential(
            nn.AdaptiveAvgPool2d(1),
            nn.Conv2d(in_channels=dw_channel // 2, out_channels=dw_channel // 2, kernel_size=1, padding=0, stride=1,
                      groups=1, bias=True),
        )

        # SimpleGate
        self.sg = SimpleGate()

        ffn_channel = FFN_Expand * c
        self.conv4 = nn.Conv2d(in_channels=c, out_channels=ffn_channel, kernel_size=1, padding=0, stride=1, groups=1, bias=True)
        self.conv5 = nn.Conv2d(in_channels=ffn_channel // 2, out_channels=c, kernel_size=1, padding=0, stride=1, groups=1, bias=True)

        self.norm1 = LayerNorm2d(c)
        self.norm2 = LayerNorm2d(c)

        self.dropout1 = nn.Dropout(drop_out_rate) if drop_out_rate > 0. else nn.Identity()
        self.dropout2 = nn.Dropout(drop_out_rate) if drop_out_rate > 0. else nn.Identity()

        self.beta = nn.Parameter(torch.zeros((1, c, 1, 1)), requires_grad=True)
        self.gamma = nn.Parameter(torch.zeros((1, c, 1, 1)), requires_grad=True)

    def forward(self, inp):
        x = inp

        x = self.norm1(x)

        x = self.conv1(x)
        x = self.conv2(x)
        x = self.sg(x)
        x = x * self.sca(x)
        x = self.conv3(x)

        x = self.dropout1(x)

        y = inp + x * self.beta

        x = self.conv4(self.norm2(y))
        x = self.sg(x)
        x = self.conv5(x)

        x = self.dropout2(x)

        return y + x * self.gamma


class NAFNet(nn.Module):

    def __init__(self, img_channel=3, width=16, middle_blk_num=1, enc_blk_nums=[], dec_blk_nums=[]):
        super().__init__()

        self.intro = nn.Conv2d(in_channels=img_channel, out_channels=width, kernel_size=3, padding=1, stride=1, groups=1,
                              bias=True)
        self.ending = nn.Conv2d(in_channels=width, out_channels=img_channel, kernel_size=3, padding=1, stride=1, groups=1,
                              bias=True)

        self.encoders = nn.ModuleList()
        self.decoders = nn.ModuleList()
        self.middle_blks = nn.ModuleList()
        self.ups = nn.ModuleList()
        self.downs = nn.ModuleList()

        chan = width
        for num in enc_blk_nums:
            self.encoders.append(
                nn.Sequential(
                    *[NAFBlock(chan) for _ in range(num)]
                )
            )
            self.downs.append(
                nn.Conv2d(chan, 2*chan, 2, 2)
            )
            chan = chan * 2

        self.middle_blks = \
            nn.Sequential(
                *[NAFBlock(chan) for _ in range(middle_blk_num)]
            )

        for num in dec_blk_nums:
            self.ups.append(
                nn.Sequential(
                    nn.Conv2d(chan, chan * 2, 1, bias=False),
                    nn.PixelShuffle(2)
                )
            )
            chan = chan // 2
            self.decoders.append(
                nn.Sequential(
                    *[NAFBlock(chan) for _ in range(num)]
                )
            )

        self.padder_size = 2 ** len(self.encoders)

    def forward(self, inp):
        B, C, H, W = inp.shape
        inp = self.check_image_size(inp)

        x = self.intro(inp)

        encs = []

        for encoder, down in zip(self.encoders, self.downs):
            x = encoder(x)
            encs.append(x)
            x = down(x)

        x = self.middle_blks(x)

        for decoder, up, enc_skip in zip(self.decoders, self.ups, encs[::-1]):
            x = up(x)
            x = x + enc_skip
            x = decoder(x)

        x = self.ending(x)
        x = x + inp

        return x[:, :, :H, :W]

    def check_image_size(self, x):
        _, _, h, w = x.size()
        mod_pad_h = (self.padder_size - h % self.padder_size) % self.padder_size
        mod_pad_w = (self.padder_size - w % self.padder_size) % self.padder_size
        x = F.pad(x, (0, mod_pad_w, 0, mod_pad_h))
        return x



class ResidualDenseBlock(nn.Module):
    def __init__(self, num_features: int, growth_channels: int):
        super().__init__()
        self.conv1 = nn.Conv2d(num_features, growth_channels, 3, 1, 1)
        self.conv2 = nn.Conv2d(num_features + growth_channels, growth_channels, 3, 1, 1)
        self.conv3 = nn.Conv2d(num_features + 2 * growth_channels, growth_channels, 3, 1, 1)
        self.conv4 = nn.Conv2d(num_features + 3 * growth_channels, growth_channels, 3, 1, 1)
        self.conv5 = nn.Conv2d(num_features + 4 * growth_channels, num_features, 3, 1, 1)
        self.lrelu = nn.LeakyReLU(0.2, inplace=True)

    def forward(self, x):
        f1 = self.lrelu(self.conv1(x))
        f2 = self.lrelu(self.conv2(torch.cat((x, f1), 1)))
        f3 = self.lrelu(self.conv3(torch.cat((x, f1, f2), 1)))
        f4 = self.lrelu(self.conv4(torch.cat((x, f1, f2, f3), 1)))
        f5 = self.conv5(torch.cat((x, f1, f2, f3, f4), 1))
        return x + 0.2 * f5


class RRDB(nn.Module):
    def __init__(self, num_features: int, growth_channels: int):
        super().__init__()
        # attribute names match the official Real-ESRGAN state_dict (rdb1/2/3)
        self.rdb1 = ResidualDenseBlock(num_features, growth_channels)
        self.rdb2 = ResidualDenseBlock(num_features, growth_channels)
        self.rdb3 = ResidualDenseBlock(num_features, growth_channels)

    def forward(self, x):
        out = self.rdb3(self.rdb2(self.rdb1(x)))
        return x + 0.2 * out


class RRDBNet(nn.Module):
    def __init__(self, num_features=64, num_blocks=16, growth_channels=32, scale=4):
        super().__init__()
        assert scale == 4, "This head is built for 4x"
        self.conv_first = nn.Conv2d(3, num_features, 3, 1, 1)
        self.body = nn.Sequential(*[RRDB(num_features, growth_channels) for _ in range(num_blocks)])
        self.conv_body = nn.Conv2d(num_features, num_features, 3, 1, 1)
        self.conv_up1 = nn.Conv2d(num_features, num_features, 3, 1, 1)
        self.conv_up2 = nn.Conv2d(num_features, num_features, 3, 1, 1)
        self.conv_hr = nn.Conv2d(num_features, num_features, 3, 1, 1)
        self.conv_last = nn.Conv2d(num_features, 3, 3, 1, 1)
        self.lrelu = nn.LeakyReLU(0.2, inplace=True)

    def forward(self, x):
        shallow = self.conv_first(x)
        deep = self.conv_body(self.body(shallow))
        features = shallow + deep
        features = self.lrelu(self.conv_up1(F.interpolate(features, scale_factor=2, mode="nearest")))
        features = self.lrelu(self.conv_up2(F.interpolate(features, scale_factor=2, mode="nearest")))
        return self.conv_last(self.lrelu(self.conv_hr(features)))




class CascadeModel(nn.Module):
    """Pretrained NAFNet-SIDD denoises the LR image; pretrained RealESRNet upscales 4x.
    Fine-tuned jointly end-to-end."""

    def __init__(self):
        super().__init__()
        self.denoiser = NAFNet(
            img_channel=3, width=32,
            enc_blk_nums=[2, 2, 4, 8], middle_blk_num=12, dec_blk_nums=[2, 2, 2, 2],
        )
        self.sr = RRDBNet(num_features=64, num_blocks=23, growth_channels=32, scale=4)

    def forward(self, x):
        # NAFNet's custom LayerNorm overflows fp16 and crawls in T4-emulated bf16 -> fp32.
        # RRDBNet is fp16-safe (proven in the RealESRNet fine-tune run) -> inherits outer autocast.
        with torch.amp.autocast("cuda", enabled=False):
            denoised = self.denoiser(x.float())
        return self.sr(denoised)