import torch
import torch.nn as nn
import torch.nn.functional as F
import sys

sys.path.insert(0, '.')  # nopep8
from ldm.modules.losses_audio.vqperceptual import *
from ldm.modules.discriminator.multi_window_disc import Discriminator

class LPAPSWithDiscriminator(nn.Module):
    def __init__(self, disc_start, logvar_init=0.0, kl_weight=1.0, pixelloss_weight=1.0,
                 time_lengths = [16,32,64], disc_factor=1.0, disc_weight=1.0,
                 perceptual_weight=1.0,  disc_conditional=False,
                 disc_loss="hinge"):

        super().__init__()
        assert disc_loss in ["hinge", "vanilla"]
        self.kl_weight = kl_weight
        self.pixel_weight = pixelloss_weight
        self.perceptual_weight = perceptual_weight
        if self.perceptual_weight > 0:
            self.perceptual_loss = LPAPS().eval()# LPIPS用于日常图像，而LPAPS用于梅尔谱图
        
        # output log variance
        self.logvar = nn.Parameter(torch.ones(size=()) * logvar_init)

        self.discriminator = Discriminator(time_lengths=time_lengths,reduction='stack').apply(weights_init) # h=8,w/(2**disc_num_layers) - 2
        self.discriminator_iter_start = disc_start
        if disc_loss == "hinge":
            self.disc_loss = hinge_d_loss
        elif disc_loss == "vanilla":
            self.disc_loss = vanilla_d_loss
        else:
            raise ValueError(f"Unknown GAN loss '{disc_loss}'.")
        print(f"LPAPSWithDiscriminator running with {disc_loss} loss.")
        self.disc_factor = disc_factor
        self.discriminator_weight = disc_weight
        self.disc_conditional = disc_conditional


    def calculate_adaptive_weight(self, nll_loss, g_loss, last_layer=None):
        if last_layer is not None:
            nll_grads = torch.autograd.grad(nll_loss, last_layer, retain_graph=True)[0]
            g_grads = torch.autograd.grad(g_loss, last_layer, retain_graph=True)[0]
        else:
            nll_grads = torch.autograd.grad(nll_loss, self.last_layer[0], retain_graph=True)[0]
            g_grads = torch.autograd.grad(g_loss, self.last_layer[0], retain_graph=True)[0]

        d_weight = torch.norm(nll_grads) / (torch.norm(g_grads) + 1e-4)
        d_weight = torch.clamp(d_weight, 0.0, 1e4).detach()
        d_weight = d_weight * self.discriminator_weight
        return d_weight

    def forward(self, inputs, reconstructions, posteriors, optimizer_idx,
                global_step, last_layer=None, cond=None, split="train", weights=None):
        if len(inputs.shape) == 3: # (B,melbins,T)
            inputs,reconstructions = inputs.unsqueeze(1),reconstructions.unsqueeze(1)            
        rec_loss = torch.abs(inputs.contiguous() - reconstructions.contiguous())
        if self.perceptual_weight > 0:
            p_loss = self.perceptual_loss(inputs.contiguous(), reconstructions.contiguous())
            rec_loss = rec_loss + self.perceptual_weight * p_loss
        else:
            p_loss = torch.tensor([0.0])

        nll_loss = rec_loss / torch.exp(self.logvar) + self.logvar
        weighted_nll_loss = nll_loss
        if weights is not None:
            weighted_nll_loss = weights*nll_loss
        weighted_nll_loss = torch.sum(weighted_nll_loss) / weighted_nll_loss.shape[0]
        nll_loss = torch.sum(nll_loss) / nll_loss.shape[0]
        kl_loss = posteriors.kl()
        kl_loss = torch.sum(kl_loss) / kl_loss.shape[0]


        inputs,reconstructions = inputs.squeeze(1).transpose(1,2),reconstructions.squeeze(1).transpose(1,2)   # (B,T,melbins)
        # now the GAN part
        if optimizer_idx == 0:
            # generator update
            if cond is None:
                assert not self.disc_conditional
                logits_fake = self.discriminator(reconstructions.contiguous())['y']
            else:
                assert self.disc_conditional
                logits_fake = self.discriminator(torch.cat((reconstructions.contiguous(), cond), dim=1))['y']
            g_loss = -torch.mean(logits_fake) # logits_fake the higher the better

            try:
                d_weight = self.calculate_adaptive_weight(nll_loss, g_loss, last_layer=last_layer)
            except RuntimeError:
                assert not self.training
                d_weight = torch.tensor(0.0)

            disc_factor = adopt_weight(self.disc_factor, global_step, threshold=self.discriminator_iter_start)
            loss = weighted_nll_loss + self.kl_weight * kl_loss + d_weight * disc_factor * g_loss

            log = {"{}/total_loss".format(split): loss.clone().detach().mean(),
                   "{}/logvar".format(split): self.logvar.detach(),
                   "{}/kl_loss".format(split): kl_loss.detach().mean(),
                   "{}/nll_loss".format(split): nll_loss.detach().mean(),
                   "{}/rec_loss".format(split): rec_loss.detach().mean(),
                   "{}/d_weight".format(split): d_weight.detach(),
                   "{}/disc_factor".format(split): torch.tensor(disc_factor),
                   "{}/g_loss".format(split): g_loss.detach().mean(),
                   }
            return loss, log

        if optimizer_idx == 1:
            # second pass for discriminator update
            if cond is None:
                logits_real = self.discriminator(inputs.contiguous().detach())['y']
                logits_fake = self.discriminator(reconstructions.contiguous().detach())['y']
            else:
                logits_real = self.discriminator(torch.cat((inputs.contiguous().detach(), cond), dim=1))['y']
                logits_fake = self.discriminator(torch.cat((reconstructions.contiguous().detach(), cond), dim=1))['y']

            disc_factor = adopt_weight(self.disc_factor, global_step, threshold=self.discriminator_iter_start)
            d_loss = disc_factor * self.disc_loss(logits_real, logits_fake)

            log = {"{}/disc_loss".format(split): d_loss.clone().detach().mean(),
                   "{}/logits_real".format(split): logits_real.detach().mean(),
                   "{}/logits_fake".format(split): logits_fake.detach().mean()
                   }
            return d_loss, log