import torch
import torch.nn as nn
import torch.optim as optim
import torch.nn.functional as F
import torch.backends.cudnn as cudnn

import torchvision
import torchvision.transforms as transforms

import os
import numpy as np

from models import *

learning_rate = 0.1
epsilon = 0.0314
k = 7
alpha = 0.00784
file_name = 'interpolated_adversarial_training'
mixup_alpha = 1.0

device = 'cuda:2'

transform_train = transforms.Compose([
    transforms.RandomCrop(32, padding=4),
    transforms.RandomHorizontalFlip(),
    transforms.ToTensor(),
])

transform_test = transforms.Compose([
    transforms.ToTensor(),
])

train_dataset = torchvision.datasets.CIFAR10(root='./data', train=True, download=True, transform=transform_train)
train_loader = torch.utils.data.DataLoader(train_dataset, batch_size=128, shuffle=True, num_workers=4)

def mixup_data(x, y):
    lam = np.random.beta(mixup_alpha, mixup_alpha)
    batch_size = x.size()[0]
    index = torch.randperm(batch_size).cuda()
    mixed_x = lam * x + (1 - lam) * x[index, :]
    y_a, y_b = y, y[index]
    return mixed_x, y_a, y_b, lam

def mixup_criterion(criterion, pred, y_a, y_b, lam):
    return lam * criterion(pred, y_a) + (1 - lam) * criterion(pred, y_b)

class LinfPGDAttack(object):
    def __init__(self, model):
        self.model = model

    def perturb(self, x_natural, y):
        x = x_natural.detach()
        x = x + torch.zeros_like(x).uniform_(-epsilon, epsilon)
        for i in range(k):
            x.requires_grad_()
            with torch.enable_grad():
                logits = self.model(x)
                loss = F.cross_entropy(logits, y)
            grad = torch.autograd.grad(loss, [x])[0]
            x = x.detach() + alpha * torch.sign(grad.detach())
            x = torch.min(torch.max(x, x_natural - epsilon), x_natural + epsilon)
            x = torch.clamp(x, 0, 1)
        return x

net = ResNet18()
net = net.to(device)
adversary = LinfPGDAttack(net)
criterion = nn.CrossEntropyLoss()
optimizer = optim.SGD(net.parameters(), lr=learning_rate, momentum=0.9, weight_decay=0.0002)

def train(epoch):
    net.train()
    benign_loss = 0
    adv_loss = 0
    for batch_idx, (inputs, targets) in enumerate(train_loader):
        inputs, targets = inputs.to(device), targets.to(device)
        optimizer.zero_grad()

        mix_input, targets_a, targets_b, lambdas = mixup_data(inputs, targets)
        benign_outputs = net(mix_input)
        loss1 = mixup_criterion(criterion, benign_outputs, targets_a, targets_b, lambdas)

        adv = adversary.perturb(inputs, targets)
        adv_inputs, adv_targets_a, adv_targets_b, adv_lam = mixup_data(adv, targets)
        adv_outputs = net(adv_inputs)
        loss2 = mixup_criterion(criterion, adv_outputs, adv_targets_a, adv_targets_b, adv_lam)

        loss = (loss1 + loss2) / 2
        loss.backward()
        optimizer.step()

for epoch in range(0, 200):
    train(epoch)