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

class VGGish(nn.Module):
    """
    PyTorch implementation of the VGGish model.
    Adapted from: https://github.com/harritaylor/torch-vggish
    The following modifications were made: (i) correction for the missing ReLU layers, (ii) correction for the
    improperly formatted data when transitioning from NHWC --> NCHW in the fully-connected layers, and (iii)
    correction for flattening in the fully-connected layers.
    """

    def __init__(self, params):
        super(VGGish, self).__init__()
        self.features = nn.Sequential(
            nn.Conv2d(1, 64, 3, stride=1, padding=1),
            nn.ReLU(inplace=True),
            nn.MaxPool2d(2, stride=2),

            nn.Conv2d(64, 128, 3, stride=1, padding=1),
            nn.ReLU(inplace=True),
            nn.MaxPool2d(2, stride=2),

            nn.Conv2d(128, 256, 3, stride=1, padding=1),
            nn.ReLU(inplace=True),
            nn.Conv2d(256, 256, 3, stride=1, padding=1),
            nn.ReLU(inplace=True),
            nn.MaxPool2d(2, stride=2),

            nn.Conv2d(256, 512, 3, stride=1, padding=1),
            nn.ReLU(inplace=True),
            nn.Conv2d(512, 512, 3, stride=1, padding=1),
            nn.ReLU(inplace=True),
            nn.MaxPool2d(2, stride=2)
        )
        self.fc = nn.Sequential(
            nn.Linear(512 * 31 * 4, 4096),
            nn.ReLU(inplace=True),
            nn.Linear(4096, 4096),
            nn.ReLU(inplace=True),
            nn.Linear(4096, 128),
            nn.ReLU(inplace=True)
        )
        self.final_fc = nn.Linear(128, params.num_class, bias=True)

    def forward(self, x):
        x = self.features(x).permute(0, 2, 3, 1).contiguous()
        x = x.view(x.size(0), -1)
        x = self.fc(x)
        x = self.final_fc(x)
        x = torch.sigmoid(x)

        return x