import gradio as gr
import torch
import torch.nn as nn
import torchvision.transforms as transforms
from PIL import Image
from safetensors.torch import load_file
from huggingface_hub import hf_hub_download

# Model architecture
class AdvancedCNN(nn.Module):
    def __init__(self, num_classes=10):
        super(AdvancedCNN, self).__init__()
        self.conv1 = nn.Conv2d(in_channels=3, out_channels=32 * 2, kernel_size=3, padding=1)
        self.conv2 = nn.Conv2d(in_channels=32 * 2, out_channels=64 * 2, kernel_size=3, padding=1)
        self.conv3 = nn.Conv2d(in_channels=64 * 2, out_channels=64 * 2, kernel_size=3, padding=1)
        self.conv4 = nn.Conv2d(in_channels=64 * 2, out_channels=128 * 2, kernel_size=3, padding=1)
        self.conv5 = nn.Conv2d(in_channels=128 * 2, out_channels=128 * 2, kernel_size=3, padding=1)
        self.conv6 = nn.Conv2d(in_channels=128 * 2, out_channels=128 * 2, kernel_size=3, padding=1)
        self.conv7 = nn.Conv2d(in_channels=128 * 2, out_channels=256 * 2, kernel_size=3, padding=1)
        self.conv8 = nn.Conv2d(in_channels=256 * 2, out_channels=256 * 2, kernel_size=3, padding=1)
        self.conv9 = nn.Conv2d(in_channels=256 * 2, out_channels=256 * 2, kernel_size=3, padding=1)
        self.bn1 = nn.BatchNorm2d(32 * 2)
        self.bn2 = nn.BatchNorm2d(128 * 2)
        self.bn3 = nn.BatchNorm2d(256 * 2)
        self.maxpool = nn.MaxPool2d(kernel_size=2, stride=2)
        self.dropout = nn.Dropout2d(0.2)
        self.fc1 = nn.Linear(4096 * 2, 4096 * 2)
        self.fc2 = nn.Linear(4096 * 2, 2048 * 2)
        self.fc3 = nn.Linear(2048 * 2, num_classes)
        self.relu = nn.ReLU()

    def forward(self, x):
        x = self.relu(self.bn1(self.conv1(x)))
        x = self.relu(self.conv2(x))
        x = self.relu(self.conv3(x))
        x = self.maxpool(x)
        x = self.relu(self.bn2(self.conv4(x)))
        x = self.relu(self.conv5(x))
        x = self.relu(self.conv6(x))
        x = self.maxpool(x)
        x = self.dropout(x)
        x = self.relu(self.bn3(self.conv7(x)))
        x = self.relu(self.conv8(x))
        x = self.relu(self.conv9(x))
        x = self.maxpool(x)
        x = self.dropout(x)
        x = torch.flatten(x, start_dim=1)
        x = self.relu(self.fc1(x))
        x = self.relu(self.fc2(x))
        x = self.dropout(x)
        x = self.fc3(x)
        return x

# Load model with try/except
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
try:
    model = AdvancedCNN(num_classes=10).to(device)
    model_path = hf_hub_download(repo_id="chandu1617/CIFAR10-CNN_Model", filename="model.safetensors")
    state_dict = load_file(model_path)
    model.load_state_dict(state_dict)
    model.eval()
    load_status = "Model loaded successfully!"
except Exception as e:
    load_status = f"Model load error: {str(e)}"

# CIFAR-10 classes
classes = ['airplane', 'automobile', 'bird', 'cat', 'deer', 'dog', 'frog', 'horse', 'ship', 'truck']

# Image Preprocessing
def preprocess_image(image):
    transform = transforms.Compose([
        transforms.Resize((32, 32)),
        transforms.ToTensor(),
        transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))
    ])
    image = transform(image).unsqueeze(0)
    return image.to(device)

# Predict function
def predict(image):
    if image is None:
        return "Please upload an image."
    try:
        image = preprocess_image(image)
        with torch.no_grad():
            output = model(image)
            prob = torch.softmax(output, dim=1)
            conf, predicted = torch.max(prob, 1)
        return f"Predicted: {classes[predicted.item()]} --> (Confidence: {conf.item()*100:.2f}%) => Overall Model Acc: 92.59%"
    except Exception as e:
        return f"Prediction error: {str(e)}\\n{load_status}"

# Gradio interface
interface = gr.Interface(
    fn=predict,
    inputs=gr.Image(type="pil", label="Upload an image (32x32 RGB recommended)"),
    outputs=gr.Textbox(label="Prediction"),
    title="🚀 CIFAR-10 CNN Classifier Demo",
    description="Upload a small RGB image to classify it. Model Acc: 92.59% on CIFAR-10 test set!",
    examples=[
        ["https://www.cs.toronto.edu/~kriz/cifar-10-sample/airplane1.png"],
        ["https://www.cs.toronto.edu/~kriz/cifar-10-sample/cat1.png"],
        ["https://www.cs.toronto.edu/~kriz/cifar-10-sample/dog1.png"]
    ]
)

if __name__ == "__main__":
    interface.launch()