import gradio as gr
import torch
from torch import nn
from torchvision import models, transforms
from PIL import Image
import requests
import base64
from io import BytesIO
import os

# Define the number of classes
num_classes = 2  # Update with the actual number of classes in your dataset

# Load the model (assuming you've already downloaded it)
def load_model():
    try:
        model = models.resnet50(pretrained=False)
        model.fc = nn.Linear(model.fc.in_features, num_classes)
        model.load_state_dict(torch.load("path_to_your_model.pth", map_location=torch.device("cpu")))
        model.eval()
        return model
    except Exception as e:
        print(f"Error loading model: {e}")
        return None

model = load_model()

# Define the transformation for the input image
transform = transforms.Compose([
    transforms.Resize(256),
    transforms.CenterCrop(224),
    transforms.ToTensor(),
    transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225]),
])

# Prediction function
def process_image(data):
    try:
        # Check if the input contains a base64-encoded string
        if isinstance(data, dict):
            if "data" in data:
                # Base64 decoding
                image_data = base64.b64decode(data["data"])
                image = Image.open(BytesIO(image_data))
            elif "url" in data:
                # URL-based image loading
                response = requests.get(data["url"])
                image = Image.open(BytesIO(response.content))
            elif "path" in data:
                # Local path image loading
                image = Image.open(data["path"])
            else:
                return "Invalid input data structure."

        # Validate image
        if not isinstance(image, Image.Image):
            return "Invalid image format received."

        # Apply transformations
        image = transform(image).unsqueeze(0)

        # Prediction
        image = image.to(torch.device("cuda" if torch.cuda.is_available() else "cpu"))
        with torch.no_grad():
            outputs = model(image)
            predicted_class = torch.argmax(outputs, dim=1).item()

        if predicted_class == 0:
            return "The photo you've sent is of fall army worm with problem ID 126."
        elif predicted_class == 1:
            return "The photo you've sent is of a healthy maize image."
        else:
            return "Unexpected class prediction."
    except Exception as e:
        return f"Error processing image: {e}"

# Create the Gradio interface
iface = gr.Interface(
    fn=process_image,
    inputs=gr.JSON(label="Upload an image (URL or Local Path)"),  # Input: JSON to handle URL or path
    outputs=gr.Textbox(label="Prediction Result"),  # Output: Prediction result
    live=True,
    title="Maize Anomaly Detection",
    description="Upload an image of maize to detect anomalies like disease or pest infestation. You can provide local paths, URLs, or base64-encoded images."
)

# Launch the Gradio interface
iface.launch(share=True, show_error=True)