import gradio as gr
import torch
import numpy as np
from PIL import Image
from sklearn.preprocessing import StandardScaler
from transformers import AutoImageProcessor
import json

# Load configuration
with open('config.json') as f:
    config = json.load(f)

# Expected features
feature_names = config['feature_columns']
class_names = list(config['class_labels'].values())

# Interface Gradio
with gr.Blocks(title="Cervical Cancer Multimodal Classifier") as demo:
    gr.Markdown("""
    # 🔬 Cervical Cancer Multimodal Classifier
    
    This model classifies cervical cell samples using both **histopathological images** 
    and **morphological features**.
    
    Provide both inputs for best results!
    """)
    
    with gr.Row():
        with gr.Column():
            gr.Markdown("### Input Image")
            image_input = gr.Image(label="Histopathological Image", type="pil")
            
        with gr.Column():
            gr.Markdown("### Morphological Features")
            feature_inputs = []
            for fname in feature_names[:10]:
                feature_inputs.append(
                    gr.Number(label=fname, value=0.0)
                )
            gr.Markdown("### More Features")
            feature_inputs2 = []
            for fname in feature_names[10:]:
                feature_inputs2.append(
                    gr.Number(label=fname, value=0.0)
                )
    
    submit_btn = gr.Button("🔍 Classify", variant="primary", size="lg")
    
    with gr.Row():
        output_label = gr.Label(label="Prediction", num_top_classes=7)
        output_plot = gr.Plot(label="Confidence Distribution")
    
    # Callback funtion (simplified)
    def classify(image, *features):
        # Load model from Hub
        # Make prediction
        # Return results
        return "normal_superficiel", None
    
    submit_btn.click(
        classify,
        inputs=[image_input] + feature_inputs + feature_inputs2,
        outputs=[output_label, output_plot]
    )
    
    gr.Examples(
        examples=[
            ["sample1.BMP", 803.5, 27804.125, 0.028, 85.86, 192.52] + [0]*15,
            ["sample2.BMP", 610.1, 18067.8, 0.032, 81.53, 153.43] + [0]*15,
        ],
        inputs=[image_input] + feature_inputs + feature_inputs2,
        label="Example Inputs"
    )

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