import gradio as gr
import pandas as pd
import numpy as np
import tensorflow as tf
from tensorflow import keras
from sklearn.preprocessing import StandardScaler
import pickle

# Load model and scaler (replace paths if needed)
model = keras.models.load_model('eligibility_predictor.h5')  # Changed to match your saved model name
with open('scaler.pkl', 'rb') as f:
    scaler = pickle.load(f)

def predict_eligibility(theory_score, practice_score):
    """Predicts eligibility given theory and practice scores."""
    try:
        # Preprocess input - create DataFrame to avoid feature name warning
        input_df = pd.DataFrame({
            'theory': [theory_score],
            'practice': [practice_score]
        })
        scaled_input = scaler.transform(input_df)
        
        # Predict
        proba = model.predict(scaled_input, verbose=0)[0][0]
        prediction = int(proba > 0.5)
        class_name = "Eligible" if prediction else "Not eligible"
        
        # Confidence score
        confidence = proba if prediction else (1 - proba)
        
        # Return 3 separate values (not a dictionary)
        return (
            class_name,  # First output: Prediction label
            f"{confidence * 100:.2f}%",  # Second output: Confidence
            "✅ Yes" if prediction else "❌ No"  # Third output: Eligibility status
        )
        
    except Exception as e:
        # Return error message for all outputs
        error_msg = f"Error: {str(e)}"
        return error_msg, "0%", "❌ Error"

# Gradio Interface
demo = gr.Interface(
    fn=predict_eligibility,
    inputs=[
        gr.Slider(0, 20, value=0, step=1, label="Theory Score"),
        gr.Slider(0, 30, value=0, step=1, label="Practice Score")
    ],
    outputs=[
        gr.Textbox(label="Prediction"),
        gr.Textbox(label="Confidence"),
        gr.Textbox(label="Eligibility Status")
    ],
    title="🎓 Eligibility Predictor",
    description="Enter theory and practice scores to check eligibility.",
    examples=[[85, 90], [45, 60], [70, 75]],
    theme="soft"
)

# Launch
demo.launch(share=True)