import numpy as np
import gradio as gr
import pickle
import os

# Try to load model, fallback to dummy if fails
def load_or_create_model():
    try:
        if os.path.exists('diabetes_model.pkl'):
            with open('diabetes_model.pkl', 'rb') as f:
                model = pickle.load(f)
            with open('scaler.pkl', 'rb') as f:
                scaler = pickle.load(f)
            print("✓ Loaded trained model")
            return model, scaler, False
    except:
        print("⚠ Could not load model, using demo mode")
    
    # Create simple rule-based model
    class SimpleModel:
        def predict(self, X):
            # Simple rule: glucose > 140 = diabetic
            return [1 if x[1] > 140 else 0 for x in X]
        def predict_proba(self, X):
            probs = []
            for x in X:
                glucose = x[1]
                prob_diabetic = min(0.95, max(0.05, glucose / 200))
                probs.append([1 - prob_diabetic, prob_diabetic])
            return probs
    
    class SimpleScaler:
        def transform(self, X):
            return X
    
    return SimpleModel(), SimpleScaler(), True

model, scaler, is_demo = load_or_create_model()

def predict(pregnancies, glucose, bp, skin, insulin, bmi, pedigree, age):
    try:
        input_data = np.array([[pregnancies, glucose, bp, skin, insulin, bmi, pedigree, age]])
        
        # Scale
        if hasattr(scaler, 'transform'):
            input_data = scaler.transform(input_data)
        
        # Predict
        prediction = model.predict(input_data)[0]
        
        # Get probability
        if hasattr(model, 'predict_proba'):
            proba = model.predict_proba(input_data)[0]
            confidence = proba[1] if prediction == 1 else proba[0]
        else:
            confidence = 0.5
        
        result = "Diabetic" if prediction == 1 else "Non-Diabetic"
        risk = "High" if confidence > 0.7 else "Medium" if confidence > 0.4 else "Low"
        
        return {
            "result": result,
            "confidence": f"{confidence * 100:.1f}%",
            "risk_level": risk,
            "demo_mode": is_demo,
            "glucose_status": "High" if glucose > 140 else "Normal" if glucose > 70 else "Low"
        }
    except Exception as e:
        return {"error": str(e)}

# Create Gradio interface
iface = gr.Interface(
    fn=predict,
    inputs=[
        gr.Number(label="Pregnancies", value=2),
        gr.Number(label="Glucose (mg/dL)", value=148),
        gr.Number(label="Blood Pressure (mmHg)", value=72),
        gr.Number(label="Skin Thickness (mm)", value=35),
        gr.Number(label="Insulin (mu U/ml)", value=0),
        gr.Number(label="BMI (kg/m²)", value=33.6),
        gr.Number(label="Diabetes Pedigree", value=0.627),
        gr.Number(label="Age (years)", value=50)
    ],
    outputs=gr.JSON(label="Prediction Result"),
    title="Diabetes Risk Predictor",
    description="Enter health metrics to assess diabetes risk",
    examples=[
        [2, 148, 72, 35, 0, 33.6, 0.627, 50],
        [1, 85, 66, 29, 0, 26.6, 0.351, 31],
        [5, 116, 74, 0, 0, 25.6, 0.201, 30]
    ],
    theme="soft"
)

if __name__ == "__main__":
    iface.launch(server_name="0.0.0.0", server_port=7860)