import pandas as pd
import numpy as np
import pickle
import os
import xgboost

# Fonction pour charger le modèle directement depuis le fichier local
def load_model():
    try:
        # Charger le modèle depuis le fichier local
        model_path = os.path.join(os.path.dirname(__file__), "xgboots_regressor.pkl")
        
        if os.path.exists(model_path):
            with open(model_path, "rb") as f:
                model = pickle.load(f)
            return model
        else:
            raise FileNotFoundError(f"Le fichier modèle '{model_path}' n'a pas été trouvé.")
    except Exception as e:
        raise ValueError(f"Erreur lors du chargement du modèle: {str(e)}")

def predict_price(model, input_df, with_confidence_interval=False, confidence_level=0.95, error_margin=0.15):
    try:
        # Make prediction
        prediction = model.predict(input_df)
        predicted_price = float(prediction[0])
        
        if with_confidence_interval:
            # Map confidence levels to z-scores
            z_scores = {
                0.50: 0.674,
                0.80: 1.282,
                0.90: 1.645,
                0.95: 1.96,
                0.99: 2.576,
                0.999: 3.291
            }
            
            # Get the appropriate z-score (default to 95% if not found)
            z_score = z_scores.get(confidence_level, 1.96)
            
            # Calculate the margin of error
            margin = predicted_price * error_margin
            
            # Calculate the bounds of the confidence interval
            lower_bound = max(0, predicted_price - (z_score * margin))
            upper_bound = predicted_price + (z_score * margin)
            
            return predicted_price, lower_bound, upper_bound
        else:
            return predicted_price
    except Exception as e:
        raise ValueError(f"Failed to make prediction: {str(e)}")