import uvicorn
import pandas as pd
from pydantic import BaseModel
from fastapi import FastAPI
import numpy as np
import joblib
from typing import List

# ==== FastAPI Description ====

description = """
# 🚗 GetAround Rental Price Predictor API

This API predicts the **rental price per day (in €)** for a car based on various features.
"""

app = FastAPI(
    title="GetAround Price Prediction API",
    description=description,
    version="1.0",
    contact={"name": "Ton Nom"},
)

# ==== Input Data Models ====

class CarCriteria(BaseModel):
    model_key: str
    mileage: int
    engine_power: int
    fuel: str
    paint_color: str
    car_type: str
    private_parking_available: bool
    has_gps: bool
    has_air_conditioning: bool
    automatic_car: bool
    has_getaround_connect: bool
    has_speed_regulator: bool
    winter_tires: bool

class CarOptions(BaseModel):
    car_options: List[CarCriteria]

# ==== Load pipeline (model + preprocessor ensemble) ====

def load_model():
    model = joblib.load("model.joblib")  # Pipeline: preprocessor + LinearRegression
    return model

# ==== Predict endpoint ====

@app.post("/predict", tags=["Machine Learning"])
async def predict(car_options: CarOptions):
    model = load_model()

    # Convertir les données en DataFrame
    df_input = pd.DataFrame([option.dict() for option in car_options.car_options])

    # Prédiction directe (le modèle contient déjà le préprocesseur)
    predictions = model.predict(df_input)

    # Retourner les résultats formatés
    formatted = [f"Option {i+1}: {round(pred)} €" for i, pred in enumerate(predictions)]
    return {"predictions": formatted}

# ==== Exécution locale ====

if __name__ == "__main__":
    uvicorn.run(app, host="0.0.0.0", port=4000)