import pandas as pd
import joblib
import sys
import random

# Define features used in training (must match train_model.py)
FEATURES = [
    "Year",
    "Mileage",
    "brand",
    "new&used",
    "Exterior color",
    "Interior color",
    "Drivetrain",
    "MPG",
    "Fuel type",
    "Transmission",
    "Engine",
    "Convenience",
    "Entertainment",
    "Exterior",
    "Safety",
    "Seating",
    "Accidents or damage",
    "Clean title",
    "1-owner vehicle",
    "Personal use only",
    "Model",
]


def predict_random_row():
    try:
        model = joblib.load("car_price_model.joblib")
    except FileNotFoundError:
        print(
            "Error: Model file 'car_price_model.joblib' not found. Please run train_model.py first."
        )
        return

    print("Loading data to pick a random row...")
    try:
        df = pd.read_csv("New_York_cars.csv")
    except FileNotFoundError:
        print("Error: 'New_York_cars.csv' not found.")
        return

    # Pick a random row
    random_index = random.randint(0, len(df) - 1)
    row = df.iloc[[random_index]]

    print(f"\nSelected Row Index: {random_index}")
    print("-" * 30)

    # Display selected features
    for feature in FEATURES:
        val = row[feature].values[0]
        print(f"{feature}: {val}")

    actual_price = row["money"].values[0]
    print("-" * 30)
    print(f"Actual Price: ${actual_price:,.2f}")

    # Predict
    try:
        # Ensure we only pass the features the model expects
        input_data = row[FEATURES]
        prediction = model.predict(input_data)[0]
        print(f"Predicted Price: ${prediction:,.2f}")

        diff = prediction - actual_price
        percent_diff = (diff / actual_price) * 100
        print(f"Difference: ${diff:,.2f} ({percent_diff:+.2f}%)")

    except Exception as e:
        print(f"Error during prediction: {e}")


def predict_price(year, mileage, brand):
    # NOTE: This manual function is now limited compared to the full model.
    # The model now expects many more features.
    # For now, we will warn the user or try to fill others with defaults/unknowns if possible,
    # but realistically, manual entry of 20+ features is hard.
    # We will just try to predict with what we have and let the model pipeline handle missing cols if it can,
    # or error out.

    print(
        "Warning: The model now uses many features. Manual entry only supports Year, Mileage, Brand."
    )
    print(
        "Other features will be set to default/unknown values, which may affect accuracy."
    )

    try:
        model = joblib.load("car_price_model.joblib")
    except FileNotFoundError:
        print("Error: Model file 'car_price_model.joblib' not found.")
        return

    # Create dataframe with all expected features initialized to NaN or appropriate defaults
    input_data = pd.DataFrame(columns=FEATURES)
    input_data.loc[0] = [None] * len(FEATURES)  # Initialize with None

    input_data["Year"] = year
    input_data["Mileage"] = mileage
    input_data["brand"] = brand

    # Fill others if necessary (the pipeline handles NaNs for some, but let's see)
    # The training pipeline uses SimpleImputer, so NaNs should be handled.

    # Predict
    try:
        prediction = model.predict(input_data)[0]
        print(f"\nEstimated Price for {year} {brand} with {mileage} miles:")
        print(f"${prediction:,.2f}")
    except Exception as e:
        print(f"Error during prediction: {e}")


def main():
    print("--- Car Price Guesser ---")

    if len(sys.argv) > 1 and sys.argv[1] == "--random":
        predict_random_row()
        return

    if len(sys.argv) == 4:
        year = int(sys.argv[1])
        mileage = float(sys.argv[2])
        brand = sys.argv[3]
        predict_price(year, mileage, brand)
    else:
        print("Options:")
        print("1. Random Row Verification: uv run predict.py --random")
        print("2. Manual Entry: uv run predict.py <Year> <Mileage> <Brand>")
        print("\nEntering interactive manual mode...")
        try:
            year = int(input("Year (e.g., 2020): "))
            mileage = float(input("Mileage (e.g., 50000): "))
            brand = input("Brand (e.g., Toyota): ")
            predict_price(year, mileage, brand)
        except ValueError:
            print("Invalid input. Please enter numbers for Year and Mileage.")


if __name__ == "__main__":
    main()