import os
import json
import joblib
import numpy as np
os.environ["CUDA_VISIBLE_DEVICES"] = "-1"
import yfinance as yf
import features
import matplotlib
matplotlib.use("Agg")

def Predict():
    # Define paths (consistent with your previous script)
    SCRIPT_DIR = os.path.dirname(os.path.abspath(__file__))
    DATA_DIR = os.path.join(SCRIPT_DIR, "data")
    MODEL_PATH = os.path.join(DATA_DIR, "model.keras")

    # Pull 1 month of current data to make prediction against | for volatility 20
    df = yf.download("SPY", period="2mo", auto_adjust=True)
    if not df.empty:
        df = features.MakeFeatures(df, 1)
        df = features.CleanDF(df)

    # Drop our predictor
    df.drop('Target_Close', axis=1, inplace=True)

    # Lazy load this so it doesnt interfere with yfinance
    from keras.models import load_model

    # Load the model
    reconstructed_model = load_model(MODEL_PATH)

    # Verify it loaded correctly
    reconstructed_model.summary()

    # Load the scalers
    feature_scaler = joblib.load(os.path.join(DATA_DIR, "feature_scaler.pkl"))
    target_scaler = joblib.load(os.path.join(DATA_DIR, "target_scaler.pkl"))

    # Scale the data
    scaled_data = feature_scaler.transform(df)

    # Predict
    scaled_predictions = reconstructed_model.predict(scaled_data)

    # Use the loaded target scaler to get back to % change
    actual_prediction = target_scaler.inverse_transform(scaled_predictions)

    # 'predictions' will be a 2D array, flatten it if you want a simple list
    flat_predictions = actual_prediction.flatten().tolist()

    print(f"Predicted Target_Close: {flat_predictions}")

    return json.dumps(flat_predictions)

if __name__ == "__main__":
    Predict()