from fastapi import FastAPI
from apscheduler.schedulers.background import BackgroundScheduler
import pandas as pd
import requests
import xgboost as xgb
from datetime import datetime, timedelta
import os
import matplotlib.pyplot as plt

app = FastAPI()

# Scheduler
scheduler = BackgroundScheduler()
scheduler.start()

# Load the trained model
model = xgb.XGBRegressor()
model.load_model("electricity_price_model.json")

# Constants
WEATHER_API = "https://api.open-meteo.com/v1/forecast"
ELECTRICITY_PRICE_API = "https://www.elprisetjustnu.se/api/v1/prices"
ENERGY_CHARTS_API = "https://api.energy-charts.info/public_power?"
PREDICTIONS_FILE = "predicted_prices.csv"

# Fetch weather data for the next day
def fetch_weather_data_for_tomorrow():
    tomorrow = (datetime.now() + timedelta(days=1)).strftime('%Y-%m-%d')
    params = {
        "latitude": 59.3293,
        "longitude": 18.0686,
        "daily": "temperature_2m_mean,precipitation_sum,wind_speed_10m_max,wind_direction_10m_dominant",
        "start_date": tomorrow,
        "end_date": tomorrow,
        "timezone": "Europe/Stockholm"
    }
    response = requests.get(WEATHER_API, params=params)
    response.raise_for_status()
    data = response.json()["daily"]
    return pd.DataFrame(data)

# Fetch energy production data for the current day
def fetch_energy_production_data():
    today = datetime.now().strftime('%Y-%m-%d')
    params = {"country": "se", "start": today, "end": today}
    response = requests.get(ENERGY_CHARTS_API, params=params)
    response.raise_for_status()
    data = response.json()

    if "production_types" in data:
        production_data = {
            "unix_seconds": data["unix_seconds"],
            **{ptype["name"]: ptype["data"] for ptype in data["production_types"]}
        }
        energy_df = pd.DataFrame(production_data)
        energy_df = energy_df.rename(columns={"unix_seconds": "time"})
        energy_df["time"] = pd.to_datetime(energy_df["time"], unit="s", errors="coerce").dt.tz_localize(None)
        return energy_df
    else:
        return pd.DataFrame()

# Fetch electricity prices for the current day
def fetch_current_electricity_prices():
    today = datetime.now().strftime('%Y/%m-%d')
    url = f"{ELECTRICITY_PRICE_API}/{today}_SE3.json"
    response = requests.get(url)
    response.raise_for_status()
    data = response.json()

    electricity_df = pd.DataFrame(data)
    electricity_df = electricity_df.rename(columns={"time_start": "time"})
    electricity_df["time"] = pd.to_datetime(electricity_df["time"], errors="coerce").dt.tz_localize(None)
    return electricity_df

# Prepare dataset for prediction
def prepare_prediction_data():
    energy_data = fetch_energy_production_data()
    electricity_data = fetch_current_electricity_prices()
    weather_data = fetch_weather_data_for_tomorrow()

    dataset = pd.merge(energy_data, electricity_data, on="time", how="inner")
    dataset = pd.merge(dataset, weather_data, on="time", how="outer")
    dataset = dataset.dropna()
    return dataset

# Predict electricity prices for the next day
def predict_next_day_price():
    dataset = prepare_prediction_data()
    X = dataset.drop(["SEK_per_kWh", "time"], axis=1, errors="ignore")
    predictions = model.predict(X)
    dataset["predicted_price"] = predictions
    dataset.to_csv(PREDICTIONS_FILE, index=False)
    generate_dashboard(dataset)
    print("Predictions saved to 'predicted_prices.csv'.")

# Generate a dashboard for visualization
def generate_dashboard(data):
    plt.figure(figsize=(10, 6))
    plt.plot(data["time"], data["predicted_price"], label="Predicted Price", linestyle="--")
    if "SEK_per_kWh" in data.columns:
        plt.plot(data["time"], data["SEK_per_kWh"], label="Actual Price")
    plt.xlabel("Time")
    plt.ylabel("Electricity Price (SEK/kWh)")
    plt.title("Electricity Prices: Predicted vs Actual")
    plt.legend()
    plt.grid()
    plt.savefig("dashboard.png")
    plt.close()

# Schedule daily updates
scheduler.add_job(predict_next_day_price, "cron", hour=23, minute=59)

# API: Get predictions
@app.get("/predictions")
def get_predictions():
    if not os.path.exists(PREDICTIONS_FILE):
        return {"error": "Predictions not available"}
    predictions = pd.read_csv(PREDICTIONS_FILE)
    return predictions.to_dict()

# API: Get dashboard
@app.get("/dashboard")
def get_dashboard():
    if not os.path.exists("dashboard.png"):
        return {"error": "Dashboard not available"}
    return {"dashboard_url": "/dashboard.png"}