def download_forcing(data_path: Path | str, forcing: str, date_start: datetime | int, date_end: datetime | int) -> None:
    """Download forcing netCDF files for a date range.

    Args:
        data_path: Directory where forcing files are stored.
        forcing: Name of reanalysis product.
        date_start: Start datetime or start year.
        date_end: End datetime or end year.
    """
    # Create data directory if it doesn't exist
    data_path = Path(data_path)
    data_path.mkdir(parents=True, exist_ok=True)

    if isinstance(date_start, int) and isinstance(date_end, int):
        date_start = datetime(date_start, 1, 1)
        date_end = datetime(date_end, 12, 31)

    assert isinstance(date_start, datetime) and isinstance(date_end, datetime)

    reanalysis = REANALYSIS[forcing]
    if reanalysis is REANALYSIS.gridMET:
        _download_gridmet(data_path, reanalysis, date_start.year, date_end.year)
    else:
        _download_xldas(data_path, reanalysis, date_start, date_end)

def find_grids(forcing: str, *, locations: LocationInput=None, screen_output: bool=True, remove_duplicates: bool=True) -> str | list[str]:
    """Find nearest valid weather grids for provided locations.

    Args:
        forcing: Name of reanalysis product.
        locations: Optional coordinates or named coordinates.
        screen_output: If True, print selected grid summary.
        remove_duplicates: If True, collapse duplicate grid selections.

    Returns:
        Weather filename for one location or list of filenames.
    """
    df = _find_grids(REANALYSIS[forcing], locations, screen_output=screen_output, remove_duplicates=remove_duplicates)

    return df['weather_file'].iloc[0] if len(df) == 1 else df['weather_file'].tolist()

def generate_weather_files(data_path: Path | str, weather_path: Path | str, forcing: str, date_start: datetime, date_end: datetime, *,
    hourly: bool=False, locations: LocationInput=None, header: bool=True) -> None:
    """Generate Cycles weather files for selected locations and dates.

    Args:
        data_path: Directory containing downloaded forcing files.
        weather_path: Output directory for generated weather files.
        forcing: Name of reanalysis product.
        date_start: Start date of generated records.
        date_end: End date of generated records.
        hourly: If True, write hourly weather files.
        locations: Optional coordinates or named coordinates.
        header: If True, write weather file headers.
    """
    reanalysis = REANALYSIS[forcing]
    resolution = Resolution.HOURLY if hourly else Resolution.DAILY

    Path(weather_path).mkdir(parents=True, exist_ok=True)

    grid_df = _find_grids(reanalysis, locations, screen_output=False, remove_duplicates=True)

    if reanalysis is REANALYSIS.gridMET:
        weather_data = _process_gridmet(Path(data_path), date_start, date_end, grid_df)
    elif reanalysis in [REANALYSIS.GLDAS, REANALYSIS.NLDAS]:
        weather_data = _process_xldas(Path(data_path), reanalysis, date_start, date_end, grid_df, resolution)

    if resolution is Resolution.HOURLY:
        start = max(date_start, reanalysis.start_time)
        freq = '1h'
    elif resolution is Resolution.DAILY:
        start = date_start
        freq = '1d'

    time_index = pd.date_range(start=start, end=date_end + timedelta(days=1), freq=freq, inclusive='left')

    weather_data = {key: _interpolate_to_hourly(reanalysis, np.array(value)) if hourly and reanalysis.data_interval > 1 else np.array(value) for key, value in weather_data.items()}

    _write_weather_files(Path(weather_path), time_index, weather_data, grid_df, header, resolution)