def add_bg_model(generic_model: GenericLineModel, vary_slope: bool = False) -> GenericLineModel:
    """Adds a LinearBackgroundModel to a generic lmfit model

    Parameters
    ----------
    generic_model : GenericLineModel
        object to which to add a linear background model
    vary_slope : bool, optional
        allows a varying linear slope rather than just constant value, by default False

    Returns
    -------
    GenericLineModel
        The input model, with background componets added
    """
    # composite_name = generic_model._name
    # bg_prefix = f"{composite_name}_".replace(" ", "_").replace("J=", "").replace("/", "_").replace("*", "").replace(".", "")
    raise NotImplementedError("No LinearBackgroundModel still exists in mass2")

def initialize_HLike_2P_model(
    element: str, conf: str, has_linear_background: bool = False, has_tails: bool = False, vary_amp_ratio: bool = False
) -> GenericLineModel:
    """Initializes H-like 2P models consisting of J=1/2 and J=3/2 lines

    Parameters
    ----------
    element : str
        atomic symbol as str, e.g. 'Ne' or 'Ar'
    conf : str
        nuclear configuration as str, e.g. '2p' or '3p'
    has_linear_background : bool, optional
        include a single linear background on top of the 2 Lorentzians, by default False
    has_tails : bool, optional
        include low energy tail in the model, by default False
    vary_amp_ratio : bool, optional
        allow the ratio of the J=3/2 to J=1/2 states to vary away from 2, by default False

    Returns
    -------
    GenericLineModel
        The new composite line
    """

    # Set up line names and lmfit prefixes
    charge = int(xraydb.atomic_number(element))
    line_name_1_2 = f"{element}{charge} {conf} 2P* J=1/2"
    line_name_3_2 = f"{element}{charge} {conf} 2P* J=3/2"
    prefix_1_2 = f"{line_name_1_2}_".replace(" ", "_").replace("J=", "").replace("/", "_").replace("*", "").replace(".", "")
    prefix_3_2 = f"{line_name_3_2}_".replace(" ", "_").replace("J=", "").replace("/", "_").replace("*", "").replace(".", "")
    # Initialize individual lines and models
    line_1_2 = spectra[line_name_1_2]
    line_3_2 = spectra[line_name_3_2]
    model_1_2 = line_1_2.model(has_linear_background=False, has_tails=has_tails, prefix=prefix_1_2)
    model_3_2 = line_3_2.model(has_linear_background=False, has_tails=has_tails, prefix=prefix_3_2)
    # Initialize composite model and set addition H-like constraints
    composite_name = f"{element}{charge} {conf}"
    composite_model = initialize_hci_composite_model(
        composite_name=composite_name,
        individual_models=[model_1_2, model_3_2],
        has_linear_background=has_linear_background,
        peak_component_name=line_name_3_2,
    )
    amp_ratio_param_name = f"{element}{charge}_{conf}_amp_ratio"
    composite_model.set_param_hint(name=amp_ratio_param_name, value=0.5, min=0.0, vary=vary_amp_ratio)
    composite_model.set_param_hint(f"{prefix_1_2}integral", expr=f"{prefix_3_2}integral * {amp_ratio_param_name}")
    return composite_model

def initialize_HeLike_complex_model(
    element: str, has_linear_background: bool = False, has_tails: bool = False, additional_line_names: list = []
) -> GenericLineModel:
    """Initializes 1s2s,2p He-like complexes for a given element.

    By default, uses only the 1s.2s 3S J=1, 1s.2p 3P* J=1, and 1s.2p 1P* J=1 lines.

    Parameters
    ----------
    element : str
        atomic symbol as str, e.g. 'Ne' or 'Ar'
    has_linear_background : bool, optional
        include a single linear background on top of the Lorentzian models, by default False
    has_tails : bool, optional
        include low energy tail in the model, by default False
    additional_line_names : list, optional
        additional line names to include in model, e.g. low level Li/Be-like features, by default []

    Returns
    -------
    GenericLineModel
        A model of the given HCI complex.
    """

    # Set up line names
    charge = int(xraydb.atomic_number(element) - 1)
    line_name_1s2s_3S = f"{element}{charge} 1s.2s 3S J=1"
    line_name_1s2p_3P = f"{element}{charge} 1s.2p 3P* J=1"
    line_name_1s2p_1P = f"{element}{charge} 1s.2p 1P* J=1"
    line_names = np.hstack([[line_name_1s2s_3S, line_name_1s2p_3P, line_name_1s2p_1P], additional_line_names])
    # Set up lines and models based on line_names
    # individual_lines = [spectra[i_line_name]() for i_line_name in line_names]
    individual_models = [
        initialize_hci_line_model(i_line_name, has_linear_background=False, has_tails=has_tails) for i_line_name in line_names
    ]
    # Set up composite model
    composite_name = f"{element}{charge} 1s2s_2p Complex"
    composite_model = initialize_hci_composite_model(
        composite_name=composite_name,
        individual_models=individual_models,
        has_linear_background=has_linear_background,
        peak_component_name=line_name_1s2p_1P,
    )
    return composite_model

def initialize_hci_composite_model(
    composite_name: str,
    individual_models: list[GenericLineModel],
    has_linear_background: bool = False,
    peak_component_name: str | None = None,
) -> GenericLineModel:
    """Initializes composite lmfit model from the sum of input models

    Parameters
    ----------
    composite_name : str
        name given to composite line model
    individual_models : list[GenericLineModel]
        Models to sum into a composite
    has_linear_background : bool, optional
        include a single linear background on top of group of lorentzians, by default False
    peak_component_name : str | None, optional
        designate a component to be a peak for energy, all expressions are referenced to this component, by default None

    Returns
    -------
    GenericLineModel
        The new composite line
    """

    composite_model: GenericLineModel = np.sum(individual_models)
    composite_model.name = composite_name
    if has_linear_background:
        composite_model = add_bg_model(composite_model)
    # Workaround for energy calibration using composite models, pick 1st GenericLineModel component
    line_model_components = [
        i_comp for i_comp in composite_model.components if isinstance(i_comp, mass2.calibration.line_models.GenericLineModel)
    ]
    if peak_component_name is None:
        peak_component_name = line_model_components[0]._name
    peak_component_index = [i_comp._name for i_comp in line_model_components].index(peak_component_name)
    peak_component = line_model_components[peak_component_index]
    composite_model.peak_prefix = peak_component.prefix
    composite_model.peak_energy = peak_component.spect.peak_energy
    # Set up some constraints relative to peak_component
    num_line_components = len(line_model_components)
    line_component_prefixes = [iComp.prefix for iComp in line_model_components]
    line_component_energies = [iComp.spect.peak_energy for iComp in line_model_components]
    for i in np.arange(num_line_components):
        if i != peak_component_index:
            # Single fwhm across model
            composite_model.set_param_hint(f"{line_component_prefixes[i]}fwhm", expr=f"{composite_model.peak_prefix}fwhm")
            # Single dph_de across model
            composite_model.set_param_hint(f"{line_component_prefixes[i]}dph_de", expr=f"{composite_model.peak_prefix}dph_de")
            # Fixed energy separation based on database values
            separation = line_component_energies[i] - composite_model.peak_energy
            hint = f"({separation} * {composite_model.peak_prefix}dph_de) + {composite_model.peak_prefix}peak_ph"
            composite_model.set_param_hint(f"{line_component_prefixes[i]}peak_ph", expr=hint)
    composite_model.shortname = composite_name
    return composite_model

def initialize_hci_line_model(line_name: str, has_linear_background: bool = False, has_tails: bool = False) -> GenericLineModel:
    """Initializes a single lorentzian HCI line model. Reformats line_name to create a lmfit valid prefix.

    Parameters
    ----------
    line_name : str
        name of line to use in mass2.spectra
    has_linear_background : bool, optional
        include linear background in the model, by default False
    has_tails : bool, optional
        include low-energy tail in the model, by default False

    Returns
    -------
    GenericLineModel
        New HCI line.
    """
    line = spectra[line_name]
    prefix = f"{line_name}_".replace(" ", "_").replace("J=", "").replace("/", "_").replace("*", "").replace(".", "")
    line_model = line.model(has_linear_background=has_linear_background, has_tails=has_tails, prefix=prefix)
    line_model.shortname = line_name
    return line_model

def models(
    has_linear_background: bool = False,
    has_tails: bool = False,
    vary_Hlike_amp_ratio: bool = False,
    additional_Helike_complex_lines=[],
) -> dict:
    """Generates some commonly used HCI line models that can be used for energy calibration, etc.

    Parameters
    ----------
    has_linear_background : bool, optional
        include a single linear background on top of the 2 Lorentzians, by default False
    has_tails : bool, optional
        nclude low-energy tail in the model, by default False
    vary_Hlike_amp_ratio : bool, optional
        allow the ratio of the J=3/2 to J=1/2 H-like states to vary, by default False
    additional_Helike_complex_lines : list, optional
        additional line names to include inHe-like complex model, e.g. low level Li/Be-like features, by default []

    Returns
    -------
    _type_
        Dictionary of mass2.calibration.fluorescence_lines.SpectralLine objects, containing commonly used HCI lines.
    """

    models_dict = {}
    # Make some common H-like 2P* models
    conf_Hlike_2P_dict = {}
    conf_Hlike_2P_dict["N"] = ["3p", "4p", "5p"]
    conf_Hlike_2P_dict["O"] = ["3p", "4p", "5p"]
    conf_Hlike_2P_dict["Ne"] = ["2p", "3p", "4p", "5p"]
    conf_Hlike_2P_dict["Ar"] = ["2p", "3p", "4p", "5p"]
    for i_element in list(conf_Hlike_2P_dict.keys()):
        for i_conf in conf_Hlike_2P_dict[i_element]:
            Hlike_model = initialize_HLike_2P_model(
                i_element,
                i_conf,
                has_linear_background=has_linear_background,
                has_tails=has_tails,
                vary_amp_ratio=vary_Hlike_amp_ratio,
            )
            models_dict[Hlike_model._name] = Hlike_model

    # Make some common He-like 1s2s,2p complex and higher order 1p* models
    # He-like lines
    Helike_complex_elements = ["N", "O", "Ne", "Ar"]
    for i_element in Helike_complex_elements:
        Helike_model = initialize_HeLike_complex_model(
            i_element,
            has_linear_background=has_linear_background,
            has_tails=has_tails,
            additional_line_names=additional_Helike_complex_lines,
        )
        models_dict[Helike_model._name] = Helike_model
    # 1s.np 1P* lines for n>=3
    conf_Helike_1P_dict = {}
    conf_Helike_1P_dict["N"] = ["1s.4p", "1s.5p"]
    conf_Helike_1P_dict["O"] = ["1s.4p", "1s.5p"]
    conf_Helike_1P_dict["Ne"] = ["1s.3p", "1s.4p", "1s.5p"]
    conf_Helike_1P_dict["Ar"] = ["1s.3p", "1s.4p", "1s.5p"]
    for i_element in list(conf_Helike_1P_dict.keys()):
        i_charge = int(xraydb.atomic_number(i_element) - 1)
        for i_conf in conf_Helike_1P_dict[i_element]:
            Helike_line_name = f"{i_element}{i_charge} {i_conf} 1P* J=1"
            Helike_model = initialize_hci_line_model(
                Helike_line_name, has_linear_background=has_linear_background, has_tails=has_tails
            )
            models_dict[Helike_model._name] = Helike_model

    # Some more complicated cases
    # 500 eV region of H-/He-like N
    N6_1s3p_model = initialize_hci_line_model("N6 1s.3p 1P* J=1", has_linear_background=False, has_tails=has_tails)
    N7_2p_model = initialize_HLike_2P_model(
        "N", "2p", has_linear_background=False, has_tails=has_tails, vary_amp_ratio=vary_Hlike_amp_ratio
    )
    N_500eV_model = initialize_hci_composite_model(
        "N 500eV Region",
        [N6_1s3p_model, N7_2p_model],
        has_linear_background=has_linear_background,
        peak_component_name="N7 2p 2P* J=3/2",
    )
    models_dict[N_500eV_model._name] = N_500eV_model
    # 660 eV region of H-/He-like O
    O8_2p_model = initialize_HLike_2P_model(
        "O", "2p", has_linear_background=False, has_tails=has_tails, vary_amp_ratio=vary_Hlike_amp_ratio
    )
    O7_1s3p_model = initialize_hci_line_model("O7 1s.3p 1P* J=1", has_linear_background=False, has_tails=has_tails)
    O_660eV_model = initialize_hci_composite_model(
        "O 660eV Region",
        [O8_2p_model, O7_1s3p_model],
        has_linear_background=has_linear_background,
        peak_component_name="O8 2p 2P* J=3/2",
    )
    models_dict[O_660eV_model._name] = O_660eV_model

    return models_dict