Filepaths

bnode_core.filepaths

Utilities for constructing, naming, and resolving files and directories used by bnode projects. This module centralizes path conventions and discovery for:

• Auto-recognition of the configuration directory based on the current working directory or CLI overrides. Using little "hint" files, the code can determine if it is running inside a bnode package repository or a standalone project.
• Creation and discovery of data folders for raw data and datasets.
• Canonical naming for raw-data and dataset artifacts derived from the data generation configuration. This includes naming conventions that reflect model name, version, if initial states, parameters, and controls are sampled, sampling strategies itself, and dataset sizes.
• Paths to Hydra runtime output artifacts (models, optimizers, datasets).
• Resolution of MLflow artifact URIs to local filesystem paths via the MLFLOW_ARTIFACTS_DESTINATION environment variable.

config_dir_auto_recognize() -> Path

Auto-recognize and return the project's configuration directory.

The current working directory is inspected for known locations of the configuration directory. CLI flags can be used to override discovery and delegate resolution to Hydra.

Returns:

Type	Description
Path	Path | None: Path to the discovered configuration directory. Returns
Path	None when --help is requested, or when CLI flags indicate that a config
Path	path will be provided externally (so Hydra may handle it).

Raises:

Type	Description
ValueError	If no configuration directory can be found and no CLI flag suggests that the path will be provided manually.

Notes

Search order: 1. If "--help" or "-h" present, print help and return None. 2. If ".bnode_package_repo" exists and "resources/config/" exists, return it. 3. Else if "./config/" exists, return it. 4. Else, if "-cp"/"--config-path" and / or "-cn/ --config-name" are present, return None to allow Hydra to handle the path. 5. Otherwise, log an error and raise ValueError.

Source code in src/bnode_core/filepaths.py

def config_dir_auto_recognize() -> Path:
    """Auto-recognize and return the project's configuration directory.

    The current working directory is inspected for known locations of the configuration directory. 
    CLI flags can be used to override discovery and delegate resolution to Hydra.

    Returns:
        Path | None: Path to the discovered configuration directory. Returns
        None when --help is requested, or when CLI flags indicate that a config
        path will be provided externally (so Hydra may handle it).

    Raises:
        ValueError: If no configuration directory can be found and no CLI flag suggests that the path will be provided manually.

    Notes:
        Search order:
        1. If "--help" or "-h" present, print help and return None.
        2. If ".bnode_package_repo" exists and "resources/config/" exists, return it.
        3. Else if "./config/" exists, return it.
        4. Else, if "-cp"/"--config-path" and / or "-cn/ --config-name" are present,
           return None to allow Hydra to handle the path.
        5. Otherwise, log an error and raise ValueError.
    """
    msg = ''
    return_path = None
    if Path('.bnode_package_repo').exists():
        if Path('resources/config/').exists():
            return_path = Path('resources/config/')
        else:
            msg += 'Even though .bnode_package_repo file exists, no config directory found in resources/config/.\n'
    else:
        if Path('./config/').exists():
            return_path =  Path('./config/')
        else:
            msg += 'No config directory found in the standard ./config/ location.\n'

    if '--help' in sys.argv or '-h' in sys.argv or '--hydra-help' in sys.argv:
        print('The config directory is auto-recognized based on the current working directory.')
        print('You can override this behavior by providing the config path via --config-path or --config-dir CLI arguments.')
        print('To access hydra help, use --hydra-help.')

    if return_path is not None:
        return return_path

    # Check if user provided config path via CLI args
    raise_error = True
    if '-cp' in sys.argv or '--config-path' in sys.argv:
        raise_error = False
    elif '--config-dir' in sys.argv or '-cd' in sys.argv:
        # we assume that the user provided also a config name in this case
        raise_error = False
    if raise_error:
        msg += 'Please ensure you are in a correct working directory or provide the config path manually.'
        logging.error(msg)
        raise ValueError(msg)
    else:
        return Path("")  # dummy return to satisfy function signature

create_path(path: Path, log: bool) -> None

Create a directory if it does not exist.

Parameters:

Name	Type	Description	Default
path	Path	Directory path to create.	required
log	bool	Whether to log the creation/exists message.	required

Returns:

Type	Description
None	None

Source code in src/bnode_core/filepaths.py

def create_path(path: Path, log: bool) -> None:
    """Create a directory if it does not exist.

    Args:
        path (Path): Directory path to create.
        log (bool): Whether to log the creation/exists message.

    Returns:
        None
    """
    if not path.exists():
        path.mkdir(parents=True)
        if log:
            logging.info('Created path: {}'.format(path))
    else:
        if log:
            logging.info('Path already exists: {}'.format(path))

log_overwriting_file(path: Path) -> None

Log whether a file will be written or overwritten.

Parameters:

Name	Type	Description	Default
path	Path	File path for which to log the action.	required

Returns:

Type	Description
None	None

Source code in src/bnode_core/filepaths.py

def log_overwriting_file(path: Path) -> None:
    """Log whether a file will be written or overwritten.

    Args:
        path (Path): File path for which to log the action.

    Returns:
        None
    """
    if not path.exists():
        logging.info('Writing on file: {}'.format(path))
    else:
        logging.info('Overwriting file: {}'.format(path))

raw_data_name(cfg: data_gen_config) -> str

Build the canonical base name for a raw-data artifact.

The name is derived from the model name/version and optional sampling strategy flags present in the configuration.

Behavior

• Always includes model name and version.
• Appends initial states sampling strategy if initial states are included. (E.g., '_s-R')
• Appends parameters sampling strategy if parameters are included. (E.g., '_p-R')
• Appends controls sampling strategy if controls are included. (E.g., '_c-RROCS')
• Appends '_c-Mo' suffix if controls are only for sampling and the actual used controls are extracted from the model.

Parameters:

Name	Type	Description	Default
cfg	data_gen_config	Data generation configuration.	required

Returns:

Name	Type	Description
str	str	Base name for raw-data artifacts.

Source code in src/bnode_core/filepaths.py

def raw_data_name(cfg: data_gen_config) -> str:
    """Build the canonical base name for a raw-data artifact.

    The name is derived from the model name/version and optional sampling
    strategy flags present in the configuration.

    Behavior:
        - Always includes model name and version.
        - Appends initial states sampling strategy if initial states are included. (E.g., '_s-R')
        - Appends parameters sampling strategy if parameters are included. (E.g., '_p-R')
        - Appends controls sampling strategy if controls are included. (E.g., '_c-RROCS')
        - Appends '_c-Mo' suffix if controls are only for sampling and the actual used controls are extracted
        from the model.

    Args:
        cfg (data_gen_config): Data generation configuration.

    Returns:
        str: Base name for raw-data artifacts.
    """
    RawData = cfg.pModel.RawData
    name = RawData.modelName + '_' + RawData.versionName
    if RawData.initial_states_include:
        name = name + '_s-' + RawData.initial_states_sampling_strategy
    if RawData.parameters_include:
        name = name + '_p-' + RawData.parameters_sampling_strategy
    if RawData.controls_include:
        name = name + '_c-' + RawData.controls_sampling_strategy
    if RawData.controls_only_for_sampling_extract_actual_from_model:
        name = name + '_c-Mo'
    return name

dataset_name(cfg: data_gen_config, n_samples: int) -> str

Build the canonical dataset name for a given configuration and size.

Parameters:

Name	Type	Description	Default
cfg	data_gen_config	Data generation configuration.	required
n_samples	int	Number of samples in the dataset.	required

Returns:

Name	Type	Description
str	str	Dataset name including sample count and optional suffix.

Source code in src/bnode_core/filepaths.py

def dataset_name(cfg: data_gen_config, n_samples: int) -> str:
    """Build the canonical dataset name for a given configuration and size.

    Args:
        cfg (data_gen_config): Data generation configuration.
        n_samples (int): Number of samples in the dataset.

    Returns:
        str: Dataset name including sample count and optional suffix.
    """
    name = raw_data_name(cfg) + '__n-' + str(n_samples)
    if cfg.pModel.dataset_prep.dataset_suffix is not None:
        name = name + '_' + cfg.pModel.dataset_prep.dataset_suffix
    return name

dir_data(log: bool = False) -> Path

Resolve the root data directory for the project.

Search order: 1) If ".bnode_package_repo" exists: "./resources/data". 2) Else if "../../.surrogate_test_data_repo" exists: "../../data". 3) Else: "./data".

The directory is created if missing.

Note

In the future, it might be a good idea to switch to a more flexible directory assignment mechanism, e.g., via environment variables or configuration files.

Parameters:

Name	Type	Description	Default
log	bool	Whether to log directory creation/existence.	False

Returns:

Name	Type	Description
Path	Path	Root data directory.

Source code in src/bnode_core/filepaths.py

def dir_data(log: bool = False) -> Path:
    """Resolve the root data directory for the project.

    Search order:
    1) If ".bnode_package_repo" exists: "./resources/data".
    2) Else if "../../.surrogate_test_data_repo" exists: "../../data".
    3) Else: "./data".

    The directory is created if missing.

    Note:
        In the future, it might be a good idea to switch to a more flexible directory assignment mechanism,
        e.g., via environment variables or configuration files.

    Args:
        log (bool): Whether to log directory creation/existence.

    Returns:
        Path: Root data directory.
    """
    if Path('.bnode_package_repo').exists():
        path = Path('./resources/data')
    else:
        path = Path('./data')
    create_path(path, log)
    return path

dir_raw_data(cfg: data_gen_config, log: bool = False) -> Path

Return the directory in which raw data for the config is stored.

The path includes a subdirectory named by raw_data_name(cfg) and is created on demand.

Parameters:

Name	Type	Description	Default
cfg	data_gen_config	Data generation configuration.	required
log	bool	Whether to log directory creation/existence.	False

Returns:

Name	Type	Description
Path	Path	Directory for raw data artifacts.

Source code in src/bnode_core/filepaths.py

def dir_raw_data(cfg: data_gen_config, log: bool = False) -> Path:
    """Return the directory in which raw data for the config is stored.

    The path includes a subdirectory named by ``raw_data_name(cfg)`` and is
    created on demand.

    Args:
        cfg (data_gen_config): Data generation configuration.
        log (bool): Whether to log directory creation/existence.

    Returns:
        Path: Directory for raw data artifacts.
    """
    path = dir_data() / 'raw_data' / raw_data_name(cfg)
    create_path(path, log)
    return path

filepath_raw_data(cfg: data_gen_config) -> Path

Return the path to the raw-data file for a configuration.

If cfg.pModel.RawData.raw_data_from_external_source is True, the path is resolved inside the raw-data directory using the external file name. Otherwise, a default file name of the form <raw_data_name>_raw_data.hdf5 is used.

Parameters:

Name	Type	Description	Default
cfg	data_gen_config	Data generation configuration.	required

Returns:

Name	Type	Description
Path	Path	Path to the raw-data file.

Source code in src/bnode_core/filepaths.py

def filepath_raw_data(cfg: data_gen_config) -> Path:
    """Return the path to the raw-data file for a configuration.

    If ``cfg.pModel.RawData.raw_data_from_external_source`` is True, the path is
    resolved inside the raw-data directory using the external file name. Otherwise,
    a default file name of the form ``<raw_data_name>_raw_data.hdf5`` is used.

    Args:
        cfg (data_gen_config): Data generation configuration.

    Returns:
        Path: Path to the raw-data file.
    """
    if cfg.pModel.RawData.raw_data_from_external_source:
        file = dir_raw_data(cfg) / cfg.pModel.RawData.raw_data_path
    else:
        file = dir_raw_data(cfg) / (raw_data_name(cfg) + '_raw_data.hdf5')
    return file

filepath_raw_data_config(cfg: data_gen_config) -> Path

Return the path to the RawData configuration YAML file.

Parameters:

Name	Type	Description	Default
cfg	data_gen_config	Data generation configuration.	required

Returns:

Name	Type	Description
Path	Path	Path to the YAML configuration stored alongside raw-data.

Source code in src/bnode_core/filepaths.py

def filepath_raw_data_config(cfg: data_gen_config) -> Path:
    """Return the path to the RawData configuration YAML file.

    Args:
        cfg (data_gen_config): Data generation configuration.

    Returns:
        Path: Path to the YAML configuration stored alongside raw-data.
    """
    file = dir_raw_data(cfg) / (raw_data_name(cfg) + '_RawData_config.yaml')
    return file

dir_datasets(log: bool = False) -> Path

Return the root directory for datasets, creating it if missing.

Parameters:

Name	Type	Description	Default
log	bool	Whether to log directory creation/existence.	False

Returns:

Name	Type	Description
Path	Path	Root datasets directory.

Source code in src/bnode_core/filepaths.py

def dir_datasets(log: bool = False) -> Path:
    """Return the root directory for datasets, creating it if missing.

    Args:
        log (bool): Whether to log directory creation/existence.

    Returns:
        Path: Root datasets directory.
    """
    path = dir_data() / 'datasets'
    create_path(path, log)
    return path

dir_specific_dataset_from_cfg(cfg: data_gen_config, n_samples: int, log: bool = False) -> Path

Return the directory for a specific dataset derived from a config.

The directory name is computed via :func:dataset_name and created if missing.

Parameters:

Name	Type	Description	Default
cfg	data_gen_config	Data generation configuration.	required
n_samples	int	Number of samples in the dataset.	required
log	bool	Whether to log directory creation/existence.	False

Returns:

Name	Type	Description
Path	Path	Directory for the specific dataset.

Source code in src/bnode_core/filepaths.py

def dir_specific_dataset_from_cfg(cfg: data_gen_config, n_samples: int, log: bool = False) -> Path:
    """Return the directory for a specific dataset derived from a config.

    The directory name is computed via :func:`dataset_name` and created if missing.

    Args:
        cfg (data_gen_config): Data generation configuration.
        n_samples (int): Number of samples in the dataset.
        log (bool): Whether to log directory creation/existence.

    Returns:
        Path: Directory for the specific dataset.
    """
    path = dir_datasets() / dataset_name(cfg, n_samples)
    create_path(path, log)
    return path

dir_specific_dataset_from_name(name: str, log: bool = False) -> Path

Return the directory for a specific dataset by name.

Parameters:

Name	Type	Description	Default
name	str	Dataset name.	required
log	bool	Unused flag kept for API symmetry; directory is not created here.	False

Returns:

Name	Type	Description
Path	Path	Directory path for the dataset name.

Source code in src/bnode_core/filepaths.py

def dir_specific_dataset_from_name(name: str, log: bool = False) -> Path:
    """Return the directory for a specific dataset by name.

    Args:
        name (str): Dataset name.
        log (bool): Unused flag kept for API symmetry; directory is not created here.

    Returns:
        Path: Directory path for the dataset name.
    """
    path = dir_datasets() / name
    return path

filepath_dataset(cfg: data_gen_config, n_samples: int) -> Path

Return the path to a dataset file for a given config and size.

Parameters:

Name	Type	Description	Default
cfg	data_gen_config	Data generation configuration.	required
n_samples	int	Number of samples in the dataset.	required

Returns:

Name	Type	Description
Path	Path	HDF5 dataset file path.

Source code in src/bnode_core/filepaths.py

def filepath_dataset(cfg: data_gen_config, n_samples: int) -> Path:
    """Return the path to a dataset file for a given config and size.

    Args:
        cfg (data_gen_config): Data generation configuration.
        n_samples (int): Number of samples in the dataset.

    Returns:
        Path: HDF5 dataset file path.
    """
    file = dir_specific_dataset_from_cfg(cfg, n_samples) / (dataset_name(cfg, n_samples) + '_dataset.hdf5')
    return file

filepath_dataset_config(cfg: data_gen_config, n_samples: int) -> Path

Return the path to the pModel configuration YAML for the dataset.

Parameters:

Name	Type	Description	Default
cfg	data_gen_config	Data generation configuration.	required
n_samples	int	Number of samples in the dataset.	required

Returns:

Name	Type	Description
Path	Path	YAML configuration file path stored with the dataset.

Source code in src/bnode_core/filepaths.py

def filepath_dataset_config(cfg: data_gen_config, n_samples: int) -> Path:
    """Return the path to the pModel configuration YAML for the dataset.

    Args:
        cfg (data_gen_config): Data generation configuration.
        n_samples (int): Number of samples in the dataset.

    Returns:
        Path: YAML configuration file path stored with the dataset.
    """
    file = dir_specific_dataset_from_cfg(cfg, n_samples) / (dataset_name(cfg, n_samples) + '_pModel_config.yaml')
    return file

filepath_dataset_from_name(name: str) -> Path

Return the HDF5 dataset file path for a given dataset name. Assumes that the dataset is located in the standard dataset directory structure of this package.

Parameters:

Name	Type	Description	Default
name	str	Dataset name.	required

Returns:

Name	Type	Description
Path	Path	HDF5 dataset file path.

Source code in src/bnode_core/filepaths.py

def filepath_dataset_from_name(name: str) -> Path:
    """Return the HDF5 dataset file path for a given dataset name. Assumes that the dataset is located
    in the standard dataset directory structure of this package.

    Args:
        name (str): Dataset name.

    Returns:
        Path: HDF5 dataset file path.
    """
    return dir_specific_dataset_from_name(name) / (name + '_dataset.hdf5')

filepath_dataset_from_config(dataset_name: str, dataset_path: str) -> Path

Return the HDF5 dataset file path for a given dataset name or explicit path.

If dataset_path is provided, it is used directly. Otherwise, the path is constructed based on the standard dataset directory structure.

Parameters:

Name	Type	Description	Default
dataset_name	str	Dataset name.	required
dataset_path	str | None	Optional explicit dataset file path.	required

Returns:

Name	Type	Description
Path	Path	HDF5 dataset file path.

Source code in src/bnode_core/filepaths.py

def filepath_dataset_from_config(dataset_name: str, dataset_path: str) -> Path:
    """Return the HDF5 dataset file path for a given dataset name or explicit path.

    If ``dataset_path`` is provided, it is used directly. Otherwise, the path is
    constructed based on the standard dataset directory structure.

    Args:
        dataset_name (str): Dataset name.
        dataset_path (str | None): Optional explicit dataset file path.

    Returns:
        Path : HDF5 dataset file path.
    """
    if dataset_path is not None:
        return Path(dataset_path)
    else:
        return filepath_dataset_from_name(dataset_name)

filepath_dataset_config_from_name(name: str) -> Path

Return the pModel configuration YAML path for a dataset name.

Parameters:

Name	Type	Description	Default
name	str	Dataset name.	required

Returns:

Name	Type	Description
Path	Path	YAML configuration file path.

Source code in src/bnode_core/filepaths.py

def filepath_dataset_config_from_name(name: str) -> Path:
    """Return the pModel configuration YAML path for a dataset name.

    Args:
        name (str): Dataset name.

    Returns:
        Path: YAML configuration file path.
    """
    file = dir_specific_dataset_from_name(name) / (name + '_pModel_config.yaml')
    return file

dir_current_hydra_output() -> Path

Return the current Hydra runtime output directory.

Returns:

Name	Type	Description
Path	Path	Path to Hydra's current run output directory.

Source code in src/bnode_core/filepaths.py

def dir_current_hydra_output() -> Path:
    """Return the current Hydra runtime output directory.

    Returns:
        Path: Path to Hydra's current run output directory.
    """
    return Path(hydra.core.hydra_config.HydraConfig.get().runtime.output_dir)

filepath_model_current_hydra_output(phase: int | None = None) -> Path

Return the model checkpoint path in the current Hydra output directory.

Parameters:

Name	Type	Description	Default
phase	int | None	Optional training phase index. When provided, the filename is "model_phase_{phase}.pt"; otherwise "model.pt".	None

Returns:

Name	Type	Description
Path	Path	Model checkpoint file path.

Source code in src/bnode_core/filepaths.py

def filepath_model_current_hydra_output(phase: int | None = None) -> Path:
    """Return the model checkpoint path in the current Hydra output directory.

    Args:
        phase (int | None): Optional training phase index. When provided,
            the filename is "model_phase_{phase}.pt"; otherwise "model.pt".

    Returns:
        Path: Model checkpoint file path.
    """
    if phase is not None:
        return dir_current_hydra_output() / 'model_phase_{}.pt'.format(phase)
    else:
        return dir_current_hydra_output() / 'model.pt'

filepath_pretrained_model_current_hydra_output() -> Path

Return the pretrained model file path in the current Hydra output dir.

Returns:

Name	Type	Description
Path	Path	"model_pretrained.pt" path.

Source code in src/bnode_core/filepaths.py

def filepath_pretrained_model_current_hydra_output() -> Path:
    """Return the pretrained model file path in the current Hydra output dir.

    Returns:
        Path: "model_pretrained.pt" path.
    """
    return dir_current_hydra_output() / 'model_pretrained.pt'

filepath_dataset_current_hydra_output() -> Path

Return the dataset file path in the current Hydra output directory.

Returns:

Name	Type	Description
Path	Path	"dataset.hdf5" path.

Source code in src/bnode_core/filepaths.py

def filepath_dataset_current_hydra_output() -> Path:
    """Return the dataset file path in the current Hydra output directory.

    Returns:
        Path: "dataset.hdf5" path.
    """
    return dir_current_hydra_output() / 'dataset.hdf5'

filepath_optimizer_current_hydra_output(phase: int | None = None) -> Path

Return the optimizer state dict path in the current Hydra output dir.

Parameters:

Name	Type	Description	Default
phase	int | None	Optional training phase index. When provided, the filename is "optimizer_phase_{phase}.pt"; otherwise "optimizer.pt".	None

Returns:

Name	Type	Description
Path	Path	Optimizer checkpoint file path.

Source code in src/bnode_core/filepaths.py

def filepath_optimizer_current_hydra_output(phase: int | None = None) -> Path:
    """Return the optimizer state dict path in the current Hydra output dir.

    Args:
        phase (int | None): Optional training phase index. When provided,
            the filename is "optimizer_phase_{phase}.pt"; otherwise "optimizer.pt".

    Returns:
        Path: Optimizer checkpoint file path.
    """
    if phase is not None:
        return dir_current_hydra_output() / 'optimizer_phase_{}.pt'.format(phase)
    else:
        return dir_current_hydra_output() / 'optimizer.pt'

filepath_from_ml_artifacts_uri(mlflow_uri: str) -> Path

Resolve an MLflow artifacts URI to a local filesystem path.

The base directory is read from the MLFLOW_ARTIFACTS_DESTINATION environment variable. The leading "file://" is stripped from the env var value if present. The "mlflow-artifacts:/" prefix in mlflow_uri is also removed before joining.

Parameters:

Name	Type	Description	Default
mlflow_uri	str	An MLflow artifacts URI (e.g., "mlflow-artifacts:///artifacts/..."), or a relative path component under the artifacts root.	required

Returns:

Name	Type	Description
Path	Path	Resolved local filesystem path.

Raises:

Type	Description
ValueError	If MLFLOW_ARTIFACTS_DESTINATION is not set.

Source code in src/bnode_core/filepaths.py

def filepath_from_ml_artifacts_uri(mlflow_uri: str) -> Path:
    """Resolve an MLflow artifacts URI to a local filesystem path.

    The base directory is read from the ``MLFLOW_ARTIFACTS_DESTINATION``
    environment variable. The leading "file://" is stripped from the env var
    value if present. The "mlflow-artifacts:/" prefix in ``mlflow_uri`` is also
    removed before joining.

    Args:
        mlflow_uri (str): An MLflow artifacts URI (e.g.,
            "mlflow-artifacts:/<experiment>/<run>/artifacts/..."), or a relative
            path component under the artifacts root.

    Returns:
        Path: Resolved local filesystem path.

    Raises:
        ValueError: If ``MLFLOW_ARTIFACTS_DESTINATION`` is not set.
    """
    try:
        # try resolving the path from the environment variable
        _dir = os.environ['MLFLOW_ARTIFACTS_DESTINATION']
        _dir = _dir.replace('file://', '')
    except:
        logging.error('MLFLOW_ARTIFACTS_DESTINATION not set')
        logging.error('please set the environment variable MLFLOW_ARTIFACTS_DESTINATION to the path where the mlflow artifacts are stored')
        logging.error('or provide the full path to the dataset')
        raise ValueError('MLFLOW_ARTIFACTS_DESTINATION not set as environment variable')
    mlflow_uri = _dir + os.sep + mlflow_uri.replace('mlflow-artifacts:/', '')
    mlflow_uri = Path(mlflow_uri)
    return mlflow_uri

filepath_from_local_or_ml_artifacts(mlflow_path: str) -> Path

Return a local Path from either a local path or an MLflow artifacts URI. If the provided path starts with "mlflow-artifacts:", it is resolved via :func:filepath_from_ml_artifacts_uri. Otherwise, it is treated as a local filesystem path.

Parameters:

Name	Type	Description	Default
mlflow_path	str	Local filesystem path or an MLflow artifacts URI beginning with "mlflow-artifacts:".	required

Returns:

Name	Type	Description
Path	Path	Local filesystem path.

Source code in src/bnode_core/filepaths.py

def filepath_from_local_or_ml_artifacts(mlflow_path: str) -> Path:
    """Return a local Path from either a local path or an MLflow artifacts URI.
    If the provided path starts with "mlflow-artifacts:", it is resolved via
    :func:`filepath_from_ml_artifacts_uri`. Otherwise, it is treated as a local
    filesystem path.

    Args:
        mlflow_path (str): Local filesystem path or an MLflow artifacts URI
            beginning with "mlflow-artifacts:".

    Returns:
        Path: Local filesystem path.
    """
    if mlflow_path.startswith('mlflow-artifacts:'):
        _path = filepath_from_ml_artifacts_uri(mlflow_path)
    else:
        _path = mlflow_path
    return Path(_path)