A Python function to extract specification names and versions from NWB files.

get_nwb_extension_names_versions.py

"""
A Python function to extract specification names and versions from NWB files.

Can be easily modified to extract only extensions. core, hdmf-common, hdmf-experimental are not extensions.

The script uses h5py to open the NWB file instead of PyNWB because it is faster.
"""

import h5py
from packaging import version


def get_nwb_specifications_names_versions(nwb_file_path):
    """
    Extract all specification names and versions from an NWB file.

    Args:
        nwb_file_path (str): The path to the NWB file.

    Returns:
        dict: A dictionary containing the specification names and versions.
    """
    specifications = {}
    with h5py.File(nwb_file_path, 'r') as f:
        if 'specifications' in f:
            specs_group = f['specifications']
            if len(specs_group) > 0:
                for name in specs_group:
                    extension_group = specs_group[name]
                    assert isinstance(extension_group, h5py.Group), f"{name} is not a group"
                    sorted_versions = sorted(extension_group, key=version.parse)
                    assert len(sorted_versions) > 0, f"Extension {name} has no versions"
                    # Use only the latest version
                    latest_version = sorted_versions[-1]
                    specifications[name] = latest_version
            else:
                print("Warning: No extensions found in the specifications group")
        else:
            print("Warning: No specifications group found in the NWB file")
    return specifications



if __name__ == "__main__":
    nwb_file_path = "/Users/rly/Documents/NWB_Data/dandisets/000409/sub-CSHL047/sub-CSHL047_ses-b52182e7-39f6-4914-9717-136db589706e_behavior+ecephys+image.nwb"
    specifications_names_versions = get_nwb_specifications_names_versions(nwb_file_path)

    # Remove core, hdmf-common, and hdmf-experimental from the specifications
    extensions_names_versions = specifications_names_versions.copy()
    extensions_names_versions.pop("core")
    extensions_names_versions.pop("hdmf-common")
    extensions_names_versions.pop("hdmf-experimental")

    print(extensions_names_versions)