[Bug]: .specloc not saved in consolidated metadata

[stephprince]

What happened?

For context, I was trying to update the ZarrIO.load_namespaces method to accept the ZarrIO.file property as an input. I planned to update these lines below to define f = io.file if a path was not provided:

hdmf-zarr/src/hdmf_zarr/backend.py

Lines 238 to 243 in 8376fe4

	def load_namespaces(cls, namespace_catalog, path, storage_options, namespaces=None):
	'''
	Load cached namespaces from a file.
	'''
	# TODO: how to use storage_options here?
	f = zarr.open(path, mode='r', storage_options=storage_options)

I thought the object returned from zarr.open would be the same as the one obtained from io.file. However, when trying to obtain the cached spec from the io.file object, there is no 'specifications' group in io.file and .specloc is not listed in the attrs. I think this may be because the io.file property can be set using zarr.open_consolidated, and looking at the file it seems .specloc is not saved in the consolidated metadata.

I wasn't sure if this was actually a bug or I may be misunderstanding how the consolidated metadata should work.

Steps to Reproduce

import zarr

from pynwb.testing.mock.ecephys import mock_Device, mock_ElectrodeGroup, mock_ElectrodeTable
from pynwb.testing.mock.file import mock_NWBFile

from hdmf.common import DynamicTableRegion
from hdmf_zarr.nwb import NWBZarrIO

# create nwb file
device = mock_Device()
group = mock_ElectrodeGroup(device=device)
electrodes_table = mock_ElectrodeTable(group=group)
electrodes = DynamicTableRegion(name="electrodes", data=list(range(5)), table=electrodes_table, description="electrodes")
nwbfile = mock_NWBFile(electrode_groups=[group], 
                       electrodes=electrodes_table, 
                       devices=[device])

path = "zarr_testing.nwb.zarr"
with NWBZarrIO(path=path, mode="w") as io:
    io.write(nwbfile)

with NWBZarrIO(path=path, mode="r") as io:
    # group from io.file
    print(f'{io.file.attrs.keys() = }')
    
    # group from zarr.open
    f = zarr.open(path, mode='r')
    print(f'{f.attrs.keys() = }')

    # group from zarr.open_consolidated
    f = zarr.open_consolidated(path, mode='r')
    print(f'{f.attrs.keys() = }')

Traceback

io.file.attrs.keys() = dict_keys(['namespace', 'neurodata_type', 'nwb_version', 'object_id'])

f.attrs.keys() = dict_keys(['.specloc', 'namespace', 'neurodata_type', 'nwb_version', 'object_id'])

f.attrs.keys() = dict_keys(['namespace', 'neurodata_type', 'nwb_version', 'object_id'])

Operating System

macOS

Python Executable

Conda

Python Version

3.11

Package Versions

No response

Code of Conduct

• I agree to follow this project's Code of Conduct
• Have you checked the Contributing document?
• Have you ensured this bug was not already reported?