[Bug]: Weird representation of tags in TimeIntervals

[gviejo]

What happened?

Hello, I am trying to parse a NWB file with pynapple. I noticed that the tag of a TimeIntervals objects is weirdly formatted.
The file is this one : https://dandiarchive.org/dandiset/000939/0.240528.1542/files?location=sub-A3707&page=1
When opening and getting the tags for the object 'nrem' for example, the tags appears differently depending on how I access it. Also it should probably appears as an array of strings.

First steps

import pynapple as nap
a = nap.load_file("/mnt/home/gviejo/Downloads/sub-A3707_behavior+ecephys.nwb")

If I do
1.

a.nwb.objects[a.key_to_id['nrem']].tags[:]

I get :

array(['0', '1', '2', '3', '4', '5', '6', '7', '8', '9', '1', '0', '1',
       '1', '1', '2', '1', '3', '1', '4', '1', '5', '1', '6', '1', '7',
       '1', '8', '1', '9', '2', '0', '2', '1', '2', '2', '2', '3', '2',
       '4', '2', '5', '2', '6', '2', '7', '2', '8', '2', '9', '3', '0',
       '3', '1', '3', '2', '3', '3', '3', '4', '3', '5', '3', '6', '3',
       '7', '3', '8', '3', '9', '4', '0', '4', '1', '4', '2', '4', '3',
       '4', '4', '4', '5', '4', '6', '4', '7', '4', '8', '4', '9', '5',
       '0', '5', '1', '5', '2', '5', '3', '5', '4', '5', '5', '5', '6',
       '5', '7', '5', '8', '5', '9', '6', '0'], dtype=object)

a.nwb.objects[a.key_to_id['nrem']]['tags'][:]

I get :

[array(['0'], dtype=object),
 array(['1'], dtype=object),
 array(['2'], dtype=object),
 array(['3'], dtype=object),
 array(['4'], dtype=object),
 array(['5'], dtype=object),
 array(['6'], dtype=object),
 array(['7'], dtype=object),
 array(['8'], dtype=object),
 array(['9'], dtype=object),
 array(['1', '0'], dtype=object),
 array(['1', '1'], dtype=object),
 array(['1', '2'], dtype=object),
 array(['1', '3'], dtype=object),
 array(['1', '4'], dtype=object),
 array(['1', '5'], dtype=object),
 array(['1', '6'], dtype=object),
 array(['1', '7'], dtype=object),
 array(['1', '8'], dtype=object),
 array(['1', '9'], dtype=object),
 array(['2', '0'], dtype=object),
 array(['2', '1'], dtype=object),
 array(['2', '2'], dtype=object),
 array(['2', '3'], dtype=object),
 array(['2', '4'], dtype=object),
 array(['2', '5'], dtype=object),
 array(['2', '6'], dtype=object),
 array(['2', '7'], dtype=object),
 array(['2', '8'], dtype=object),
 array(['2', '9'], dtype=object),
 array(['3', '0'], dtype=object),
 array(['3', '1'], dtype=object),
 array(['3', '2'], dtype=object),
 array(['3', '3'], dtype=object),
 array(['3', '4'], dtype=object),
 array(['3', '5'], dtype=object),
 array(['3', '6'], dtype=object),
 array(['3', '7'], dtype=object),
 array(['3', '8'], dtype=object),
 array(['3', '9'], dtype=object),
 array(['4', '0'], dtype=object),
 array(['4', '1'], dtype=object),
 array(['4', '2'], dtype=object),
 array(['4', '3'], dtype=object),
 array(['4', '4'], dtype=object),
 array(['4', '5'], dtype=object),
 array(['4', '6'], dtype=object),
 array(['4', '7'], dtype=object),
 array(['4', '8'], dtype=object),
 array(['4', '9'], dtype=object),
 array(['5', '0'], dtype=object),
 array(['5', '1'], dtype=object),
 array(['5', '2'], dtype=object),
 array(['5', '3'], dtype=object),
 array(['5', '4'], dtype=object),
 array(['5', '5'], dtype=object),
 array(['5', '6'], dtype=object),
 array(['5', '7'], dtype=object),
 array(['5', '8'], dtype=object),
 array(['5', '9'], dtype=object),
 array(['6', '0'], dtype=object)]

The correct output should be :

['0' '1' '2' '3' '4' '5' '6' '7' '8' '9' '10' '11' '12' '13' '14' '15'
 '16' '17' '18' '19' '20' '21' '22' '23' '24' '25' '26' '27' '28' '29'
 '30' '31' '32' '33' '34' '35' '36' '37' '38' '39' '40' '41' '42' '43'
 '44' '45' '46' '47' '48' '49' '50' '51' '52' '53' '54' '55' '56' '57'
 '58' '59' '60']

Steps to Reproduce

import pynapple as nap
import numpy as np

a = nap.load_file("/mnt/home/gviejo/Downloads/sub-A3707_behavior+ecephys.nwb")

a.nwb.objects[a.key_to_id['nrem']]['tags'][:]
a.nwb.objects[a.key_to_id['nrem']].tags[:]

Traceback

No response

Operating System

Linux

Python Executable

Python

Python Version

3.10

Package Versions

PyNWB 2.8.1

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?

[stephprince]

Hi @gviejo, thanks for the issue.

I noticed that the tag of a TimeIntervals objects is weirdly formatted.

Unfortunately, in the example you shared this appears to be a bug with how the data was originally written to the file. It seems the tags were incorrectly parsed into separate strings when the data was written. We can contact the authors of this dandiset to determine where this issue might have originated in PyNWB/MatNWB.

When opening and getting the tags for the object 'nrem' for example, the tags appears differently depending on how I access it.

These differences are due to how ragged arrays are stored, and .tags[:] is accessing the raw VectorData object. This is potentially confusing behavior and ideally .tags[:] and ['tags'][:] should return the same output. We've opened up an issue in hdmf to address this: hdmf-dev/hdmf#1210

I believe the intended behavior would be what is returned currently by ['tags'][:], since this representation will handle any ragged arrays where there might be multiple tags per row in the DynamicTable.

[gviejo]

Thanks. Yes I think it would be nice to update the dandiset if possible.

[stephprince]

I will close this issue for now in favor of the relevant hdmf issue and dandiset issue discussion post. I emailed the corresponding author of the dandiset and they plan to upload an updated version in January.