BIDSlayout.get_sessions() fails on openneuro ds001541

[poldrack]

here is code to reproduce:

import os
from bids import BIDSLayout

os.system('datalad clone [email protected]:OpenNeuroDatasets/ds001541.git')

layout = BIDSLayout('ds001541')
layout.get_sessions()

should raise the following error:

  File "pybids_ds001541.py", line 9, in <module>
    layout.get_sessions()
  File "/Users/poldrack/anaconda3/envs/py38/lib/python3.8/site-packages/bids/layout/layout.py", line 970, in get
    results = list(set([x.entities[target] for x in results]))
TypeError: unhashable type: 'dict'

I wonder if it's related to different numbers of sessions across subjects:

ls ds001541/sub*
ds001541/sub-562:
ses-1	ses-2	ses-3	ses-4	ses-5	ses-6

ds001541/sub-572:
ses-4	ses-5	ses-7	ses-8	ses-9

ds001541/sub-573:
ses-1	ses-2	ses-3

ds001541/sub-584:
ses-1

ds001541/sub-589:
ses-1

ds001541/sub-597:
ses-1	ses-2	ses-4

ds001541/sub-598:
ses-1	ses-2	ses-3	ses-4	ses-5	ses-6	ses-7

ds001541/sub-600:
ses-1

ds001541/sub-603:
ses-1	ses-2	ses-3	ses-4	ses-5	ses-6	ses-7

ds001541/sub-620:
ses-1	ses-2	ses-3	ses-4	ses-5	ses-6	ses-7

ds001541/sub-647:
ses-1

ds001541/sub-652:
ses-1	ses-2

ds001541/sub-662:
ses-1	ses-2	ses-3

ds001541/sub-663:
ses-2

ds001541/sub-666:
ses-1

ds001541/sub-667:
ses-1	ses-2

[effigies]

I think it's probably that there's metadata in sessions.tsv.

[poldrack]

I am not seeing a sessions.tsv file in the tree...

…

On Fri, Jun 18, 2021 at 2:53 PM Chris Markiewicz ***@***.***> wrote: I think it's probably that there's metadata in sessions.tsv. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#747 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAGUVEDZDIOVXBQDRJIDFTLTTO56TANCNFSM466KTKYA> .

-- Russell A. Poldrack Albert Ray Lang Professor of Psychology Associate Director, Stanford Data Science Director, SDS Center for Open and Reproducible Science Building 420 Stanford University Stanford, CA 94305 ***@***.*** http://www.poldracklab.org/

[adelavega]

I'm having the same issue with get_tasks on narratives.

What is happening is the by default pybids has index_metadata=True. What this does is index the .json sidecar meta-data as entities for the corresponding files. This is fine for .json side cars for images, but for .tsv side cars, this doesn't make sense. .tsv sidecars are descriptions of columns, not additional "entities" for the corresponding files.

Since the columns in .tsv files are not protected, then one could use "sessions" or "tasks" and describe those columns in a .json side car with an object: {"session": {"Description": "session name"}}. This object is then entered into the set() operation used in get_sessions.

A quick fix is to set index_metadata=False, but this is a bug that should be fixed.