It's probably easier for the average CuBIDS user with BIDS noncompliant data to write a PyBIDS JSON config file than a set of BIDS schema files. However, PyBIDS is going to swap out its JSON-based config files for the BIDS schema files at some point, though I don't know when that'll actually happen.

@mattcieslak BIDS makes the schema available as a single JSON file (https://bids-specification.readthedocs.io/en/stable/schema.json). I wonder if we could use that here.

Hi all, after running cubids apply I ran into the same issue as above where only some bids keys were kept in the filename instead of all keys present. For example, this sub-01_ses-02_acq-TSE_ce-gad_rec-NORM_run-001_T1w.nii.gz was renamed to sub-01_ses-02_acq-TSEVARIANTEchoTime_rec-NORM_run-001_T1w.nii.gz where the ce key is dropped but the acq and rec keys are retained.

• CuBIDS version: 1.1.1.dev7+g3f35046
• Datalad version: Not using datalad

I was expecting any key found in the bids schema to pass through the apply step since these keys are retained in the RenameKeyGroup column of the summary.tsv file. The scan from the example is classified as acquisition-TSEVARIANTEchoTime_ceagent-gad_datatype-anat_reconstruction-NORM_run-001_suffix_T1w in the RenameKeyGroup column.

Thanks @dabrielz! Agreed 100%.

Trying to come up with a plan for this. What about the following?

1. Download a copy of the schema JSON to cubids/data/.
   • We'll need a way to update this regularly. That's the part that worries me.
2. Add schema argument to CuBIDS command-line interfaces so users can provide their own version.
   • One drawback is that users will almost certainly have a hard time creating custom schema JSONs.
3. Use the schema JSON to determine what entities are allowed and in what order they should appear.
   • The entities are in schema['objects']['entities'].
   • The order of entities is in schema['rules']['entities'].