Integrate updated HED validation library call

[christinerogers]

details to follow here, by me or Kay Robbins @VisLab HED lead

in a nutshell:

currently one file in the javascript file serves as the interface and calls HED Tools (MPM pkg currently)

the idea is to replace this with the HED tools python package at the same time, to include in the new BIDS validator.
Probably straightforward to port over this file used in the javascript validator, for use in python

The HED team is very happy to contribute to this effort, let's coordinate in March for followup.

background:

• hedtags.org
• hedtools.org
• About HED tags in BIDS https://bids-specification.readthedocs.io/en/stable/appendices/hed.html

[VisLab]

The Python hedtools (available on PyPI) provide methods very similar to those used in the HED JavaScript validator (available on npm). We would be happy to adapt if needed. It would be appreciated if you would ping us on any details about the interface when they become available.

[yarikoptic]

@effigies where/how do you see integrations with external validators should be implemented here? (asking for a friend)

[effigies]

#26 has the start of the data structures needed to begin validating beyond filenames. Next call is on Monday at 11am EDT.

It would be good to discuss what the interface is between the validator and specific checks. The validator will walk the file tree and construct context objects (https://bids-validator.readthedocs.io/en/latest/validation-model/context.html). We will need some method to determine whether a check should be run on a given file and then to perform the validation. This could just be a simple function that accepts a context, checks for applicability and returns a (possibly empty) list of issues.

def validateHED(context: Context) -> list[Issue]:
    if not usesHED(context):
        return []

    # Extract relevant context fields
    # Call external validator
    return [wrap(external_issue) for external_issue in ...]

But possibly the context is insufficient, or possibly we need to provide caching file content access to avoid excessive reads. I haven't yet looked into the HED tools to see what they accept and emit.

[yarikoptic]

#26 has the start of the data structures needed to begin validating beyond filenames. Next call is on Monday at 11am EDT.

cool! I will be at NWB Community Hackathon and thought to work on adding NWB validation (#27) at it so would love to join.

We will need some method to determine whether a check should be run on a given file and then to perform the validation

Insofar for NWB it would be sufficient to just base it on extension (.nwb or smth like .nwb.zarr), for Zarr would be just on .zarr and being a folder I guess (is that info part of context or need to query filesystem).