Allow datasets to specify HED versions or individual files as validation schema

[happy5214]

closes #942

This implements the changes proposed in bids-standard/bids-specification#527 to allow datasets to specify their HED schema version, or a specific schema file in sourcedata/, to run HED validation against. I have tested both ways with a sample dataset.

[sappelhoff]

Thanks @happy5214. I'll leave a full review of this to @rwblair ... and we should also probably wait with merging this until the PR bids-standard/bids-specification#527 is merged.

But I'll leave a few comments:

• You need to update https://github.com/bids-standard/bids-validator/blob/master/bids-validator/validators/json/schemas/dataset_description.json so that it includes HEDVersion
• in [ENH] Add the ability of users to specify an explicit HED.xml schema for validation. bids-specification#527 we are discussing that HEDVersion should be a recommended field. So perhaps it'd be nice to make BIDS-validator warn users if they are using HED tags ... but not supplying an HEDVersion field.
• I think we should have a very basic example dataset with HED tags in https://github.com/bids-standard/bids-examples/ ... it could be as simple as a single subject single session EEG ... with HED tags in the events.json. One idea could be to update https://github.com/bids-standard/bids-examples/tree/master/eeg_matchingpennies which is a very basic example.

edit: tagging @VisLab for attention

[happy5214]

Thanks @happy5214. I'll leave a full review of this to @rwblair ... and we should also probably wait with merging this until the PR bids-standard/bids-specification#527 is merged.

But I'll leave a few comments:

• You need to update https://github.com/bids-standard/bids-validator/blob/master/bids-validator/validators/json/schemas/dataset_description.json so that it includes HEDVersion

• in bids-standard/bids-specification#527 we are discussing that HEDVersion should be a recommended field. So perhaps it'd be nice to make BIDS-validator warn users if they are using HED tags ... but not supplying an HEDVersion field.

• I think we should have a very basic example dataset with HED tags in https://github.com/bids-standard/bids-examples/ ... it could be as simple as a single subject single session EEG ... with HED tags in the events.json. One idea could be to update https://github.com/bids-standard/bids-examples/tree/master/eeg_matchingpennies which is a very basic example.

edit: tagging @VisLab for attention

The issues listed have been addressed (or will be after bids-standard/bids-examples#191 is merged). I left in the local schema code for now, but I can remove it if anyone has a strong opinion on that.

[happy5214]

Is this waiting on the example dataset? That will probably take a while to finish, so perhaps we should merge this first.

[sappelhoff]

It'd be cool to give @rwblair some more time to review this - and I think bids-standard/bids-specification#527 should be merged first once discussion is done over there.

[rwblair]

I'm ok with leaving in the local schema ability, can we add a catch statement on this promise chain? If HEDVersion is not a semantic version and the file doesn't exist it produces an uncaught error. I think generating a new issue here would be good. I figure most end users will have mistyped the semantic version rather than be trying to use their own schema.

I think the catch is also good in case the hed validator ever rejects its promise during build schema for whatever reason: https://github.com/hed-standard/hed-javascript/blob/f9674dfc6a18ec068260559b6d485b6ca61d0298/validators/schema.js#L374

[rwblair]

Sorry I didn't realize that I had left this unsubmitted @happy5214 I'm going to go ahead and fix the merge conflict and merge this, and create a new issue for my concern here.