Add regular expression jsons to be used by bids-validator and pybids

[tsalo]

@rwblair has been working on using the new YAML files in the specification to build regular expressions to replace those currently in place in bids-validator. These regular expressions should also be useful for pybids. One idea we briefly talked about recently would be to either move the regex-generation code from bids-validator to here, or to push the generated regex jsons here while keeping the code in the bids-validator repo. In either case, both bids-validator and pybids could pull the jsons from this repository for their respective uses.

I have a couple of questions about this:

1. @yarikoptic, does that fit with your plans for this repository? Basically, the schema itself would remain in bids-specification (at least for now), but several automatically-derived files, like these regex jsons, would live here.
2. @rwblair, would you prefer to have the code here or just the generated files?

[rwblair]

My current conversion code is all written in JS so dumping a node project in here feels dirty. If I ever covert it to python I'd be happy with it living in this workload, especially if this repo exposes different versions of the schemas across specification releases.

[tsalo]

That makes sense. Python is probably the way to go for this repo, since so few BIDS folks (myself included) know JS. Thanks for clarifying!

[yarikoptic]

FWIW, I think minting regexes for legit filenames so they could be re-used by validators etc is a great starting point.
Probably should be JS/Python specific, so they would immediately include the group names for each entity, so they could be used not only for validation but for parsing as well.

[yarikoptic]

we already got jsonschema dump archived as schema.json right in the root of the copy. Anything else to go there?
I am on cross-roads on either to bother with subfolders. I feel like we better do, but then - what should they be, what about

• source/ or yaml/ or bids-specification/ or orig/ or schema/? for original schema from bids-specification
• jsonschema/? for that single ATM schema.json
• what else?

[yarikoptic]

FWIW, taking an executive decision to make it schema/ , schema.json will go to the top level directory there nearby schema/

[yarikoptic]

FWIW, done as of 1f60004 . Check it out.

[yarikoptic]

@tsalo @rwblair should we keep this issue open?

I still would not object any other useful "extract" from the schema to accompany it here but indeed unclear if we have a clear utility/purpose here since regexes on their own would have only very limited utility but could still help to decide for tools on what is "legit" part of a bids dataset.