bids extractor - sidecar .json files have no bids info

[yarikoptic]

with current master 0.2.0-11-gb3d8897 (and new reworked metadata handling in revolution 0.9.0-183-g8765949, pybids seems to be 0.8.0) saw that side car files have only datalad_core extractor records:

{
  "datalad_core": {
    "@id": "SHA1-s1835--d1bd1ee3b7131c44ad2190558c428fda02b630b6",
    "contentbytesize": 1835
  },
  "path": "sub-9013/ses-2/func/sub-9013_ses-2_task-sleepiness_bold.json"
}

ideally they should carry bids extractor metadata such as subject/session/task. Didn't look into it, i.e. either it is something on pybids or our side to do to address it

[mih]

Why would you would like to see that? Those are metadata. They describe subject/session/task, but they are not itself part of the data that makes up subject/session/task.

[yarikoptic]

Those are files potentially needed to get "full" set of (meta)data for a specific subject/session. So, e.g. if I am interested "give me files for subject X", I would like to see .json and .tsv and ... whatnot which is relevant to analysis of this subject. If I am interested in particular type of file(s), then extension and whatever other metadata fields could be used to restrict the query. I really see no reason why sidecar files shouldn't be annotated with metadata on where they belong (in terms of subject/session)

[mih]

I dont see how this should be done for a semantic markup, please make a proposal.

That being said, I think we will end up with separate extractors for metadata dumps (like the current bids and nifti1) extractors, and extractors for linked data. What you want is easily done in the former.