It works! Some feedback and kudos

[yarikoptic]

Just a "feedback" item to state that I have tried on a sample dataset we generated here and it worked nicely!

Some notes from our quick experience which might lead to tunes up at some point (by me or anyone else) to make bids2nda even better ;)

• Thanks @chrisfilo ! The tool is a great help!
• output directory had to be created manually or it fails to create output file. may be should be pre-created if not exists
• we were preparing to "upload" using their web app. Since that one is not sharing "current working directory" with the bids2nda, we had to specify full paths (for input/output directories) so that web app could find the files.
  • if there is a cmdline client for that, it would be cool if bids2nda could invoke it and point to the files to be uploaded, and then relative paths could be used (probably). I would have preferred to upload using relative paths since IIRC those do get recorded in NDA DB
  • issue a warning if relative path to dataset/output is provided describing that most likely upload via web app would fail due to different working path
• in our heudiconv'erted dataset we have __dup-01 files which are the canceled runs. We had to remove them before running bids2nda since it was picking them up as well and (IIRC) getting confused or something. I wonder if there could be an option or .bidsignore support to ignore such files.
• introduce checks for the upload of any .tsv or .json file if it has TODO in them. Our use case is upload of heudiconv produced dataset. We provide templates for all the needed files, including _events.tsv files. Since those are uploaded as well, ideally they should be populated before the upload. Such a check with a warning issued would help to avoid uploading "too raw to share" datasets ;)