Assets summary is faulty for a BIDS dandiset with raw and derived data

[yarikoptic]

happens on an embargoed dandiset 000874 for which we show

which has one participant (according to metadata):

❯ grep Participant -A1 response_1706639141849.json | grep identifier | uniq -c
     29             "identifier": "SP002"

with following folders for that subject

❯ grep path response_1706639141849.json | awk '/sub-SP002/' | sed -e 's,.*"\(.*sub-SP002/\).*,\1,g' | sort | uniq -c
     40 derivatives/MRI-pipeline/sub-SP002/
      8 derivatives/OCT-pipeline/sub-SP002/
     58 rawdata/sub-SP002/

❯ grep path response_1706639141849.json | awk '/sub-SP002/' | sed -e 's,.*"\(.*sub-SP002/[^/]\+/\).*,\1,g' | sort | uniq -c
     40 derivatives/MRI-pipeline/sub-SP002/anat/
      8 derivatives/OCT-pipeline/sub-SP002/micr/
     56 rawdata/sub-SP002/ses-MRI/
      2 rawdata/sub-SP002/ses-OCT/

and some files do have _sample- entities with 4 different values. But overall not clear if 4 is really want we want there.

[yarikoptic]

ok, troubleshooted to the data bug -- has - instead of _ to separate session for subject id in sub-SP002-ses-MRI, thus making that file to define having subject id SP002-ses-MRI. We really should look into using proper bids-validator here :-/

note to future myself -- underlying function is actually within dandischema. Here is the code snippet to load dumped in swagger result and run summary on it to see what is going etc

❯ cat try_stats.py
import json

resp = json.loads(open('/tmp/response_1706639141849.json').read())

from dandischema.metadata import *

aggregate_assets_summary(_['metadata'] for _ in resp['results'])