dandi ls on bids .zarr directory name : 'NoneType' object has no attribute 'items'

[yarikoptic]

(git-annex)lena:~/proj/dandi/dandisets/000108[draft]git
$> DANDI_CACHE=ignore dandi ls sub-MITU01/ses-20210521h17m17s06/micr/sub-MITU01_ses-20210521h17m17s06_sample-178_stain-NN_run-1_chunk-1_SPIM.ome.zarr
- errors: 1
  path: sub-MITU01/ses-20210521h17m17s06/micr/sub-MITU01_ses-20210521h17m17s06_sample-178_stain-NN_run-1_chunk-1_SPIM.ome.zarr
2022-06-30 16:39:13,717 [ WARNING] Failed to operate on some paths (empty records were listed):
 AttributeError: 1 paths
2022-06-30 16:39:13,718 [    INFO] Logs saved in /home/yoh/.cache/dandi-cli/log/20220630203911Z-3420683.log

and in the log I see

2022-06-30T16:39:13-0400 [DEBUG   ] dandi 3420683:140027539093312 Failed to get metadata for sub-MITU01/ses-20210521h17m17s06/micr/sub-MITU01_ses-20210521h17m17s06_sample-178_stain-NN_run-1_chunk-1_SPIM.ome.zarr: No dandiset at sub-MITU01/ses-20210521h17m17s06/micr/sub-MITU01_ses-20210521h17m17s06_sample-178_stain-NN_run-1_chunk-1_SPIM.ome.zarr
2022-06-30T16:39:13-0400 [DEBUG   ] dandi 3420683:140027539093312 Problem obtaining metadata for sub-MITU01/ses-20210521h17m17s06/micr/sub-MITU01_ses-20210521h17m17s06_sample-178_stain-NN_run-1_chunk-1_SPIM.ome.zarr: 'NoneType' object has no attribute 'items'

[yarikoptic]

ping on this issue -- remains as of 0.39.6+39.g8139966 of #1011

[TheChymera]

The issue has evolved a bit since this was reported, sadly not towards a solution. ls seems all-round broken for BIDS and does not use the object logic that was recently introduced. Some refactoring is needed, probably best discussed/reviewed in the (currently draft) PR linked above.

[TheChymera]

@yarikoptic are you satisfied with this? efab43e (the fix being not just in that one commit, it's more the whole PR, but that commit tests it.)

[yarikoptic]

looks ok, thanks!

Unrelated (I hope) issue although I do not see digest calculated for .nwb -- why do we seems to calculate (but not even output) digest for the .zarr -- any idea?

❯ dandi ls -f yaml micr_SEMzarr/sub-01/ses-01/micr/sub-01_ses-01_sample-A_SPIM.ome.zarr
2022-12-05 20:37:11,908 [    INFO] Calculating digest for micr_SEMzarr/sub-01/ses-01/micr/sub-01_ses-01_sample-A_SPIM.ome.zarr
2022-12-05 20:37:12,144 [ WARNING] BIDSVersion `1.7.0` is less than the minimal working `schema`. Falling back to `schema`. To force the usage of earlier versions specify them explicitly when calling the validator.
- age: null
  genotype: null
  identifier: '01'
  path: micr_SEMzarr/sub-01/ses-01/micr/sub-01_ses-01_sample-A_SPIM.ome.zarr
  sex: null
  species: null
2022-12-05 20:37:13,356 [    INFO] Logs saved in /home/yoh/.cache/dandi-cli/log/20221206013708Z-357541.log
dandi ls -f yaml   5.68s user 1.45s system 102% cpu 6.975 total

[TheChymera]

I was unaware that we would ever want to display checksums in ls, we don't display them for non-zarr files either. I am pretty sure the digest computation is just an internal requirement of the function. @jwodder is that right, or should be outputting it somehow?