For me this error occurs when:

bids_analysis = Analysis(bids_layout, model)
bids_analysis.setup(**entities)

TypeError: unhashable type: 'list'

Looking at the traceback:

collections = self.layout.get_collections(self.level, drop_na=drop_na, **kwargs)

seems to be what's failing because in:

> /celery_worker/src/pybids/bids/variables/entities.py(161)get_nodes()
    159         sort_cols = [sc for sc in sort_cols if sc in set(rows.columns)]
    160         sort_cols += list(set(rows.columns) - set(sort_cols))
--> 161         rows = rows.sort_values(sort_cols)
    162         inds = rows['node_index'].astype(int)
    163         return [self.nodes[i] for i in inds]

In this dataaset, rows contains some values that are lists because of structured meta-data.

For example:

rows['ImageType'] 
0    [ORIGINAL, PRIMARY, FMRI, NONE, ND, NORM, MOSA]
Name: ImageType, dtype: object

I can't really make sense of why that requires sorting (if its not being returned), and if only node_index is being accesed. But either way, I think we need to add meta-data with lists to the tests, as that's probably why we didn't catch this.

This is a near-total refactoring of the core BIDSLayout object. The main point of this exercise is to increase maintainability (I don't want to be in charge of this forever) by removing the grabbit dependency and increasing code clarity. With a few very minor exceptions, the user-facing API remains unperturbed (as evidence of that, the tests for most modules passed with virtually no modification). Two side benefits of the overhaul are that (a) indexing should be substantially more efficient, and (b) there's a new (currently undocumented) object-oriented API that allows one to walk down a BIDS project (each BIDSLayout has a root_node attribute, which is a BIDSNode object that has helpful attributes like .children, .parent, .path, entities, etc.).

API changes: I believe the only common case where this PR breaks the public 0.7 API is in the handling of derivatives in get() calls. Previously, get took a derivatives argument that indicates whether to search derivatives. This argument has been replaced with a new scope argument. The behavior of scope is quite similar (see docstring), but I think a name change is in order because the scope argument now allows one to scan only derivatives folders, which was previously impossible.

There are quite a few changes that could potentially break things for people who were already reaching into the internals. Most notably, _get_nearest_helper is gone. If you were using this, you need to switch to get_nearest, and you'll need to update the calling arguments (see examples throughout the codebase, but most likely you need to set all_=True and possibly pass ignore_strict_entities). Also, the Domain class is gone. Every Entity must now be unique across the entire BIDSLayout and any derivative layouts. This has some downsides, but I decided we could live with them, as it simplifies the code quite a bit. The upshot is that if the word 'domain' occurs anywhere in a workflow, you'll probably need to make minor changes to get things to work again (in most function calls, you may just be able to delete the domain argument without any problems).

Beyond that, it's hard for me to anticipate where else changes with downstream impacts might have happened, so @effigies and @adelavega, maybe try running with this branch and let me know if anything breaks in a way that isn't trivially fixable. Keep in mind that I'm still porting over peripheral functionality from grabbit.

Remaining items:

• [x] Add back all functionality from the old Layout (currently the BIDSLayout is stripped down to bare essentials)
• [x] Port the extensions in grabbit (e.g., file-writing and path-building)
• [x] Port grabbit tests to pybids (basically, all functionality that isn't currently covered by the pybids tests)
• [x] Add docstrings and update docs to reflect the changes
• [x] Add back support for include and exclude directives to allow file filtration at a more fine-grained level than entire derivatives pipelines.

Closes #350. Closes #285.

Adds files needed for submission to JOSS (see #394). All: please look over paper.md and feel free to provide comments/edits. Keep in mind that this is meant to be "a summary describing the high-level functionality and purpose of the software for a diverse, non-specialist audience" (see author guidelines); it is not a tutorial on how to use the software. It's supposed to be 250 - 1,000 words (we're currently around 600).

I'll give this a few days, then will probably submit early next week. Please provide comments before then.

This is a complete refactoring of the analysis module. Nearly all of the change are internal, so I won't go into detail, but I'll summarize some of the key changes, plus the (fairly small) changes in the API:

• The analysis module has been split up into separate variables and analysis modules. variables contains functionality for working with variables read from BIDSLayout objects. No analysis is implied by anything here; it's just a set of tools for loading, representing, and accessing variables from BIDS projects.
• The variables module is structured into entities, variables, collections, and io submodules:
  • entities contains classes that try to mirror the BIDS structure relatively closely. Basically, it's a big tree that begins with a single Dataset node and then branches into Subject, Session, and Run levels. Users will rarely if ever have any reason to use anything inside entities directly.
  • variables contains classes that represent individual variables.
  • collections contains classes that represent collections of variables.
  • io contains a single load_variables function (plus some private handlers) that is the workhorse for extracting variables from a BIDSLayout. However, it doesn't need to be called directly very often, because...
• The BIDSLayout class now implements a get_collections class method that returns BIDSVariableCollections containing all kinds of variable goodies. For most users, interactions with variables can probably be accomplished exclusively through this function.
• There are actually some tests this time. Not a lot, but some. This is progress.
• The Analysis API has been simultaneously simplified and made more flexible. The minimum needed to run an analysis is now:

layout = BIDSLayout('./ds000114')
analysis = Analysis(layout, 'model.json')
analysis.setup(task='fingerfootlips')

# Then we can do things like...
dm = analysis[0].get_design_matrix(format='wide', entities=False, subject='01', run=1)[0]
cm = analysis[0].get_contrast_matrix(subject='01', run=1)[0]

The problems with generating contrast matrices are now solved; the returned contrast matrix should have contrasts in rows and images in columns, with appropriate coding.

• The Analysis interface is also now more efficient in some ways. In particular, variable loading is now done on a per-node basis, so that passing **selector constraints won't require loading of all variables. This means that the following works efficiently now:

layout = BIDSLayout('./ds000114')
analysis = Analysis(layout, 'model.json')
analysis.setup(task='fingerfootlips', subject='01')
...

Because variable loading is now deferred to the Block-level setup, the above snippet would only read variables for subject '01'. This makes it efficient to loop over subjects, sessions, runs, etc., which is probably a more natural way to interact with the Analysis object in many cases.

That all said, this isn't quite finished. Things left to do (I should be able to get to these on Wednesday):

• [x] More testing (@effigies, please flag problems you find)
• [x] I haven't tested reading from scans.tsv, sessions.tsv, or participants.tsv at all. This will be much easier with the new synthetic dataset.
• [x] Propagating the results of one Block onto the next isn't working yet. This should be a quick fix though. For now, the first level should work fine, but the rest won't.
• [x] Cleanup and harmonization of the entire codebase, which probably has some jagged edges after this major refactoring (again, @effigies, feel free to take a pass through)
• [ ] Reimplement aggregation of Collections into a higher level.
• [ ] Merging Collections at different levels doesn't work yet. This is not a huge deal since it's not necessary for standard summary statistics-style workflows. It will be necessary however for more complex analyses (e.g., fitting one big model for all subjects at once).

This replaces much of the code in get_metadata() with a cleaner approach that leverages the latest version of grabbit. Also provides a generic get_nearest() method that can be easily used to walk up the hierarchy and get the appropriate file for any given input file. E.g.,

layout = BIDSLayout(path)
event_file = layout.get_nearest('path/to/sub-03_ses-1_run-1_bold.nii.gz', type='event', all_=True)

...would return all matching event files in the hierarchy for the given functional.

I'm trying to pass variable names to a bids model with wildcards like so:

"Model": { "X": [ 'stimulus_type.neutral_anticipation', 'stimulus_type.neutral_feedback', 'stimulus_type.loss_anticipation', 'stimulus_type.avoided_loss_feedback', 'stimulus_type.gain_anticipation', 'stimulus_type.gain_feedback', 'stimulus_type.missed_gain_feedback', 'stimulus_type.loss_feedback', 'trans_', 'rot_', 'cosine*', 'a_comp_cor_', 't_comp_cor_', 'aroma_motion_', 'censor_' ], },

This is throwing a key error trans_*

Traceback (most recent call last):
  File "/data/nielsond/python/envs/fitlins_dev/lib/python3.6/site-packages/nipype/pipeline/plugins/multiproc.py", line 69, in run_node
    result['result'] = node.run(updatehash=updatehash)
  File "/data/nielsond/python/envs/fitlins_dev/lib/python3.6/site-packages/nipype/pipeline/engine/nodes.py", line 473, in run
    result = self._run_interface(execute=True)
  File "/data/nielsond/python/envs/fitlins_dev/lib/python3.6/site-packages/nipype/pipeline/engine/nodes.py", line 564, in _run_interface
    return self._run_command(execute)
  File "/data/nielsond/python/envs/fitlins_dev/lib/python3.6/site-packages/nipype/pipeline/engine/nodes.py", line 649, in _run_command
    result = self._interface.run(cwd=outdir)
  File "/data/nielsond/python/envs/fitlins_dev/lib/python3.6/site-packages/nipype/interfaces/base/core.py", line 376, in run
    runtime = self._run_interface(runtime)
  File "/gpfs/gsfs11/users/MBDU/midla/notebooks/code/fitlins/fitlins/interfaces/bids.py", line 235, in _run_interface
    analysis.setup(drop_na=False, desc='preproc', **selectors)
  File "/gpfs/gsfs11/users/MBDU/midla/notebooks/code/pybids/bids/analysis/analysis.py", line 89, in setup
    b.setup(input_nodes, drop_na=drop_na, **selectors)
  File "/gpfs/gsfs11/users/MBDU/midla/notebooks/code/pybids/bids/analysis/analysis.py", line 213, in setup
    transform.Select(coll, X)
  File "/gpfs/gsfs11/users/MBDU/midla/notebooks/code/pybids/bids/analysis/transformations/base.py", line 87, in __new__
    return t.transform()
  File "/gpfs/gsfs11/users/MBDU/midla/notebooks/code/pybids/bids/analysis/transformations/base.py", line 223, in transform
    self._clone_variables()
  File "/gpfs/gsfs11/users/MBDU/midla/notebooks/code/pybids/bids/analysis/transformations/base.py", line 123, in _clone_variables
    for v in self.variables}
  File "/gpfs/gsfs11/users/MBDU/midla/notebooks/code/pybids/bids/analysis/transformations/base.py", line 123, in <dictcomp>
    for v in self.variables}
  File "/gpfs/gsfs11/users/MBDU/midla/notebooks/code/pybids/bids/variables/kollekshuns.py", line 184, in __getitem__
    return self.variables[var]
KeyError: 'trans_*'

I get the same thing if i use trans_.* I'm pretty sure this is happening because no matching is being done on X here: https://github.com/bids-standard/pybids/blob/15aa08fb2b146e3fcc62f359ea4a4d810d38591a/bids/analysis/analysis.py#L212-L213

I think there is currently some serious performance issues with BIDSLayout. Using a somewhat average database of 132 subjects (1 session and 1 run per subject), it needs about 1:15 minute to get a layout object. Using the following code:

from bids import BIDSLayout
%lprun -f BIDSLayout.__init__ BIDSLayout("/media/christian/ElementsSE/MPI-Leipzig_Mind-Brain-Body-LEMON/BIDS_LEMON/")

I get the following profiling report :

Total time: 76.4714 s
File: /home/christian/pybids/bids/layout/layout.py
Function: __init__ at line 196

Line #      Hits         Time  Per Hit   % Time  Line Contents
==============================================================
   196                                               def __init__(self, root, validate=True, absolute_paths=True,
   197                                                            derivatives=False, config=None, sources=None, ignore=None,
   198                                                            force_index=None, config_filename='layout_config.json',
   199                                                            regex_search=False, database_path=None, database_file=None,
   200                                                            reset_database=False, index_metadata=True):
   201                                                   """Initialize BIDSLayout."""
   202         1          4.0      4.0      0.0          self.root = str(root)
   203         1          2.0      2.0      0.0          self.validate = validate
   204         1          1.0      1.0      0.0          self.absolute_paths = absolute_paths
   205         1          2.0      2.0      0.0          self.derivatives = {}
   206         1          2.0      2.0      0.0          self.sources = sources
   207         1          3.0      3.0      0.0          self.regex_search = regex_search
   208         1          2.0      2.0      0.0          self.config_filename = config_filename
   209                                                   # Store original init arguments as dictionary
   210         1          3.0      3.0      0.0          self._init_args = self._sanitize_init_args(
   211         1          2.0      2.0      0.0              root=root, validate=validate, absolute_paths=absolute_paths,
   212         1          2.0      2.0      0.0              derivatives=derivatives, ignore=ignore, force_index=force_index,
   213         1         91.0     91.0      0.0              index_metadata=index_metadata, config=config)
   214                                           
   215         1          4.0      4.0      0.0          if database_path is None and database_file is not None:
   216                                                       database_path = database_file
   217                                                       warnings.warn(
   218                                                           'In pybids 0.10 database_file argument was deprecated in favor'
   219                                                           ' of database_path, and will be removed in 0.12. '
   220                                                           'For now, treating database_file as a directory.',
   221                                                           DeprecationWarning)
   222         1          4.0      4.0      0.0          if database_path:
   223                                                       database_path = str(Path(database_path).absolute())
   224                                           
   225         1         47.0     47.0      0.0          self.session = None
   226                                           
   227         1      25891.0  25891.0      0.0          index_dataset = self._init_db(database_path, reset_database)
   228                                           
   229                                                   # Do basic BIDS validation on root directory
   230         1        488.0    488.0      0.0          self._validate_root()
   231                                           
   232         1          4.0      4.0      0.0          if ignore is None:
   233         1          3.0      3.0      0.0              ignore = self._default_ignore
   234                                           
   235                                                   # Instantiate after root validation to ensure os.path.join works
   236         1          3.0      3.0      0.0          self.ignore = [os.path.abspath(os.path.join(self.root, patt))
   237                                                                  if isinstance(patt, str) else patt
   238         1        102.0    102.0      0.0                         for patt in listify(ignore or [])]
   239         1          3.0      3.0      0.0          self.force_index = [os.path.abspath(os.path.join(self.root, patt))
   240                                                                       if isinstance(patt, str) else patt
   241         1          4.0      4.0      0.0                              for patt in listify(force_index or [])]
   242                                           
   243                                                   # Initialize the BIDS validator and examine ignore/force_index args
   244         1          4.0      4.0      0.0          self._validate_force_index()
   245                                           
   246         1          1.0      1.0      0.0          if index_dataset:
   247                                                       # Create Config objects
   248         1          2.0      2.0      0.0              if config is None:
   249         1          2.0      2.0      0.0                  config = 'bids'
   250         1          1.0      1.0      0.0              config = [Config.load(c, session=self.session)
   251         1      62271.0  62271.0      0.1                        for c in listify(config)]
   252         1         15.0     15.0      0.0              self.config = {c.name: c for c in config}
   253                                                       # Missing persistence of configs to the database
   254         2          6.0      3.0      0.0              for config_obj in self.config.values():
   255         1        308.0    308.0      0.0                  self.session.add(config_obj)
   256         1      27372.0  27372.0      0.0                  self.session.commit()
   257                                           
   258                                                       # Index files and (optionally) metadata
   259         1         35.0     35.0      0.0              indexer = BIDSLayoutIndexer(self)
   260         1   28127988.0 28127988.0     36.8              indexer.index_files()
   261         1          3.0      3.0      0.0              if index_metadata:
   262         1   48226769.0 48226769.0     63.1                  indexer.index_metadata()
   263                                                   else:
   264                                                       # Load Configs from DB
   265                                                       self.config = {c.name: c for c in self.session.query(Config).all()}
   266                                           
   267                                                   # Add derivatives if any are found
   268         1          3.0      3.0      0.0          if derivatives:
   269                                                       if derivatives is True:
   270                                                           derivatives = os.path.join(root, 'derivatives')
   271                                                       self.add_derivatives(
   272                                                           derivatives, parent_database_path=database_path,
   273                                                           validate=validate, absolute_paths=absolute_paths,
   274                                                           derivatives=None, sources=self, ignore=ignore,  config=None,
   275                                                           force_index=force_index, config_filename=config_filename,
   276                                                           regex_search=regex_search, index_metadata=index_metadata,
   277                                                           reset_database=index_dataset or reset_database
   278                                                           )

For day-to-day interaction with a dataset, development tests, etc., this kind of delay seems prohibitive to me...

I have a set of derivatives files that look like this:

derivatives/rt/sub-02/func/sub-02_task-dts_meanRT.tsv

they seem to pass the validator fine, but I get nothing when I use .get(scope='derivatives'):

In [35]: layout2 = BIDSLayout(args.bids_dir, derivatives=True)

In [36]: layout2.get(scope='derivatives')
Out[36]: []

When trying to resample a dense variable (physio) with a sampling rate of 0.49999984750004656 the length of the x given to interp1d is one longer than values. My guess is that this has to do with a rounding error with the funky sampling_rate.

#464 breaks FitLins. I recognize that these derivatives aren't standardized, but neither is "contrast" a standardized name for the "ce" entity. As it is, trying to parse a FitLins derivatives structure yields:

*** sqlalchemy.orm.exc.FlushError: New instance <Entity at 0x2b8e1e860c18> with identity key (<class 'bids.layout.models.Entity'>, ('contrast',), None) conflicts with persistent instance <Entity at 0x2b8e1d8da208>

@tyarkoni @oesteban any objections to reverting this?

Hey - I was trying to use pybids in a pipeline of mine, but found that it takes a very long time to create a layout on large datasets. I tested this using the public NKI-RS dataset from FCP-INDI in BIDS format, and ran using subsets of the dataset of various size: 2 subjects, 90, and the full thing, 963.

My script is the following:

from bids.layout import BIDSLayout
import time

start1 = time.time()
bl = BIDSLayout('/project/6008063/gkiar/data/smallRS')
dur1 = time.time() - start1
print("{0}:{1}".format(len(bl.get_subjects()), dur1))


start2 = time.time()
bl = BIDSLayout('/project/6008063/gkiar/data/medRS/')
dur2 = time.time() - start2
print("{0}:{1}".format(len(bl.get_subjects()), dur2))

start3 = time.time()
bl = BIDSLayout('/project/6008063/gkiar/data/RocklandSample/')
dur3 = time.time() - start3
print("{0}:{1}".format(len(bl.get_subjects()), dur3))

The output, in the form of {n subs}:{time elapsed in seconds}, is:

$ date
Wed Nov 14 15:40:58 EST 2018
$ python profilebl.py
2:0.39104771614074707
90:14.413928270339966
963:631.153669834137

With this being the first step of my pipeline, and I'm almost always specifying a --participant_label or --session_label for a bunch of these launched in parallel, it would be great if it didn't take upwards of 10 minutes per task to find the data. I'd still like to have the BIDSLayout as part of the pipeline itself so I can grab various pieces of metadata as I need them.

Any ideas why this gets so slow as sample size increases, or places you suggest I could make a PR to speed things up (likely in grabbit)?

• closes #661

Summary

Add en EventPlotter class to handle plotting of events file with plotly.

• plots time course of each trial type
• plots inter stimulus interval histogram
• plots responses and response time histogram if there is a response_time column
• allows to select which column should define the events to include

Possible addition

• [x] duration histogram
• [x] pass list of trial types to include in plot
• [ ] accept dataframe and BIDSDataFile as input ?

TODO

• documentation ?

Hello: I'm trying to use PyBIDS in Google Collab. However, after installing it:

!pip3 install git+https://github.com/bids-standard/pybids.git

Looking in indexes: https://pypi.org/simple, https://us-python.pkg.dev/colab-wheels/public/simple/ Collecting git+https://github.com/bids-standard/pybids.git Cloning https://github.com/bids-standard/pybids.git to /tmp/pip-req-build-ho9uyjz5 Running command git clone --filter=blob:none --quiet https://github.com/bids-standard/pybids.git /tmp/pip-req-build-ho9uyjz5 Resolved https://github.com/bids-standard/pybids.git to commit 18230cfe0facbc3b21f978613e20e5c53f6fd5f3 Running command git submodule update --init --recursive -q Installing build dependencies ... done Getting requirements to build wheel ... done Installing backend dependencies ... done Preparing metadata (pyproject.toml) ... done Requirement already satisfied: num2words in /usr/local/lib/python3.8/site-packages (from pybids==0.15.5.post0.dev16) (0.5.12) Requirement already satisfied: sqlalchemy<1.4.0.dev0 in /usr/local/lib/python3.8/site-packages (from pybids==0.15.5.post0.dev16) (1.3.24) Requirement already satisfied: scipy in /usr/local/lib/python3.8/site-packages (from pybids==0.15.5.post0.dev16) (1.9.3) Requirement already satisfied: nibabel>=2.1 in /usr/local/lib/python3.8/site-packages (from pybids==0.15.5.post0.dev16) (4.0.2) Requirement already satisfied: formulaic<0.6,>=0.2.4 in /usr/local/lib/python3.8/site-packages (from pybids==0.15.5.post0.dev16) (0.2.4) Requirement already satisfied: click>=8.0 in /usr/local/lib/python3.8/site-packages (from pybids==0.15.5.post0.dev16) (8.1.3) Requirement already satisfied: numpy in /usr/local/lib/python3.8/site-packages (from pybids==0.15.5.post0.dev16) (1.24.1) Requirement already satisfied: bids-validator in /usr/local/lib/python3.8/site-packages (from pybids==0.15.5.post0.dev16) (1.9.9) Requirement already satisfied: pandas>=0.23 in /usr/local/lib/python3.8/site-packages (from pybids==0.15.5.post0.dev16) (1.5.2) Requirement already satisfied: interface-meta>=1.2 in /usr/local/lib/python3.8/site-packages (from formulaic<0.6,>=0.2.4->pybids==0.15.5.post0.dev16) (1.3.0) Requirement already satisfied: wrapt in /usr/local/lib/python3.8/site-packages (from formulaic<0.6,>=0.2.4->pybids==0.15.5.post0.dev16) (1.14.1) Requirement already satisfied: astor in /usr/local/lib/python3.8/site-packages (from formulaic<0.6,>=0.2.4->pybids==0.15.5.post0.dev16) (0.8.1) Requirement already satisfied: packaging>=17.0 in /usr/local/lib/python3.8/site-packages (from nibabel>=2.1->pybids==0.15.5.post0.dev16) (22.0) Requirement already satisfied: setuptools in /usr/local/lib/python3.8/site-packages (from nibabel>=2.1->pybids==0.15.5.post0.dev16) (65.5.0) Requirement already satisfied: python-dateutil>=2.8.1 in /usr/local/lib/python3.8/site-packages (from pandas>=0.23->pybids==0.15.5.post0.dev16) (2.8.2) Requirement already satisfied: pytz>=2020.1 in /usr/local/lib/python3.8/site-packages (from pandas>=0.23->pybids==0.15.5.post0.dev16) (2022.7) Requirement already satisfied: docopt>=0.6.2 in /usr/local/lib/python3.8/site-packages (from num2words->pybids==0.15.5.post0.dev16) (0.6.2) Requirement already satisfied: six>=1.5 in /usr/local/lib/python3.8/site-packages (from python-dateutil>=2.8.1->pandas>=0.23->pybids==0.15.5.post0.dev16) (1.16.0)

• If I try to import:

from bids import BIDSLayout

I get module not found error:

ModuleNotFoundError: No module named 'bids'

• Also, in the next case:

from pybids import BIDSLayout

ModuleNotFoundError: No module named 'pybids'

Please, could you clarify how to do the import? Thank you in advance, María

It looks like newer version of pandas and pybids are having issues with some syntax in the Scale and Convolve transformations.

In particular, Scale has an issue when performing a boolean operation on a Series

Convolve requires the oversampling parameter to be coerced to an int

Dear developers,

Is it possible to select a specific session and a specific run for a specific subject? Could it be possible to input 3 lists with the desired combinations (participant_id, session_id, run_id) and obtain the selected files (e.g T1w) ? See an example in the attached csv file.

Best and many thanks in advance for your help and your advice Yasser Alemán Gómez

If presented with a subject with an empty anat/ directory, PyBIDS will raise an unhandled StopIteration:

  File "/opt/conda/lib/python3.9/site-packages/fmriprep/config.py", line 472, in init
    cls._layout = BIDSLayout(
  File "/opt/conda/lib/python3.9/site-packages/bids/layout/layout.py", line 152, in __init__
    indexer(self)
  File "/opt/conda/lib/python3.9/site-packages/bids/layout/index.py", line 109, in __call__
    self._index_dir(self._layout._root, self._config)
  File "/opt/conda/lib/python3.9/site-packages/bids/layout/index.py", line 193, in _index_dir
    self._index_dir(d, list(config), default_action=default)
  File "/opt/conda/lib/python3.9/site-packages/bids/layout/index.py", line 193, in _index_dir
    self._index_dir(d, list(config), default_action=default)
  File "/opt/conda/lib/python3.9/site-packages/bids/layout/index.py", line 173, in _index_dir
    _, dirnames, filenames = next(os.walk(path))

Running through pdb:

StopIteration
Uncaught exception. Entering post mortem debugging
Running 'cont' or 'step' will restart the program
> /opt/conda/lib/python3.9/site-packages/bids/layout/index.py(173)_index_dir()
-> _, dirnames, filenames = next(os.walk(path))
(Pdb) print(path)
/data/sub-050202002B/anat
(Pdb) up
> /opt/conda/lib/python3.9/site-packages/bids/layout/index.py(193)_index_dir()
-> self._index_dir(d, list(config), default_action=default)
(Pdb) print(d)
/data/sub-050202002B/anat
(Pdb) up
> /opt/conda/lib/python3.9/site-packages/bids/layout/index.py(193)_index_dir()
-> self._index_dir(d, list(config), default_action=default)
(Pdb) print(d)
/data/sub-050202002B

Initially reported in https://github.com/nipreps/fmriprep/issues/2902