read_raw_snirf fails with TypeError when landmarkLabels is empty #13627

Description of the problem

mne.io.read_raw_snirf raises TypeError: iteration over a 0-d array when reading SNIRF files where /nirs/probe/landmarkLabels exists but is empty or scalar.

Steps to reproduce

import mne  # Download test file from OpenNeuro ds005929 
# File: sub-6016_ses-1_task-MA_run-01_nirs.snirf raw = mne.io.read_raw_snirf("sub-6016_ses-1_task-MA_run-01_nirs.snirf")

Link to data

aws s3 sync --no-sign-request s3://openneuro.org/ds005929/sub-6016 ./ds005929/sub-6016   

Expected results

File loads successfully, treating empty landmark labels as no landmarks.

Actual results

Actual Behavior

TypeError: iteration over a 0-d array

Full traceback:

File "mne/io/snirf/_snirf.py", line 393, 
in __init__     for idx, dign in enumerate(digname): 
TypeError: iteration over a 0-d array

Root Cause

In _snirf.py line 390, landmarkLabels is converted to a numpy array without checking dimensions:

digname = np.array(dat.get("/nirs/probe/landmarkLabels"))

When landmarkLabels is empty or scalar, this creates a 0-d array that cannot be iterated.

Suggested Fix

Apply _correct_shape() (already defined in the file) to digname, or add a dimension check:

digname = np.array(dat.get("/nirs/probe/landmarkLabels"))
if digname.ndim == 0:   
     digname = digname.reshape(-1) if digname.size > 0 else [] 

Additional information

Platform

• MNE: 1.12.0.dev (bug also present in main branch) - Python: 3.12 - OS: macOS