Merge pull request Open-Minds-Lab#32 from sinhaharsh/main

Bugfixes for BidsDataset

Author: raamana

MRdataset/bids.py

@@ -4,7 +4,7 @@
 
 from MRdataset import logger
 from MRdataset.base import BaseDataset
-from MRdataset.config import VALID_BIDS_DATATYPES
+from MRdataset.config import VALID_BIDS_DATATYPES, SUPPORTED_BIDS_DATATYPES
 from MRdataset.dicom_utils import is_bids_file
 from MRdataset.utils import folders_with_min_files, valid_dirs, read_json
 from protocol import BidsImagingSequence
@@ -129,6 +129,13 @@ def _process(self, folder):
                                  name=name)
             if seq.is_valid():
                 sequences.append(seq)
+            else:
+                if name not in SUPPORTED_BIDS_DATATYPES:
+                    logger.error(f'MRdataset primarily supports '
+                                 f'{SUPPORTED_BIDS_DATATYPES}'
+                                 f'It seems the parameters in '
+                                 f'this sequence are invalid or '
+                                 f'not supported yet. Skipping it.')
         return sequences
 
     @staticmethod
@@ -138,7 +145,7 @@ def get_run_id(filename, last_id):
         Example filename : sub-01_ses-imagery01_task-imagery_run-01_bold.json
         """
         # Regular expression pattern
-        pattern = r'run-\d+'
+        pattern = r'run-[^_]+'
         # Extracting substring using regex
         match = search(pattern, str(filename))
 

MRdataset/config.py

@@ -30,8 +30,8 @@
     'pet'
 ]
 
+SUPPORTED_BIDS_DATATYPES = ['func', 'anat', 'dwi', 'fmap']
 
-# Suppress duplicated warnings
 
 class MRException(Exception):
     """
@@ -59,7 +59,7 @@ def __init__(self):
                          '--data_source. Got 0 DICOM/JSON files.')
 
 
-def configure_logger(log, output_dir, mode='w', level='WARNING'):
+def configure_logger(log, output_dir, mode='w', level='ERROR'):
     """
     Initiate log files.
 

MRdataset/dicom_utils.py

@@ -29,9 +29,12 @@ def is_bids_file(filename: Union[str, Path]):
     # TODO: Add some criteria to skip certain files
     if 'derivatives' in str(filename):
         return False
-
+    if 'bidsignore' in str(filename):
+        return False
+    if 'sourcedata' in str(filename):
+        return False
     # Regular expression pattern
-    pattern = r'sub-\d+'
+    pattern = r'sub-[^_]+'
     # Extracting substring using regex
     match = search(pattern, str(filename))
     if not match: