Merge pull request #647 from dbic/pr-540

strip non-alphanumeric from session ids too

heudiconv/bids.py

@@ -14,6 +14,7 @@
 from random import sample
 from glob import glob
 import errno
+import warnings
 
 from .external.pydicom import dcm
 
@@ -502,7 +503,7 @@ def get_formatted_scans_key_row(dcm_fn):
 
 
 def convert_sid_bids(subject_id):
-    """Strips any non-BIDS compliant characters within subject_id
+    """Shim for stripping any non-BIDS compliant characters within subject_id
 
     Parameters
     ----------
@@ -515,15 +516,10 @@ def convert_sid_bids(subject_id):
     subject_id : string
         Original subject ID
     """
-    cleaner = lambda y: ''.join([x for x in y if x.isalnum()])
-    sid = cleaner(subject_id)
-    if not sid:
-        raise ValueError(
-            "Subject ID became empty after cleanup.  Please provide manually "
-            "a suitable alphanumeric subject ID")
-    lgr.warning('{0} contained nonalphanumeric character(s), subject '
-                'ID was cleaned to be {1}'.format(subject_id, sid))
-    return sid, subject_id
+    warnings.warn('convert_sid_bids() is deprecated, '
+                  'please use sanitize_label() instead.',
+                  DeprecationWarning)
+    return sanitize_label(subject_id)
 
 
 def get_shim_setting(json_file):
@@ -1028,3 +1024,26 @@ def suffix(self):
     @property
     def extension(self):
         return self._extension
+
+
+def sanitize_label(label):
+    """Strips any non-BIDS compliant characters within label
+
+    Parameters
+    ----------
+    label : string
+
+    Returns
+    -------
+    clean_label : string
+        New, sanitized label
+    """
+    clean_label = ''.join(x for x in label if x.isalnum())
+    if not clean_label:
+        raise ValueError(
+            "Label became empty after cleanup.  Please provide manually "
+            "a suitable alphanumeric label.")
+    if clean_label != label:
+        lgr.warning('%r label contained non-alphanumeric character(s), it '
+                    'was cleaned to be %r', label, clean_label)
+    return clean_label

heudiconv/convert.py

@@ -26,7 +26,7 @@
     file_md5sum
 )
 from .bids import (
-    convert_sid_bids,
+    sanitize_label,
     populate_bids_templates,
     populate_intended_for,
     save_scans_key,
@@ -102,8 +102,9 @@ def prep_conversion(sid, dicoms, outdir, heuristic, converter, anon_sid,
         if not sid:
             raise ValueError(
                 "BIDS requires alphanumeric subject ID. Got an empty value")
-        if not sid.isalnum():  # alphanumeric only
-            sid, old_sid = convert_sid_bids(sid)
+        sid = sanitize_label(sid)
+        if ses:
+            ses = sanitize_label(ses)
 
     if not anon_sid:
         anon_sid = sid

heudiconv/tests/test_bids.py

@@ -42,6 +42,7 @@
     AllowedCriteriaForFmapAssignment,
     KeyInfoForForce,
     BIDSFile,
+    sanitize_label,
 )
 from heudiconv.cli.run import main as runner
 
@@ -1155,3 +1156,8 @@ def test_ME_mag_phase_conversion(tmpdir, subID='MEGRE', heuristic='bids_ME.py'):
                     % (subID, subID, e, part, ext)
                 )
 
+
+def test_sanitize_label():
+    assert sanitize_label('az XZ-@09') == 'azXZ09'
+    with pytest.raises(ValueError):
+        sanitize_label(' @ ')