fixed sanitization syntax, added more robust tests

nwb2bids/base.py

@@ -145,7 +145,7 @@ def reposit(in_dir, out_dir, *, no_copy=False):
 
 
 def sanitize_bids_value(in_string, pattern=r"[^a-zA-Z0-9]", replacement="X"):
-    out_string = re.sub(in_string, pattern, replacement)
+    out_string = re.sub(pattern, replacement, in_string)
     return out_string
 
 

nwb2bids/conftest.py

@@ -30,3 +30,28 @@ def nwb_testdata(
         io.write(nwbfile)
 
     return filename
+
+@pytest.fixture(scope="session")
+def nwb_testdata_nosessionid(
+    tmp_path_factory,
+    ):
+    from pynwb.testing.mock.file import mock_NWBFile
+    from pynwb.file import Subject
+
+    nwbfile = mock_NWBFile(
+        session_start_time = datetime(2024, 3, 9, 22, 30, 3, tzinfo=tz.gettz("US/Eastern")),
+        session_id = "",
+        )
+    time_series = pynwb.TimeSeries(name="Test", data=numpy.array([], dtype="uint8"), unit="n.a.", rate=1.0)
+    nwbfile.add_acquisition(time_series)
+
+    subject = Subject(
+        subject_id="12_34",
+        sex="male",
+        )
+    nwbfile.subject = subject
+    filename = tmp_path_factory.mktemp("test_nwb2bids") / "testfile.nwb"
+    with pynwb.NWBHDF5IO(path=filename, mode="w") as io:
+        io.write(nwbfile)
+
+    return filename

nwb2bids/test/test_base.py

@@ -2,24 +2,52 @@
 import os
 
 def test_reposit(nwb_testdata, tmp_path):
+   tmp_path = str(tmp_path)
    nwb_testdir = os.path.dirname(nwb_testdata)
    base.reposit(nwb_testdir, tmp_path)
    # This should be done by invoking the validator once BEP032 is in the BIDS schema:
    for root, dirs, files in os.walk(tmp_path):
       if root == tmp_path:
-         assert dirs == ['sub-1234']
+         assert dirs == ['sub-12X34']
          assert set(files) == set(['participants.json', 'participants.tsv'])
-      if root == os.path.join(tmp_path, 'sub-1234'):
+      elif root == os.path.join(tmp_path, 'sub-12X34'):
          assert dirs == ['ses-20240309']
          assert set(files) == set(['sessions.json', 'sessions.tsv'])
-      if root == os.path.join(tmp_path, 'sub-1234', 'ses-20240309'):
+      elif root == os.path.join(tmp_path, 'sub-12X34', 'ses-20240309'):
          assert dirs == ['ephys']
          assert files == []
-      if root == os.path.join(tmp_path, 'sub-1234', 'ses-20240309', 'ephys'):
+      elif root == os.path.join(tmp_path, 'sub-12X34', 'ses-20240309', 'ephys'):
          assert dirs == []
          assert set(files) == set([
-            'sub-1234_contacts.tsv',
-            'sub-1234_channels.tsv',
-            'sub-1234_probes.tsv',
-            'sub-1234_ses-20240309_ephys.nwb',
+            'sub-12X34_contacts.tsv',
+            'sub-12X34_channels.tsv',
+            'sub-12X34_probes.tsv',
+            'sub-12X34_ses-20240309_ephys.nwb',
             ])
+      else:
+         print(root, files, dirs)
+         raise
+
+def test_reposit_nosessionid(nwb_testdata_nosessionid, tmp_path):
+   tmp_path = str(tmp_path)
+   nwb_testdir = os.path.dirname(nwb_testdata_nosessionid)
+   base.reposit(nwb_testdir, tmp_path)
+   # This should be done by invoking the validator once BEP032 is in the BIDS schema:
+   for root, dirs, files in os.walk(tmp_path):
+      if root == tmp_path:
+         assert dirs == ['sub-12X34']
+         assert set(files) == set(['participants.json', 'participants.tsv'])
+      elif root == os.path.join(tmp_path, 'sub-12X34'):
+         assert dirs == ['ephys']
+         assert set(files) == set(['sessions.json', 'sessions.tsv'])
+      elif root == os.path.join(tmp_path, 'sub-12X34', 'ephys'):
+         assert dirs == []
+         assert set(files) == set([
+            'sub-12X34_channels.tsv',
+            'sub-12X34__ephys.nwb',
+            'sub-12X34_contacts.tsv',
+            'sub-12X34_probes.tsv',
+            ])
+      else:
+         print(root, files, dirs)
+         raise