[MRG] Update empty-room matching to look for and prioritize same-session recordings with task "noise" (closes #1354)

mne_bids/path.py

@@ -56,12 +56,62 @@
     datatype = "meg"  # We're only concerned about MEG data here
     bids_fname = bids_path.update(suffix=datatype).fpath
     _, ext = _parse_ext(bids_fname)
+    # Create a path for the empty-room directory to be used for matching.
     emptyroom_dir = BIDSPath(root=bids_root, subject="emptyroom").directory
 
-    if not emptyroom_dir.exists():
+    # Find matching "task-noise" files in the same directory as the recording.
+    noisetask_path = bids_path.update(
+        split=None, run=None, task="noise", datatype=datatype, suffix=datatype
+    )
+
+    allowed_extensions = list(reader.keys())
+    # `.pdf` is just a "virtual" extension for BTi data (which is stored inside
+    # a dedicated directory that doesn't have an extension)
+    del allowed_extensions[allowed_extensions.index(".pdf")]
+
+    # Get possible noise task files in the same directory as the recording.
+    noisetask_tmp = [
+        candidate
+        for candidate in noisetask_path.match()
+        if candidate.extension in allowed_extensions
+    ]
+    # For some reason a single file can produce multiple hits in the match function.
+    # Remove dups
+    noisetask_fns = []
+    for i in range(len(noisetask_tmp)):
+        fn = noisetask_tmp.pop()
+        if len(noisetask_tmp) == 0 or not any(
+            fn.fpath == f.fpath for f in noisetask_fns
+        ):
+            noisetask_fns.append(fn)
+
+    # If we have more than one noise task file, we need to disambiguate.
+    # It might be that it's a
+    # split recording.
+    if len(noisetask_fns) > 1 and any(path.split is not None for path in noisetask_fns):
+        noisetask_path.update(spilt="01")
+        noisetask_fns = [
+            candidate
+            for candidate in noisetask_path.match()
+            if candidate.extension in allowed_extensions
+        ]
+
+    # If it wasn't a split recording, warn the user that something is wonky,
+    # then resort to looking for sub-emptyroom recordings and date-matching.
+    if len(noisetask_fns) > 1:
+        msg = (
+            "Found more than one matching noise task file."
+            " Falling back to looking for sub-emptyroom recordings and date-matching."
+        )
+        warn(msg)
+        noisetask_fns = []
+    elif len(noisetask_fns) == 1:
+        return noisetask_fns[0]
+
+    if not emptyroom_dir.exists() and not noisetask_fns:
         return list()
 
-    # Find the empty-room recording sessions.
+    # Check the 'emptyroom' subject for empty-room recording sessions.
     emptyroom_session_dirs = [
         x
         for x in emptyroom_dir.iterdir()
@@ -71,12 +121,6 @@
         emptyroom_session_dirs = [emptyroom_dir]
 
     # Now try to discover all recordings inside the session directories.
-
-    allowed_extensions = list(reader.keys())
-    # `.pdf` is just a "virtual" extension for BTi data (which is stored inside
-    # a dedicated directory that doesn't have an extension)
-    del allowed_extensions[allowed_extensions.index(".pdf")]
-
     candidate_er_fnames = []
     for session_dir in emptyroom_session_dirs:
         dir_contents = glob.glob(
@@ -105,6 +149,10 @@
     from mne_bids import read_raw_bids  # avoid circular import.
 
     candidates = _find_empty_room_candidates(bids_path)
+    # If a single candidate is returned, then there's a same-session noise
+    # task recording that takes priority.
+    if not isinstance(candidates, list):
+        return candidates
 
     # Walk through recordings, trying to extract the recording date:
     # First, from the filename; and if that fails, from `info['meas_date']`.

mne_bids/tests/test_path.py

@@ -1146,6 +1146,7 @@ def test_find_empty_room(return_bids_test_dir, tmp_path):
         task="audiovisual",
         run="01",
         root=bids_root,
+        datatype="meg",
         suffix="meg",
     )
     write_raw_bids(raw, bids_path, overwrite=True, verbose=False)
@@ -1173,6 +1174,30 @@ def test_find_empty_room(return_bids_test_dir, tmp_path):
     recovered_er_bids_path = bids_path.find_empty_room()
     assert er_bids_path == recovered_er_bids_path
 
+    # Test that when there is a noise task file in the subject directory it will take
+    # precedence over the emptyroom directory file
+    os.remove(er_bids_path.fpath)
+    er_noise_task_path = bids_path.copy().update(
+        run=None,
+        task="noise",
+    )
+
+    write_raw_bids(er_raw, er_noise_task_path, overwrite=True, verbose=False)
+    recovered_er_bids_path = bids_path.find_empty_room()
+    assert er_noise_task_path == recovered_er_bids_path
+
+    # Check that when there are multiple matches that cannot be resolved via assigning
+    # split=01 that the sub-emptyroom is the fallback
+    dup_noise_task_path = er_noise_task_path.copy()
+    dup_noise_task_path.update(run="100", split=None)
+    write_raw_bids(er_raw, dup_noise_task_path, overwrite=True, verbose=False)
+    write_raw_bids(er_raw, er_bids_path, overwrite=True, verbose=False)
+    with pytest.warns(RuntimeWarning):
+        recovered_er_bids_path = bids_path.find_empty_room()
+    assert er_bids_path == recovered_er_bids_path
+    os.remove(er_noise_task_path.fpath)
+    os.remove(dup_noise_task_path.fpath)
+
     # assert that we get best emptyroom if there are multiple available
     sh.rmtree(op.join(bids_root, "sub-emptyroom"))
     dates = ["20021204", "20021201", "20021001"]