sane behavior, add changelog

doc/whats_new.rst

@@ -39,6 +39,7 @@ Detailed list of changes
 
 - :func:`mne_bids.read_raw_bids` no longer warns about unit changes in channels upon reading, as that information is taken from ``channels.tsv`` and judged authorative, by `Stefan Appelhoff`_ (:gh:`1282`)
 - MEG OPM channels are now experimentally included, by `Amaia Benitez`_ (:gh:`1222`)
+- :func:`mne_bids.mark_channels` will no longer created a ``status_description`` column filled with ``n/a`` in the ``channels.tsv`` file, by `Stefan Appelhoff`_ (:gh:`1293`)
 
 🛠 Requirements
 ^^^^^^^^^^^^^^^

mne_bids/write.py

@@ -2460,17 +2460,24 @@ def mark_channels(bids_path, *, ch_names, status, descriptions=None, verbose=Non
         type (e.g., only EEG or MEG data) is present in the dataset, it will be
         selected automatically.
     ch_names : str | list of str
-        The names of the channel(s) to mark with a ``status`` and possibly a
+        The names of the channel(s) to mark with a ``status`` and optionally a
         ``description``. Can be an empty list to indicate all channel names.
     status : 'good' | 'bad' | list of str
         The status of the channels ('good', or 'bad'). If it is a list, then must be a
         list of 'good', or 'bad' that has the same length as ``ch_names``.
     descriptions : None | str | list of str
-        Descriptions of the reasons that lead to the exclusion of the
+        Descriptions of the reasons that lead to the marking ('good' or 'bad') of the
         channel(s). If a list, it must match the length of ``ch_names``.
         If ``None``, no descriptions are added.
     %(verbose)s
 
+    Notes
+    -----
+    If the 'status' or 'status_description' columns were not present in the
+    corresponding tsv file before using this function, they may be created with default
+    values ('good' for status, 'n/a' for status_description) for all channels that are
+    not differently specified (by using ``ch_names``, ``status``, and ``descriptions``).
+
     Examples
     --------
     Mark a single channel as bad.
@@ -2525,7 +2532,9 @@ def mark_channels(bids_path, *, ch_names, status, descriptions=None, verbose=Non
     # set descriptions based on how it's passed in
     if isinstance(descriptions, str):
         descriptions = [descriptions] * len(ch_names)
+        write_descriptions = True
     elif not descriptions:
+        write_descriptions = False
         descriptions = [None] * len(ch_names)
 
     # make sure statuses is a list of strings
@@ -2549,11 +2558,17 @@ def mark_channels(bids_path, *, ch_names, status, descriptions=None, verbose=Non
 
     # Read sidecar and create required columns if they do not exist.
     if "status" not in tsv_data:
-        logger.info('No "status" column found in input file. Creating.')
+        logger.info(
+            'No "status" column found in input file.'
+            'Creating it with default value "good".'
+        )
         tsv_data["status"] = ["good"] * len(tsv_data["name"])
 
-    if "status_description" not in tsv_data:
-        logger.info('No "status_description" column found in input file. Creating.')
+    if "status_description" not in tsv_data and write_descriptions:
+        logger.info(
+            'No "status_description" column found in input file. '
+            'Creating it with default value "n/a".'
+        )
         tsv_data["status_description"] = ["n/a"] * len(tsv_data["name"])
 
     # Now actually mark the user-requested channels as bad.
@@ -2570,7 +2585,7 @@ def mark_channels(bids_path, *, ch_names, status, descriptions=None, verbose=Non
         tsv_data["status"][idx] = status_
 
         # only write if the description was passed in
-        if description is not None:
+        if description:
             tsv_data["status_description"][idx] = description
 
     _write_tsv(channels_fname, tsv_data, overwrite=True)