[MRG] FIX: Support commas in BrainVision channel names, event types, and event descriptions

[sappelhoff]

This came up in bids-standard/pybv#53

In the BrainVision format channel names, event types, and event descriptions that contain a comma , are encoded such that the comma is written as \1.

I add a test here to show that MNE-Python does currently not adequately replace the \1 characters to , during reading.

This is potentially broken for event type and event description as well, but I haven't looked into that yet.

Should we fix it?

[larsoner]

Should we fix it?

Yes I think fixing the channel names and the event type and descriptions would be good

[sappelhoff]

Would this PR be eligible for a backport? It may come in handy for the pybv package.

[larsoner]

Sure but we just cut 0.21.1 yesterday so it might be a few weeks or so until we do a 0.21.2

[sappelhoff]

indeed, on Windows, the temporary file-replacement for testing is not working as expected:

The windows tests only fail with:

mne\io\brainvision\tests\test_brainvision.py:304: in test_ch_names_comma
    assert nperformed_replacements == len(replace_dict)
E   assert 1 == 2
E     +1
E     -2

[sappelhoff]

@cbrnr could you please have a look at this one and merge if you are happy?

Then I can switch on the corresponding test on pybv with @requires_version("mne", "0.22")

[cbrnr]

I'll merge once CIs come back green.

[cbrnr]

Not sure what's up with the failing test. It stopped after mne/tests/test_report.py::test_render_add_sections[mayavi] PASSED [ 75%], but it doesn't say which test failed and why.

[sappelhoff]

I don't know either. 🤷‍♂️ but it passed before and it passes locally

[cbrnr]

I'll re-run and see if this magically resolves the issue (it often does).

[cbrnr]

Now Travis fails at the same test (mne/tests/test_report.py::test_render_add_sections[pyvista] Traceback (most recent call last):)? I'd like to clarify this before merging. @larsoner do you have any idea?

[GuillaumeFavelier]

Now Travis fails at the same test

It's related to #8494

I'm working on a fix.

[sappelhoff]

@cbrnr I rebased on Guillaume's fix and CIs pass now :-)

[cbrnr]

Thanks @sappelhoff!