Fixes behavior of stc.save() as mentioned in issue #13158

[shresth-keshari]

Fixes #13158 .

Doing stc.save("test.h5") saves a single test.h5 file, and doing stc.save("test.stc") saves to the two files test-lh.stc and test-rh.stc.

Please let me know if any more changes required in the current iteration.

[drammock]

CI errors look related:

https://github.com/mne-tools/mne-python/actions/runs/13924372063/job/38964972043?pr=13165#step:16:6285

FAILED mne/tests/test_source_estimate.py::test_io_stc - OSError: SourceEstimate File(s) not found for: PosixPath('/private/var/folders/0j/bwqcs4y508s2n4ck4dhf3rpc0000gn/T/pytest-of-runner/pytest-0/test_io_stc0/tmp.stc')

Does mne/tests/test_source_estimate.py::test_io_stc pass for you locally?

[shresth-keshari]

@drammock I have changed ftype from None to auto coz it was fairly easy.
I also put it to test by saving a sample of both .stc as well as .h5 filetypes. You can see them in the screenshot, that they're being saved in the directory, named as it was suggested by @larsoner in the issue. (will be deleting these files before commit)

But, while running tests, it shows the following error:

which says: FAILED test_source_estimate.py::test_io_stc - OSError: SourceEstimate File(s) not found for: PosixPath('/tmp/pytest-of-shresth/pytest-1/test_io_stc0/tmp.stc')

I'm not aware about how to fix this error. Please help.

[drammock]

I had noticed that a couple of days back, added "*" as suggested, but still, when I run pytest, it fails with the same message as mentioned above.

The * is not related to the pytest failures

Well, if I test the function using a sample stc file, it works perfectly.

Our tests often check for "corner cases" that are probably different than the approach you're trying. I suggest running pytest with the --pdb flag so you can examine what the problematic values are when the test fails, and see if that helps you figure out what else needs to be changed.

[shresth-keshari]

Hey, @drammock, I implemented your suggestions, and ran pytest with --pdb flag.
The issue seems to be coming from "read_source_estimate" in source_estimate.py, which looks like this in the terminal:

test_source_estimate.py:484: in test_io_stc
    stc2 = read_source_estimate(tmp_path / "tmp.stc")

..\source_estimate.py:330: in read_source_estimate
    raise OSError(f"SourceEstimate File(s) not found for: {fname_arg!r}")

E   OSError: SourceEstimate File(s) not found for: WindowsPath('C:/Users/kesha/AppData/Local/Temp/pytest-of-kesha/pytest-3/test_io_stc0/tmp.stc')

Now that I look at test_source_estimate.py, in the function definition of test_io_stc, the stc file is saved first, using stc.save() function. Later, the same file is being read using read _source_estimate. Since we changed the naming of the files being saved, I believe that's what causes the issue. Will try fixing that and will get back to you with a solution soon.

have attached screenshots as well, please have a look:

[shresth-keshari]

I was able to successfully run " pytest test_source_estimate.py" with no fails, locally.
I feel the issue has been solved and desired "intuitive naming" of stc files while saving them has been achieved.
Please review the pull request once.

[drammock]

This can be simplified, as the first element in each endswith tuple already captures all cases we care about.

[drammock]

is overwrite=True really needed here? I wouldn't have expected it to be, since it's the first .save action in the test.

[shresth-keshari]

Hello, @drammock and @larsoner
I believe that most of the changes have passed checks, but after numerous pytests, I can't pinpoint where does the issue lie that is stopping the remaining checks from passing.

Please guide me ahead..

[larsoner]

I merged main into your branch to see if it fixes the CIs

[larsoner]

@shresth-keshari still interested in working on this one or would it help if someone else took over?

[larsoner]

I think the changes here aren't quite extensive enough to cover the behaviors mentioned in #13158