Proj id and proj name access

[LaurentLM]

Reference issue

Fixes #13252.

What does this implement/fix?

Allows user to set proj_idand proj_name through direct dict access.
Ensures that proj_id is either an integer or None.
Ensures that proj_name is either a string or None.

Additional information

Example:

import mne
info = mne.create_info(
    ch_names=['EEG 001', 'EEG 002'],
    sfreq=1000,
    ch_types='eeg',
)

info['proj_name'] = None
info['proj_name'] = 123 #TypeError: 123 is not a string
info['proj_name'] = 'test' 

info['proj_id'] = None
info['proj_id'] = 123
info['proj_id'] = '123' #TypeError: '123' is not an int

[welcome]

Hello! 👋 Thanks for opening your first pull request here! ❤️ We will try to get back to you soon. 🚴

[larsoner]

This looks like it's headed in the right direction!

If you look at one CI failure, you'll see:

________________ ERROR collecting mne/stats/tests/test_erp.py _________________
mne\stats\tests\test_erp.py:14: in <module>
    raw = read_raw_fif(base_dir / "test_raw.fif")
mne\io\fiff\raw.py:543: in read_raw_fif
    return Raw(
<decorator-gen-364>:12: in __init__
    ???
mne\io\fiff\raw.py:103: in __init__
    raw, next_fname, buffer_size_sec = self._read_raw_file(
<decorator-gen-365>:12: in _read_raw_file
    ???
mne\io\fiff\raw.py:209: in _read_raw_file
    info, meas = read_meas_info(fid, tree, clean_bads=True)
<decorator-gen-31>:12: in read_meas_info
    ???
mne\_fiff\meas_info.py:2540: in read_meas_info
    info["proj_id"] = proj_id
mne\_fiff\meas_info.py:960: in __setitem__
    val = self._attributes[key](
mne\_fiff\meas_info.py:1008: in _check_types
    _validate_type(x, types, name)
mne\utils\check.py:640: in _validate_type
    raise TypeError(
E   TypeError: proj_id must be an instance of int or None, got <class 'numpy.ndarray'> instead.

I would suggest to modify mne\_fiff\meas_info.py:2540: in read_meas_info to have a .item() to get it to be a float rather than an ndarray with one element.

Any new change would ideally include some changed test, could you modify mne/_fiff/tests/test_meas_info.py somewhere to make sure you can set these two new values? And maybe even a couple with pytest.raises(TypeError, match="...") that you can't set them to something of the wrong type?

Finally, we'll need a doc/changes/devel/13261.newfeature.rst that mentions the ability to set these values. In there you should use the :newcontrib: directive with your name and add a link to doc/changes/names.inc as well.

[LaurentLM]

Hi @larsoner,

Thanks for your guidance, I'm very new to this.

Still a bit confused as to why the test failed. proj_id should be either an int or None. It makes sense to me that an error is returned if someone tries to set proj_id with a numpy.ndarray.
Having said that, I made the change you suggested and it seems to work fine.

I have added tests and contribution information, hopefully it's all in the right place and the right format. Let me know if not :-)

[LaurentLM]

I think I am a bit stuck with this. I fail to understand why info["proj_id"] is converted into an ndarray here:

mne-python/mne/_fiff/meas_info.py

Line 3501 in 68b5523

info["proj_id"] = np.zeros_like(info["proj_id"])

It would make more sense to me to set it as 0 (as an integer), the same way as subject_info["id"] is anonymized.