Improve read_raw file extension detection (mne-tools#11521)

Co-authored-by: Eric Larson <[email protected]>
Co-authored-by: Daniel McCloy <[email protected]>

Author: 3 people

doc/changes/latest.inc

@@ -54,6 +54,7 @@ Bugs
 - All functions accepting paths can now correctly handle :class:`~pathlib.Path` as input. Historically, we expected strings (instead of "proper" path objects), and only added :class:`~pathlib.Path` support in a few select places, leading to inconsistent behavior. (:gh:`11473` and :gh:`11499` by `Mathieu Scheltienne`_)
 - Fix visualization dialog compatibility with matplotlib 3.7 (:gh:`11409` by `Daniel McCloy`_ and `Eric Larson`_)
 - Expand tilde (user directory) in config keys (:gh:`11537` by `Clemens Brunner`_)
+- Fix :func:`mne.io.read_raw` for file names containing multiple dots (:gh:`11521` by `Clemens Brunner`_)
 
 API changes
 ~~~~~~~~~~~

mne/io/_read_raw.py

@@ -12,7 +12,8 @@
                read_raw_fif, read_raw_eeglab, read_raw_cnt, read_raw_egi,
                read_raw_eximia, read_raw_nirx, read_raw_fieldtrip,
                read_raw_artemis123, read_raw_nicolet, read_raw_kit,
-               read_raw_ctf, read_raw_boxy, read_raw_snirf, read_raw_fil)
+               read_raw_ctf, read_raw_boxy, read_raw_snirf, read_raw_fil,
+               read_raw_nihon)
 from ..utils import fill_doc
 
 
@@ -29,6 +30,7 @@ def _read_unsupported(fname, **kwargs):
 # supported read file formats
 supported = {
     ".edf": dict(EDF=read_raw_edf),
+    ".eeg": dict(NihonKoden=read_raw_nihon),
     ".bdf": dict(BDF=read_raw_bdf),
     ".gdf": dict(GDF=read_raw_gdf),
     ".vhdr": dict(brainvision=read_raw_brainvision),
@@ -57,13 +59,23 @@ def _read_unsupported(fname, **kwargs):
 suggested = {
     ".vmrk": dict(brainvision=partial(_read_unsupported, suggest=".vhdr")),
     ".amrk": dict(brainvision=partial(_read_unsupported, suggest=".ahdr")),
-    ".eeg": dict(brainvision=partial(_read_unsupported, suggest=".vhdr")),
 }
 
 # all known file formats
 readers = {**supported, **suggested}
 
 
+def split_name_ext(fname):
+    """Return name and supported file extension."""
+    maxsuffixes = max(ext.count(".") for ext in supported)
+    suffixes = Path(fname).suffixes
+    for si in range(-maxsuffixes, 0):
+        ext = "".join(suffixes[si:]).lower()
+        if ext in readers:
+            return Path(fname).name[:-len(ext)], ext
+    return fname, None  # unknown file extension
+
+
 @fill_doc
 def read_raw(fname, *, preload=False, verbose=None, **kwargs):
     """Read raw file.
@@ -99,7 +111,7 @@ def read_raw(fname, *, preload=False, verbose=None, **kwargs):
     raw : mne.io.Raw
         Raw object.
     """
-    ext = "".join(Path(fname).suffixes)
+    _, ext = split_name_ext(fname)
     kwargs['verbose'] = verbose
     kwargs['preload'] = preload
     if ext not in readers:

mne/io/tests/test_read_raw.py

@@ -5,11 +5,13 @@
 # License: BSD-3-Clause
 
 from pathlib import Path
+from shutil import copyfile
 
 import pytest
 
-from mne.io import read_raw
 from mne.datasets import testing
+from mne.io import read_raw
+from mne.io._read_raw import split_name_ext, readers
 
 
 base = Path(__file__).parent.parent
@@ -32,7 +34,7 @@ def test_read_raw_unsupported_multi(fname, tmp_path):
         read_raw(fname)
 
 
-@pytest.mark.parametrize('fname', ['x.vmrk', 'x.eeg'])
+@pytest.mark.parametrize('fname', ['x.vmrk', 'y.amrk'])
 def test_read_raw_suggested(fname):
     """Test handling of unsupported file types with suggested alternatives."""
     with pytest.raises(ValueError, match='Try reading'):
@@ -43,6 +45,8 @@ def test_read_raw_suggested(fname):
 
 
 @pytest.mark.parametrize('fname', [
+    base / 'tests/data/test_raw.fif',
+    base / 'tests/data/test_raw.fif.gz',
     base / 'edf/tests/data/test.edf',
     base / 'edf/tests/data/test.bdf',
     base / 'brainvision/tests/data/test.vhdr',
@@ -66,3 +70,22 @@ def test_read_raw_supported(fname):
     read_raw(fname, verbose=False)
     raw = read_raw(fname, preload=True)
     assert "data loaded" in str(raw)
+
+
+def test_split_name_ext():
+    """Test file name extension splitting."""
+    # test known extensions
+    for ext in readers:
+        assert split_name_ext(f"test{ext}")[1] == ext
+
+    # test unsupported extensions
+    for ext in ("this.is.not.supported", "a.b.c.d.e", "fif.gz.xyz"):
+        assert split_name_ext(f"test{ext}")[1] is None
+
+
+def test_read_raw_multiple_dots(tmp_path):
+    """Test if file names with multiple dots work correctly."""
+    src = base / 'edf/tests/data/test.edf'
+    dst = tmp_path / "test.this.file.edf"
+    copyfile(src, dst)
+    read_raw(dst)