Read a blank dat_path as no raw data file (#61)

A dataset with no raw data file on disk has a blank dat_path in params.py.
SpikeInterface writes one for recordings with an unknown source, such as
simulated recordings.

Path('') is Path('.'), so a blank entry was resolved against dir_path and
silently became the dataset directory itself. The model then reported the
directory as its raw data file, and get_ephys_reader() warned twice about an
unknown file extension before giving up.

Drop blank entries instead, so dat_path ends up empty, which is the path the
model already supports for datasets with no raw data.

Author: adityasingh2400

CHANGELOG.md

@@ -14,6 +14,8 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - Development checkouts now report a development version with a Git commit suffix.
 - Template datasets without waveform templates can be reloaded after cluster assignments change,
   and stored spike-waveform subsets provide their waveform sample count.
+- #57 a blank `dat_path` in `params.py` is now read as "no raw data file" instead of resolving to
+  the dataset directory.
 
 ## [2.7.0] 2025-12-10
 

phylib/io/model.py

@@ -322,13 +322,10 @@ def __init__(self, **kwargs):
         assert isinstance(self.dir_path, Path)
         assert self.dir_path.exists()
 
-        # Set dat_path.
-        if not self.dat_path:  # pragma: no cover
-            self.dat_path = []
-        elif not isinstance(self.dat_path, (list, tuple)):
-            self.dat_path = [self.dat_path]
-        assert isinstance(self.dat_path, (list, tuple))
-        self.dat_path = [Path(p).resolve() if not Path(p).is_symlink() else p for p in self.dat_path]
+        # Set dat_path. Blank entries mean "no raw data file", see _clean_dat_path().
+        self.dat_path = [
+            Path(p).resolve() if not Path(p).is_symlink() else p
+            for p in _clean_dat_path(self.dat_path)]
 
         self.dtype = getattr(self, 'dtype', np.int16)
         if not self.sample_rate:  # pragma: no cover
@@ -1416,6 +1413,23 @@ def close(self):
             _close_memmap(k, v)
 
 
+def _clean_dat_path(dat_path):
+    """Normalize a raw data path specification into a list of non-empty paths.
+
+    A blank `dat_path` means the dataset has no raw data file at all. Some sorters write
+    `dat_path = ''` in `params.py` when the recording has no file on disk, for instance
+    SpikeInterface with a simulated or in-memory recording. Such entries have to be dropped
+    rather than resolved, because `Path('')` is `Path('.')`, which would silently turn the
+    missing raw data file into the dataset directory itself.
+
+    """
+    if not dat_path:
+        return []
+    if not isinstance(dat_path, (list, tuple)):
+        dat_path = [dat_path]
+    return [p for p in dat_path if str(p).strip()]
+
+
 def _make_abs_path(p, dir_path):
     p = Path(p)
     if not op.isabs(p):
@@ -1438,9 +1452,8 @@ def get_template_params(params_path):
     assert params['dir_path'].is_dir()
     assert params['dir_path'].exists()
 
-    if isinstance(params['dat_path'], str):
-        params['dat_path'] = [params['dat_path']]
-    params['dat_path'] = [_make_abs_path(_, params['dir_path']) for _ in params['dat_path']]
+    params['dat_path'] = [
+        _make_abs_path(_, params['dir_path']) for _ in _clean_dat_path(params['dat_path'])]
     return params
 
 

phylib/io/tests/test_model.py

@@ -14,7 +14,7 @@
 
 # from phylib.utils import Bunch
 from phylib.utils.testing import captured_output
-from ..model import from_sparse, load_model
+from ..model import from_sparse, get_template_params, load_model
 
 logger = logging.getLogger(__name__)
 
@@ -135,6 +135,46 @@ def test_model_save(template_model_full):
     m.save_spike_clusters(m.spike_clusters)
 
 
+def _set_dat_path(params_path, value):
+    """Replace the `dat_path` assignment of a `params.py` file by a literal value."""
+    lines = params_path.read_text().splitlines()
+    lines = [
+        'dat_path = %s' % value if line.startswith('dat_path') else line for line in lines]
+    params_path.write_text('\n'.join(lines) + '\n')
+
+
+def test_model_blank_dat_path(template_path):
+    # Regression test for #57: a dataset with no raw data file on disk has a blank `dat_path`
+    # in params.py. `Path('')` is `Path('.')`, so a blank entry used to resolve to the dataset
+    # directory instead of being dropped.
+    _set_dat_path(template_path, "''")
+    assert get_template_params(template_path)['dat_path'] == []
+
+    model = load_model(template_path)
+    try:
+        assert model.dat_path == []
+        assert model.traces is None
+        with captured_output() as (stdout, stderr):
+            model.describe()
+        assert stdout.getvalue().splitlines()[0].strip() == 'Data files'
+    finally:
+        model.close()
+
+
+def test_model_blank_dat_path_in_list(template_path):
+    # Blank entries are dropped from a list too, and the real file is kept.
+    _set_dat_path(template_path, "['', 'sim_binary.dat']")
+    dat_path = get_template_params(template_path)['dat_path']
+    assert [p.name for p in dat_path] == ['sim_binary.dat']
+
+    model = load_model(template_path)
+    try:
+        assert [p.name for p in model.dat_path] == ['sim_binary.dat']
+        assert model.traces is not None
+    finally:
+        model.close()
+
+
 def test_model_reload_curated_dataset_without_templates(template_path):
     model = load_model(template_path)
     model.save_spikes_subset_waveforms(10, 16)