Merge pull request #1087 from cookpa/traverse_link_dirs

BUG: symlink dirs treated as files during indexing

bids/layout/index.py

@@ -207,6 +207,21 @@ def _index_dir(self, path, config, force=None):
         # Get lists of 1st-level subdirectories and files in the path directory
         _, dirnames, filenames = next(path.fs.walk(path.path))
 
+        # Symbolic links are returned as filenames even if they point to directories
+        # Temporary list to store symbolic links that need to be removed from filenames
+        links_to_dirs = []
+
+        # Find directories in filenames
+        for possible_link in filenames:
+            full_path = path / possible_link
+            if full_path.is_dir():
+                links_to_dirs.append(possible_link)
+
+        dirnames.extend(links_to_dirs)
+
+        for link in links_to_dirs:
+            filenames.remove(link)
+
         # Move any directories suffixed with .zarr to filenames
         zarrnames = [entry for entry in dirnames if Path(entry).suffix == '.zarr']
         dirnames = [entry for entry in dirnames if Path(entry).suffix != '.zarr']
@@ -332,7 +347,7 @@ def load_json(path):
 
                 to_store = (file_ents, payload, bf.path)
                 file_data[key][bf._dirname].append(to_store)
-                
+
         # To avoid integrity errors, track primary keys we've seen
         seen_assocs = set()
 
@@ -494,7 +509,7 @@ def create_association_pair(src, dst, kind, kind2=None):
                     self.session.add(all_entities[md_key])
                 tag = _create_tag_dict(bf, all_entities[md_key], md_val, is_metadata=True)
                 all_tag_dicts.append(tag)
-                
+
         self.session.bulk_save_objects(all_objs)
         self.session.bulk_insert_mappings(Tag, all_tag_dicts)
         self.session.commit()

bids/layout/tests/test_layout.py

@@ -1140,3 +1140,22 @@ def test_indexer_patterns(fname):
         [re.compile(r"/\.")],
         root=root,
     ) is (".datalad" in fname)
+
+
+def test_symlinks_in_path(tmp_path):
+
+    src_ds = os.path.abspath(join(get_test_data_path(), '7t_trt'))
+    src_sub = join(src_ds, 'sub-04')
+
+    # This will be a normal directory with symlink contents
+    ds_with_links = Path(tmp_path / '7t_trt').absolute()
+
+    os.makedirs(ds_with_links, exist_ok=False)
+
+    link_ds_description = ds_with_links / 'dataset_description.json'
+    link_sub = ds_with_links / 'sub-04'
+
+    os.symlink(join(src_ds, 'dataset_description.json'), link_ds_description)
+    os.symlink(src_sub, link_sub)
+
+    assert "Subjects: 1 | Sessions: 2 | Runs: 2" in str(BIDSLayout(tmp_path / "7t_trt"))