fix(hns): cache root bucket properly in _get_dirs_and_update_cache

gcsfs/core.py

@@ -1902,29 +1902,31 @@ def _get_dirs_and_update_cache(self, path, objects, prefix="", update_cache=True
 
             while parent:
                 dir_key = self.split_path(parent)[1]
-                if not dir_key or len(parent) < len(path.rstrip("/")):
+                if len(parent) < len(path.rstrip("/")):
                     break
 
                 if prefix and not parent.startswith(full_prefix):
                     # If this parent doesn't match the prefix, neither will its parents.
                     break
 
-                dirs[parent] = {
-                    "Key": dir_key,
-                    "Size": 0,
-                    "name": parent,
-                    "StorageClass": "DIRECTORY",
-                    "type": "directory",
-                    "size": 0,
-                }
+                if dir_key:
+                    dirs[parent] = {
+                        "Key": dir_key,
+                        "Size": 0,
+                        "name": parent,
+                        "StorageClass": "DIRECTORY",
+                        "type": "directory",
+                        "size": 0,
+                    }
 
                 if not prefix and update_cache:
                     listing = cache_entries.setdefault(parent, {})
                     name = previous["name"]
                     if name not in listing:
                         listing[name] = previous
 
-                previous = dirs[parent]
+                if parent in dirs:
+                    previous = dirs[parent]
                 parent = self._parent(parent)
         if not prefix and update_cache:
             cache_entries_list = {k: list(v.values()) for k, v in cache_entries.items()}

gcsfs/tests/integration/test_extended_hns.py

@@ -1339,6 +1339,21 @@ def test_find_updates_dircache_without_prefix(
         }
         assert not empty_dir_listing
 
+    @pytest.mark.parametrize("withdirs_param", [True, False])
+    def test_find_updates_dircache_for_root_bucket(
+        self, gcs_hns, test_structure, withdirs_param
+    ):
+        """Test that find() populates the dircache for the root bucket itself."""
+        root_bucket = TEST_HNS_BUCKET
+        gcs_hns.invalidate_cache()
+        assert not gcs_hns.dircache
+
+        # Run find on the root bucket to populate the cache
+        gcs_hns.find(root_bucket, withdirs=withdirs_param)
+
+        # Verify that the cache is now populated for the root bucket
+        assert root_bucket in gcs_hns.dircache
+
     def test_find_maxdepth_updates_cache(self, gcs_hns, test_structure):
         """Test that find with maxdepth updates cache for deeper objects."""
         base_dir = test_structure["base_dir"]

gcsfs/tests/test_core.py

@@ -1946,10 +1946,10 @@ def test_attrs(gcs):
     assert gcs.getxattr(a, "something") == "not"
 
 
-def test_request_user_project(gcs_factory):
-    gcs = gcs_factory(requester_pays=True, project=TEST_PROJECT)
+def test_request_user_project(gcs, gcs_factory):
+    gcs_inst = gcs_factory(requester_pays=True, project=TEST_PROJECT)
     # test directly against `_call` to inspect the result
-    r = gcs.call(
+    r = gcs_inst.call(
         "GET",
         "b/{}/o",
         TEST_BUCKET,
@@ -1963,11 +1963,11 @@ def test_request_user_project(gcs_factory):
     assert result["userProject"] == [TEST_PROJECT]
 
 
-def test_request_user_project_string(gcs_factory):
-    gcs = gcs_factory(requester_pays=TEST_PROJECT)
-    assert gcs.requester_pays == TEST_PROJECT
+def test_request_user_project_string(gcs, gcs_factory):
+    gcs_inst = gcs_factory(requester_pays=TEST_PROJECT)
+    assert gcs_inst.requester_pays == TEST_PROJECT
     # test directly against `_call` to inspect the result
-    r = gcs.call(
+    r = gcs_inst.call(
         "GET",
         "b/{}/o",
         TEST_BUCKET,
@@ -1981,10 +1981,10 @@ def test_request_user_project_string(gcs_factory):
     assert result["userProject"] == [TEST_PROJECT]
 
 
-def test_request_header(gcs_factory):
-    gcs = gcs_factory(requester_pays=True)
+def test_request_header(gcs, gcs_factory):
+    gcs_inst = gcs_factory(requester_pays=True)
     # test directly against `_call` to inspect the result
-    r = gcs.call(
+    r = gcs_inst.call(
         "GET",
         "b/{}/o",
         TEST_BUCKET,
@@ -2025,7 +2025,7 @@ def test_requester_pays_fails_without_user_project(requester_pays_bucket, gcs_fa
         fs.ls(requester_pays_bucket)
 
 
-def test_fs_requester_pays_on_bucket_without_requester_pays(gcs_factory):
+def test_fs_requester_pays_on_bucket_without_requester_pays(gcs, gcs_factory):
     """Test that metadata and data operations work when fs has requester_pays=True
     but the bucket does not have requester-pays enabled."""
     fs = gcs_factory(requester_pays=True)