Make illegal path-like variable names when constructing a DataTree from a Dataset (pydata#9378)

* Make illegal path-like variable names when constructing a DataTree from a Dataset

* Updated whats-new.rst

* PR comments

* Revert diff

* Update xarray/core/datatree.py

Co-authored-by: Tom Nicholas <[email protected]>

* Update xarray/core/datatree.py

Co-authored-by: Tom Nicholas <[email protected]>

* Update xarray/tests/test_datatree.py

Co-authored-by: Tom Nicholas <[email protected]>

* Update expected Exception message in test

* Merge changes from pydata#9476

* Fix

---------

Co-authored-by: Tom Nicholas <[email protected]>

Author: 2 people

doc/whats-new.rst

@@ -100,6 +100,9 @@ Breaking changes
 Bug fixes
 ~~~~~~~~~
 
+- Make illegal path-like variable names when constructing a DataTree from a Dataset
+  (:issue:`9339`, :pull:`9378`)
+  By `Etienne Schalk <https://github.com/etienneschalk>`_.
 - Fix bug with rechunking to a frequency when some periods contain no data (:issue:`9360`).
   By `Deepak Cherian <https://github.com/dcherian>`_.
 - Fix bug causing `DataTree.from_dict` to be sensitive to insertion order (:issue:`9276`, :pull:`9292`).

xarray/core/datatree.py

@@ -156,6 +156,18 @@ def check_alignment(
             check_alignment(child_path, child_ds, base_ds, child.children)
 
 
+def _check_for_slashes_in_names(variables: Iterable[Hashable]) -> None:
+    offending_variable_names = [
+        name for name in variables if isinstance(name, str) and "/" in name
+    ]
+    if len(offending_variable_names) > 0:
+        raise ValueError(
+            "Given variables have names containing the '/' character: "
+            f"{offending_variable_names}. "
+            "Variables stored in DataTree objects cannot have names containing '/' characters, as this would make path-like access to variables ambiguous."
+        )
+
+
 class DatasetView(Dataset):
     """
     An immutable Dataset-like view onto the data in a single DataTree node.
@@ -453,6 +465,7 @@ def __init__(
         super().__init__(name=name, children=children)
 
     def _set_node_data(self, dataset: Dataset):
+        _check_for_slashes_in_names(dataset.variables)
         data_vars, coord_vars = _collect_data_and_coord_variables(dataset)
         self._data_variables = data_vars
         self._node_coord_variables = coord_vars

xarray/tests/test_datatree.py

@@ -83,6 +83,23 @@ def test_child_gets_named_on_attach(self):
         mary = DataTree(children={"Sue": sue})  # noqa
         assert mary.children["Sue"].name == "Sue"
 
+    def test_dataset_containing_slashes(self):
+        xda: xr.DataArray = xr.DataArray(
+            [[1, 2]],
+            coords={"label": ["a"], "R30m/y": [30, 60]},
+        )
+        xds: xr.Dataset = xr.Dataset({"group/subgroup/my_variable": xda})
+        with pytest.raises(
+            ValueError,
+            match=re.escape(
+                "Given variables have names containing the '/' character: "
+                "['R30m/y', 'group/subgroup/my_variable']. "
+                "Variables stored in DataTree objects cannot have names containing '/' characters, "
+                "as this would make path-like access to variables ambiguous."
+            ),
+        ):
+            DataTree(xds)
+
 
 class TestPaths:
     def test_path_property(self):