Avoid a crash if dataset has supplementary variables (#2198)

Co-authored-by: Valeriu Predoi <[email protected]>

esmvalcore/_recipe/from_datasets.py

@@ -160,6 +160,23 @@ def _group_identical_facets(variable: Mapping[str, Any]) -> Recipe:
     return result
 
 
+class _SortableDict(dict):
+    """A `dict` class that can be sorted."""
+
+    def __lt__(self, other):
+        return tuple(self.items()) < tuple(other.items())
+
+
+def _change_dict_type(item, dict_type):
+    """Change the dict type in a nested structure."""
+    change_dict_type = partial(_change_dict_type, dict_type=dict_type)
+    if isinstance(item, dict):
+        return dict_type((k, change_dict_type(v)) for k, v in item.items())
+    if isinstance(item, (list, tuple, set)):
+        return type(item)(change_dict_type(elem) for elem in item)
+    return item
+
+
 def _group_ensemble_members(dataset_facets: Iterable[Facets]) -> list[Facets]:
     """Group ensemble members.
 
@@ -171,6 +188,7 @@ def grouper(facets):
         return tuple((k, facets[k]) for k in sorted(facets) if k != 'ensemble')
 
     result = []
+    dataset_facets = _change_dict_type(dataset_facets, _SortableDict)
     dataset_facets = sorted(dataset_facets, key=grouper)
     for group_facets, group in itertools.groupby(dataset_facets, key=grouper):
         ensembles = [f['ensemble'] for f in group if 'ensemble' in f]
@@ -181,6 +199,7 @@ def grouper(facets):
                 facets = dict(group_facets)
                 facets['ensemble'] = ensemble
                 result.append(facets)
+    result = _change_dict_type(result, dict)
     return result
 
 

tests/unit/recipe/test_from_datasets.py

@@ -8,6 +8,7 @@
     _group_ensemble_names,
     _group_identical_facets,
     _move_one_level_up,
+    _SortableDict,
     _to_frozen,
     datasets_to_recipe,
 )
@@ -136,6 +137,12 @@ def test_supplementary_datasets_to_recipe():
     assert datasets_to_recipe([dataset]) == recipe
 
 
+def test_sortable_dict():
+    assert _SortableDict({'a': 1}) < _SortableDict({'a': 2})
+    assert _SortableDict({'a': 1}) < _SortableDict({'a': 1, 'b': 1})
+    assert _SortableDict({'a': 1}) < _SortableDict({'b': 1})
+
+
 def test_datasets_to_recipe_group_ensembles():
     datasets = [
         Dataset(