Do not overwrite facets from recipe with CMOR table facets for derive…

…d variables (#2255)

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

esmvalcore/_recipe/to_datasets.py

@@ -483,8 +483,12 @@ def _get_input_datasets(dataset: Dataset) -> list[Dataset]:
     # idea: add option to specify facets in list of dicts that is value of
     # 'derive' in the recipe and use that instead of get_required?
     for input_facets in required_vars:
-        input_dataset = dataset.copy(**input_facets)
-        _update_cmor_facets(input_dataset.facets, override=True)
+        input_dataset = dataset.copy()
+        keep = {'alias', 'recipe_dataset_index', *dataset.minimal_facets}
+        input_dataset.facets = {
+            k: v for k, v in input_dataset.facets.items() if k in keep
+        }
+        input_dataset.facets.update(input_facets)
         input_dataset.augment_facets()
         _fix_cmip5_fx_ensemble(input_dataset)
         if input_facets.get('optional') and not input_dataset.files:

esmvalcore/cmor/table.py

@@ -36,7 +36,7 @@
 )
 
 
-def _update_cmor_facets(facets, override=False):
+def _update_cmor_facets(facets):
     """Update `facets` with information from CMOR table."""
     project = facets['project']
     mip = facets['mip']
@@ -53,7 +53,7 @@ def _update_cmor_facets(facets, override=False):
             f"'{short_name}' with mip '{mip}'")
     facets['original_short_name'] = table_entry.short_name
     for key in _CMOR_KEYS:
-        if key not in facets or override:
+        if key not in facets:
             value = getattr(table_entry, key, None)
             if value is not None:
                 facets[key] = value

tests/unit/recipe/test_to_datasets.py

@@ -277,6 +277,30 @@ def test_merge_supplementaries_missing_short_name_fails(session):
         Dataset.from_recipe(recipe_txt, session)
 
 
+def test_get_input_datasets_derive(session):
+    dataset = Dataset(
+        dataset='ERA5',
+        project='native6',
+        mip='E1hr',
+        short_name='rlus',
+        alias='ERA5',
+        derive=True,
+        force_derivation=True,
+        frequency='1hr',
+        recipe_dataset_index=0,
+        tier='3',
+        type='reanaly',
+        version='v1',
+    )
+    rlds, rlns = to_datasets._get_input_datasets(dataset)
+    assert rlds['short_name'] == 'rlds'
+    assert rlds['long_name'] == 'Surface Downwelling Longwave Radiation'
+    assert rlds['frequency'] == '1hr'
+    assert rlns['short_name'] == 'rlns'
+    assert rlns['long_name'] == 'Surface Net downward Longwave Radiation'
+    assert rlns['frequency'] == '1hr'
+
+
 def test_max_years(session):
     recipe_txt = textwrap.dedent("""
     diagnostics:
@@ -300,6 +324,7 @@ def test_max_years(session):
 
 @pytest.mark.parametrize('found_files', [True, False])
 def test_dataset_from_files_fails(monkeypatch, found_files):
+
     def from_files(_):
         file = LocalFile('/path/to/file')
         file.facets = {'facets1': 'value1'}
@@ -323,6 +348,7 @@ def from_files(_):
 
 
 def test_fix_cmip5_fx_ensemble(monkeypatch):
+
     def find_files(self):
         if self.facets['ensemble'] == 'r0i0p0':
             self._files = ['file1.nc']
@@ -343,6 +369,7 @@ def find_files(self):
 
 
 def test_get_supplementary_short_names(monkeypatch):
+
     def _update_cmor_facets(facets):
         facets['modeling_realm'] = 'atmos'