Gracefully handle the case of missing facets required by esmvalcore.io.local.LocalDataSource

esmvalcore/io/local.py

@@ -58,7 +58,6 @@
 from netCDF4 import Dataset
 
 import esmvalcore.io.protocol
-from esmvalcore.exceptions import RecipeError
 from esmvalcore.iris_helpers import ignore_warnings_context
 
 if TYPE_CHECKING:
@@ -421,6 +420,30 @@ def _select_files(
     return selection
 
 
+class _MissingFacetError(KeyError):
+    """Error raised when a facet required for filling the template is missing."""
+
+
+def _format_iterable(iterable: Iterable[Any]) -> str:
+    """Format an iterable as a string for use in messages.
+
+    Parameters
+    ----------
+    iterable:
+        The iterable to format.
+
+    Returns
+    -------
+    :
+        The formatted string.
+    """
+    items = [f"'{item}'" for item in sorted(iterable)]
+    if len(items) > 1:
+        items[-1] = f"and {items[-1]}"
+    txt = " ".join(items) if len(items) == 2 else ", ".join(items)
+    return f"s {txt}" if len(items) > 1 else f" {txt}"
+
+
 def _replace_tags(
     paths: str | list[str],
     variable: Facets,
@@ -446,6 +469,7 @@ def _replace_tags(
             tlist.add("sub_experiment")
         pathset = new_paths
 
+    failed = set()
     for original_tag in tlist:
         tag, _, _ = _get_caps_options(original_tag)
 
@@ -454,12 +478,17 @@ def _replace_tags(
         elif tag == "version":
             replacewith = "*"
         else:
-            msg = (
-                f"Dataset key '{tag}' must be specified for {variable}, check "
-                f"your recipe entry and/or extra facet file(s)"
-            )
-            raise RecipeError(msg)
+            failed.add(tag)
+            continue
         pathset = _replace_tag(pathset, original_tag, replacewith)
+    if failed:
+        msg = (
+            f"Unable to complete path{_format_iterable(pathset)} because "
+            f"the facet{_format_iterable(failed)}"
+            + (" has" if len(failed) == 1 else " have")
+            + " not been specified."
+        )
+        raise _MissingFacetError(msg)
     return [Path(p) for p in pathset]
 
 
@@ -566,7 +595,11 @@ def find_data(self, **facets: FacetValue) -> list[LocalFile]:
         if "original_short_name" in facets:
             facets["short_name"] = facets["original_short_name"]
 
-        globs = self._get_glob_patterns(**facets)
+        try:
+            globs = self._get_glob_patterns(**facets)
+        except _MissingFacetError as exc:
+            self.debug_info = exc.args[0]
+            return []
         self.debug_info = "No files found matching glob pattern " + "\n".join(
             str(g) for g in globs
         )

esmvalcore/local.py

@@ -21,10 +21,12 @@
     get_project_config,
     load_config_developer,
 )
+from esmvalcore.exceptions import RecipeError
 from esmvalcore.io.local import (
     LocalDataSource,
     LocalFile,
     _filter_versions_called_latest,
+    _MissingFacetError,
     _replace_tags,
     _select_latest_version,
 )
@@ -300,7 +302,10 @@ def _get_output_file(variable: dict[str, Any], preproc_dir: Path) -> Path:
     if isinstance(variable.get("exp"), (list, tuple)):
         variable = dict(variable)
         variable["exp"] = "-".join(variable["exp"])
-    outfile = _replace_tags(cfg["output_file"], variable)[0]
+    try:
+        outfile = _replace_tags(cfg["output_file"], variable)[0]
+    except _MissingFacetError as exc:
+        raise RecipeError(exc.args[0]) from exc
     if "timerange" in variable:
         timerange = variable["timerange"].replace("/", "-")
         outfile = Path(f"{outfile}_{timerange}")

tests/integration/io/test_local.py

@@ -14,6 +14,7 @@
 import esmvalcore.config._config
 import esmvalcore.local
 from esmvalcore.config import CFG
+from esmvalcore.exceptions import RecipeError
 from esmvalcore.io.local import (
     LocalDataSource,
     LocalFile,
@@ -83,6 +84,30 @@ def test_get_output_file(monkeypatch, cfg):
     assert output_file == expected
 
 
+def test_get_output_file_missing_facets(
+    monkeypatch: pytest.MonkeyPatch,
+) -> None:
+    """Test that a RecipeError is raised if a required facet is missing."""
+    monkeypatch.setattr(esmvalcore.cmor.table, "CMOR_TABLES", {})
+    monkeypatch.setitem(
+        CFG,
+        "config_developer_file",
+        Path(esmvalcore.__path__[0], "config-developer.yml"),
+    )
+    facets = {
+        "project": "CMIP6",
+        "mip": "Amon",
+        "short_name": "tas",
+    }
+    expected_message = (
+        "Unable to complete path 'CMIP6_{dataset}_Amon_{exp}_{ensemble}_tas"
+        "_{grid}' because the facets 'dataset', 'ensemble', 'exp', and 'grid' "
+        "have not been specified."
+    )
+    with pytest.raises(RecipeError, match=expected_message):
+        _get_output_file(facets, Path("/preproc/dir"))
+
+
 @pytest.mark.parametrize("cfg", CONFIG["get_output_file"])
 def test_get_output_file_no_config_developer(monkeypatch, cfg):
     """Test getting output name for preprocessed files."""
@@ -216,6 +241,31 @@ def test_find_data(root, cfg):
         assert str(pattern) in data_source.debug_info
 
 
+def test_find_data_facet_missing() -> None:
+    """Test that a MissingFacetError is raised if a required facet is missing."""
+    data_source = LocalDataSource(
+        name="test-data-source",
+        project="CMIP6",
+        rootpath=Path("/data/cmip6"),
+        priority=1,
+        dirname_template="{dataset}/{exp}/{ensemble}",
+        filename_template="{short_name}.nc",
+    )
+    facets = {
+        "short_name": "tas",
+        "dataset": "test-dataset",
+        "exp": ["historical", "ssp585"],
+    }
+    expected_message = (
+        "Unable to complete paths 'test-dataset/historical/{ensemble}' and "
+        "'test-dataset/ssp585/{ensemble}' because the facet 'ensemble' has "
+        "not been specified."
+    )
+    files = data_source.find_data(**facets)
+    assert not files
+    assert data_source.debug_info == expected_message
+
+
 def test_select_invalid_drs_structure(monkeypatch: pytest.MonkeyPatch) -> None:
     monkeypatch.setattr(esmvalcore.cmor.table, "CMOR_TABLES", {})
     monkeypatch.setitem(

tests/unit/io/local/test_replace_tags.py

@@ -1,11 +1,14 @@
 """Tests for `_replace_tags` in `esmvalcore.io.local`."""
 
+import re
 from pathlib import Path
 
 import pytest
 
-from esmvalcore.exceptions import RecipeError
-from esmvalcore.io.local import _replace_tags
+from esmvalcore.io.local import (
+    _MissingFacetError,
+    _replace_tags,
+)
 
 VARIABLE = {
     "project": "CMIP6",
@@ -58,13 +61,28 @@ def test_replace_tags_with_caps():
 
 
 def test_replace_tags_missing_facet():
-    """Check that a RecipeError is raised if a required facet is missing."""
+    """Check that a MissingFacetError is raised if a required facet is missing."""
     paths = ["{short_name}_{missing}_*.nc"]
     variable = {"short_name": "tas"}
-    with pytest.raises(RecipeError) as exc:
+    expected_message = (
+        "Unable to complete path 'tas_{missing}_*.nc' because the facet "
+        "'missing' has not been specified."
+    )
+    with pytest.raises(_MissingFacetError, match=re.escape(expected_message)):
         _replace_tags(paths, variable)
 
-    assert "Dataset key 'missing' must be specified" in exc.value.message
+
+def test_replace_tags_missing_facets():
+    """Check that a MissingFacetError is raised if multiple facets are missing."""
+    paths = ["{missing1}_{short_name}_{missing2}_{missing3}_*.nc"]
+    variable = {"short_name": "tas"}
+    expected_message = (
+        "Unable to complete path '{missing1}_tas_{missing2}_{missing3}_*.nc' "
+        "because the facets 'missing1', 'missing2', and 'missing3' have not "
+        "been specified."
+    )
+    with pytest.raises(_MissingFacetError, match=re.escape(expected_message)):
+        _replace_tags(paths, variable)
 
 
 def test_replace_tags_list_of_str():