Check if constituents of common regions are in native regions (IAMcon…

…sortium#463)

* Check if constituents of common regions are in native regions

* Update hash to fix tests

* Fix error message

* Remove redundant check

* Check excluded regions

---------

Co-authored-by: David Almeida <almeida@iiasa.ac.at>

nomenclature/error.py

@@ -40,6 +40,10 @@
         "region_not_defined",
         "Region(s)\n{regions}\nin {file}\nnot found in RegionCodeList",
     ),
+    "ConstituentsNotNativeError": (
+        "constituents_not_native",
+        "Constituent region(s)\n{regions}\nin {file} not found in native region(s)",
+    ),
 }
 
 PydanticCustomErrors = namedtuple("PydanticCustomErrors", pydantic_custom_error_config)

nomenclature/processor/region.py

@@ -220,6 +220,21 @@ def check_exclude_common_region_overlap(
     ) -> "RegionAggregationMapping":
         return _check_exclude_region_overlap(v, "common_regions")
 
+    @model_validator(mode="after")
+    @classmethod
+    def check_constituent_regions_in_native_regions(
+        cls, v: "RegionAggregationMapping"
+    ) -> "RegionAggregationMapping":
+        if v.common_regions and v.native_regions:
+            if missing := set(
+                [cr for r in v.common_regions for cr in r.constituent_regions]
+            ).difference([r.name for r in v.native_regions] + v.exclude_regions):
+                raise PydanticCustomError(
+                    *custom_pydantic_errors.ConstituentsNotNativeError,
+                    {"regions": missing, "file": v.file},
+                )
+        return v
+
     @classmethod
     def from_file(cls, file: Path | str) -> "RegionAggregationMapping":
         """Initialize a RegionAggregationMapping from a file.

tests/data/cli/structure_validation_fails/mappings/mapping_2.yaml

@@ -4,4 +4,3 @@ native_regions:
 common_regions:
   - World:
     - region_a
-    - region_b

tests/data/region_processing/external_repo_test/nomenclature.yaml

@@ -4,7 +4,7 @@ dimensions:
 repositories:
   common-definitions:
     url: https://github.com/IAMconsortium/common-definitions.git/
-    hash: cb85704
+    hash: 091c0fe
 definitions:
   region:
     repository: common-definitions

tests/data/region_processing/external_repo_test_missing_region/nomenclature.yaml

@@ -4,7 +4,7 @@ dimensions:
 repositories:
   common-definitions:
     url: https://github.com/IAMconsortium/common-definitions.git/
-    hash: cb85704
+    hash: 091c0fe
 definitions:
   region:
     repository:

tests/data/region_processing/region_aggregation/illegal_mapping_constituent_native_missing.yaml

@@ -0,0 +1,10 @@
+# Constituent region is not defined in native region
+model: model_a
+native_regions:
+  - region_a: alternative_name_a
+  - region_b: alternative_name_b
+common_regions:
+  - common_region_1:
+    - region_a
+    - region_b
+    - region_c

tests/test_region_aggregation.py

@@ -79,6 +79,10 @@ def test_mapping():
             "illegal_mapping_model_only.yaml",
             "one of 'native_regions' and 'common_regions'",
         ),
+        (
+            "illegal_mapping_constituent_native_missing.yaml",
+            "Constituent region\(s\)\n.*\n",
+        ),
     ],
 )
 def test_illegal_mappings(file, error_msg_pattern):
@@ -196,7 +200,7 @@ def test_region_processor_wrong_args():
 
 def test_region_processor_multiple_wrong_mappings(simple_definition):
     # Read in the entire region_aggregation directory and return **all** errors
-    msg = "Collected 9 errors"
+    msg = "Collected 10 errors"
 
     with pytest.raises(ValueError, match=msg):
         RegionProcessor.from_directory(