Add parsing of data-validation yaml files (#366)

Co-authored-by: Philip Hackstock <[email protected]>

nomenclature/processor/data_validator.py

@@ -0,0 +1,43 @@
+from pathlib import Path
+from typing import List, Union
+
+import yaml
+
+from nomenclature import DataStructureDefinition
+from nomenclature.error import ErrorCollector
+from nomenclature.processor.iamc import IamcDataFilter
+from nomenclature.processor import Processor
+from nomenclature.processor.utils import get_relative_path
+
+
+class DataValidationCriteria(IamcDataFilter):
+    """Data validation criteria"""
+
+    upper_bound: float = None
+    lower_bound: float = None
+
+
+class DataValidator(Processor):
+    """Processor for validating IAMC datapoints"""
+
+    criteria_items: List[DataValidationCriteria]
+    file: Path
+
+    @classmethod
+    def from_file(cls, file: Union[Path, str]) -> "DataValidator":
+        with open(file, "r") as f:
+            content = yaml.safe_load(f)
+        return cls(file=file, criteria_items=content)
+
+    def apply(self):
+        pass
+
+    def validate_with_definition(self, dsd: DataStructureDefinition) -> None:
+        errors = ErrorCollector()
+        for data in self.criteria_items:
+            try:
+                data.validate_with_definition(dsd)
+            except ValueError as value_error:
+                errors.append(value_error)
+        if errors:
+            raise ValueError(f"In file {get_relative_path(self.file)}:\n{errors}")

nomenclature/processor/iamc.py

@@ -1,4 +1,4 @@
-from typing import List
+from typing import List, Tuple, Any
 from pydantic import BaseModel, field_validator
 
 from pyam import IAMC_IDX
@@ -14,7 +14,7 @@ class IamcDataFilter(BaseModel):
     unit: List[str] | None = None
     year: List[int] | None = None
 
-    @field_validator("*", mode="before")
+    @field_validator(*IAMC_IDX + ["year"], mode="before")
     @classmethod
     def single_input_to_list(cls, v):
         return v if isinstance(v, list) else [v]
@@ -24,12 +24,13 @@ def validate_with_definition(self, dsd: DataStructureDefinition) -> None:
 
         # check for filter-items that are not defined in the codelists
         for dimension in IAMC_IDX:
-            if codelist := getattr(dsd, dimension, None) is None:
+            codelist = getattr(dsd, dimension, None)
+            if codelist is None:
                 continue
-            if invalid := codelist.validate_items(getattr(self, dimension) or []):
+            if invalid := codelist.validate_items(getattr(self, dimension, [])):
                 error_msg += (
-                    f"The following {dimension}s were not found in the "
-                    f"DataStructureDefinition:\n{invalid}\n"
+                    f"The following {dimension}s are not defined in the "
+                    f"DataStructureDefinition:\n   {', '.join(invalid)}\n"
                 )
 
         if error_msg:

tests/data/validation/definition/region/region.yaml

@@ -0,0 +1,2 @@
+- Common:
+  - World

tests/data/validation/definition/variable/variable.yaml

@@ -0,0 +1,8 @@
+- Final Energy:
+    unit: EJ/yr
+- Primary Energy:
+    unit: EJ/yr
+- Emissions|CO2:
+    unit: Mt CO2/yr
+- Emissions|CH4:
+    unit: Mt CH4/yr

tests/data/validation/validate_data/fail_unknown_region.yaml

@@ -0,0 +1,3 @@
+ - variable: Final Energy
+   region: Asia
+

tests/data/validation/validate_data/simple_validation.yaml

@@ -0,0 +1,6 @@
+ - region: World
+   variable: Final Energy
+   year: 2010
+   upper_bound: 2.5
+   lower_bound: 1
+

tests/data/validation/validate_data/validation_unknown_region.yaml

@@ -0,0 +1,5 @@
+ - region: Asia
+   variable: Final Energy
+   year: 2010
+   upper_bound: 2.5
+   lower_bound: 1

tests/data/validation/validate_data/validation_unknown_variable.yaml

@@ -0,0 +1,5 @@
+ - region: World
+   variable: Final Energy|Industry
+   year: 2010
+   upper_bound: 2.5
+   lower_bound: 1

tests/test_data_validation.py

@@ -0,0 +1,59 @@
+import pytest
+from conftest import TEST_DATA_DIR
+
+from nomenclature import DataStructureDefinition
+from nomenclature.processor.data_validator import DataValidator
+
+DATA_VALIDATION_TEST_DIR = TEST_DATA_DIR / "validation" / "validate_data"
+
+
+def test_DataValidator_from_file():
+    exp = DataValidator(
+        **{
+            "criteria_items": [
+                {
+                    "region": ["World"],
+                    "variable": "Final Energy",
+                    "year": [2010],
+                    "upper_bound": 2.5,
+                    "lower_bound": 1.0,  # test that integer in yaml is cast to float
+                }
+            ],
+            "file": DATA_VALIDATION_TEST_DIR / "simple_validation.yaml",
+        }
+    )
+    obs = DataValidator.from_file(DATA_VALIDATION_TEST_DIR / "simple_validation.yaml")
+    assert obs == exp
+
+    dsd = DataStructureDefinition(TEST_DATA_DIR / "validation" / "definition")
+    assert obs.validate_with_definition(dsd) is None
+
+
+@pytest.mark.parametrize(
+    "dimension, match",
+    [
+        ("region", r"regions.*not defined.*\n.*Asia"),
+        ("variable", r"variables.*not defined.*\n.*Final Energy\|Industry"),
+    ],
+)
+def test_DataValidator_validate_with_definition_raises(dimension, match):
+    # Testing two different failure cases
+    # 1. Undefined region
+    # 2. Undefined variable
+    # TODO Undefined unit
+
+    data_validator = DataValidator.from_file(
+        DATA_VALIDATION_TEST_DIR / f"validation_unknown_{dimension}.yaml"
+    )
+
+    # validating against a DataStructure with all dimensions raises
+    dsd = DataStructureDefinition(TEST_DATA_DIR / "validation" / "definition")
+    with pytest.raises(ValueError, match=match):
+        data_validator.validate_with_definition(dsd)
+
+    # validating against a DataStructure without the offending dimension passes
+    dsd = DataStructureDefinition(
+        TEST_DATA_DIR / "validation" / "definition",
+        dimensions=[dim for dim in ["region", "variable"] if dim != dimension],
+    )
+    assert data_validator.validate_with_definition(dsd) is None