Define input_data as Processor property

nomenclature/processor/data_validator.py

@@ -388,3 +388,24 @@ def validate_with_definition(self, dsd: DataStructureDefinition) -> None:
                 f"Error in DataValidator (file {get_relative_path(self.file)})",
                 errors,
             )
+
+    @property
+    def input_data(self):
+        return merge_criteria_items_filter_args(self.criteria_items)
+
+
+def merge_criteria_items_filter_args(list_of_items: list[DataValidationItem]) -> dict:
+    """Concatenate filter-arguments"""
+    merged_dict = {}
+    for dct in list_of_items:
+        for key, value in dct.filter_args.items():
+            if key in merged_dict:
+                merged_dict[key].extend(value)
+            else:
+                merged_dict[key] = value
+
+    # drop duplicates
+    for key, value in merged_dict.items():
+        merged_dict[key] = list(set(merged_dict[key]))
+
+    return merged_dict

nomenclature/processor/processor.py

@@ -6,12 +6,16 @@
 
 
 class Processor(BaseModel, abc.ABC):
-    input_data: Annotated[dict[str, list[str]] | None, Field(frozen=True)] = None
+#    input_data: Annotated[dict[str, list[str]] | None, Field(frozen=True)] = None
     input_meta: Annotated[list[str] | None, Field(frozen=True)] = None
     output_data: Annotated[dict[str, list[str]] | None, Field(frozen=True)] = None
     output_meta: Annotated[list[str] | None, Field(frozen=True)] = None
     fail_ok: bool = False
 
+    @property
+    def input_meta(self) -> Annotated[list[str] | None, Field(frozen=True)]:
+        return
+
     @abc.abstractmethod
     def apply(self, df: IamDataFrame) -> IamDataFrame:
         raise NotImplementedError

tests/test_aggregator.py

@@ -34,8 +34,7 @@ def test_aggregator_from_file():
         "rename": [
             {"name": "Resource|Extraction|Petrol", "rename": "Resource|Extraction|Oil"}
         ],
-        "input_data": None,
-        "input_meta": None,
+#        "input_data": None,
         "output_data": None,
         "output_meta": None,
         "fail_ok": False,

tests/test_processor.py

@@ -1,4 +1,4 @@
 from pydantic import ValidationError
 import pytest
 from pyam import IamDataFrame
 
@@ -10,22 +10,22 @@
         def apply(self, df: IamDataFrame) -> IamDataFrame:
             return df
 
-    input_data = {"variable": ["Emissions|CO2"], "region": ["World"]}
+#    input_data = {"variable": ["Emissions|CO2"], "region": ["World"]}
     output_meta = ["Emissions Diagnostics|Year of Net Zero|CO2"]
     processor = ProcessorSubclass(
-        input_data=input_data,
+#        input_data=input_data,
         output_meta=output_meta,
         fail_ok=True,
     )
 
-    assert processor.input_data == input_data
+#    assert processor.input_data == input_data
     assert processor.input_meta is None
     assert processor.output_data is None
     assert processor.output_meta == output_meta
 
     # check that frozen fields cannot be modified after instantiation
-    with pytest.raises(ValidationError, match="Field is frozen"):
-        processor.input_data = {"variable": ["Emissions|Kyoto Gases"]}
-    with pytest.raises(ValidationError, match="Field is frozen"):
-        processor.input_meta = ["Emissions Diagnostics|Cumulative CO2 [2020-2100"]
+#    with pytest.raises(ValidationError, match="Field is frozen"):
+#        processor.input_data = {"variable": ["Emissions|Kyoto Gases"]}
+#    with pytest.raises(ValidationError, match="Field is frozen"):
+#        processor.input_meta = ["Emissions Diagnostics|Cumulative CO2 [2020-2100"]
     assert processor.fail_ok is True

tests/test_validate_data.py

@@ -42,6 +42,8 @@ def test_DataValidator_simple_from_file():
     dsd = DataStructureDefinition(TEST_DATA_DIR / "validation" / "definitions")
     assert obs.validate_with_definition(dsd) is None
 
+    assert obs.input_data == {"variable": ["Final Energy"], "year": [2010]}
+
 
 def test_DataValidator_structured_from_file():
     exp = DataValidator(
@@ -98,6 +100,10 @@ def test_DataValidator_from_codelist():
     )
     assert variable_codelist.data_validator == exp
 
+    assert variable_codelist.data_validator.input_data == {
+        "variable": ["Final Energy", "Final Energy|Electricity [Share]"]
+    }
+
 
 @pytest.mark.parametrize(
     "name, match",