Add CLI command to validate scenario data file from definitions

nomenclature/cli.py

@@ -277,3 +277,30 @@ def cli_run_workflow(
     df = getattr(workflow, workflow_function)(IamDataFrame(input_file))
     if output_file is not None:
         df.to_excel(output_file)
+
+
+@cli.command("validate-scenarios")
+@click.argument("input_file", type=click.Path(exists=True, path_type=Path))
+@click.option(
+    "--definitions",
+    help="Optional name for definitions folder",
+    type=click.Path(exists=True, path_type=Path),
+    default="definitions",
+)
+def cli_validate_scenarios(input_file: Path, definitions: Path):
+    """Run a given input file through a DataStructureDefinition validation,
+    checking it against defined codelist(s)
+
+    Parameters
+    ----------
+    input_file : Path
+        Input data file, must be IAMC format, .xlsx or .csv
+    definitions : Path
+        Definitions folder with codelists, by default "definitions"
+
+    Raises
+    ------
+    ValueError
+        If input_file validation fails against specified codelist(s).
+    """
+    DataStructureDefinition(definitions).validate(IamDataFrame(input_file))

tests/data/cli/validate_scenarios/definitions/region/regions.yaml

@@ -0,0 +1,4 @@
+- common:
+    - World
+- country:
+    - Austria

tests/data/cli/validate_scenarios/definitions/variable/variables.yaml

@@ -0,0 +1,3 @@
+- Primary Energy:
+    definition: Total primary energy consumption
+    unit: EJ/yr

tests/test_cli.py

@@ -420,3 +420,28 @@ def test_cli_run_workflow(tmp_path, simple_df):
     )
 
     assert_iamframe_equal(simple_df, IamDataFrame(tmp_path / "output.xlsx"))
+
+
+@pytest.mark.parametrize(
+    "status, unit, exit_code", [("valid", "EJ/yr", 0), ("invalid", "EJ", 1)]
+)
+def test_cli_valid_scenarios(status, unit, exit_code, tmp_path):
+    """Check that CLI validates an IAMC dataset according to defined codelist."""
+    IamDataFrame(
+        pd.DataFrame(
+            [
+                ["m_a", "s_a", "World", "Primary Energy", unit, 1, 2],
+            ],
+            columns=IAMC_IDX + [2005, 2010],
+        )
+    ).to_excel(tmp_path / f"{status}_data.xlsx")
+    result_valid = runner.invoke(
+        cli,
+        [
+            "validate-scenarios",
+            str(tmp_path / f"{status}_data.xlsx"),
+            "--definitions",
+            str(MODULE_TEST_DATA_DIR / "validate_scenarios" / "definitions"),
+        ],
+    )
+    assert result_valid.exit_code == exit_code