From 4667a278b039535cb1fe7b75f02948e5a4cbcfb6 Mon Sep 17 00:00:00 2001
From: David Almeida <58078834+dc-almeida@users.noreply.github.com>
Date: Thu, 31 Oct 2024 15:45:02 +0100
Subject: [PATCH] Add dimensions option to CLI validate-scenarios (#421)

* Add dimensions option to CLI validate-scenarios

* Add dimensions from nomenclature.yaml to CLI validate-project

* Update tests

* Revert changes to validate-project testing

* Revert default dimensions checks for CLI validate-scenarios and update tests
---
 nomenclature/cli.py                           | 22 ++++++++-
 .../structure_validation/nomenclature.yaml    |  3 ++
 tests/test_cli.py                             | 48 +++++++++++++++----
 3 files changed, 62 insertions(+), 11 deletions(-)
 create mode 100644 tests/data/cli/structure_validation/nomenclature.yaml

diff --git a/nomenclature/cli.py b/nomenclature/cli.py
index 18ef895d..1a0f45e5 100644
--- a/nomenclature/cli.py
+++ b/nomenclature/cli.py
@@ -287,19 +287,37 @@ def cli_run_workflow(
     type=click.Path(exists=True, path_type=Path),
     default="definitions",
 )
-def cli_validate_scenarios(input_file: Path, definitions: Path):
+@click.option(
+    "--dimension",
+    "dimensions",
+    help="Optional list of dimensions",
+    type=str,
+    multiple=True,
+    default=None,
+)
+def cli_validate_scenarios(input_file: Path, definitions: Path, dimensions: List[str]):
     """Validate a scenario file against the codelists of a project
 
+    Example
+    -------
+    $ nomenclature validate-scenarios <input-file>
+                        --definitions <def-folder>
+                        --dimension <folder1>
+                        --dimension <folder2>
+                        --dimension <folder3>
+
     Parameters
     ----------
     input_file : Path
         Input data file, must be IAMC format, .xlsx or .csv
     definitions : Path
         Definitions folder with codelists, by default "definitions"
+    dimensions : List[str], optional
+        Dimensions to be checked, defaults to all sub-folders of `definitions`
 
     Raises
     ------
     ValueError
         If input_file validation fails against specified codelist(s).
     """
-    DataStructureDefinition(definitions).validate(IamDataFrame(input_file))
+    DataStructureDefinition(definitions, dimensions).validate(IamDataFrame(input_file))
diff --git a/tests/data/cli/structure_validation/nomenclature.yaml b/tests/data/cli/structure_validation/nomenclature.yaml
new file mode 100644
index 00000000..83535f55
--- /dev/null
+++ b/tests/data/cli/structure_validation/nomenclature.yaml
@@ -0,0 +1,3 @@
+dimensions:
+  - region
+  - variable
diff --git a/tests/test_cli.py b/tests/test_cli.py
index 9eaee1ff..ce698579 100644
--- a/tests/test_cli.py
+++ b/tests/test_cli.py
@@ -423,10 +423,15 @@ def test_cli_run_workflow(tmp_path, simple_df):
 
 
 @pytest.mark.parametrize(
-    "status, unit, exit_code", [("valid", "EJ/yr", 0), ("invalid", "EJ", 1)]
+    "status, unit, dimensions, exit_code",
+    [
+        ("valid_1", "EJ/yr", ["region", "variable"], 0),
+        ("invalid", "EJ", "variable", 1),
+        ("valid_2", "EJ", "region", 0),
+    ],
 )
-def test_cli_valid_scenarios(status, unit, exit_code, tmp_path):
-    """Check that CLI validates an IAMC dataset according to defined codelist."""
+def test_cli_valid_scenarios(status, unit, exit_code, dimensions, tmp_path):
+    """Check that CLI validates an IAMC dataset according to defined codelists."""
     IamDataFrame(
         pd.DataFrame(
             [
@@ -435,17 +440,42 @@ def test_cli_valid_scenarios(status, unit, exit_code, tmp_path):
             columns=IAMC_IDX + [2005, 2010],
         )
     ).to_excel(tmp_path / f"{status}_data.xlsx")
+    dimensions = [dimensions] if isinstance(dimensions, str) else dimensions
+    dimension_args = []
+    for dim in dimensions:
+        dimension_args.append("--dimension")
+        dimension_args.append(dim)
+
     result_valid = runner.invoke(
         cli,
         [
             "validate-scenarios",
             str(tmp_path / f"{status}_data.xlsx"),
             "--definitions",
-            str(
-                MODULE_TEST_DATA_DIR
-                / "structure_validation_no_mappings"
-                / "definitions"
-            ),
-        ],
+            str(MODULE_TEST_DATA_DIR / "structure_validation" / "definitions"),
+        ]
+        + dimension_args,
     )
     assert result_valid.exit_code == exit_code
+
+
+def test_cli_valid_scenarios_implicit_dimensions(tmp_path):
+    """Check that CLI validates an IAMC dataset according to implicit dimensions codelists."""
+    IamDataFrame(
+        pd.DataFrame(
+            [
+                ["m_a", "s_a", "World", "Primary Energy", "EJ/yr", 1, 2],
+            ],
+            columns=IAMC_IDX + [2005, 2010],
+        )
+    ).to_excel(tmp_path / "valid_data.xlsx")
+    result_valid = runner.invoke(
+        cli,
+        [
+            "validate-scenarios",
+            str(tmp_path / "valid_data.xlsx"),
+            "--definitions",
+            str(MODULE_TEST_DATA_DIR / "structure_validation" / "definitions"),
+        ],
+    )
+    assert result_valid.exit_code == 0