Generate excel test-data on demand for fmudesign

pyproject.toml

@@ -61,7 +61,8 @@ dev = [
     'types-openpyxl',
     'types-seaborn',
     'types-setuptools',
-    'xlwt'
+    'xlsxwriter>=3.2.3',
+    'xlwt',
 ]
 
 [project.urls]

tests/fmudesign/conftest.py

@@ -0,0 +1,62 @@
+import csv
+from pathlib import Path
+
+import pytest
+from xlsxwriter import Workbook  # type: ignore[import-untyped]
+from xlsxwriter.worksheet import Worksheet  # type: ignore[import-untyped]
+
+from tests.fmudesign.workbook_specs import WORKBOOK_SPECS, CellValue, WorkbookSpec
+
+SOURCE_DATA = Path(__file__).parent / "data"
+
+
+def _write_cell(worksheet: Worksheet, row: int, column: int, value: CellValue) -> None:
+    if isinstance(value, str):
+        worksheet.write_string(row, column, value)
+    elif value is not None:
+        worksheet.write_number(row, column, value)
+
+
+def _write_workbook(path: Path, spec: WorkbookSpec) -> None:
+    path.parent.mkdir(parents=True, exist_ok=True)
+    with Workbook(path) as workbook:
+        for sheet_name, rows in spec.items():
+            worksheet = workbook.add_worksheet(sheet_name)
+            for row_number, values in rows.items():
+                for column, value in enumerate(values):
+                    _write_cell(worksheet, row_number - 1, column, value)
+
+
+def _parse_csv_cell(value: str) -> CellValue:
+    if not value:
+        return None
+    try:
+        return int(value)
+    except ValueError:
+        try:
+            return float(value)
+        except ValueError:
+            return value
+
+
+def _write_design_summary_workbook(path: Path) -> None:
+    with (SOURCE_DATA / "distributions/design.csv").open(
+        encoding="utf-8", newline=""
+    ) as stream:
+        rows = {
+            row_number: tuple(_parse_csv_cell(value) for value in values)
+            for row_number, values in enumerate(csv.reader(stream), start=1)
+        }
+    # The legacy Excel fixture uses the full parameter name; the CSV abbreviates it.
+    rows[1] = (*rows[1][:-1], "RELP_GO_ILETOFTE")
+    _write_workbook(path, {"DesignSheet01": rows})
+
+
+@pytest.fixture(scope="session")
+def fmudesign_test_data(tmp_path_factory: pytest.TempPathFactory) -> Path:
+    test_data = tmp_path_factory.mktemp("fmudesign_test_data")
+    config_dir = test_data / "config"
+    for filename, spec in WORKBOOK_SPECS.items():
+        _write_workbook(config_dir / filename, spec)
+    _write_design_summary_workbook(test_data / "distributions/design.xlsx")
+    return test_data

tests/fmudesign/data/README.md

@@ -1,4 +1,7 @@
-Testdata for tornadoplots from one by one sensitivities (design matrix)
-distributions: contains design matrix on fmu standard format in excel and .csv format
-results: contains in place volumes exported from RMS in fmu standard csv format
-config: contains yaml config files for add_webviz_tornado_onebyone.py
+Text test data for FMU-design.
+
+- `distributions` contains a design matrix in FMU-standard CSV format.
+- `results` contains in-place volumes exported from RMS in FMU-standard CSV format.
+- `config` contains external seed data used by generated workbook fixtures.
+
+Excel workbooks are generated from `tests/fmudesign/workbook_specs.py` during tests.

tests/fmudesign/test_create_design.py

@@ -18,8 +18,6 @@
 from semeio.fmudesign.create_design import MonteCarloSensitivity, _derive_rng
 from semeio.fmudesign.quality_report import print_corrmat
 
-TESTDATA = Path(__file__).parent / "data"
-
 
 @pytest.mark.integration_test
 @pytest.mark.parametrize("correlations", [True, False])
@@ -209,10 +207,10 @@ def gl(paramname, distname, p1, p2, p3="", p4=""):
         assert np.sqrt(np.mean((obs_corr - corr_values) ** 2)) < 0.02
 
 
-def test_generate_onebyone(tmpdir):
+def test_generate_onebyone(tmpdir, fmudesign_test_data):
     """Test generation of onebyone design"""
 
-    inputfile = TESTDATA / "config/design_input_example1.xlsx"
+    inputfile = fmudesign_test_data / "config/design_input_example1.xlsx"
 
     input_dict = excel_to_dict(inputfile)
 
@@ -379,14 +377,14 @@ def test_generate_onebyone(tmpdir):
         pytest.fail("Timestamp in Metadata sheet is not in expected format")
 
 
-def test_generate_full_mc_snapshot(snapshot):
+def test_generate_full_mc_snapshot(snapshot, fmudesign_test_data):
     """Test that full monte carlo design matrix generation remains consistent.
 
     This is a snapshot test that verifies the entire output of the design matrix
     generation process, including both the design values and default values.
     """
     # Setup
-    inputfile = TESTDATA / "config/design_input_mc_with_correls.xlsx"
+    inputfile = fmudesign_test_data / "config/design_input_mc_with_correls.xlsx"
     input_dict = excel_to_dict(inputfile)
     design = DesignMatrix()
 
@@ -415,11 +413,11 @@ def test_generate_full_mc_snapshot(snapshot):
     snapshot.assert_match(snapshot_str, "design_output_mc_with_correls.json")
 
 
-def test_generate_full_mc_snapshot_independent(snapshot):
+def test_generate_full_mc_snapshot_independent(snapshot, fmudesign_test_data):
     """Same config as test_generate_full_mc_snapshot, but with the opt-in
     'independent' seed strategy.
     """
-    inputfile = TESTDATA / "config/design_input_mc_with_correls.xlsx"
+    inputfile = fmudesign_test_data / "config/design_input_mc_with_correls.xlsx"
     input_dict = excel_to_dict(inputfile)
     input_dict["seed_strategy"] = "independent"
     design = DesignMatrix()
@@ -440,9 +438,9 @@ def test_generate_full_mc_snapshot_independent(snapshot):
     snapshot.assert_match(snapshot_str, "design_output_mc_with_correls.json")
 
 
-def test_generate_full_mc(tmpdir):
+def test_generate_full_mc(tmpdir, fmudesign_test_data):
     """Test generation of full monte carlo"""
-    inputfile = TESTDATA / "config/design_input_mc_with_correls.xlsx"
+    inputfile = fmudesign_test_data / "config/design_input_mc_with_correls.xlsx"
     input_dict = excel_to_dict(inputfile)
 
     design = DesignMatrix()
@@ -517,10 +515,10 @@ def test_generate_full_mc(tmpdir):
 
 
 @pytest.mark.integration_test
-def test_generate_background(tmpdir):
-    inputfile = TESTDATA / "config/design_input_background.xlsx"
+def test_generate_background(tmpdir, fmudesign_test_data):
+    inputfile = fmudesign_test_data / "config/design_input_background.xlsx"
     input_dict = excel_to_dict(inputfile)
-    source_file = TESTDATA / "config/doe1.xlsx"
+    source_file = fmudesign_test_data / "config/doe1.xlsx"
     dest_file = tmpdir.join("doe1.xlsx")
     shutil.copy2(source_file, dest_file)
 

tests/fmudesign/test_designmatrix.py

@@ -10,8 +10,6 @@
 
 from semeio.fmudesign import DesignMatrix
 
-TESTDATA = Path(__file__).parent / "data"
-
 
 def matches(pattern: str, text: str) -> bool:
     """Match text against a pattern where <ANY> acts as a wildcard.
@@ -74,13 +72,13 @@ def test_designmatrix():
 
 
 @pytest.mark.integration_test
-def test_endpoint(tmpdir, monkeypatch):
+def test_endpoint(tmpdir, monkeypatch, fmudesign_test_data):
     """Test the installed endpoint
 
     Will write generated design matrices to the pytest tmpdir directory,
     usually /tmp/pytest-of-<username>/
     """
-    designfile = TESTDATA / "config/design_input_onebyone.xlsx"
+    designfile = fmudesign_test_data / "config/design_input_onebyone.xlsx"
 
     # The xlsx file contains a relative path, relative to the input design sheet:
     dependency = (

tests/fmudesign/test_designsummary.py

@@ -2,14 +2,14 @@
 
 from semeio.fmudesign import summarize_design
 
-TESTDATA = Path(__file__).parent / "data"
+SOURCE_DATA = Path(__file__).parent / "data"
 
 
-def test_designsummary():
+def test_designsummary(fmudesign_test_data):
     """Test import and summary of design matrix"""
 
     snorrebergdesign = summarize_design(
-        TESTDATA / "distributions/design.xlsx", "DesignSheet01"
+        fmudesign_test_data / "distributions/design.xlsx", "DesignSheet01"
     )
     # checking dimensions and some values in summary of design matrix
     assert snorrebergdesign.shape == (7, 9)
@@ -49,5 +49,5 @@ def test_designsummary():
     assert snorrebergdesign["endreal1"].sum() == 333
 
     # Test same also when design matrix is in .csv format
-    designcsv = summarize_design(TESTDATA / "distributions/design.csv")
+    designcsv = summarize_design(SOURCE_DATA / "distributions/design.csv")
     assert snorrebergdesign.equals(designcsv)

tests/fmudesign/test_use_cases.py

@@ -10,12 +10,10 @@
 
 from semeio.fmudesign import DesignMatrix, excel_to_dict
 from semeio.fmudesign.fmudesignrunner import EXAMPLES
+from tests.fmudesign.workbook_specs import DESIGN_INPUT_FILES, DESIGN_INPUT_IDS
 
 EXAMPLE_FILES = [ex.filename for ex in EXAMPLES]
 
-TESTDATA = Path(__file__).parent / "data"
-TEST_FILES = list((TESTDATA / "config").glob("design_input*.xlsx"))
-
 
 def test_prediction_rejection_sampled_ensemble(tmpdir, monkeypatch):
     """Test making a design matrix for prediction realizations based on an
@@ -163,11 +161,14 @@ def test_constant_distribution(tmpdir, monkeypatch, gen_input_sheet):
 
 
 @pytest.mark.integration_test
-@pytest.mark.parametrize("designfile", TEST_FILES, ids=[p.stem for p in TEST_FILES])
+@pytest.mark.parametrize("designfile", DESIGN_INPUT_FILES, ids=DESIGN_INPUT_IDS)
 @pytest.mark.parametrize("verbosity", [0, 1, 2])
-def test_all_input_files(tmpdir, monkeypatch, designfile, verbosity):
+def test_all_input_files(
+    tmpdir, monkeypatch, fmudesign_test_data, designfile, verbosity
+):
     """Smoketest all files."""
 
+    designfile = fmudesign_test_data / "config" / designfile
     monkeypatch.chdir(tmpdir)
 
     # Copy all example files over, to guarantee existence of dependency files
@@ -203,11 +204,14 @@ def test_all_example_files_cmd_init(tmpdir, monkeypatch, designfile, verbosity):
 
 
 @pytest.mark.integration_test
-@pytest.mark.parametrize("designfile", TEST_FILES, ids=[p.stem for p in TEST_FILES])
-def test_all_input_files_relative_paths(tmpdir, monkeypatch, designfile):
+@pytest.mark.parametrize("designfile", DESIGN_INPUT_FILES, ids=DESIGN_INPUT_IDS)
+def test_all_input_files_relative_paths(
+    tmpdir, monkeypatch, fmudesign_test_data, designfile
+):
     """Smoketest all files, but invoke them from a directory above.
     This tests that relative paths in the Excel files work correctly."""
 
+    designfile = fmudesign_test_data / "config" / designfile
     monkeypatch.chdir(tmpdir)
     copy_to = os.path.join(".", "path", "going", "down")