Improve validation of design matrix

This commit adds some validation to design matrix, and also makes the active_realizations_field
in the gui only accept realizations actually found in design matrix.
The designmatrix validation added in this commit:
* parameter names cannot contain multiple words
* parameter names cannot be numerical

src/ert/config/design_matrix.py

@@ -191,7 +191,7 @@ def read_design_matrix(
 
         if error_list := DesignMatrix._validate_design_matrix(design_matrix_df):
             error_msg = "\n".join(error_list)
-            raise ValueError(f"Design matrix is not valid, error:\n{error_msg}")
+            raise ValueError(f"Design matrix is not valid, error(s):\n{error_msg}")
 
         defaults_to_use = DesignMatrix._read_defaultssheet(
             self.xls_filename, self.default_sheet, design_matrix_df.columns.to_list()
@@ -259,13 +259,6 @@ def _validate_design_matrix(design_matrix: pd.DataFrame) -> list[str]:
             return []
         errors = []
         column_na_mask = design_matrix.columns.isna()
-        column_indexes_unnamed = [
-            index for index, value in enumerate(column_na_mask) if value
-        ]
-        if len(column_indexes_unnamed) > 0:
-            errors.append(
-                f"Column headers not present in column {column_indexes_unnamed}"
-            )
         if not design_matrix.columns[~column_na_mask].is_unique:
             errors.append("Duplicate parameter names found in design sheet")
         empties = [
@@ -274,6 +267,16 @@ def _validate_design_matrix(design_matrix: pd.DataFrame) -> list[str]:
         ]
         if len(empties) > 0:
             errors.append(f"Design matrix contains empty cells {empties}")
+
+        for column_num, param_name in enumerate(design_matrix.columns):
+            if pd.isna(param_name) or len(param_name.split()) == 0:
+                errors.append(f"Empty parameter name found in column {column_num}.")
+            elif len(param_name.split()) > 1:
+                errors.append(
+                    f"Multiple words in parameter name found in column {column_num} ({param_name})."
+                )
+            elif param_name.isnumeric():
+                errors.append(f"Numeric parameter name found in column {column_num}.")
         return errors
 
     @staticmethod

src/ert/gui/simulation/ensemble_experiment_panel.py

@@ -15,7 +15,7 @@
 from ert.gui.tools.design_matrix.design_matrix_panel import DesignMatrixPanel
 from ert.mode_definitions import ENSEMBLE_EXPERIMENT_MODE
 from ert.run_models import EnsembleExperiment
-from ert.validation import ActiveRange, RangeStringArgument
+from ert.validation import ActiveRange, RangeStringArgument, RangeSubsetStringArgument
 from ert.validation.proper_name_argument import ExperimentValidation, ProperNameArgument
 
 from .experiment_config_panel import ExperimentConfigPanel
@@ -88,6 +88,11 @@ def __init__(
             self._active_realizations_field.setText(
                 ActiveRange(design_matrix.active_realizations).rangestring
             )
+            self._active_realizations_field.setValidator(
+                RangeSubsetStringArgument(
+                    ActiveRange(design_matrix.active_realizations)
+                )
+            )
             show_dm_param_button = QPushButton("Show parameters")
             show_dm_param_button.setObjectName("show-dm-parameters")
             show_dm_param_button.setMinimumWidth(50)

src/ert/validation/__init__.py

@@ -5,7 +5,7 @@
 from .number_list_string_argument import NumberListStringArgument
 from .proper_name_argument import ExperimentValidation, ProperNameArgument
 from .proper_name_format_argument import ProperNameFormatArgument
-from .range_string_argument import RangeStringArgument
+from .range_string_argument import RangeStringArgument, RangeSubsetStringArgument
 from .rangestring import mask_to_rangestring, rangestring_to_list, rangestring_to_mask
 from .runpath_argument import RunPathArgument
 from .string_definition import StringDefinition
@@ -21,6 +21,7 @@
     "ProperNameArgument",
     "ProperNameFormatArgument",
     "RangeStringArgument",
+    "RangeSubsetStringArgument",
     "RunPathArgument",
     "StringDefinition",
     "ValidationStatus",

src/ert/validation/active_range.py

@@ -8,6 +8,7 @@ def __init__(
         rangestring: str | None = None,
         length: int | None = None,
     ):
+        self.length = length
         if mask is None and rangestring is None and length is None:
             raise ValueError("Supply mask or rangestring and length to IndexRange.")
         if mask is None:
@@ -74,3 +75,13 @@ def validate_rangestring_vs_length(
                     f"for size {length}"
                 )
         return (rangestring, length)
+
+    def validate_range_is_subset(self, other: "ActiveRange") -> None:
+        if (
+            not [a and b for a, b in zip(self.mask, other.mask, strict=False)]
+            == other.mask
+        ):
+            raise ValueError(
+                f"Specified rangestring ({other.rangestring}) is not a subset "
+                f"of the active realizations ({self.rangestring})"
+            )

src/ert/validation/range_string_argument.py

@@ -41,3 +41,24 @@ def validate(self, token: str) -> ValidationStatus:
         validation_status.setValue(token)
 
         return validation_status
+
+
+class RangeSubsetStringArgument(RangeStringArgument):
+    def __init__(self, source_active_range: ActiveRange) -> None:
+        super().__init__()
+        self._source_range = source_active_range
+
+    def validate(self, token: str) -> ValidationStatus:
+        validation_status = super().validate(token)
+
+        if not validation_status:
+            return validation_status
+        try:
+            other_range = ActiveRange(rangestring=token, length=len(self._source_range))
+            self._source_range.validate_range_is_subset(other_range)
+        except ValueError as e:
+            validation_status.setFailed()
+            validation_status.addToMessage(str(e))
+
+        validation_status.setValue(token)
+        return validation_status

tests/ert/unit_tests/gui/ide/test_range_string_argument.py

@@ -1,4 +1,5 @@
-from ert.validation import RangeStringArgument
+from ert.validation import RangeStringArgument, RangeSubsetStringArgument
+from ert.validation.active_range import ActiveRange
 
 
 def test_proper_name_argument():
@@ -23,3 +24,15 @@ def test_proper_name_argument():
 
     assert argument.validate("1-5, 9")
     assert not argument.validate("10")
+
+
+def test_range_subset():
+    source_range = ActiveRange(rangestring="0-3,5,7-9", length=10)
+    argument_range = RangeSubsetStringArgument(source_active_range=source_range)
+
+    assert argument_range.validate("1")
+    assert argument_range.validate("0-3")
+    assert argument_range.validate("0-3,5,7-9")
+    assert not argument_range.validate("10")
+    assert not argument_range.validate("1-10")
+    assert not argument_range.validate("0-4")

tests/ert/unit_tests/sensitivity_analysis/test_design_matrix.py

@@ -238,9 +238,25 @@ def test_reading_design_matrix_validate_reals(tmp_path, real_column, error_msg):
         ),
         pytest.param(
             ["a", "b  ", ""],
-            r"Column headers not present in column \[2\]",
+            r"Empty parameter name found in column 2",
             id="missing entries",
         ),
+        pytest.param(
+            ["a", "b", "parameter name with spaces"],
+            "Multiple words in parameter name found in column 2.",
+            id="multiple words in parameter name",
+        ),
+        pytest.param(
+            ["a", "b", " "],
+            "Empty parameter name found in column 2",
+            id="dataframe loads parameter name as whitespace",
+        ),
+        pytest.param(["a", "b", "3"], "Numeric parameter name found in column 2"),
+        pytest.param(
+            ["a", "b c d e", 33],
+            r"Multiple words in parameter name found in column 1 \(b c d e\)\.\nNumeric parameter name found in column 2",
+            id="multiple errors",
+        ),
     ],
 )
 def test_reading_design_matrix_validate_headers(tmp_path, column_names, error_msg):