1274 Add minus operation for set difference to support CORE-000334 expected-variable checks

cdisc_rules_engine/models/operation_params.py

@@ -58,6 +58,7 @@ class OperationParams:
     returntype: str = None
     source: str = None
     target: str = None
+    subtract: str = None
     value_is_reference: bool = False
     namespace: str = None
     delimiter: str = None

cdisc_rules_engine/operations/minus.py

@@ -0,0 +1,46 @@
+"""
+Set difference operation: name minus subtract.
+Returns elements in name that are not in subtract, preserving order from name.
+"""
+
+from cdisc_rules_engine.operations.base_operation import BaseOperation
+
+
+def _normalize_to_list(val):
+    """Convert value to a list for set operations."""
+    if val is None:
+        return []
+    if isinstance(val, list):
+        return val
+    if isinstance(val, (set, tuple)):
+        return list(val)
+    return [val]
+
+
+def _set_difference_preserve_order(list_a: list, list_b: list) -> list:
+    """
+    Compute set difference A \\ B (elements in A not in B).
+    Preserves order from list_a.
+    """
+    set_b = set(_normalize_to_list(list_b))
+    return [x for x in _normalize_to_list(list_a) if x not in set_b]
+
+
+class Minus(BaseOperation):
+    """
+    Operation that computes set difference: name minus subtract.
+    name (minuend) and subtract (subtrahend) reference other operation results.
+    Returns elements in name that are not in subtract.
+    """
+
+    def _execute_operation(self):
+        name_ref = self.params.target
+        subtract_ref = self.params.subtract
+
+        if not name_ref or name_ref not in self.evaluation_dataset.columns:
+            return []
+        list_a = self.evaluation_dataset[name_ref].iloc[0]
+        if not subtract_ref or subtract_ref not in self.evaluation_dataset.columns:
+            return _normalize_to_list(list_a)
+        list_b = self.evaluation_dataset[subtract_ref].iloc[0]
+        return _set_difference_preserve_order(list_a, list_b)

cdisc_rules_engine/operations/operations_factory.py

@@ -41,6 +41,7 @@
     MedDRATermReferencesValidator,
 )
 from cdisc_rules_engine.operations.min_date import MinDate
+from cdisc_rules_engine.operations.minus import Minus
 from cdisc_rules_engine.operations.minimum import Minimum
 from cdisc_rules_engine.operations.record_count import RecordCount
 from cdisc_rules_engine.operations.split_by import SplitBy
@@ -109,6 +110,7 @@ class OperationsFactory(FactoryInterface):
         "mean": Mean,
         "min": Minimum,
         "min_date": MinDate,
+        "minus": Minus,
         "record_count": RecordCount,
         "valid_meddra_code_references": MedDRACodeReferencesValidator,
         "valid_whodrug_references": WhodrugReferencesValidator,

cdisc_rules_engine/utilities/rule_processor.py

@@ -391,6 +391,7 @@ def perform_rule_operations(
                 operation_id=operation.get("id"),
                 operation_name=operation.get("operator"),
                 original_target=original_target,
+                subtract=operation.get("subtract"),
                 regex=operation.get("regex"),
                 returntype=operation.get("returntype"),
                 source=operation.get("source"),

resources/schema/rule/Operations.json

@@ -50,42 +50,54 @@
     },
     {
       "properties": {
-        "operator": { "const": "domain_is_custom" }
+        "operator": {
+          "const": "domain_is_custom"
+        }
       },
       "required": ["id", "operator"],
       "type": "object"
     },
     {
       "properties": {
-        "operator": { "const": "domain_label" }
+        "operator": {
+          "const": "domain_label"
+        }
       },
       "required": ["id", "operator"],
       "type": "object"
     },
     {
       "properties": {
-        "operator": { "const": "dy" }
+        "operator": {
+          "const": "dy"
+        }
       },
       "required": ["id", "operator", "name"],
       "type": "object"
     },
     {
       "properties": {
-        "operator": { "const": "extract_metadata" }
+        "operator": {
+          "const": "extract_metadata"
+        }
       },
       "required": ["id", "operator", "name"],
       "type": "object"
     },
     {
       "properties": {
-        "operator": { "const": "expected_variables" }
+        "operator": {
+          "const": "expected_variables"
+        }
       },
       "required": ["id", "operator"],
       "type": "object"
     },
     {
       "properties": {
-        "operator": { "const": "get_codelist_attributes" }
+        "operator": {
+          "const": "get_codelist_attributes"
+        }
       },
       "required": ["id", "operator", "name", "ct_attribute", "version"],
       "type": "object"
@@ -162,46 +174,67 @@
     },
     {
       "properties": {
-        "operator": { "const": "map" }
+        "operator": {
+          "const": "map"
+        }
       },
       "required": ["id", "operator", "map"],
       "type": "object"
     },
     {
       "properties": {
-        "operator": { "const": "max" }
+        "operator": {
+          "const": "max"
+        }
       },
       "required": ["id", "operator", "name"],
       "type": "object"
     },
     {
       "properties": {
-        "operator": { "const": "max_date" }
+        "operator": {
+          "const": "max_date"
+        }
       },
       "required": ["id", "operator", "name"],
       "type": "object"
     },
     {
       "properties": {
-        "operator": { "const": "mean" }
+        "operator": {
+          "const": "mean"
+        }
       },
       "required": ["id", "operator", "name"],
       "type": "object"
     },
     {
       "properties": {
-        "operator": { "const": "min" }
+        "operator": {
+          "const": "min"
+        }
       },
       "required": ["id", "operator", "name"],
       "type": "object"
     },
     {
       "properties": {
-        "operator": { "const": "min_date" }
+        "operator": {
+          "const": "min_date"
+        }
       },
       "required": ["id", "operator", "name"],
       "type": "object"
     },
+    {
+      "properties": {
+        "operator": {
+          "const": "minus"
+        }
+      },
+      "required": ["id", "operator", "name", "subtract"],
+      "type": "object"
+    },
     {
       "properties": {
         "operator": {
@@ -222,42 +255,54 @@
     },
     {
       "properties": {
-        "operator": { "const": "record_count" }
+        "operator": {
+          "const": "record_count"
+        }
       },
       "required": ["id", "operator"],
       "type": "object"
     },
     {
       "properties": {
-        "operator": { "const": "required_variables" }
+        "operator": {
+          "const": "required_variables"
+        }
       },
       "required": ["id", "operator"],
       "type": "object"
     },
     {
       "properties": {
-        "operator": { "const": "split_by" }
+        "operator": {
+          "const": "split_by"
+        }
       },
       "required": ["id", "operator", "delimiter", "name"],
       "type": "object"
     },
     {
       "properties": {
-        "operator": { "const": "study_domains" }
+        "operator": {
+          "const": "study_domains"
+        }
       },
       "required": ["id", "operator"],
       "type": "object"
     },
     {
       "properties": {
-        "operator": { "const": "dataset_names" }
+        "operator": {
+          "const": "dataset_names"
+        }
       },
       "required": ["id", "operator"],
       "type": "object"
     },
     {
       "properties": {
-        "operator": { "const": "standard_domains" }
+        "operator": {
+          "const": "standard_domains"
+        }
       },
       "required": ["id", "operator"],
       "type": "object"
@@ -280,7 +325,6 @@
       "required": ["id", "operator", "external_dictionary_type"],
       "type": "object"
     },
-
     {
       "properties": {
         "operator": {
@@ -364,28 +408,36 @@
     },
     {
       "properties": {
-        "operator": { "const": "variable_count" }
+        "operator": {
+          "const": "variable_count"
+        }
       },
       "required": ["id", "operator"],
       "type": "object"
     },
     {
       "properties": {
-        "operator": { "const": "variable_exists" }
+        "operator": {
+          "const": "variable_exists"
+        }
       },
       "required": ["id", "operator"],
       "type": "object"
     },
     {
       "properties": {
-        "operator": { "const": "variable_is_null" }
+        "operator": {
+          "const": "variable_is_null"
+        }
       },
       "required": ["id", "operator"],
       "type": "object"
     },
     {
       "properties": {
-        "operator": { "const": "variable_names" }
+        "operator": {
+          "const": "variable_names"
+        }
       },
       "required": ["id", "operator"],
       "type": "object"
@@ -410,7 +462,9 @@
     },
     {
       "properties": {
-        "operator": { "const": "get_xhtml_errors" }
+        "operator": {
+          "const": "get_xhtml_errors"
+        }
       },
       "required": ["id", "operator", "name", "namespace"],
       "type": "object"
@@ -457,7 +511,9 @@
       ]
     },
     "ct_package_types": {
-      "items": { "$ref": "Operations.json#/properties/ct_package_type" },
+      "items": {
+        "$ref": "Operations.json#/properties/ct_package_type"
+      },
       "type": "array"
     },
     "ct_packages": {
@@ -488,7 +544,9 @@
     "external_dictionary_type": {
       "enum": ["meddra"]
     },
-    "filter": { "type": "object" },
+    "filter": {
+      "type": "object"
+    },
     "filter_key": {
       "type": "string"
     },
@@ -536,7 +594,9 @@
       "items": {
         "type": "object",
         "properties": {
-          "output": { "type": "string" }
+          "output": {
+            "type": "string"
+          }
         },
         "required": ["output"]
       }
@@ -560,6 +620,9 @@
     "source": {
       "type": "string"
     },
+    "subtract": {
+      "type": "string"
+    },
     "term_value": {
       "type": "string"
     },