#1248 Specify define xml path in operations

cdisc_rules_engine/models/operation_params.py

@@ -62,3 +62,4 @@ class OperationParams:
     value_is_reference: bool = False
     namespace: str = None
     delimiter: str = None
+    define_xml_path: str = None

cdisc_rules_engine/operations/define_dictionary_version_validator.py

@@ -36,8 +36,15 @@ def _execute_operation(self) -> bool:
             whodrug_path=self.params.whodrug_path,
             loinc_path=self.params.loinc_path,
         )
+        define_path = (
+            self.params.define_xml_path
+            if self.params.define_xml_path
+            else os.path.join(self.params.directory_path, DEFINE_XML_FILE_NAME)
+        )
+        if not os.path.exists(define_path):
+            raise FileNotFoundError(f"Define XML file {define_path} not found")
         define_contents = self.data_service.get_define_xml_contents(
-            dataset_name=os.path.join(self.params.directory_path, DEFINE_XML_FILE_NAME)
+            dataset_name=define_path
         )
         define_reader = DefineXMLReaderFactory.from_file_contents(define_contents)
         define_dictionary_version = define_reader.get_external_dictionary_version(

cdisc_rules_engine/operations/define_variable_metadata.py

@@ -1,9 +1,10 @@
-from cdisc_rules_engine.constants.define_xml_constants import DEFINE_XML_FILE_NAME
+import os.path
+
 from cdisc_rules_engine.services.define_xml.define_xml_reader_factory import (
     DefineXMLReaderFactory,
 )
 from .base_operation import BaseOperation
-import os
+from cdisc_rules_engine.constants.define_xml_constants import DEFINE_XML_FILE_NAME
 
 
 class DefineVariableMetadata(BaseOperation):
@@ -32,8 +33,15 @@ def _execute_operation(self):
                     ...
                 }
         """
+        define_path = (
+            self.params.define_xml_path
+            if self.params.define_xml_path
+            else os.path.join(self.params.directory_path, DEFINE_XML_FILE_NAME)
+        )
+        if not os.path.exists(define_path):
+            raise FileNotFoundError(f"Define XML file {define_path} not found")
         define_contents = self.data_service.get_define_xml_contents(
-            dataset_name=os.path.join(self.params.directory_path, DEFINE_XML_FILE_NAME)
+            dataset_name=define_path
         )
         define_reader = DefineXMLReaderFactory.from_file_contents(define_contents)
         variables_metadata = define_reader.extract_variables_metadata(

cdisc_rules_engine/rules_engine.py

@@ -399,6 +399,7 @@ def execute_rule(
             standard_substandard=self.standard_substandard,
             external_dictionaries=self.external_dictionaries,
             ct_packages=ct_packages,
+            define_xml_path=self.define_xml_path,
         )
         dataset_variable = DatasetVariable(
             dataset,

cdisc_rules_engine/utilities/rule_processor.py

@@ -234,11 +234,12 @@ def rule_applies_to_class(
         excluded_classes = classes.get("Exclude", [])
         is_included = True
         is_excluded = False
+        dataset_name = dataset_metadata.full_path
         if included_classes:
             if ALL_KEYWORD in included_classes:
                 return True
             variables = self.data_service.get_variables_metadata(
-                dataset_name=dataset_metadata.full_path, datasets=datasets
+                dataset_name=dataset_name, datasets=datasets
             ).data.variable_name
             class_name = self.data_service.get_dataset_class(
                 variables,
@@ -252,7 +253,7 @@ def rule_applies_to_class(
                 is_included = False
         if excluded_classes:
             variables = self.data_service.get_variables_metadata(
-                dataset_name=dataset_metadata.full_path, datasets=datasets
+                dataset_name=dataset_name, datasets=datasets
             ).data.variable_name
             class_name = self.data_service.get_dataset_class(
                 variables,
@@ -367,6 +368,7 @@ def perform_rule_operations(
                     for ct_package_type in operation.get("ct_package_types", [])
                 ],
                 ct_version=operation.get("version"),
+                define_xml_path=kwargs.get("define_xml_path"),
                 dataframe=dataset_copy,
                 dataset_path=dataset_path,
                 datasets=datasets,

tests/QARegressionTests/test_Issues/test_CoreIssue1248.py

@@ -0,0 +1,186 @@
+import os
+import subprocess
+
+import pytest
+import json
+from conftest import get_python_executable
+
+
+@pytest.mark.regression
+class TestCoreIssue1248:
+    @pytest.mark.parametrize(
+        "command,rules_report,num_issues",
+        [
+            # define path provided will lead to successful execution
+            (
+                [
+                    f"{get_python_executable()}",
+                    "-m",
+                    "core",
+                    "validate",
+                    "-s",
+                    "sdtmig",
+                    "-v",
+                    "3-2",
+                    "-of",
+                    "JSON",
+                    "-lr",
+                    os.path.join("tests", "resources", "CoreIssue1248", "sample.yml"),
+                    "-cs",
+                    "-dxp",
+                    os.path.join(
+                        "tests",
+                        "resources",
+                        "CoreIssue1248",
+                        "define_subfolder",
+                        "define.xml",
+                    ),
+                    "-ps",
+                    "1",
+                    "-dp",
+                    os.path.join("tests", "resources", "CoreIssue1248", "data.xlsx"),
+                ],
+                [
+                    {
+                        "core_id": "SD1129",
+                        "version": "1",
+                        "cdisc_rule_id": "",
+                        "fda_rule_id": "",
+                        "message": "TEST",
+                        "status": "ISSUE REPORTED",
+                    }
+                ],
+                2,
+            ),
+            # JSON data file and no define.xml in same folder and no -dxp param will provide error
+            (
+                [
+                    f"{get_python_executable()}",
+                    "-m",
+                    "core",
+                    "validate",
+                    "-s",
+                    "sdtmig",
+                    "-v",
+                    "3-2",
+                    "-of",
+                    "JSON",
+                    "-lr",
+                    os.path.join("tests", "resources", "CoreIssue1248", "sample.yml"),
+                    "-cs",
+                    "-dp",
+                    os.path.join("tests", "resources", "CoreIssue1248", "relrec.json"),
+                    "-ps",
+                    "1",
+                ],
+                [
+                    {
+                        "core_id": "SD1129",
+                        "version": "1",
+                        "cdisc_rule_id": "",
+                        "fda_rule_id": "",
+                        "message": "TEST",
+                        "status": "EXECUTION ERROR",
+                    }
+                ],
+                1,
+            ),
+            # no define.xml in same path as data.xlsx file will provide error
+            (
+                [
+                    f"{get_python_executable()}",
+                    "-m",
+                    "core",
+                    "validate",
+                    "-s",
+                    "sdtmig",
+                    "-v",
+                    "3-2",
+                    "-of",
+                    "JSON",
+                    "-lr",
+                    os.path.join("tests", "resources", "CoreIssue1248", "sample.yml"),
+                    "-cs",
+                    "-dp",
+                    os.path.join("tests", "resources", "CoreIssue1248", "data.xlsx"),
+                    "-ps",
+                    "1",
+                ],
+                [
+                    {
+                        "core_id": "SD1129",
+                        "version": "1",
+                        "cdisc_rule_id": "",
+                        "fda_rule_id": "",
+                        "message": "TEST",
+                        "status": "EXECUTION ERROR",
+                    }
+                ],
+                1,
+            ),
+            # define.xml in same folder as the data.xls and no -dxp provided will provide error until
+            # in ExcelDataService dataset metadata creation full_path=dataset_name
+            (
+                [
+                    f"{get_python_executable()}",
+                    "-m",
+                    "core",
+                    "validate",
+                    "-s",
+                    "sdtmig",
+                    "-v",
+                    "3-2",
+                    "-of",
+                    "JSON",
+                    "-lr",
+                    os.path.join("tests", "resources", "CoreIssue1248", "sample.yml"),
+                    "-cs",
+                    "-dp",
+                    os.path.join(
+                        "tests",
+                        "resources",
+                        "CoreIssue1248",
+                        "data_and_define",
+                        "data.xlsx",
+                    ),
+                    "-ps",
+                    "1",
+                ],
+                [
+                    {
+                        "core_id": "SD1129",
+                        "version": "1",
+                        "cdisc_rule_id": "",
+                        "fda_rule_id": "",
+                        "message": "TEST",
+                        "status": "EXECUTION ERROR",
+                    }
+                ],
+                1,
+            ),
+        ],
+    )
+    def test_define_path_used(self, command, rules_report, num_issues):
+        subprocess.run(command, check=True)
+
+        # Get the latest created report file
+        files = os.listdir()
+        json_files = [
+            file
+            for file in files
+            if file.startswith("CORE-Report-") and file.endswith(".json")
+        ]
+        json_report_path = sorted(json_files)[-1]
+        # Open the JSON report file
+        json_report = json.load(open(json_report_path))
+        assert {
+            "Conformance_Details",
+            "Dataset_Details",
+            "Issue_Summary",
+            "Issue_Details",
+            "Rules_Report",
+        }.issubset(json_report.keys())
+        assert len(json_report["Issue_Details"]) == num_issues
+        assert json_report["Rules_Report"] == rules_report
+        if os.path.exists(json_report_path):
+            os.remove(json_report_path)