From 14afcc4511344515e3734d3210cd0cdb2556d291 Mon Sep 17 00:00:00 2001
From: "Hassan D. M. Sambo" <hassandeme.mamasambo@gsa.gov>
Date: Thu, 17 Aug 2023 19:47:22 -0400
Subject: [PATCH] #1848 Added cross validation for award reference uniqueness

---
 backend/audit/cross_validation/__init__.py    |  2 +
 .../check_award_reference_uniqueness.py       | 33 +++++++++++++
 backend/audit/cross_validation/errors.py      |  7 +++
 .../test_check_award_reference_uniqueness.py  | 46 +++++++++++++++++++
 4 files changed, 88 insertions(+)
 create mode 100644 backend/audit/cross_validation/check_award_reference_uniqueness.py
 create mode 100644 backend/audit/cross_validation/test_check_award_reference_uniqueness.py

diff --git a/backend/audit/cross_validation/__init__.py b/backend/audit/cross_validation/__init__.py
index 4a61bf46f9..e8561ef20d 100644
--- a/backend/audit/cross_validation/__init__.py
+++ b/backend/audit/cross_validation/__init__.py
@@ -55,6 +55,7 @@
     ]
 
 """
+from .check_award_reference_uniqueness import check_award_reference_uniqueness
 from .number_of_findings import number_of_findings
 from .additional_ueis import additional_ueis
 from .auditee_ueis_match import auditee_ueis_match
@@ -69,6 +70,7 @@
     auditee_ueis_match,
     additional_ueis,
     award_ref_and_references_uniqueness,
+    check_award_reference_uniqueness,
     number_of_findings,
     submission_progress_check,
     tribal_data_sharing_consent,
diff --git a/backend/audit/cross_validation/check_award_reference_uniqueness.py b/backend/audit/cross_validation/check_award_reference_uniqueness.py
new file mode 100644
index 0000000000..833edc3796
--- /dev/null
+++ b/backend/audit/cross_validation/check_award_reference_uniqueness.py
@@ -0,0 +1,33 @@
+from .errors import (
+    err_duplicate_award_reference,
+)
+from collections import Counter
+
+
+def check_award_reference_uniqueness(sac_dict, *_args, **_kwargs):
+    """
+    Check that all award references in the Federal Award workbook are distinct.
+    """
+
+    all_sections = sac_dict.get("sf_sac_sections", {})
+    federal_awards_section = all_sections.get("federal_awards") or {}
+    federal_awards = federal_awards_section.get("federal_awards", [])
+    award_refs = []
+    errors = []
+
+    for award in federal_awards:
+        award_reference = award.get("award_reference", None)
+        if award_reference:
+            award_refs.append(award_reference)
+
+    duplicated_award_refs = _find_duplicates(award_refs)
+
+    for dup_award_ref in duplicated_award_refs:
+        errors.append({"error": err_duplicate_award_reference(dup_award_ref)})
+
+    return errors
+
+
+def _find_duplicates(lst):
+    count = Counter(lst)
+    return [item for item, count in count.items() if count > 1]
diff --git a/backend/audit/cross_validation/errors.py b/backend/audit/cross_validation/errors.py
index 0e2fc8499c..6a7c0ccc14 100644
--- a/backend/audit/cross_validation/errors.py
+++ b/backend/audit/cross_validation/errors.py
@@ -45,3 +45,10 @@ def err_number_of_findings_inconsistent(total_expected, total_counted, award_ref
         f"You reported {total_expected} findings for award {award_ref} in the {SECTION_NAMES.FEDERAL_AWARDS} workbook, "
         f"but declared {total_counted} findings for the same award in the {SECTION_NAMES.FEDERAL_AWARDS_AUDIT_FINDINGS} workbook."
     )
+
+
+def err_duplicate_award_reference(award_ref):
+    return (
+        f"The award reference {award_ref} shows up more than once. "
+        f"This should not be possible. Please contact customer support."
+    )
diff --git a/backend/audit/cross_validation/test_check_award_reference_uniqueness.py b/backend/audit/cross_validation/test_check_award_reference_uniqueness.py
new file mode 100644
index 0000000000..65aa872a35
--- /dev/null
+++ b/backend/audit/cross_validation/test_check_award_reference_uniqueness.py
@@ -0,0 +1,46 @@
+from django.test import TestCase
+from audit.models import SingleAuditChecklist
+from .check_award_reference_uniqueness import check_award_reference_uniqueness
+from .sac_validation_shape import sac_validation_shape
+from .errors import err_duplicate_award_reference
+from model_bakery import baker
+
+
+class CheckAwardReferenceUniquenessTests(TestCase):
+    def setUp(self):
+        """Set up the common variables for the test cases."""
+        self.award1 = {"award_reference": "AWARD-0001"}
+        self.award2 = {"award_reference": "AWARD-2219"}
+        self.award3 = {"award_reference": "AWARD-2910"}
+
+    def _make_federal_awards(self, award_refs) -> dict:
+        return {
+            "FederalAwards": {
+                "federal_awards": award_refs,
+            }
+        }
+
+    def _make_sac(self, award_refs) -> SingleAuditChecklist:
+        sac = baker.make(SingleAuditChecklist)
+        sac.federal_awards = self._make_federal_awards(award_refs)
+        return sac
+
+    def test_unique_award_references_raise_no_errors(self):
+        """When all award references are unique, no errors should be raised."""
+        sac = self._make_sac([self.award1, self.award2, self.award3])
+        errors = check_award_reference_uniqueness(sac_validation_shape(sac))
+        self.assertEqual(errors, [])
+
+    def test_duplicate_award_references_raise_errors(self):
+        """When there are duplicate award references, the appropriate errors should be raised."""
+        sac = self._make_sac(
+            [self.award1, self.award1, self.award3, self.award3, self.award3]
+        )
+        errors = errors = check_award_reference_uniqueness(sac_validation_shape(sac))
+        self.assertEqual(len(errors), 2)
+
+        errors = check_award_reference_uniqueness(sac_validation_shape(sac))
+        expected_error1 = err_duplicate_award_reference(self.award1["award_reference"])
+        expected_error2 = err_duplicate_award_reference(self.award3["award_reference"])
+        self.assertIn({"error": expected_error1}, errors)
+        self.assertIn({"error": expected_error2}, errors)