Replacing missing ZIPCODE, CPAZIPCODE in historical audits

backend/audit/intakelib/checks/check_gsa_migration_keyword.py

@@ -33,6 +33,7 @@ def check_for_gsa_migration_keyword(ir):
         "contains_chart_or_table",
         "is_minimis_rate_used",
         "compliance_requirement",
+        "secondary_auditor_address_zipcode",
         "secondary_auditor_name",
         "secondary_auditor_address_state",
     ]

backend/audit/test_schemas.py

@@ -257,7 +257,7 @@ def test_invalid_zip(self):
 
                     with self.assertRaisesRegex(
                         exceptions.ValidationError,
-                        "does not match",
+                        "is not valid under any of the given schemas",
                         msg=f"ValidationError not raised with zip = {bad_zip}",
                     ):
                         validate(instance, schema)
@@ -290,7 +290,7 @@ def test_invalid_zip_plus_4(self):
 
                     with self.assertRaisesRegex(
                         exceptions.ValidationError,
-                        "does not match",
+                        "is not valid under any of the given schemas",
                         msg=f"ValidationError not raised with zip = {bad_zip}",
                     ):
                         validate(instance, schema)

backend/audit/validators.py

@@ -250,6 +250,8 @@ def validate_use_of_gsa_migration_keyword(general_information, is_data_migration
         general_information.get("auditee_uei", ""),
         general_information.get("ein", ""),
         general_information.get("auditor_ein", ""),
+        general_information.get("auditee_zip", ""),
+        general_information.get("auditor_zip", ""),
     ]:
         raise ValidationError(
             _(f"{settings.GSA_MIGRATION} not permitted outside of migrations"),

backend/census_historical_migration/sac_general_lib/general_information.py

@@ -441,6 +441,32 @@ def xform_replace_empty_or_invalid_auditee_ein_with_gsa_migration(general_inform
     return general_information
 
 
+def xform_replace_empty_zips(general_information):
+    """Replaces empty auditor and auditee zipcodes with GSA Migration keyword"""
+    # Transformation recorded.
+    if not general_information.get("auditor_zip"):
+        general_information["auditor_zip"] = settings.GSA_MIGRATION
+        track_transformations(
+            "CPAZIPCODE",
+            "",
+            "auditor_zip",
+            general_information["auditor_zip"],
+            "xform_replace_empty_zips",
+        )
+
+    if not general_information.get("auditee_zip"):
+        general_information["auditee_zip"] = settings.GSA_MIGRATION
+        track_transformations(
+            "ZIPCODE",
+            "",
+            "auditee_zip",
+            general_information["auditee_zip"],
+            "xform_replace_empty_zips",
+        )
+
+    return general_information
+
+
 def general_information(audit_header):
     """Generates general information JSON."""
     xform_update_multiple_eins_flag(audit_header)
@@ -458,6 +484,7 @@ def general_information(audit_header):
         xform_replace_empty_auditee_email,
         xform_replace_empty_or_invalid_auditor_ein_with_gsa_migration,
         xform_replace_empty_or_invalid_auditee_ein_with_gsa_migration,
+        xform_replace_empty_zips,
     ]
 
     for transform in transformations:

backend/census_historical_migration/test_general_information_xforms.py

@@ -19,6 +19,7 @@
     xform_replace_empty_or_invalid_auditee_uei_with_gsa_migration,
     xform_replace_empty_or_invalid_auditor_ein_with_gsa_migration,
     xform_replace_empty_or_invalid_auditee_ein_with_gsa_migration,
+    xform_replace_empty_zips,
 )
 from .exception_utils import (
     DataMigrationError,
@@ -431,3 +432,25 @@ def test_auditee_ein_empty_replaced(self):
             result = xform_replace_empty_or_invalid_auditee_ein_with_gsa_migration(info)
             mock_track.assert_called_once()
             self.assertEqual(result["ein"], settings.GSA_MIGRATION)
+
+
+class TestXformReplaceEmptyAuditorZip(SimpleTestCase):
+    def test_empty_auditor_zip(self):
+        """Test that an empty auditor_zip and auditee_zip are replaced with 'GSA_MIGRATION'"""
+        input_data = {
+            "auditee_zip": "",
+            "auditor_zip": "",
+        }
+        expected_output = {
+            "auditee_zip": settings.GSA_MIGRATION,
+            "auditor_zip": settings.GSA_MIGRATION,
+        }
+        self.assertEqual(xform_replace_empty_zips(input_data), expected_output)
+
+    def test_non_empty_auditor_zip(self):
+        """Test that a non-empty auditor_zip and auditee_zip remain unchanged"""
+        input_data = {
+            "auditee_zip": "10108",
+            "auditor_zip": "12345",
+        }
+        self.assertEqual(xform_replace_empty_zips(input_data), input_data)

backend/census_historical_migration/test_secondary_auditors_xforms.py

@@ -3,11 +3,12 @@
 
 from .workbooklib.secondary_auditors import (
     xform_address_state,
+    xform_address_zipcode,
     xform_cpafirmname,
 )
 
 
-class TestXformSecondaryAuditors(SimpleTestCase):
+class TestXformSecondaryAuditorsState(SimpleTestCase):
     class SecondaryAuditor:
         def __init__(self, state):
             self.CPASTATE = state
@@ -33,6 +34,32 @@ def test_missing_address_state(self):
         self.assertEqual(secondary_auditors[0].CPASTATE, settings.GSA_MIGRATION)
 
 
+class TestXformSecondaryAuditorsZipcode(SimpleTestCase):
+    class SecondaryAuditor:
+        def __init__(self, zipcode):
+            self.CPAZIPCODE = zipcode
+
+    def test_normal_address_zipcode(self):
+        # Setup specific test data
+        secondary_auditors = [
+            self.SecondaryAuditor("10108"),
+        ]
+
+        xform_address_zipcode(secondary_auditors)
+
+        self.assertEqual(secondary_auditors[0].CPAZIPCODE, "10108")
+
+    def test_missing_address_zipcode(self):
+        # Setup specific test data
+        secondary_auditors = [
+            self.SecondaryAuditor(""),
+        ]
+
+        xform_address_zipcode(secondary_auditors)
+
+        self.assertEqual(secondary_auditors[0].CPAZIPCODE, settings.GSA_MIGRATION)
+
+
 class TestXformCpaFirmName(SimpleTestCase):
     class MockSecondaryAuditorHeader:
         def __init__(

backend/census_historical_migration/transforms/xform_remove_hyphen_and_pad_zip.py

@@ -1,16 +1,22 @@
 from ..exception_utils import DataMigrationError
 from .xform_string_to_string import string_to_string
+from django.conf import settings
 
 
 def xform_remove_hyphen_and_pad_zip(zip):
     """
     Transform a ZIP code string by removing hyphen if present and adding a leading zero when necessary.
+    - If the ZIP is settings.GSA_MIGRATION, return it
     - If the ZIP code has hyphen, remove that character.
     - If the ZIP code has 4 or 8 digits, pads with a leading zero.
     - Returns the ZIP code if it has 5 digits or 9 digitis (after padding if needed).
     - Raises an error for other cases.
     """
     # Transformation to be documented.
+
+    if zip == settings.GSA_MIGRATION:
+        return zip
+
     strzip = string_to_string(zip)
     if "-" in strzip:
         parts = strzip.split("-")

backend/census_historical_migration/workbooklib/secondary_auditors.py

@@ -94,6 +94,33 @@ def xform_address_state(secondary_auditors):
         InspectionRecord.append_secondary_auditor_changes(change_records)
 
 
+def xform_address_zipcode(secondary_auditors):
+    """Default missing address_zipcode to GSA_MIGRATION"""
+    change_records = []
+    is_empty_address_zipcode_found = False
+
+    for secondary_auditor in secondary_auditors:
+        address_zipcode = string_to_string(secondary_auditor.CPAZIPCODE)
+        if not address_zipcode:
+            is_empty_address_zipcode_found = True
+            address_zipcode = settings.GSA_MIGRATION
+
+        track_transformations(
+            "CPAZIPCODE",
+            secondary_auditor.CPAZIPCODE,
+            "address_zipcode",
+            address_zipcode,
+            ["xform_address_zipcode"],
+            change_records,
+        )
+
+        secondary_auditor.CPAZIPCODE = address_zipcode
+
+    # See Transformation Method Change Recording comment at the top of this file
+    if change_records and is_empty_address_zipcode_found:
+        InspectionRecord.append_secondary_auditor_changes(change_records)
+
+
 def _get_secondary_auditors(dbkey, year):
     results = Caps.objects.filter(DBKEY=dbkey, AUDITYEAR=year)
 
@@ -142,6 +169,7 @@ def generate_secondary_auditors(audit_header, outfile):
         audit_header.DBKEY, audit_header.AUDITYEAR
     )
     xform_address_state(secondary_auditors)
+    xform_address_zipcode(secondary_auditors)
     xform_cpafirmname(secondary_auditors)
     map_simple_columns(wb, mappings, secondary_auditors)
     wb.save(outfile)

backend/schemas/output/sections/GeneralInformation.schema.json

@@ -162,8 +162,16 @@
          ]
       },
       "auditee_zip": {
-         "pattern": "^[0-9]{5}(?:[0-9]{4})?$",
-         "type": "string"
+         "oneOf": [
+            {
+               "pattern": "^[0-9]{5}(?:[0-9]{4})?$",
+               "type": "string"
+            },
+            {
+               "const": "GSA_MIGRATION",
+               "type": "string"
+            }
+         ]
       },
       "auditor_address_line_1": {
          "maxLength": 100,
@@ -298,8 +306,16 @@
          ]
       },
       "auditor_zip": {
-         "pattern": "^[0-9]{5}(?:[0-9]{4})?$",
-         "type": "string"
+         "oneOf": [
+            {
+               "pattern": "^[0-9]{5}(?:[0-9]{4})?$",
+               "type": "string"
+            },
+            {
+               "const": "GSA_MIGRATION",
+               "type": "string"
+            }
+         ]
       },
       "ein": {
          "oneOf": [

backend/schemas/output/sections/GeneralInformationRequired.schema.json

@@ -272,8 +272,16 @@
          ]
       },
       "auditee_zip": {
-         "pattern": "^[0-9]{5}(?:[0-9]{4})?$",
-         "type": "string"
+         "oneOf": [
+            {
+               "pattern": "^[0-9]{5}(?:[0-9]{4})?$",
+               "type": "string"
+            },
+            {
+               "const": "GSA_MIGRATION",
+               "type": "string"
+            }
+         ]
       },
       "auditor_address_line_1": {
          "maxLength": 100,
@@ -408,8 +416,16 @@
          ]
       },
       "auditor_zip": {
-         "pattern": "^[0-9]{5}(?:[0-9]{4})?$",
-         "type": "string"
+         "oneOf": [
+            {
+               "pattern": "^[0-9]{5}(?:[0-9]{4})?$",
+               "type": "string"
+            },
+            {
+               "const": "GSA_MIGRATION",
+               "type": "string"
+            }
+         ]
       },
       "ein": {
          "oneOf": [

backend/schemas/output/sections/SecondaryAuditors.schema.json

@@ -136,8 +136,16 @@
                         "type": "string"
                      },
                      "secondary_auditor_address_zipcode": {
-                        "pattern": "^[0-9]{5}(?:[0-9]{4})?$",
-                        "type": "string"
+                        "oneOf": [
+                           {
+                              "pattern": "^[0-9]{5}(?:[0-9]{4})?$",
+                              "type": "string"
+                           },
+                           {
+                              "const": "GSA_MIGRATION",
+                              "type": "string"
+                           }
+                        ]
                      },
                      "secondary_auditor_contact_email": {
                         "format": "email",

backend/schemas/source/sections/GeneralInformation.schema.jsonnet

@@ -43,7 +43,14 @@ Typechecks fields, but allows for empty data as well. Contains conditional check
       maxLength: 100,
     },
     auditee_state: Base.Enum.UnitedStatesStateAbbr,
-    auditee_zip: Base.Compound.Zip,
+    auditee_zip: {
+      oneOf: [
+        Base.Compound.Zip,
+        Types.string {
+          const: Base.Const.GSA_MIGRATION,
+        },
+      ],
+    },
 
     auditee_contact_name: Types.string {
       maxLength: 100,
@@ -91,7 +98,14 @@ Typechecks fields, but allows for empty data as well. Contains conditional check
       maxLength: 100,
     },
     auditor_state: Base.Enum.UnitedStatesStateAbbr,
-    auditor_zip: Base.Compound.Zip,
+    auditor_zip: {
+      oneOf: [
+        Base.Compound.Zip,
+        Types.string {
+          const: Base.Const.GSA_MIGRATION,
+        },
+      ],
+    },
 
     auditor_contact_name: Types.string {
       maxLength: 100,

backend/schemas/source/sections/SecondaryAuditors.schema.jsonnet

@@ -35,7 +35,14 @@ local SecondaryAuditorsEntry = {
         },
       ],
     },
-    secondary_auditor_address_zipcode: Base.Compound.Zip,
+    secondary_auditor_address_zipcode: {
+      oneOf: [
+        Base.Compound.Zip,
+        Types.string {
+          const: Base.Const.GSA_MIGRATION,
+        },
+      ],
+    },
     secondary_auditor_contact_name: Types.string,
     secondary_auditor_contact_title: Types.string,
     secondary_auditor_contact_phone: Base.Compound.UnitedStatesPhone,