fix: resolve parser organizations from directory only

2026-06-07 16:28:57 +02:00
parent 6ba8fa8d88
commit 802257a757
19 changed files with 184 additions and 83 deletions
--- a/tests/apps/organizations/test_resolver.py
+++ b/tests/apps/organizations/test_resolver.py
@@ -1,6 +1,7 @@
 """Tests for organization directory resolver."""

 from django.test import TestCase
+from django.utils import timezone
 from organizations.models import Organization
 from organizations.resolver import OrganizationDirectoryResolver

@@ -8,15 +9,20 @@ from organizations.resolver import OrganizationDirectoryResolver
 class OrganizationDirectoryResolverTest(TestCase):
    """Checks matching parser rows to the organization directory."""

+    @staticmethod
+    def _directory_organization(**kwargs) -> Organization:
+        kwargs.setdefault("directory_imported_at", timezone.now())
+        return Organization.objects.create(**kwargs)
+
    def test_resolves_by_rn_okpo_and_exact_pairs(self):
-        by_rn = Organization.objects.create(name="RN", rn=100)
-        by_okpo = Organization.objects.create(name="OKPO", okpo="001")
-        by_inn_kpp = Organization.objects.create(
+        by_rn = self._directory_organization(name="RN", rn=100)
+        by_okpo = self._directory_organization(name="OKPO", okpo="001")
+        by_inn_kpp = self._directory_organization(
            name="INN KPP",
            inn="7701001001",
            kpp="770101001",
        )
-        by_ogrn_kpp = Organization.objects.create(
+        by_ogrn_kpp = self._directory_organization(
            name="OGRN KPP",
            ogrn="1027700100001",
            kpp="770101002",
@@ -54,13 +60,13 @@ class OrganizationDirectoryResolverTest(TestCase):
        )

    def test_duplicate_inn_ogrn_uses_single_head(self):
-        head = Organization.objects.create(
+        head = self._directory_organization(
            name="Head",
            inn="7701001001",
            ogrn="1027700100001",
            is_branch=False,
        )
-        Organization.objects.create(
+        self._directory_organization(
            name="Branch",
            inn="7701001001",
            ogrn="1027700100001",
@@ -78,13 +84,13 @@ class OrganizationDirectoryResolverTest(TestCase):
        self.assertEqual(result.organization, head)

    def test_duplicate_inn_ogrn_without_single_head_is_ambiguous(self):
-        Organization.objects.create(
+        self._directory_organization(
            name="Head 1",
            inn="7701001001",
            ogrn="1027700100001",
            is_branch=False,
        )
-        Organization.objects.create(
+        self._directory_organization(
            name="Head 2",
            inn="7701001001",
            ogrn="1027700100001",
@@ -100,3 +106,20 @@ class OrganizationDirectoryResolverTest(TestCase):

        self.assertEqual(result.status, "ambiguous")
        self.assertIsNone(result.organization)
+
+    def test_ignores_legacy_organization_matches(self):
+        Organization.objects.create(
+            name="Legacy",
+            inn="7701001001",
+            kpp="770101001",
+        )
+
+        result = OrganizationDirectoryResolver.resolve(
+            OrganizationDirectoryResolver.identity(
+                inn="7701001001",
+                kpp="770101001",
+            )
+        )
+
+        self.assertEqual(result.status, "unmatched")
+        self.assertIsNone(result.organization)
--- a/tests/apps/organizations/test_source_backfill.py
+++ b/tests/apps/organizations/test_source_backfill.py
@@ -8,6 +8,7 @@ from apps.parsers.models import (
 )
 from django.core.management import call_command
 from django.test import TestCase
+from django.utils import timezone
 from organizations.models import (
    DefenseSupplierExtension,
    FinancialIndicatorsExtension,
@@ -28,8 +29,13 @@ from tests.apps.parsers.factories import (
 class OrganizationSourceBackfillServiceTest(TestCase):
    """Checks idempotent migration from legacy source records."""

+    @staticmethod
+    def _directory_organization(**kwargs) -> Organization:
+        kwargs.setdefault("directory_imported_at", timezone.now())
+        return Organization.objects.create(**kwargs)
+
    def test_backfills_inspection_records_into_planned_inspection_extension(self):
-        organization = Organization.objects.create(
+        organization = self._directory_organization(
            name='ООО "Проверка"',
            inn="7707083801",
            ogrn="1027700132001",
@@ -71,7 +77,7 @@ class OrganizationSourceBackfillServiceTest(TestCase):
        self.assertEqual(OrganizationSourceRecord.objects.count(), 1)

    def test_backfill_payload_serializes_registry_organization_fk_as_id(self):
-        organization = Organization.objects.create(
+        organization = self._directory_organization(
            name='ООО "FK Payload"',
            inn="7707083815",
            ogrn="1027700132015",
@@ -94,7 +100,7 @@ class OrganizationSourceBackfillServiceTest(TestCase):
        self.assertEqual(record.payload["registry_organization"], str(organization.pk))

    def test_backfills_financial_report_lines(self):
-        organization = Organization.objects.create(
+        organization = self._directory_organization(
            name='ООО "Финансы"',
            inn="7707083802",
            ogrn="1027700132002",
@@ -132,7 +138,7 @@ class OrganizationSourceBackfillServiceTest(TestCase):
        self.assertEqual(record.financial_lines.get().period_end, 200)

    def test_backfills_generic_defense_supplier_records(self):
-        organization = Organization.objects.create(
+        organization = self._directory_organization(
            name='ООО "ГОЗ"',
            inn="7707083803",
        )
@@ -189,7 +195,7 @@ class OrganizationSourceBackfillServiceTest(TestCase):
        )

    def test_management_command_runs_source_backfill(self):
-        organization = Organization.objects.create(
+        organization = self._directory_organization(
            name='ООО "Команда"',
            inn="7707083804",
            ogrn="1027700132004",
--- a/tests/apps/organizations/test_source_ingestion.py
+++ b/tests/apps/organizations/test_source_ingestion.py
@@ -4,6 +4,7 @@ from decimal import Decimal

 from apps.parsers.models import FinancialReport, GenericParserRecord, ParserLoadLog
 from django.test import TestCase
+from django.utils import timezone
 from organizations.models import (
    DefenseSupplierExtension,
    FinancialIndicatorsExtension,
@@ -22,8 +23,13 @@ from organizations.source_ingestion import (
 class OrganizationSourceIngestionServiceTest(TestCase):
    """Checks runtime parser writes bypass legacy parser record tables."""

+    @staticmethod
+    def _directory_organization(**kwargs) -> Organization:
+        kwargs.setdefault("directory_imported_at", timezone.now())
+        return Organization.objects.create(**kwargs)
+
    def test_save_generic_records_writes_source_records_without_legacy_rows(self):
-        Organization.objects.create(
+        self._directory_organization(
            name='ООО "ГОЗ"',
            inn="7707083803",
        )
@@ -70,7 +76,7 @@ class OrganizationSourceIngestionServiceTest(TestCase):
        self.assertEqual(record.legacy_pk, "")

    def test_save_records_is_idempotent_by_source_external_id(self):
-        Organization.objects.create(
+        self._directory_organization(
            name='ООО "Идемпотентность"',
            inn="7707083810",
        )
@@ -140,7 +146,7 @@ class OrganizationSourceIngestionServiceTest(TestCase):
        self.assertFalse(name_only.source_extensions.exists())

    def test_save_records_does_not_fill_missing_identity_on_resolved_organization(self):
-        partial = Organization.objects.create(
+        partial = self._directory_organization(
            name="Acme Security",
            inn="7713497980",
            ogrn="1237700253306",
@@ -168,14 +174,14 @@ class OrganizationSourceIngestionServiceTest(TestCase):
        self.assertEqual(record.extension.organization, partial)

    def test_save_records_uses_single_head_organization_for_duplicate_inn_ogrn(self):
-        head = Organization.objects.create(
+        head = self._directory_organization(
            name="Head",
            inn="7713497980",
            ogrn="1237700253306",
            filial=".F.",
            is_branch=False,
        )
-        Organization.objects.create(
+        self._directory_organization(
            name="Branch",
            inn="7713497980",
            ogrn="1237700253306",
@@ -202,14 +208,14 @@ class OrganizationSourceIngestionServiceTest(TestCase):
        self.assertEqual(record.extension.organization, head)

    def test_save_records_skips_ambiguous_duplicate_inn_ogrn(self):
-        Organization.objects.create(
+        self._directory_organization(
            name="Head 1",
            inn="7713497980",
            ogrn="1237700253306",
            filial=".F.",
            is_branch=False,
        )
-        Organization.objects.create(
+        self._directory_organization(
            name="Head 2",
            inn="7713497980",
            ogrn="1237700253306",
@@ -259,7 +265,7 @@ class OrganizationSourceIngestionServiceTest(TestCase):
        self.assertFalse(Organization.objects.filter(inn="7720525156").exists())

    def test_save_records_does_not_add_ogrip_to_existing_legal_entity(self):
-        organization = Organization.objects.create(
+        organization = self._directory_organization(
            name='ООО "Металл-Завод"',
            inn="7720525156",
            ogrn="1057746645395",
@@ -286,7 +292,7 @@ class OrganizationSourceIngestionServiceTest(TestCase):
        self.assertEqual(organization.ogrip, "")

    def test_save_financial_report_writes_financial_lines_without_legacy_report(self):
-        Organization.objects.create(
+        self._directory_organization(
            name='ООО "Финансовый отчет"',
            ogrn="1027700132002",
        )
--- a/tests/apps/organizations/test_tasks.py
+++ b/tests/apps/organizations/test_tasks.py
@@ -6,6 +6,7 @@ from apps.parsers.models import ParserLoadLog
 from django.apps import apps as django_apps
 from django.core.cache import cache
 from django.test import TestCase
+from django.utils import timezone
 from django_celery_beat.models import PeriodicTask
 from organizations.cache import get_organization_api_cache_version
 from organizations.models import (
@@ -24,8 +25,13 @@ from tests.apps.parsers.factories import IndustrialCertificateRecordFactory
 class OrganizationSourceBackfillTasksTest(TestCase):
    """Checks Celery tasks that maintain API v2 organization source extensions."""

+    @staticmethod
+    def _directory_organization(**kwargs) -> Organization:
+        kwargs.setdefault("directory_imported_at", timezone.now())
+        return Organization.objects.create(**kwargs)
+
    def test_backfill_all_task_rebuilds_sources_and_invalidates_api_cache(self):
-        organization = Organization.objects.create(
+        organization = self._directory_organization(
            name='ООО "Источник"',
            inn="7800000401",
            ogrn="1027700144401",
@@ -57,7 +63,7 @@ class OrganizationSourceBackfillTasksTest(TestCase):
        )

    def test_backfill_parser_batch_task_limits_source_and_batch(self):
-        organization = Organization.objects.create(
+        organization = self._directory_organization(
            name='ООО "Пакет источника"',
            inn="7800000402",
            ogrn="1027700144402",