perf(organizations): speed up filtered API lists

tests/apps/organizations/test_api_v2.py

@@ -12,6 +12,7 @@ from django.test import override_settings
 from django.test.utils import CaptureQueriesContext
 from django.urls import reverse
 from organizations.cache import invalidate_organization_api_cache
+from organizations.filters import OrganizationFilter
 from organizations.models import Organization, OrganizationDataSnapshot
 from organizations.services import OrganizationDataSnapshotRefreshService
 from rest_framework import status
@@ -195,6 +196,46 @@ class OrganizationsApiV2Test(APITestCase):
             ],
         )
 
+    def test_list_default_uses_snapshot_summary_without_loading_full_data(self):
+        organization = Organization.objects.create(
+            name='ООО "Легкий снапшот"',
+            inn="7712345685",
+            kpp="771201008",
+            ogrn="1027700132203",
+        )
+        OrganizationDataSnapshot.objects.create(
+            organization=organization,
+            data={
+                "industrial": [{"id": index} for index in range(100)],
+                "fns_reports": [{"id": index} for index in range(50)],
+            },
+            registries=[],
+        )
+
+        with CaptureQueriesContext(connection) as captured:
+            response = self.client.get(
+                reverse("api_v2:organizations:organizations-list"),
+                {"inn": organization.inn, "has_registry": "false"},
+            )
+
+        self.assertEqual(response.status_code, status.HTTP_200_OK)
+        item = response.data["data"][0]
+        self.assertEqual(item["data"], {})
+        self.assertEqual(
+            item["data_sources"],
+            [
+                {"source": "fns_reports", "count": 50},
+                {"source": "industrial", "count": 100},
+            ],
+        )
+        full_snapshot_data_queries = [
+            query["sql"]
+            for query in captured
+            if "ORGANIZATIONS_DATA_SNAPSHOT" in query["sql"].upper()
+            and '"data"' in query["sql"]
+        ]
+        self.assertEqual(full_snapshot_data_queries, [])
+
     def test_list_returns_snapshot_data_when_sources_are_requested(self):
         organization = Organization.objects.create(
             name='ООО "Явные данные"',
@@ -740,6 +781,18 @@ class OrganizationsApiV2Test(APITestCase):
         self.assertEqual(no_registry.data["meta"]["pagination"]["total_count"], 1)
         self.assertEqual(no_registry.data["data"][0]["uid"], str(without_registry.uid))
 
+    def test_has_registry_filter_uses_uncorrelated_identity_subqueries(self):
+        filterset = OrganizationFilter(
+            data={"has_registry": "true"},
+            queryset=Organization.objects.all(),
+        )
+
+        self.assertTrue(filterset.is_valid(), filterset.errors)
+        sql = str(filterset.qs.query).upper()
+
+        self.assertIn(" IN ", sql)
+        self.assertNotIn("EXISTS", sql)
+
     def test_list_defaults_to_has_registry_true(self):
         with_registry = Organization.objects.create(
             name='ООО "Дефолтный реестр"',
@@ -852,6 +905,39 @@ class OrganizationsApiV2Test(APITestCase):
         self.assertEqual(has_fns.data["meta"]["pagination"]["total_count"], 1)
         self.assertEqual(has_fns.data["data"][0]["uid"], str(with_fns.uid))
 
+    def test_has_fns_reports_filter_does_not_preload_report_identities(self):
+        organization = Organization.objects.create(
+            name='ООО "Отчетность без preload"',
+            inn="7800000004",
+            kpp="780001004",
+            ogrn="1027700133004",
+        )
+        FinancialReport.objects.create(
+            external_id="fin-presence-no-preload",
+            ogrn=organization.ogrn,
+            file_name="fin_presence_no_preload.xlsx",
+            file_hash="d" * 64,
+            load_batch=1,
+            status=FinancialReport.Status.SUCCESS,
+            source=FinancialReport.SourceType.API,
+        )
+
+        with CaptureQueriesContext(connection) as captured:
+            response = self.client.get(
+                reverse("api_v2:organizations:organizations-list"),
+                {"has_fns_reports": "true", "has_registry": "false"},
+            )
+
+        self.assertEqual(response.status_code, status.HTTP_200_OK)
+        self.assertEqual(response.data["meta"]["pagination"]["total_count"], 1)
+        distinct_report_queries = [
+            query["sql"]
+            for query in captured
+            if "PARSERS_FINANCIAL_REPORT" in query["sql"].upper()
+            and "SELECT DISTINCT" in query["sql"].upper()
+        ]
+        self.assertEqual(distinct_report_queries, [])
+
     def test_limits_response_data_sources(self):
         organization = Organization.objects.create(
             name='ООО "Источник"',

tests/apps/organizations/test_services.py

@@ -99,6 +99,10 @@ class OrganizationDataSnapshotRefreshServiceTest(TestCase):
             snapshot.data["industrial"][0]["certificate_number"],
             "SNAPSHOT-BATCH-CERT",
         )
+        self.assertEqual(
+            snapshot.data_source_counts,
+            [{"source": "industrial", "count": 1}],
+        )
 
     def test_refresh_for_fns_batch_matches_by_ogrn(self):
         organization = Organization.objects.create(