fix: align media upload and gosedo records

2026-08-20 11:43:00 +02:00
parent d95a002d84
commit f2e36a8991
5 changed files with 207 additions and 15 deletions
--- a/src/apps/parsers/api_result_urls.py
+++ b/src/apps/parsers/api_result_urls.py
@@ -6,6 +6,7 @@ from apps.parsers.serializers import (
 )
 from apps.parsers.source_registry import PARSER_SOURCES, ParserSourceDescriptor
 from apps.parsers.views import (
+    MEDIA_NEWS_UPLOAD_FILE_PARAM,
    RESULT_DETAIL_PARAMS,
    RESULT_LIST_PARAMS,
    UPLOAD_FILE_PARAM,
@@ -123,7 +124,11 @@ def _upload_view(descriptor: ParserSourceDescriptor):
                f"Ручная загрузка файла для источника: {descriptor.data_scope}. "
                "Файл обрабатывается через Celery."
            ),
-            manual_parameters=[UPLOAD_FILE_PARAM],
+            manual_parameters=[
+                MEDIA_NEWS_UPLOAD_FILE_PARAM
+                if descriptor.key == "media_news"
+                else UPLOAD_FILE_PARAM
+            ],
            consumes=["multipart/form-data"],
            tags=[tag],
            responses={202: ParserRunResponseSerializer, 400: "Ошибка валидации"},
--- a/src/apps/parsers/views.py
+++ b/src/apps/parsers/views.py
@@ -19,6 +19,7 @@ from apps.core.serializers import BackgroundJobListSerializer
 from apps.core.services import BackgroundJobService
 from apps.parsers import tasks
 from apps.parsers.fns_upload import FNSUploadService
+from apps.parsers.media_news import MEDIA_NEWS_MAX_BYTES
 from apps.parsers.models import (
    VACANCY_RECORD_SOURCES,
    FinancialReport,
@@ -374,6 +375,13 @@ UPLOAD_FILE_PARAM = openapi.Parameter(
    type=openapi.TYPE_FILE,
    required=True,
 )
+MEDIA_NEWS_UPLOAD_FILE_PARAM = openapi.Parameter(
+    "file",
+    openapi.IN_FORM,
+    description="XLSX-файл новостей СМИ размером не более 25 МиБ",
+    type=openapi.TYPE_FILE,
+    required=True,
+)
 RESULT_LIST_PARAMS = [
    PAGE_PARAM,
    PAGE_SIZE_PARAM,
@@ -2793,19 +2801,30 @@ class ParserUploadView(APIView):
            )
        serializer = ParserUploadRequestSerializer(data=request.data)
        serializer.is_valid(raise_exception=True)
-        if source_key == "media_news" and not serializer.validated_data[
-            "file"
-        ].name.lower().endswith(".xlsx"):
-            return api_error_response(
-                [
-                    {
-                        "code": "invalid_file_type",
-                        "message": "Для новостей требуется XLSX",
-                    }
-                ],
-                status_code=status.HTTP_400_BAD_REQUEST,
-            )
        uploaded_file = serializer.validated_data["file"]
+        if source_key == "media_news":
+            if not uploaded_file.name.lower().endswith(".xlsx"):
+                return api_error_response(
+                    [
+                        {
+                            "code": "invalid_file_type",
+                            "field": "file",
+                            "message": "Для новостей требуется XLSX",
+                        }
+                    ],
+                    status_code=status.HTTP_400_BAD_REQUEST,
+                )
+            if uploaded_file.size > MEDIA_NEWS_MAX_BYTES:
+                return api_error_response(
+                    [
+                        {
+                            "code": "file_too_large",
+                            "field": "file",
+                            "message": "Размер XLSX для новостей не должен превышать 25 МиБ",
+                        }
+                    ],
+                    status_code=status.HTTP_400_BAD_REQUEST,
+                )
        file_path = _save_uploaded_parser_file(uploaded_file)
        run_serializer = ParserRunRequestSerializer(data={"file_path": file_path})
        run_serializer.is_valid(raise_exception=True)
--- a/src/organizations/views.py
+++ b/src/organizations/views.py
@@ -94,6 +94,8 @@ SOURCE_GROUP_VALUES = [choice.value for choice in SourceGroup]

 SOURCE_RECORD_ORDERING_FIELDS = (
    "record_date",
+    "record_type",
+    "external_id",
    "extension__organization__name",
    "extension__organization__full_name",
    "created_at",
@@ -105,6 +107,9 @@ SOURCE_RECORD_ORDERING_FIELDS = (
    "extension__organization__okpo",
    "status",
    "payload__attestation_status",
+    "payload__full_name",
+    "payload__medo_address",
+    "payload__registration_number",
    "payload__sentiment",
    "payload__news_source",
 )
@@ -225,11 +230,14 @@ SOURCE_RECORD_LIST_PARAMS = [
    _query_parameter(
        "ordering",
        description=(
-            "Сортировка по полям: record_date, extension__organization__name, "
+            "Сортировка по полям: record_date, record_type, external_id, "
+            "extension__organization__name, "
            "extension__organization__full_name, created_at, updated_at, title, "
            "uid, extension__organization__inn, extension__organization__ogrn, "
            "extension__organization__okpo, status, "
-            "payload__attestation_status, payload__sentiment, "
+            "payload__attestation_status, payload__full_name, "
+            "payload__medo_address, payload__registration_number, "
+            "payload__sentiment, "
            "payload__news_source. Для обратной сортировки используйте префикс -. "
            "Значения record_date с null сортируются последними."
        ),
--- a/tests/apps/organizations/test_api_v2_source_extensions.py
+++ b/tests/apps/organizations/test_api_v2_source_extensions.py
@@ -389,6 +389,113 @@ class OrganizationSourceExtensionsApiV2Test(APITestCase):
                    expected,
                )

+    def test_flat_gosedo_records_include_all_linked_types_and_record_sorting(self):
+        organization = create_frontend_organization(
+            name="Связанная организация",
+            inn="7707083828",
+            ogrn="1027700132028",
+        )
+        extension = ElectronicDocumentExchangeExtension.objects.create(
+            organization=organization,
+            title="Электронный документооборот",
+        )
+        for (
+            record_type,
+            external_id,
+            title,
+            full_name,
+            medo_address,
+            registration_number,
+        ) in (
+            (
+                "participant",
+                "GOSEDO-PARTICIPANT",
+                "Участник",
+                "Полное имя участника",
+                "MEDO-3",
+                "3",
+            ),
+            (
+                "operator",
+                "GOSEDO-OPERATOR",
+                "Оператор",
+                "Полное имя оператора",
+                "MEDO-1",
+                "1",
+            ),
+            (
+                "organizer",
+                "GOSEDO-ORGANIZER",
+                "Организатор",
+                "Полное имя организатора",
+                "MEDO-2",
+                "2",
+            ),
+        ):
+            OrganizationSourceRecord.objects.create(
+                extension=extension,
+                record_type=record_type,
+                source="gosedo_address_directory",
+                external_id=external_id,
+                title=title,
+                payload={
+                    "full_name": full_name,
+                    "medo_address": medo_address,
+                    "registration_number": registration_number,
+                },
+            )
+
+        endpoint = reverse("api_v2:organizations:organization-source-records-list")
+        params = {
+            "source_group": "electronic_document_exchange",
+            "source": "gosedo_address_directory",
+            "organization": str(organization.uid),
+        }
+        response = self.client.get(endpoint, params)
+
+        self.assertEqual(response.status_code, status.HTTP_200_OK)
+        self.assertEqual(
+            {item["record_type"] for item in response.data["data"]},
+            {"participant", "operator", "organizer"},
+        )
+
+        expected_by_ordering = {
+            "record_type": [
+                "GOSEDO-OPERATOR",
+                "GOSEDO-ORGANIZER",
+                "GOSEDO-PARTICIPANT",
+            ],
+            "external_id": [
+                "GOSEDO-OPERATOR",
+                "GOSEDO-ORGANIZER",
+                "GOSEDO-PARTICIPANT",
+            ],
+            "payload__full_name": [
+                "GOSEDO-OPERATOR",
+                "GOSEDO-ORGANIZER",
+                "GOSEDO-PARTICIPANT",
+            ],
+            "payload__medo_address": [
+                "GOSEDO-OPERATOR",
+                "GOSEDO-ORGANIZER",
+                "GOSEDO-PARTICIPANT",
+            ],
+            "payload__registration_number": [
+                "GOSEDO-OPERATOR",
+                "GOSEDO-ORGANIZER",
+                "GOSEDO-PARTICIPANT",
+            ],
+        }
+        for ordering, expected in expected_by_ordering.items():
+            with self.subTest(ordering=ordering):
+                response = self.client.get(endpoint, {**params, "ordering": ordering})
+
+                self.assertEqual(response.status_code, status.HTTP_200_OK)
+                self.assertEqual(
+                    [item["external_id"] for item in response.data["data"]],
+                    expected,
+                )
+
    def test_flat_media_records_support_frontend_sorting_fields(self):
        organization = create_frontend_organization(
            name="Media sorting organization",
--- a/tests/apps/parsers/test_gosedo_media_news.py
+++ b/tests/apps/parsers/test_gosedo_media_news.py
@@ -18,6 +18,7 @@ from apps.parsers.gosedo import (
    stable_gosedo_uid,
 )
 from apps.parsers.media_news import (
+    MEDIA_NEWS_MAX_BYTES,
    import_media_news,
    normalize_news_text,
    stable_media_external_id,
@@ -619,6 +620,58 @@ class MediaNewsPermissionsTest(APITestCase):
            task_id,
        )

+    def test_upload_rejects_non_xlsx_before_saving_or_queueing(self):
+        self.client.force_authenticate(self.admin)
+        with patch("apps.parsers.views._save_uploaded_parser_file") as save_file, patch(
+            "apps.parsers.tasks.parse_media_news.apply_async"
+        ) as apply_async:
+            response = self.client.post(
+                self.url,
+                {"file": SimpleUploadedFile("media.csv", b"headline")},
+                format="multipart",
+            )
+
+        self.assertEqual(response.status_code, status.HTTP_400_BAD_REQUEST)
+        self.assertEqual(response.data["errors"][0]["code"], "invalid_file_type")
+        save_file.assert_not_called()
+        apply_async.assert_not_called()
+
+    def test_upload_rejects_oversized_xlsx_before_saving_or_queueing(self):
+        self.client.force_authenticate(self.admin)
+        with patch("apps.parsers.views.MEDIA_NEWS_MAX_BYTES", 1), patch(
+            "apps.parsers.views._save_uploaded_parser_file"
+        ) as save_file, patch(
+            "apps.parsers.tasks.parse_media_news.apply_async"
+        ) as apply_async:
+            response = self.client.post(
+                self.url,
+                {"file": SimpleUploadedFile("media.xlsx", b"xx")},
+                format="multipart",
+            )
+
+        self.assertEqual(response.status_code, status.HTTP_400_BAD_REQUEST)
+        self.assertEqual(response.data["errors"][0]["code"], "file_too_large")
+        save_file.assert_not_called()
+        apply_async.assert_not_called()
+
+    def test_upload_accepts_xlsx_at_size_limit(self):
+        self.client.force_authenticate(self.admin)
+        with patch("apps.parsers.views.MEDIA_NEWS_MAX_BYTES", 1), patch(
+            "apps.parsers.views._save_uploaded_parser_file",
+            return_value="parser_uploads/media.xlsx",
+        ), patch(
+            "apps.parsers.tasks.parse_media_news.apply_async",
+            side_effect=lambda **kwargs: SimpleNamespace(id=kwargs["task_id"]),
+        ):
+            response = self.client.post(
+                self.url,
+                {"file": SimpleUploadedFile("media.xlsx", b"x")},
+                format="multipart",
+            )
+
+        self.assertEqual(response.status_code, status.HTTP_202_ACCEPTED)
+        self.assertEqual(MEDIA_NEWS_MAX_BYTES, 25 * 1024 * 1024)
+
    def test_gosedo_manual_run_is_admin_only_and_returns_task_ids(self):
        url = reverse(
            "api_v1:parsers:run-parser",