fix: exclude provider records from source exports

docs/source-record-export-matrix-ru.md

@@ -38,6 +38,9 @@ JavaScript. Ticket не попадает в URL и после первого з
 объединяет три таблицы, а поле `record_type` различает тип строки. Все строки
 содержат реквизиты организации, включая ОКПО.
 
+Записи технического внешнего поставщика в публичные файлы не включаются; в БД
+они сохраняются для работы интеграции и дедупликации.
+
 Физических XLSX-файлов может быть больше: по умолчанию один файл содержит не
 более 100 000 строк данных и получает суффикс `-part-001`, `-part-002` и далее.
 

src/apps/external_data/source_record_export.py

@@ -54,6 +54,11 @@ DEFAULT_XLSX_DATA_ROWS_PER_FILE = 100_000
 DEFAULT_DOWNLOAD_TICKET_TTL_SECONDS = 5 * 60
 SOURCE_RECORD_EXPORT_TICKET_CACHE_PREFIX = "external-data:source-record-exports:ticket"
 SOURCE_RECORD_EXPORT_TICKET_PATTERN = re.compile(r"[A-Za-z0-9_-]{43}")
+EXCLUDED_EXPORT_PROVIDER_PATTERN = re.compile(
+    r"(?<![A-Za-zА-Яа-яЁё0-9])(?:checko(?:\.ru)?|чекало|чекко|чеко)"
+    r"(?![A-Za-zА-Яа-яЁё0-9])",
+    flags=re.IGNORECASE,
+)
 
 ORGANIZATION_EXPORT_FIELDS = [
     "Наименование",
@@ -737,6 +742,8 @@ def _spool_source_group_rows(
         is_first_row = True
         for model_spec in source_spec.models:
             for record in _iter_source_model_records(model_spec):
+                if _is_excluded_export_provider_record(record, model_spec=model_spec):
+                    continue
                 row = _build_record_row(
                     record,
                     source_spec=source_spec,
@@ -911,6 +918,32 @@ def _build_record_row(
     return {key: _serialize_json_value(value) for key, value in row.items()}
 
 
+def _is_excluded_export_provider_record(
+    record: Any,
+    *,
+    model_spec: SourceModelExportSpec,
+) -> bool:
+    """Return whether a provider-backed record must stay out of public files."""
+    return any(
+        _contains_excluded_export_provider(getattr(record, field_name))
+        for field_name in model_spec.fields
+    )
+
+
+def _contains_excluded_export_provider(value: object) -> bool:
+    if isinstance(value, str):
+        return EXCLUDED_EXPORT_PROVIDER_PATTERN.search(value) is not None
+    if isinstance(value, dict):
+        return any(
+            _contains_excluded_export_provider(key)
+            or _contains_excluded_export_provider(item)
+            for key, item in value.items()
+        )
+    if isinstance(value, list | tuple):
+        return any(_contains_excluded_export_provider(item) for item in value)
+    return False
+
+
 def _serialize_flat_value(value: Any) -> str | int | float | bool:
     if value is None:
         return ""

tests/apps/external_data/test_source_record_export.py

@@ -16,6 +16,7 @@ from rest_framework import status
 from rest_framework.test import APITestCase
 
 from tests.apps.external_data.factories import (
+    BankruptcyProcedureFactory,
     FinancialReportFactory,
     FinancialReportLineFactory,
     IndustrialCertificateFactory,
@@ -106,6 +107,42 @@ class SourceRecordExportApiTest(APITestCase):
         financial_rows = json.loads(financial_path.read_text(encoding="utf-8"))
         self.assertEqual(financial_rows[0]["financial_lines"][0]["line_code"], "1600")
 
+    def test_generation_excludes_provider_records_and_keeps_okpo(self):
+        organization = OrganizationFactory.create(okpo="11223344")
+        excluded_record = BankruptcyProcedureFactory.create(
+            organization=organization,
+            external_id="checko-fedresurs:123",
+            source_url="https://api.checko.ru/v2/bankruptcy/123",
+        )
+        included_record = BankruptcyProcedureFactory.create(
+            organization=organization,
+            external_id="checkout-reference:456",
+            source_url="https://fedresurs.ru/message/456",
+        )
+
+        generation = build_source_record_export_artifacts()
+
+        self.assertEqual(generation.records_count, 1)
+        artifacts = [
+            artifact
+            for artifact in generation.artifacts
+            if artifact.source_group == "bankruptcy"
+        ]
+        self.assertEqual(
+            {artifact.file_format for artifact in artifacts}, {"csv", "xlsx", "json"}
+        )
+        for artifact in artifacts:
+            if artifact.file_format == "json":
+                exported_values = json.loads(artifact.path.read_text(encoding="utf-8"))
+            elif artifact.file_format == "csv":
+                exported_values = artifact.path.read_text(encoding="utf-8-sig")
+            else:
+                workbook = load_workbook(artifact.path, read_only=True)
+                exported_values = list(workbook["data"].iter_rows(values_only=True))
+            self.assertNotIn(str(excluded_record.id), str(exported_values))
+            self.assertIn(str(included_record.id), str(exported_values))
+            self.assertIn("11223344", str(exported_values))
+
     def test_management_command_bootstraps_first_generation(self):
         command_output = StringIO()