fix: align source exports with mostovik
All checks were successful
CI/CD Pipeline / Code Quality Checks (push) Successful in 2m28s
CI/CD Pipeline / Run Tests (push) Successful in 4m47s
CI/CD Pipeline / Build and Push Dev Images (push) Successful in 32s
CI/CD Pipeline / Deploy Dev via Compose (push) Successful in 32s

This commit is contained in:
2026-08-04 12:24:09 +02:00
parent c481d35b8c
commit c2a09a0403
3 changed files with 291 additions and 36 deletions

View File

@@ -36,10 +36,12 @@ JavaScript. Ticket не попадает в URL и после первого з
Итого формируется 25 логических артефактов. Финансовые показатели всегда Итого формируется 25 логических артефактов. Финансовые показатели всегда
выгружаются в JSON с вложенным массивом `financial_lines`. Промышленная группа выгружаются в JSON с вложенным массивом `financial_lines`. Промышленная группа
объединяет три таблицы, а поле `record_type` различает тип строки. Все строки объединяет три таблицы, а поле `record_type` различает тип строки. Все строки
содержат реквизиты организации, включая ОКПО. используют тот же контракт, что и Mostovik: реквизиты организации, включая ОКПО,
общие поля записи источника и специфичные поля в `payload.*`.
Записи технического внешнего поставщика в публичные файлы не включаются; в БД Все записи включаются в публичные файлы, а техническое наименование внешнего
они сохраняются для работы интеграции и дедупликации. поставщика нейтрализуется. Исходные значения в БД сохраняются для работы
интеграции и дедупликации.
Физических XLSX-файлов может быть больше: по умолчанию один файл содержит не 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. более 100 000 строк данных и получает суффикс `-part-001`, `-part-002` и далее.

View File

@@ -54,7 +54,11 @@ DEFAULT_XLSX_DATA_ROWS_PER_FILE = 100_000
DEFAULT_DOWNLOAD_TICKET_TTL_SECONDS = 5 * 60 DEFAULT_DOWNLOAD_TICKET_TTL_SECONDS = 5 * 60
SOURCE_RECORD_EXPORT_TICKET_CACHE_PREFIX = "external-data:source-record-exports:ticket" 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}") SOURCE_RECORD_EXPORT_TICKET_PATTERN = re.compile(r"[A-Za-z0-9_-]{43}")
EXCLUDED_EXPORT_PROVIDER_PATTERN = re.compile( EXPORT_PROVIDER_URL_PATTERN = re.compile(
r"https?://(?:api\.)?checko\.ru[^\s\"'<>]*",
flags=re.IGNORECASE,
)
EXPORT_PROVIDER_NAME_PATTERN = re.compile(
r"(?<![A-Za-zА-Яа-яЁё0-9])(?:checko(?:\.ru)?|чекало|чекко|чеко)" r"(?<![A-Za-zА-Яа-яЁё0-9])(?:checko(?:\.ru)?|чекало|чекко|чеко)"
r"(?![A-Za-zА-Яа-яЁё0-9])", r"(?![A-Za-zА-Яа-яЁё0-9])",
flags=re.IGNORECASE, flags=re.IGNORECASE,
@@ -66,12 +70,19 @@ ORGANIZATION_EXPORT_FIELDS = [
"ОГРН", "ОГРН",
"КПП", "КПП",
"ОКПО", "ОКПО",
"organization",
] ]
SOURCE_RECORD_EXPORT_FIELDS = [ SOURCE_RECORD_EXPORT_FIELDS = [
"uid", "uid",
"source_group", "source_group",
"source",
"record_type", "record_type",
"external_id",
"title",
"record_date",
"amount",
"status",
"url",
"load_batch",
"created_at", "created_at",
"updated_at", "updated_at",
] ]
@@ -93,6 +104,18 @@ class SourceModelExportSpec:
model: Any model: Any
record_type: str record_type: str
fields: tuple[str, ...] fields: tuple[str, ...]
source: str
source_field: str | None = None
external_id_field: str | None = None
title_field: str | None = None
record_date_field: str | None = None
amount_field: str | None = None
status_field: str | None = None
url_field: str | None = None
load_batch_field: str | None = None
title_prefix: str = ""
payload_aliases: tuple[tuple[str, str], ...] = ()
payload_export_fields: tuple[str, ...] | None = None
prefetch_related: tuple[str, ...] = () prefetch_related: tuple[str, ...] = ()
@@ -123,6 +146,12 @@ SOURCE_GROUP_EXPORT_SPECS: dict[str, SourceGroupExportSpec] = {
"source", "source",
"error_message", "error_message",
), ),
source=FINANCIAL_SOURCE_GROUP,
source_field="source",
external_id_field="external_id",
title_field="file_name",
status_field="status",
load_batch_field="load_batch",
prefetch_related=("lines",), prefetch_related=("lines",),
), ),
), ),
@@ -144,6 +173,12 @@ SOURCE_GROUP_EXPORT_SPECS: dict[str, SourceGroupExportSpec] = {
"execution_end_date", "execution_end_date",
"purchase_name", "purchase_name",
), ),
source="government_procurements",
external_id_field="purchase_number",
title_field="purchase_name",
record_date_field="contract_date",
amount_field="contract_amount",
status_field="status",
), ),
), ),
), ),
@@ -161,6 +196,9 @@ SOURCE_GROUP_EXPORT_SPECS: dict[str, SourceGroupExportSpec] = {
"tnved_code", "tnved_code",
"registry_number", "registry_number",
), ),
source="industrial_production",
external_id_field="registry_number",
title_field="product_name",
), ),
SourceModelExportSpec( SourceModelExportSpec(
model=IndustrialCertificate, model=IndustrialCertificate,
@@ -173,11 +211,20 @@ SOURCE_GROUP_EXPORT_SPECS: dict[str, SourceGroupExportSpec] = {
"organisation_name", "organisation_name",
"ogrn", "ogrn",
), ),
source="industrial_production",
external_id_field="certificate_number",
title_field="organisation_name",
record_date_field="issue_date",
url_field="certificate_file_url",
payload_aliases=(("certificate_file_url", "url"),),
), ),
SourceModelExportSpec( SourceModelExportSpec(
model=ManufacturerRegistryEntry, model=ManufacturerRegistryEntry,
record_type="manufacturer_registry_entry", record_type="manufacturer_registry_entry",
fields=("full_legal_name", "inn", "ogrn", "address"), fields=("full_legal_name", "inn", "ogrn", "address"),
source="industrial_production",
external_id_field="ogrn",
title_field="full_legal_name",
), ),
), ),
), ),
@@ -196,6 +243,11 @@ SOURCE_GROUP_EXPORT_SPECS: dict[str, SourceGroupExportSpec] = {
"start_date", "start_date",
"status", "status",
), ),
source="planned_inspections",
external_id_field="registration_number",
title_field="control_authority",
record_date_field="start_date",
status_field="status",
), ),
), ),
), ),
@@ -214,6 +266,26 @@ SOURCE_GROUP_EXPORT_SPECS: dict[str, SourceGroupExportSpec] = {
"status", "status",
"source_url", "source_url",
), ),
source="bankruptcy",
external_id_field="external_id",
title_field="message_type",
record_date_field="message_date",
status_field="status",
url_field="source_url",
payload_aliases=(("source_url", "url"),),
payload_export_fields=(
"inn",
"url",
"ogrn",
"source",
"messages",
"load_batch",
"case_number",
"message_date",
"message_type",
"messages_count",
"organisation_name",
),
), ),
), ),
), ),
@@ -234,6 +306,14 @@ SOURCE_GROUP_EXPORT_SPECS: dict[str, SourceGroupExportSpec] = {
"status", "status",
"source_url", "source_url",
), ),
source="defense_suppliers",
source_field="registry_source",
external_id_field="external_id",
title_field="supplier_name",
record_date_field="included_at",
status_field="status",
url_field="source_url",
payload_aliases=(("source_url", "url"),),
), ),
), ),
), ),
@@ -251,6 +331,26 @@ SOURCE_GROUP_EXPORT_SPECS: dict[str, SourceGroupExportSpec] = {
"status", "status",
"decision_date", "decision_date",
), ),
source="arbitration",
external_id_field="case_number",
title_field="case_number",
record_date_field="decision_date",
status_field="status",
title_prefix="Дело ",
payload_aliases=(("court_name", "court"), ("party_role", "role")),
payload_export_fields=(
"inn",
"url",
"ogrn",
"role",
"court",
"source",
"status",
"load_batch",
"case_number",
"claim_amount",
"organisation_name",
),
), ),
), ),
), ),
@@ -269,6 +369,11 @@ SOURCE_GROUP_EXPORT_SPECS: dict[str, SourceGroupExportSpec] = {
"issued_at", "issued_at",
"expires_at", "expires_at",
), ),
source="security_registries",
external_id_field="external_id",
title_field="registry_name",
record_date_field="issued_at",
status_field="presence_status",
), ),
), ),
), ),
@@ -288,6 +393,15 @@ SOURCE_GROUP_EXPORT_SPECS: dict[str, SourceGroupExportSpec] = {
"salary_amount", "salary_amount",
"source_url", "source_url",
), ),
source="vacancies",
source_field="vacancy_source",
external_id_field="external_id",
title_field="title",
record_date_field="published_at",
amount_field="salary_amount",
status_field="status",
url_field="source_url",
payload_aliases=(("source_url", "url"),),
), ),
), ),
), ),
@@ -717,17 +831,19 @@ def _iter_source_model_records(
def _source_group_headers(source_spec: SourceGroupExportSpec) -> list[str]: def _source_group_headers(source_spec: SourceGroupExportSpec) -> list[str]:
model_fields: list[str] = [] payload_fields: list[str] = []
for model_spec in source_spec.models: for model_spec in source_spec.models:
for field_name in model_spec.fields: for field_name in _payload_export_field_names(model_spec):
if field_name not in model_fields: payload_field_name = f"payload.{field_name}"
model_fields.append(field_name) if payload_field_name not in payload_fields:
payload_fields.append(payload_field_name)
payload_fields.sort()
if source_spec.source_group == FINANCIAL_SOURCE_GROUP: if source_spec.source_group == FINANCIAL_SOURCE_GROUP:
model_fields.append("financial_lines") payload_fields.append("financial_lines")
return [ return [
*ORGANIZATION_EXPORT_FIELDS, *ORGANIZATION_EXPORT_FIELDS,
*SOURCE_RECORD_EXPORT_FIELDS, *SOURCE_RECORD_EXPORT_FIELDS,
*model_fields, *payload_fields,
] ]
@@ -742,8 +858,6 @@ def _spool_source_group_rows(
is_first_row = True is_first_row = True
for model_spec in source_spec.models: for model_spec in source_spec.models:
for record in _iter_source_model_records(model_spec): 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( row = _build_record_row(
record, record,
source_spec=source_spec, source_spec=source_spec,
@@ -899,13 +1013,28 @@ def _build_record_row(
"ОГРН": organization.ogrn, "ОГРН": organization.ogrn,
"КПП": organization.kpp, "КПП": organization.kpp,
"ОКПО": organization.okpo, "ОКПО": organization.okpo,
"organization": organization.id,
"uid": record.id, "uid": record.id,
"source_group": source_spec.source_group, "source_group": source_spec.source_group,
"source": _source_value(record, model_spec=model_spec),
"record_type": model_spec.record_type, "record_type": model_spec.record_type,
"external_id": _record_field_value(record, model_spec.external_id_field),
"title": _title_value(record, model_spec=model_spec),
"record_date": _record_field_value(record, model_spec.record_date_field),
"amount": _record_field_value(record, model_spec.amount_field),
"status": _record_field_value(record, model_spec.status_field),
"url": _record_field_value(record, model_spec.url_field),
"load_batch": _record_field_value(record, model_spec.load_batch_field),
"created_at": record.created_at, "created_at": record.created_at,
"updated_at": record.updated_at, "updated_at": record.updated_at,
**{field_name: getattr(record, field_name) for field_name in model_spec.fields}, **{
f"payload.{field_name}": _payload_field_value(
record,
organization=organization,
model_spec=model_spec,
field_name=field_name,
)
for field_name in _payload_export_field_names(model_spec)
},
} }
if source_spec.source_group == FINANCIAL_SOURCE_GROUP: if source_spec.source_group == FINANCIAL_SOURCE_GROUP:
row["financial_lines"] = [ row["financial_lines"] = [
@@ -915,33 +1044,78 @@ def _build_record_row(
} }
for line in record.lines.all() for line in record.lines.all()
] ]
return {key: _serialize_json_value(value) for key, value in row.items()} public_row = _sanitize_export_provider_value(row)
return {key: _serialize_json_value(value) for key, value in public_row.items()}
def _is_excluded_export_provider_record( def _source_value(record: Any, *, model_spec: SourceModelExportSpec) -> Any:
source_value = _record_field_value(record, model_spec.source_field)
return source_value or model_spec.source
def _title_value(record: Any, *, model_spec: SourceModelExportSpec) -> Any:
title_value = _record_field_value(record, model_spec.title_field)
if not title_value or not model_spec.title_prefix:
return title_value
return f"{model_spec.title_prefix}{title_value}"
def _payload_field_name(model_spec: SourceModelExportSpec, field_name: str) -> str:
return dict(model_spec.payload_aliases).get(field_name, field_name)
def _payload_export_field_names(model_spec: SourceModelExportSpec) -> tuple[str, ...]:
if model_spec.payload_export_fields is not None:
return model_spec.payload_export_fields
field_names = ["inn", "ogrn", "organisation_name", "source"]
for field_name in model_spec.fields:
payload_field_name = _payload_field_name(model_spec, field_name)
if payload_field_name not in field_names:
field_names.append(payload_field_name)
return tuple(field_names)
def _payload_field_value(
record: Any, record: Any,
*, *,
organization: Any,
model_spec: SourceModelExportSpec, model_spec: SourceModelExportSpec,
) -> bool: field_name: str,
"""Return whether a provider-backed record must stay out of public files.""" ) -> Any:
return any( if field_name == "inn":
_contains_excluded_export_provider(getattr(record, field_name)) return organization.inn
for field_name in model_spec.fields if field_name == "ogrn":
) return organization.ogrn
if field_name == "organisation_name":
return organization.full_name or organization.short_name or organization.name
if field_name == "source":
return _source_value(record, model_spec=model_spec)
source_field_by_alias = {
alias: source_field for source_field, alias in model_spec.payload_aliases
}
source_field = source_field_by_alias.get(field_name, field_name)
return getattr(record, source_field, None)
def _contains_excluded_export_provider(value: object) -> bool: def _record_field_value(record: Any, field_name: str | None) -> Any:
return getattr(record, field_name) if field_name else None
def _sanitize_export_provider_value(value: Any) -> Any:
"""Remove provider wording from files without changing stored source data."""
if isinstance(value, str): if isinstance(value, str):
return EXCLUDED_EXPORT_PROVIDER_PATTERN.search(value) is not None sanitized = EXPORT_PROVIDER_URL_PATTERN.sub("external-source", value)
return EXPORT_PROVIDER_NAME_PATTERN.sub("external-source", sanitized)
if isinstance(value, dict): if isinstance(value, dict):
return any( return {
_contains_excluded_export_provider(key) _sanitize_export_provider_value(key): _sanitize_export_provider_value(item)
or _contains_excluded_export_provider(item)
for key, item in value.items() for key, item in value.items()
) }
if isinstance(value, list | tuple): if isinstance(value, list | tuple):
return any(_contains_excluded_export_provider(item) for item in value) return [_sanitize_export_provider_value(item) for item in value]
return False return value
def _serialize_flat_value(value: Any) -> str | int | float | bool: def _serialize_flat_value(value: Any) -> str | int | float | bool:

View File

@@ -6,6 +6,10 @@ from io import BytesIO, StringIO
from tempfile import TemporaryDirectory from tempfile import TemporaryDirectory
from apps.external_data.source_record_export import ( from apps.external_data.source_record_export import (
ORGANIZATION_EXPORT_FIELDS,
SOURCE_GROUP_EXPORT_SPECS,
SOURCE_RECORD_EXPORT_FIELDS,
_source_group_headers,
build_source_record_export_artifacts, build_source_record_export_artifacts,
load_current_source_record_export_generation, load_current_source_record_export_generation,
) )
@@ -16,6 +20,7 @@ from rest_framework import status
from rest_framework.test import APITestCase from rest_framework.test import APITestCase
from tests.apps.external_data.factories import ( from tests.apps.external_data.factories import (
ArbitrationCaseFactory,
BankruptcyProcedureFactory, BankruptcyProcedureFactory,
FinancialReportFactory, FinancialReportFactory,
FinancialReportLineFactory, FinancialReportLineFactory,
@@ -74,6 +79,7 @@ class SourceRecordExportApiTest(APITestCase):
IndustrialCertificateFactory.create(organization=organization) IndustrialCertificateFactory.create(organization=organization)
ManufacturerRegistryEntryFactory.create(organization=organization) ManufacturerRegistryEntryFactory.create(organization=organization)
ProsecutorCheckFactory.create(organization=organization) ProsecutorCheckFactory.create(organization=organization)
arbitration_case = ArbitrationCaseFactory.create(organization=organization)
report = FinancialReportFactory.create(organization=organization) report = FinancialReportFactory.create(organization=organization)
FinancialReportLineFactory.create(report=report, line_code="1600") FinancialReportLineFactory.create(report=report, line_code="1600")
@@ -81,7 +87,13 @@ class SourceRecordExportApiTest(APITestCase):
self.assertEqual(generation.artifacts_count, 25) self.assertEqual(generation.artifacts_count, 25)
self.assertEqual(generation.files_count, 25) self.assertEqual(generation.files_count, 25)
self.assertEqual(generation.records_count, 5) self.assertEqual(generation.records_count, 6)
expected_prefix = [*ORGANIZATION_EXPORT_FIELDS, *SOURCE_RECORD_EXPORT_FIELDS]
for source_spec in SOURCE_GROUP_EXPORT_SPECS.values():
self.assertEqual(
_source_group_headers(source_spec)[: len(expected_prefix)],
expected_prefix,
)
industrial_path = next( industrial_path = next(
artifact.path artifact.path
for artifact in generation.artifacts for artifact in generation.artifacts
@@ -98,6 +110,12 @@ class SourceRecordExportApiTest(APITestCase):
}, },
) )
self.assertEqual({row["ОКПО"] for row in industrial_rows}, {"12345678"}) self.assertEqual({row["ОКПО"] for row in industrial_rows}, {"12345678"})
self.assertEqual(
list(industrial_rows[0])[: len(expected_prefix)],
expected_prefix,
)
self.assertNotIn("organization", industrial_rows[0])
self.assertIn("payload.product_name", industrial_rows[0])
financial_path = next( financial_path = next(
artifact.path artifact.path
@@ -107,7 +125,46 @@ class SourceRecordExportApiTest(APITestCase):
financial_rows = json.loads(financial_path.read_text(encoding="utf-8")) financial_rows = json.loads(financial_path.read_text(encoding="utf-8"))
self.assertEqual(financial_rows[0]["financial_lines"][0]["line_code"], "1600") self.assertEqual(financial_rows[0]["financial_lines"][0]["line_code"], "1600")
def test_generation_excludes_provider_records_and_keeps_okpo(self): arbitration_path = next(
artifact.path
for artifact in generation.artifacts
if artifact.source_group == "arbitration" and artifact.file_format == "json"
)
arbitration_row = json.loads(arbitration_path.read_text(encoding="utf-8"))[0]
self.assertEqual(arbitration_row["source"], "arbitration")
self.assertEqual(
arbitration_row["external_id"], arbitration_row["payload.case_number"]
)
self.assertEqual(
arbitration_row["title"],
f"Дело {arbitration_row['payload.case_number']}",
)
self.assertEqual(
arbitration_row["record_date"], arbitration_case.decision_date.isoformat()
)
self.assertEqual(arbitration_row["status"], arbitration_row["payload.status"])
self.assertEqual(arbitration_row["payload.role"], "defendant")
self.assertIn("payload.court", arbitration_row)
self.assertEqual(arbitration_row["payload.source"], "arbitration")
self.assertEqual(arbitration_row["payload.inn"], organization.inn)
self.assertEqual(
[key for key in arbitration_row if key.startswith("payload.")],
[
"payload.inn",
"payload.url",
"payload.ogrn",
"payload.role",
"payload.court",
"payload.source",
"payload.status",
"payload.load_batch",
"payload.case_number",
"payload.claim_amount",
"payload.organisation_name",
],
)
def test_generation_keeps_provider_records_but_hides_provider_mentions(self):
organization = OrganizationFactory.create(okpo="11223344") organization = OrganizationFactory.create(okpo="11223344")
excluded_record = BankruptcyProcedureFactory.create( excluded_record = BankruptcyProcedureFactory.create(
organization=organization, organization=organization,
@@ -122,7 +179,7 @@ class SourceRecordExportApiTest(APITestCase):
generation = build_source_record_export_artifacts() generation = build_source_record_export_artifacts()
self.assertEqual(generation.records_count, 1) self.assertEqual(generation.records_count, 2)
artifacts = [ artifacts = [
artifact artifact
for artifact in generation.artifacts for artifact in generation.artifacts
@@ -134,14 +191,36 @@ class SourceRecordExportApiTest(APITestCase):
for artifact in artifacts: for artifact in artifacts:
if artifact.file_format == "json": if artifact.file_format == "json":
exported_values = json.loads(artifact.path.read_text(encoding="utf-8")) exported_values = json.loads(artifact.path.read_text(encoding="utf-8"))
self.assertEqual(
[key for key in exported_values[0] if key.startswith("payload.")],
[
"payload.inn",
"payload.url",
"payload.ogrn",
"payload.source",
"payload.messages",
"payload.load_batch",
"payload.case_number",
"payload.message_date",
"payload.message_type",
"payload.messages_count",
"payload.organisation_name",
],
)
elif artifact.file_format == "csv": elif artifact.file_format == "csv":
exported_values = artifact.path.read_text(encoding="utf-8-sig") exported_values = artifact.path.read_text(encoding="utf-8-sig")
else: else:
workbook = load_workbook(artifact.path, read_only=True) workbook = load_workbook(artifact.path, read_only=True)
exported_values = list(workbook["data"].iter_rows(values_only=True)) exported_values = list(workbook["data"].iter_rows(values_only=True))
self.assertNotIn(str(excluded_record.id), str(exported_values)) serialized_values = str(exported_values)
self.assertIn(str(excluded_record.id), serialized_values)
self.assertIn(str(included_record.id), str(exported_values)) self.assertIn(str(included_record.id), str(exported_values))
self.assertIn("11223344", str(exported_values)) self.assertIn("11223344", str(exported_values))
self.assertNotRegex(
serialized_values,
r"(?i)(?<![a-zа-яё0-9])(?:checko(?:\.ru)?|чеко|чекало|чекко)"
r"(?![a-zа-яё0-9])",
)
def test_management_command_bootstraps_first_generation(self): def test_management_command_bootstraps_first_generation(self):
command_output = StringIO() command_output = StringIO()
@@ -190,7 +269,7 @@ class SourceRecordExportApiTest(APITestCase):
rows = list(workbook["data"].iter_rows(values_only=True)) rows = list(workbook["data"].iter_rows(values_only=True))
self.assertEqual( self.assertEqual(
rows[0][:6], rows[0][:6],
("Наименование", "ИНН", "ОГРН", "КПП", "ОКПО", "organization"), ("Наименование", "ИНН", "ОГРН", "КПП", "ОКПО", "uid"),
) )
self.assertEqual(rows[1][4], "87654321") self.assertEqual(rows[1][4], "87654321")