feat: align source data and nightly exports

src/apps/parsers/checko_collection.py

@@ -1,4 +1,4 @@
-"""Quota-safe monthly claims for organization lookups in Checko."""
+"""Quota-safe monthly claims for external organization lookups."""
 
 from datetime import date, datetime
 
@@ -24,7 +24,7 @@ def claim_monthly_collection(
     source: str,
     at: date | datetime | None = None,
 ) -> CheckoCollectionAttempt | None:
-    """Atomically claim this month's only allowed Checko collection attempt."""
+    """Atomically claim this month's only allowed external collection attempt."""
     attempt, created = CheckoCollectionAttempt.objects.get_or_create(
         organization_id=organization_id,
         source=source,

src/apps/parsers/clients/checko/client.py

@@ -346,14 +346,18 @@ class CheckoClient:
                     # Preserve the existing classification for non-quota HTTP
                     # errors while still recovering Checko quota metadata.
                     pass
-            logger.error("Checko HTTP request failed with status=%s", e.status_code)
+            logger.error(
+                "External provider HTTP request failed with status=%s", e.status_code
+            )
             raise CheckoConnectionError(
-                "Checko API request failed",
+                "External provider API request failed",
                 url=e.url,
             ) from e
         except Exception as e:
             logger.error("Connection error: %s", e)
-            raise CheckoConnectionError(f"Failed to connect to Checko API: {e}") from e
+            raise CheckoConnectionError(
+                f"Failed to connect to external provider API: {e}"
+            ) from e
 
         self._raise_api_error(data)
 

src/apps/parsers/migrations/0029_neutralize_external_collection_labels.py

@@ -0,0 +1,18 @@
+from django.db import migrations
+
+
+class Migration(migrations.Migration):
+    dependencies = [
+        ("parsers", "0028_registry_procurement_claims"),
+    ]
+
+    operations = [
+        migrations.AlterModelOptions(
+            name="checkocollectionattempt",
+            options={
+                "ordering": ["-period_month", "source", "organization_id"],
+                "verbose_name": "попытка сбора внешних данных",
+                "verbose_name_plural": "попытки сбора внешних данных",
+            },
+        ),
+    ]

src/apps/parsers/models.py

@@ -118,7 +118,7 @@ class ParserBatchSequence(TimestampMixin, models.Model):
 
 
 class CheckoCollectionAttempt(TimestampMixin, models.Model):
-    """Monthly Checko collection claim for one organization and source."""
+    """Monthly external collection claim for one organization and source."""
 
     class Source(models.TextChoices):
         ARBITRATION = "arbitration", _("Арбитражные дела")
@@ -160,8 +160,8 @@ class CheckoCollectionAttempt(TimestampMixin, models.Model):
 
     class Meta:
         db_table = "parsers_checko_collection_attempt"
-        verbose_name = _("попытка сбора Checko")
-        verbose_name_plural = _("попытки сбора Checko")
+        verbose_name = _("попытка сбора внешних данных")
+        verbose_name_plural = _("попытки сбора внешних данных")
         ordering = ["-period_month", "source", "organization_id"]
         constraints = [
             models.UniqueConstraint(

src/apps/parsers/source_registry.py

@@ -236,10 +236,10 @@ PARSER_SOURCES: dict[str, ParserSourceDescriptor] = {
         status="implemented",
         upstream_url="https://kad.arbitr.ru/",
         access_method="official_search_api",
-        parser_strategy="checko_legal_cases_by_inn_ogrn",
+        parser_strategy="external_legal_cases_by_inn_ogrn",
         source_notes=(
             "Поиск дел выполняется по ИНН/ОГРН активных организаций из реестров. "
-            "Checko отдаёт карточки со ссылками на КАД Арбитр."
+            "Внешний сервис данных отдаёт карточки со ссылками на КАД Арбитр."
         ),
         api_route="arbitration/cases",
     ),
@@ -258,7 +258,7 @@ PARSER_SOURCES: dict[str, ParserSourceDescriptor] = {
         parser_strategy="fedresurs_bankruptcy_search",
         source_notes=(
             "Официальный ЕФРСБ; может отдавать anti-bot challenge worker'ам. "
-            "Если официальный портал недоступен, используется Checko API по "
+            "Если официальный портал недоступен, используется внешний сервис данных по "
             "организациям из реестров. "
             "Ручная загрузка разрешена только для выгрузок, переданных Сергеем."
         ),

src/apps/parsers/tasks.py

@@ -470,7 +470,7 @@ def _fetch_fedresurs_bankruptcy_records(
     file_path: str | None,
     proxies: list[str] | None,
 ) -> list[GenericParserItem]:
-    """Загрузить банкротства: официальный портал, затем fallback через Checko."""
+    """Загрузить банкротства: официальный портал, затем внешний fallback."""
     official_error: Exception | None = None
     try:
         official_records = _fetch_structured_records(
@@ -482,14 +482,14 @@ def _fetch_fedresurs_bankruptcy_records(
         if official_records or file_url or file_path:
             return official_records
         logger.warning(
-            "Fedresurs official source returned no records, falling back to Checko"
+            "Fedresurs official source returned no records, using external fallback"
         )
     except Exception as exc:
         if file_url or file_path:
             raise
         official_error = exc
         logger.warning(
-            "Fedresurs official source failed, falling back to Checko: %s",
+            "Fedresurs official source failed, using external fallback: %s",
             exc,
         )
     records = _fetch_checko_bankruptcy_records(proxies=proxies)
@@ -498,12 +498,12 @@ def _fetch_fedresurs_bankruptcy_records(
     if official_error is None:
         raise ParserSourceSkipped(
             "fedresurs official source returned no bankruptcy records; "
-            "Checko fallback returned no bankruptcy records"
+            "external fallback returned no bankruptcy records"
         )
     if isinstance(official_error, HTTPClientError):
         raise ParserSourceSkipped(
             "fedresurs upstream is unavailable or blocked; "
-            "Checko fallback returned no bankruptcy records"
+            "external fallback returned no bankruptcy records"
         ) from official_error
     raise official_error
 
@@ -580,7 +580,7 @@ def _enrich_fstec_record_identities(
 
     logger.info(
         "FSTEC identity enrichment completed: enriched=%d ambiguous=%d "
-        "local_candidates=%d checko_candidates=%d",
+        "local_candidates=%d external_candidates=%d",
         enriched_count,
         ambiguous_count,
         len(local_candidates),
@@ -720,7 +720,7 @@ def _fstec_checko_identity_candidates(
                 )
             except CheckoError as exc:
                 logger.info(
-                    "Checko FSTEC identity lookup skipped for %s: %s",
+                    "External FSTEC identity lookup skipped for %s: %s",
                     applicant_name,
                     exc,
                 )
@@ -846,10 +846,10 @@ def _fetch_checko_bankruptcy_records(
     *,
     proxies: list[str] | None,
 ) -> list[GenericParserItem]:
-    """Получить ЕФРСБ-сообщения по организациям из наших реестров через Checko."""
+    """Получить ЕФРСБ-сообщения по организациям через внешний сервис данных."""
     api_key = getattr(settings, "CHECKO_API_KEY", "")
     if not api_key:
-        logger.warning("CHECKO_API_KEY is empty; Fedresurs fallback skipped")
+        logger.warning("External provider API key is empty; Fedresurs fallback skipped")
         return []
 
     limit = _resolve_lookup_limit(
@@ -861,7 +861,7 @@ def _fetch_checko_bankruptcy_records(
         default=FEDRESURS_CHECKO_FALLBACK_LIMIT,
     )
     if limit <= 0:
-        logger.info("Fedresurs Checko fallback is disabled by limit=%s", limit)
+        logger.info("Fedresurs external fallback is disabled by limit=%s", limit)
         return []
     targets = _active_registry_lookup_targets(
         limit=limit,
@@ -893,7 +893,7 @@ def _fetch_checko_bankruptcy_records(
         except CheckoRateLimitError as exc:
             finish_collection(attempt, records_count=0, error=exc)
             logger.warning(
-                "Checko bankruptcy fallback stopped: quota/rate limit reached "
+                "External bankruptcy fallback stopped: quota/rate limit reached "
                 "(status_code=%s)",
                 exc.status_code,
             )
@@ -901,7 +901,7 @@ def _fetch_checko_bankruptcy_records(
         except CheckoError as exc:
             finish_collection(attempt, records_count=0, error=exc)
             logger.info(
-                "Checko bankruptcy lookup skipped for target=%s: %s",
+                "External bankruptcy lookup skipped for target=%s: %s",
                 target.inn or target.ogrn,
                 exc,
             )
@@ -920,7 +920,7 @@ def _fetch_checko_bankruptcy_records(
             attempt,
             records_count=len(records) - records_before,
         )
-    logger.info("Fetched %d bankruptcy records through Checko fallback", len(records))
+    logger.info("Fetched %d bankruptcy records through external fallback", len(records))
     return records
 
 
@@ -931,7 +931,7 @@ def _checko_bankruptcy_items(
     fallback_ogrn: str,
     fallback_name: str,
 ) -> list[GenericParserItem]:
-    """Преобразовать банкротные сообщения Checko в generic records."""
+    """Преобразовать банкротные сообщения внешнего сервиса в generic records."""
     inn = str(getattr(company, "inn", "") or fallback_inn)
     ogrn = str(getattr(company, "ogrn", "") or fallback_ogrn)
     name = getattr(company, "short_name", None) or fallback_name
@@ -1084,14 +1084,16 @@ def _fetch_checko_arbitration_records(
     limit: int | None,
     proxies: list[str] | None,
 ) -> list[GenericParserItem]:
-    """Получить арбитражные дела по ИНН/ОГРН через Checko legal-cases API."""
+    """Получить арбитражные дела по ИНН/ОГРН через внешний API."""
     api_key = getattr(settings, "CHECKO_API_KEY", "")
     if not api_key:
-        raise ParserSourceSkipped("CHECKO_API_KEY is empty; arbitration parser skipped")
+        raise ParserSourceSkipped(
+            "External provider API key is empty; arbitration parser skipped"
+        )
 
     resolved_limit = _resolve_arbitration_limit(limit)
     if resolved_limit <= 0:
-        logger.info("Arbitration Checko parser is disabled by limit=%s", limit)
+        logger.info("Arbitration external parser is disabled by limit=%s", limit)
         return []
 
     subjects = _arbitration_subjects(resolved_limit)
@@ -1128,7 +1130,7 @@ def _fetch_checko_arbitration_records(
             failed_lookups += 1
             finish_collection(attempt, records_count=0, error=exc)
             logger.info(
-                "Checko arbitration lookup skipped for subject=%s: %s",
+                "External arbitration lookup skipped for subject=%s: %s",
                 _arbitration_subject_key(subject),
                 exc,
             )
@@ -1139,10 +1141,12 @@ def _fetch_checko_arbitration_records(
             )
 
     if attempted_lookups and failed_lookups == attempted_lookups and not records:
-        raise ParserSourceSkipped("Checko arbitration lookups failed for all subjects")
+        raise ParserSourceSkipped(
+            "External arbitration lookups failed for all subjects"
+        )
 
     logger.info(
-        "Fetched %d arbitration records through Checko for %d subjects",
+        "Fetched %d arbitration records through external service for %d subjects",
         len(records),
         len(subjects),
     )
@@ -1150,7 +1154,7 @@ def _fetch_checko_arbitration_records(
 
 
 def _checko_arbitration_item(case, *, subject: ArbitrationSubject) -> GenericParserItem:
-    """Преобразовать дело Checko в generic record."""
+    """Преобразовать дело внешнего сервиса в generic record."""
     case_number = getattr(case, "case_number", "") or ""
     filing_date = getattr(case, "filing_date", "") or ""
     role = _case_role_for_subject(case, subject)
@@ -1335,11 +1339,11 @@ def _fetch_checko_registry_inspections(
     limit: int | None,
     proxies: list[str] | None,
 ) -> list[ProverkiInspection]:
-    """Получить проверки по активным организациям из реестров через Checko."""
+    """Получить проверки по активным организациям через внешний сервис данных."""
     api_key = getattr(settings, "CHECKO_API_KEY", "")
     if not api_key:
         raise ParserSourceSkipped(
-            "CHECKO_API_KEY is empty; registry inspections parser skipped"
+            "External provider API key is empty; registry inspections parser skipped"
         )
 
     resolved_limit = _resolve_lookup_limit(
@@ -1351,7 +1355,9 @@ def _fetch_checko_registry_inspections(
         ),
     )
     if resolved_limit <= 0:
-        logger.info("Registry inspections Checko parser is disabled by limit=%s", limit)
+        logger.info(
+            "Registry inspections external parser is disabled by limit=%s", limit
+        )
         return []
 
     targets = _active_registry_lookup_targets(
@@ -1391,7 +1397,7 @@ def _fetch_checko_registry_inspections(
             failed_lookups += 1
             finish_collection(attempt, records_count=0, error=exc)
             logger.info(
-                "Checko inspections lookup skipped for target=%s: %s",
+                "External inspections lookup skipped for target=%s: %s",
                 target.inn or target.ogrn,
                 exc,
             )
@@ -1402,10 +1408,10 @@ def _fetch_checko_registry_inspections(
             )
 
     if attempted_lookups and failed_lookups == attempted_lookups and not records:
-        raise ParserSourceSkipped("Checko inspections lookups failed for all targets")
+        raise ParserSourceSkipped("External inspections lookups failed for all targets")
 
     logger.info(
-        "Fetched %d inspections through Checko for %d registry organizations",
+        "Fetched %d inspections through external service for %d registry organizations",
         len(records),
         len(targets),
     )
@@ -1488,7 +1494,7 @@ def _checko_unfair_supplier_items(
     company,
     target: RegistryLookupTarget,
 ) -> list[GenericParserItem]:
-    """Convert Checko НедобПостЗап values into supplier-bound source records."""
+    """Convert external НедобПостЗап values into supplier-bound source records."""
     company_inn = _normalize_identifier(getattr(company, "inn", ""))
     company_ogrn = _normalize_identifier(getattr(company, "ogrn", ""))
     if target.inn and company_inn != target.inn:
@@ -1563,10 +1569,12 @@ def _fetch_checko_unfair_supplier_records(  # noqa: C901
     proxies: list[str] | None,
     organization_ids: list[str] | None = None,
 ) -> list[GenericParserItem]:
-    """Fetch RNP entries through one Checko /company request per OПК organization."""
+    """Fetch RNP entries through one external request per OПК organization."""
     api_key = getattr(settings, "CHECKO_API_KEY", "")
     if not api_key:
-        raise ParserSourceSkipped("CHECKO_API_KEY is empty; RNP parser skipped")
+        raise ParserSourceSkipped(
+            "External provider API key is empty; RNP parser skipped"
+        )
 
     resolved_limit = _resolve_registry_enrichment_limit(limit)
     if resolved_limit <= 0:
@@ -1604,13 +1612,13 @@ def _fetch_checko_unfair_supplier_records(  # noqa: C901
             finish_collection(attempt, records_count=0, error=exc)
             failed_lookups += 1
             rate_limited = True
-            logger.warning("Checko RNP lookup stopped: quota/rate limit reached")
+            logger.warning("External RNP lookup stopped: quota/rate limit reached")
             break
         except CheckoError as exc:
             finish_collection(attempt, records_count=0, error=exc)
             failed_lookups += 1
             logger.info(
-                "Checko RNP lookup failed for target=%s: %s",
+                "External RNP lookup failed for target=%s: %s",
                 target.inn or target.ogrn,
                 exc,
             )
@@ -1628,7 +1636,7 @@ def _fetch_checko_unfair_supplier_records(  # noqa: C901
             )
 
     if not records and (rate_limited or failed_lookups == len(targets)):
-        raise ParserSourceSkipped("Checko RNP lookups failed for all targets")
+        raise ParserSourceSkipped("External RNP lookups failed for all targets")
     return records
 
 
@@ -1729,11 +1737,11 @@ def _fetch_checko_registry_contract_records(
     limit: int | None,
     proxies: list[str] | None,
 ) -> list[GenericParserItem]:
-    """Получить контракты по активным организациям из реестров через Checko."""
+    """Получить контракты по активным организациям через внешний сервис данных."""
     api_key = getattr(settings, "CHECKO_API_KEY", "")
     if not api_key:
         raise ParserSourceSkipped(
-            "CHECKO_API_KEY is empty; registry contracts parser skipped"
+            "External provider API key is empty; registry contracts parser skipped"
         )
 
     resolved_limit = _resolve_lookup_limit(
@@ -1745,7 +1753,7 @@ def _fetch_checko_registry_contract_records(
         ),
     )
     if resolved_limit <= 0:
-        logger.info("Registry contracts Checko parser is disabled by limit=%s", limit)
+        logger.info("Registry contracts external parser is disabled by limit=%s", limit)
         return []
 
     targets = _active_registry_lookup_targets(
@@ -1790,7 +1798,7 @@ def _fetch_checko_registry_contract_records(
                 failed_lookups += 1
                 target_failures.append(exc)
                 logger.info(
-                    "Checko contracts lookup skipped for target=%s law=%s: %s",
+                    "External contracts lookup skipped for target=%s law=%s: %s",
                     target.inn or target.ogrn,
                     law.value,
                     exc,
@@ -1803,10 +1811,10 @@ def _fetch_checko_registry_contract_records(
 
     expected_lookups = attempted_lookups * 2
     if expected_lookups and failed_lookups == expected_lookups and not records:
-        raise ParserSourceSkipped("Checko contracts lookups failed for all targets")
+        raise ParserSourceSkipped("External contracts lookups failed for all targets")
 
     logger.info(
-        "Fetched %d contracts through Checko for %d registry organizations",
+        "Fetched %d contracts through external service for %d registry organizations",
         len(records),
         len(targets),
     )
@@ -3374,7 +3382,7 @@ def parse_unfair_suppliers(
     organization_ids: list[str] | None = None,
     requested_by_id: int | None = None,
 ) -> dict:
-    """Checko RNP lookup by default; explicit files remain a manual tool."""
+    """External RNP lookup by default; explicit files remain a manual tool."""
     proxies = _resolve_proxies(proxies)
     if file_url or file_path:
 
@@ -3478,7 +3486,7 @@ def parse_registry_inspections(
     proxies: list[str] | None = None,
     requested_by_id: int | None = None,
 ) -> dict:
-    """Lookup проверок по активным организациям из реестров через Checko."""
+    """Lookup проверок по активным организациям через внешний сервис данных."""
     proxies = _resolve_proxies(proxies)
     return _run_inspection_parser(
         self,