api/tests/test_mapping_golden.py

@@ -29,13 +29,14 @@ from sqlalchemy import select
 from sqlalchemy.ext.asyncio import AsyncSession
 
 from src.contract.result import RecognitionResult
+from src.contract.result_v2 import RecognitionResultV2
 from src.domain.osv_mapping import normalize_account_code
 from src.models.accounting import ChartOfAccount
 from src.models.enums import DocumentType, FileStatus, RecognitionStatus
 from src.models.files import Document, StoredFile
 from src.models.osv import OsvEntry
 from src.repositories.counterparties import CounterpartyIndex
-from src.services.result_mapping import map_file
+from src.services.result_mapping import _osv_documents, _split_duplicates, map_file
 from tests.factories import make_legal_entity, make_stored_file
 from tests.fixtures_helpers import golden_result
 
@@ -141,7 +142,7 @@ async def _form_problems(session: AsyncSession, file_id: uuid.UUID, raw: dict[st
     ]
 
 
-async def test_all_golden_snapshots_mapped(seeded: AsyncSession, fixtures_dir: Path) -> None:  # noqa: PLR0914
+async def test_all_golden_snapshots_mapped(seeded: AsyncSession, fixtures_dir: Path) -> None:  # noqa: PLR0914, PLR0915
     """«Все golden-снимки парсера» (не меньше 113) маппятся в БД, инварианты спеки на каждом документе, ≤ 60 с."""
     session = seeded
     snapshots = _snapshots(fixtures_dir)
@@ -155,7 +156,7 @@ async def test_all_golden_snapshots_mapped(seeded: AsyncSession, fixtures_dir: P
     assert chart, "план счетов seed пуст — сверка account_id ничего не проверит"
     index = await CounterpartyIndex.load(session)
 
-    mapped: list[tuple[Path, uuid.UUID, RecognitionResult]] = []
+    mapped: list[tuple[Path, uuid.UUID, RecognitionResult | RecognitionResultV2]] = []
     mapping_seconds = 0.0
     for path in snapshots:
         file, job = await make_stored_file(
@@ -174,7 +175,7 @@ async def test_all_golden_snapshots_mapped(seeded: AsyncSession, fixtures_dir: P
     problems: list[str] = []
     missing_codes: set[str] = set()
     documents_total = entries_total = with_account = with_kind = 0
-    for path, file_id, _ in mapped:
+    for path, file_id, result in mapped:
         name = str(path.relative_to(fixtures_dir / GOLDEN_DIR))
         raw = json.loads(path.read_text("utf-8"))
         documents = {
@@ -189,7 +190,14 @@ async def test_all_golden_snapshots_mapped(seeded: AsyncSession, fixtures_dir: P
             ).scalars()
         }
         problems.extend(await _form_problems(session, file_id, raw, name))
-        sheets = [sheet for sheet in raw["documents"] if sheet["document_type"] == "OSV"]
+        # D-7: повтор счёта/периода в одном файле не становится отдельным документом
+        # (services/result_mapping._split_duplicates) — синтетика s48/s49/s50/s53 (tp#81, AURORA-1717/1732) намеренно
+        # кладёт несколько листов под один счёт в одном файле, проверяя распознавание разных названий ведомости.
+        kept, _duplicate_warnings, _warnings = _split_duplicates(_osv_documents(result))
+        kept_locators = {sheet.locator for sheet in kept}
+        sheets = [
+            sheet for sheet in raw["documents"] if sheet["document_type"] == "OSV" and sheet["locator"] in kept_locators
+        ]
         if set(documents) != {sheet["locator"] for sheet in sheets}:
             problems.append(f"{name}: документы {sorted(documents)} не совпадают с листами снимка")
             continue