Гудвилл: единицы ББ по словарю arm_statements

StatementsEngine.canonical_units — нормализация подписи единиц шаблонами пакета;
УК ДО переводится в рубли и при единицах, исправленных аналитиком вручную.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

api/src/services/goodwill_source.py

@@ -13,6 +13,9 @@ from src.domain.goodwill import Investment
 from src.domain.statement_aggregate import AggLine, aggregate_by_code
 from src.repositories import goodwill as repo
 from src.services.consolidated_osv import ConsolidatedOsvBuildResult, ConsolidatedOsvService
+from src.services.statements_engine import default_engine
+
+UNIT_FACTORS: dict[str | None, Decimal] = {"руб.": Decimal(1), "тыс. руб.": Decimal(1000), "млн руб.": Decimal(1000000)}
 
 
 @dataclass
@@ -26,7 +29,7 @@ class Sources:
 
 def capital_amounts(meta_id, units, columns, values, date) -> tuple[Decimal | None, Decimal | None]:
     selected = [c.id for c in columns if c.meta_id == meta_id and c.period_date == date]
-    factor = {"руб.": Decimal(1), "тыс. руб.": Decimal(1000), "млн руб.": Decimal(1000000)}.get(units)
+    factor = UNIT_FACTORS.get(default_engine().canonical_units(units))
     if len(selected) != 1 or factor is None:
         return None, None
     lines = {}

api/src/services/statements_engine.py

@@ -29,6 +29,7 @@ from arm_statements import (
     statements_version,
     status_input_from_contract,
 )
+from arm_statements.text import normalize_caption
 
 __all__ = [
     "ArmStatementsEngine",
@@ -62,6 +63,7 @@ class StatementsEngine(Protocol):
     def editions(self) -> tuple[Edition, ...]: ...
     def resolve_relative(self, kind: PeriodKind, effective_date: datetime.date | None) -> datetime.date | None: ...
     def tolerance(self) -> Decimal: ...
+    def canonical_units(self, text: str | None) -> str | None: ...
 
 
 class ArmStatementsEngine:
@@ -105,6 +107,13 @@ class ArmStatementsEngine:
     def tolerance(self) -> Decimal:
         return self._rules.checks.tolerance
 
+    def canonical_units(self, text: str | None) -> str | None:
+        """Единица формы по словарю пакета («тыс.руб.», «в тысячах рублей» → «тыс. руб.»); не узнали — None."""
+        if text is None:
+            return None
+        caption = normalize_caption(text)
+        return next((canonical for pattern, canonical in self._rules.units.spellings if pattern.search(caption)), None)
+
 
 @functools.cache
 def default_engine() -> StatementsEngine:

api/tests/fake_engine.py

@@ -58,6 +58,9 @@ class FakeEngine:
     def version(self) -> str | None:
         return self._version
 
+    def canonical_units(self, text: str | None) -> str | None:
+        return text
+
     def recompute(self, form: StoredForm) -> RecomputeResult:
         self.forms.append(form)
         return self.result

api/tests/test_goodwill_source.py

@@ -57,3 +57,29 @@ def test_capital_aggregates_duplicates_at_period_end() -> None:
     assert capital_amounts(mid, "тыс. руб.", columns, values, dt.date(2025, 9, 30)) == (Decimal("120000"), None)
     assert capital_amounts(mid, "тыс. руб.", columns, values, dt.date(2025, 6, 30)) == (None, None)
     assert capital_amounts(mid, None, columns, values, dt.date(2025, 9, 30)) == (None, None)
+
+
+@pytest.mark.parametrize(
+    ("units", "factor"),
+    [
+        ("тыс. руб.", Decimal(1000)),
+        ("тыс.руб.", Decimal(1000)),
+        ("в тысячах рублей", Decimal(1000)),
+        ("руб", Decimal(1)),
+        ("млн. руб.", Decimal(1000000)),
+    ],
+)
+def test_capital_units_spellings(units: str, factor: Decimal) -> None:
+    """Единицы, исправленные аналитиком вручную, читаются словарём единиц arm_statements."""
+    mid, col = UUID(int=1), UUID(int=2)
+    columns = [N(id=col, meta_id=mid, period_date=dt.date(2025, 9, 30), is_current=True)]
+    line = N(id=UUID(int=4), meta_id=mid, line_code="1310", row_index=0, line_status=StatementLineStatus.ok)
+    values = [(line, col, Decimal("7"))]
+    assert capital_amounts(mid, units, columns, values, dt.date(2025, 9, 30)) == (Decimal("7") * factor, None)
+
+
+def test_capital_unknown_units() -> None:
+    mid, col = UUID(int=1), UUID(int=2)
+    columns = [N(id=col, meta_id=mid, period_date=dt.date(2025, 9, 30), is_current=True)]
+    line = N(id=UUID(int=4), meta_id=mid, line_code="1310", row_index=0, line_status=StatementLineStatus.ok)
+    assert capital_amounts(mid, "штук", columns, [(line, col, Decimal("7"))], dt.date(2025, 9, 30)) == (None, None)