api/src/domain/quality_facts.py

@@ -10,7 +10,7 @@
 Объекты строк, значений и нарушений читаются атрибутами без привязки к сессии: отсутствующий атрибут даёт `None`.
 """
 
-from collections.abc import Iterable, Sequence
+from collections.abc import Iterable, Mapping, Sequence
 from datetime import date, datetime
 from decimal import Decimal
 from enum import Enum
@@ -37,22 +37,33 @@ _VALUE_FIELDS = (
 _ISSUE_FIELDS = ("kind", "code", "severity", "source", "field", "relation", "codes", "line_ids")
 
 
-def build_intake_payload(
+def build_intake_payload(  # noqa: PLR0913 - сигнатура снимка фиксирована (design §4.1)
     document: Document,
     *,
     lines: Sequence[Any] | None = None,
     values: Sequence[Any] | None = None,
     issues: Iterable[Any] = (),
     osv_warnings: Iterable[str] = (),
+    header_source: Mapping[str, Any] | None = None,
+    osv_row_count: int | None = None,
 ) -> dict[str, Any]:
     """Снимок «как распознано» для `quality_intake_fact.payload` (§14.1, design §4.1).
 
-    `BALANCE` / `INCOME`: `{"header", "lines", "values", "issues"}`. `OSV`: `{"osv_warnings"}` без `lines` и `values`.
+    `BALANCE` / `INCOME`: `{"header", "lines", "values", "issues"}`; при переданном `header_source` (сырой текст и
+    уверенность по полям шапки, `financial_statement_meta.header_source`) он кладётся в `header["header_source"]`.
+    `OSV`: `{"osv_warnings"}` без `lines` и `values`; при переданном `osv_row_count` добавляется `row_count`.
+    Без необязательных аргументов ключей `header_source` и `row_count` в payload нет.
     """
     if document.document_type is DocumentType.OSV:
-        return {"osv_warnings": list(osv_warnings)}
+        payload: dict[str, Any] = {"osv_warnings": list(osv_warnings)}
+        if osv_row_count is not None:
+            payload["row_count"] = osv_row_count
+        return payload
+    header = _pick(document, _HEADER_FIELDS)
+    if header_source is not None:
+        header["header_source"] = {key: _plain_mapping(entry) for key, entry in header_source.items()}
     return {
-        "header": _pick(document, _HEADER_FIELDS),
+        "header": header,
         "lines": [_line(line) for line in lines or ()],
         "values": [_value(value) for value in values or ()],
         "issues": [_pick(issue, _ISSUE_FIELDS) for issue in issues],
@@ -67,13 +78,24 @@ def build_outcome_payload(  # noqa: PLR0913 - сигнатура снимка ф
     issues: Iterable[Any] = (),
     osv_warnings: Iterable[str] = (),
     fixes: Iterable[Any] = (),
+    header_source: Mapping[str, Any] | None = None,
+    osv_row_count: int | None = None,
 ) -> dict[str, Any]:
     """Подтверждённый результат и правки для `quality_outcome_fact.payload` (§14.1, design §4.2).
 
-    Строение — как у снимка приёма (`build_intake_payload`), плюс `fixes`: список правок с отнесением, каждая запись
-    строится вызывающим (состав записи — задача отнесения правок, не этой функции).
+    Строение — как у снимка приёма (`build_intake_payload`, с теми же `header_source` и `osv_row_count`), плюс `fixes`:
+    список правок с отнесением, каждая запись строится вызывающим (состав записи — задача отнесения правок, не этой
+    функции).
     """
-    payload = build_intake_payload(document, lines=lines, values=values, issues=issues, osv_warnings=osv_warnings)
+    payload = build_intake_payload(
+        document,
+        lines=lines,
+        values=values,
+        issues=issues,
+        osv_warnings=osv_warnings,
+        header_source=header_source,
+        osv_row_count=osv_row_count,
+    )
     payload["fixes"] = list(fixes)
     return payload
 
@@ -86,6 +108,13 @@ def _line(line: Any) -> dict[str, Any]:
     }
 
 
+def _plain_mapping(entry: Any) -> Any:
+    """Приводит запись `header_source` (`{"text", "confidence"}`) к JSON-совместимому виду; Decimal → строка."""
+    if isinstance(entry, Mapping):
+        return {key: _plain(item) for key, item in entry.items()}
+    return _plain(entry)
+
+
 def _value(value: Any) -> dict[str, Any]:
     return _pick(value, _VALUE_FIELDS)
 

api/tests/test_domain_quality_facts.py

@@ -15,6 +15,8 @@ from src.models.enums import (
 from src.models.files import Document
 from src.models.statements import DocumentIssue, FinancialStatementLine, FinancialStatementValue
 
+ROW_COUNT = 4_200
+
 
 def make_document(document_type: DocumentType = DocumentType.BALANCE) -> Document:
     return Document(
@@ -125,3 +127,39 @@ def test_build_outcome_payload_osv_has_fixes_and_no_lines():
     assert payload["osv_warnings"] == ["OSV_SOMETHING"]
     assert payload["fixes"] == [fix]
     assert "lines" not in payload and "values" not in payload
+
+
+def test_build_intake_payload_includes_header_source_for_statement():
+    document = make_document(document_type=DocumentType.BALANCE)
+    header_source = {"inn": {"text": "7701234567", "confidence": Decimal("100.00")}}
+    payload = build_intake_payload(document, lines=[], values=[], header_source=header_source)
+    assert payload["header"]["header_source"] == {"inn": {"text": "7701234567", "confidence": "100.00"}}
+
+
+def test_build_outcome_payload_includes_header_source_for_statement():
+    document = make_document(document_type=DocumentType.INCOME)
+    header_source = {"okpo": {"text": "12345678", "confidence": None}}
+    payload = build_outcome_payload(document, lines=[], values=[], fixes=[], header_source=header_source)
+    assert payload["header"]["header_source"] == {"okpo": {"text": "12345678", "confidence": None}}
+
+
+def test_build_intake_payload_includes_osv_row_count():
+    document = make_document(document_type=DocumentType.OSV)
+    payload = build_intake_payload(document, osv_warnings=["OSV_SOMETHING"], osv_row_count=ROW_COUNT)
+    assert payload["row_count"] == ROW_COUNT
+    assert payload["osv_warnings"] == ["OSV_SOMETHING"]
+
+
+def test_build_outcome_payload_includes_osv_row_count():
+    document = make_document(document_type=DocumentType.OSV)
+    payload = build_outcome_payload(document, osv_warnings=[], fixes=[], osv_row_count=0)
+    assert payload["row_count"] == 0
+
+
+def test_header_source_and_row_count_absent_when_kwargs_omitted():
+    balance = build_intake_payload(make_document(document_type=DocumentType.BALANCE), lines=[], values=[])
+    assert "header_source" not in balance["header"]
+    osv = build_intake_payload(make_document(document_type=DocumentType.OSV), osv_warnings=[])
+    assert "row_count" not in osv
+    outcome_osv = build_outcome_payload(make_document(document_type=DocumentType.OSV), osv_warnings=[], fixes=[])
+    assert "row_count" not in outcome_osv