feat: перевод документа и предупреждений в форму контракта v2.0, вход статуса из документа контракта

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01BJKMLHdxMwSUeYcdkwqX4W

arm_statements/__init__.py

@@ -1,12 +1,12 @@
 """Общий код нормализации форм ББ/ОФР.
 
-Публичный API — только имена `__all__` (раздел «Публичный API пакета» плана 1). Имена задачи 19
-(`to_contract`, `warning_to_contract`, `status_input_from_contract`) добавляются ими.
+Публичный API — только имена `__all__` (раздел «Публичный API пакета» плана 1). Контрактные имена — `contract.py`.
 """
 
 from arm_statements.amounts import parse_amount
 from arm_statements.codes import parse_code, sheet_joins
 from arm_statements.columns import amount_columns
+from arm_statements.contract import status_input_from_contract, to_contract, warning_to_contract
 from arm_statements.dates import parse_date, resolve_relative
 from arm_statements.header import continuation_type, is_continuation, parse_header, type_hint
 from arm_statements.issues import WarningCode, make_warning
@@ -130,6 +130,9 @@ __all__ = [
     "sheet_joins",
     "statements_version",
     "status_input",
+    "status_input_from_contract",
+    "to_contract",
     "to_jsonable",
     "type_hint",
+    "warning_to_contract",
 ]

arm_statements/contract.py

@@ -0,0 +1,151 @@
+"""Перевод выхода пакета в форму контракта v2.0 и вход статуса из документа контракта.
+
+Один перевод на оба сервиса (P1-26): хосты свой `contract_map` не пишут. Схему пакет не читает —
+форма держится тестами и разовой сверкой со `statementDocument` схемы v2.0.
+"""
+
+from collections.abc import Mapping
+from decimal import Decimal
+from typing import Any
+
+from arm_statements.model import (
+    HEADER_FIELDS,
+    DocumentType,
+    FailedRelation,
+    LineStatus,
+    PeriodKind,
+    Severity,
+    StatementDocument,
+    StatementHeader,
+    StatementWarning,
+)
+from arm_statements.status import StatusInput, StatusLine, StatusValue
+
+
+def _money(value: Decimal | None) -> str | None:
+    return None if value is None else format(value, "f")
+
+
+def warning_to_contract(warning: StatementWarning, *, file_level: bool = False) -> dict[str, Any]:
+    """`warning` схемы; при `file_level` — `fileWarning` (с `locator`). Ключи без значения опускаются."""
+    data: dict[str, Any] = {
+        "code": warning.code,
+        "severity": str(warning.severity),
+        "message": warning.message,
+    }
+    if warning.line_refs:
+        data["line_refs"] = list(warning.line_refs)
+    if warning.column_index is not None:
+        data["column_index"] = warning.column_index
+    if warning.field is not None:
+        data["field"] = warning.field
+    if file_level and warning.locator is not None:
+        data["locator"] = warning.locator
+    return data
+
+
+def _header(header: StatementHeader) -> dict[str, Any]:
+    data: dict[str, Any] = {"form_name": header.form_name}
+    for name in HEADER_FIELDS:
+        field = getattr(header, name)
+        value = field.value
+        data[name] = value.isoformat() if name == "report_date" and value is not None else value
+        data[f"{name}_text"] = field.text
+        data[f"{name}_confidence"] = field.confidence
+    data["needs_review_fields"] = list(header.needs_review_fields)
+    return data
+
+
+def _relation(relation: FailedRelation) -> dict[str, Any]:
+    if relation.column_index is None:
+        raise ValueError(f"у соотношения {relation.relation!r} нет column_index: у сервиса номер колонки есть всегда")
+    return {
+        "column_index": relation.column_index,
+        "relation": relation.relation,
+        "codes": list(relation.codes),
+        "expected": _money(relation.expected),
+        "actual": _money(relation.actual),
+    }
+
+
+def to_contract(document: StatementDocument) -> dict[str, Any]:
+    """`statementDocument` схемы v2.0 как json-совместимый `dict`."""
+    data: dict[str, Any] = {
+        "document_type": str(document.document_type),
+        "form_variant": None if document.form_variant is None else str(document.form_variant),
+        "locator": document.locator,
+        "status": str(document.status),
+        "warnings": [warning_to_contract(w) for w in document.warnings],
+        "header": _header(document.header),
+        "columns": [
+            {
+                "column_index": c.column_index,
+                "label": c.label,
+                "label_confidence": c.label_confidence,
+                "period_date": None if c.period_date is None else c.period_date.isoformat(),
+                "period_kind": str(c.period_kind),
+            }
+            for c in document.columns
+        ],
+        "lines": [
+            {
+                "row_index": ln.row_index,
+                "ref": ln.ref,
+                "line_status": str(ln.line_status),
+                "line_code": ln.line_code,
+                "line_code_text": ln.line_code_text,
+                "line_code_confidence": ln.line_code_confidence,
+                "line_name": ln.line_name,
+                "line_name_confidence": ln.line_name_confidence,
+                "parent_ref": ln.parent_ref,
+                "needs_review": ln.needs_review,
+                "values": [
+                    {
+                        "column_index": v.column_index,
+                        "amount": _money(v.amount),
+                        "string_value": v.string_value,
+                        "text": v.text,
+                        "confidence": v.confidence,
+                        "is_recognized": v.is_recognized,
+                        "needs_review": v.needs_review,
+                    }
+                    for v in ln.values
+                ],
+            }
+            for ln in document.lines
+        ],
+    }
+    if document.checks is not None:
+        data["checks"] = {
+            "relations_failed": [_relation(r) for r in document.checks.relations_failed],
+            "soft_relations_failed": [_relation(r) for r in document.checks.soft_relations_failed],
+        }
+    return data
+
+
+def status_input_from_contract(document: Mapping[str, Any]) -> StatusInput:
+    """Вход функции статуса из `statementDocument` контракта v2.0 (то же правило, что `status_input`).
+
+    Схему не проверяет: битый документ — `KeyError` / `ValueError`. Отсутствующий необязательный ключ и `null` равны.
+    """
+    header = document["header"]
+    lines = document["lines"]
+    columns = document["columns"]
+    period_known = header.get("report_date") is not None or any(
+        c["period_kind"] == PeriodKind.ABSOLUTE and c.get("period_date") for c in columns
+    )
+    checks = document.get("checks")
+    return StatusInput(
+        document_type=DocumentType(document["document_type"]),
+        layout_found=len(lines) > 0,
+        has_organization_or_inn=bool(header.get("organization_name") or header.get("inn")),
+        period_known=period_known,
+        needs_review_fields=tuple(header.get("needs_review_fields") or ()),
+        lines=tuple(StatusLine(LineStatus(ln["line_status"]), bool(ln["needs_review"])) for ln in lines),
+        values=tuple(
+            StatusValue(LineStatus(ln["line_status"]), bool(v["needs_review"])) for ln in lines for v in ln["values"]
+        ),
+        hard_failures=0 if checks is None else len(checks["relations_failed"]),
+        accepted=False,
+        warning_severities=tuple(Severity(w["severity"]) for w in document["warnings"]),
+    )

arm_statements/tests/test_contract.py

@@ -0,0 +1,99 @@
+import json
+from dataclasses import replace
+from decimal import Decimal as D
+
+import pytest
+
+from arm_statements import (
+    DocStatus,
+    DocumentType,
+    document_status,
+    make_warning,
+    normalize_candidate,
+    status_input,
+    status_input_from_contract,
+    to_contract,
+    warning_to_contract,
+)
+from arm_statements.loader import Rules
+from arm_statements.model import FailedRelation, StatementChecks, StatementDocument
+from arm_statements.tests.builders import hdr, ocr
+
+HEAD = ["Наименование показателя", "Код", "На 30.09.2025"]
+FULL_HEADER = hdr(
+    organization_name="ООО «Пример»", inn=("7700000000", 80.0), report_date="30.09.2025", units="тыс. руб."
+)
+HEADER_KEYS = {"form_name", "needs_review_fields"} | {
+    f"{name}{suffix}"
+    for name in ("organization_name", "inn", "report_date", "units", "okud", "okpo", "okopf", "okfs", "okei", "address")
+    for suffix in ("", "_text", "_confidence")
+}
+
+
+def balance(rules: Rules, *rows: list[str]) -> StatementDocument:
+    cand = ocr([HEAD, *rows], header=FULL_HEADER, type_hint=DocumentType.BALANCE, title="Бухгалтерский баланс")
+    doc = normalize_candidate(cand, rules).document
+    assert doc is not None
+    return doc
+
+
+def test_document_in_contract_form(rules: Rules) -> None:
+    doc = balance(rules, ["Запасы", "1210", "60"], ["Итого II", "1200", "60"], ["НИ", "", "7"])
+    data = json.loads(json.dumps(to_contract(doc), ensure_ascii=False))
+    assert set(data["header"]) == HEADER_KEYS
+    assert (data["header"]["form_name"], data["header"]["inn"], data["header"]["inn_confidence"]) == (
+        "Бухгалтерский баланс",
+        "7700000000",
+        80.0,
+    )
+    assert data["header"]["report_date"] == "2025-09-30" and data["header"]["okei"] is None
+    assert data["lines"][0]["values"][0]["amount"] == "60.00"
+    assert {ln["line_status"] for ln in data["lines"]} >= {"ok"}
+    assert "column_key" not in json.dumps(data) and "row:" not in json.dumps(data["header"])
+    assert "locator" not in json.dumps(data["warnings"])
+    assert set(data["checks"]) == {"relations_failed", "soft_relations_failed"}
+
+
+def test_empty_target_is_null_in_contract(rules: Rules) -> None:
+    doc = balance(rules, ["Запасы", "1210", "60"], ["Итого II", "1200", "60"], ["Итого I", "1100", "40"])
+    failed = to_contract(doc)["checks"]["relations_failed"]
+    target = next(f for f in failed if f["relation"] == "1600 = 1100 + 1200")
+    assert (target["expected"], target["actual"], target["column_index"]) == (None, "100.00", 2)
+
+
+def test_no_lines_no_checks_key(rules: Rules) -> None:
+    doc = normalize_candidate(
+        ocr([["Итого", "15 000"]], type_hint=DocumentType.BALANCE, header=FULL_HEADER), rules
+    ).document
+    assert doc is not None and "checks" not in to_contract(doc) and to_contract(doc)["lines"] == []
+
+
+def test_relation_without_column_index_is_error(rules: Rules) -> None:
+    doc = balance(rules, ["Запасы", "1210", "60"], ["Итого II", "1200", "60"])
+    bad = FailedRelation(None, "c1", "1600 = 1700", ("1600", "1700"), D("1"), None)
+    with pytest.raises(ValueError, match="column_index"):
+        to_contract(replace(doc, checks=StatementChecks((bad,), ())))
+
+
+def test_warning_keys(rules: Rules) -> None:
+    w = make_warning("ROW_SPLIT", "разлеплено", rules, line_refs=("row:3#1", "row:3#2"), locator="file:a.json")
+    assert warning_to_contract(w) == {
+        "code": "ROW_SPLIT",
+        "severity": "info",
+        "message": "разлеплено",
+        "line_refs": ["row:3#1", "row:3#2"],
+    }
+    lost = make_warning("TABLE_LOST", "таблица", rules, locator="file:a.json;table:1")
+    assert warning_to_contract(lost, file_level=True) == {
+        "code": "TABLE_LOST",
+        "severity": "error",
+        "message": "таблица",
+        "locator": "file:a.json;table:1",
+    }
+
+
+def test_status_input_roundtrip(rules: Rules) -> None:
+    doc = balance(rules, ["Запасы", "1210", "60"], ["Итого II", "1200", "60"], ["НИ", "", "7"])
+    data = status_input_from_contract(json.loads(json.dumps(to_contract(doc), ensure_ascii=False)))
+    assert data == status_input(doc)
+    assert document_status(data) is doc.status is DocStatus.NEEDS_CHECK