- I-1 (R-23): readers/_limits.py — сумма ZipInfo.file_size ≤ 10 × PARSE_MAX_BYTES (параметр read_book от OsvParser), потоковый скан <mergeCell ref> до openpyxl и площадь merged после чтения любого формата (MERGED_AREA_LIMIT = 200 000; реальный максимум 10 090) → UNSUPPORTED_FORMAT; дедлайн в циклах detect (поиск строки групп, дополнительные колонки); README «Лимиты» исправлен, «Известные ограничения» (defusedxml). - I-2: OsvParser логирует стек непредвиденной ошибки (loguru ERROR), UNREADABLE — WARNING без стека. - M-1: ACCOUNT_RE с re.ASCII | re.DOTALL (многострочное название после запятой, только цифры 0–9). - M-3 (R-24): правило детектора ИНН не дополняет 9/11 цифр нулём; дополнение — только при извлечении (numbers.inn_digits) из колонки, выбранной по заголовку или по строгому правилу. - T9.1: тесты R-21 (свёрнутый родитель на противоположной стороне; (0,0) при детях (X, X)). - T1.1: хелперы фикстур пропускают скрытые файлы/каталоги и lock-файлы Excel (~$…). - T12.3: golden-update не удаляет снимки части без исходного каталога (osv/synthetic или реальные файлы). Golden-снимки не изменились (82/82). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_012vJkmBGKKqMoEMzvsXG3y2
282 lines
11 KiB
Python
282 lines
11 KiB
Python
"""Контрольные суммы §1.4 (spec «Контрольные суммы определяют статус», `05-domain-rules.md` §1.4, R-13, R-20).
|
||
|
||
Строки собираются напрямую (`ClassifiedRow` над `BodyRow`): проверки — чистые функции над деревом строк, им не нужна
|
||
книга. Номера строк произвольные; суммы — `Decimal` или `None` (пусто = 0).
|
||
"""
|
||
|
||
from decimal import Decimal
|
||
|
||
from src.contract.common import DataType
|
||
from src.parser.checks import TOLERANCE, balance_identity, parent_sums, run_checks, total_matches_root
|
||
from src.parser.classify import ClassifiedRow, RowGroupType
|
||
from src.parser.hierarchy import BodyRow
|
||
|
||
Sums = tuple[str | None, str | None, str | None, str | None, str | None, str | None]
|
||
EMPTY: Sums = (None, None, None, None, None, None)
|
||
|
||
|
||
def row( # noqa: PLR0913 - строка теста описывается полностью одним вызовом
|
||
number: int,
|
||
sums: Sums,
|
||
*,
|
||
level: int = 0,
|
||
parent: int | None = None,
|
||
group_type: RowGroupType = "COUNTERPARTY",
|
||
data_type: DataType | None = "BU",
|
||
text: str = "строка",
|
||
) -> ClassifiedRow:
|
||
body = BodyRow(
|
||
row=number,
|
||
text=text,
|
||
is_account=group_type == "ACCOUNT",
|
||
account_rank=1 if group_type == "ACCOUNT" else None,
|
||
bold=False,
|
||
indent=0,
|
||
outline=level,
|
||
sums=tuple(None if value is None else Decimal(value) for value in sums),
|
||
)
|
||
return ClassifiedRow(
|
||
body=body,
|
||
level=level,
|
||
parent=parent,
|
||
group_type=group_type,
|
||
group_name=None,
|
||
account_code_raw="60",
|
||
counterparty_name=None,
|
||
counterparty_inn=None,
|
||
data_type=data_type,
|
||
extra=None,
|
||
)
|
||
|
||
|
||
def account(number: int, sums: Sums) -> ClassifiedRow:
|
||
return row(number, sums, group_type="ACCOUNT", text="60")
|
||
|
||
|
||
def total(number: int, sums: Sums) -> ClassifiedRow:
|
||
return row(number, sums, group_type="TOTAL", text="Итого")
|
||
|
||
|
||
# Корень 60 с двумя контрагентами, «Итого» равно корню; тождество выполняется во всех строках
|
||
ROOT: Sums = ("100", None, "50", "30", "120", None)
|
||
CHILD_A: Sums = ("60", None, "20", "10", "70", None)
|
||
CHILD_B: Sums = ("40", None, "30", "20", "50", None)
|
||
|
||
|
||
def tree() -> list[ClassifiedRow]:
|
||
return [
|
||
account(7, ROOT),
|
||
row(8, CHILD_A, level=1, parent=7),
|
||
row(9, CHILD_B, level=1, parent=7),
|
||
total(10, ROOT),
|
||
]
|
||
|
||
|
||
def test_tolerance_is_one_kopeck() -> None:
|
||
assert Decimal("0.01") == TOLERANCE
|
||
|
||
|
||
def test_consistent_tree_has_no_violations() -> None:
|
||
checks, warnings = run_checks(tree())
|
||
|
||
assert checks.total_row_matches_root is True
|
||
assert checks.balance_identity_violations == []
|
||
assert checks.parent_sum_violations == []
|
||
assert warnings == []
|
||
|
||
|
||
def test_balance_identity_treats_null_as_zero() -> None:
|
||
rows = [
|
||
row(1, (None, None, None, None, None, None)),
|
||
row(2, (None, "5", None, None, None, "5")), # только кредит: −5 = −5
|
||
row(3, ("10", None, None, "10", None, None)), # 10 − 10 = 0 при пустом конце
|
||
]
|
||
|
||
assert balance_identity(rows) == []
|
||
|
||
|
||
def test_balance_identity_reports_row_numbers_beyond_tolerance() -> None:
|
||
rows = [
|
||
row(1, ("100", None, "50", "30", "120.01", None)), # в пределах ±0.01
|
||
row(2, ("100", None, "50", "30", "121", None)), # конец увеличен на 1
|
||
row(3, (None, None, None, None, None, "0.02")),
|
||
]
|
||
|
||
assert balance_identity(rows) == [2, 3]
|
||
|
||
|
||
def test_parent_differs_from_sum_of_children() -> None:
|
||
rows = tree()
|
||
rows[2] = row(9, ("40", None, "30", "20", "51", None), level=1, parent=7)
|
||
|
||
assert parent_sums(rows) == [7]
|
||
|
||
|
||
def test_parent_sum_is_checked_per_column_with_tolerance() -> None:
|
||
rows = [
|
||
account(7, ("100", None, "50", "30", "120", None)),
|
||
row(8, ("99.995", None, "50", "30", "119.995", None), level=1, parent=7),
|
||
]
|
||
|
||
assert parent_sums(rows) == []
|
||
|
||
|
||
def test_collapsed_parent_balance_is_compared_by_net() -> None:
|
||
"""Счёт со свёрнутым сальдо (одна сторона), субсчета — с развёрнутым: сверка сальдо по нетто (как в 1С на всех
|
||
реальных файлах с субсчетами разных сторон); обороты — по колонкам.
|
||
"""
|
||
rows = [
|
||
account(12, ("16", None, "134", "111", "39", None)),
|
||
row(13, ("40", None, "131", "111", "60", None), level=1, parent=12, group_type="ACCOUNT"),
|
||
row(22, (None, "24", "3", None, None, "21"), level=1, parent=12, group_type="ACCOUNT"),
|
||
]
|
||
|
||
assert parent_sums(rows) == []
|
||
|
||
|
||
def test_collapsed_parent_with_wrong_net_is_violation() -> None:
|
||
rows = [
|
||
account(12, ("17", None, "134", "111", "39", None)),
|
||
row(13, ("40", None, "131", "111", "60", None), level=1, parent=12, group_type="ACCOUNT"),
|
||
row(22, (None, "24", "3", None, None, "21"), level=1, parent=12, group_type="ACCOUNT"),
|
||
]
|
||
|
||
assert parent_sums(rows) == [12]
|
||
|
||
|
||
def test_collapsed_parent_on_the_opposite_side_is_violation() -> None:
|
||
"""T9.1 (R-21): свёрнутое сальдо родителя на противоположной стороне (16 Кт при нетто детей +16 Дт) — нарушение:
|
||
нетто сравнивается со знаком, совпадения модуля мало."""
|
||
rows = [
|
||
account(12, (None, "16", "134", "111", "39", None)),
|
||
row(13, ("40", None, "131", "111", "60", None), level=1, parent=12, group_type="ACCOUNT"),
|
||
row(22, (None, "24", "3", None, None, "21"), level=1, parent=12, group_type="ACCOUNT"),
|
||
]
|
||
|
||
assert parent_sums(rows) == [12]
|
||
|
||
|
||
def test_zero_parent_with_children_netting_to_zero_is_accepted() -> None:
|
||
"""T9.1 (R-21, осознанное поведение): родитель (0, 0) при детях (X Дт, X Кт) — без нарушения: нулевое сальдо
|
||
родителя — свёрнутое (не больше одной ненулевой стороны), нетто детей X − X = 0. Остаточный риск описан в спеке
|
||
(§1.4): потерянная пара сальдо у родителя этой проверкой не ловится."""
|
||
rows = [
|
||
account(12, ("0", "0", "134", "111", "23", None)),
|
||
row(13, ("50", None, "131", "111", "70", None), level=1, parent=12, group_type="ACCOUNT"),
|
||
row(22, (None, "50", "3", None, None, "47"), level=1, parent=12, group_type="ACCOUNT"),
|
||
]
|
||
|
||
assert balance_identity(rows) == []
|
||
assert parent_sums(rows) == []
|
||
|
||
|
||
def test_expanded_parent_balance_must_match_by_columns() -> None:
|
||
"""Родитель с обеими сторонами сальдо — развёрнутое сальдо: совпадение одного нетто недостаточно."""
|
||
rows = [
|
||
account(12, ("41", "25", "134", "111", "60", "21")),
|
||
row(13, ("40", None, "131", "111", "60", None), level=1, parent=12, group_type="ACCOUNT"),
|
||
row(22, (None, "24", "3", None, None, "21"), level=1, parent=12, group_type="ACCOUNT"),
|
||
]
|
||
|
||
assert parent_sums(rows) == [12]
|
||
|
||
|
||
def test_turnovers_are_compared_by_columns_even_when_net_matches() -> None:
|
||
rows = [
|
||
account(12, ("16", None, "135", "112", "39", None)),
|
||
row(13, ("40", None, "131", "111", "60", None), level=1, parent=12, group_type="ACCOUNT"),
|
||
row(22, (None, "24", "3", None, None, "21"), level=1, parent=12, group_type="ACCOUNT"),
|
||
]
|
||
|
||
assert parent_sums(rows) == [12]
|
||
|
||
|
||
def test_currency_and_quantity_children_are_excluded() -> None:
|
||
rows = [
|
||
account(7, ROOT),
|
||
row(8, ROOT, level=1, parent=7),
|
||
row(9, ("1", None, "2", None, "3", None), level=2, parent=8, group_type="OTHER", data_type="CURRENCY"),
|
||
row(10, ("7", None, None, None, "7", None), level=2, parent=8, group_type="OTHER", data_type="QUANTITY"),
|
||
total(11, ROOT),
|
||
]
|
||
|
||
assert parent_sums(rows) == []
|
||
|
||
|
||
def test_children_of_other_data_type_are_not_summed() -> None:
|
||
"""Σ детей «того же data_type»: ребёнок НУ под родителем БУ в сумму не входит."""
|
||
rows = [
|
||
account(7, ("10", None, None, None, "10", None)),
|
||
row(8, ("10", None, None, None, "10", None), level=1, parent=7),
|
||
row(9, ("5", None, None, None, "5", None), level=1, parent=7, data_type="NU"),
|
||
]
|
||
|
||
assert parent_sums(rows) == []
|
||
|
||
|
||
def test_rows_without_children_are_not_parents() -> None:
|
||
rows = [account(7, ROOT), account(8, ("1", None, None, None, "1", None))]
|
||
|
||
assert parent_sums(rows) == []
|
||
|
||
|
||
def test_total_matches_first_root_account() -> None:
|
||
rows = tree()
|
||
|
||
assert total_matches_root(rows) is True
|
||
rows[-1] = total(10, ("100", None, "50", "30", "121", None))
|
||
assert total_matches_root(rows) is False
|
||
|
||
|
||
def test_empty_osv_without_root_matches_when_total_is_empty_or_zero() -> None:
|
||
assert total_matches_root([total(8, EMPTY)]) is True
|
||
assert total_matches_root([total(8, ("0", None, "0.00", None, None, "0"))]) is True
|
||
assert total_matches_root([total(8, ("1", None, None, None, "1", None))]) is False
|
||
|
||
|
||
def test_missing_total_fails_root_check_with_warning() -> None:
|
||
"""R-20: без «Итого» сверку выполнить нельзя — `false` и предупреждение."""
|
||
rows = tree()[:-1]
|
||
|
||
checks, warnings = run_checks(rows)
|
||
|
||
assert total_matches_root(rows) is False
|
||
assert checks.total_row_matches_root is False
|
||
assert "строка «Итого» не найдена" in warnings
|
||
|
||
|
||
def test_run_checks_warnings_name_row_numbers() -> None:
|
||
rows = tree()
|
||
rows[2] = row(9, ("40", None, "30", "20", "51", None), level=1, parent=7)
|
||
|
||
checks, warnings = run_checks(rows)
|
||
|
||
assert checks.balance_identity_violations == [9]
|
||
assert checks.parent_sum_violations == [7]
|
||
assert checks.total_row_matches_root is True
|
||
assert any("строки 9" in warning for warning in warnings), warnings
|
||
assert any("строки 7" in warning for warning in warnings), warnings
|
||
|
||
|
||
def test_run_checks_reports_total_mismatch_with_row_numbers() -> None:
|
||
rows = tree()
|
||
rows[-1] = total(10, ("100", None, "50", "30", "121", None))
|
||
|
||
checks, warnings = run_checks(rows)
|
||
|
||
assert checks.total_row_matches_root is False
|
||
assert any("10" in warning and "7" in warning and "Итого" in warning for warning in warnings), warnings
|
||
|
||
|
||
def test_run_checks_warning_lists_are_capped() -> None:
|
||
"""Список нарушений в `checks` полный, текст предупреждения — первые номера и общее число."""
|
||
broken: Sums = ("1", None, None, None, "2", None)
|
||
rows = [row(number, broken) for number in range(1, 101)]
|
||
|
||
checks, warnings = run_checks(rows)
|
||
|
||
assert checks.balance_identity_violations == list(range(1, 101))
|
||
(identity,) = [warning for warning in warnings if "сальдо" in warning]
|
||
assert "всего 100" in identity
|
||
assert "100," not in identity
|