arm-tabular-parser/api/tests/unit/parser/statements/test_forms.py

"""Блоки форм, вид листа и столбцы на словарях `arm_statements` (план 4, задача 4)."""

import pytest
from gridbuilder import at, sheet_from_rows

from arm_statements import DocumentType, continuation_type, default_rules
from src.parser.statements.forms import (
    StatementRules,
    caption_matches,
    classify_statement_sheet,
    codes_of,
    continuation_marker,
    find_blocks,
    is_weak_continuation,
    matches,
    sheet_amount_columns,
    sheet_code_column,
)

CODE_COL = 3


@pytest.fixture(scope="module")
def rules() -> StatementRules:
    return default_rules()


def test_title_and_okud_without_codes_between_form_one_block(rules: StatementRules) -> None:
    sheet = sheet_from_rows(
        [
            ["Бухгалтерская отчетность"],
            [],
            [None, "Бухгалтерский баланс"],
            [],
            [*at({3: "Форма по ОКУД", 5: "0710001"})],
            [*at({1: "Актив", 3: "Код", 4: "На 30.09.2025"})],
            [*at({1: "Основные средства", 3: "1150", 4: 10})],
        ]
    )
    [block] = find_blocks(sheet, rules)
    assert (block.type_hint, block.title, block.okud, block.first_row, block.last_row) == (
        DocumentType.BALANCE,
        (3, 2),
        (5, 5),
        1,
        7,
    )


def test_second_title_after_codes_opens_new_block(rules: StatementRules) -> None:
    balance = [at({1: name, 3: code, 4: 10}) for name, code in (("а", 1110), ("б", 1150), ("в", 1170), ("г", 1600))]
    income = [at({1: name, 3: code, 4: 10}) for name, code in (("д", 2110), ("е", 2120), ("ж", 2400))]
    sheet = sheet_from_rows(
        [["Бухгалтерский баланс"], *balance, ["Отчет о финансовых результатах"], *income],
    )
    blocks = find_blocks(sheet, rules)
    assert [(b.type_hint, b.first_row, b.last_row) for b in blocks] == [
        (DocumentType.BALANCE, 1, 5),
        (DocumentType.INCOME, 6, 9),
    ]


def test_okud_wins_over_title(rules: StatementRules) -> None:
    blocks = find_blocks(sheet_from_rows([["Бухгалтерский баланс"], ["по ОКУД", "0710002"]]), rules)
    assert [block.type_hint for block in blocks] == [DocumentType.INCOME]


def test_report_title_alone_is_not_a_form(rules: StatementRules) -> None:
    assert find_blocks(sheet_from_rows([["Бухгалтерская отчетность"]]), rules) == []


def test_continuation_marker_is_not_a_block(rules: StatementRules) -> None:
    sheet = sheet_from_rows([["Форма 0710001 с.2"], ["Пассив", "Код", "На 30.09.2025"], ["Капитал", "1310", 10]])
    assert find_blocks(sheet, rules) == []
    assert classify_statement_sheet(sheet, rules) == "continuation"


def test_code_column_ignores_title_codes_block(rules: StatementRules) -> None:
    titles = ["Коды", "0710001", 30, "09", 2025, 12345678, 7700000000, 12267, 16, 384]
    codes = [1110, 1150, 1170, 1100, 1210, 1230]
    rows = [at({6: value}) for value in titles]
    rows += [at({3: code, 4: 10}) for code in codes]
    sheet = sheet_from_rows(rows)
    assert sheet_code_column(sheet, rules, 1, sheet.n_rows) == CODE_COL
    assert codes_of(sheet, rules, CODE_COL, 1, sheet.n_rows) == [str(code) for code in codes]


def test_amount_columns_counts_numeric_columns_right_of_code(rules: StatementRules) -> None:
    rows = [at({1: "Актив", 3: "Код", 4: "На 30.09.2025", 7: "Изменение"})]
    rows += [at({1: "Статья", 3: code, 4: 1, 5: 2, 6: 3, 7: "рост"}) for code in (1110, 1150, 1170, 1100)]
    sheet = sheet_from_rows(rows)
    assert sheet_amount_columns(sheet, rules, CODE_COL, 1, sheet.n_rows) == CODE_COL


def test_weak_and_other_sheets(rules: StatementRules) -> None:
    weak = sheet_from_rows([["Чистая прибыль", "2400", 10, 8]])
    assert classify_statement_sheet(weak, rules) == "other"
    assert is_weak_continuation(weak, rules, DocumentType.INCOME)
    assert not is_weak_continuation(weak, rules, DocumentType.BALANCE)
    notes = sheet_from_rows([["Примечания"], ["Текст пояснения к отчетности"]])
    assert classify_statement_sheet(notes, rules) == "other"
    assert not is_weak_continuation(notes, rules, DocumentType.INCOME)


def test_marker_with_unknown_okud_is_continuation(rules: StatementRules) -> None:
    rows = [["Форма 0710000 с.2"]] + [at({1: "Статья", 3: code, 4: 10}) for code in (1110, 1150, 1170)]
    sheet = sheet_from_rows(rows)
    assert find_blocks(sheet, rules) == []
    assert classify_statement_sheet(sheet, rules) == "continuation"
    marker = continuation_marker(sheet, rules)
    assert marker is not None
    assert marker.text == "Форма 0710000 с.2"
    assert continuation_type(marker.text, rules) is None


def test_labels_with_punctuation_match(rules: StatementRules) -> None:
    assert matches(rules.headers.labels["address"], "Местонахождение (адрес)")
    assert matches(rules.headers.labels["okopf_okfs"], "по ОКОПФ / ОКФС")
    assert matches(rules.headers.legal_form, "АО «Пример»")
    assert caption_matches([rules.headers.code_caption], "Код")


def test_column_caption_in_table_header_is_not_a_form_title(rules: StatementRules) -> None:
    rows = [
        ["Форма 0710001 с.2"],
        at({1: "Наименование", 2: "Пояснения", 3: "Код", 4: "На 30.09.2025"}),
        at({1: "Капитал", 3: 1310, 4: 10}),
    ]
    sheet = sheet_from_rows(rows)
    assert find_blocks(sheet, rules) == []
    assert classify_statement_sheet(sheet, rules) == "continuation"


def test_real_title_above_same_table_header_stays_a_block(rules: StatementRules) -> None:
    rows = [
        ["Бухгалтерский баланс"],
        at({1: "Наименование", 2: "Пояснения", 3: "Код", 4: "На 30.09.2025"}),
        at({1: "Капитал", 3: 1310, 4: 10}),
    ]
    [block] = find_blocks(sheet_from_rows(rows), rules)
    assert block.type_hint is DocumentType.BALANCE