feat(statements): строки с кодом, значения Decimal, дубликаты кодов, предупреждение о единицах

Refs #14

Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
Claude-Session: cedd7915-2efe-41e5-988c-a9117033598b

api/src/parser/statements/lines.py

@@ -0,0 +1,116 @@
+"""Строки формы с кодом и их значения (design D-6).
+
+Строка с кодом в колонке кодов → `StatementLine`; строки без кода не отдаются (ASSUMPTION(OQ-36)), но если среди них
+есть суммы — одно предупреждение. Повтор кода — остаётся первая строка, код в `duplicates`. Числа — `numbers.to_money`
+(Decimal, скобки → минус, прочерк → null); нечисловой текст — `string_value`, `is_recognized = false`. Единицы не
+нормализуются: при «тыс.» в единицах и доле значений с дробной частью ≥ `FRACTIONAL_SHARE` — предупреждение.
+"""
+
+from dataclasses import dataclass, field
+
+from src.contract.result import StatementLine, StatementValue
+from src.parser.grid import Sheet
+from src.parser.numbers import to_money
+from src.parser.rules import Rules, normalize_text
+from src.parser.statements.columns import TableLayout, cell_label, line_code
+
+__all__ = ["FRACTIONAL_SHARE", "LinesResult", "collect_lines"]
+
+_FULL_CONFIDENCE = 100
+FRACTIONAL_SHARE = 0.1
+_THOUSANDS = "тыс"
+
+
+@dataclass
+class LinesResult:
+    lines: list[StatementLine] = field(default_factory=list)
+    duplicates: list[str] = field(default_factory=list)
+    skipped_with_sums: int = 0
+    fractional_values: int = 0
+    numeric_values: int = 0
+    warnings: list[str] = field(default_factory=list)
+
+
+def _value(sheet: Sheet, row: int, col: int, column_index: int) -> tuple[StatementValue, bool]:
+    """Значение ячейки и признак «число с дробной частью»."""
+    cell = sheet.cell(*sheet.merged_origin(row, col))
+    parsed = to_money(cell)
+    if parsed.value is not None:
+        has_fractional = parsed.value % 1 != 0
+        value = StatementValue(
+            column_index=column_index,
+            amount=parsed.value,
+            confidence=_FULL_CONFIDENCE,
+            is_recognized=True,
+        )
+        return value, has_fractional
+    if parsed.warning is None:  # пусто или прочерк
+        value = StatementValue(
+            column_index=column_index,
+            amount=None,
+            confidence=_FULL_CONFIDENCE,
+            is_recognized=True,
+        )
+        return value, False
+    value = StatementValue(
+        column_index=column_index,
+        amount=None,
+        string_value=cell.text,
+        confidence=_FULL_CONFIDENCE,
+        is_recognized=False,
+    )
+    return value, False
+
+
+def collect_lines(  # noqa: PLR0913
+    sheet: Sheet,
+    rules: Rules,
+    layout: TableLayout,
+    first_row: int,
+    last_row: int,
+    *,
+    units: str | None = None,
+) -> LinesResult:
+    result = LinesResult()
+    seen: set[str] = set()
+    for row in range(first_row, last_row + 1):
+        if sheet.is_row_empty(row):
+            continue
+        code = line_code(sheet.cell(row, layout.code_col), rules)
+        if code is None:
+            if any(to_money(sheet.cell(row, column.col)).value is not None for column in layout.periods):
+                result.skipped_with_sums += 1
+            continue
+        if code in seen:
+            result.duplicates.append(code)
+            continue
+        seen.add(code)
+        values: list[StatementValue] = []
+        for index, column in enumerate(layout.periods):
+            value, fractional = _value(sheet, row, column.col, index)
+            values.append(value)
+            if value.amount is not None:
+                result.numeric_values += 1
+                result.fractional_values += int(fractional)
+        name = cell_label(sheet, row, layout.name_col) if layout.name_col is not None else ""
+        result.lines.append(
+            StatementLine(
+                line_code=code,
+                line_code_confidence=_FULL_CONFIDENCE,
+                line_name=name,
+                line_name_confidence=_FULL_CONFIDENCE,
+                values=values,
+            )
+        )
+    if result.skipped_with_sums:
+        result.warnings.append(f"строк без кода с суммами пропущено: {result.skipped_with_sums}")
+    if (
+        units
+        and _THOUSANDS in normalize_text(units)
+        and result.numeric_values
+        and result.fractional_values / result.numeric_values >= FRACTIONAL_SHARE
+    ):
+        result.warnings.append(
+            f"значений с дробной частью при единице тыс. руб.: {result.fractional_values} из {result.numeric_values}"
+        )
+    return result

api/tests/unit/parser/statements/test_lines.py

@@ -0,0 +1,92 @@
+"""Строки с кодом и значения (design D-6; spec «Строки с кодом и значения»)."""
+
+from decimal import Decimal
+
+import pytest
+from gridbuilder import sheet_from_rows
+
+from src.parser.rules import Rules, load_rules
+from src.parser.statements.columns import find_layout
+from src.parser.statements.lines import collect_lines
+
+_FULL_CONFIDENCE = 100
+_FRACTIONAL_VALUES_COUNT = 4
+_NUMERIC_VALUES_COUNT = 9
+_SKIPPED_WITH_SUMS_COUNT = 1
+
+
+@pytest.fixture(scope="module")
+def rules() -> Rules:
+    return load_rules()
+
+
+ROWS = [
+    ["Пояснения", "Наименование показателя", "Код", "На 31 декабря 2024 г.", "На 31 декабря 2023 г."],
+    [None, "АКТИВ"],
+    [None, "Нематериальные активы", "1110", Decimal("5971264.14"), 27405634],
+    [None, "Нематериальные активы\nв организации", "11101", Decimal("5852464.14"), "27 214 614,65"],
+    [None, None, None, 0, 0],
+    [None, "в том числе:"],
+    [None, "Финансовые вложения", "1170", None, 10],
+    [None, "Прочие", "1190", "–", "(2 100)"],
+    [None, "Запасы", "1210", "н/д", 5],
+    [None, "Запасы", "1210", 1, 1],
+    [None, "Итого по разделу I", "1100", Decimal("11823728.28"), 27405644],
+]
+
+
+def test_lines_values_and_skips(rules: Rules) -> None:
+    sheet = sheet_from_rows(ROWS)
+    layout = find_layout(sheet, rules, 1, sheet.n_rows)
+
+    result = collect_lines(sheet, rules, layout, layout.first_data_row, sheet.n_rows)
+
+    codes = [line.line_code for line in result.lines]
+    assert codes == ["1110", "11101", "1170", "1190", "1210", "1100"]
+    by_code = {line.line_code: line for line in result.lines}
+    assert by_code["11101"].line_name == "Нематериальные активы в организации"
+    assert [value.amount for value in by_code["11101"].values] == [Decimal("5852464.14"), Decimal("27214614.65")]
+    assert by_code["1170"].values[0].amount is None and by_code["1170"].values[0].is_recognized is True
+    assert by_code["1190"].values[0].amount is None and by_code["1190"].values[1].amount == Decimal("-2100.00")
+    unrecognised = by_code["1210"].values[0]
+    assert unrecognised.amount is None and unrecognised.string_value == "н/д" and unrecognised.is_recognized is False
+    assert all(value.confidence == _FULL_CONFIDENCE for line in result.lines for value in line.values)
+    assert all(
+        line.line_code_confidence == _FULL_CONFIDENCE and line.line_name_confidence == _FULL_CONFIDENCE
+        for line in result.lines
+    )
+    assert [value.column_index for value in by_code["1110"].values] == [0, 1]
+    assert result.duplicates == ["1210"] and by_code["1210"].values[1].amount == Decimal("5.00")
+    assert result.skipped_with_sums == _SKIPPED_WITH_SUMS_COUNT  # строка нулей без кода
+    assert result.fractional_values == _FRACTIONAL_VALUES_COUNT and result.numeric_values == _NUMERIC_VALUES_COUNT
+
+
+def test_fractional_warning_when_units_thousands(rules: Rules) -> None:
+    sheet = sheet_from_rows(ROWS)
+    layout = find_layout(sheet, rules, 1, sheet.n_rows)
+
+    result = collect_lines(sheet, rules, layout, layout.first_data_row, sheet.n_rows, units="в тыс. рублей")
+
+    expected_warning = (
+        f"значений с дробной частью при единице тыс. руб.: {_FRACTIONAL_VALUES_COUNT} из {_NUMERIC_VALUES_COUNT}"
+    )
+    assert any(warning.startswith(expected_warning) for warning in result.warnings)
+    assert any("строк без кода с суммами пропущено: 1" in warning for warning in result.warnings)
+
+
+def test_no_fractional_warning_without_thousands(rules: Rules) -> None:
+    sheet = sheet_from_rows(ROWS)
+    layout = find_layout(sheet, rules, 1, sheet.n_rows)
+
+    result = collect_lines(sheet, rules, layout, layout.first_data_row, sheet.n_rows, units="руб.")
+
+    assert not any("дробной частью" in warning for warning in result.warnings)
+
+
+def test_line_without_name_column(rules: Rules) -> None:
+    sheet = sheet_from_rows([["Код", "На 31.12.2024"], ["1150", 5], ["1210", 6], ["1600", 11]])
+    layout = find_layout(sheet, rules, 1, 4)
+
+    result = collect_lines(sheet, rules, layout, layout.first_data_row, 4)
+
+    assert [line.line_name for line in result.lines] == ["", "", ""]