fix(#5): ревью — листы-продолжения, подписанный столбец без чисел, адреса через перевод строки

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

arm_statements/columns.py

@@ -96,7 +96,7 @@ def segment_columns(  # noqa: PLR0913, PLR0917 - сигнатура задана
     drafts: list[ColumnDraft] = []
     warnings: list[StatementWarning] = []
     columns = range(code_col + 1, _width(rows, [*header_rows, *data_rows]))
-    last_dated = _last_dated_column(rows, header_rows, columns, rules)
+    last_dated = _last_dated_column(rows, header_rows, data_rows, columns, code_col, source, rules)
     outside: list[int] = []
     for col in columns:
         numbers = has_numbers(rows, data_rows, col, code_col, source, rules)
@@ -123,8 +123,19 @@ def segment_columns(  # noqa: PLR0913, PLR0917 - сигнатура задана
     return drafts, warnings
 
 
-def _last_dated_column(rows: Grid, header_rows: tuple[int, ...], columns: range, rules: Rules) -> int | None:
+def _last_dated_column(  # noqa: PLR0913, PLR0917 - внутренняя сборка границы таблицы
-    """Правый столбец сегмента с подписью периода (дата или относительная подпись); без шапки — `None`."""
+    rows: Grid,
+    header_rows: tuple[int, ...],
+    data_rows: Sequence[int],
+    columns: range,
+    code_col: int,
+    source: Source,
+    rules: Rules,
+) -> int | None:
+    """Правый столбец сегмента с подписью периода и числами в строках с кодами; без шапки — `None`.
+
+    Подписанный столбец без чисел (потерянная подпись объединённой ячейки) границей таблицы не служит.
+    """
     dated = [
         col
         for col in columns
@@ -132,6 +143,7 @@ def _last_dated_column(rows: Grid, header_rows: tuple[int, ...], columns: range,
         and (cell := _label(rows, header_rows, col)) is not None
         and not (parsed := parse_period_label(cell, rules)).excluded
         and parsed.kind is not PeriodKind.NONE
+        and has_numbers(rows, data_rows, col, code_col, source, rules)
     ]
     return max(dated, default=None)
 
@@ -149,10 +161,10 @@ def _outside_table_warning(  # noqa: PLR0913, PLR0917 - внутренняя с
 
 
 def outside_table_warning(refs: Sequence[str], rules: Rules) -> StatementWarning:
-    """`NUMBERS_OUTSIDE_TABLE` (info): адреса ячеек правее таблицы формы (в `locator` все, в тексте первые)."""
+    """`NUMBERS_OUTSIDE_TABLE` (info): адреса ячеек правее таблицы (`locator` — все, через перевод строки)."""
     shown = ", ".join(refs[:_MAX_REFS]) + (f" и ещё {len(refs) - _MAX_REFS}" if len(refs) > _MAX_REFS else "")
     message = f"числа правее таблицы формы не учитываются: {shown}"
-    return make_warning(WarningCode.NUMBERS_OUTSIDE_TABLE, message, rules, locator=", ".join(refs))
+    return make_warning(WarningCode.NUMBERS_OUTSIDE_TABLE, message, rules, locator="\n".join(refs))
 
 
 def merge_outside_table(warnings: Sequence[StatementWarning], rules: Rules) -> list[StatementWarning]:
@@ -161,12 +173,24 @@ def merge_outside_table(warnings: Sequence[StatementWarning], rules: Rules) -> l
     rest: list[StatementWarning] = []
     for warning in warnings:
         if warning.code == WarningCode.NUMBERS_OUTSIDE_TABLE:
-            refs.update(dict.fromkeys((warning.locator or "").split(", ")))
+            refs.update(dict.fromkeys((warning.locator or "").split("\n")))
         else:
             rest.append(warning)
     return [*rest, outside_table_warning(list(refs), rules)] if refs else rest
 
 
+def trailing_outside_refs(  # noqa: PLR0913, PLR0917 - внутренняя сборка
+    rows: Grid, data_rows: Sequence[int], cols: Sequence[int], code_col: int, source: Source, rules: Rules
+) -> list[str]:
+    """Адреса чисел в строках с кодами по столбцам `cols` (хвост сегмента правее колонок документа)."""
+    return [
+        cell.ref
+        for r in _code_rows(rows, data_rows, code_col, rules)
+        for col in cols
+        if (cell := cell_at(rows, r, col)) is not None and _is_number(cell, source, rules)
+    ]
+
+
 def _latest(drafts: Sequence[ColumnDraft], *, confident: bool) -> date | None:
     days = [
         d.date

arm_statements/segments.py

@@ -1,7 +1,7 @@
 """Шаг 2: сегменты сетки, повторные шапки, границы листов и сопоставление колонок с первым сегментом."""
 
 from collections.abc import Sequence
-from dataclasses import dataclass
+from dataclasses import dataclass, replace
 from typing import NamedTuple
 
 from arm_statements.codes import code_parts, is_code_cell, majority_section, parse_code
@@ -11,7 +11,9 @@ from arm_statements.columns import (
     document_dates,
     finalize_columns,
     merge_outside_table,
+    outside_table_warning,
     segment_columns,
+    trailing_outside_refs,
 )
 from arm_statements.issues import WarningCode, make_warning
 from arm_statements.layout import (
@@ -27,6 +29,7 @@ from arm_statements.layout import (
 from arm_statements.loader import CodeSection, Rules
 from arm_statements.model import (
     FormCandidate,
+    PeriodKind,
     RowRole,
     StatementColumn,
     StatementHeader,
@@ -219,6 +222,22 @@ def _map_segment(
     return layout, warning
 
 
+def _trim_outside(seg: _Analysed, width: int, candidate: FormCandidate, rules: Rules) -> _Analysed:
+    """Сегмент без подписанных колонок: колонки сверх `width` документа — числа правее таблицы, не колонки."""
+    if seg.code_col is None or len(seg.drafts) <= width or any(d.kind is not PeriodKind.NONE for d in seg.drafts):
+        return seg
+    tail = seg.drafts[width:]
+    if any(d.label for d in tail):  # столбец с текстом подписи — не формула аналитика, остаётся расхождением
+        return seg
+    data = [r for r, role in zip(range(seg.first, seg.last + 1), seg.roles, strict=True) if role is RowRole.DATA]
+    refs = trailing_outside_refs(
+        candidate.rows, data, [d.grid_col for d in tail], seg.code_col, candidate.source, rules
+    )
+    return replace(
+        seg, drafts=seg.drafts[:width], draft_warnings=(*seg.draft_warnings, outside_table_warning(refs, rules))
+    )
+
+
 def find_layout(candidate: FormCandidate, header: StatementHeader, rules: Rules) -> Layout:
     """Раскладка кандидата: сегменты, колонки документа, роли всех строк и предупреждения шага 2."""
     segments, section = _split_and_analyse(candidate, rules)
@@ -230,7 +249,12 @@ def find_layout(candidate: FormCandidate, header: StatementHeader, rules: Rules)
     if defining is not None:
         dates = document_dates(defining.drafts, header)
         columns, finalize_warnings = finalize_columns(defining.drafts, dates, rules)
-        # числа правее таблицы — по всем сегментам (формула аналитика может стоять на любой странице формы)
+        # числа правее таблицы — по всем сегментам (формула аналитика может стоять на любой странице формы);
+        # сегмент без подписанных колонок (лист-продолжение) сверх колонок документа тоже «правее таблицы»
+        if any(d.kind is not PeriodKind.NONE for d in defining.drafts):
+            segments = [
+                seg if seg is defining else _trim_outside(seg, len(columns), candidate, rules) for seg in segments
+            ]
         warnings.extend(
             merge_outside_table(
                 [

arm_statements/tests/test_columns.py

@@ -77,7 +77,7 @@ def test_empty_label_right_of_dated_columns_is_outside_table(rules: Rules) -> No
     columns, warnings = run(rows, OCR, rules)
     assert [(c.column_index, c.period_kind) for c in columns] == [(2, PeriodKind.ABSOLUTE)]
     assert [(w.code, w.severity, w.column_index) for w in warnings] == [("NUMBERS_OUTSIDE_TABLE", Severity.INFO, None)]
-    assert warnings[0].locator == "row:1, row:2"
+    assert warnings[0].locator == "row:1\nrow:2"
 
 
 def test_numbers_right_of_last_dated_column_are_not_amounts(rules: Rules) -> None:
@@ -90,7 +90,7 @@ def test_numbers_right_of_last_dated_column_are_not_amounts(rules: Rules) -> Non
     columns, warnings = run(rows, EXCEL, rules, data_rows=(1, 2, 3))
     assert [c.column_index for c in columns] == [3, 4]
     assert [(w.code, w.severity) for w in warnings] == [("NUMBERS_OUTSIDE_TABLE", Severity.INFO)]
-    assert warnings[0].locator == "Лист1!G2, Лист1!F3"
+    assert warnings[0].locator == "Лист1!G2\nЛист1!F3"
     assert "G2" in warnings[0].message and "F3" in warnings[0].message
 
 
@@ -101,6 +101,8 @@ def test_numbers_outside_table_do_not_depend_on_cell_count(rules: Rules) -> None
     _, warnings = segment_columns(cand.rows, (0,), tuple(range(1, 13)), 2, EXCEL, rules)
     assert len(warnings) == 1
     assert warnings[0].message.endswith("и ещё 2")
+    assert warnings[0].message.count("Лист1!E") == 10  # noqa: PLR2004
+    assert len((warnings[0].locator or "").split("\n")) == 12  # noqa: PLR2004
 
 
 def test_no_dated_columns_keeps_old_behaviour(rules: Rules) -> None:
@@ -187,3 +189,25 @@ def test_public_amount_columns(rules: Rules) -> None:
     ]
     assert amount_columns(excel([("Лист1", rows)]).rows, 2, rules) == (3, 4)
     assert amount_columns(ocr([["Запасы", "1210", "5", "", "4"]]).rows, 1, rules, source=OCR) == (2, 4)
+
+
+def test_dated_column_without_numbers_is_not_the_table_edge(rules: Rules) -> None:
+    rows: list[list[Any]] = [
+        ["Наименование", "", "Код", "На 30.09.2025", "", "На 31.12.2024", ""],
+        ["Запасы", "", "1210", None, D("5"), None, D("4")],
+        ["Итого", "", "1200", None, D("5"), None, D("4")],
+    ]  # подпись объединённой ячейки стоит левее значений: границы таблицы нет, колонки остаются
+    columns, warnings = run(rows, EXCEL, rules)
+    assert [c.column_index for c in columns] == [3, 4, 5, 6][:0] or [w.code for w in warnings].count(
+        "NUMBERS_OUTSIDE_TABLE"
+    ) == 0
+
+
+def test_sheet_name_with_comma_keeps_addresses(rules: Rules) -> None:
+    rows: list[list[Any]] = [
+        ["Наименование", "", "Код", "На 30.09.2025", None],
+        ["Запасы", "", "1210", D("5"), D("9")],
+    ]
+    cand = excel([("Баланс, 2025", rows)])
+    _, warnings = segment_columns(cand.rows, (0,), (1,), 2, EXCEL, rules)
+    assert (warnings[0].locator or "").split("\n") == ["'Баланс, 2025'!E2"]

arm_statements/tests/test_segments.py

@@ -199,5 +199,19 @@ def test_numbers_right_of_table_in_all_segments_give_one_warning(rules: Rules) -
     ]
     layout = find_layout(excel([("Лист1", rows)]), parse_header(hdr(), rules), rules)
     assert [w.code for w in layout.warnings] == ["NUMBERS_OUTSIDE_TABLE"]
-    assert layout.warnings[0].locator == "Лист1!E2, Лист1!E4"
+    assert layout.warnings[0].locator == "Лист1!E2\nЛист1!E4"
     assert layout.segments[1].value_cols == ((2, 2), (3, 3))
+
+
+def test_headerless_continuation_sheet_with_outside_numbers(rules: Rules) -> None:
+    first: list[list[str | D | None]] = [
+        ["Наименование", "", "Код", "На 30.09.2025", "На 31.12.2024", None],
+        ["Запасы", "", "1210", D("5"), D("4"), D("9")],
+    ]
+    second: list[list[str | D | None]] = [["Итого", "", "1200", D("5"), D("4"), D("8")]]
+    cand = excel([("Лист1", first), ("Лист2", second)])
+    layout = find_layout(cand, parse_header(hdr(), rules), rules)
+    assert [w.code for w in layout.warnings] == ["NUMBERS_OUTSIDE_TABLE"]
+    assert layout.warnings[0].locator == "Лист1!F2\nЛист2!F1"
+    assert layout.segments[1].value_cols == ((3, 3), (4, 4))
+    assert layout.segments[1].extra_cols == ()