"""Раскладка колонок таблицы OCR (§11.1 п. 3): шапка «Код» и вывод по данным.

Перенос `_build_layout`/`_infer_layout` из старого ядра."""

import pytest

from src.normalizer.ocr.layout import Cell, ColumnLayout, build_layout, cell_text, cells_by_index, infer_layout

KNOWN = frozenset({"1110", "1120", "1130", "1150", "2110", "2120"})


def _row(*cells: tuple[int, str, float]) -> dict:
    return {"row_index": 0, "cells": [{"column_index": i, "value": v, "confidence": c} for i, v, c in cells]}


def test_cells_by_index_normalizes_types() -> None:
    cells = cells_by_index(_row((0, "Выручка", 71.4), (1, None, None)))
    assert cells == {0: Cell("Выручка", 71.4), 1: Cell("", 0.0)}
    assert cell_text(cells, 0) == "Выручка"
    assert cell_text(cells, 7) == ""
    assert cell_text(cells, None) == ""


@pytest.mark.parametrize(
    ("raw", "expected"),
    [(float("nan"), 0.0), (150, 100.0), (-5, 0.0), (float("inf"), 0.0), (10**400, 0.0), ("90", 0.0), (None, 0.0)],
)
def test_cells_by_index_clamps_confidence(raw: object, expected: float) -> None:
    """Решение №30: `json.loads` принимает NaN/Infinity — уверенность ячейки приводится к [0, 100], NaN → 0."""
    row = {"cells": [{"column_index": 0, "value": "x", "confidence": raw}]}
    assert cells_by_index(row)[0].confidence == expected


class TestBuildLayout:
    def test_header_with_code_and_years(self) -> None:
        cells = cells_by_index(
            _row(
                (0, "Наименование показателя", 100),
                (1, "Код", 100),
                (2, "На 30 июня\n2025 г", 100),
                (4, "На 31 декабря\n2024 г.", 100),
                (5, "На 31 декабря\n2023 г.", 100),
            )
        )
        assert build_layout(cells) == ColumnLayout(label_idx=0, code_idx=1, value_idxs=(2, 4, 5))

    def test_label_defaults_to_last_filled_left_of_code(self) -> None:
        cells = cells_by_index(_row((0, "Пояснения", 100), (1, "Показатель", 100), (2, "Код", 100), (3, "2024", 100)))
        assert build_layout(cells) == ColumnLayout(label_idx=1, code_idx=2, value_idxs=(3,))

    def test_no_code_cell_is_not_a_header(self) -> None:
        assert build_layout(cells_by_index(_row((0, "Выручка", 100), (1, "2110", 100), (2, "13 070", 100)))) is None


class TestInferLayout:
    def test_aep3_like_rows_without_header(self) -> None:
        rows = [
            cells_by_index(
                _row(
                    (1, "НИ", 0),
                    (0, "Нематериальные активы", 100),
                    (2, "30127", 60),
                    (4, "| |  г", 16),
                    (5, "", 100),
                )
            ),
            cells_by_index(
                _row(
                    (1, "1120", 50),
                    (0, "Результаты исследований и", 94),
                    (2, ":", 0),
                    (4, "--", 0),
                    (5, ":", 0),
                )
            ),
            cells_by_index(
                _row(
                    (1, "1130", 80),
                    (0, "Нематериальные поисковые", 87),
                    (2, "не ****", 5),
                    (4, "3", 0),
                    (5, "+ :", 0),
                )
            ),
            cells_by_index(
                _row(
                    (1, "1150", 100),
                    (0, "Основные средства", 100),
                    (2, "1 234", 100),
                    (4, "5 678", 100),
                    (5, "", 100),
                )
            ),
        ]
        layout = infer_layout(rows, KNOWN)
        assert layout is not None
        assert layout.code_idx == 1
        assert layout.label_idx == 0
        # R-1: в колонке 5 только мусор (пусто, ':', '+ :'), настоящих сумм нет
        # в колонке 4 и мусор, и суммы ('--', '3', '5 678'), но сумм не меньше, чем мусора
        assert layout.value_idxs == (2, 4)

    def test_no_known_codes_is_none(self) -> None:
        rows = [cells_by_index(_row((0, "текст", 100), (1, "мусор", 100)))]
        assert infer_layout(rows, KNOWN) is None

    def test_empty_rows_is_none(self) -> None:
        assert infer_layout([], KNOWN) is None