"""decision 包测试。"""

from __future__ import annotations

import numpy as np
import pytest

pytestmark = pytest.mark.skip(reason="未实现：见 docs/03-roadmap.md L1")


def test_topsis_ranks_dominated_alternative_last():
    """一个在所有指标上都最差的方案，TOPSIS 必须排最后。"""
    from mcm_toolkit.decision import topsis

    X = np.array(
        [
            [9.0, 8.0, 7.0],   # 全面最优
            [5.0, 5.0, 5.0],   # 中等
            [1.0, 1.0, 1.0],   # 全面最差
        ]
    )
    r = topsis(X, benefit_columns=[0, 1, 2])
    ranks = dict(zip(range(3), r.data["rank"], strict=False))
    assert ranks[0] == 1
    assert ranks[2] == 3


def test_topsis_direction_flips_result():
    """把某列从效益型改成成本型，排名必须变——方向不能静默搞错。"""
    from mcm_toolkit.decision import topsis

    X = np.array([[9.0, 1.0], [1.0, 9.0]])
    a = topsis(X, benefit_columns=[0, 1])
    b = topsis(X, benefit_columns=[0])
    assert list(a.data["rank"]) != list(b.data["rank"])


def test_entropy_weights_sum_to_one_and_prefers_dispersed_column(rng):
    from mcm_toolkit.decision import entropy_weights

    X = np.column_stack([rng.normal(size=50), np.ones(50)])  # 第二列无差异
    r = entropy_weights(X)
    assert np.isclose(r.params["weights"].sum(), 1.0)
    assert r.params["weights"][0] > r.params["weights"][1]


def test_entropy_weights_handles_zero_and_negative():
    """熵权法的三个坑之一：零值不能产生 NaN。"""
    from mcm_toolkit.decision import entropy_weights

    r = entropy_weights(np.array([[0.0, 1.0], [1.0, 2.0], [2.0, 0.0]]))
    assert np.isfinite(r.params["weights"]).all()


def test_ahp_consistency_for_consistent_matrix():
    """完全一致的判断矩阵 CR 应接近 0。"""
    from mcm_toolkit.decision import ahp_weights

    A = np.array([[1, 2, 4], [0.5, 1, 2], [0.25, 0.5, 1]], dtype=float)
    r = ahp_weights(A)
    assert r.metrics["CR"] < 0.01


def test_ahp_weights_match_geometric_mean_for_consistent_matrix():
    from mcm_toolkit.decision import ahp_weights

    A = np.array([[1, 2, 4], [0.5, 1, 2], [0.25, 0.5, 1]], dtype=float)
    r = ahp_weights(A)
    geo = (A.prod(axis=1)) ** (1 / 3)
    assert np.allclose(r.params["weights"], geo / geo.sum(), atol=1e-3)


def test_gm11_fits_exponential_sequence():
    """GM(1,1) 对纯指数序列应拟合得极好。"""
    from mcm_toolkit.decision import gm11_fit_predict

    series = [2.0 * 1.1**k for k in range(6)]
    r = gm11_fit_predict(series, steps=2)
    assert r.metrics["c_ratio"] < 0.35  # 后验差一级