tests/test_compatibility_rules.py

"""确定性相容性规则引擎 ``skills.compatibility.rules`` 的单元 / 属性测试（任务 18）。

覆盖需求 8.2：相容性「计算」为确定性官能团 → 风险规则引擎（纯 Python，风险分级）。

测试取向：
- 规则正确性：已知官能团 × 辅料特性组合命中预期反应与风险等级（红 / 黄 / 绿）。
- 确定性：相同输入恒得相同输出（无随机 / 无网络 / 无 LLM）。
- 聚合正确：整体风险 = 各辅料风险的最坏者（worse 聚合）。
- 鲁棒性：未知辅料 / 空输入 / dict 与 str 混合输入均稳健处理。

导入路径由 tests/conftest.py 统一设置（仓库根加入 sys.path，顶层 skills 包可导入）。
"""

from __future__ import annotations

import pytest

from skills.compatibility import rules


# ---------------------------------------------------------------------------
# 辅料解析
# ---------------------------------------------------------------------------

def test_resolve_known_excipient_lactose_is_reducing_sugar():
    profile = rules.resolve_excipient("乳糖")
    assert profile.known is True
    assert "reducing_sugar" in profile.tags


def test_resolve_excipient_english_alias_case_insensitive():
    profile = rules.resolve_excipient("Magnesium Stearate")
    assert profile.known is True
    assert "alkaline" in profile.tags


def test_resolve_unknown_excipient_marked_not_known_with_no_tags():
    profile = rules.resolve_excipient("某种未收录辅料XYZ")
    assert profile.known is False
    assert profile.tags == frozenset()


def test_resolve_empty_name_returns_unknown_placeholder():
    profile = rules.resolve_excipient("")
    assert profile.known is False
    assert profile.tags == frozenset()


# ---------------------------------------------------------------------------
# 规则命中：五大反应维度
# ---------------------------------------------------------------------------

def test_maillard_high_risk_amine_plus_reducing_sugar():
    """伯胺 + 还原糖（乳糖）→ 美拉德高风险。"""
    assessment = rules.evaluate(
        functional_groups=[{"id": "primary_amine"}],
        excipients=["乳糖"],
    )
    assert assessment.overall_risk == rules.RISK_HIGH
    er = assessment.excipient_results[0]
    reactions = {r.reaction: r.risk for r in er.reactions}
    assert reactions.get("maillard") == rules.RISK_HIGH


def test_oxidation_high_risk_thiol_plus_peroxide():
    """巯基 + 过氧化物（交联聚维酮）→ 氧化高风险。"""
    assessment = rules.evaluate(["thiol"], ["交联聚维酮"])
    er = assessment.excipient_results[0]
    reactions = {r.reaction: r.risk for r in er.reactions}
    assert reactions.get("oxidation") == rules.RISK_HIGH
    assert assessment.overall_risk == rules.RISK_HIGH


def test_oxidation_medium_risk_phenol_plus_peroxide():
    """酚羟基 + 过氧化物 → 氧化中风险（非高）。"""
    assessment = rules.evaluate(["phenol"], ["聚维酮"])
    er = assessment.excipient_results[0]
    reactions = {r.reaction: r.risk for r in er.reactions}
    assert reactions.get("oxidation") == rules.RISK_MEDIUM


def test_hydrolysis_high_risk_ester_plus_alkaline():
    """酯基 + 碱性辅料（硬脂酸镁）→ 碱催化水解高风险。"""
    assessment = rules.evaluate(["ester"], ["硬脂酸镁"])
    er = assessment.excipient_results[0]
    reactions = {r.reaction: r.risk for r in er.reactions}
    assert reactions.get("hydrolysis") == rules.RISK_HIGH


def test_acid_base_medium_risk_carboxyl_plus_alkaline():
    """羧基 + 碱性辅料 → 酸碱相互作用中风险。"""
    assessment = rules.evaluate(["carboxylic_acid"], ["磷酸氢钙"])
    er = assessment.excipient_results[0]
    reactions = {r.reaction: r.risk for r in er.reactions}
    assert reactions.get("acid_base") == rules.RISK_MEDIUM


def test_adsorption_low_risk_any_group_plus_high_surface_area():
    """任意官能团 + 高比表面积辅料（胶态二氧化硅）→ 吸附低风险。"""
    assessment = rules.evaluate(["alcohol"], ["胶态二氧化硅"])
    er = assessment.excipient_results[0]
    reactions = {r.reaction: r.risk for r in er.reactions}
    assert reactions.get("adsorption") == rules.RISK_LOW


# ---------------------------------------------------------------------------
# 无反应 / 惰性辅料
# ---------------------------------------------------------------------------

def test_inert_excipient_yields_no_significant_risk():
    """甘露醇（非还原糖醇，惰性）+ 伯胺 → 无显著相互作用。"""
    assessment = rules.evaluate(["primary_amine"], ["甘露醇"])
    er = assessment.excipient_results[0]
    assert er.reactions == []
    assert er.risk == rules.RISK_NONE
    assert assessment.overall_risk == rules.RISK_NONE


def test_no_functional_groups_adds_note():
    """无官能团时仍可评估辅料，但附低置信度提示。"""
    assessment = rules.evaluate([], ["乳糖"])
    # 无官能团 → 美拉德等规则不命中。
    assert assessment.excipient_results[0].reactions == []
    assert any("未识别到 API 官能团" in n for n in assessment.notes)


# ---------------------------------------------------------------------------
# 整体风险聚合（取最坏）
# ---------------------------------------------------------------------------

def test_overall_risk_is_worst_of_all_excipients():
    """多辅料：整体风险取各辅料的最坏者。"""
    assessment = rules.evaluate(
        functional_groups=["primary_amine", "ester"],
        excipients=["甘露醇", "胶态二氧化硅", "乳糖"],  # none, low(adsorption), high(maillard)
    )
    risks = {er.display_zh: er.risk for er in assessment.excipient_results}
    assert risks["甘露醇"] == rules.RISK_NONE
    assert risks["胶态二氧化硅"] == rules.RISK_LOW
    assert risks["乳糖"] == rules.RISK_HIGH
    assert assessment.overall_risk == rules.RISK_HIGH


def test_counts_and_summary_structure():
    assessment = rules.evaluate(["primary_amine", "ester"], ["乳糖", "甘露醇"])
    summary = assessment.summary()
    assert summary["n_excipients"] == 2
    assert summary["overall_risk"] == rules.RISK_HIGH
    assert summary["n_high"] >= 1
    assert isinstance(summary["excipients"], list)
    # 风险色块包含整体 + 每个辅料。
    blocks = assessment.risk_blocks()
    assert len(blocks) == 1 + 2
    assert blocks[0]["color"] == rules.QBD_RISK_COLORS[rules.RISK_HIGH]


# ---------------------------------------------------------------------------
# 输入归一鲁棒性
# ---------------------------------------------------------------------------

def test_accepts_string_and_dict_inputs_equivalently():
    """str 与 dict 形式的官能团 / 辅料输入应等价。"""
    a = rules.evaluate(["primary_amine"], ["乳糖"])
    b = rules.evaluate([{"id": "primary_amine"}], [{"name": "乳糖"}])
    assert a.overall_risk == b.overall_risk
    assert a.excipient_results[0].risk == b.excipient_results[0].risk


def test_duplicate_excipients_deduplicated():
    assessment = rules.evaluate(["primary_amine"], ["乳糖", "乳糖", "Lactose"])
    # 乳糖 与 Lactose 命中同一档；去重按名称小写。
    assert len(assessment.excipient_results) == 2  # "乳糖" 与 "Lactose" 名称不同


def test_empty_excipients_returns_empty_results_with_note():
    assessment = rules.evaluate(["primary_amine"], [])
    assert assessment.excipient_results == []
    assert assessment.overall_risk == rules.RISK_NONE
    assert any("未提供任何辅料" in n for n in assessment.notes)


# ---------------------------------------------------------------------------
# 确定性（属性式）：相同输入恒得相同输出
# ---------------------------------------------------------------------------

@pytest.mark.parametrize(
    "groups,excips",
    [
        (["primary_amine"], ["乳糖"]),
        (["thiol", "ester"], ["交联聚维酮", "硬脂酸镁"]),
        (["phenol"], ["聚维酮", "甘露醇"]),
        ([], ["未知辅料A", "微晶纤维素"]),
    ],
)
def test_determinism_same_input_same_output(groups, excips):
    """规则引擎确定性：多次评估结果完全一致。"""
    first = rules.evaluate(groups, excips).summary()
    for _ in range(3):
        again = rules.evaluate(groups, excips).summary()
        assert again == first


def test_worse_helper_picks_more_severe():
    assert rules.worse(rules.RISK_LOW, rules.RISK_HIGH) == rules.RISK_HIGH
    assert rules.worse(rules.RISK_MEDIUM, rules.RISK_LOW) == rules.RISK_MEDIUM
    assert rules.worse(rules.RISK_NONE, rules.RISK_NONE) == rules.RISK_NONE


def test_risk_color_maps_to_qbd_palette():
    assert rules.risk_color(rules.RISK_HIGH) == "#dc3545"
    assert rules.risk_color(rules.RISK_MEDIUM) == "#ffc107"
    assert rules.risk_color(rules.RISK_LOW) == "#28a745"
    # 未知等级回退绿。
    assert rules.risk_color("???") == "#28a745"


def test_merged_reaction_takes_worst_risk_and_merges_groups():
    """同一反应被多条规则命中时，取最坏风险并合并触发官能团。

    氧化：thiol→高、phenol→中；二者同存时该反应应为高，且两官能团都被记录。
    """
    reactions = rules.assess_pair(
        group_ids=["thiol", "phenol"], tags=["peroxide"]
    )
    ox = next(r for r in reactions if r.reaction == "oxidation")
    assert ox.risk == rules.RISK_HIGH
    assert "thiol" in ox.group_ids and "phenol" in ox.group_ids