"""Tests for the deterministic regulation differ.

These tests pin the behaviour that the previous cosine-similarity implementation
could not deliver: real regulatory edits (numeric limits, deontic modals) must be
detected, and inserting a paragraph must not report unrelated paragraphs as
changed. Everything here runs offline — no LLM, no network, no embeddings.
"""

from __future__ import annotations

from app.infrastructure.perception.regulation_differ import RegulationDiffer


def _differ() -> RegulationDiffer:
    """Build a differ with an explicit ratio so tests never depend on .env."""
    return RegulationDiffer(min_change_ratio=0.02)


def test_identical_documents_report_no_changes():
    """An unchanged regulation must produce an empty change list."""
    text = "第一条 车辆制动系统应在时速50公里条件下于30米内完全停止。\n第二条 驾驶员座椅面料的阻燃性能应符合附录B的规定。"
    assert _differ().diff(text, text) == []


def test_numeric_tightening_is_detected():
    """A changed numeric limit is the case cosine similarity scored 0.9153 and missed."""
    old = "第一条 车辆制动系统应在时速50公里条件下于30米内完全停止。"
    new = "第一条 车辆制动系统应在时速50公里条件下于20米内完全停止。"

    changes = _differ().diff(old, new)

    assert len(changes) == 1
    change = changes[0]
    assert change.change_type == "modified"
    assert change.numeric_changed is True
    assert change.needs_llm is True


def test_deontic_relaxation_is_detected():
    """Weakening 应当 to 宜 changes the legal force and must be flagged."""
    old = "第三条 生产企业应当每年开展一次安全评估。"
    new = "第三条 生产企业宜每年开展一次安全评估。"

    changes = _differ().diff(old, new)

    assert len(changes) == 1
    assert changes[0].deontic_changed is True
    assert changes[0].needs_llm is True


def test_prohibition_removal_is_detected():
    """Dropping 不得 flips a prohibition into a permission."""
    old = "第四条 车辆不得使用未经认证的电池组。"
    new = "第四条 车辆可以使用经备案的电池组。"

    changes = _differ().diff(old, new)

    assert len(changes) == 1
    assert changes[0].deontic_changed is True


def test_inserted_paragraph_does_not_shift_the_rest():
    """Regression test for positional alignment.

    The previous implementation compared old[i] to new[i], so inserting one
    paragraph at the top reported every following paragraph as changed. With
    sequence alignment only the inserted paragraph is new.
    """
    old = "\n".join([
        "第一条 本标准规定了车辆制动系统的技术要求。",
        "第二条 车辆制动系统应在时速50公里条件下于30米内完全停止。",
        "第三条 驾驶员座椅面料的阻燃性能应符合附录B的规定。",
    ])
    new = "\n".join([
        "第零条 本标准适用于所有M1类车辆。",
        "第一条 本标准规定了车辆制动系统的技术要求。",
        "第二条 车辆制动系统应在时速50公里条件下于30米内完全停止。",
        "第三条 驾驶员座椅面料的阻燃性能应符合附录B的规定。",
    ])

    changes = _differ().diff(old, new)

    assert len(changes) == 1, f"expected only the inserted paragraph, got {changes}"
    assert changes[0].change_type == "added"
    assert "第零条" in changes[0].new_text
    assert changes[0].old_text == ""


def test_deleted_paragraph_is_reported_once():
    """Removing a provision yields exactly one 'removed' record."""
    old = "\n".join([
        "第一条 本标准规定了车辆制动系统的技术要求。",
        "第二条 车辆制动系统应在时速50公里条件下于30米内完全停止。",
        "第三条 驾驶员座椅面料的阻燃性能应符合附录B的规定。",
    ])
    new = "\n".join([
        "第一条 本标准规定了车辆制动系统的技术要求。",
        "第三条 驾驶员座椅面料的阻燃性能应符合附录B的规定。",
    ])

    changes = _differ().diff(old, new)

    assert len(changes) == 1
    assert changes[0].change_type == "removed"
    assert "第二条" in changes[0].old_text
    assert changes[0].new_text == ""
    assert changes[0].needs_llm is True


def test_trivial_edit_is_not_sent_to_the_llm():
    """A cosmetic edit with no number or modal change must not cost an LLM call."""
    old = "第五条 本标准由全国汽车标准化技术委员会归口管理。"
    new = "第五条 本标准由全国汽车标准化技术委员会归口管理"

    changes = _differ().diff(old, new)

    for change in changes:
        assert change.numeric_changed is False
        assert change.deontic_changed is False
        assert change.needs_llm is False, f"trivial edit was gated to the LLM: {change}"


def test_large_rewrite_is_sent_to_the_llm():
    """A substantial rewrite clears the change-ratio gate even without numbers or modals."""
    old = "第六条 本标准参考了国际同类标准的相关内容。"
    new = "第六条 本条款描述了完全不同的主题内容，涉及整车认证流程与型式试验的组织安排。"

    changes = _differ().diff(old, new)

    assert len(changes) == 1
    assert changes[0].change_ratio >= 0.02
    assert changes[0].needs_llm is True


def test_empty_old_text_yields_no_changes():
    """A first crawl has no baseline, so there is nothing to diff."""
    assert _differ().diff("", "第一条 任意内容。") == []


def test_unchanged_paragraphs_are_never_returned():
    """Only changed paragraphs appear; equal ones are dropped."""
    old = "\n".join([
        "第一条 保持不变的条款。",
        "第二条 车辆制动距离不得超过30米。",
        "第三条 另一条保持不变的条款。",
    ])
    new = "\n".join([
        "第一条 保持不变的条款。",
        "第二条 车辆制动距离不得超过20米。",
        "第三条 另一条保持不变的条款。",
    ])

    changes = _differ().diff(old, new)

    assert len(changes) == 1
    assert changes[0].numeric_changed is True
    assert "第二条" in changes[0].old_text