# -*- coding: utf-8 -*-
"""pipeline 模块测试：排序、成本计算、缓存路径。"""
from violation_detector.pipeline import (
    _cache_path, _cat_num, deepseek_cost, natural_key,
)


def R(cat):
    return {"category": cat, "attr": "a", "logic": "l", "file": "x.jpg",
            "usage": {"prompt": 2199, "completion": 2282, "cached": 1792}}


def test_natural_key_sort():
    names = ["10_a.jpg", "2_a.jpg", "1_a.jpg", "21_a.jpg", "3_a.jpg"]
    assert sorted(names, key=natural_key) == ["1_a.jpg", "2_a.jpg", "3_a.jpg",
                                              "10_a.jpg", "21_a.jpg"]


def test_cat_num():
    assert _cat_num("12. 侵权 - 其他") == 12
    assert _cat_num("1. 烟草") == 1
    # 无编号的结论统一排到最后（无违规/违规不明不变动分类）
    assert _cat_num("无违规") == 99
    assert _cat_num("违规不明") == 99


def test_deepseek_cost_known_usage():
    # miss 407 + cached 1792 + completion 2282
    results = {"a": R("无违规")}
    peak, idle, comp = deepseek_cost(results)
    expect_peak = (407 * 3.0 + 1792 * 0.10 + 2282 * 9.0) / 1e6
    expect_idle = (407 * 1.5 + 1792 * 0.05 + 2282 * 4.5) / 1e6
    assert abs(peak - expect_peak) < 1e-9
    assert abs(idle - expect_idle) < 1e-9
    assert comp == 2282


def test_deepseek_cost_empty():
    assert deepseek_cost({}) == (0.0, 0.0, 0.0)


def test_cache_path_sanitizes_folder(tmp_path):
    p = _cache_path("ark", str(tmp_path / "we ird\\name+:x"))
    assert p.parent.exists()
    assert p.name.startswith("ark_")
    for ch in ' :+':
        assert ch not in p.name