UHD_Media_Tool/packages/domain/test/differential/rules.test.ts

/**
 * MoviePilot 规则串 / 规则组链 差分测试(W9-D)。
 *
 * rules-cases.json 由 scripts/mp-sync/gen-rules-differential.py 用 MP 的 Python 实现生成:
 * - parse:每个规则串按 `>` 切层后 pyparsing RuleParser 的 as_list();
 * - cases:FilterModule.filter_torrents(多组串联,pri 取最后一组)的 pri_order 与 debug trace。
 * 我们用 parseRuleString → evaluateRuleChain 求同样的 pri。
 * 断言:解析结构全部一致;pri 不一致集合 == deviations 登记集合(补迁消除差异后测试会失败,提醒删登记)。
 */
import { readFileSync } from "node:fs";
import { describe, expect, it } from "vitest";
import {
  BUILTIN_ATOMS,
  catalogOf,
  customAtomFromMp,
  evaluateRuleChain,
  fromMpParseList,
  normalizeExpr,
  parseRelease,
  parseRuleString,
  type RuleChainResult,
  type RuleGroupSpec,
  serializeLayers,
} from "../../src/index.ts";

interface Torrent {
  id: string;
  title: string;
  description: string;
  size: number;
  seeders: number;
  labels?: string[];
  downloadvolumefactor?: number;
}
interface Rule {
  id: string;
  groups: string[];
  mediainfo?: { original_language?: string };
}
interface Case {
  caseId: string;
  ruleId: string;
  torrentId: string;
  mp: { pri: number | null; trace: string[]; rustTrace?: string[] };
}
type DeviationKind = "intentional" | "missing" | "our_bug" | "mp_bug" | "not_expressible";
interface Deviation {
  kind: DeviationKind;
  note: string;
  mp?: number | null;
  ours?: number | null;
}
interface Fixture {
  generatedFrom: { commit: string };
  customRules: Record<string, Record<string, string>>;
  torrents: Torrent[];
  rules: Rule[];
  parse: Record<string, unknown[]>;
  cases: Case[];
  deviations: Record<string, Deviation>;
}

const fx = JSON.parse(readFileSync(new URL("./rules-cases.json", import.meta.url), "utf8")) as Fixture;
const custom = Object.entries(fx.customRules).map(([id, r]) => customAtomFromMp(id, r));
const catalog = catalogOf(custom);
const customIds = custom.map((a) => a.id);
const torrents = new Map(fx.torrents.map((t) => [t.id, t]));
const rules = new Map(fx.rules.map((r) => [r.id, r]));

function groupsOf(rule: Rule): RuleGroupSpec[] {
  return rule.groups.map((s, i) => {
    const p = parseRuleString(s, { customAtoms: customIds, allowUnknown: true });
    if (p.errors.length) throw new Error(`${rule.id}: ${JSON.stringify(p.errors)}`);
    return { name: `${rule.id}#${i}`, layers: p.layers };
  });
}

function ours(c: Case): RuleChainResult {
  const rule = rules.get(c.ruleId)!;
  const t = torrents.get(c.torrentId)!;
  const parsed = { ...parseRelease(t.title, t.description ? { subtitle: t.description } : {}), sizeBytes: t.size };
  return evaluateRuleChain(
    groupsOf(rule),
    parsed,
    {
      ...(t.description ? { subtitle: t.description } : {}),
      labels: t.labels ?? [],
      seeders: t.seeders,
      downloadVolumeFactor: t.downloadvolumefactor ?? 1,
      ...(rule.mediainfo?.original_language ? { originalLanguage: rule.mediainfo.original_language } : {}),
    },
    catalog,
  );
}
const priOf = (r: RuleChainResult) => (r.passed ? (r.pri ?? 100) : null);

describe("MP 规则串解析差分(rules-cases.json#parse)", () => {
  it("≥40 条规则链,覆盖全部支持的内置关键字", () => {
    expect(fx.rules.length).toBeGreaterThanOrEqual(40);
    const used = new Set(Object.keys(fx.parse).flatMap((s) => s.match(/[A-Za-z0-9]+/g) ?? []));
    const supported = BUILTIN_ATOMS.filter((a) => a.supported).map((a) => a.id);
    expect(supported.filter((id) => !used.has(id))).toEqual([]);
  });

  it("每个规则串的分层与层内结构 == MP RuleParser", () => {
    for (const [s, mpLayers] of Object.entries(fx.parse)) {
      const p = parseRuleString(s, { customAtoms: customIds, allowUnknown: true });
      expect(p.errors, s).toEqual([]);
      expect(
        p.layers.map((l) => l.expr),
        s,
      ).toEqual(mpLayers.map((l) => normalizeExpr(fromMpParseList(l))));
    }
  });

  it("序列化后再解析结构不变(规则串双向映射)", () => {
    for (const s of Object.keys(fx.parse)) {
      const a = parseRuleString(s, { customAtoms: customIds, allowUnknown: true }).layers;
      const text = serializeLayers(a);
      const b = parseRuleString(text, { customAtoms: customIds, allowUnknown: true }).layers;
      expect(b, `${s} → ${text}`).toEqual(a);
    }
  });
});

describe("MP 规则组过滤差分(rules-cases.json#cases)", () => {
  it("pri 不一致集合 == deviations 登记集合", () => {
    const actual: Record<string, { mp: number | null; ours: number | null }> = {};
    for (const c of fx.cases) {
      const o = priOf(ours(c));
      if (o !== c.mp.pri) actual[c.caseId] = { mp: c.mp.pri, ours: o };
    }
    if (process.env.UMT_RULES_DUMP) console.log(`UMT_RULES_DUMP ${JSON.stringify(actual)}`);
    expect(Object.keys(actual).sort()).toEqual(Object.keys(fx.deviations).sort());
    for (const [k, v] of Object.entries(fx.deviations)) {
      expect(["intentional", "missing", "our_bug", "mp_bug", "not_expressible"]).toContain(v.kind);
      expect(actual[k], k).toEqual({ mp: v.mp ?? null, ours: v.ours ?? null });
    }
    // 一致率(报告用):登记外全部一致
    expect(fx.cases.length - Object.keys(actual).length).toBeGreaterThan(fx.cases.length * 0.9);
  });

  it("trace 可读:过滤给出未命中层与原子依据,放行给出命中层与 pri,未知条件写明不存在", () => {
    for (const c of fx.cases) {
      const r = ours(c);
      const text = r.trace.join("\n");
      if (!r.passed) expect(text, c.caseId).toMatch(/未命中|硬排除|过滤/);
      const withLayers = r.groups.some((g) => g.pri !== null);
      if (r.passed && withLayers) expect(text, c.caseId).toMatch(/命中,pri=\d+/);
      // MP trace 里提到的"不包含任何项"对应我们 ✗ 的原子依据
      if (c.mp.trace.some((x) => x.startsWith("不包含任何项")) && !r.passed) expect(text).toMatch(/✗/);
      if (c.mp.trace.some((x) => /规则 \w+ 不存在/.test(x))) expect(text).toMatch(/条件不存在/);
    }
  });
});

describe("规则串错误", () => {
  it("未知条件给位置与相近名;音乐条件不支持;括号/空层", () => {
    const a = parseRuleString("4K & CNSUBB > 1080P");
    expect(a.errors[0]).toMatchObject({ code: "unknown_atom", position: 5, layer: 0, atom: "CNSUBB" });
    expect(a.errors[0]!.suggestions).toContain("CNSUB");
    expect(parseRuleString("FLAC").errors[0]).toMatchObject({ code: "unsupported_atom", atom: "FLAC" });
    expect(parseRuleString("(4K | 1080P").errors[0]).toMatchObject({ code: "unbalanced_paren", position: 0 });
    expect(parseRuleString("4K)").errors[0]).toMatchObject({ code: "unbalanced_paren", position: 2 });
    expect(parseRuleString("4K >  > 720P").errors[0]).toMatchObject({ code: "empty_layer", layer: 1, position: 4 });
    expect(parseRuleString("4K 1080P").errors[0]).toMatchObject({ code: "unexpected_token", position: 3 });
    expect(parseRuleString("4K & $").errors[0]).toMatchObject({ code: "invalid_char", position: 5 });
    expect(parseRuleString("").errors[0]!.code).toBe("empty");
    expect(parseRuleString("4k & !remux").layers.map((l) => l.expr)).toEqual([
      {
        op: "and",
        args: [
          { op: "atom", id: "4K" },
          { op: "not", arg: { op: "atom", id: "REMUX" } },
        ],
      },
    ]);
  });

  it("序列化只加必要括号", () => {
    const s = "SPECSUB & CNSUB > (4K | 1080P) & !(BLU | REMUX) > !!3D | 720P & H265";
    expect(serializeLayers(parseRuleString(s).layers)).toBe(s);
  });
});