UHD_Media_Tool/packages/adapters/test/differential/indexer.test.ts

/**
 * 站点索引解析差分:MP `SiteSpider.parse`(PyQuery 回退)vs 本项目 `SiteSpider.parse`。
 * 同一 indexer 配置(本项目 definitions/*.json 或内联配置)+ 同一 HTML;期望由 scripts/mp-sync/gen-differential.py rss|indexer 生成(indexer.json)。
 * search_type=subtitles 的样本走字幕分支(MP get_subtitle_info)。
 */
import { describe, expect, it } from "vitest";
import { BUILTIN_SITE_DEFINITIONS, type IndexerDefinition, SiteSpider } from "../../src/sites/index.ts";
import {
  diffAgainstRegistry,
  dumpMismatches,
  FieldCollector,
  formatMismatches,
  loadDoc,
  readInput,
} from "./diff-util.ts";

interface IndexerSample {
  id: string;
  definition:
    | { source: "definition"; id: string }
    | {
        source: "inline";
        config: unknown;
      };
  input: { file?: string; inline?: string };
  query: { keyword: string | null; search_type: "torrents" | "subtitles" };
  mp: {
    is_error?: boolean;
    torrents?: Array<Record<string, unknown>>;
    status?: string;
    error?: string;
  };
}

const doc = loadDoc<IndexerSample>("indexer.json");

function definitionOf(s: IndexerSample): IndexerDefinition {
  if (s.definition.source === "definition") {
    return BUILTIN_SITE_DEFINITIONS[s.definition.id] as IndexerDefinition;
  }
  return s.definition.config as IndexerDefinition;
}

/**
 * 行对齐:条数相同按下标;条数不同(MP 整行丢弃等)且双方都有 page_url 时按 page_url 配对,
 * 避免一处丢行导致后续所有行错位。行号 i 取 MP 侧下标。
 */
function alignRows(
  mp: Array<Record<string, unknown>>,
  ours: Array<Record<string, unknown>>,
): Array<[number, Record<string, unknown>, Record<string, unknown>]> {
  const byUrl = mp.length !== ours.length && [...mp, ...ours].every((r) => typeof r.page_url === "string");
  if (!byUrl) {
    return mp.slice(0, Math.min(mp.length, ours.length)).map((m, i) => [i, m, ours[i]!]);
  }
  const index = new Map(ours.map((o) => [o.page_url as string, o]));
  const out: Array<[number, Record<string, unknown>, Record<string, unknown>]> = [];
  mp.forEach((m, i) => {
    const o = index.get(m.page_url as string);
    if (o) out.push([i, m, o]);
  });
  return out;
}

describe("站点索引差分(MP SiteSpider Python 回退)", () => {
  const c = new FieldCollector();
  for (const s of doc.samples) {
    if (s.mp.status === "error") {
      throw new Error(`${s.id}: MP 运行失败 ${s.mp.error}`);
    }
    const subtitles = s.query.search_type === "subtitles";
    const html = s.input.file ? readInput(s.input.file) : (s.input.inline ?? "");
    const spider = new SiteSpider(definitionOf(s), {
      ...(s.query.keyword ? { keyword: s.query.keyword } : {}),
      ...(subtitles ? { searchType: "subtitles" as const } : {}),
    });
    const ours = spider.parse(html);
    const mpRows = s.mp.torrents ?? [];
    if (Boolean(s.mp.is_error) !== ours.isError) {
      c.mismatches.push({
        key: `${s.id}.status`,
        mp: s.mp.is_error,
        ours: ours.isError,
      });
    }
    if (mpRows.length !== ours.torrents.length) {
      c.mismatches.push({
        key: `${s.id}.count`,
        mp: mpRows.length,
        ours: ours.torrents.length,
      });
    }
    for (const [i, m, o] of alignRows(mpRows, ours.torrents as Array<Record<string, unknown>>)) {
      // 字幕分支输出字段与种子不同(language_icon/report_url/uploader/torrent_id/subtitle_id…),按双方键并集比较
      const fields = subtitles ? [...new Set([...Object.keys(m), ...Object.keys(o)])].sort() : doc.fields;
      for (const f of fields) c.check(`${s.id}#${i}.${f}`, m[f], o[f]);
    }
  }

  it("有样本且记录了生成来源", () => {
    expect(doc.samples.length).toBeGreaterThan(10);
    expect(doc.generatedFrom.commit).toMatch(/^[0-9a-f]{40}$/);
  });

  it("实际不一致集合 == 登记的 deviations", () => {
    dumpMismatches("indexer", c.mismatches, {
      compared: c.compared,
      equal: c.equal,
    });
    const { unregistered, stale } = diffAgainstRegistry(c.mismatches, doc.deviations);
    expect(formatMismatches(unregistered), "未登记的不一致").toBe("");
    expect(stale.join("\n"), "登记了但已不再出现的偏差(请从 indexer.json deviations 删除)").toBe("");
  });

  it("登记项分类合法", () => {
    for (const [key, d] of Object.entries(doc.deviations)) {
      expect(["intentional", "missing", "our_bug", "mp_bug"], key).toContain(d.kind);
      expect(d.note.length, key).toBeGreaterThan(0);
    }
  });
});