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

/**
 * RSS 解析差分:MP `RssHelper._parse_impl`(lxml 回退)vs 本项目 `parseFeed`。
 * 期望由 scripts/mp-sync/gen-differential.py rss|indexer 生成(rss.json);不一致必须登记在 rss.json `deviations`。
 */
import { describe, expect, it } from "vitest";
import { FeedError, type FeedItem, parseFeed } from "../../src/rss/index.ts";
import {
  diffAgainstRegistry,
  dumpMismatches,
  FieldCollector,
  formatMismatches,
  loadDoc,
  readInput,
} from "./diff-util.ts";

interface MpPubdate {
  kind: "absent" | "unparsed" | "aware" | "naive";
  text?: string;
  wall?: string;
  utc?: string;
}
interface MpRssItem {
  title: string;
  description: string;
  link: string;
  enclosure: string;
  size: number;
  nickname?: string;
  pubdate: MpPubdate;
}
interface RssSample {
  id: string;
  input: { file: string; sha256: string };
  mp: { status: "ok" | "false_error" | "none_expired"; items: MpRssItem[] };
}

const doc = loadDoc<RssSample>("rss.json");

/** MP 字段 → 可与本项目比较的值:空串即缺省;pubdate 统一为 UTC ISO */
function mpPubdate(p: MpPubdate): string | undefined {
  if (p.kind === "aware") return p.utc;
  // naive:MP astimezone(None) 以宿主本地时区解释墙上时间;JS 对无偏移的日期时间同样按本地解释
  if (p.kind === "naive") return new Date(p.wall!).toISOString();
  return undefined;
}
const blank = (s: string | null | undefined) => (s === "" || s == null ? undefined : s);

function runOurs(xml: string): { status: string; items: FeedItem[] } {
  try {
    return {
      status: "ok",
      items: parseFeed(xml, { fetchedAt: new Date("2026-09-27T00:00:00Z") }).items,
    };
  } catch (e) {
    if (e instanceof FeedError) return { status: `error:${e.kind}`, items: [] };
    throw e;
  }
}

function statusEquivalent(mp: RssSample["mp"]["status"], ours: string): boolean {
  if (mp === "ok") return ours === "ok";
  if (mp === "none_expired") return ours === "error:expired";
  return ours.startsWith("error:") && ours !== "error:expired";
}

describe("RSS 差分(MP RssHelper Python 回退)", () => {
  const c = new FieldCollector();
  for (const s of doc.samples) {
    const ours = runOurs(readInput(s.input.file));
    if (!statusEquivalent(s.mp.status, ours.status)) {
      c.mismatches.push({
        key: `${s.id}.status`,
        mp: s.mp.status,
        ours: ours.status,
      });
    }
    if (s.mp.status !== "ok" || ours.status !== "ok") continue;
    if (s.mp.items.length !== ours.items.length) {
      c.mismatches.push({
        key: `${s.id}.count`,
        mp: s.mp.items.length,
        ours: ours.items.length,
      });
    }
    const n = Math.min(s.mp.items.length, ours.items.length);
    for (let i = 0; i < n; i++) {
      const m = s.mp.items[i]!;
      const o = ours.items[i]!;
      const k = (f: string) => `${s.id}#${i}.${f}`;
      c.check(k("title"), m.title, o.title);
      c.check(k("description"), blank(m.description), blank(o.description));
      c.check(k("link"), blank(m.link), blank(o.link));
      c.check(k("enclosure"), blank(m.enclosure), blank(o.enclosureUrl));
      c.check(k("size"), m.size, o.size);
      c.check(k("pubdate"), mpPubdate(m.pubdate), o.publishedAt);
      c.check(k("nickname"), blank(m.nickname), blank(o.nickname));
    }
  }

  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("rss", 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 rss.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);
    }
  });
});