UHD_Media_Tool/packages/domain/src/scrape/nfo.ts

// SCR-1 刮削:作品 / 季 / 集元数据 → NFO XML(Emby / Jellyfin / Kodi 通用)。纯函数,无 IO。
// 对照 MP app/modules/themoviedb/scraper.py(__gen_common_nfo / __gen_movie_nfo_file / __gen_tv_nfo_file /
// __gen_tv_season_nfo_file / __gen_tv_episode_nfo_file)的元素名、顺序与属性;输出格式对照 minidom
// toprettyxml(indent="  ", encoding="utf-8"):UTF-8 声明、两空格缩进、只有文本的元素写在一行、简介用 CDATA。
// 与 MP 的差异(差分登记见 apps/server/test/scrape/nfo-deviations.json):
// - 缺值不写元素(MP 写空元素 `<premiered/>`、评分缺省写 0、空 CDATA);
// - 简介里的 `]]>` 拆成两段 CDATA(MP 的 minidom 直接抛错);
// - 额外写 豆瓣 id(doubanid + uniqueid type=douban)、标语 tagline、集时长 runtime;
// - 集 NFO 没有 TMDB 集 id(本地不存),不写 uniqueid / tmdbid,媒体服务器按季号集号匹配。

export const TMDB_IMAGE_BASE = "https://image.tmdb.org/t/p/original";

/** 人物:演员(role = 角色名)或导演 */
export interface NfoPerson {
  tmdbId?: number | null;
  name: string;
  role?: string | null;
  /** TMDB profile_path(`/xx.jpg`)或完整 URL */
  profilePath?: string | null;
}

export interface NfoIds {
  tmdb?: string | null;
  imdb?: string | null;
  tvdb?: string | null;
  douban?: string | null;
}

/** 电影 / 剧集(tvshow.nfo) */
export interface NfoWork {
  kind: "movie" | "tvshow";
  title: string;
  originalTitle?: string | null;
  year?: number | null;
  /** 首映 / 首播日期 YYYY-MM-DD */
  premiered?: string | null;
  overview?: string | null;
  tagline?: string | null;
  /** 0–10 */
  rating?: number | null;
  genres?: readonly string[];
  /** 制片国(显示名) */
  countries?: readonly string[];
  /** 出品 / 播出平台 */
  studios?: readonly string[];
  /** 分钟 */
  runtime?: number | null;
  directors?: readonly NfoPerson[];
  actors?: readonly NfoPerson[];
  ids: NfoIds;
}

export interface NfoSeason {
  season: number;
  title?: string | null;
  overview?: string | null;
  airDate?: string | null;
}

export interface NfoEpisode {
  season: number;
  episode: number;
  title?: string | null;
  overview?: string | null;
  airDate?: string | null;
  rating?: number | null;
  runtime?: number | null;
  directors?: readonly NfoPerson[];
  actors?: readonly NfoPerson[];
}

// ── 极简 XML 树(只覆盖 NFO 用到的形态) ──

type Node = { tag: string; attrs?: [string, string][]; text?: string; cdata?: string; children?: Node[] };

const escText = (s: string) => s.replace(/&/g, "&amp;").replace(/</g, "&lt;").replace(/>/g, "&gt;");
const escAttr = (s: string) =>
  escText(s).replace(/"/g, "&quot;").replace(/\r/g, "&#13;").replace(/\n/g, "&#10;").replace(/\t/g, "&#9;");
/** XML 1.0 不允许的控制字符去掉(TMDB 简介偶见) */
// biome-ignore lint/suspicious/noControlCharactersInRegex: 就是要删控制字符
const clean = (s: string) => s.replace(/[\u0000-\u0008\u000B\u000C\u000E-\u001F￾￿]/g, "");
const cdata = (s: string) => `<![CDATA[${s.replace(/]]>/g, "]]]]><![CDATA[>")}]]>`;

function render(n: Node, indent: string, out: string[]) {
  const attrs = (n.attrs ?? []).map(([k, v]) => ` ${k}="${escAttr(clean(v))}"`).join("");
  if (n.children?.length) {
    out.push(`${indent}<${n.tag}${attrs}>`);
    for (const c of n.children) render(c, `${indent}  `, out);
    out.push(`${indent}</${n.tag}>`);
  } else if (n.cdata !== undefined) out.push(`${indent}<${n.tag}${attrs}>${cdata(clean(n.cdata))}</${n.tag}>`);
  else if (n.text !== undefined) out.push(`${indent}<${n.tag}${attrs}>${escText(clean(n.text))}</${n.tag}>`);
  else out.push(`${indent}<${n.tag}${attrs}/>`);
}

function toXml(root: Node): string {
  const out = ['<?xml version="1.0" encoding="utf-8"?>'];
  render(root, "", out);
  return `${out.join("\n")}\n`;
}

const has = (v: string | null | undefined): v is string => typeof v === "string" && v.trim().length > 0;
const num = (v: number | null | undefined): v is number => typeof v === "number" && Number.isFinite(v);

/** TMDB 图片路径 → 完整 URL(MP settings.TMDB_IMAGE_URL,原图) */
export function tmdbImageFullUrl(path: string): string {
  return /^https?:\/\//.test(path) ? path : `${TMDB_IMAGE_BASE}/${path.replace(/^\/+/, "")}`;
}

/** 评分:MP 写 round(x, 1) 的 Python 字符串形式(8.0 → "8.0") */
const ratingText = (v: number) => {
  const r = Math.round(v * 10) / 10;
  return Number.isInteger(r) ? r.toFixed(1) : String(r);
};

class Builder {
  readonly nodes: Node[] = [];
  text(tag: string, v: string | number | null | undefined, attrs?: [string, string][]) {
    if (typeof v === "number") {
      if (num(v)) this.nodes.push({ tag, text: String(v), ...(attrs ? { attrs } : {}) });
    } else if (has(v)) this.nodes.push({ tag, text: v.trim(), ...(attrs ? { attrs } : {}) });
    return this;
  }
  plot(v: string | null | undefined) {
    if (has(v)) {
      this.nodes.push({ tag: "plot", cdata: v.trim() });
      this.nodes.push({ tag: "outline", cdata: v.trim() });
    }
    return this;
  }
  directors(list: readonly NfoPerson[] | undefined) {
    for (const d of list ?? []) {
      if (!has(d.name)) continue;
      this.nodes.push({ tag: "director", text: d.name, attrs: [["tmdbid", num(d.tmdbId) ? String(d.tmdbId) : ""]] });
    }
    return this;
  }
  actors(list: readonly NfoPerson[] | undefined, withRole: boolean) {
    for (const a of list ?? []) {
      if (!has(a.name)) continue;
      const children: Node[] = [
        { tag: "name", text: a.name },
        { tag: "type", text: "Actor" },
      ];
      if (withRole && has(a.role)) children.push({ tag: "role", text: a.role });
      if (num(a.tmdbId)) children.push({ tag: "tmdbid", text: String(a.tmdbId) });
      if (has(a.profilePath)) children.push({ tag: "thumb", text: tmdbImageFullUrl(a.profilePath) });
      if (num(a.tmdbId)) children.push({ tag: "profile", text: `https://www.themoviedb.org/person/${a.tmdbId}` });
      this.nodes.push({ tag: "actor", children });
    }
    return this;
  }
  list(tag: string, values: readonly string[] | undefined) {
    const seen = new Set<string>();
    for (const v of values ?? []) {
      if (!has(v) || seen.has(v.trim())) continue;
      seen.add(v.trim());
      this.nodes.push({ tag, text: v.trim() });
    }
    return this;
  }
}

/**
 * 外部 id 块(MP __gen_common_nfo):tmdbid + uniqueid(tmdb);有 TVDB / IMDb 时追加,IMDb 存在时 IMDb 为 default。
 * 豆瓣是我们加的(uniqueid type=douban,不做 default)。
 */
function idBlock(b: Builder, ids: NfoIds) {
  const tmdb: Node | null = has(ids.tmdb)
    ? {
        tag: "uniqueid",
        text: ids.tmdb,
        attrs: [
          ["type", "tmdb"],
          ["default", "true"],
        ],
      }
    : null;
  if (tmdb) {
    b.text("tmdbid", ids.tmdb);
    b.nodes.push(tmdb);
  }
  if (has(ids.tvdb)) {
    b.text("tvdbid", ids.tvdb);
    b.nodes.push({ tag: "uniqueid", text: ids.tvdb, attrs: [["type", "tvdb"]] });
  }
  if (has(ids.imdb)) {
    b.text("imdbid", ids.imdb);
    b.nodes.push({
      tag: "uniqueid",
      text: ids.imdb,
      attrs: [
        ["type", "imdb"],
        ["default", "true"],
      ],
    });
    if (tmdb)
      tmdb.attrs = [
        ["type", "tmdb"],
        ["default", "false"],
      ];
  }
  if (has(ids.douban)) {
    b.text("doubanid", ids.douban);
    b.nodes.push({ tag: "uniqueid", text: ids.douban, attrs: [["type", "douban"]] });
  }
}

/** movie.nfo(`<视频名>.nfo`)/ tvshow.nfo */
export function workNfo(w: NfoWork): string {
  const b = new Builder();
  idBlock(b, w.ids);
  b.plot(w.overview);
  b.directors(w.directors);
  b.actors(w.actors, true);
  b.list("genre", w.genres);
  if (num(w.rating) && w.rating > 0) b.text("rating", ratingText(w.rating));
  b.list("studio", w.studios);
  b.list("country", w.countries);
  if (num(w.runtime) && w.runtime > 0) b.text("runtime", w.runtime);
  b.text("title", w.title);
  b.text("originaltitle", w.originalTitle);
  b.text("premiered", w.premiered);
  const year = num(w.year)
    ? w.year
    : has(w.premiered) && /^\d{4}/.test(w.premiered)
      ? Number(w.premiered.slice(0, 4))
      : null;
  b.text("year", year);
  if (w.kind === "tvshow") {
    b.text("season", "-1");
    b.text("episode", "-1");
  }
  b.text("tagline", w.tagline);
  return toXml({ tag: w.kind === "movie" ? "movie" : "tvshow", children: b.nodes });
}

/** season.nfo(季目录内) */
export function seasonNfo(s: NfoSeason): string {
  const b = new Builder();
  b.plot(s.overview);
  b.text("title", has(s.title) ? s.title : `季 ${s.season}`);
  b.text("premiered", s.airDate);
  b.text("releasedate", s.airDate);
  b.text("year", has(s.airDate) && /^\d{4}/.test(s.airDate) ? s.airDate.slice(0, 4) : null);
  b.text("seasonnumber", String(s.season));
  return toXml({ tag: "season", children: b.nodes });
}

/** 每集 `<视频名>.nfo` */
export function episodeNfo(e: NfoEpisode): string {
  const b = new Builder();
  b.text("title", has(e.title) ? e.title : `第 ${e.episode} 集`);
  b.plot(e.overview);
  b.text("aired", e.airDate);
  b.text("year", has(e.airDate) && /^\d{4}/.test(e.airDate) ? e.airDate.slice(0, 4) : null);
  b.text("season", String(e.season));
  b.text("episode", String(e.episode));
  if (num(e.rating) && e.rating > 0) b.text("rating", ratingText(e.rating));
  b.directors(e.directors);
  b.actors(e.actors, false);
  if (num(e.runtime) && e.runtime > 0) b.text("runtime", e.runtime);
  return toXml({ tag: "episodedetails", children: b.nodes });
}