/**
 * 通用站点爬虫(配置驱动)。
 *
 * 迁移自 MoviePilot e6820a8 `app/modules/indexer/spider/__init__.py`:
 * - `select_media_categories` / `resolve_category_media_type` → 同名函数
 * - `SiteSpider.__get_search_url` / `__format_search_word` → `SiteSpider.getSearchUrl` / `formatSearchWord`
 * - `SiteSpider.get_info` 及 `__get_title/__get_description/__get_detail/__get_download/__get_grabs/__get_leechers/
 *   __get_seeders/__get_size/__get_imdbid/__get_downloadvolumefactor/__get_uploadvolumefactor/__get_pubdate/__get_date/
 *   __get_date_from_cell/__get_date_elapsed/__get_free_date/__get_labels/__get_hit_and_run/__get_category`
 *   → `SiteSpider.getInfo` 与同名私有方法
 * - `_safe_query / __filter_text / __remove / __attribute_or_text / __index` → 同名方法
 * - `__is_login_or_permission_page` → `isLoginOrPermissionPage`
 * - `parse` → `SiteSpider.parse`(不含 Rust 加速分支);`search_type=subtitles` 分支:`get_subtitle_info /
 *   __get_language_icon / __get_report_url / __get_subtitle_field / __fill_subtitle_ids` → `getSubtitleInfo` 等同名方法
 * - `app/foundation/size.py parse_size` → `parseSizeMp`(Python round 为银行家舍入,同样实现)
 *
 * 与 MP 的差异(均在 README 与迁移账本登记):
 * 1. 额外输出 `category_id`(原始分类 ID),`category` 仍是 MP 的媒体类型投影(REQ §4.5 原始分类不得丢失)。
 * 2. 系数 `selector` 模式用浮点而非 int()(MP 对 0.5 抛异常导致整条丢失,迁移账本 SA4)。
 * 3. 不发请求:HTTP/Cookie/UA 由 adapter 注入,本类只负责 URL 构造与 HTML 解析(便于限流与脱敏统一)。
 * 4. 查询范围:PyQuery 在行元素上查询是 descendant-or-self(行自身也会命中),MP Rust 快路径与 cheerio `.find` 只查后代。
 *    本项目取折中:促销 case / hr / labels 这三类"行级标记"(NexusPHP 把 free_bg、sticky 等 class 挂在 <tr> 上)
 *    包含行自身,与 Python 一致;取值字段(safeQuery)只查后代,避免 `div:nth-child(3)` 这类选择器命中行本身取到整行文本。
 */
import { type CheerioAPI, load } from "cheerio";
import type {
  IndexerDefinition,
  SpiderCategoryConfig,
  SpiderFieldSelector,
  SpiderFields,
  SpiderFilter,
  SpiderListConfig,
} from "./schema.ts";
import { renderTemplate } from "./template.ts";

type Sel = ReturnType<CheerioAPI>;

/** MP `MediaType` 取值 */
export type MpMediaType = "电影" | "电视剧" | "音乐" | "未知";
export type MediaKind = "movie" | "tv" | "music";

const MEDIA_VALUE: Record<MediaKind, MpMediaType> = { movie: "电影", tv: "电视剧", music: "音乐" };

/**
 * 以 htmlparser2(HTML 模式)加载:与 MP 的 lxml.html 一致不自动插入 <tbody>,
 * 使 `table.torrents > tr` 这类 MP 风格选择器保持原语义。
 */
export function loadHtml(html: string): CheerioAPI {
  return load(html, { xml: { xmlMode: false, decodeEntities: true } });
}

/**
 * PyQuery(cssselect)接受 `:has("sel")` 引号写法,cheerio-select 不接受;统一去引号以兼容 MP 风格定义。
 */
export function normalizeSelector(sel: string): string {
  return sel.replace(/:has\((["'])(.*?)\1\)/g, ":has($2)");
}

/** 解析结果:键名与 MP `TorrentInfo` 字段一致,`category_id` 为本项目扩展。 */
export interface SpiderTorrent {
  title?: string;
  description?: string;
  page_url?: string;
  enclosure?: string;
  grabs?: number;
  peers?: number;
  seeders?: number;
  size?: number;
  imdbid?: string;
  downloadvolumefactor?: number;
  uploadvolumefactor?: number;
  pubdate?: string;
  date_elapsed?: string;
  freedate?: string;
  labels?: string[];
  hit_and_run?: boolean;
  category?: MpMediaType;
  category_id?: string;
  // ---- 字幕分支(search_type=subtitles)专有字段,键名与 MP 一致 ----
  language_icon?: string;
  report_url?: string;
  language?: string;
  uploader?: string;
  torrent_id?: string;
  subtitle_id?: string;
  file_name?: string;
}

export type SpiderSearchType = "torrents" | "subtitles";

/** MP `select_media_categories` */
export function selectMediaCategories(category: IndexerDefinition["category"], mtype?: MediaKind) {
  if (!category) return [];
  if (mtype === "tv") return category.tv ?? [];
  if (mtype === "movie") return category.movie ?? [];
  if (mtype === "music") return category.music ?? [];
  return [...(category.movie ?? []), ...(category.tv ?? []), ...(category.music ?? [])];
}

/** MP `resolve_category_media_type`:同一 ID 只落在一个媒体类型里才投影,否则未知。 */
export function resolveCategoryMediaType(value: unknown, category: IndexerDefinition["category"]): MpMediaType {
  if (value == null || !category) return "未知";
  const id = String(value);
  const matches = (["movie", "tv", "music"] as const).filter((k) =>
    (category[k] ?? []).some((item) => item.id != null && String(item.id) === id),
  );
  return matches.length === 1 ? MEDIA_VALUE[matches[0]!] : "未知";
}

/** Python 3 `round()`:恰好 .5 时取偶数(银行家舍入) */
function roundHalfEven(v: number): number {
  const f = Math.floor(v);
  const diff = v - f;
  if (diff === 0.5) return f % 2 === 0 ? f : f + 1;
  return Math.round(v);
}

/** MP `size_tools.parse_size`:二进制单位,round(银行家舍入)。 */
export function parseSizeMp(text: string | number | null | undefined): number {
  if (!text) return 0;
  const s = String(text);
  if (/^\d+$/.test(s)) return Number(s);
  const normalized = s.replace(/,/g, "").replace(/ /g, "").toUpperCase();
  const sizeText = normalized.replace(/[KMGTPI]*B?/gi, "");
  const size = Number(sizeText);
  if (sizeText === "" || !Number.isFinite(size)) return 0;
  let v = size;
  if (normalized.includes("PB") || normalized.includes("PIB")) v *= 1024 ** 5;
  else if (normalized.includes("TB") || normalized.includes("TIB")) v *= 1024 ** 4;
  else if (normalized.includes("GB") || normalized.includes("GIB")) v *= 1024 ** 3;
  else if (normalized.includes("MB") || normalized.includes("MIB")) v *= 1024 ** 2;
  else if (normalized.includes("KB") || normalized.includes("KIB")) v *= 1024;
  return roundHalfEven(v);
}

/** strptime 结果(MP dateparse 过滤器返回 datetime) */
class ParsedDateTime {
  constructor(readonly text: string) {}
}

const STRPTIME_TOKENS: Record<string, string> = {
  "%Y": "(\\d{4})",
  "%m": "(\\d{1,2})",
  "%d": "(\\d{1,2})",
  "%H": "(\\d{1,2})",
  "%M": "(\\d{1,2})",
  "%S": "(\\d{1,2})",
};

/** Python strptime 的最小子集(%Y %m %d %H %M %S),失败抛错(与 MP 一致由调用方吞掉)。 */
function strptime(text: string, fmt: string): ParsedDateTime {
  const order: string[] = [];
  const re = fmt.replace(/[.*+?^${}()|[\]\\]/g, "\\$&").replace(/%[YmdHMS]/g, (t) => {
    order.push(t);
    return STRPTIME_TOKENS[t]!;
  });
  const m = new RegExp(`^${re}$`).exec(text);
  if (!m) throw new Error(`time data '${text}' does not match format '${fmt}'`);
  const v: Record<string, number> = { "%Y": 1900, "%m": 1, "%d": 1, "%H": 0, "%M": 0, "%S": 0 };
  order.forEach((t, i) => {
    v[t] = Number(m[i + 1]);
  });
  const p = (n: number) => String(n).padStart(2, "0");
  return new ParsedDateTime(`${v["%Y"]}-${p(v["%m"]!)}-${p(v["%d"]!)} ${p(v["%H"]!)}:${p(v["%M"]!)}:${p(v["%S"]!)}`);
}

const STD_TIME = /^\d{4}-\d{2}-\d{2} \d{2}:\d{2}:\d{2}$/;

/** MP `time_tools.normalize_datetime` 的保守子集:只接受可确定解析的绝对时间。 */
function normalizeDatetime(value: string): string {
  const m = /^(\d{4})[-/年.](\d{1,2})[-/月.](\d{1,2})日?(?:[ T](\d{1,2}):(\d{1,2})(?::(\d{1,2}))?)?$/.exec(
    value.trim(),
  );
  if (!m) return value;
  const p = (n: string | undefined) => String(Number(n ?? 0)).padStart(2, "0");
  return `${m[1]}-${p(m[2])}-${p(m[3])} ${p(m[4])}:${p(m[5])}:${p(m[6])}`;
}

function isRelativePubdateText(pubdate: string | null | undefined): boolean {
  if (!pubdate) return false;
  const text = String(pubdate).trim().toLowerCase();
  if (/\d{4}[-/年]\d{1,2}/.test(text)) return false;
  if (text.includes("ago")) return true;
  return /\d+\s*(秒|分钟|分|小时|天|周|月|年)/.test(text);
}

function isInvalidPubdateText(pubdate: string | null | undefined): boolean {
  if (!pubdate) return true;
  const text = String(pubdate).trim();
  if (text.toLowerCase() === "now" || text === "0") return true;
  if (isRelativePubdateText(text)) return true;
  return !STD_TIME.test(text);
}

function pyIndex<T>(arr: T[], i: number): T | undefined {
  return i < 0 ? arr[arr.length + i] : arr[i];
}

/** Python `parse_qs(urlparse(url).query).get(key)[0]` */
function querystringValue(text: string, key: string): string {
  const q = text.includes("?") ? text.slice(text.indexOf("?") + 1) : "";
  const v = new URLSearchParams(q.split("#")[0]).getAll(key).filter((x) => x !== "");
  return v[0] ?? "";
}

/** PyQuery `.text()` 近似:`<br>` 视为换行,行内空白折叠。 */
function pqText(el: Sel): string {
  const c = el.clone();
  c.find("br").replaceWith("\n");
  return c
    .text()
    .split("\n")
    .map((l) => l.replace(/[ \t\r\f\v\u00a0]+/g, " ").trim())
    .filter((l) => l !== "")
    .join("\n");
}

export interface SiteSpiderQuery {
  keyword?: string | string[];
  mtype?: MediaKind;
  /** 逗号分隔的分类 ID 白名单(MP `cat`) */
  cat?: string;
  /** 站点内部页码(MP 0 起) */
  page?: number;
  /** MP `search_type`:subtitles 时改用 `def.subtitles` 的配置 */
  searchType?: SpiderSearchType;
}

export class SiteSpider {
  readonly def: IndexerDefinition;
  readonly keyword: string | string[] | undefined;
  readonly mtype: MediaKind | undefined;
  readonly cat: string | undefined;
  readonly page: number;
  readonly domain: string;
  readonly resultNum: number;
  readonly siteMediaType: MediaKind | undefined;
  readonly searchType: SpiderSearchType;
  private list: SpiderListConfig;
  private fields: SpiderFields;
  private searchCfg: IndexerDefinition["search"];
  private batchCfg: IndexerDefinition["batch"];
  private browseCfg: IndexerDefinition["browse"];
  private categoryCfg: SpiderCategoryConfig | undefined;
  private requestedResultMediaType: MediaKind | undefined;
  private $!: CheerioAPI;
  private info: SpiderTorrent = {};

  constructor(def: IndexerDefinition, q: SiteSpiderQuery = {}) {
    this.def = def;
    this.keyword = q.keyword;
    this.mtype = q.mtype;
    this.cat = q.cat;
    this.page = q.page ?? 0;
    this.siteMediaType = fromAgent(def.media_type);
    this.searchType = q.searchType ?? "torrents";
    if (this.searchType === "subtitles") {
      const sc = def.subtitles ?? {};
      this.searchCfg = sc.search;
      this.batchCfg = sc.batch;
      this.browseCfg = sc.browse;
      this.categoryCfg = sc.category;
      this.list = sc.list ?? { selector: "" };
      this.fields = sc.fields ?? {};
      this.resultNum = Number(sc.result_num ?? def.result_num ?? 100);
    } else {
      this.searchCfg = def.search;
      this.batchCfg = def.batch;
      this.browseCfg = def.browse;
      this.categoryCfg = def.category;
      this.list = def.torrents.list;
      this.fields = def.torrents.fields ?? {};
      if (!q.keyword && def.browse) {
        this.list = def.browse.list ?? this.list;
        this.fields = def.browse.fields ?? this.fields;
      }
      this.resultNum = Number(def.result_num ?? 100);
    }
    this.domain = def.domain.endsWith("/") ? def.domain : `${def.domain}/`;
  }

  /** MP `__can_search` */
  canSearch(): boolean {
    if (this.siteMediaType && this.mtype && this.siteMediaType !== this.mtype) return false;
    return Boolean(this.searchCfg && this.domain);
  }

  /** MP `__format_search_word` */
  private formatSearchWord(word: string): string {
    if (word && /^tt\d+$/.test(word) && this.searchCfg?.imdbid_format) {
      return this.searchCfg.imdbid_format
        .replace(/\{keyword\}/g, word)
        .replace(/\{imdbid\}/g, word)
        .replace(/\{imdbid_num\}/g, word.slice(2));
    }
    return word;
  }

  /** MP `__get_search_url` */
  getSearchUrl(): string {
    const search = this.searchCfg ?? { paths: [] };
    const paths = search.paths ?? [];
    let torrentspath = "";
    let typedPathSelected = false;
    let categoryFilterSelected = false;
    if (paths.length === 1) {
      torrentspath = paths[0]!.path ?? "";
    } else {
      let fallback = "";
      const expected = this.mtype;
      for (const p of paths) {
        if (p.type === "all" && !fallback) fallback = p.path ?? "";
        if ((expected && p.type === expected) || (!expected && p.type === "all")) {
          torrentspath = p.path ?? "";
          typedPathSelected = Boolean(expected && p.type === expected);
          break;
        }
      }
      if (!torrentspath) torrentspath = fallback;
    }

    let searchurl: string;
    if (this.keyword && (typeof this.keyword === "string" || this.keyword.length > 0)) {
      let searchWord: string;
      let searchMode: string;
      if (Array.isArray(this.keyword)) {
        if (this.batchCfg) {
          const delimiter = this.batchCfg.delimiter || " ";
          const spaceReplace = this.batchCfg.space_replace || " ";
          searchWord = this.keyword.map((k) => String(k).replace(/ /g, spaceReplace)).join(delimiter);
        } else {
          searchWord = this.keyword.join(" ");
        }
        searchMode = "1";
      } else {
        searchWord = this.keyword;
        searchMode = "0";
      }
      const isImdb = typeof this.keyword === "string" && /^tt\d+$/.test(this.keyword);
      searchWord = this.formatSearchWord(searchWord);
      const indexerParams: Record<string, string | number> = { ...(search.params ?? {}) };
      if (Object.keys(indexerParams).length > 0) {
        if (indexerParams.search_area && !isImdb) delete indexerParams.search_area;
        const params: Record<string, string | number | Array<string | number>> = {
          search_mode: searchMode,
          search_area: 0,
          page: this.page || 0,
          notnewword: 1,
        };
        for (const [k, v] of Object.entries(indexerParams)) params[k] = String(v).replace(/\{keyword\}/g, searchWord);
        const category = this.categoryCfg;
        if (category) {
          const cats = selectMediaCategories(category, this.mtype);
          const allowed = this.cat ? new Set(this.cat.split(",")) : null;
          for (const c of cats) {
            if (allowed && !allowed.has(String(c.id))) continue;
            if (category.field) {
              categoryFilterSelected = true;
              const cur = String(params[category.field] ?? "");
              params[category.field] = cur + (category.delimiter ?? " ") + String(c.id);
            } else {
              const categoryParam = c.param || category.param;
              if (categoryParam) {
                categoryFilterSelected = true;
                const cid = c.value ?? c.id;
                const cur = params[categoryParam];
                if (cur === undefined) params[categoryParam] = cid;
                else if (Array.isArray(cur)) cur.push(cid);
                else params[categoryParam] = [cur, cid];
              } else {
                categoryFilterSelected = true;
                params[`cat${c.id}`] = 1;
              }
            }
            if (c.params) Object.assign(params, c.params);
          }
        }
        searchurl = combineUrl(this.domain, torrentspath, params);
      } else {
        searchurl =
          this.domain +
          torrentspath
            .replace(/\{keyword\}/g, encodeURIComponent(searchWord))
            .replace(/\{page\}/g, String(this.page || 0));
      }
    } else {
      let page = this.page || 0;
      if (this.browseCfg && !typedPathSelected) {
        torrentspath = this.browseCfg.path;
        if (this.browseCfg.start) page = Number(this.browseCfg.start) + Number(this.page || 0);
      } else if (this.page && !torrentspath.includes("{page}")) {
        torrentspath += `${torrentspath.includes("?") ? "&" : "?"}page=${this.page}`;
      }
      searchurl = this.domain + torrentspath.replace(/\{page\}/g, String(page)).replace(/\{keyword\}/g, "");
    }

    this.requestedResultMediaType =
      search.result_media_type === "requested" && this.mtype && (typedPathSelected || categoryFilterSelected)
        ? this.mtype
        : undefined;
    return searchurl;
  }

  /** MP `__is_login_or_permission_page` */
  static isLoginOrPermissionPage($: CheerioAPI): boolean {
    const title = $("title").first().text().trim();
    const pageText = $.root().text().split(/\s+/).filter(Boolean).join(" ").slice(0, 1000);
    if (title === "登录" || title.includes(":: 登录")) return true;
    return ["未登录", "登录 / 注册", "必须在登录后才能访问", "你需要启用cookies才能登录"].some((m) =>
      pageText.includes(m),
    );
  }

  /** MP `parse`:登录/权限页返回 isError=true 与空列表。 */
  parse(html: string): { isError: boolean; loginPage: boolean; torrents: SpiderTorrent[]; $: CheerioAPI | null } {
    if (!html) return { isError: true, loginPage: false, torrents: [], $: null };
    const $ = loadHtml(html);
    if (SiteSpider.isLoginOrPermissionPage($)) return { isError: true, loginPage: true, torrents: [], $ };
    this.$ = $;
    const out: SpiderTorrent[] = [];
    const rows = $(normalizeSelector(this.list.selector ?? ""));
    rows.each((i, el) => {
      if (i >= this.resultNum) return false;
      const info = this.searchType === "subtitles" ? this.getSubtitleInfo($(el)) : this.getInfo($(el));
      if (Object.keys(info).length > 0) out.push(this.applyRequestedMediaType(info));
      return undefined;
    });
    return { isError: false, loginPage: false, torrents: out, $ };
  }

  /** 对任意片段(如详情页头部)按字段配置取值 */
  infoFrom($: CheerioAPI, el: Sel): SpiderTorrent {
    this.$ = $;
    return this.getInfo(el);
  }

  private applyRequestedMediaType(t: SpiderTorrent): SpiderTorrent {
    if (this.requestedResultMediaType) t.category = MEDIA_VALUE[this.requestedResultMediaType];
    return t;
  }

  /** MP `get_info`:任一字段异常则整条返回 {}。 */
  getInfo(row: Sel): SpiderTorrent {
    this.info = {};
    try {
      this.getTitle(row);
      this.getDescription(row);
      this.getDetail(row);
      this.getDownload(row);
      this.getGrabs(row);
      this.getLeechers(row);
      this.getSeeders(row);
      this.getSize(row);
      this.getImdbid(row);
      this.getVolumeFactor(row, "downloadvolumefactor");
      this.getVolumeFactor(row, "uploadvolumefactor");
      this.getPubdate(row);
      this.getDateElapsed(row);
      this.getFreeDate(row);
      this.getLabels(row);
      this.getHitAndRun(row);
      this.getCategory(row);
      return { ...this.info };
    } catch {
      return {};
    }
  }

  /** MP `get_subtitle_info`:缺标题或下载链接的行丢弃;任一字段异常整条返回 {}。 */
  getSubtitleInfo(row: Sel): SpiderTorrent {
    this.info = {};
    try {
      this.getTitle(row);
      this.getDescription(row);
      this.getDetail(row);
      this.getDownload(row);
      this.getSize(row);
      this.getPubdate(row);
      this.getDateElapsed(row);
      this.getGrabs(row);
      this.getLinkField(row, "language_icon", "language_icon");
      this.getLinkField(row, "report", "report_url");
      for (const name of ["language", "uploader", "torrent_id", "subtitle_id", "file_name"] as const) {
        this.getSubtitleField(row, name);
      }
      this.fillSubtitleIds();
      if (!this.info.title || !this.info.enclosure) return {};
      return { ...this.info };
    } catch {
      return {};
    }
  }

  /** MP `__get_language_icon` / `__get_report_url`:相对链接补全为绝对链接 */
  private getLinkField(row: Sel, name: string, key: "language_icon" | "report_url") {
    if (!(name in this.fields)) return;
    const selector = this.fields[name] ?? {};
    const link = this.filterText(this.safeQuery(row, selector), selector.filters);
    if (link) this.info[key] = this.normalizeLink(String(link));
  }

  /** MP `__get_subtitle_field` */
  private getSubtitleField(row: Sel, name: "language" | "uploader" | "torrent_id" | "subtitle_id" | "file_name") {
    const selector = this.fields[name];
    if (!selector || Object.keys(selector).length === 0) return;
    const v = this.filterText(this.safeQuery(row, selector), selector.filters);
    if (v != null) this.info[name] = v instanceof ParsedDateTime ? v.text : String(v);
  }

  /** MP `__fill_subtitle_ids`:从下载链接 query 补种子 ID(torrentid/torrent_id)与字幕 ID(subid/subtitle) */
  private fillSubtitleIds() {
    const enclosure = this.info.enclosure;
    if (!enclosure) return;
    const q = enclosure.includes("?") ? enclosure.slice(enclosure.indexOf("?") + 1).split("#")[0]! : "";
    const params = new URLSearchParams(q);
    const first = (...keys: string[]) => {
      for (const k of keys) {
        const v = params.getAll(k).find((x) => x !== "");
        if (v !== undefined) return v;
      }
      return undefined;
    };
    if (!this.info.torrent_id) {
      const v = first("torrentid", "torrent_id");
      if (v !== undefined) this.info.torrent_id = v;
    }
    if (!this.info.subtitle_id) {
      const v = first("subid", "subtitle");
      if (v !== undefined) this.info.subtitle_id = v;
    }
  }

  // ---- 字段 ----

  private getTitle(row: Sel) {
    if (!("title" in this.fields)) return;
    const selector = this.fields.title ?? {};
    let title: string | null | undefined;
    if (selector.selector) {
      title = this.safeQuery(row, selector);
    } else if (selector.text) {
      const render: Record<string, unknown> = {};
      if ("title_default" in this.fields) render.title_default = this.safeQuery(row, this.fields.title_default);
      if ("title_optional" in this.fields) render.title_optional = this.safeQuery(row, this.fields.title_optional);
      title = renderTemplate(selector.text, render);
    }
    const v = this.filterText(title, selector.filters);
    if (v != null) this.info.title = String(v);
  }

  private getDescription(row: Sel) {
    if (!("description" in this.fields)) return;
    const selector = this.fields.description ?? {};
    let desc: string | null | undefined;
    if (selector.selector || selector.selectors) {
      const s: SpiderFieldSelector = { ...selector };
      if (selector.selectors && !selector.selector) s.selector = selector.selectors;
      desc = this.safeQuery(row, s);
    } else if (selector.text) {
      const render: Record<string, unknown> = {};
      for (const k of ["tags", "subject", "description_free_forever", "description_normal"]) {
        if (k in this.fields) render[k] = this.safeQuery(row, this.fields[k]);
      }
      desc = renderTemplate(selector.text, render);
    }
    const v = this.filterText(desc, selector.filters);
    if (v != null) this.info.description = String(v);
  }

  private normalizeLink(link: string): string {
    if (link.startsWith("http")) return link;
    if (link.startsWith("//")) return `${this.domain.split(":")[0]}:${link}`;
    if (link.startsWith("/")) return this.domain + link.slice(1);
    return this.domain + link;
  }

  private getDetail(row: Sel) {
    if (!("details" in this.fields)) return;
    const selector = this.fields.details ?? {};
    const link = this.filterText(this.safeQuery(row, selector), selector.filters);
    if (link) this.info.page_url = this.normalizeLink(String(link));
  }

  private getDownload(row: Sel) {
    if (!("download" in this.fields)) return;
    const selector = this.fields.download ?? {};
    const raw = this.filterText(this.safeQuery(row, selector), selector.filters);
    if (!raw) return;
    const link = String(raw);
    if (!link.startsWith("http") && !link.startsWith("magnet")) {
      const [scheme, host] = splitNetloc(this.domain);
      if (host && link.includes(host)) {
        this.info.enclosure = link.startsWith("/") ? `${scheme}:${link}` : `${scheme}://${link}`;
      } else {
        this.info.enclosure = link.startsWith("/") ? `${this.domain}${link.slice(1)}` : `${this.domain}${link}`;
      }
    } else {
      this.info.enclosure = link;
    }
  }

  private countField(row: Sel, name: string, key: "grabs" | "peers" | "seeders", allowZeroText: boolean) {
    if (!(name in this.fields)) return;
    const selector = this.fields[name] ?? {};
    const item = this.safeQuery(row, selector);
    const present = allowZeroText ? item != null && item !== "" : Boolean(item);
    if (present) {
      let v: unknown = String(item).split("/")[0]!.replace(/,/g, "");
      v = this.filterText(v as string, selector.filters);
      this.info[key] = typeof v === "string" && /^\d+$/.test(v) ? Number(v) : 0;
    } else {
      this.info[key] = 0;
    }
  }

  private getGrabs(row: Sel) {
    this.countField(row, "grabs", "grabs", true);
  }
  private getLeechers(row: Sel) {
    this.countField(row, "leechers", "peers", false);
  }
  private getSeeders(row: Sel) {
    this.countField(row, "seeders", "seeders", false);
  }

  private getSize(row: Sel) {
    if (!("size" in this.fields)) return;
    const selector = this.fields.size ?? {};
    const item = this.safeQuery(row, selector);
    if (item != null && item !== "") {
      const v = this.filterText(item.replace(/\n/g, "").trim(), selector.filters);
      this.info.size = parseSizeMp(v == null ? "" : String(v));
    } else {
      this.info.size = 0;
    }
  }

  private getImdbid(row: Sel) {
    if (!("imdbid" in this.fields)) return;
    const selector = this.fields.imdbid ?? {};
    const v = this.filterText(this.safeQuery(row, selector), selector.filters);
    if (v != null) this.info.imdbid = String(v);
  }

  private getVolumeFactor(row: Sel, key: "downloadvolumefactor" | "uploadvolumefactor") {
    const selector = this.fields[key];
    if (!selector || Object.keys(selector).length === 0) return;
    this.info[key] = 1;
    if (selector.case) {
      for (const [css, value] of Object.entries(selector.case)) {
        if (this.rowScope(row, css).length > 0) {
          this.info[key] = value;
          break;
        }
      }
    } else if (selector.selector) {
      const item = this.safeQuery(row, selector);
      const m = item ? /(\d+\.?\d*)/.exec(item) : null;
      // 偏离 MP:MP 用 int() 转换,"0.5" 抛异常导致整条丢失(迁移账本 SA4)
      if (m) this.info[key] = Number.parseFloat(m[1]!);
    }
  }

  private getPubdate(row: Sel) {
    if (!("date_added" in this.fields) && !("date" in this.fields)) return;
    let selector: SpiderFieldSelector = this.fields.date_added ?? {};
    let pubdate: string | null | undefined = this.safeQuery(row, selector);
    if (!pubdate) {
      selector = this.fields.date ?? {};
      pubdate = this.getDate(row, selector);
    }
    if (pubdate) pubdate = pubdate.replace(/\n/g, " ").trim();
    const filtered = this.filterText(pubdate, selector.filters);
    let value: string | undefined;
    if (filtered instanceof ParsedDateTime) value = filtered.text;
    else if (filtered) value = STD_TIME.test(String(filtered)) ? String(filtered) : normalizeDatetime(String(filtered));
    if (value && !isInvalidPubdateText(value)) this.info.pubdate = value;
    else delete this.info.pubdate;
  }

  private getDate(row: Sel, selector: SpiderFieldSelector): string | null {
    if (!selector || Object.keys(selector).length === 0) return null;
    if (selector.selector) return this.safeQuery(row, selector) ?? null;
    if (!selector.text) return null;
    const render: Record<string, string | null> = {};
    for (const name of ["date_elapsed", "date_added"]) {
      const fs = this.fields[name] ?? {};
      let v = this.safeQuery(row, fs) ?? null;
      if (!v) v = this.getDateFromCell(row, fs);
      render[name] = v;
    }
    if (!Object.values(render).some(Boolean)) return null;
    const s = renderTemplate(selector.text, render);
    if (s === "now" || isRelativePubdateText(s)) return null;
    return s;
  }

  private getDateFromCell(row: Sel, selector: SpiderFieldSelector): string | null {
    const sel = selector.selector?.trim();
    if (!sel?.includes("> span")) return null;
    return this.safeQuery(row, { selector: sel.split("> span")[0]!.trim() }) ?? null;
  }

  private getDateElapsed(row: Sel) {
    if (!("date_elapsed" in this.fields)) return;
    const selector = this.fields.date_elapsed ?? {};
    const v = this.filterText(this.safeQuery(row, selector), selector.filters);
    if (v != null) this.info.date_elapsed = String(v);
  }

  private getFreeDate(row: Sel) {
    if (!("freedate" in this.fields)) return;
    const selector = this.fields.freedate ?? {};
    const v = this.filterText(this.safeQuery(row, selector), selector.filters);
    if (v != null) this.info.freedate = v instanceof ParsedDateTime ? v.text : String(v);
  }

  private getLabels(row: Sel) {
    if (!("labels" in this.fields)) return;
    const selector = this.fields.labels ?? {};
    if (!selector.selector) {
      this.info.labels = [];
      return;
    }
    const q = this.rowScope(row, selector.selector).clone();
    this.remove(q, selector);
    this.info.labels = this.attributeOrText(q, selector).filter((x): x is string => Boolean(x));
  }

  private getHitAndRun(row: Sel) {
    if (!("hr" in this.fields)) return;
    const selector = this.fields.hr ?? {};
    this.info.hit_and_run = selector.selector ? this.rowScope(row, selector.selector).length > 0 : false;
  }

  private getCategory(row: Sel) {
    if (this.requestedResultMediaType) {
      this.info.category = MEDIA_VALUE[this.requestedResultMediaType];
      return;
    }
    if (!("category" in this.fields)) {
      if (this.siteMediaType) this.info.category = MEDIA_VALUE[this.siteMediaType];
      return;
    }
    const selector = this.fields.category ?? {};
    const value = this.filterText(this.safeQuery(row, selector), selector.filters);
    if (value != null && value !== "") this.info.category_id = String(value);
    let resolved = resolveCategoryMediaType(value, this.categoryCfg);
    if (resolved === "未知" && this.siteMediaType) resolved = MEDIA_VALUE[this.siteMediaType];
    this.info.category = resolved;
  }

  // ---- 通用查询 ----

  /** 行级标记查询:行自身 + 后代(PyQuery descendant-or-self,见文件头差异 4);`:scope` 选择器只查后代 */
  private rowScope(row: Sel, selector: string): Sel {
    const css = normalizeSelector(selector);
    const desc = row.find(css);
    if (css.includes(":scope")) return desc;
    return row.filter(css).add(desc);
  }

  /** MP `_safe_query` */
  safeQuery(row: Sel, config: SpiderFieldSelector | undefined): string | null | undefined {
    if (!config?.selector) return null;
    let q = row.find(normalizeSelector(config.selector));
    if (config.remove) q = q.clone();
    this.remove(q, config);
    return this.index(this.attributeOrText(q, config), config);
  }

  private remove(q: Sel, config: SpiderFieldSelector) {
    if (!config.remove) return;
    for (const v of config.remove.split(", ")) q.find(normalizeSelector(v)).remove();
  }

  private attributeOrText(q: Sel, config: SpiderFieldSelector): Array<string | undefined> {
    if (q.length === 0) return [];
    const out: Array<string | undefined> = [];
    q.each((_, el) => {
      const item = this.$(el);
      out.push(config.attribute ? item.attr(config.attribute) : pqText(item));
    });
    return out;
  }

  private index(items: Array<string | undefined>, config: SpiderFieldSelector): string | undefined {
    if (items.length === 0) return undefined;
    if (config.contents !== undefined && items.length > Number(config.contents)) {
      return pyIndex((items[0] ?? "").split("\n"), Number(config.contents));
    }
    if (config.index !== undefined && items.length > Number(config.index)) return pyIndex(items, Number(config.index));
    return items[0];
  }

  /** MP `__filter_text`;单个过滤器失败只跳过该过滤器。 */
  filterText(
    text: string | null | undefined,
    filters: SpiderFilter[] | undefined,
  ): string | ParsedDateTime | null | undefined {
    if (!text || !filters || !Array.isArray(filters)) return text;
    let t: string | ParsedDateTime = String(text);
    for (const f of filters) {
      if (!t) break;
      try {
        if (t instanceof ParsedDateTime) break;
        const s: string = t;
        switch (f.name) {
          case "re_search": {
            if (!Array.isArray(f.args)) break;
            const m = new RegExp(f.args[0]).exec(s);
            if (m) t = m[Number(f.args[f.args.length - 1])] ?? "";
            break;
          }
          case "split":
            if (Array.isArray(f.args)) {
              const parts = s.split(f.args[0]);
              const v = pyIndex(parts, Number(f.args[f.args.length - 1]));
              if (v === undefined) throw new Error("list index out of range");
              t = v;
            }
            break;
          case "replace":
            if (Array.isArray(f.args)) t = s.split(String(f.args[0])).join(String(f.args[f.args.length - 1]));
            break;
          case "dateparse":
            if (typeof f.args === "string") t = strptime(s.replace(/\n/g, " ").trim(), f.args);
            break;
          case "strip":
            t = s.trim();
            break;
          case "appendleft":
            t = `${f.args}${s}`;
            break;
          case "querystring":
            t = querystringValue(s, String(f.args));
            break;
        }
      } catch {
        // MP 仅 debug 日志,继续下一个过滤器
      }
    }
    return typeof t === "string" ? t.trim() : t;
  }
}

/** MP `MediaType.from_agent` 的子集 */
function fromAgent(v: string | undefined): MediaKind | undefined {
  if (!v) return undefined;
  const s = v.toLowerCase();
  if (s === "movie" || v === "电影") return "movie";
  if (s === "tv" || v === "电视剧") return "tv";
  if (s === "music" || v === "音乐") return "music";
  return undefined;
}

/** MP `url.split_netloc` */
function splitNetloc(url: string): [string, string] {
  if (!url) return ["", ""];
  if (!url.startsWith("http")) return ["http", url];
  const u = new URL(url);
  return [u.protocol.replace(":", ""), u.host];
}

/** MP `UrlUtils.combine_url`:urljoin + 合并已有 query + urlencode(doseq=True)。 */
export function combineUrl(
  host: string,
  path: string,
  query: Record<string, string | number | Array<string | number>>,
): string {
  const base = host.endsWith("/") ? host : `${host}/`;
  const url = new URL(path || "/", base);
  for (const [k, v] of Object.entries(query)) {
    url.searchParams.delete(k);
    if (Array.isArray(v)) for (const x of v) url.searchParams.append(k, String(x));
    else url.searchParams.set(k, String(v));
  }
  return url.toString();
}