UHD_Media_Tool/packages/adapters/src/sites/spider/detail-page.ts

/**
 * 详情页扩展内容(简介 / MediaInfo / 截图 / IMDb·豆瓣·TMDB / 文件清单 / 标签 / 上传者)—— 本项目扩展。
 * 以 MovieClaw detail_*_css(description / imdb / douban / file_list)为起点,其余按 NexusPHP details.php 公开结构
 * 做通用规则:字段标签(td.rowhead 或站点配置的 div 标签)的下一个兄弟元素是值。站点特化写在定义 `umt.detail.page`。
 * 文件清单只取页面上已有的内容,不为它额外请求 viewfilelist.php(请求克制)。
 */
import type { CheerioAPI } from "cheerio";
import {
  emptyDetailPage,
  extractTmdb,
  looksLikeMediaInfo,
  MAX_FILES,
  MAX_SCREENSHOTS,
  screenshotUrl,
  tidyText,
} from "../detail-text.ts";
import type { SiteDetailFile, SiteDetailPage } from "../types.ts";
import { extractDoubanId, extractImdbId } from "../util.ts";
import { mcSelect } from "./mc-selector.ts";
import type { IndexerDefinition } from "./schema.ts";
import { parseSizeMp } from "./spider.ts";

type DetailConf = NonNullable<IndexerDefinition["umt"]["detail"]>;
type El = ReturnType<CheerioAPI>;

const LABELS = {
  description: ["简介", "Description"],
  mediaInfo: ["MediaInfo", "媒体信息", "BDInfo"],
  tags: ["标签", "Tags"],
  uploader: ["上传者", "发布者", "发布人", "上传人", "Uploader"],
  files: ["文件列表", "文件清单", "File list", "Files"],
};

/** 简介里可能装 MediaInfo 的块 */
const MI_BLOCKS = "fieldset, div.codemain, pre, code, blockquote, .spoiler-content, .hide, div.nexus-media-info-raw";

const normLabel = (t: string) =>
  t
    .replace(/\s+/g, "")
    .replace(/[:：]+$/, "")
    .toLowerCase();

/** 保留换行的文本:<br> 与块级元素换行 */
function blockText($: CheerioAPI, el: El): string {
  const c = el.clone();
  c.find("script, style").remove();
  c.find("br").replaceWith("\n");
  c.find("div, p, li, tr, pre, fieldset, blockquote, h1, h2, h3, h4, h5, h6").each((_, n) => {
    $(n).append("\n");
  });
  return tidyText(c.text());
}

const SIZE_TAIL = /^(.*?)\s*[(（[]\s*([\d.,]+\s*[KMGTP]?i?B)\s*[)）\]]\s*$/i;
const HAS_EXT = /\.[A-Za-z0-9]{2,5}$/;

function fileFromText(text: string): SiteDetailFile | null {
  const t = text.replace(/\s+/g, " ").trim();
  if (!t) return null;
  const m = SIZE_TAIL.exec(t);
  if (m?.[1]) return { name: m[1], sizeBytes: parseSizeMp(m[2]!) || null };
  return HAS_EXT.test(t) ? { name: t, sizeBytes: null } : null;
}

function filesFrom($: CheerioAPI, rows: El): SiteDetailFile[] {
  const out: SiteDetailFile[] = [];
  rows.each((_, n) => {
    if (out.length >= MAX_FILES) return;
    const row = $(n);
    const cells = row.children("td");
    if (cells.length >= 2) {
      const name = cells.first().text().replace(/\s+/g, " ").trim();
      const size = cells.eq(1).text().trim();
      if (name && HAS_EXT.test(name) && /\d/.test(size)) out.push({ name, sizeBytes: parseSizeMp(size) || null });
      return;
    }
    const f = fileFromText(row.text());
    if (f) out.push(f);
  });
  return out;
}

export function extractDetailPage($: CheerioAPI, conf: DetailConf, base: string): SiteDetailPage {
  const page = emptyDetailPage();
  const p = conf.page ?? {};

  // 字段标签 → 值元素
  const pairs: { label: string; value: El }[] = [];
  const addLabels = (sel: string, valueFor: (e: El) => El) => {
    $(sel).each((_, n) => {
      const e = $(n);
      const label = normLabel(e.text());
      if (!label || label.length > 16) return;
      const value = valueFor(e);
      if (value.length) pairs.push({ label, value });
    });
  };
  addLabels(conf.rowLabel, (e) => e.nextAll(conf.rowValue).first());
  for (const sel of p.labelSelectors ?? []) addLabels(sel, (e) => e.next());
  const byLabel = (names: string[]): El | null =>
    pairs.find((x) => names.some((n) => x.label.startsWith(normLabel(n))))?.value ?? null;
  page.probe.labels = [...new Set(pairs.map((x) => x.label))].slice(0, 40);

  // 简介容器
  const descSel = conf.fields?.description ?? conf.description;
  let desc: El | null = descSel ? $(descSel).first() : null;
  if (!desc?.length) desc = byLabel(LABELS.description);
  page.probe.hasDescription = !!desc?.length;

  // MediaInfo:站点选择器 → 字段标签 → 简介里的引用 / 代码块(取最内层文本,整块从简介里去掉)
  let mediaInfoEl: El | null = null;
  let miBlockIndex = -1;
  if (p.mediaInfo) {
    const el = $(p.mediaInfo).first();
    if (el.length) mediaInfoEl = el;
  }
  mediaInfoEl ??= byLabel(LABELS.mediaInfo);
  const blocks = desc?.length ? desc.find(MI_BLOCKS) : null;
  if (!mediaInfoEl && blocks) {
    blocks.each((i, n) => {
      if (miBlockIndex >= 0) return;
      if (looksLikeMediaInfo(blockText($, $(n)))) miBlockIndex = i;
    });
    if (miBlockIndex >= 0) {
      const outer = blocks.eq(miBlockIndex);
      const inner = outer
        .find(MI_BLOCKS)
        .filter((_, n) => looksLikeMediaInfo(blockText($, $(n))))
        .last();
      mediaInfoEl = inner.length ? inner : outer;
    }
  }
  if (mediaInfoEl) {
    const t = blockText($, mediaInfoEl);
    if (t) page.mediaInfo = t.slice(0, 50_000);
  }

  // 截图
  const imgs: string[] = [];
  const imgEls = p.screenshots ? $(p.screenshots) : desc?.length ? desc.find("img") : null;
  imgEls?.each((_, n) => {
    const e = $(n);
    const u = screenshotUrl(e.attr("data-src") ?? e.attr("data-original") ?? e.attr("src"), base);
    if (u && !imgs.includes(u)) imgs.push(u);
  });
  page.probe.imageCount = imgs.length;
  page.screenshots = imgs.slice(0, MAX_SCREENSHOTS);

  // 简介纯文本:去掉 MediaInfo 块与图片
  if (desc?.length) {
    const c = desc.clone();
    if (miBlockIndex >= 0) c.find(MI_BLOCKS).eq(miBlockIndex).remove();
    c.find("img").remove();
    const t = blockText($, c);
    if (t) page.descriptionText = t.slice(0, 20_000);
  }

  // 外部 ID:页面上的链接优先,其次简介正文
  const descHtml = desc?.html() ?? "";
  const f = conf.fields ?? {};
  page.imdbId =
    extractImdbId(mcSelect($, null, f.imdb ?? "a[href*='imdb.com']::attr(href)")) ?? extractImdbId(descHtml) ?? null;
  page.doubanId =
    extractDoubanId(mcSelect($, null, f.douban ?? "a[href*='douban.com']::attr(href)")) ??
    extractDoubanId(descHtml) ??
    null;
  page.tmdb = extractTmdb($("a[href*='themoviedb.org']").first().attr("href")) ?? extractTmdb(descHtml);

  // 文件清单:站点选择器 → MovieClaw file_list(去伪元素)→ 字段标签 → #filelist 表格
  const fileSel = p.files ?? f.fileList?.replace(/::text$/, "");
  let files = fileSel ? filesFrom($, $(fileSel)) : [];
  if (!files.length) {
    const v = byLabel(LABELS.files);
    if (v) files = filesFrom($, v.find("li, tr"));
  }
  if (!files.length) files = filesFrom($, $("#filelist tr"));
  page.files = files;

  // 标签
  const tagEl = p.tags ? $(p.tags) : byLabel(LABELS.tags);
  if (tagEl?.length) {
    const parts = tagEl.find("span, a").length
      ? tagEl
          .find("span, a")
          .map((_, n) => $(n).text().trim())
          .get()
      : tagEl.text().split(/[\s,，|/]+/);
    page.tags = [...new Set(parts.map((t) => t.trim()).filter((t) => t && t.length <= 20))].slice(0, 20);
  }

  // 上传者
  const upEl = p.uploader ? $(p.uploader).first() : byLabel(LABELS.uploader);
  if (upEl?.length) {
    const a = upEl.is("a") ? upEl : upEl.find("a[href*='userdetails.php']").first();
    const name = (a.length ? a.text() : upEl.text()).replace(/\s+/g, " ").trim();
    if (name) page.uploader = name.slice(0, 64);
  }
  if (!page.uploader) {
    for (const x of pairs) {
      const m = /(?:上传者|发布者|发布人|Uploader)\s*[:：]\s*(\S+)/i.exec(x.value.text());
      if (m?.[1]) {
        page.uploader = m[1].slice(0, 64);
        break;
      }
    }
  }
  return page;
}