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

/**
 * 详情页解析 —— 本项目扩展。MP 通用 SiteSpider 不解析详情页(MP 只在列表/搜索结果上工作),
 * 这里按 NexusPHP details.php 公开结构(h1#top + td.rowhead/td.rowfollow)实现,选择器在站点定义 `umt.detail` 中配置。
 * 促销/H&R 复用站点定义中 downloadvolumefactor / uploadvolumefactor / freedate / hr 字段(与列表同语义)。
 */
import type { CheerioAPI } from "cheerio";
import type { SiteReleaseDetail } from "../types.ts";
import { extractDoubanId, extractImdbId, parseCount, parseSiteTime } from "../util.ts";
import { extractDetailPage } from "./detail-page.ts";
import { categoryName, promotionFromFactors } from "./map.ts";
import { mcSelect } from "./mc-selector.ts";
import type { IndexerDefinition } from "./schema.ts";
import { loadHtml, parseSizeMp, SiteSpider } from "./spider.ts";

export type DetailPageKind = "ok" | "login" | "not_found" | "unknown";

export function parseDetailPage(
  html: string,
  def: IndexerDefinition,
  id: string,
  observedAt: string,
): { kind: DetailPageKind; detail: SiteReleaseDetail | null } {
  const conf = def.umt.detail;
  if (!conf) return { kind: "unknown", detail: null };
  const $ = loadHtml(html);
  if (SiteSpider.isLoginOrPermissionPage($)) return { kind: "login", detail: null };
  const bodyHead = $("body").text().replace(/\s+/g, " ").slice(0, 2000);
  if (conf.notFoundMarkers.some((m) => bodyHead.includes(m))) return { kind: "not_found", detail: null };
  const h1 = $(conf.title).first();
  if (h1.length === 0) return { kind: "unknown", detail: null };

  const row = (labels: string[]) => {
    let found: ReturnType<CheerioAPI> | null = null;
    $(conf.rowLabel).each((_, el) => {
      const label = $(el).text().replace(/\s+/g, "");
      if (!found && labels.some((l) => label.startsWith(l.replace(/\s+/g, "")))) {
        found = $(el).nextAll(conf.rowValue).first();
      }
    });
    return found as ReturnType<CheerioAPI> | null;
  };

  // 逐字段选择器(MovieClaw detail_*_css)优先于 rowhead 表格模型
  const f = conf.fields ?? {};
  const title = (
    f.title
      ? (mcSelect($, null, f.title) ?? "")
      : h1
          .contents()
          .filter((_, n) => n.type === "text")
          .text()
  )
    .replace(/\s+/g, " ")
    // 促销到期的「剩余时间：」是 h1 的直接文本节点(观众),不属于标题
    .replace(/\s*(?:剩余时间|剩餘時間|Remaining(?: time)?)\s*[:：]?\s*$/i, "")
    .trim();
  const subtitleEl = f.subtitle ? null : row(conf.labels.subtitle)?.clone();
  // 副标题行右侧的豆瓣 / IMDb / 烂番茄等评分徽标(观众)不算副标题
  subtitleEl?.find("[title*='评分'], [title*='Rating'], img").remove();
  const subtitle = f.subtitle
    ? (mcSelect($, null, f.subtitle) ?? undefined)
    : subtitleEl?.text().replace(/\s+/g, " ").trim();
  const basicText = row(conf.labels.basicInfo)?.text().replace(/\s+/g, " ") ?? "";
  const sizeText = f.size
    ? (/([\d.,]+\s*[KMGTP]?i?B)/i.exec(mcSelect($, null, f.size) ?? "")?.[1] ?? null)
    : (/(?:大小|Size)\s*[:：]\s*([\d.,]+\s*[KMGTP]?i?B)/i.exec(basicText)?.[1] ?? null);
  const catText = /(?:类型|Type)\s*[:：]\s*(\S+)/i.exec(basicText)?.[1] ?? null;
  const catId = catText
    ? (Object.entries(def.umt.categoryNames ?? {}).find(([, name]) => name === catText)?.[0] ?? null)
    : null;
  const peersText = row(conf.labels.peers)?.text().replace(/\s+/g, " ") ?? "";
  const seeders = /(\d[\d,]*)\s*个?做种者|(\d[\d,]*)\s*seeder/i.exec(peersText);
  const leechers = /(\d[\d,]*)\s*个?下载者|(\d[\d,]*)\s*leecher/i.exec(peersText);
  // 站点把做种 / 下载 / 完成数放在其它位置时(观众:基本信息行右侧的 #seeders / #leechers / 完成链接)按选择器取
  const countOf = (sel: string | undefined) => (sel ? parseCount(mcSelect($, null, sel) ?? undefined) : null);
  const infoHash = mcSelect($, null, f.infoHash)?.toLowerCase();
  const descEl = $(conf.description).first();
  const descText = descEl.text().trim();
  const descHtml = descEl.html() ?? "";

  // 促销/HR:只用这四个字段在详情页头部取值
  const fields = def.torrents.fields;
  const promoDef: IndexerDefinition = {
    ...def,
    torrents: {
      list: def.torrents.list,
      fields: Object.fromEntries(
        (["downloadvolumefactor", "uploadvolumefactor", "freedate", "hr"] as const)
          .map((k) => [k, conf.promo?.[k] ?? fields[k]] as const)
          .filter(([, v]) => v),
      ),
    },
    browse: undefined,
  };
  const promo = new SiteSpider(promoDef).infoFrom($, $(conf.headerScope).first());
  // 发布时间:带完整时间的 span[title] → 「发布时间 / 上传时间 / 添加时间」字段的文本(憨憨是 div 标签 + 值)
  // 标题区(headerScope)里的 span[title] 是促销到期时间,不是发布时间
  const timeTitle =
    $("span[title]")
      .filter(
        (_, el) =>
          /\d{4}-\d{2}-\d{2} \d{2}:\d{2}/.test($(el).attr("title") ?? "") &&
          $(el).closest(conf.headerScope).length === 0,
      )
      .first()
      .attr("title") ??
    /(?:发布时间|上传时间|添加时间|Added)\s*[:：]?\s*(\d{4}-\d{2}-\d{2} \d{2}:\d{2}(?::\d{2})?)/.exec(
      $("body").text().replace(/\s+/g, " "),
    )?.[1];

  const detail: SiteReleaseDetail = {
    siteId: def.id,
    siteReleaseId: id,
    title,
    category: { id: catId, name: catText ?? categoryName(def, catId ?? undefined) },
    sizeBytes: sizeText ? parseSizeMp(sizeText) : null,
    seeders: seeders ? parseCount(seeders[1] ?? seeders[2]) : countOf(f.seeders),
    leechers: leechers ? parseCount(leechers[1] ?? leechers[2]) : countOf(f.leechers),
    snatched: countOf(f.snatched),
    promotion: promotionFromFactors(
      promo.downloadvolumefactor,
      promo.uploadvolumefactor,
      promo.freedate ? parseSiteTime(promo.freedate, def.umt.timezone) : null,
    ),
    hr: promo.hit_and_run === undefined ? "unknown" : promo.hit_and_run,
    uploadedAt: parseSiteTime(timeTitle, def.umt.timezone),
    detailUrl: new URL(`${def.umt.detailPath}?id=${id}`, def.domain).toString(),
    downloadUrl: new URL(`${def.umt.downloadPath}?id=${id}`, def.domain).toString(),
    tags: [],
    raw: { basicInfoText: basicText.trim().slice(0, 500), peersText: peersText.trim(), ...promo },
    observedAt,
  };
  if (subtitle) detail.subtitle = subtitle;
  if (infoHash && /^[0-9a-f]{40}$/.test(infoHash)) detail.infoHash = infoHash;
  if (descText) detail.descriptionText = descText.slice(0, 20_000);
  const imdb = extractImdbId(mcSelect($, null, f.imdb)) ?? extractImdbId(descHtml) ?? extractImdbId(descText);
  if (imdb) detail.imdbId = imdb;
  const douban = extractDoubanId(mcSelect($, null, f.douban)) ?? extractDoubanId(descHtml);
  if (douban) detail.doubanId = douban;
  const page = extractDetailPage($, conf, def.domain);
  page.imdbId ??= detail.imdbId ?? null;
  page.doubanId ??= detail.doubanId ?? null;
  // 详情页扩展内容里兜底出来的外部 ID(MediaInfo 的 IMDB 行、豆瓣信息块)也回填到资源本身
  if (!detail.imdbId && page.imdbId) detail.imdbId = page.imdbId;
  if (!detail.doubanId && page.doubanId) detail.doubanId = page.doubanId;
  if (page.tags.length) detail.tags = page.tags;
  detail.page = page;
  return { kind: "ok", detail };
}