/**
 * @mp-source app/modules/indexer/__init__.py @ 3b156a5
 * 通用 spider 站点适配器(NexusPHP 等 HTML 站)。
 *
 * 对应 MoviePilot v3.2.4:
 * - `app/modules/indexer/__init__.py IndexerModule.search_torrents / refresh_torrents` → `search` / `list`
 *   (`__search_check` 的英文站不搜中文 → search 前检查;`__clear_search_text` → `clearSearchText`;
 *    `SitesHelper.check(domain)` 流控 → 站点令牌桶)
 * - `SiteSpider.get_torrents`(RequestUtils GET + cookie + ua)→ `fetchPage`
 * - `app/chain/download/submission.py download_torrent`(带 cookie GET enclosure,磁力原样返回)→ `downloadTarget`
 *
 * 本项目调整:登录页不再"返回空列表",而是抛 `auth_expired`(REQ §4.7.1 不展示空列表冒充无内容);
 * 3xx 不自动跟随(避免把 cookie 带到站外,并把跳转 login 识别为失效);凭证经 SecretResolver 注入。
 */

import { inspectTorrentIsolated } from "../../isolation.ts";
import { redactText } from "../../rss/redact.ts";
import { challengeOutcomeText, challengeRuleOfUnrecognizedPage } from "../challenge.ts";
import { buildHttpContext, type SiteAdapterDeps } from "../deps.ts";
import { buildFailureSample, type SiteRequestKind } from "../failure-sample.ts";
import { readBody, readText, siteFetch } from "../http.ts";
import {
  type DownloadTarget,
  type SiteAdapter,
  type SiteCallOptions,
  SiteError,
  type SiteHealth,
  type SiteHealthDetail,
  type SiteListFilters,
  type SitePage,
  type SiteReleaseDetail,
} from "../types.ts";
import { parseDetailPage } from "./detail.ts";
import { filesFromFileListHtml } from "./detail-page.ts";
import { spiderTorrentToRelease } from "./map.ts";
import type { IndexerDefinition } from "./schema.ts";
import { loadHtml, normalizeSelector, SiteSpider } from "./spider.ts";

/** MP `text_tools.remove_punctuation(text, replacement=" ", allow_space=True)` */
export function clearSearchText(text: string): string {
  return text
    .replace(/[​-‍]/g, "")
    .replace(/[、.。,，·:：;；!！?？'’"“”()（）[\]【】「」\-—―+|\\_/&#～~]/g, " ")
    .replace(/\s+/g, " ")
    .trim();
}

function containsChinese(s: string): boolean {
  return /[一-鿿]/.test(s);
}

export function healthFromError(siteId: string, e: unknown, checkedAt: string): SiteHealth {
  if (e instanceof SiteError) {
    const status: SiteHealth["status"] =
      e.kind === "network"
        ? "unreachable"
        : e.kind === "not_found" || e.kind === "unsupported"
          ? "parse_error"
          : e.kind;
    const detail = e.detail ?? (e.httpStatus !== undefined ? { reason: "http", httpStatus: e.httpStatus } : undefined);
    return {
      siteId,
      status,
      ok: false,
      checkedAt,
      latencyMs: null,
      message: e.message,
      ...(detail ? { detail } : {}),
    };
  }
  return { siteId, status: "unreachable", ok: false, checkedAt, latencyMs: null, message: redactText(String(e)) };
}

/**
 * 健康检查的首页登录判定:登录页 / 权限页 → 未登录;登录标志选择器命中 → 已登录;
 * 否则区分"疑似人机验证"与"页面里找不到登录标志"。detail 是原始判定(不含页面里的个人数据)。
 */
export function judgeLoginPage(
  html: string,
  loginSuccess: string | undefined,
): { ok: true; detail: SiteHealthDetail } | { ok: false; message: string; detail: SiteHealthDetail } {
  const $ = loadHtml(html);
  const base = { path: "index.php", title: $("title").first().text().trim().slice(0, 80) };
  const login = SiteSpider.loginPageRule($);
  if (login) {
    return {
      ok: false,
      message:
        login.kind === "login_page"
          ? "健康检查:首页返回登录页,Cookie 已失效"
          : "健康检查:首页返回权限提示页(要求先登录)",
      detail: { ...base, reason: login.kind, rule: login.rule },
    };
  }
  if (!loginSuccess || $(loginSuccess).length > 0) {
    return { ok: true, detail: { ...base, reason: "logged_in", ...(loginSuccess ? { selector: loginSuccess } : {}) } };
  }
  const challenge = SiteSpider.challengeRule($, html);
  if (challenge) {
    return {
      ok: false,
      message: "健康检查:疑似人机验证(Cloudflare),无法确认登录态",
      detail: { ...base, reason: "challenge", rule: challenge },
    };
  }
  return {
    ok: false,
    message: `健康检查:首页里找不到登录标志(选择器 ${loginSuccess})`,
    detail: { ...base, reason: "marker_missing", selector: loginSuccess, length: html.length },
  };
}

/** 抓到的页面:正文与留样用的响应信息 */
interface FetchedPage {
  html: string;
  url: string;
  status: number;
  headers: Headers;
}

/** 错误文案里的页面名 */
const PAGE_LABEL: Record<SiteRequestKind, string> = {
  search: "搜索页",
  list: "列表页",
  detail: "详情页",
  health: "首页",
  other: "页面",
};

export function createSpiderSiteAdapter(input: IndexerDefinition, deps: SiteAdapterDeps): SiteAdapter {
  const def: IndexerDefinition = deps.baseUrl ? { ...input, domain: deps.baseUrl } : input;
  const ext = def.umt;
  const http = buildHttpContext(def.id, deps, ext.rateLimit);
  const base = def.domain.endsWith("/") ? def.domain : `${def.domain}/`;

  async function cookie(): Promise<string> {
    const c = await deps.secrets.resolve({ siteId: def.id, kind: "cookie" });
    if (!c && def.public) return "";
    if (!c) throw new SiteError(def.id, "auth_missing", "未配置站点 Cookie");
    return c;
  }

  /** 留样:交给调用方保存;回调出错不影响请求结果 */
  function report(e: SiteError): SiteError {
    if (e.sample && deps.onFailureSample) {
      try {
        deps.onFailureSample(e.sample);
      } catch {
        // 留样失败不影响请求
      }
    }
    return e;
  }

  /** 页面认不出来时的错误:像人机验证就报 challenge,否则 parse_error;都附页面样本 */
  function unrecognized(page: FetchedPage, what: string, kind: SiteRequestKind, parseMessage: string): SiteError {
    const rule = challengeRuleOfUnrecognizedPage(page.html);
    const title = redactText(loadHtml(page.html)("title").first().text().replace(/\s+/g, " ").trim()).slice(0, 120);
    const message = rule
      ? `${what}：${PAGE_LABEL[kind]}需要人机验证（Cloudflare 等），${challengeOutcomeText("unhandled")}`
      : `${parseMessage}（HTTP ${page.status}${title ? `，页面标题：${title}` : ""}；页面样本见站点页技术信息）`;
    const sample = buildFailureSample({
      kind: rule ? "challenge" : "parse_error",
      requestKind: kind,
      url: page.url,
      httpStatus: page.status,
      headers: page.headers,
      message,
      rule,
      html: page.html,
      now: http.now(),
    });
    return report(
      new SiteError(def.id, rule ? "challenge" : "parse_error", message, {
        httpStatus: page.status,
        ...(rule ? { detail: { reason: "challenge", rule, httpStatus: page.status } } : {}),
        sample,
      }),
    );
  }

  async function fetchPage(
    url: string,
    what: string,
    opts: SiteCallOptions,
    kind: SiteRequestKind,
  ): Promise<FetchedPage> {
    let res: Response;
    try {
      res = await siteFetch(http, url, {
        method: "GET",
        headers: {
          cookie: await cookie(),
          accept: "text/html,application/xhtml+xml,application/rss+xml,application/xml",
          referer: base,
        },
        redirect: "manual",
        ...(opts.signal ? { signal: opts.signal } : {}),
      });
    } catch (e) {
      if (e instanceof SiteError && e.kind === "challenge") {
        // 网络层判出的人机验证:换成带请求类型的文案,样本补上请求类型
        const outcome = typeof e.detail?.outcome === "string" ? e.detail.outcome : null;
        const message = `${what}：${PAGE_LABEL[kind]}需要人机验证（Cloudflare 等），${challengeOutcomeText(outcome)}`;
        throw report(
          new SiteError(def.id, "challenge", message, {
            ...(e.httpStatus !== undefined ? { httpStatus: e.httpStatus } : {}),
            ...(e.detail ? { detail: e.detail } : {}),
            ...(e.sample ? { sample: { ...e.sample, requestKind: kind, message } } : {}),
          }),
        );
      }
      throw e;
    }
    if (res.status >= 300 && res.status < 400) {
      const loc = res.headers.get("location") ?? "";
      await res.body?.cancel().catch(() => {});
      if (/login/i.test(loc)) {
        throw new SiteError(def.id, "auth_expired", `${what}:跳转登录页,Cookie 已失效`, {
          httpStatus: res.status,
          detail: { reason: "redirect_login", httpStatus: res.status, location: redactText(loc).slice(0, 200) },
        });
      }
      throw new SiteError(def.id, "parse_error", redactText(`${what}:意外重定向到 ${loc}`), { httpStatus: res.status });
    }
    // Cloudflare 挑战响应带 cf-mitigated: challenge(403 / 503)
    if (res.headers.get("cf-mitigated") === "challenge") {
      await res.body?.cancel().catch(() => {});
      throw new SiteError(def.id, "challenge", `${what}:疑似人机验证(Cloudflare)`, {
        httpStatus: res.status,
        detail: { reason: "challenge", rule: "header cf-mitigated=challenge", httpStatus: res.status },
      });
    }
    if (res.status === 401 || res.status === 403) {
      await res.body?.cancel().catch(() => {});
      throw new SiteError(def.id, "auth_invalid", `${what}:HTTP ${res.status}`, { httpStatus: res.status });
    }
    if (!res.ok) {
      await res.body?.cancel().catch(() => {});
      throw new SiteError(def.id, res.status === 404 ? "not_found" : "network", `${what}:HTTP ${res.status}`, {
        httpStatus: res.status,
      });
    }
    return { html: await readText(http, res), url, status: res.status, headers: res.headers };
  }

  async function run(
    page: number,
    filters: SiteListFilters,
    keyword: string | undefined,
    opts: SiteCallOptions,
  ): Promise<SitePage> {
    const what = keyword ? "搜索" : `列表第 ${page} 页`;
    const spider = new SiteSpider(def, {
      ...(keyword ? { keyword } : {}),
      ...(filters.mtype ? { mtype: filters.mtype } : {}),
      ...(filters.categoryIds?.length ? { cat: filters.categoryIds.join(",") } : {}),
      page: Math.max(1, page) - 1, // 对外 1 起,MP 内部 0 起
    });
    if (!spider.canSearch()) throw new SiteError(def.id, "unsupported", `${what}:站点定义不支持该媒体类型`);
    const kind: SiteRequestKind = keyword ? "search" : "list";
    const page$ = await fetchPage(spider.getSearchUrl(), what, opts, kind);
    const parsed = spider.parse(page$.html);
    if (parsed.loginPage) throw new SiteError(def.id, "auth_expired", `${what}:返回登录或权限提示页`);
    if (parsed.isError || !parsed.$) throw unrecognized(page$, what, kind, `${what}:响应为空`);
    const $ = parsed.$;
    if (parsed.torrents.length === 0) {
      const text = $("body").text();
      const hasList =
        $(normalizeSelector(spider.def.torrents.list.selector)).length > 0 ||
        $("table.torrents, rss > channel, feed").length > 0;
      const empty = (ext.emptyMarkers ?? []).some((m) => text.includes(m));
      // 认不出列表:先看是不是人机验证页(搜索页单独加盾、间歇触发),不是才报改版
      if (!hasList && !empty) throw unrecognized(page$, what, kind, `${what}:页面结构无法识别(可能改版)`);
      if (ext.loginSuccess && $(ext.loginSuccess).length === 0 && !empty) {
        throw new SiteError(def.id, "auth_expired", `${what}:未找到登录标志`);
      }
    }
    const observedAt = http.now().toISOString();
    let items = parsed.torrents
      .map((t) => spiderTorrentToRelease(t, def, observedAt))
      .filter((r): r is NonNullable<typeof r> => r !== null);
    if (filters.freeOnly) items = items.filter((r) => r.promotion.downloadFactor === 0);
    if (!keyword && filters.categoryIds?.length) {
      // MP 浏览模式不带分类参数;本地按原始分类过滤
      const set = new Set(filters.categoryIds);
      items = items.filter((r) => r.category.id !== null && set.has(r.category.id));
    }
    let maxRaw: number | null = null;
    if (ext.pagination) {
      $(ext.pagination).each((_, el) => {
        const p = Number(/[?&]page=(\d+)/.exec($(el).attr("href") ?? "")?.[1]);
        if (Number.isInteger(p) && (maxRaw === null || p > maxRaw)) maxRaw = p;
      });
    }
    const start = keyword ? 0 : Number(def.browse?.start ?? 0);
    return {
      items,
      page: Math.max(1, page),
      totalPages: parsed.totalPages ?? (maxRaw === null ? null : Math.max(page, (maxRaw as number) - start + 1)),
      totalPagesEstimated: parsed.totalPages === undefined,
      fetchedAt: observedAt,
    };
  }

  return {
    id: def.id,
    kind: ext.kind,
    displayName: ext.displayName ?? def.name,
    capabilities: ext.capabilities,
    definitionVersion: ext.version,

    list(page, filters = {}, opts = {}) {
      return run(page, filters, undefined, opts);
    },

    async search(query, filters = {}, opts = {}) {
      const kw = /^tt\d+$/.test(query.trim()) ? query.trim() : clearSearchText(query);
      if (kw && def.language === "en" && containsChinese(kw)) {
        throw new SiteError(def.id, "unsupported", "该站点不支持中文搜索");
      }
      return run(filters.page ?? 1, filters, kw, opts);
    },

    async detail(siteReleaseId, opts = {}): Promise<SiteReleaseDetail> {
      if (!/^\d+$/.test(siteReleaseId)) throw new SiteError(def.id, "not_found", "资源 ID 非法");
      if (!ext.detail) throw new SiteError(def.id, "unsupported", "站点定义未配置详情页解析");
      const url = new URL(`${ext.detail.path}?id=${siteReleaseId}&hit=1`, base).toString();
      const detailPage = await fetchPage(url, `详情 ${siteReleaseId}`, opts, "detail");
      const r = parseDetailPage(detailPage.html, def, siteReleaseId, http.now().toISOString());
      if (r.kind === "login") throw new SiteError(def.id, "auth_expired", `详情 ${siteReleaseId}:返回登录页`);
      if (r.kind === "not_found") throw new SiteError(def.id, "not_found", `详情 ${siteReleaseId}:资源不存在`);
      if (!r.detail) {
        throw unrecognized(detailPage, `详情 ${siteReleaseId}`, "detail", `详情 ${siteReleaseId}:页面结构无法识别`);
      }
      // 文件清单不在页面上(NexusPHP 点「查看列表」才 AJAX 取)→ 按需 1 次;失败只是没有文件清单
      const flPath = ext.detail.page?.fileListPath;
      if (flPath && r.detail.page && !r.detail.page.files.length) {
        try {
          const flUrl = new URL(flPath.replace("{id}", siteReleaseId), base).toString();
          const frag = await fetchPage(flUrl, `文件清单 ${siteReleaseId}`, opts, "detail");
          r.detail.page.files = filesFromFileListHtml(loadHtml(frag.html));
        } catch (e) {
          if (e instanceof SiteError && e.isAuthError) throw e;
        }
      }
      return r.detail;
    },

    async downloadTarget(siteReleaseId, opts = {}): Promise<DownloadTarget> {
      if (!/^\d+$/.test(siteReleaseId)) throw new SiteError(def.id, "not_found", "资源 ID 非法");
      const what = `下载 ${siteReleaseId}`;
      const url = new URL(`${ext.downloadPath}?id=${siteReleaseId}`, base).toString();
      const res = await siteFetch(http, url, {
        method: "GET",
        headers: { cookie: await cookie(), referer: base },
        redirect: "manual",
        ...(opts.signal ? { signal: opts.signal } : {}),
      });
      const verifiedAt = http.now().toISOString();
      if (res.status >= 300 && res.status < 400) {
        const loc = res.headers.get("location") ?? "";
        await res.body?.cancel().catch(() => {});
        // MP download_torrent:磁力链原样返回
        if (loc.startsWith("magnet:")) return { kind: "magnet", uri: loc, verifiedAt };
        if (/login/i.test(loc)) throw new SiteError(def.id, "auth_expired", `${what}:跳转登录页,Cookie 已失效`);
        throw new SiteError(def.id, "parse_error", `${what}:意外重定向`, { httpStatus: res.status });
      }
      if (res.headers.get("cf-mitigated") === "challenge") {
        await res.body?.cancel().catch(() => {});
        throw new SiteError(def.id, "challenge", `${what}:疑似人机验证(Cloudflare)`, {
          httpStatus: res.status,
          detail: { reason: "challenge", rule: "header cf-mitigated=challenge", httpStatus: res.status },
        });
      }
      if (!res.ok) {
        await res.body?.cancel().catch(() => {});
        throw new SiteError(def.id, res.status === 404 ? "not_found" : "network", `${what}:HTTP ${res.status}`, {
          httpStatus: res.status,
        });
      }
      const bytes = await readBody(http, res);
      const info = await inspectTorrentIsolated(bytes);
      if (!info) {
        // 站点在 cookie 失效时常返回 200 + 登录页 HTML
        const $ = loadHtml(new TextDecoder().decode(bytes.subarray(0, 200_000)));
        if (SiteSpider.isLoginOrPermissionPage($)) throw new SiteError(def.id, "auth_expired", `${what}:返回登录页`);
        const text = $("body").text();
        if (ext.detail?.notFoundMarkers.some((m) => text.includes(m))) {
          throw new SiteError(def.id, "not_found", `${what}:资源不存在`);
        }
        throw new SiteError(def.id, "parse_error", `${what}:返回内容不是 .torrent`);
      }
      const disposition = res.headers.get("content-disposition") ?? "";
      const fileName =
        /filename\*=UTF-8''([^;]+)/i.exec(disposition)?.[1] ?? /filename="?([^";]+)"?/i.exec(disposition)?.[1] ?? null;
      return {
        kind: "torrent",
        bytes,
        infoHash: info.infoHash,
        name: info.name,
        fileName: fileName ? decodeURIComponent(fileName) : null,
        verifiedAt,
      };
    },

    async healthCheck(opts = {}): Promise<SiteHealth> {
      const started = Date.now();
      const checkedAt = http.now().toISOString();
      try {
        const home = await fetchPage(new URL("index.php", base).toString(), "健康检查", opts, "health");
        const v = judgeLoginPage(home.html, ext.loginSuccess);
        const latencyMs = Date.now() - started;
        if (v.ok) return { siteId: def.id, status: "ok", ok: true, checkedAt, latencyMs, detail: v.detail };
        return {
          siteId: def.id,
          status: v.detail.reason === "challenge" ? "challenge" : "auth_expired",
          ok: false,
          checkedAt,
          latencyMs,
          message: v.message,
          detail: v.detail,
        };
      } catch (e) {
        return healthFromError(def.id, e, checkedAt);
      }
    },
  };
}