import { redactString } from "@umt/contracts";
import { load } from "cheerio";

// 网页文本工具:HTML → 纯文本(去脚本、样式、表单等非正文元素)、实体解码、截断与脱敏。

const NAMED: Record<string, string> = { amp: "&", lt: "<", gt: ">", quot: '"', apos: "'", nbsp: " " };

/** 去掉摘要里的内联标签(Brave 用 <strong> 标关键字)并解码实体 */
export function stripInlineHtml(s: string): string {
  return s
    .replace(/<[^>]*>/g, "")
    .replace(/&(#x[0-9a-f]+|#\d+|[a-z]+);/gi, (m, e: string) => {
      if (e[0] === "#") {
        const code = e[1] === "x" || e[1] === "X" ? Number.parseInt(e.slice(2), 16) : Number(e.slice(1));
        return Number.isFinite(code) && code > 0 && code < 0x110000 ? String.fromCodePoint(code) : m;
      }
      return NAMED[e.toLowerCase()] ?? m;
    })
    .replace(/\s+/g, " ")
    .trim();
}

/** 按字符截断,末尾加 … */
export function truncate(s: string, max: number): { text: string; truncated: boolean } {
  if (s.length <= max) return { text: s, truncated: false };
  return { text: `${s.slice(0, Math.max(0, max - 1))}…`, truncated: true };
}

/** 交给模型前的清理:脱敏(URL 参数里的 passkey / token 等)+ 截断 */
export function sanitizeForModel(s: string, max: number): { text: string; truncated: boolean } {
  return truncate(redactString(s), max);
}

const DROP = "script,style,noscript,svg,canvas,iframe,object,embed,form,input,select,button,link,meta";
const BLOCK =
  "p,div,section,article,main,header,footer,aside,nav,li,ul,ol,table,tr,h1,h2,h3,h4,h5,h6,pre,blockquote,dd,dt,br,hr";

/** HTML → { title, text }:去脚本样式等,块级元素换行,合并空白 */
export function htmlToText(html: string): { title: string | null; text: string } {
  // <template> 的内容不在 DOM 文本里,但 Wiki.js 等把正文放在 <template slot="contents">:先改成普通元素
  const $ = load(html.replace(/<template\b/gi, "<div").replace(/<\/template>/gi, "</div>"));
  const title = $("title").first().text().replace(/\s+/g, " ").trim() || null;
  $(DROP).remove();
  $("head").remove();
  $(BLOCK).each((_i, el) => {
    $(el).prepend("\n").append("\n");
  });
  const body = $("body").length ? $("body").text() : $.root().text();
  const text = body
    .split("\n")
    .map((l) => l.replace(/[ \t\f\v ]+/g, " ").trim())
    .filter((l) => l.length > 0)
    .join("\n");
  return { title, text };
}