#!/usr/bin/env node
/**
 * 生成 public/llms.txt（llmstxt.org 标准）——AI 爬虫的站点摘要索引。
 *
 * 数据源：zh 目录下全部 Markdown 的 H1 + :::tip 概述首句（与 meta
 * description 同源，复用 .vitepress/config/extract.mjs）。按侧边栏章节分组。
 *
 * 用法：node scripts/generate-llms-txt.mjs
 * 已挂入 package.json 的 docs:build 前置步骤。
 */
import fs from "node:fs";
import path from "node:path";
import { fileURLToPath } from "node:url";
import {
  SITE_URL,
  SITE_DESCRIPTION,
  extractDescription,
} from "../.vitepress/config/extract.mjs";

const ROOT = path.resolve(path.dirname(fileURLToPath(import.meta.url)), "..");
const ZH = path.join(ROOT, "zh");

/** 顶层目录 → 侧边栏章节名（与 .vitepress/config/zh.ts 保持一致） */
const SECTIONS = [
  ["", "1. 了解项目"],
  ["volume1", "2. 快速入门"],
  ["volume2", "3. 模块开发实战"],
  ["volume3", "4. 架构设计"],
  ["volume4", "5. 开发指南"],
  ["volume5", "6. 代码生成器"],
  ["volume6", "7. 运维部署"],
  ["volume7", "8. 规范标准"],
  ["other", "附录"],
];

function walk(dir, out = []) {
  for (const entry of fs.readdirSync(dir, { withFileTypes: true })) {
    const p = path.join(dir, entry.name);
    if (entry.isDirectory()) walk(p, out);
    else if (entry.name.endsWith(".md")) out.push(p);
  }
  return out;
}

/** 提取首个 H1 文本 */
function h1Of(md) {
  // 围栏感知：跳过 ``` 代码块内的 # 注释，取真正的页面 H1
  let inFence = false;
  for (const line of md.split("\n")) {
    if (line.startsWith("```")) { inFence = !inFence; continue; }
    if (!inFence && line.startsWith("# ") && !line.startsWith("## ")) {
      return line.slice(2).trim();
    }
  }
  return "";
}

/** H1 编号排序键："5.10 x" → [5,10]；无编号排最后 */
function sortKey(h1) {
  const m = h1.match(/^(\d+(?:\.\d+)*)/);
  if (m) return m[1].split(".").map(Number);
  const letter = h1.match(/^([A-Z])\./);
  if (letter) return [1000 + letter[1].charCodeAt(0)];
  return [9999];
}

function cmpKey(a, b) {
  const len = Math.max(a.length, b.length);
  for (let i = 0; i < len; i++) {
    const x = a[i] ?? 0;
    const y = b[i] ?? 0;
    if (x !== y) return x - y;
  }
  return 0;
}

const files = walk(ZH);
const groups = new Map(SECTIONS.map(([dir, label]) => [label, []]));

for (const file of files) {
  const rel = path.relative(ZH, file).replace(/\\/g, "/");
  const topDir = rel.includes("/") ? rel.split("/")[0] : "";
  const base = rel.split("/").pop();
  const label = (topDir === "" && base === "faq.md") ? "附录"
    : SECTIONS.find(([dir]) => dir === topDir)?.[1] ?? "附录";

  const md = fs.readFileSync(file, "utf8");
  const h1 = h1Of(md);
  if (!h1) continue;
  const desc = extractDescription(md);
  const url = `${SITE_URL}/zh/${rel.replace(/\.md$/, ".html")}`;
  groups.get(label).push({ h1, desc, url, key: sortKey(h1) });
}

const lines = [
  "# DjangoAdmin Tornado+ElementPlus 开发文档",
  "",
  `> ${SITE_DESCRIPTION}。内置 RBAC 权限体系、代码生成器等开箱即用功能，支持五种数据库一键切换。本文档共 ${files.length} 篇，按章节组织。`,
  "",
];

// 按章节顺序输出，合并重复 label（appendix/other 同为「附录」，只输出一次）
const printed = new Set();
for (const [, label] of SECTIONS) {
  if (printed.has(label)) continue;
  const items = groups.get(label);
  if (!items?.length) continue;
  printed.add(label);

  lines.push(`## ${label}`, "");
  items.sort((a, b) => cmpKey(a.key, b.key));
  for (const { h1, desc, url } of items) {
    lines.push(desc ? `- [${h1}](${url}): ${desc}` : `- [${h1}](${url})`);
  }
  lines.push("");
}

const out = path.join(ROOT, "public", "llms.txt");
fs.writeFileSync(out, lines.join("\n"), "utf8");
console.log(
  `llms.txt: ${files.length} 页 → ${path.relative(ROOT, out)} (${printed.size} 个章节)`,
);