// หน้าแรก > งานบริการ > ศูนย์ข้อมูลข่าวสาร

const { execSync } = require("child_process");
const cheerio = require("cheerio");
const fs = require("fs");
const path = require("path");
const axios = require("axios").default;

const BASE = "https://ladsawai.go.th";
const OUT = path.join(process.cwd(), "ศูนย์ข้อมูลข่าวสาร");
fs.mkdirSync(OUT, { recursive: true });

function curlHtml(url) {
  return execSync(
    `curl -L -s "${url}" -H "User-Agent: Mozilla/5.0" -H "Accept-Language: th-TH,th;q=0.9"`,
    { encoding: "utf8", maxBuffer: 30 * 1024 * 1024 }
  );
}

function absUrl(href) {
  if (!href) return null;
  if (href.startsWith("http")) return href;
  if (href.startsWith("/")) return BASE + href;
  return BASE + "/" + href;
}

function buildUrl(menuId, page) {
  return `${BASE}/public/list/data/index/menu/${menuId}/page/${page}`;
}

function detectTotalPages($) {
  let maxPage = 1;
  $("a").each((_, a) => {
    const t = $(a).text().trim();
    if (/^\d+$/.test(t)) maxPage = Math.max(maxPage, Number(t));
  });
  return maxPage;
}

function extractFileLinksFromDetail(detailUrl) {
  const html = curlHtml(detailUrl);
  const $ = cheerio.load(html);

  const files = [];

  $("a.uploadconfig_link").each((_, a) => {
    const el = $(a);
    const href = el.attr("href");
    const dataHref = el.attr("data-href");
    const fileUrl = absUrl(dataHref || href);
    if (!fileUrl) return;

    const text = el.text().replace(/\s+/g, " ").trim() || null
    let title = text
    let downloadCount = 0
    if(text && text.includes('ดาวน์โหลดแล้ว')){
        try {
            const splitList = text.split(' ดาวน์โหลดแล้ว ')
            title = splitList[0]
            downloadCount = parseInt(splitList[1].replace('ครั้ง', '').trim())
        } catch (error) {
            title = text
            downloadCount = 0
        }
    }

    files.push({
      text: title,
      url: fileUrl,
      downloadCount: downloadCount
    });
  });

  // fallback: ลิงก์ไฟล์แบบตรง ๆ
  $("a[href]").each((_, a) => {
    const href = $(a).attr("href");
    const u = absUrl(href);
    if (!u) return;

    if (/\.(pdf|doc|docx|xls|xlsx|ppt|pptx|zip|rar)(\?|$)/i.test(u)) {
      if (!files.some((f) => f.url === u)) {
        files.push({ text: $(a).text().trim() || null, url: u });
      }
    }
  });

  return files;
}

// ✅ ยิง api /status/1/ เพื่อเอา path จริง
async function resolveRealFilePath(fileUrl) {
  try {
    // กันกรณีมี / ท้ายอยู่แล้ว
    const statusUrl = fileUrl.replace(/\/$/, "") + "/status/1/";
    const res = await axios.get(statusUrl, { timeout: 30000 });
    return res?.data?.path || null;
  } catch (e) {
    return null;
  }
}

async function scrapeOnePage(menuId, page, saveHtml = false) {
  const url = buildUrl(menuId, page);
  const html = curlHtml(url);

  if (saveHtml) {
    fs.writeFileSync(
      path.join(OUT, `debug-menu-${menuId}-page-${page}.html`),
      html,
      "utf8"
    );
  }

  const $ = cheerio.load(html);

    let items = []

    try {
        const files = extractFileLinksFromDetail(url);
        for(let i = 0; i < files.length; i++){
            const file = files[i]
            let realPath = null;
            let fileObject = {
                fileName: file.text,
                fileUrl: file.url, // ไฟล์จากหน้า detail
                filePath: "", // ✅ ของจริงจาก api /status/1/
                downloadCount: file.downloadCount
            }
            try {
                const fileUrl = file?.url ? absUrl(file.url) : null;
                if (fileUrl) {
                    realPath = await resolveRealFilePath(fileUrl);
                    fileObject.filePath = `https://ladsawai.go.th/public/${realPath}`
                }
            } catch (error) {
                realPath = null;
            }

            items.push(fileObject)
        }
    } catch (e) {
    }

  const output = {
    source: url,
    scrapedAt: new Date().toISOString(),
    menuId,
    page,
    count: items.length,
    items,
  };

  fs.writeFileSync(
    path.join(OUT, `menu-${menuId}-page-${page}.json`),
    JSON.stringify(output, null, 2),
    "utf8"
  );

  console.log(`✅ page ${page} -> items ${items.length}`);
  return { $, items };
}

(async function main() {
  const menuId = 1585;

  const first = await scrapeOnePage(menuId, 1, true);
  const totalPages = detectTotalPages(first.$);
  console.log("✅ totalPages =", totalPages);

  const all = [];
  const seen = new Set();

  function addItems(items) {
    for (const it of items) {
      const key = `${it.title}|${it.detailUrl || ""}|${it.filePath || ""}`;
      if (seen.has(key)) continue;
      seen.add(key);
      all.push(it);
    }
  }

  addItems(first.items);

  for (let p = 2; p <= totalPages; p++) {
    const { items } = await scrapeOnePage(menuId, p, false);
    addItems(items);
  }

  const merged = {
    menuId,
    totalPages,
    scrapedAt: new Date().toISOString(),
    totalItems: all.length,
    items: all,
  };

  const outAll = path.join(OUT, `menu-${menuId}-all.json`);
  fs.writeFileSync(outAll, JSON.stringify(merged, null, 2), "utf8");

  console.log("🎉 Saved all:", outAll);
  console.log("🎉 Total unique:", all.length);
})();