// ประกาศจัดซื้อจัดจ้างภาครัฐ (egp)

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

const BASE = "https://ladsawai.go.th";
const OUT = path.join(process.cwd(), "ประกาศจัดซื้อจัดจ้างภาครัฐ (egp)");
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(src) {
  if (!src) return null;
  if (src.startsWith("http")) return src;
  if (src.startsWith("/")) return BASE + src;
  return BASE + "/" + src; // กัน "public/..."
}

function scrapeOnePage(menuId, page, saveHtml = false) {
  const url = `${BASE}/public/rss/egp/listegp/menu/${menuId}/page/${page}`;
  const html = curlHtml(url);

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

  const $ = cheerio.load(html);
  const items = [];

  // ✅ ตารางรายการ
  $("table tbody tr").each((_, tr) => {
    const tds = $(tr).find("td");
    if (tds.length < 3) return;

    const date = $(tds[0]).text().replace(/\s+/g, " ").trim();
    const category = $(tds[1]).text().replace(/\s+/g, " ").trim();

    const a = $(tds[2]).find("a[href]").first();
    const title = a.text().replace(/\s+/g, " ").trim();
    const link = absUrl(a.attr("href"));

    if (!title) return;

    items.push({
      title,
      date: date || null,
      category: category || null,
      link: link || null,
      sourcePage: page,
      sourceUrl: url,
    });
  });

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

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

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

(function main() {
  const menuId = 1564;
  const totalPages = 240;

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

  const saveHtml = false;

  for (let page = 1; page <= totalPages; page++) {
    const items = scrapeOnePage(menuId, page, saveHtml);

    for (const it of items) {
      // ✅ FIX: EGP ไม่มี image → ใช้ title+date+category+link
      const key = `${it.title}|${it.date || ""}|${it.category || ""}|${it.link || ""}`;
      if (seen.has(key)) continue;
      seen.add(key);
      all.push(it);
    }
  }

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

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

  console.log("✅ Saved merged JSON:", outAll);
  console.log("✅ Total unique items:", all.length);
})();