migrate-lsw/public-relations.js

// ข่าวประชาสัมพันธ์
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(), "ข่าวประชาสัมพันธ์");
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; // กันเคส data-href = "public/...."
}

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

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

  // ---------- files ----------
  const fileSet = new Set();

  $("a.uploadconfig_link").each((_, a) => {
    const $a = $(a);
    const raw = ($a.attr("data-href") || $a.attr("href") || "").trim();
    const full = absUrl(raw);
    if (full) fileSet.add(full);
  });

  if (fileSet.size === 0) {
    $("a[href], a[data-href]").each((_, a) => {
      const $a = $(a);
      const raw = ($a.attr("data-href") || $a.attr("href") || "").trim();
      const full = absUrl(raw);
      if (!full) return;

      if (/\.(pdf|doc|docx|xls|xlsx|ppt|pptx|zip|rar)(\?|$)/i.test(full)) {
        fileSet.add(full);
      }
    });
  }

  // ✅ Set ต้องแปลงเป็น Array ก่อนเอา [0]
  const firstFileUrl = Array.from(fileSet)[0] || null;

  let realPath = null;
  if (firstFileUrl) {
    const p = await resolveRealFilePath(firstFileUrl);
    realPath = p ? `https://ladsawai.go.th/public/${p}` : null;
  }

  // ---------- images ----------
  const imgSet = new Set();

  $(".maingroup.gallery a[href]").each((_, a) => {
    const href = ($(a).attr("href") || "").trim();
    const full = absUrl(href);
    if (full) imgSet.add(full);
  });

  if (imgSet.size === 0) {
    $("a[href]").each((_, a) => {
      const href = ($(a).attr("href") || "").trim();
      const full = absUrl(href);
      if (full && /\.(jpg|jpeg|png|webp|gif)(\?|$)/i.test(full)) imgSet.add(full);
    });
  }

  // ---------- content ----------
  const candidates = $(".col-12.maingroup").not(".gallery");

  let bestBox = null;
  let bestScore = -1;

  candidates.each((_, el) => {
    const $el = $(el);
    const text = $el
      .clone()
      .find("img, script, style")
      .remove()
      .end()
      .text()
      .replace(/\s+/g, " ")
      .trim();

    const pCount = $el.find("p").length;
    const score = (text ? text.length : 0) + pCount * 50;

    if (score > bestScore) {
      bestScore = score;
      bestBox = $el;
    }
  });

  let content = "";
  if (bestBox && bestBox.length) {
    const lines = [];
    bestBox.find("p").each((_, p) => {
      const t = $(p)
        .clone()
        .find("img")
        .remove()
        .end()
        .text()
        .replace(/\s+/g, " ")
        .trim();
      if (t) lines.push(t);
    });

    content = lines.length
      ? lines.join("\n")
      : bestBox
          .clone()
          .find("img, script, style")
          .remove()
          .end()
          .text()
          .replace(/\s+/g, " ")
          .trim();
  }

  return { imgs: [...imgSet], text: content, files: realPath };
}

// ✅ ต้องเป็น async เพื่อ await detail
async function scrapeOnePage(menuId, page, saveHtml = false) {
  const url = `${BASE}/public/list/data/index/menu/${menuId}/page/${page}`;
  const html = curlHtml(url);

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

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

  // ✅ ห้ามใช้ .each(async ...) ให้ใช้ loop ปกติแทน
  const rows = $(".row.data-row").toArray();
  for (const row of rows) {
    const el = $(row);

    const title = el.find(".col-sm-8").text().replace(/\s+/g, " ").trim();
    if (!title) continue;

    const href = (el.find("a.listdataconfig_link").attr("href") || "").trim();
    if (!href) continue;

    const linkD = absUrl(href);

    const date = el.find(".col-sm-2").last().text().trim();
    const imgSrc = el.find("img").attr("src");

    let detail = { imgs: [], text: "", files: null };
    try {
      detail = linkD ? await scrapeDetailImagesContent(linkD) : detail;
    } catch {
      // กันหน้า detail บางอันพัง
    }

    items.push({
      title,
      detailRef: linkD,
      detail: {
        img: detail.imgs,
        content: detail.text,
        link: detail.files,
      },
      date: date || null,
      image: absUrl(imgSrc),
      sourcePage: page,
      sourceUrl: url,
    });
  }

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

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

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

// ✅ main ต้อง async เพื่อ await scrapeOnePage
(async function main() {
  const menuId = 1554;
  const totalPages = 53;

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

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

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

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

  const outAll = path.join(OUT, `list-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);
})();