umdate
This commit is contained in:
parent
3b0f4992a2
commit
b4142df483
|
|
@ -15,7 +15,7 @@ async function safeGetContent(page) {
|
|||
} catch (err) {
|
||||
// Nếu navigation xảy ra → chờ page ổn định lại rồi đọc tiếp
|
||||
if (err.message.includes("Execution context was destroyed")) {
|
||||
await page.waitForNetworkIdle({ idleTime: 1000 }).catch(() => {});
|
||||
await page.waitForNetworkIdle({ idleTime: 1000 }).catch(() => { });
|
||||
return await page.content();
|
||||
}
|
||||
throw err;
|
||||
|
|
@ -117,11 +117,12 @@ async function scrapeWithPuppeteer(store) {
|
|||
nodes.forEach((node) => {
|
||||
const payload = {};
|
||||
// ---------------- LINK + ID ----------------
|
||||
const linkEl = node.querySelector("div.su-media__image a");
|
||||
const linkEl = node.querySelector("div.su-image a");
|
||||
// if (!linkEl) return;
|
||||
|
||||
payload.link_detail = linkEl?.href || "";
|
||||
const idMatch = linkEl.href.match(/\/itm\/(\d+)/);
|
||||
const linkDetail = linkEl && linkEl?.href ? linkEl?.href : ""
|
||||
if (!linkDetail) return;
|
||||
payload.link_detail = linkDetail;
|
||||
const idMatch = linkDetail.match(/\/itm\/(\d+)/);
|
||||
if (!idMatch) return;
|
||||
|
||||
payload.id = idMatch[1];
|
||||
|
|
|
|||
Loading…
Reference in New Issue