Update scraper.js

scraper.js

@@ -15,34 +15,36 @@ async function getPinterestVideos(query) {
         const searchUrl = `https://www.pinterest.com/search/pins/?q=${encodeURIComponent(query)}&rs=typed`;
         await page.goto(searchUrl, { waitUntil: 'networkidle2', timeout: 30000 });
 
-        // Scroll a bit
         await page.evaluate(async () => {
-            window.scrollBy(0, 1000);
-            await new Promise(r => setTimeout(r, 1000));
+            for (let i = 0; i < 2; i++) {
+                window.scrollBy(0, window.innerHeight * 2);
+                await new Promise(r => setTimeout(r, 1000));
+            }
         });
 
-        const pinUrls = await page.evaluate(() => {
-            return Array.from(document.querySelectorAll('a[href*="/pin/"]'))
-                .map(a => a.href).filter(href => href.includes('/pin/'));
+        const pins = await page.evaluate(() => {
+            const results = [];
+            const anchors = document.querySelectorAll('a[href*="/pin/"]');
+            anchors.forEach(a => {
+                const img = a.querySelector('img');
+                results.append({
+                    url: a.href,
+                    title: a.getAttribute('aria-label') || (img ? img.getAttribute('alt') : "")
+                });
+            });
+            return results;
         });
 
-        const uniquePins = [...new Set(pinUrls)].slice(0, 40);
+        const uniquePins = Array.from(new Set(pins.map(p => p.url)))
+            .map(url => pins.find(p => p.url === url)).slice(0, 30);
 
-        for (const pinUrl of uniquePins) {
+        for (const pin of uniquePins) {
             const pPage = await browser.newPage();
             try {
-                await pPage.goto(pinUrl, { waitUntil: 'domcontentloaded', timeout: 10000 });
-                
-                const data = await pPage.evaluate(() => {
-                    // Extract Title and Description
-                    const title = document.querySelector('h1')?.innerText || "";
-                    const desc = document.querySelector('div[data-test-id="main-pin-description-text"]')?.innerText || "";
-                    const html = document.body.innerHTML;
-                    return { title, desc, html };
-                });
-
+                await pPage.goto(pin.url, { waitUntil: 'domcontentloaded', timeout: 8000 });
+                const content = await pPage.content();
                 const videoRegex = /https:\/\/v1\.pinimg\.com\/videos\/mc\/[^\s"']+/g;
-                const matches = data.html.match(videoRegex);
+                const matches = content.match(videoRegex);
 
                 if (matches) {
                     let rawUrl = matches[0].replace(/\\u002F/g, '/').replace(/[奖励"']/g, '');
@@ -50,8 +52,8 @@ async function getPinterestVideos(query) {
                     if (hashMatch) {
                         const h = hashMatch[1];
                         const finalUrl = `https://v1.pinimg.com/videos/mc/720p/${h.substring(0,2)}/${h.substring(2,4)}/${h.substring(4,6)}/${h}.mp4`;
-                        // Output format: URL | TITLE | DESCRIPTION
-                        process.stdout.write(`${finalUrl}|${data.title.replace(/\n/g, ' ')}|${data.desc.replace(/\n/g, ' ')}\n`);
+                        // Output: URL|METADATA for Python to split
+                        process.stdout.write(`${finalUrl}|${pin.title}\n`);
                     }
                 }
             } catch (e) {}