Update scraper.js

scraper.js

@@ -15,11 +15,11 @@ 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 to load ~40-60 pins
+        // --- SCROLLING LOGIC TO FIND MORE PINS ---
         await page.evaluate(async () => {
-            for (let i = 0; i < 2; i++) {
+            for (let i = 0; i < 3; i++) { // Scroll 3 times to load ~60 pins
                 window.scrollBy(0, window.innerHeight * 2);
-                await new Promise(r => setTimeout(r, 1000));
+                await new Promise(r => setTimeout(r, 1500));
             }
         });
 
@@ -28,44 +28,33 @@ async function getPinterestVideos(query) {
                 .map(a => a.href).filter(href => href.includes('/pin/'));
         });
 
-        const uniquePins = [...new Set(pinUrls)].slice(0, 40);
+        // Get up to 60 unique pins to check
+        const uniquePins = [...new Set(pinUrls)].slice(0, 60);
+        console.error(`Found ${uniquePins.length} potential pins. Checking for video files...`);
 
         for (const pinUrl of uniquePins) {
             const pPage = await browser.newPage();
             try {
+                // We go to each pin and try to find the video source
                 await pPage.goto(pinUrl, { waitUntil: 'domcontentloaded', timeout: 10000 });
-                
-                // Extract high-quality metadata from the __PWS_DATA__ script tag
-                const videoData = await pPage.evaluate(() => {
-                    const script = document.getElementById('__PWS_DATA__');
-                    if (!script) return null;
-                    const data = JSON.parse(script.innerText);
-                    const pinId = Object.keys(data.props.initialReduxState.pins)[0];
-                    const pin = data.props.initialReduxState.pins[pinId];
-                    
-                    if (pin && pin.videos && pin.videos.video_list) {
-                        const list = pin.videos.video_list;
-                        // Pick the best MP4 or HLS variant
-                        // Priority: V_720P > V_HLSV4 > others
-                        const best = list.V_720P || list.V_HLSV4 || Object.values(list)[0];
-                        return {
-                            url: best.url,
-                            width: best.width,
-                            height: best.height
-                        };
+                const content = await pPage.content();
+                const videoRegex = /https:\/\/v1\.pinimg\.com\/videos\/mc\/[^\s"']+/g;
+                const matches = content.match(videoRegex);
+
+                if (matches) {
+                    let rawUrl = matches[0].replace(/\\u002F/g, '/').replace(/[奖励"']/g, '');
+                    const hashMatch = rawUrl.match(/([a-f0-9]{32})/);
+                    if (hashMatch) {
+                        const h = hashMatch[1];
+                        // Output the 720p URL for Python to check
+                        process.stdout.write(`https://v1.pinimg.com/videos/mc/720p/${h.substring(0,2)}/${h.substring(2,4)}/${h.substring(4,6)}/${h}.mp4\n`);
                     }
-                    return null;
-                });
-
-                if (videoData) {
-                    // Output format: URL | WIDTH | HEIGHT
-                    process.stdout.write(`${videoData.url}|${videoData.width}|${videoData.height}\n`);
                 }
-            } catch (e) {}
+            } catch (e) {} 
             await pPage.close();
         }
     } catch (error) {
-        console.error(error);
+        console.error("Scraper Error: " + error.message);
     } finally {
         if (browser) await browser.close();
     }