Update scraper.js
Browse files- scraper.js +32 -23
scraper.js
CHANGED
|
@@ -1,6 +1,6 @@
|
|
| 1 |
const puppeteer = require('puppeteer-core');
|
| 2 |
|
| 3 |
-
async function
|
| 4 |
let browser;
|
| 5 |
try {
|
| 6 |
browser = await puppeteer.launch({
|
|
@@ -8,52 +8,61 @@ async function scrapeAll(query) {
|
|
| 8 |
headless: 'new',
|
| 9 |
args: ['--no-sandbox', '--disable-setuid-sandbox', '--disable-dev-shm-usage']
|
| 10 |
});
|
| 11 |
-
|
| 12 |
const page = await browser.newPage();
|
| 13 |
await page.setUserAgent('Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/121.0.0.0 Safari/537.36');
|
| 14 |
|
| 15 |
-
|
| 16 |
-
|
| 17 |
-
|
| 18 |
-
await page.
|
| 19 |
-
|
|
|
|
|
|
|
|
|
|
| 20 |
return Array.from(document.querySelectorAll('a[href*="/pin/"]')).map(a => a.href);
|
| 21 |
});
|
| 22 |
-
const uniquePins = [...new Set(
|
| 23 |
|
| 24 |
-
for (const
|
| 25 |
const pPage = await browser.newPage();
|
| 26 |
try {
|
| 27 |
-
await pPage.goto(
|
| 28 |
const content = await pPage.content();
|
| 29 |
const match = content.match(/https:\/\/v1\.pinimg\.com\/videos\/mc\/[^\s"']+/);
|
| 30 |
if (match) {
|
| 31 |
const hashMatch = match[0].match(/([a-f0-9]{32})/);
|
| 32 |
if (hashMatch) {
|
| 33 |
const h = hashMatch[1];
|
| 34 |
-
|
| 35 |
}
|
| 36 |
}
|
| 37 |
} catch (e) {}
|
| 38 |
await pPage.close();
|
| 39 |
}
|
| 40 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 41 |
|
| 42 |
-
|
| 43 |
-
|
| 44 |
-
|
| 45 |
-
|
| 46 |
-
|
| 47 |
-
return Array.from(document.querySelectorAll('video source')).map(s => s.src);
|
| 48 |
});
|
| 49 |
-
|
| 50 |
-
|
|
|
|
| 51 |
|
| 52 |
-
} catch (
|
| 53 |
-
process.stderr.write(
|
| 54 |
} finally {
|
| 55 |
if (browser) await browser.close();
|
| 56 |
}
|
| 57 |
}
|
| 58 |
|
| 59 |
-
|
|
|
|
|
|
|
|
|
| 1 |
const puppeteer = require('puppeteer-core');
|
| 2 |
|
| 3 |
+
async function scrape(query, mode, orientation) {
|
| 4 |
let browser;
|
| 5 |
try {
|
| 6 |
browser = await puppeteer.launch({
|
|
|
|
| 8 |
headless: 'new',
|
| 9 |
args: ['--no-sandbox', '--disable-setuid-sandbox', '--disable-dev-shm-usage']
|
| 10 |
});
|
|
|
|
| 11 |
const page = await browser.newPage();
|
| 12 |
await page.setUserAgent('Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/121.0.0.0 Safari/537.36');
|
| 13 |
|
| 14 |
+
if (mode === "pinterest") {
|
| 15 |
+
const searchUrl = `https://www.pinterest.com/search/pins/?q=${encodeURIComponent(query)}&rs=typed`;
|
| 16 |
+
await page.goto(searchUrl, { waitUntil: 'networkidle2', timeout: 30000 });
|
| 17 |
+
await page.evaluate(async () => {
|
| 18 |
+
window.scrollBy(0, 1500);
|
| 19 |
+
await new Promise(r => setTimeout(r, 1000));
|
| 20 |
+
});
|
| 21 |
+
const pinUrls = await page.evaluate(() => {
|
| 22 |
return Array.from(document.querySelectorAll('a[href*="/pin/"]')).map(a => a.href);
|
| 23 |
});
|
| 24 |
+
const uniquePins = [...new Set(pinUrls)].slice(0, 30);
|
| 25 |
|
| 26 |
+
for (const pinUrl of uniquePins) {
|
| 27 |
const pPage = await browser.newPage();
|
| 28 |
try {
|
| 29 |
+
await pPage.goto(pinUrl, { waitUntil: 'domcontentloaded', timeout: 10000 });
|
| 30 |
const content = await pPage.content();
|
| 31 |
const match = content.match(/https:\/\/v1\.pinimg\.com\/videos\/mc\/[^\s"']+/);
|
| 32 |
if (match) {
|
| 33 |
const hashMatch = match[0].match(/([a-f0-9]{32})/);
|
| 34 |
if (hashMatch) {
|
| 35 |
const h = hashMatch[1];
|
| 36 |
+
console.log(`https://v1.pinimg.com/videos/mc/720p/${h.substring(0,2)}/${h.substring(2,4)}/${h.substring(4,6)}/${h}.mp4`);
|
| 37 |
}
|
| 38 |
}
|
| 39 |
} catch (e) {}
|
| 40 |
await pPage.close();
|
| 41 |
}
|
| 42 |
+
}
|
| 43 |
+
else if (mode === "pexels") {
|
| 44 |
+
// Pexels supports orientation filter directly in the URL!
|
| 45 |
+
const pexelsOri = orientation.toLowerCase() === "any" ? "" : orientation.toLowerCase();
|
| 46 |
+
const searchUrl = `https://www.pexels.com/search/videos/${encodeURIComponent(query)}/?orientation=${pexelsOri}`;
|
| 47 |
+
await page.goto(searchUrl, { waitUntil: 'networkidle2', timeout: 30000 });
|
| 48 |
|
| 49 |
+
const videoUrls = await page.evaluate(() => {
|
| 50 |
+
// Find all video source tags on Pexels search page
|
| 51 |
+
return Array.from(document.querySelectorAll('video source'))
|
| 52 |
+
.map(s => s.src)
|
| 53 |
+
.filter(src => src.includes('.mp4') || src.includes('video-preview'));
|
|
|
|
| 54 |
});
|
| 55 |
+
|
| 56 |
+
videoUrls.forEach(url => console.log(url));
|
| 57 |
+
}
|
| 58 |
|
| 59 |
+
} catch (error) {
|
| 60 |
+
process.stderr.write(error.message);
|
| 61 |
} finally {
|
| 62 |
if (browser) await browser.close();
|
| 63 |
}
|
| 64 |
}
|
| 65 |
|
| 66 |
+
const args = process.argv.slice(2);
|
| 67 |
+
// args[0] = query, args[1] = mode, args[2] = orientation
|
| 68 |
+
scrape(args[0], args[1] || "pinterest", args[2] || "any");
|