Convert multiple URLs into PDF using puppeteer

puppeteer.js

const puppeteer = require('puppeteer'); // [email protected]

const urls = [
  'https://en.wikipedia.org/wiki/TypeScript',
  'https://en.wikipedia.org/wiki/JavaScript',
];

const urlCount = urls.length;

const delay = 5000; // Delay between each save

(async () => {
  const browser = await puppeteer.launch();

  const page = await browser.newPage();

  // Set multiple cookies (domain required)
  await page.setCookie(...[
    {
      name: 'name',
      value: 'value',
      domain: '.wikipedia.org',
    },
  ]);

  const savePageAsPDF = async (url) => {
    await page.goto(url, { waitUntil: 'networkidle2' });

    await page.evaluate(() => {
      // Remove HTML elements
      [
        'header',
        'progress',
        'footer',
      ].forEach((selector) => {
        const element = document.querySelector(selector);
        if (element) element.parentNode.removeChild(element);
      });
    });

    const title = await page.title(); // Website title

    // Save as PDF
    await page.pdf({
      path: `${title}.pdf`,
      format: 'a4',
      margin: { top: '1.27cm', left: '1.27cm', right: '1.27cm', bottom: '1.27cm' },
    });

    return title;
  };

  for (const [index, url] of urls.entries()) {
    console.log(`Saving ${index + 1} / ${urlCount}`);
    const title = await savePageAsPDF(url);
    console.log(`Saved '${title}'`);
    if (index + 1 !== urlCount) {
      console.log(`Waiting for ${delay}ms`);
      await new Promise((resolve) => setTimeout(resolve, delay));
    } else {
      console.log('Batch completed');
    }
  }

  await browser.close();
})();