Scrap a webpage with Infinite scrolling using phantomjs

phantonjs-infinite-scroll.js

const phantom = require('phantom');

async function wait(timeInMills) {
    return new Promise((resolve) => {
        setTimeout(() => {
            resolve();
        }, timeInMills);
    });
}

// Scrolls the page till new content is available
async function scrollPage(page) {
    const currentContentLength = (await page.property('content')).length;
    await page.evaluate(function () {
        window.document.body.scrollTop = document.body.scrollHeight;
    });
    await wait(Math.max(5000, 10000 * Math.random()));
    const nextContentLength = (await page.property('content')).length;
    if (currentContentLength != nextContentLength) {
        console.log("Scrolling page:", await page.property('url'), "for more content");
        await scrollPage(page);
    }
}

// Scrolls the page and gets the page content using PhantomJS
async function getPageData(pageUrl, shouldScrollPage) {
    const instance = await phantom.create();
    const page = await instance.createPage();
    await page.open(pageUrl);
    if (shouldScrollPage) {
        await scrollPage(page);
    }
    const pageContent = await page.property('content');
    await page.close();
    await instance.exit();
    return pageContent;
};