Html-data-scraper NPM

An efficient wrapper around puppeteer for data scraping web pages.

Install

yarn add html-data-scraper

npm install html-data-scraper

API

htmlDataScraper(urls, configurations, customBrowser)

urls: string[] An array of urls
configurations?: CustomConfigurations An CustomConfigurations object to configure everything.
customBrowser?: Browser An instance of Browser created outside the library, this instance will also be given back in browserInstance.
returns Promise<{results:PageResult[], browserInstance: Browser}> Promise which resolves an array of PageResult objects and the Browser instance used during the process.

This main function will distribute the scraping process regarding all minus one cpu cores number available ( if the computer have 4 cores, it will distribute on 3 cores ). The distribution mean opening a page for each available core.

Usage

import htmlDataScraper, {PageResult} from 'html-data-scraper';    

const progress: Record<string, string[]> = {};

const urls = [];
const urlNumber = 7;
const maxSimultaneousBrowser = 3;

for (let i = 0; i < urlNumber; i++) {
    urls.push('https://fr.wikipedia.org/wiki/World_Wide_Web');
}

htmlDataScraper(urls, {
    maxSimultaneousBrowser,
    onEvaluateForEachUrl: {
        title: (): string => {
            const titleElement: HTMLElement | null = document.getElementById('firstHeading');
            const innerElement: HTMLCollectionOf<HTMLElement> | null = titleElement.getElementsByTagName('span');
            return innerElement && innerElement.length ? innerElement[0].innerText : '';
        },
    },
    onProgress: (resultNumber: number, totalNumber: number, internalPageIndex: number) => {
        console.log('Scraping page n°',internalPageIndex, '->' , resultNumber + '/' + totalNumber);
    },
})
    .then(({results}: {results:PageResult[], browserInstance: Browser}) => {

        console.log(results);
        // [
        //    {
        //        pageData: '<!DOCTYPE html>.....',
        //        evaluates: { 
        //            title: 'World Wide Web'
        //        }
        //    },
        //    ...
        // ]


    });

CustomConfigurations

This object is use to setup the scraping process and puppeteer itself. The following object show the default values :

const configurations = {
    maxSimultaneousBrowser  : 1,
    additionalWaitSeconds   : 1,
    puppeteerOptions        : {
        browser : {
            args : [
                '--no-sandbox',
                '--disable-setuid-sandbox',
                '--disable-dev-shm-usage',
            ],
        },
        pageGoTo : { waitUntil: 'networkidle2' },
    },
}

if maxSimultaneousBrowser is not set during the initialization, it will use all available core minus one :

if (!customConfigurations.hasOwnProperty('maxSimultaneousBrowser')){
    const cpuCoreCount = os.cpus().length;
    configuration.maxSimultaneousBrowser = cpuCoreCount > 2 ? cpuCoreCount - 1 : 1;
}

Additionally you can use the following keys:

onPageRequest

onPageRequest: (request) => void

request: Request Represents a page HTTPRequest.

Whenever the page sends a request, such as for a network resource, the following function is triggered.

onPageLoadedForEachUrl

onPageLoadedForEachUrl: (puppeteerPage, currentUrl) => {}

puppeteerPage: Page A reference to the current puppeteer Page.
currentUrl: string The current url of the page.
return any You can return whatever you need.

The returned value is set in PageResult.pageData for each url. If this function is not set, by default PageResult.pageData is set with page.content().

onEvaluateForEachUrl

onEvaluateForEachUrl: {}

This key contain an object that register functions. Those functions are use by page.evaluate. Each function return is set to the corresponding name in PageResult.evaluates :

import htmlDataScraper, {PageResult} from 'html-data-scraper';    

htmlDataScraper([
    'https://www.bbc.com',
], {
    onEvaluateForEachUrl: {
        title: (): string => {
            const titleElement: HTMLElement | null = document.getElementById('page-title');
            return titleElement ? titleElement.innerText : '';
        },
    },
})
.then((results: {results:PageResult[], browserInstance: Browser}) => {
    console.log(results[0]);
    // {
    //     pageData: "....",
    //     evaluates: {
    //         title: "..." 
    //     } 
    // }
});

onProgress

onProgress: (resultNumber, totalNumber, internalPageIndex) => {}:

resultNumber: number The number of processed urls.
totalNumber: number The total number of urls to process.
internalPageIndex: number An index representing the page that process urls.
return void No return needed.

This function is run each time a page processing finish.

import htmlDataScraper from 'html-data-scraper';    

const progress: Record<string,string[]> = {};

htmlDataScraper([
    // ...
], {
    onProgress: (resultNumber: number, totalNumber: number, internalPageIndex: number) => {
        const status = resultNumber + '/' + totalNumber;
        if (progress[internalPageIndex]){
            progress[internalPageIndex].push(status);
        } else {
            progress[internalPageIndex] = [status];
        }
    },
})

PageResult

interface PageResult{
    pageData: any;
    evaluates: null | {
        [k: string]: any;
    };
}

pageData

This key will contain the html content of the webpage. But if you set the onPageLoadedForEachUrl, this will contain de returned value on the function.