README.md

crawler.newride.tech

How to start crawling?

To start crawling and extract data from sites you need to create a data extractor that is capable of handling given resource and plug it into the crawler engine. Then you can read incoming data from dataBus.

code your data extractor

This extractor reads h1 title from example.com and pushes it into the dataBus as 'website header'.

const Extractor = require('crawler.newride.tech/Extractor');

const MyExtractor extends Extractor {
  canCrawlUrl(url) {
    return Promise.resolve(/example.com\/index.html/.test(url));
  }

  extractFromUrl(urlListDuplexStream, dataBus, url) {
    return new Nightmare()
      .goto(url)
      .evaluate(() => document.querySelector('h1').textContent)
      .end()
      .then(textContent => dataBus.pushData('website header', textContent))
    ;
  }
}

module.exports = MyExtractor;

setup crawler engine

To work, crawler needs UrlListDuplexStream, DataBus and ExtractorSet. Those are basic and necessary objects for the mechanism to work and each one of them plays its own role in the crawling process.

why not just one Crawler object?

Using several classess, their concerns are separated and can operate independently. Only combined together they make a Crawler that is able to extract data from various sources efficiently.

what is the exact role of each of those objects?

Crawler - orchestrates and manages DataBus, ExtractorSet and UrlListDuplexStream to work together. It's the only object able to keep everyting together and start the crawling process.

DataBus - its sole purpose is to transfer data. Since all crawling happens asynchronously you can never know when the desired data arrives. DataBus will notify you at the exact moment any extractor finds the information you are waiting for. For DataBus by design there is no difference how the data is obtained and when it's published. Thanks to that it does not enforce any implementation on Extractors.

ExtractorSet - each Extractor must specify what kind of resource it can handle. ExtractorSet keeps all the Extractors together and when asked it points the Extractors that can handle the given resource (website for example).

UrlListDuplexStream - keeps track of the links that are going to be crawled. If you start crawling with empty UrlListDuplexStream then Crawler will stop immediately since it has nothing to do. UrlListDuplexStream is a FIFO queueu that feeds links to the Crawler until it's empty. You can push more links into UrlListDuplexStream while crawling. Thanks to that you can crawl resources list using one extractor and then push the links to individual resources into UrlListDuplexStream so other Extractors can crawl them. Crawling stops when UrlListDuplexStream is empty.

how to implement this?

const Crawler = require('crawler.newride.tech/Crawler');
const DataBus = require('crawler.newride.tech/EventEmitter/DataBus');
const ExtractorSet = require('crawler.newride.tech/ExtractorSet');
const UrlListDuplexStream = require('crawler.newride.tech/UrlListDuplexStream');

// your extractor
const MyExtractor = require('./MyExtractor');

const dataBus = new DataBus();

dataBus.addListener('website header', datagram => {
    console.log(datagram.data); // 'Example Domain'
    datagram.resolve();
});

const crawler = new Crawler(dataBus, new ExtractorSet([
    new MyExtractor()
]));

const urlListDuplexStream = new UrlListDuplexStream();

urlListDuplexStream.feed([
    'http://example.com'
]);

crawler.run(urlListDuplexStream);