jobscraper/scrapers/rss.technojobs.js

/**
 * Created by WebStorm.
 * User: martin
 * Date: 24/05/2020
 * Time: 23:04

 */

const MasterRSS = require('../lib/rss');

class RssTechnojobs extends MasterRSS {
  constructor() {
    super();

    this.siteurl = 'www.technojobs.co.uk';
    this.siteid = 'technojobs';
    this.useStone = false;
    this.requestOptions = {
      'url'    : ''
    };

    this.guidRegex = /\/(\d+)/;
    this.imgRegex = /src="(.+?)"/g;
    this.locationRegEx = /Location:(.*?)?</;
    this.companyRegEx = /Advertiser:(?:&nbsp;)?(.*?)(?:Reference|Start|Email)/;
    this.rateRegEx = /Salary\/Rate:(.*?)?</;
  }

  reduceItem(item) {
    const newObj = this.newRecord();

    /*
    console.log('++ reduce');
    console.log(item);
*/

    const location = this.locationRegEx.exec(item.content);
    const rate = this.rateRegEx.exec(item.content);
    const id = this.guidRegex.exec(item.link);

    const _content = item.content.replace(this.imgRegex, (full, part) => {
      return `src="${this.makeImg(part)}"`;
    });

    if (location !== null) {
      const _location = location[1].trim();
      if (_location.length <= 30)
        newObj.location = _location;
    }

    if (rate !== null)
      newObj.salary = rate[1].trim().slice(0, 55);

    if (id !== null)
      newObj.id = id[1].trim();

    newObj.title = item.title;
    newObj.postDate = item.isoDate;
    newObj.url = item.link;
    newObj.summary = _content;

    /*
     console.log(newObj);
    console.log('-- reduce');
*/

    return newObj;
  }

  async go(url) {
    await super.go();
    this.setStartUrl(url);

    await this.processFeed();
  }
}

module.exports = RssTechnojobs;