HhUaProvider.php

gistfile1.txt

<?php

namespace AppBundle\Provider;

use AppBundle\Entity\JobTitle;
use AppBundle\Entity\Provider;
use GuzzleHttp\Client;
use Symfony\Component\DomCrawler\Crawler;
use Symfony\Component\HttpFoundation\Response;

class HHUaProvider extends AbstractProvider
{
    /**
     * @var string
     */
    public $url = 'https://hh.ua/catalog/Informacionnye-tehnologii-Internet-Telekom';

    /**
     * {@inheritdoc}
     */
    public function parse()
    {
        $client = new Client();

        $i = 0;
        $array = [];

        for ($count = 0; $count < 100000; $count++) {

            //var_dump($client->request('GET', $this->url . '/page-' . $count)->getStatusCode());
            if ($client->request('GET', $this->url . '/page-' . $count)->getStatusCode() == Response::HTTP_NOT_FOUND) {
                var_dump('break');
                break;
            }
            $response = $client->request('GET', $this->url . '/page-' . $count);
            $crawler = new Crawler($response->getBody()->getContents());
            $vacations = $crawler->filter('div.search-result-item__head')->children()->first();
            foreach ($vacations as $el) {
                // set all params nullable
                $salary = null;
                $companyName = null;
                $location = null;

                $vacation_url = $el->getAttribute('href');
                //var_dump($vacation_url);

                // check for existing vacation
                if ($this->em->getRepository(JobTitle::class)->findByUrl($vacation_url)) {
                    var_dump(123);
                    continue;
                }
                $vacation_response = $client->request('GET', $vacation_url);
                $vacation_crawler = new Crawler($vacation_response->getBody()->getContents());
                $jobName = $vacation_crawler->filter('h1')->text();

                $subject = $vacation_crawler->filter('div.l-paddings')->getNode(3)->textContent;
                //var_dump((str_replace(' ', '', $subject)));
                var_dump($subject);

                if ($subject == ' з/п не указана') {
                    $salary = null;
                } else {
                    $salary = $subject;
                }
//                if (ctype_alnum((str_replace(' ', '', $subject)))) {
//                    $salary = $vacation_crawler->filter('div.l-paddings')->getNode(3)->textContent;
//                }

                $location = $vacation_crawler->filter('div.l-paddings')->getNode(4)->textContent;
//
//                $c=0;
//                var_dump($otherData);
//                foreach ($otherData as $info) {
//                    $c++;
//                    if ($c++ == 3) {
//                        $salary = $info->textContent;
//                        var_dump($salary);
//                    }
//                    if ($c == 4) {
//                        $location = $info->textContent;
//                        //var_dump($location);
//                        break;
//                    }
//                }
                $companyName = $vacation_crawler->filter('div.companyname')->text();

                $array[$i]['company'] = $companyName;
                $array[$i]['jobtitle'] = $jobName;
                $array[$i]['location'] = $location;
                $array[$i]['url'] = $vacation_url;
                $array[$i]['salary'] = $salary;
                $array[$i]['provider'] = Provider::HH_UA_PROVIDER;
                var_dump($array[$i]['salary']);
                $i++;
            }
        }
        $this->import($array);

        return $this;
    }
}