rOpenDev\Google\ExtractResults

Complexity

Total Complexity

9

Size/Duplication

Total Lines	55
Duplicated Lines	0 %

Importance

Changes

0

5 Methods

Rating	Name	Duplication	Size	Complexity
A	getSelector()	0	7	2
A	getUrlFromGoogleSerpFromat()	0	6	2
A	getOrganicResults()	0	16	3
A	__construct()	0	4	1
A	normalizeTextFromGoogle()	0	3	1

<?php

namespace rOpenDev\Google;

class ExtractResults
{
    /**
     * @var array from \rOpenDev\Google\ResultTypes.php
     */
    protected $types;
    /**
     * @var object dom
     */
    protected $html;

    public function __construct(string $source)
    {
        $this->html = new \simple_html_dom();
        $this->html->load($source);
    }

    public function getOrganicResults()
    {
        $results = $this->html->find(str_replace(';', ', ', $this->getSelector('organic')));
        $result = [];
        foreach ($results as $r) {
            $title = $r->find('h3, [role=heading]', 0);
            if ($title) {
                $result[] = [
                    'type' => 'organic',
                    'title' => $this->normalizeTextFromGoogle($title->innertext),
                    'link' => $this->getUrlFromGoogleSerpFromat($r->find('a', 0)->href),
                ];
            }
        }

        return $result;
    }

    protected function getSelector(string $type)
    {
        if (null === $this->types) {
            $this->types = ResultsTypes::get();
        }

        return $this->types[$type];
    }

    protected static function getUrlFromGoogleSerpFromat($str)
    {
        preg_match('/\/url\?.*(q|url)=(http.+)&/SiU', $str, $m1);
        $str = isset($m1[2]) ? $m1[2] : $str;

        return $str;
    }

    protected function normalizeTextFromGoogle($text)
    {
        return htmlspecialchars_decode(html_entity_decode(strip_tags($text)));
    }
}