Pilipinews\Website\Rappler\Crawler::crawl()

Complexity

Conditions	4
Paths	1

Size

Total Lines	33
Code Lines	15

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	13
CRAP Score	4

Importance

Changes	2
Bugs	0	Features	0

<?php
    public function crawl()
    {
        $base = 'https://rappler.com';

        $excluded = $this->excluded;

        $excluded = function ($text) use ($excluded)
        {
            preg_match('/(.*):(.*)/i', $text, $matches);

            $keyword = isset($matches[1]) ? $matches[1] : null;

            return in_array($keyword, (array) $excluded);
        };

        $callback = function (DomCrawler $node) use ($base, $excluded)
        {
            $items = explode('/', $link = $node->attr('href'));

            $allowed = $items[1] === 'nation' && ! $excluded($node->text());

            return $allowed ? $base . $node->attr('href') : null;
        };

        $crawler = new DomCrawler(Client::request($this->link));

        $news = $crawler->filter((string) $this->pattern);

        $filtered = array_filter($news->each($callback));

        $reversed = array_reverse($filtered);

The assignment to $reversed is dead and can be removed.

        return array_values(array_unique($filtered));
    }