Crawler::crawl() - Code Metrics - pilipinews/abscbn - Measure and Improve Code Quality continuously with Scrutinizer

Crawler::crawl() A
last analyzed 2020-07-13 15:57 UTC

↳ Parent: Crawler

Complexity

Conditions	1
Paths	1

Size

Total Lines	18
Code Lines	8

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	10
CRAP Score	1

Importance

Changes	1
Bugs	0	Features	0

Metric	Value
cc	1
eloc	8
c	1
b	0
f	0
nc	1
nop	0
dl	0
loc	18
ccs	10
cts	10
cp	1
crap	1
rs	10

<?php

namespace Pilipinews\Website\Abscbn;

use Pilipinews\Common\Client;
use Pilipinews\Common\Crawler as DomCrawler;
use Pilipinews\Common\Interfaces\CrawlerInterface;

/**
 * ABS-CBN News Crawler
 *
 * @package Pilipinews
 * @author  Rougin Gutib <[email protected]>
 */
class Crawler implements CrawlerInterface
{
    /**
     * @var string
     */
    protected $link = 'https://news.abs-cbn.com/news';

    /**
     * Returns an array of articles to scrape.
     *
     * @return string[]
     */
    public function crawl()
    {
        $response = Client::request($this->link);

        $callback = function (DomCrawler $node)
        {
            $url = 'https://news.abs-cbn.com';

            return $url . $node->attr('href');
        };

        $crawler = new DomCrawler((string) $response);

        $news = $crawler->filter('#latest-news li > p > a');

        $news = $this->verify($news->each($callback));

        return array_reverse(array_filter((array) $news));
    }

    /**
     * Returns the allowed article URLs to scrape.
     *
     * @param  string[] $items
     * @return string[]
     */
    protected function verify($items)
    {
        $callback = function ($link)
        {
            $news = strpos($link, '.com/news/') !== false;

            $media = strpos($link, 'news/multimedia') === false;

            return $news && $media === true ? $link : null;
        };

        $items = array_map($callback, (array) $items);

        return array_values(array_filter($items));
    }
}


1		<?php
2
3		namespace Pilipinews\Website\Abscbn;
4
5		use Pilipinews\Common\Client;
6		use Pilipinews\Common\Crawler as DomCrawler;
7		use Pilipinews\Common\Interfaces\CrawlerInterface;
8
9		/**
10		* ABS-CBN News Crawler
11		*
12		* @package Pilipinews
13		* @author Rougin Gutib <[email protected]>
14		*/
15		class Crawler implements CrawlerInterface
16		{
17		/**
18		* @var string
19		*/
20		protected $link = 'https://news.abs-cbn.com/news';
21
22		/**
23		* Returns an array of articles to scrape.
24		*
25		* @return string[]
26		*/
27	3	public function crawl()
28		{
29	3	$response = Client::request($this->link);
30
31	3	$callback = function (DomCrawler $node)
32		{
33	3	$url = 'https://news.abs-cbn.com';
34
35	3	return $url . $node->attr('href');
36	3	};
37
38	3	$crawler = new DomCrawler((string) $response);
39
40	3	$news = $crawler->filter('#latest-news li > p > a');
41
42	3	$news = $this->verify($news->each($callback));
43
44	3	return array_reverse(array_filter((array) $news));
45		}
46
47		/**
48		* Returns the allowed article URLs to scrape.
49		*
50		* @param string[] $items
51		* @return string[]
52		*/
53		protected function verify($items)
54		{
55	3	$callback = function ($link)
56		{
57	3	$news = strpos($link, '.com/news/') !== false;
58
59	3	$media = strpos($link, 'news/multimedia') === false;
60
61	3	return $news && $media === true ? $link : null;
62	3	};
63
64		$items = array_map($callback, (array) $items);
65
66		return array_values(array_filter($items));
67		}
68		}
69

pilipinews / abscbn

Crawler::crawl() A last analyzed 2020-07-13 15:57 UTC

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like

Crawler::crawl() A
last analyzed 2020-07-13 15:57 UTC