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

Crawler::items() A
last analyzed 2020-08-04 01:42 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	1
dl	0
loc	18
ccs	10
cts	10
cp	1
crap	1
rs	10

<?php

namespace Pilipinews\Website\Sunstar;

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

/**
 * Sunstar News Crawler
 *
 * @package Pilipinews
 * @author  Rougin Gutib <[email protected]>
 */
class Crawler implements CrawlerInterface
{
    /**
     * @var string[]
     */
    protected $pages = array(
        'https://www.sunstar.com.ph/morearticles/Manila/Local-news',
        'https://www.sunstar.com.ph/morearticles/Cebu/Local-news',
        'https://www.sunstar.com.ph/morearticles/Davao/Local-news',
    );

    /**
     * Returns an array of articles to scrape.
     *
     * @return string[]
     */
    public function crawl()
    {
        $result = array();

        foreach ((array) $this->pages as $page)
        {
            $items = $this->items((string) $page);

            $result = array_merge($result, $items);
        }

        return $result;
    }

    /**
     * Returns an array of articles to scrape.
     *
     * @param  string $link
     * @return string[]
     */
    protected function items($link)
    {
        $crawler = new DomCrawler(Client::request($link));

        $pattern = '.search-inner > .outer-content';

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

        $items = $news->each(function (DomCrawler $node)
        {
            $pattern = '.inner-content > .title > a';

            $result = $node->filter((string) $pattern);

            return $result->first()->attr('href');
        });

        return array_reverse($items);
    }
}


1		<?php
2
3		namespace Pilipinews\Website\Sunstar;
4
5		use Pilipinews\Common\Crawler as DomCrawler;
6		use Pilipinews\Common\Interfaces\CrawlerInterface;
7
8		/**
9		* Sunstar News Crawler
10		*
11		* @package Pilipinews
12		* @author Rougin Gutib <[email protected]>
13		*/
14		class Crawler implements CrawlerInterface
15		{
16		/**
17		* @var string[]
18		*/
19		protected $pages = array(
20		'https://www.sunstar.com.ph/morearticles/Manila/Local-news',
21		'https://www.sunstar.com.ph/morearticles/Cebu/Local-news',
22		'https://www.sunstar.com.ph/morearticles/Davao/Local-news',
23		);
24
25		/**
26		* Returns an array of articles to scrape.
27		*
28		* @return string[]
29		*/
30	3	public function crawl()
31		{
32	3	$result = array();
33
34	3	foreach ((array) $this->pages as $page)
35		{
36	3	$items = $this->items((string) $page);
37
38	3	$result = array_merge($result, $items);
39	1	}
40
41	3	return $result;
42		}
43
44		/**
45		* Returns an array of articles to scrape.
46		*
47		* @param string $link
48		* @return string[]
49		*/
50	3	protected function items($link)
51		{
52	3	$crawler = new DomCrawler(Client::request($link));
53
54	3	$pattern = '.search-inner > .outer-content';
55
56	3	$news = $crawler->filter((string) $pattern);
57
58	3	$items = $news->each(function (DomCrawler $node)
59		{
60	3	$pattern = '.inner-content > .title > a';
61
62	3	$result = $node->filter((string) $pattern);
63
64	3	return $result->first()->attr('href');
65	3	});
66
67	1	return array_reverse($items);
68		}
69		}
70

pilipinews / sunstar

Crawler::items() A last analyzed 2020-08-04 01:42 UTC

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like

Crawler::items() A
last analyzed 2020-08-04 01:42 UTC