|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Pilipinews\Website\Pna; |
|
4
|
|
|
|
|
5
|
|
|
use Pilipinews\Common\Client; |
|
6
|
|
|
use Pilipinews\Common\Crawler as DomCrawler; |
|
7
|
|
|
use Pilipinews\Common\Interfaces\CrawlerInterface; |
|
8
|
|
|
|
|
9
|
|
|
/** |
|
10
|
|
|
* Philippine News Agency 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 $categories = array( |
|
21
|
|
|
'https://www.pna.gov.ph/categories/national', |
|
22
|
|
|
'https://www.pna.gov.ph/categories/provincial', |
|
23
|
|
|
); |
|
24
|
|
|
|
|
25
|
|
|
/** |
|
26
|
|
|
* Returns an array of articles to scrape. |
|
27
|
|
|
* |
|
28
|
|
|
* @return string[] |
|
29
|
|
|
*/ |
|
30
|
3 |
|
public function crawl() |
|
31
|
|
|
{ |
|
32
|
3 |
|
list($articles, $result) = array(array(), array()); |
|
33
|
|
|
|
|
34
|
3 |
|
foreach ($this->categories as $category) |
|
35
|
|
|
{ |
|
36
|
3 |
|
$result[] = $this->items($category); |
|
37
|
|
|
} |
|
38
|
|
|
|
|
39
|
3 |
|
foreach ($result[0] as $key => $item) |
|
40
|
|
|
{ |
|
41
|
3 |
|
$articles[] = $result[0][$key]; |
|
42
|
3 |
|
$articles[] = $result[1][$key]; |
|
43
|
|
|
} |
|
44
|
|
|
|
|
45
|
3 |
|
return array_reverse((array) $articles); |
|
46
|
|
|
} |
|
47
|
|
|
|
|
48
|
|
|
/** |
|
49
|
|
|
* Returns an array of articles to scrape. |
|
50
|
|
|
* |
|
51
|
|
|
* @return string[] |
|
52
|
|
|
*/ |
|
53
|
3 |
|
protected function items($link) |
|
54
|
|
|
{ |
|
55
|
3 |
|
$pattern = '.articles > .article.media'; |
|
56
|
|
|
|
|
57
|
3 |
|
$base = 'https://www.pna.gov.ph'; |
|
58
|
|
|
|
|
59
|
3 |
|
$callback = function (DomCrawler $node) use ($base) |
|
60
|
|
|
{ |
|
61
|
3 |
|
$link = $node->filter('.media-heading > a'); |
|
62
|
|
|
|
|
63
|
3 |
|
return $base . (string) $link->attr('href'); |
|
64
|
3 |
|
}; |
|
65
|
|
|
|
|
66
|
|
|
$crawler = new DomCrawler(Client::request($link)); |
|
67
|
|
|
|
|
68
|
|
|
return $crawler->filter($pattern)->each($callback); |
|
69
|
|
|
} |
|
70
|
|
|
} |
|
71
|
|
|
|
pilipinews /
pna
