Total Complexity | 4 |
Total Lines | 57 |
Duplicated Lines | 0 % |
Coverage | 100% |
Changes | 1 | ||
Bugs | 0 | Features | 0 |
1 | <?php |
||
15 | class Crawler implements CrawlerInterface |
||
16 | { |
||
17 | /** |
||
18 | * @var string[] |
||
19 | */ |
||
20 | protected $allowed = array('news/nation', 'news/regions'); |
||
21 | |||
22 | /** |
||
23 | * @var string |
||
24 | */ |
||
25 | protected $link = 'https://data.gmanetwork.com/gno/pages/home_1a_json.gz'; |
||
26 | |||
27 | /** |
||
28 | * Returns an array of articles to scrape. |
||
29 | * |
||
30 | * @return string[] |
||
31 | */ |
||
32 | 3 | public function crawl() |
|
33 | { |
||
34 | 3 | $response = Client::request($this->link); |
|
35 | |||
36 | 3 | $json = json_decode($response, true); |
|
37 | |||
38 | 3 | $items = $json['story_lists_just_in']; |
|
39 | |||
40 | 3 | return $this->verify((array) $items); |
|
41 | } |
||
42 | |||
43 | /** |
||
44 | * Returns the allowed article URLs to scrape. |
||
45 | * |
||
46 | * @param string[] $items |
||
47 | * @return string[] |
||
48 | */ |
||
49 | 3 | protected function verify($items) |
|
72 | } |
||
73 | } |
||
74 |