Total Complexity | 5 |
Total Lines | 36 |
Duplicated Lines | 0 % |
Changes | 0 |
1 | <?php declare(strict_types=1); |
||
13 | class Helper { |
||
14 | /** |
||
15 | * @todo Re-factor result into class |
||
16 | * |
||
17 | * @param string $urlToCrawl |
||
18 | * |
||
19 | * @return object |
||
20 | */ |
||
21 | public static function getCleanedUrl($urlToCrawl) { |
||
22 | $parts = parse_url($urlToCrawl); |
||
23 | |||
24 | if ($parts === false) { |
||
25 | throw new MalformedURLException($urlToCrawl . ' - is a malformed URL and cannot be processed'); |
||
26 | } |
||
27 | |||
28 | $prefix = isset($parts['query']) && $parts['query'] ? '&' : '?'; |
||
29 | |||
30 | $finalUrl = str_replace('#!', $prefix . '_escaped_fragment_=', $urlToCrawl); |
||
31 | |||
32 | return (object)[ |
||
33 | 'url' => $urlToCrawl, |
||
34 | 'parts' => (object)$parts, |
||
35 | 'linkhash' => md5($urlToCrawl), |
||
36 | 'finalUrl' => $finalUrl, |
||
37 | ]; |
||
38 | } |
||
39 | |||
40 | /** |
||
41 | * @param string $text |
||
42 | * |
||
43 | * @return string |
||
44 | */ |
||
45 | public static function textNormalise($text) { |
||
49 | } |
||
50 | } |
||
51 |