Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
100.00% |
9 / 9 |
|
100.00% |
5 / 5 |
CRAP | |
100.00% |
1 / 1 |
| SourceUrl | |
100.00% |
9 / 9 |
|
100.00% |
5 / 5 |
6 | |
100.00% |
1 / 1 |
| __construct | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
2 | |||
| getUrl | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getParsedUrl | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getHost | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| __toString | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace FileImporter\Data; |
| 4 | |
| 5 | use FileImporter\Exceptions\InvalidArgumentException; |
| 6 | use MediaWiki\MediaWikiServices; |
| 7 | |
| 8 | /** |
| 9 | * Class representing a (possibly unnormalized) URL passed into the extension for importing. Can be |
| 10 | * any URL (e.g. a Flickr image), not necessarily pointing to a MediaWiki. |
| 11 | * |
| 12 | * Basic normalization that is true for any URL can be done here (e.g. Unicode normalization). |
| 13 | * Normalizations for specific sources (as detected by dedicated SourceUrlCheckers) need to be done |
| 14 | * in dedicated SourceUrlNormalizers. |
| 15 | * |
| 16 | * @license GPL-2.0-or-later |
| 17 | * @author Addshore |
| 18 | */ |
| 19 | class SourceUrl { |
| 20 | |
| 21 | private const ERROR_SOURCE_URL_UNPARSEABLE = 'sourceUrlUnparseable'; |
| 22 | |
| 23 | private string $url; |
| 24 | /** @var string[] */ |
| 25 | private $parsed; |
| 26 | |
| 27 | /** |
| 28 | * @param string $url For example, https://en.wikipedia.org/wiki/File:Foo.JPG |
| 29 | * |
| 30 | * @throws InvalidArgumentException When $url is not parsable |
| 31 | */ |
| 32 | public function __construct( string $url ) { |
| 33 | $this->url = trim( $url ); |
| 34 | $this->parsed = MediaWikiServices::getInstance()->getUrlUtils()->parse( $this->url ) ?? []; |
| 35 | if ( !$this->parsed ) { |
| 36 | throw new InvalidArgumentException( '$url is not parsable', |
| 37 | self::ERROR_SOURCE_URL_UNPARSEABLE ); |
| 38 | } |
| 39 | } |
| 40 | |
| 41 | /** |
| 42 | * @return string The raw URL |
| 43 | */ |
| 44 | public function getUrl(): string { |
| 45 | return $this->url; |
| 46 | } |
| 47 | |
| 48 | /** |
| 49 | * @return string[] Parsed URL array provided by wfParseUrl |
| 50 | */ |
| 51 | public function getParsedUrl(): array { |
| 52 | return $this->parsed; |
| 53 | } |
| 54 | |
| 55 | /** |
| 56 | * @return string The host, for example "en.wikipedia.org" |
| 57 | */ |
| 58 | public function getHost(): string { |
| 59 | return strtolower( $this->parsed['host'] ); |
| 60 | } |
| 61 | |
| 62 | public function __toString(): string { |
| 63 | return $this->url; |
| 64 | } |
| 65 | |
| 66 | } |