Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
100.00% |
14 / 14 |
|
100.00% |
2 / 2 |
CRAP | |
100.00% |
1 / 1 |
| WikiLinkPrefixer | |
100.00% |
14 / 14 |
|
100.00% |
2 / 2 |
9 | |
100.00% |
1 / 1 |
| __construct | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| process | |
100.00% |
12 / 12 |
|
100.00% |
1 / 1 |
8 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace FileImporter\Services\Wikitext; |
| 4 | |
| 5 | use MediaWiki\Title\MalformedTitleException; |
| 6 | use MediaWiki\Title\TitleParser; |
| 7 | |
| 8 | /** |
| 9 | * A small parser that adds an interwiki prefix to all links it can't understand. Incoming links are |
| 10 | * expected to have canonical English namespace names, or namespace names in the target wikis |
| 11 | * content language. |
| 12 | * |
| 13 | * This class uses MediaWiki's TitleParser to extract known interwiki prefixes and namespaces from |
| 14 | * the link, but doesn't use it to do any normalization. |
| 15 | * |
| 16 | * @license GPL-2.0-or-later |
| 17 | */ |
| 18 | class WikiLinkPrefixer implements WikiLinkCleaner { |
| 19 | |
| 20 | private string $interWikiPrefix; |
| 21 | private TitleParser $parser; |
| 22 | |
| 23 | /** |
| 24 | * @param string $interWikiPrefix E.g. "de" for the German Wikipedia. |
| 25 | * @param TitleParser $parser |
| 26 | */ |
| 27 | public function __construct( string $interWikiPrefix, TitleParser $parser ) { |
| 28 | $this->interWikiPrefix = $interWikiPrefix; |
| 29 | $this->parser = $parser; |
| 30 | } |
| 31 | |
| 32 | public function process( string $link ): string { |
| 33 | if ( $this->interWikiPrefix === '' |
| 34 | // Bail out early if the prefix is already there; the extra + avoid backtracking |
| 35 | || preg_match( '{^\h*+:?\h*+' . preg_quote( $this->interWikiPrefix ) . '\h*+:}iu', $link ) |
| 36 | ) { |
| 37 | return $link; |
| 38 | } |
| 39 | |
| 40 | try { |
| 41 | $title = $this->parser->parseTitle( $link ); |
| 42 | } catch ( MalformedTitleException $ex ) { |
| 43 | return $link; |
| 44 | } |
| 45 | |
| 46 | if ( $title->inNamespace( NS_CATEGORY ) |
| 47 | // The syntax for thumbnail images would break with a prefix |
| 48 | || $title->inNamespace( NS_FILE ) |
| 49 | || $title->inNamespace( NS_MEDIA ) |
| 50 | || strcasecmp( $title->getInterwiki(), $this->interWikiPrefix ) === 0 |
| 51 | ) { |
| 52 | return $link; |
| 53 | } |
| 54 | |
| 55 | return ':' . $this->interWikiPrefix . ':' . ltrim( ltrim( $link ), ':' ); |
| 56 | } |
| 57 | |
| 58 | } |