Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
30.23% |
13 / 43 |
|
33.33% |
1 / 3 |
CRAP | |
0.00% |
0 / 1 |
| ArticleCompileSnippet | |
30.23% |
13 / 43 |
|
33.33% |
1 / 3 |
128.03 | |
0.00% |
0 / 1 |
| compile | |
0.00% |
0 / 8 |
|
0.00% |
0 / 1 |
42 | |||
| generateArticleSnippet | |
0.00% |
0 / 22 |
|
0.00% |
0 / 1 |
20 | |||
| hasReferenceTag | |
100.00% |
13 / 13 |
|
100.00% |
1 / 1 |
8 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace MediaWiki\Extension\PageTriage\ArticleCompile; |
| 4 | |
| 5 | use MediaWiki\Content\TextContent; |
| 6 | use MediaWiki\MediaWikiServices; |
| 7 | use MediaWiki\Parser\Sanitizer; |
| 8 | use MediaWiki\Title\Title; |
| 9 | |
| 10 | /** |
| 11 | * Article snippet |
| 12 | */ |
| 13 | class ArticleCompileSnippet extends ArticleCompile { |
| 14 | |
| 15 | public function compile() { |
| 16 | foreach ( $this->mPageId as $pageId ) { |
| 17 | $content = $this->getContentByPageId( $pageId ); |
| 18 | if ( $content ) { |
| 19 | $text = ( $content instanceof TextContent ) ? $content->getText() : null; |
| 20 | if ( $text !== null ) { |
| 21 | $this->metadata[$pageId]['snippet'] = self::generateArticleSnippet( $text, $pageId ); |
| 22 | // Reference tag (and other tags) have text strings as the value. |
| 23 | $this->metadata[$pageId]['reference'] = $this->hasReferenceTag( $text ) ? '1' : '0'; |
| 24 | } |
| 25 | } |
| 26 | } |
| 27 | return true; |
| 28 | } |
| 29 | |
| 30 | /** |
| 31 | * Generate article snippet for listview from article text |
| 32 | * @param string $text page text |
| 33 | * @param int $pageId Id of the page |
| 34 | * @return string |
| 35 | */ |
| 36 | public static function generateArticleSnippet( string $text, int $pageId ) { |
| 37 | $text = strip_tags( $text ); |
| 38 | $attempt = 0; |
| 39 | $lang = MediaWikiServices::getInstance()->getContentLanguage(); |
| 40 | $title = Title::newFromID( $pageId ); |
| 41 | |
| 42 | // 10 attempts at most, the logic here is to find the first }} and |
| 43 | // find the matching {{ for that }} |
| 44 | while ( $attempt < 10 ) { |
| 45 | $closeCurPos = strpos( $text, '}}' ); |
| 46 | |
| 47 | if ( $closeCurPos === false ) { |
| 48 | break; |
| 49 | } |
| 50 | $tempStr = substr( $text, 0, $closeCurPos + 2 ); |
| 51 | |
| 52 | $openCurPos = strrpos( $tempStr, '{{' ); |
| 53 | if ( $openCurPos === false ) { |
| 54 | $text = substr_replace( $text, '', $closeCurPos, 2 ); |
| 55 | } else { |
| 56 | $text = substr_replace( $text, '', $openCurPos, $closeCurPos - $openCurPos + 2 ); |
| 57 | } |
| 58 | $attempt++; |
| 59 | } |
| 60 | |
| 61 | $text = trim( Sanitizer::stripAllTags( |
| 62 | MediaWikiServices::getInstance()->getMessageCache()->parse( $text, $title )->getText( [ |
| 63 | 'enableSectionEditLinks' => false, |
| 64 | 'userLang' => $lang |
| 65 | ] ) |
| 66 | ) ); |
| 67 | // strip out non-useful data for snippet |
| 68 | $text = str_replace( [ '{', '}', '[edit]' ], '', $text ); |
| 69 | |
| 70 | return $lang->truncateForDatabase( $text, 255 ); |
| 71 | } |
| 72 | |
| 73 | /** |
| 74 | * Check if a page has a reference. This checks <ref> and </ref> |
| 75 | * tags along with the presence of the sfn or harvb templates |
| 76 | * |
| 77 | * @param string $wikitext Article content in wikitext format. |
| 78 | * @return bool |
| 79 | */ |
| 80 | private function hasReferenceTag( string $wikitext ): bool { |
| 81 | $closeTag = stripos( $wikitext, '</ref>' ); |
| 82 | |
| 83 | if ( $closeTag !== false ) { |
| 84 | $openTag = stripos( $wikitext, '<ref ' ); |
| 85 | if ( $openTag !== false && $openTag < $closeTag ) { |
| 86 | return true; |
| 87 | } |
| 88 | $openTag = stripos( $wikitext, '<ref>' ); |
| 89 | if ( $openTag !== false && $openTag < $closeTag ) { |
| 90 | return true; |
| 91 | } |
| 92 | } |
| 93 | |
| 94 | $refStyleArray = [ '{{sfn', '{{harvnb' ]; |
| 95 | foreach ( $refStyleArray as $refStyle ) { |
| 96 | if ( stripos( $wikitext, $refStyle ) !== false ) { |
| 97 | return true; |
| 98 | } |
| 99 | } |
| 100 | |
| 101 | return false; |
| 102 | } |
| 103 | |
| 104 | } |