Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
76.79% |
43 / 56 |
|
20.00% |
1 / 5 |
CRAP | |
0.00% |
0 / 1 |
| MoreLikeTrait | |
76.79% |
43 / 56 |
|
20.00% |
1 / 5 |
20.62 | |
0.00% |
0 / 1 |
| doExpand | |
71.43% |
5 / 7 |
|
0.00% |
0 / 1 |
3.21 | |||
| collectTitles | |
75.00% |
3 / 4 |
|
0.00% |
0 / 1 |
2.06 | |||
| collectTitlesFromElastic | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
12 | |||
| collectTitlesFromDB | |
85.71% |
18 / 21 |
|
0.00% |
0 / 1 |
7.14 | |||
| buildMoreLikeQuery | |
100.00% |
17 / 17 |
|
100.00% |
1 / 1 |
2 | |||
| getConfig | n/a |
0 / 0 |
n/a |
0 / 0 |
0 | |||||
| 1 | <?php |
| 2 | |
| 3 | namespace CirrusSearch\Query; |
| 4 | |
| 5 | use CirrusSearch\CirrusConfigNames; |
| 6 | use CirrusSearch\Connection; |
| 7 | use CirrusSearch\Hooks; |
| 8 | use CirrusSearch\SearchConfig; |
| 9 | use CirrusSearch\WarningCollector; |
| 10 | use Elastica\Query\MoreLikeThis; |
| 11 | use MediaWiki\MediaWikiServices; |
| 12 | use MediaWiki\Page\PageIdentity; |
| 13 | use MediaWiki\Title\Title; |
| 14 | |
| 15 | trait MoreLikeTrait { |
| 16 | /** |
| 17 | * @param string $key |
| 18 | * @param string $term |
| 19 | * @param WarningCollector $warningCollector |
| 20 | * @return PageIdentity[] |
| 21 | */ |
| 22 | protected function doExpand( $key, $term, WarningCollector $warningCollector ) { |
| 23 | // If no fields have been set we return no results. This can happen if |
| 24 | // the user override this setting with field names that are not allowed |
| 25 | // in $this->getConfig()->get( 'CirrusSearchMoreLikeThisAllowedFields' ) |
| 26 | // (see Hooks.php) |
| 27 | if ( !$this->getConfig()->get( CirrusConfigNames::MoreLikeThisFields ) ) { |
| 28 | $warningCollector->addWarning( "cirrussearch-mlt-not-configured", $key ); |
| 29 | return []; |
| 30 | } |
| 31 | $titles = $this->collectTitles( $term ); |
| 32 | if ( $titles === [] ) { |
| 33 | $warningCollector->addWarning( "cirrussearch-mlt-feature-no-valid-titles", $key ); |
| 34 | } |
| 35 | return $titles; |
| 36 | } |
| 37 | |
| 38 | /** |
| 39 | * @param string $term |
| 40 | * @return PageIdentity[] |
| 41 | */ |
| 42 | private function collectTitles( $term ) { |
| 43 | if ( $this->getConfig()->getElement( CirrusConfigNames::DevelOptions, |
| 44 | 'morelike_collect_titles_from_elastic' ) |
| 45 | ) { |
| 46 | return $this->collectTitlesFromElastic( $term ); |
| 47 | } else { |
| 48 | return $this->collectTitlesFromDB( $term ); |
| 49 | } |
| 50 | } |
| 51 | |
| 52 | /** |
| 53 | * Use for devel purpose only |
| 54 | * @param string $terms |
| 55 | * @return PageIdentity[] |
| 56 | */ |
| 57 | private function collectTitlesFromElastic( $terms ) { |
| 58 | $titles = []; |
| 59 | foreach ( explode( '|', $terms ) as $term ) { |
| 60 | $title = null; |
| 61 | Hooks::handleSearchGetNearMatch( $term, $title ); |
| 62 | if ( $title != null ) { |
| 63 | $titles[] = $title; |
| 64 | } |
| 65 | } |
| 66 | return $titles; |
| 67 | } |
| 68 | |
| 69 | /** |
| 70 | * @param string $term |
| 71 | * @return PageIdentity[] |
| 72 | */ |
| 73 | private function collectTitlesFromDB( $term ) { |
| 74 | $titles = []; |
| 75 | $found = []; |
| 76 | $titleFactory = MediaWikiServices::getInstance()->getTitleFactory(); |
| 77 | $redirectLookup = MediaWikiServices::getInstance()->getRedirectLookup(); |
| 78 | foreach ( explode( '|', $term ) as $title ) { |
| 79 | $title = $titleFactory->newFromText( trim( $title ) ); |
| 80 | while ( true ) { |
| 81 | if ( !$title ) { |
| 82 | continue 2; |
| 83 | } |
| 84 | $titleText = $title->getFullText(); |
| 85 | if ( isset( $found[$titleText] ) ) { |
| 86 | continue 2; |
| 87 | } |
| 88 | $found[$titleText] = true; |
| 89 | if ( !$title->exists() ) { |
| 90 | continue 2; |
| 91 | } |
| 92 | $redirectTarget = $redirectLookup->getRedirectTarget( $title ); |
| 93 | if ( !$redirectTarget ) { |
| 94 | break; |
| 95 | } |
| 96 | $title = Title::castFromLinkTarget( $redirectTarget ); |
| 97 | } |
| 98 | $titles[] = $title; |
| 99 | } |
| 100 | |
| 101 | return $titles; |
| 102 | } |
| 103 | |
| 104 | /** |
| 105 | * Builds a more like this query for the specified titles. Take care that |
| 106 | * this outputs a stable result, regardless of order of configuration |
| 107 | * parameters and input titles. The result of this is hashed to generate an |
| 108 | * application side cache key. If the result is unstable we will see a |
| 109 | * reduced hit rate, and waste cache storage space. |
| 110 | * |
| 111 | * @param PageIdentity[] $titles |
| 112 | * @return MoreLikeThis |
| 113 | */ |
| 114 | protected function buildMoreLikeQuery( array $titles ) { |
| 115 | sort( $titles, SORT_STRING ); |
| 116 | $likeDocs = []; |
| 117 | // We pull a connection object to access index names, ideally we should not require this |
| 118 | // since we actually don't make any connection but these methods have been available there |
| 119 | // for historical reasons. |
| 120 | $connection = new Connection( $this->getConfig() ); |
| 121 | $indexBaseName = $this->getConfig()->get( CirrusConfigNames::IndexBaseName ); |
| 122 | foreach ( $titles as $title ) { |
| 123 | $docId = $this->getConfig()->makeId( $title->getId() ); |
| 124 | |
| 125 | $likeDocs[] = [ |
| 126 | '_id' => $docId, |
| 127 | '_index' => $connection->getIndexName( $indexBaseName, $connection->getIndexSuffixForNamespace( $title->getNamespace() ) ) |
| 128 | ]; |
| 129 | } |
| 130 | |
| 131 | $moreLikeThisFields = $this->getConfig()->get( CirrusConfigNames::MoreLikeThisFields ); |
| 132 | sort( $moreLikeThisFields ); |
| 133 | $query = new MoreLikeThis(); |
| 134 | $query->setParams( $this->getConfig()->get( CirrusConfigNames::MoreLikeThisConfig ) ); |
| 135 | $query->setFields( $moreLikeThisFields ); |
| 136 | |
| 137 | /** @phan-suppress-next-line PhanTypeMismatchArgumentProbablyReal library is mis-annotated */ |
| 138 | $query->setLike( $likeDocs ); |
| 139 | |
| 140 | return $query; |
| 141 | } |
| 142 | |
| 143 | /** |
| 144 | * @return SearchConfig |
| 145 | */ |
| 146 | abstract public function getConfig(): SearchConfig; |
| 147 | } |