Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 67 |
|
0.00% |
0 / 3 |
CRAP | |
0.00% |
0 / 1 |
| FormSearchEntity | |
0.00% |
0 / 67 |
|
0.00% |
0 / 3 |
56 | |
0.00% |
0 / 1 |
| getElasticSearchQuery | |
0.00% |
0 / 58 |
|
0.00% |
0 / 1 |
30 | |||
| makeResultType | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
2 | |||
| getRankedSearchResults | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
2 | |||
| 1 | <?php |
| 2 | namespace Wikibase\Lexeme\Search\Elastic; |
| 3 | |
| 4 | use CirrusSearch\Search\ResultsType; |
| 5 | use CirrusSearch\Search\SearchContext; |
| 6 | use Elastica\Query\AbstractQuery; |
| 7 | use Elastica\Query\BoolQuery; |
| 8 | use Elastica\Query\DisMax; |
| 9 | use Elastica\Query\MatchNone; |
| 10 | use Elastica\Query\MatchQuery; |
| 11 | use Elastica\Query\Term; |
| 12 | use Wikibase\Lexeme\MediaWiki\Content\LexemeContent; |
| 13 | use Wikibase\Search\Elastic\EntitySearchElastic; |
| 14 | use Wikibase\Search\Elastic\EntitySearchUtils; |
| 15 | |
| 16 | /** |
| 17 | * Search for entity of type form. |
| 18 | * |
| 19 | * @license GPL-2.0-or-later |
| 20 | * @author Stas Malyshev |
| 21 | */ |
| 22 | class FormSearchEntity extends LexemeSearchEntity { |
| 23 | /** |
| 24 | * Search limit. |
| 25 | * @var int |
| 26 | */ |
| 27 | private $limit; |
| 28 | |
| 29 | /** |
| 30 | * Produce ES query that matches the arguments. |
| 31 | * This is search for forms - matches only form representations |
| 32 | * but not lexemes. |
| 33 | * |
| 34 | * @param string $text |
| 35 | * @param string $entityType |
| 36 | * @param SearchContext $context |
| 37 | * |
| 38 | * @return AbstractQuery |
| 39 | */ |
| 40 | protected function getElasticSearchQuery( |
| 41 | $text, |
| 42 | $entityType, |
| 43 | SearchContext $context |
| 44 | ) { |
| 45 | $context->setOriginalSearchTerm( $text ); |
| 46 | // TODO consider using Form::ENTITY_TYPE |
| 47 | if ( $entityType !== 'form' ) { |
| 48 | $context->setResultsPossible( false ); |
| 49 | $context->addWarning( 'wikibase-search-bad-entity-type', $entityType ); |
| 50 | return new MatchNone(); |
| 51 | } |
| 52 | // Drop only leading spaces for exact matches, and all spaces for the rest |
| 53 | $textExact = ltrim( $text ); |
| 54 | $text = trim( $text ); |
| 55 | |
| 56 | $labelsFilter = new MatchQuery( 'lexeme_forms.representation.prefix', $text ); |
| 57 | |
| 58 | $profile = $context->getConfig() |
| 59 | ->getProfileService() |
| 60 | ->loadProfile( EntitySearchElastic::WIKIBASE_PREFIX_QUERY_BUILDER, |
| 61 | self::CONTEXT_LEXEME_PREFIX ); |
| 62 | |
| 63 | $dismax = new DisMax(); |
| 64 | $dismax->setTieBreaker( $profile['tie-breaker'] ?? 0 ); |
| 65 | |
| 66 | $fields = [ |
| 67 | [ |
| 68 | "lexeme_forms.representation.near_match", |
| 69 | $profile['exact'] * $profile['form-discount'], |
| 70 | ], |
| 71 | [ |
| 72 | "lexeme_forms.representation.near_match_folded", |
| 73 | $profile['folded'] * $profile['form-discount'], |
| 74 | ], |
| 75 | ]; |
| 76 | // Fields to which query applies exactly as stated, without trailing space trimming |
| 77 | $fieldsExact = []; |
| 78 | if ( $textExact !== $text ) { |
| 79 | $fields[] = |
| 80 | [ |
| 81 | "lexeme_forms.representation.prefix", |
| 82 | $profile['prefix'] * $profile['space-discount'] * $profile['form-discount'], |
| 83 | ]; |
| 84 | $fieldsExact[] = |
| 85 | [ |
| 86 | "lexeme_forms.representation.prefix", |
| 87 | $profile['prefix'] * $profile['form-discount'], |
| 88 | ]; |
| 89 | } else { |
| 90 | $fields[] = |
| 91 | [ |
| 92 | "lexeme_forms.representation.prefix", |
| 93 | $profile['prefix'] * $profile['form-discount'], |
| 94 | ]; |
| 95 | } |
| 96 | |
| 97 | foreach ( $fields as $field ) { |
| 98 | $dismax->addQuery( EntitySearchUtils::makeConstScoreQuery( $field[0], $field[1], |
| 99 | $text ) ); |
| 100 | } |
| 101 | |
| 102 | foreach ( $fieldsExact as $field ) { |
| 103 | $dismax->addQuery( EntitySearchUtils::makeConstScoreQuery( $field[0], $field[1], |
| 104 | $textExact ) ); |
| 105 | } |
| 106 | |
| 107 | $labelsQuery = new BoolQuery(); |
| 108 | $labelsQuery->addFilter( $labelsFilter ); |
| 109 | $labelsQuery->addShould( $dismax ); |
| 110 | // lexeme_forms.id is a lowercase_keyword so use Match to apply the analyzer |
| 111 | $titleMatch = new MatchQuery( 'lexeme_forms.id', |
| 112 | EntitySearchUtils::normalizeId( $text, $this->idParser ) ); |
| 113 | |
| 114 | $query = new BoolQuery(); |
| 115 | // Match either labels or exact match to title |
| 116 | $query->addShould( $labelsQuery ); |
| 117 | $query->addShould( $titleMatch ); |
| 118 | $query->setMinimumShouldMatch( 1 ); |
| 119 | |
| 120 | // Filter to fetch only given entity type |
| 121 | $query->addFilter( new Term( [ 'content_model' => LexemeContent::CONTENT_MODEL_ID ] ) ); |
| 122 | |
| 123 | return $query; |
| 124 | } |
| 125 | |
| 126 | /** |
| 127 | * Get results type object for this search. |
| 128 | * @return ResultsType |
| 129 | */ |
| 130 | protected function makeResultType() { |
| 131 | return new FormTermResult( |
| 132 | $this->idParser, |
| 133 | $this->userLanguage, |
| 134 | $this->lookupFactory, |
| 135 | $this->limit |
| 136 | ); |
| 137 | } |
| 138 | |
| 139 | /** |
| 140 | * @inheritDoc |
| 141 | */ |
| 142 | public function getRankedSearchResults( |
| 143 | $text, |
| 144 | $languageCode, |
| 145 | $entityType, |
| 146 | $limit, |
| 147 | $strictLanguage, |
| 148 | ?string $profileContext = null |
| 149 | ) { |
| 150 | // We need to keep the limit since one document can produce several matches. |
| 151 | $this->limit = $limit; |
| 152 | return parent::getRankedSearchResults( $text, $languageCode, $entityType, $limit, |
| 153 | $strictLanguage, $profileContext ); |
| 154 | } |
| 155 | |
| 156 | } |