Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
91.49% |
43 / 47 |
|
33.33% |
1 / 3 |
CRAP | |
0.00% |
0 / 1 |
| EntitySearchElastic | |
91.49% |
43 / 47 |
|
33.33% |
1 / 3 |
9.05 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
6 / 6 |
|
100.00% |
1 / 1 |
3 | |||
| getElasticSearchQuery | |
85.71% |
18 / 21 |
|
0.00% |
0 / 1 |
2.01 | |||
| getRankedSearchResults | |
95.00% |
19 / 20 |
|
0.00% |
0 / 1 |
4 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace Wikibase\Search\Elastic; |
| 4 | |
| 5 | use CirrusSearch\CirrusDebugOptions; |
| 6 | use CirrusSearch\Search\SearchContext; |
| 7 | use Elastica\Query\AbstractQuery; |
| 8 | use Elastica\Query\MatchNone; |
| 9 | use MediaWiki\Language\Language; |
| 10 | use MediaWiki\Request\FauxRequest; |
| 11 | use MediaWiki\Request\WebRequest; |
| 12 | use Wikibase\DataModel\Entity\EntityIdParser; |
| 13 | use Wikibase\Lib\LanguageFallbackChainFactory; |
| 14 | use Wikibase\Repo\Api\EntitySearchException; |
| 15 | use Wikibase\Repo\Api\EntitySearchHelper; |
| 16 | use Wikibase\Search\Elastic\Query\LabelsCompletionQuery; |
| 17 | |
| 18 | /** |
| 19 | * Entity search implementation using ElasticSearch. |
| 20 | * Requires CirrusSearch extension and $wgEntitySearchUseCirrus to be on. |
| 21 | * |
| 22 | * @license GPL-2.0-or-later |
| 23 | * @author Stas Malyshev |
| 24 | */ |
| 25 | class EntitySearchElastic implements EntitySearchHelper { |
| 26 | /** |
| 27 | * Default rescore profile |
| 28 | */ |
| 29 | public const DEFAULT_RESCORE_PROFILE = 'wikibase_prefix'; |
| 30 | |
| 31 | /** |
| 32 | * Name of the context for profile name resolution |
| 33 | */ |
| 34 | public const CONTEXT_WIKIBASE_PREFIX = 'wikibase_prefix_search'; |
| 35 | |
| 36 | /** |
| 37 | * Name of the context for profile name resolution |
| 38 | */ |
| 39 | public const CONTEXT_WIKIBASE_FULLTEXT = 'wikibase_fulltext_search'; |
| 40 | |
| 41 | /** |
| 42 | * Name of the context for profile name resolution |
| 43 | */ |
| 44 | public const CONTEXT_WIKIBASE_IN_LABEL = 'wikibase_in_label_search'; |
| 45 | |
| 46 | /** |
| 47 | * Name of the profile type used to build the elastic query |
| 48 | */ |
| 49 | public const WIKIBASE_PREFIX_QUERY_BUILDER = 'wikibase_prefix_querybuilder'; |
| 50 | |
| 51 | /** |
| 52 | * Name of the profile type used to build the elastic query |
| 53 | */ |
| 54 | public const WIKIBASE_IN_LABEL_QUERY_BUILDER = 'wikibase_in_label_querybuilder'; |
| 55 | |
| 56 | /** |
| 57 | * Default query builder profile for prefix searches |
| 58 | */ |
| 59 | public const DEFAULT_QUERY_BUILDER_PROFILE = 'default'; |
| 60 | |
| 61 | /** |
| 62 | * Default query builder profile for fulltext searches |
| 63 | * |
| 64 | */ |
| 65 | public const DEFAULT_FULL_TEXT_QUERY_BUILDER_PROFILE = 'wikibase'; |
| 66 | |
| 67 | /** |
| 68 | * Replacement syntax for statement boosting |
| 69 | * @see \CirrusSearch\Profile\SearchProfileRepositoryTransformer |
| 70 | * and repo/config/ElasticSearchRescoreFunctions.php |
| 71 | */ |
| 72 | public const STMT_BOOST_PROFILE_REPL = 'functions.*[type=term_boost].params[statement_keywords=_statementBoost_].statement_keywords'; |
| 73 | |
| 74 | /** |
| 75 | * @var LanguageFallbackChainFactory |
| 76 | */ |
| 77 | private $languageChainFactory; |
| 78 | |
| 79 | /** |
| 80 | * @var EntityIdParser |
| 81 | */ |
| 82 | private $idParser; |
| 83 | |
| 84 | /** |
| 85 | * @var string[] |
| 86 | */ |
| 87 | private $contentModelMap; |
| 88 | |
| 89 | /** |
| 90 | * Web request context. |
| 91 | * Used for implementing debug features such as cirrusDumpQuery. |
| 92 | * @var WebRequest |
| 93 | */ |
| 94 | private $request; |
| 95 | |
| 96 | /** |
| 97 | * @var Language User language for display. |
| 98 | */ |
| 99 | private $userLang; |
| 100 | |
| 101 | /** |
| 102 | * @var CirrusDebugOptions |
| 103 | */ |
| 104 | private $debugOptions; |
| 105 | |
| 106 | /** |
| 107 | * @param LanguageFallbackChainFactory $languageChainFactory |
| 108 | * @param EntityIdParser $idParser |
| 109 | * @param Language $userLang |
| 110 | * @param array $contentModelMap Maps entity type => content model name |
| 111 | * @param WebRequest|null $request Web request context |
| 112 | * @param CirrusDebugOptions|null $options |
| 113 | */ |
| 114 | public function __construct( |
| 115 | LanguageFallbackChainFactory $languageChainFactory, |
| 116 | EntityIdParser $idParser, |
| 117 | Language $userLang, |
| 118 | array $contentModelMap, |
| 119 | ?WebRequest $request = null, |
| 120 | ?CirrusDebugOptions $options = null |
| 121 | ) { |
| 122 | $this->languageChainFactory = $languageChainFactory; |
| 123 | $this->idParser = $idParser; |
| 124 | $this->userLang = $userLang; |
| 125 | $this->contentModelMap = $contentModelMap; |
| 126 | $this->request = $request ?: new FauxRequest(); |
| 127 | $this->debugOptions = $options ?: CirrusDebugOptions::fromRequest( $this->request ); |
| 128 | } |
| 129 | |
| 130 | /** |
| 131 | * Produce ES query that matches the arguments. |
| 132 | * |
| 133 | * @param string $text |
| 134 | * @param string $languageCode |
| 135 | * @param string $entityType |
| 136 | * @param bool $strictLanguage |
| 137 | * @param SearchContext $context |
| 138 | * |
| 139 | * @return AbstractQuery |
| 140 | */ |
| 141 | protected function getElasticSearchQuery( |
| 142 | $text, |
| 143 | $languageCode, |
| 144 | $entityType, |
| 145 | $strictLanguage, |
| 146 | SearchContext $context |
| 147 | ) { |
| 148 | $context->setOriginalSearchTerm( $text ); |
| 149 | if ( empty( $this->contentModelMap[$entityType] ) ) { |
| 150 | $context->setResultsPossible( false ); |
| 151 | $context->addWarning( 'wikibasecirrus-search-bad-entity-type', $entityType ); |
| 152 | return new MatchNone(); |
| 153 | } |
| 154 | $profile = LabelsCompletionQuery::loadProfile( |
| 155 | $context->getConfig()->getProfileService(), |
| 156 | $this->languageChainFactory, |
| 157 | self::WIKIBASE_PREFIX_QUERY_BUILDER, |
| 158 | $context->getProfileContext(), |
| 159 | $context->getProfileContextParams(), |
| 160 | $languageCode |
| 161 | ); |
| 162 | return LabelsCompletionQuery::build( |
| 163 | $text, |
| 164 | $profile, |
| 165 | $this->contentModelMap[$entityType], |
| 166 | $languageCode, |
| 167 | $strictLanguage, |
| 168 | EntitySearchUtils::entityIdParserNormalizer( $this->idParser ) |
| 169 | ); |
| 170 | } |
| 171 | |
| 172 | /** |
| 173 | * @inheritDoc |
| 174 | */ |
| 175 | public function getRankedSearchResults( |
| 176 | $text, |
| 177 | $languageCode, |
| 178 | $entityType, |
| 179 | $limit, |
| 180 | $strictLanguage, |
| 181 | ?string $profileContext = null |
| 182 | ) { |
| 183 | $profileContext ??= self::CONTEXT_WIKIBASE_PREFIX; |
| 184 | $searcher = new WikibaseEntitySearcher( 0, $limit, 'wikibase_prefix', 'wikibase-prefix', $this->debugOptions ); |
| 185 | $searcher->getSearchContext()->setProfileContext( |
| 186 | $profileContext, |
| 187 | [ 'language' => $languageCode ] ); |
| 188 | $query = $this->getElasticSearchQuery( $text, $languageCode, $entityType, $strictLanguage, |
| 189 | $searcher->getSearchContext() ); |
| 190 | |
| 191 | $searcher->setResultsType( new EntityElasticTermResult( |
| 192 | $this->idParser, |
| 193 | $query instanceof LabelsCompletionQuery ? $query->getSearchLanguageCodes() : [], |
| 194 | 'prefix', |
| 195 | $this->languageChainFactory->newFromLanguage( $this->userLang ) |
| 196 | ) ); |
| 197 | |
| 198 | $result = $searcher->performSearch( $query ); |
| 199 | |
| 200 | if ( $result->isOK() ) { |
| 201 | $result = $result->getValue(); |
| 202 | } else { |
| 203 | throw new EntitySearchException( $result ); |
| 204 | } |
| 205 | |
| 206 | if ( $searcher->isReturnRaw() ) { |
| 207 | $result = $searcher->processRawReturn( $result, $this->request ); |
| 208 | } |
| 209 | |
| 210 | return $result; |
| 211 | } |
| 212 | |
| 213 | } |