Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
94.44% |
51 / 54 |
|
80.00% |
4 / 5 |
CRAP | |
0.00% |
0 / 1 |
| LangDetectFallbackMethod | |
94.44% |
51 / 54 |
|
80.00% |
4 / 5 |
18.06 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
6 / 6 |
|
100.00% |
1 / 1 |
1 | |||
| build | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
2 | |||
| successApproximation | |
100.00% |
23 / 23 |
|
100.00% |
1 / 1 |
8 | |||
| rewrite | |
85.00% |
17 / 20 |
|
0.00% |
0 / 1 |
6.12 | |||
| getMetrics | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace CirrusSearch\Fallbacks; |
| 4 | |
| 5 | use CirrusSearch\CirrusConfigNames; |
| 6 | use CirrusSearch\InterwikiResolver; |
| 7 | use CirrusSearch\LanguageDetector\Detector; |
| 8 | use CirrusSearch\LanguageDetector\LanguageDetectorFactory; |
| 9 | use CirrusSearch\Parser\BasicQueryClassifier; |
| 10 | use CirrusSearch\Search\CirrusSearchResultSet; |
| 11 | use CirrusSearch\Search\SearchMetricsProvider; |
| 12 | use CirrusSearch\Search\SearchQuery; |
| 13 | use CirrusSearch\Search\SearchQueryBuilder; |
| 14 | use CirrusSearch\SearchConfig; |
| 15 | use CirrusSearch\Searcher; |
| 16 | use MediaWiki\MainConfigNames; |
| 17 | use Wikimedia\Assert\Assert; |
| 18 | |
| 19 | class LangDetectFallbackMethod implements FallbackMethod, SearchMetricsProvider { |
| 20 | use FallbackMethodTrait; |
| 21 | |
| 22 | /** |
| 23 | * @var SearchQuery |
| 24 | */ |
| 25 | private $query; |
| 26 | |
| 27 | /** |
| 28 | * @var SearcherFactory |
| 29 | */ |
| 30 | private $searcherFactory; |
| 31 | |
| 32 | /** |
| 33 | * @var array|null |
| 34 | */ |
| 35 | private $searchMetrics = []; |
| 36 | |
| 37 | /** |
| 38 | * @var Detector[] |
| 39 | */ |
| 40 | private $detectors; |
| 41 | |
| 42 | /** |
| 43 | * @var InterwikiResolver |
| 44 | */ |
| 45 | private $interwikiResolver; |
| 46 | |
| 47 | /** |
| 48 | * @var SearchConfig|null |
| 49 | */ |
| 50 | private $detectedLangWikiConfig; |
| 51 | |
| 52 | /** |
| 53 | * @var int |
| 54 | */ |
| 55 | private $threshold; |
| 56 | |
| 57 | /** |
| 58 | * Do not use this constructor outside of tests! |
| 59 | * @param SearchQuery $query |
| 60 | * @param Detector[] $detectors |
| 61 | * @param InterwikiResolver $interwikiResolver |
| 62 | */ |
| 63 | public function __construct( |
| 64 | SearchQuery $query, |
| 65 | array $detectors, |
| 66 | InterwikiResolver $interwikiResolver |
| 67 | ) { |
| 68 | Assert::precondition( $query->getCrossSearchStrategy()->isCrossLanguageSearchSupported(), |
| 69 | "Cross language search must be supported for this query" ); |
| 70 | $this->query = $query; |
| 71 | $this->detectors = $detectors; |
| 72 | $this->interwikiResolver = $interwikiResolver; |
| 73 | $this->threshold = $query->getSearchConfig()->get( CirrusConfigNames::InterwikiThreshold ); |
| 74 | } |
| 75 | |
| 76 | /** |
| 77 | * @param SearchQuery $query |
| 78 | * @param array $params |
| 79 | * @param InterwikiResolver $interwikiResolver |
| 80 | * @return FallbackMethod|null |
| 81 | */ |
| 82 | public static function build( SearchQuery $query, array $params, InterwikiResolver $interwikiResolver ) { |
| 83 | if ( !$query->getCrossSearchStrategy()->isCrossLanguageSearchSupported() ) { |
| 84 | return null; |
| 85 | } |
| 86 | $langDetectFactory = new LanguageDetectorFactory( $query->getSearchConfig() ); |
| 87 | return new self( $query, $langDetectFactory->getDetectors(), $interwikiResolver ); |
| 88 | } |
| 89 | |
| 90 | /** |
| 91 | * @param FallbackRunnerContext $context |
| 92 | * @return float |
| 93 | */ |
| 94 | public function successApproximation( FallbackRunnerContext $context ) { |
| 95 | $firstPassResults = $context->getInitialResultSet(); |
| 96 | if ( !$this->query->isAllowRewrite() ) { |
| 97 | return 0.0; |
| 98 | } |
| 99 | if ( $this->resultsThreshold( $firstPassResults, $this->threshold ) ) { |
| 100 | return 0.0; |
| 101 | } |
| 102 | if ( !$this->query->getParsedQuery()->isQueryOfClass( BasicQueryClassifier::SIMPLE_BAG_OF_WORDS ) ) { |
| 103 | return 0.0; |
| 104 | } |
| 105 | foreach ( $this->detectors as $name => $detector ) { |
| 106 | $lang = $detector->detect( $this->query->getParsedQuery()->getRawQuery() ); |
| 107 | if ( $lang === null ) { |
| 108 | continue; |
| 109 | } |
| 110 | if ( $lang === $this->query->getSearchConfig()->get( MainConfigNames::LanguageCode ) ) { |
| 111 | // The query is in the wiki language so we |
| 112 | // don't need to actually try another wiki. |
| 113 | // Note that this may not be very accurate for |
| 114 | // wikis that use deprecated language codes |
| 115 | // but the interwiki resolver should not return |
| 116 | // ourselves. |
| 117 | continue; |
| 118 | } |
| 119 | $iwPrefixAndConfig = $this->interwikiResolver->getSameProjectConfigByLang( $lang ); |
| 120 | if ( $iwPrefixAndConfig ) { |
| 121 | // it might be more accurate to attach these to the 'next' |
| 122 | // log context? It would be inconsistent with the |
| 123 | // langdetect => false condition which does not have a next |
| 124 | // request though. |
| 125 | Searcher::appendLastLogPayload( 'langdetect', $name ); |
| 126 | $prefix = key( $iwPrefixAndConfig ); |
| 127 | $config = $iwPrefixAndConfig[$prefix]; |
| 128 | $metric = [ $config->getWikiId(), $prefix ]; |
| 129 | $this->detectedLangWikiConfig = $config; |
| 130 | return 0.5; |
| 131 | } |
| 132 | } |
| 133 | Searcher::appendLastLogPayload( 'langdetect', 'failed' ); |
| 134 | return 0.0; |
| 135 | } |
| 136 | |
| 137 | public function rewrite( FallbackRunnerContext $context ): FallbackStatus { |
| 138 | $previousSet = $context->getPreviousResultSet(); |
| 139 | Assert::precondition( $this->detectedLangWikiConfig !== null, |
| 140 | 'nothing has been detected, this should not even be tried.' ); |
| 141 | |
| 142 | if ( $this->resultsThreshold( $previousSet, $this->threshold ) ) { |
| 143 | return FallbackStatus::noSuggestion(); |
| 144 | } |
| 145 | |
| 146 | if ( !$context->costlyCallAllowed() ) { |
| 147 | return FallbackStatus::noSuggestion(); |
| 148 | } |
| 149 | |
| 150 | $crossLangQuery = SearchQueryBuilder::forCrossLanguageSearch( $this->detectedLangWikiConfig, |
| 151 | $this->query )->build(); |
| 152 | $searcher = $context->makeSearcher( $crossLangQuery ); |
| 153 | $status = $searcher->search( $crossLangQuery ); |
| 154 | if ( !$status->isOK() ) { |
| 155 | return FallbackStatus::noSuggestion(); |
| 156 | } |
| 157 | $crossLangResults = $status->getValue(); |
| 158 | if ( !$crossLangResults instanceof CirrusSearchResultSet ) { |
| 159 | // NOTE: Can/should this happen? |
| 160 | return FallbackStatus::noSuggestion(); |
| 161 | } |
| 162 | if ( $crossLangResults->numRows() > 0 ) { |
| 163 | return FallbackStatus::addInterwikiResults( $crossLangResults, |
| 164 | $this->detectedLangWikiConfig->getWikiId() ); |
| 165 | } |
| 166 | return FallbackStatus::noSuggestion(); |
| 167 | } |
| 168 | |
| 169 | /** @inheritDoc */ |
| 170 | public function getMetrics() { |
| 171 | return $this->searchMetrics; |
| 172 | } |
| 173 | } |