Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
100.00% |
32 / 32 |
|
100.00% |
4 / 4 |
CRAP | |
100.00% |
1 / 1 |
| ArticleTopicFeature | |
100.00% |
32 / 32 |
|
100.00% |
4 / 4 |
9 | |
100.00% |
1 / 1 |
| getTopicScores | |
100.00% |
7 / 7 |
|
100.00% |
1 / 1 |
2 | |||
| parseValue | |
100.00% |
9 / 9 |
|
100.00% |
1 / 1 |
2 | |||
| getKeywords | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| doApply | |
100.00% |
15 / 15 |
|
100.00% |
1 / 1 |
4 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace CirrusSearch\Query; |
| 4 | |
| 5 | use CirrusSearch\Search\SearchContext; |
| 6 | use CirrusSearch\WarningCollector; |
| 7 | use CirrusSearch\Wikimedia\WeightedTagsHooks; |
| 8 | use Elastica\Query\DisMax; |
| 9 | use Elastica\Query\Term; |
| 10 | use MediaWiki\Message\Message; |
| 11 | |
| 12 | /** |
| 13 | * Finds pages based on how well they match a given topic, based on scores provided by the |
| 14 | * (Wikimedia-specific) articletopic ORES model. |
| 15 | * @package CirrusSearch\Wikimedia |
| 16 | * @see WeightedTagsHooks |
| 17 | * @see https://www.mediawiki.org/wiki/Help:CirrusSearch#Articletopic |
| 18 | */ |
| 19 | class ArticleTopicFeature extends SimpleKeywordFeature { |
| 20 | public const ARTICLE_TOPIC_TAG_PREFIX = 'classification.ores.articletopic'; |
| 21 | public const DRAFT_TOPIC_TAG_PREFIX = 'classification.ores.drafttopic'; |
| 22 | |
| 23 | private const PREFIX_PER_KEYWORD = [ |
| 24 | 'articletopic' => self::ARTICLE_TOPIC_TAG_PREFIX, |
| 25 | 'drafttopic' => self::DRAFT_TOPIC_TAG_PREFIX |
| 26 | ]; |
| 27 | |
| 28 | public const TERMS_TO_LABELS = [ |
| 29 | 'biography' => 'Culture.Biography.Biography*', |
| 30 | 'women' => 'Culture.Biography.Women', |
| 31 | 'food-and-drink' => 'Culture.Food and drink', |
| 32 | 'internet-culture' => 'Culture.Internet culture', |
| 33 | 'linguistics' => 'Culture.Linguistics', |
| 34 | 'literature' => 'Culture.Literature', |
| 35 | 'books' => 'Culture.Media.Books', |
| 36 | 'entertainment' => 'Culture.Media.Entertainment', |
| 37 | 'films' => 'Culture.Media.Films', |
| 38 | 'media' => 'Culture.Media.Media*', |
| 39 | 'music' => 'Culture.Media.Music', |
| 40 | 'radio' => 'Culture.Media.Radio', |
| 41 | 'software' => 'Culture.Media.Software', |
| 42 | 'television' => 'Culture.Media.Television', |
| 43 | 'video-games' => 'Culture.Media.Video games', |
| 44 | 'performing-arts' => 'Culture.Performing arts', |
| 45 | 'philosophy-and-religion' => 'Culture.Philosophy and religion', |
| 46 | 'sports' => 'Culture.Sports', |
| 47 | 'architecture' => 'Culture.Visual arts.Architecture', |
| 48 | 'comics-and-anime' => 'Culture.Visual arts.Comics and Anime', |
| 49 | 'fashion' => 'Culture.Visual arts.Fashion', |
| 50 | 'visual-arts' => 'Culture.Visual arts.Visual arts*', |
| 51 | 'geographical' => 'Geography.Geographical', |
| 52 | 'africa' => 'Geography.Regions.Africa.Africa*', |
| 53 | 'central-africa' => 'Geography.Regions.Africa.Central Africa', |
| 54 | 'eastern-africa' => 'Geography.Regions.Africa.Eastern Africa', |
| 55 | 'northern-africa' => 'Geography.Regions.Africa.Northern Africa', |
| 56 | 'southern-africa' => 'Geography.Regions.Africa.Southern Africa', |
| 57 | 'western-africa' => 'Geography.Regions.Africa.Western Africa', |
| 58 | 'central-america' => 'Geography.Regions.Americas.Central America', |
| 59 | 'north-america' => 'Geography.Regions.Americas.North America', |
| 60 | 'south-america' => 'Geography.Regions.Americas.South America', |
| 61 | 'asia' => 'Geography.Regions.Asia.Asia*', |
| 62 | 'central-asia' => 'Geography.Regions.Asia.Central Asia', |
| 63 | 'east-asia' => 'Geography.Regions.Asia.East Asia', |
| 64 | 'north-asia' => 'Geography.Regions.Asia.North Asia', |
| 65 | 'south-asia' => 'Geography.Regions.Asia.South Asia', |
| 66 | 'southeast-asia' => 'Geography.Regions.Asia.Southeast Asia', |
| 67 | 'west-asia' => 'Geography.Regions.Asia.West Asia', |
| 68 | 'eastern-europe' => 'Geography.Regions.Europe.Eastern Europe', |
| 69 | 'europe' => 'Geography.Regions.Europe.Europe*', |
| 70 | 'northern-europe' => 'Geography.Regions.Europe.Northern Europe', |
| 71 | 'southern-europe' => 'Geography.Regions.Europe.Southern Europe', |
| 72 | 'western-europe' => 'Geography.Regions.Europe.Western Europe', |
| 73 | 'oceania' => 'Geography.Regions.Oceania', |
| 74 | 'business-and-economics' => 'History and Society.Business and economics', |
| 75 | 'education' => 'History and Society.Education', |
| 76 | 'history' => 'History and Society.History', |
| 77 | 'military-and-warfare' => 'History and Society.Military and warfare', |
| 78 | 'politics-and-government' => 'History and Society.Politics and government', |
| 79 | 'society' => 'History and Society.Society', |
| 80 | 'transportation' => 'History and Society.Transportation', |
| 81 | 'biology' => 'STEM.Biology', |
| 82 | 'chemistry' => 'STEM.Chemistry', |
| 83 | 'computing' => 'STEM.Computing', |
| 84 | 'earth-and-environment' => 'STEM.Earth and environment', |
| 85 | 'engineering' => 'STEM.Engineering', |
| 86 | 'libraries-and-information' => 'STEM.Libraries & Information', |
| 87 | 'mathematics' => 'STEM.Mathematics', |
| 88 | 'medicine-and-health' => 'STEM.Medicine & Health', |
| 89 | 'physics' => 'STEM.Physics', |
| 90 | 'stem' => 'STEM.STEM*', |
| 91 | 'space' => 'STEM.Space', |
| 92 | 'technology' => 'STEM.Technology', |
| 93 | ]; |
| 94 | |
| 95 | /** |
| 96 | * Helper method for turning raw ORES score data (as stored in the Cirrus document) into |
| 97 | * search terms, for analytics/debugging. |
| 98 | * @param array $rawTopicData The unprefixed content of the document's weighted_tags field |
| 99 | * @return array corresponding search term => ORES score (rounded to three decimals) |
| 100 | */ |
| 101 | public static function getTopicScores( array $rawTopicData ): array { |
| 102 | $labelsToTerms = array_flip( self::TERMS_TO_LABELS ); |
| 103 | $topicScores = []; |
| 104 | foreach ( $rawTopicData as $rawTopic ) { |
| 105 | [ $oresLabel, $scaledScore ] = explode( '|', $rawTopic ); |
| 106 | $topicId = $labelsToTerms[$oresLabel]; |
| 107 | $topicScores[$topicId] = (int)$scaledScore / 1000; |
| 108 | } |
| 109 | return $topicScores; |
| 110 | } |
| 111 | |
| 112 | /** |
| 113 | * @inheritDoc |
| 114 | * @phan-return array{topics:string[],tag_prefix:string} |
| 115 | */ |
| 116 | public function parseValue( |
| 117 | $key, $value, $quotedValue, $valueDelimiter, $suffix, WarningCollector $warningCollector |
| 118 | ) { |
| 119 | $topics = explode( '|', $value ); |
| 120 | $invalidTopics = array_diff( $topics, array_keys( self::TERMS_TO_LABELS ) ); |
| 121 | $validTopics = array_filter( array_map( static function ( $topic ) { |
| 122 | return self::TERMS_TO_LABELS[$topic]; |
| 123 | }, array_diff( $topics, $invalidTopics ) ) ); |
| 124 | |
| 125 | if ( $invalidTopics ) { |
| 126 | $warningCollector->addWarning( 'cirrussearch-articletopic-invalid-topic', |
| 127 | Message::listParam( $invalidTopics, 'comma' ), count( $invalidTopics ) ); |
| 128 | } |
| 129 | return [ 'topics' => $validTopics, 'tag_prefix' => self::PREFIX_PER_KEYWORD[$key] ]; |
| 130 | } |
| 131 | |
| 132 | /** @inheritDoc */ |
| 133 | protected function getKeywords() { |
| 134 | return array_keys( self::PREFIX_PER_KEYWORD ); |
| 135 | } |
| 136 | |
| 137 | /** @inheritDoc */ |
| 138 | protected function doApply( SearchContext $context, $key, $value, $quotedValue, $negated ) { |
| 139 | $parsed = $this->parseValue( $key, $value, $quotedValue, '', '', $context ); |
| 140 | $topics = $parsed['topics']; |
| 141 | $tagPrefix = $parsed['tag_prefix']; |
| 142 | if ( $topics === [] ) { |
| 143 | $context->setResultsPossible( false ); |
| 144 | return [ null, true ]; |
| 145 | } |
| 146 | |
| 147 | $query = new DisMax(); |
| 148 | foreach ( $topics as $topic ) { |
| 149 | $topicQuery = new Term(); |
| 150 | $topicQuery->setTerm( WeightedTagsHooks::FIELD_NAME, $tagPrefix . '/' . $topic ); |
| 151 | $query->addQuery( $topicQuery ); |
| 152 | } |
| 153 | |
| 154 | if ( !$negated ) { |
| 155 | $context->addNonTextQuery( $query ); |
| 156 | return [ null, false ]; |
| 157 | } else { |
| 158 | return [ $query, false ]; |
| 159 | } |
| 160 | } |
| 161 | |
| 162 | } |