Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
96.67% |
58 / 60 |
|
88.89% |
8 / 9 |
CRAP | |
0.00% |
0 / 1 |
| InCategoryFeature | |
96.67% |
58 / 60 |
|
88.89% |
8 / 9 |
20 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| getKeywords | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getCrossSearchStrategy | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
2 | |||
| doApply | |
80.00% |
8 / 10 |
|
0.00% |
0 / 1 |
3.07 | |||
| parseValue | |
100.00% |
21 / 21 |
|
100.00% |
1 / 1 |
5 | |||
| expand | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| doExpand | |
100.00% |
14 / 14 |
|
100.00% |
1 / 1 |
3 | |||
| matchPageCategories | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
2 | |||
| getFilterQuery | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
2 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace CirrusSearch\Query; |
| 4 | |
| 5 | use CirrusSearch\CirrusConfigNames; |
| 6 | use CirrusSearch\CrossSearchStrategy; |
| 7 | use CirrusSearch\Parser\AST\KeywordFeatureNode; |
| 8 | use CirrusSearch\Query\Builder\QueryBuildingContext; |
| 9 | use CirrusSearch\Search\SearchContext; |
| 10 | use CirrusSearch\SearchConfig; |
| 11 | use CirrusSearch\WarningCollector; |
| 12 | use Elastica\Query\AbstractQuery; |
| 13 | use MediaWiki\Config\Config; |
| 14 | use MediaWiki\MediaWikiServices; |
| 15 | use MediaWiki\Page\PageRecord; |
| 16 | use MediaWiki\Page\PageStore; |
| 17 | |
| 18 | /** |
| 19 | * Filters by one or more categories, specified either by name or by category |
| 20 | * id. Multiple categories are separated by |. Categories specified by id |
| 21 | * must follow the syntax `id:<id>`. |
| 22 | * |
| 23 | * We emulate template syntax here as best as possible, so things in NS_MAIN |
| 24 | * are prefixed with ":" and things in NS_TEMPLATE don't have a prefix at all. |
| 25 | * Since we don't actually index templates like that, munge the query here. |
| 26 | * |
| 27 | * Examples: |
| 28 | * incategory:id:12345 |
| 29 | * incategory:Music_by_genre |
| 30 | * incategory:Music_by_genre|Animals |
| 31 | * incategory:"Music by genre|Animals" |
| 32 | * incategory:Animals|id:54321 |
| 33 | * incategory::Something_in_NS_MAIN |
| 34 | */ |
| 35 | class InCategoryFeature extends SimpleKeywordFeature implements FilterQueryFeature { |
| 36 | /** |
| 37 | * @var int |
| 38 | */ |
| 39 | private $maxConditions; |
| 40 | /** |
| 41 | * @var PageStore|null |
| 42 | */ |
| 43 | private $pageStore; |
| 44 | |
| 45 | /** |
| 46 | * @param Config $config |
| 47 | * @param PageStore|null $pageStore |
| 48 | */ |
| 49 | public function __construct( Config $config, ?PageStore $pageStore = null ) { |
| 50 | $this->maxConditions = $config->get( CirrusConfigNames::MaxIncategoryOptions ); |
| 51 | $this->pageStore = $pageStore; |
| 52 | } |
| 53 | |
| 54 | /** |
| 55 | * @return string[] |
| 56 | */ |
| 57 | protected function getKeywords() { |
| 58 | return [ 'incategory' ]; |
| 59 | } |
| 60 | |
| 61 | /** |
| 62 | * @param KeywordFeatureNode $node |
| 63 | * @return CrossSearchStrategy |
| 64 | */ |
| 65 | public function getCrossSearchStrategy( KeywordFeatureNode $node ) { |
| 66 | if ( empty( $node->getParsedValue()['pageIds'] ) ) { |
| 67 | // We depend on the db to fetch the category by id |
| 68 | return CrossSearchStrategy::allWikisStrategy(); |
| 69 | } else { |
| 70 | return CrossSearchStrategy::hostWikiOnlyStrategy(); |
| 71 | } |
| 72 | } |
| 73 | |
| 74 | /** |
| 75 | * @param SearchContext $context |
| 76 | * @param string $key The keyword |
| 77 | * @param string $value The value attached to the keyword with quotes stripped |
| 78 | * @param string $quotedValue The original value in the search string, including quotes if used |
| 79 | * @param bool $negated Is the search negated? Not used to generate the returned AbstractQuery, |
| 80 | * that will be negated as necessary. Used for any other building/context necessary. |
| 81 | * @return array Two element array, first an AbstractQuery or null to apply to the |
| 82 | * query. Second a boolean indicating if the quotedValue should be kept in the search |
| 83 | * string. |
| 84 | */ |
| 85 | protected function doApply( SearchContext $context, $key, $value, $quotedValue, $negated ) { |
| 86 | $parsedValue = $this->parseValue( $key, $value, $quotedValue, '', '', $context ); |
| 87 | if ( $parsedValue === null ) { |
| 88 | $context->setResultsPossible( false ); |
| 89 | return [ null, false ]; |
| 90 | } |
| 91 | |
| 92 | $names = $this->doExpand( $key, $parsedValue, $context ); |
| 93 | |
| 94 | if ( $names === [] ) { |
| 95 | $context->setResultsPossible( false ); |
| 96 | return [ null, false ]; |
| 97 | } |
| 98 | |
| 99 | $filter = $this->matchPageCategories( $names ); |
| 100 | return [ $filter, false ]; |
| 101 | } |
| 102 | |
| 103 | /** |
| 104 | * @param string $key |
| 105 | * @param string $value |
| 106 | * @param string $quotedValue |
| 107 | * @param string $valueDelimiter |
| 108 | * @param string $suffix |
| 109 | * @param WarningCollector $warningCollector |
| 110 | * @return array|false|null |
| 111 | */ |
| 112 | public function parseValue( $key, $value, $quotedValue, $valueDelimiter, $suffix, WarningCollector $warningCollector ) { |
| 113 | // en:Programming|id:3041512\ |
| 114 | $categories = explode( '|', $value );// en:programming |
| 115 | if ( count( $categories ) > $this->maxConditions ) { |
| 116 | $warningCollector->addWarning( |
| 117 | 'cirrussearch-feature-too-many-conditions', |
| 118 | $key, |
| 119 | $this->maxConditions |
| 120 | ); |
| 121 | $categories = array_slice( |
| 122 | $categories, |
| 123 | 0, |
| 124 | $this->maxConditions |
| 125 | ); |
| 126 | } |
| 127 | |
| 128 | $pageIds = []; |
| 129 | $names = []; |
| 130 | |
| 131 | foreach ( $categories as $category ) { |
| 132 | if ( str_starts_with( $category, 'id:' ) ) { |
| 133 | $pageId = substr( $category, 3 ); |
| 134 | if ( ctype_digit( $pageId ) ) { |
| 135 | $pageIds[] = $pageId; |
| 136 | } |
| 137 | } else { |
| 138 | $names[] = $category;// en:programming |
| 139 | } |
| 140 | } |
| 141 | |
| 142 | return [ 'names' => $names, 'pageIds' => $pageIds ];// en:programming |
| 143 | } |
| 144 | |
| 145 | /** |
| 146 | * @param KeywordFeatureNode $node |
| 147 | * @param SearchConfig $config |
| 148 | * @param WarningCollector $warningCollector |
| 149 | * @return array |
| 150 | */ |
| 151 | public function expand( KeywordFeatureNode $node, SearchConfig $config, WarningCollector $warningCollector ) { |
| 152 | return $this->doExpand( $node->getKey(), $node->getParsedValue(), $warningCollector ); |
| 153 | } |
| 154 | |
| 155 | /** |
| 156 | * @param string $key |
| 157 | * @param array $parsedValue |
| 158 | * @param WarningCollector $warningCollector |
| 159 | * @return array |
| 160 | */ |
| 161 | private function doExpand( $key, array $parsedValue, WarningCollector $warningCollector ) { |
| 162 | $names = $parsedValue['names'];// en:programming |
| 163 | $pageIds = $parsedValue['pageIds']; |
| 164 | |
| 165 | $pageStore = $this->pageStore ?? MediaWikiServices::getInstance()->getPageStore(); |
| 166 | $titles = $pageStore |
| 167 | ->newSelectQueryBuilder() |
| 168 | ->wherePageIds( $pageIds ) |
| 169 | ->caller( __METHOD__ ) |
| 170 | ->fetchPageRecords(); |
| 171 | |
| 172 | $titleFormatter = MediaWikiServices::getInstance()->getTitleFormatter(); |
| 173 | |
| 174 | /** @var PageRecord $title */ |
| 175 | foreach ( $titles as $title ) { |
| 176 | $names[] = $titleFormatter->getText( $title ); |
| 177 | } |
| 178 | |
| 179 | if ( $names === [] ) { |
| 180 | $warningCollector->addWarning( 'cirrussearch-incategory-feature-no-valid-categories', $key ); |
| 181 | } |
| 182 | return $names;// en:programing |
| 183 | } |
| 184 | |
| 185 | /** |
| 186 | * Builds an or between many categories that the page could be in. |
| 187 | * |
| 188 | * @param string[] $names categories to match |
| 189 | * @return \Elastica\Query\BoolQuery|null A null return value means all values are filtered |
| 190 | * and an empty result set should be returned. |
| 191 | */ |
| 192 | private function matchPageCategories( array $names ) { |
| 193 | $filter = new \Elastica\Query\BoolQuery(); |
| 194 | |
| 195 | foreach ( $names as $name ) { |
| 196 | $filter->addShould( QueryHelper::matchCategory( 'category.lowercase_keyword', $name ) ); |
| 197 | } |
| 198 | |
| 199 | return $filter; |
| 200 | } |
| 201 | |
| 202 | /** |
| 203 | * @param KeywordFeatureNode $node |
| 204 | * @param QueryBuildingContext $context |
| 205 | * @return AbstractQuery|null |
| 206 | */ |
| 207 | public function getFilterQuery( KeywordFeatureNode $node, QueryBuildingContext $context ) { |
| 208 | $names = $context->getKeywordExpandedData( $node ); |
| 209 | if ( $names === [] ) { |
| 210 | return null; |
| 211 | } |
| 212 | return $this->matchPageCategories( $names ); |
| 213 | } |
| 214 | } |