Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
82.61% |
57 / 69 |
|
0.00% |
0 / 1 |
CRAP | |
0.00% |
0 / 1 |
| ApiQueryZObjectLabels | |
82.61% |
57 / 69 |
|
0.00% |
0 / 1 |
20.90 | |
0.00% |
0 / 1 |
| __construct | n/a |
0 / 0 |
n/a |
0 / 0 |
1 | |||||
| run | |
82.61% |
57 / 69 |
|
0.00% |
0 / 1 |
17.35 | |||
| getAllowedParams | n/a |
0 / 0 |
n/a |
0 / 0 |
1 | |||||
| getExamplesMessages | n/a |
0 / 0 |
n/a |
0 / 0 |
1 | |||||
| 1 | <?php |
| 2 | /** |
| 3 | * WikiLambda ZObject labels helper for the query API |
| 4 | * |
| 5 | * @file |
| 6 | * @ingroup Extensions |
| 7 | * @copyright 2020– Abstract Wikipedia team; see AUTHORS.txt |
| 8 | * @license MIT |
| 9 | */ |
| 10 | |
| 11 | namespace MediaWiki\Extension\WikiLambda\ActionAPI; |
| 12 | |
| 13 | use MediaWiki\Api\ApiBase; |
| 14 | use MediaWiki\Api\ApiQuery; |
| 15 | use MediaWiki\Extension\WikiLambda\HttpStatus; |
| 16 | use MediaWiki\Extension\WikiLambda\Registry\ZErrorTypeRegistry; |
| 17 | use MediaWiki\Extension\WikiLambda\WikiLambdaServices; |
| 18 | use MediaWiki\Extension\WikiLambda\ZErrorFactory; |
| 19 | use MediaWiki\Extension\WikiLambda\ZObjectUtils; |
| 20 | use MediaWiki\MediaWikiServices; |
| 21 | use MediaWiki\Title\Title; |
| 22 | use Wikimedia\ParamValidator\ParamValidator; |
| 23 | use Wikimedia\ParamValidator\TypeDef\IntegerDef; |
| 24 | |
| 25 | class ApiQueryZObjectLabels extends WikiLambdaApiQueryGeneratorBase { |
| 26 | |
| 27 | /** |
| 28 | * @codeCoverageIgnore |
| 29 | */ |
| 30 | public function __construct( ApiQuery $query, string $moduleName ) { |
| 31 | parent::__construct( $query, $moduleName, 'wikilambdasearch_' ); |
| 32 | } |
| 33 | |
| 34 | /** |
| 35 | * @inheritDoc |
| 36 | */ |
| 37 | protected function run( $resultPageSet = null ) { |
| 38 | // Exit if we're running in non-repo mode (e.g. on a client wiki) |
| 39 | if ( !$this->getConfig()->get( 'WikiLambdaEnableRepoMode' ) ) { |
| 40 | WikiLambdaApiBase::dieWithZError( |
| 41 | ZErrorFactory::createZErrorInstance( |
| 42 | ZErrorTypeRegistry::Z_ERROR_USER_CANNOT_RUN, |
| 43 | [] |
| 44 | ), |
| 45 | HttpStatus::BAD_REQUEST |
| 46 | ); |
| 47 | } |
| 48 | |
| 49 | [ |
| 50 | 'search' => $searchTerm, |
| 51 | 'exact' => $exact, |
| 52 | 'language' => $language, |
| 53 | 'type' => $types, |
| 54 | 'return_type' => $returnTypes, |
| 55 | 'limit' => $limit, |
| 56 | 'continue' => $continue, |
| 57 | ] = $this->extractRequestParams(); |
| 58 | |
| 59 | // TODO (T348545): We can reduce this control limit to 100 when we have |
| 60 | // have a system to return results already pre-ranked from the DB. |
| 61 | $controlLimit = 5000; |
| 62 | |
| 63 | $zObjectStore = WikiLambdaServices::getZObjectStore(); |
| 64 | $res = $zObjectStore->searchZObjectLabels( |
| 65 | $searchTerm, |
| 66 | $exact, |
| 67 | [], |
| 68 | $types ?? [], |
| 69 | $returnTypes ?? [], |
| 70 | null, |
| 71 | $controlLimit |
| 72 | ); |
| 73 | |
| 74 | // 1. Set match_rate for every entry and eliminate duplicates with lower match rates |
| 75 | // TODO (T349583): Improve this result sorting algorithm; e.g. should we prioritize matches with primary labels? |
| 76 | $matches = []; |
| 77 | $hasSearchTerm = ( $searchTerm !== '' ); |
| 78 | $matchField = ZObjectUtils::isValidZObjectReference( $searchTerm ) ? 'wlzl_zobject_zid' : 'wlzl_label'; |
| 79 | |
| 80 | foreach ( $res as $row ) { |
| 81 | $matchRate = $hasSearchTerm ? self::getMatchRate( $searchTerm, $row->{ $matchField }, $exact ) : 0; |
| 82 | |
| 83 | // If the current row is new or a better match, keep. Else, ignore. |
| 84 | if ( !array_key_exists( $row->wlzl_zobject_zid, $matches ) || |
| 85 | ( $matches[ $row->wlzl_zobject_zid ][ 'match_rate' ] < $matchRate ) ) { |
| 86 | $matches[ $row->wlzl_zobject_zid ] = [ |
| 87 | // TODO (T338248): Implement, otherwise the generator won't work. |
| 88 | 'page_id' => 0, |
| 89 | // TODO (T258915): When we support redirects, implement. |
| 90 | 'page_is_redirect' => false, |
| 91 | 'page_namespace' => NS_MAIN, |
| 92 | 'page_content_model' => CONTENT_MODEL_ZOBJECT, |
| 93 | 'page_title' => $row->wlzl_zobject_zid, |
| 94 | 'page_type' => $row->wlzl_type, |
| 95 | 'match_label' => $hasSearchTerm ? $row->{ $matchField } : null, |
| 96 | 'match_is_primary' => $hasSearchTerm ? $row->wlzl_label_primary : null, |
| 97 | 'match_lang' => $hasSearchTerm ? $row->wlzl_language : null, |
| 98 | 'match_rate' => $matchRate, |
| 99 | // Labels in the user language will be set after selecting the page |
| 100 | 'label' => null, |
| 101 | 'type_label' => null, |
| 102 | ]; |
| 103 | } |
| 104 | } |
| 105 | |
| 106 | // 2. Sort all results by match_rate to get best hits |
| 107 | usort( $matches, static function ( $a, $b ) { |
| 108 | return $b[ 'match_rate' ] <=> $a[ 'match_rate' ]; |
| 109 | } ); |
| 110 | |
| 111 | // 3. Prune the result set to the limit, slice to requested page, and set continue |
| 112 | $continue = $continue === null ? 0 : intval( $continue ); |
| 113 | $hits = array_slice( $matches, $continue * $limit, $limit ); |
| 114 | $pageSize = count( $matches ) - ( $continue * $limit ); |
| 115 | if ( $pageSize > $limit ) { |
| 116 | $this->setContinueEnumParameter( 'continue', strval( $continue + 1 ) ); |
| 117 | } |
| 118 | |
| 119 | // 4. Add relevant user language labels to each hit: This will be the main |
| 120 | // name shown in the selector, while the match_label set above will be used |
| 121 | // as supporting text when the search text has matched an alias or a label in |
| 122 | // a different language. |
| 123 | foreach ( $hits as $index => $hit ) { |
| 124 | $hits[ $index ][ 'label' ] = $zObjectStore->fetchZObjectLabel( $hit[ 'page_title' ], $language ); |
| 125 | $hits[ $index ][ 'type_label' ] = $zObjectStore->fetchZObjectLabel( $hit[ 'page_type' ], $language ); |
| 126 | } |
| 127 | |
| 128 | if ( $resultPageSet ) { |
| 129 | // TODO (T362192): This needs to be an IResultWrapper, not an array of assoc. objects, irritatingly. |
| 130 | // $resultPageSet->populateFromQueryResult( $dbr, $hits ); |
| 131 | foreach ( $hits as $index => $entry ) { |
| 132 | $resultPageSet->setGeneratorData( |
| 133 | Title::makeTitle( $entry['page_namespace'], $entry['page_title'] ), |
| 134 | [ 'index' => $index + $continue + 1 ] |
| 135 | ); |
| 136 | } |
| 137 | } else { |
| 138 | $result = $this->getResult(); |
| 139 | foreach ( $hits as $entry ) { |
| 140 | $result->addValue( [ 'query', $this->getModuleName() ], null, $entry ); |
| 141 | } |
| 142 | } |
| 143 | } |
| 144 | |
| 145 | /** |
| 146 | * @inheritDoc |
| 147 | * @codeCoverageIgnore |
| 148 | */ |
| 149 | protected function getAllowedParams(): array { |
| 150 | return [ |
| 151 | 'search' => [ |
| 152 | ParamValidator::PARAM_TYPE => 'string', |
| 153 | ParamValidator::PARAM_DEFAULT => '', |
| 154 | ], |
| 155 | 'language' => [ |
| 156 | ParamValidator::PARAM_TYPE => array_keys( |
| 157 | // TODO (T330033): Consider injecting this service rather than just fetching from main |
| 158 | MediaWikiServices::getInstance()->getLanguageNameUtils()->getLanguageNames() |
| 159 | ), |
| 160 | ParamValidator::PARAM_REQUIRED => true, |
| 161 | ], |
| 162 | // This is the wrong way around logically, but MediaWiki's Action API doesn't allow for |
| 163 | // default-true boolean flags to ever be set false. |
| 164 | 'nofallback' => [ |
| 165 | ParamValidator::PARAM_TYPE => 'boolean', |
| 166 | ParamValidator::PARAM_DEFAULT => false, |
| 167 | ], |
| 168 | 'exact' => [ |
| 169 | ParamValidator::PARAM_TYPE => 'boolean', |
| 170 | ParamValidator::PARAM_DEFAULT => false, |
| 171 | ], |
| 172 | 'type' => [ |
| 173 | ParamValidator::PARAM_TYPE => 'string', |
| 174 | ParamValidator::PARAM_ISMULTI => true, |
| 175 | ], |
| 176 | 'return_type' => [ |
| 177 | ParamValidator::PARAM_TYPE => 'string', |
| 178 | ParamValidator::PARAM_ISMULTI => true, |
| 179 | ], |
| 180 | 'limit' => [ |
| 181 | ParamValidator::PARAM_TYPE => 'limit', |
| 182 | ParamValidator::PARAM_DEFAULT => 10, |
| 183 | IntegerDef::PARAM_MIN => 1, |
| 184 | IntegerDef::PARAM_MAX => ApiBase::LIMIT_BIG1, |
| 185 | IntegerDef::PARAM_MAX2 => ApiBase::LIMIT_BIG2, |
| 186 | ], |
| 187 | 'continue' => [ |
| 188 | ApiBase::PARAM_HELP_MSG => 'api-help-param-continue', |
| 189 | ], |
| 190 | ]; |
| 191 | } |
| 192 | |
| 193 | /** |
| 194 | * @see ApiBase::getExamplesMessages() |
| 195 | * @return array |
| 196 | * @codeCoverageIgnore |
| 197 | */ |
| 198 | protected function getExamplesMessages() { |
| 199 | return [ |
| 200 | // search "foo" in language "en" |
| 201 | 'action=query&list=wikilambdasearch_labels&' |
| 202 | . ' wikilambdasearch_search=foo&' |
| 203 | . ' wikilambdasearch_language=en' => 'apihelp-query+wikilambda-example-simple', |
| 204 | // search "foo" in language "fr" without fallbacks |
| 205 | 'action=query&list=wikilambdasearch_labels&' |
| 206 | . 'wikilambdasearch_search=foo&' |
| 207 | . 'wikilambdasearch_language=fr&' |
| 208 | . 'wikilambdasearch_nofallback=true' => 'apihelp-query+wikilambda-example-nofallback', |
| 209 | // Search for objects of type "Z4" |
| 210 | 'action=query&list=wikilambdasearch_labels&' |
| 211 | . 'wikilambdasearch_type=Z4&' |
| 212 | . 'wikilambdasearch_language=en' => 'apihelp-query+wikilambda-example-type', |
| 213 | // Search for objects that resolve to "Z40" |
| 214 | 'action=query&list=wikilambdasearch_labels&' |
| 215 | . 'wikilambdasearch_return_type=Z40&' |
| 216 | . 'wikilambdasearch_language=en' => 'apihelp-query+wikilambda-example-return-type', |
| 217 | // Search for functions that output "Z40" or "Z1" |
| 218 | 'action=query&list=wikilambdasearch_labels&' |
| 219 | . 'wikilambdasearch_type=Z8&' |
| 220 | . 'wikilambdasearch_return_type=Z40|Z1&' |
| 221 | . 'wikilambdasearch_language=en' => 'apihelp-query+wikilambda-example-type-and-return-types', |
| 222 | // Search for function calls equivalent to "Z4" or literal "Z4" objects |
| 223 | 'action=query&list=wikilambdasearch_labels&' |
| 224 | . 'wikilambdasearch_type=Z4|Z7&' |
| 225 | . 'wikilambdasearch_return_type=Z4&' |
| 226 | . 'wikilambdasearch_language=en' => 'apihelp-query+wikilambda-example-types-and-return-type', |
| 227 | ]; |
| 228 | } |
| 229 | } |