Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
17.33% |
13 / 75 |
|
9.09% |
1 / 11 |
CRAP | |
0.00% |
0 / 1 |
| WikiProjectFullLookup | |
17.33% |
13 / 75 |
|
9.09% |
1 / 11 |
378.08 | |
0.00% |
0 / 1 |
| __construct | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
2 | |||
| getWikiProjects | |
100.00% |
13 / 13 |
|
100.00% |
1 / 1 |
6 | |||
| hasWikiProjects | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| hasWikiProjectsAfter | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
12 | |||
| isKnownEntity | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getDataForEntities | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
2 | |||
| computeDataForEntities | |
0.00% |
0 / 13 |
|
0.00% |
0 / 1 |
12 | |||
| queryWikidataAPI | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
6 | |||
| queryWikidataAPIBatch | |
0.00% |
0 / 22 |
|
0.00% |
0 / 1 |
12 | |||
| buildEntitySiteLink | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
6 | |||
| invertDirection | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
6 | |||
| 1 | <?php |
| 2 | |
| 3 | declare( strict_types=1 ); |
| 4 | |
| 5 | namespace MediaWiki\Extension\WikimediaCampaignEvents\WikiProject; |
| 6 | |
| 7 | use InvalidArgumentException; |
| 8 | use JsonException; |
| 9 | use MediaWiki\Http\HttpRequestFactory; |
| 10 | use MediaWiki\WikiMap\WikiMap; |
| 11 | use Wikimedia\ObjectCache\WANObjectCache; |
| 12 | |
| 13 | /** |
| 14 | * This is the main lookup service for WikiProject data, intended to be used with an interface that supports pagination. |
| 15 | * Given a number of results and a starting entity to enumerate from, it queries Wikidata to get additional information |
| 16 | * about WikiProjects (such as labels and descriptions). |
| 17 | */ |
| 18 | class WikiProjectFullLookup { |
| 19 | public const SERVICE_NAME = 'WikimediaCampaignEventsWikiProjectFullLookup'; |
| 20 | |
| 21 | public const DIR_FORWARDS = 1; |
| 22 | public const DIR_BACKWARDS = 2; |
| 23 | |
| 24 | private WikiProjectIDLookup $wikiProjectIDLookup; |
| 25 | private WANObjectCache $cache; |
| 26 | private HttpRequestFactory $httpRequestFactory; |
| 27 | |
| 28 | public function __construct( |
| 29 | WikiProjectIDLookup $wikiProjectIDLookup, |
| 30 | WANObjectCache $cache, |
| 31 | HttpRequestFactory $httpRequestFactory |
| 32 | ) { |
| 33 | $this->wikiProjectIDLookup = $wikiProjectIDLookup; |
| 34 | $this->cache = $cache; |
| 35 | $this->httpRequestFactory = $httpRequestFactory; |
| 36 | } |
| 37 | |
| 38 | /** |
| 39 | * @param string $languageCode |
| 40 | * @param int $limit |
| 41 | * @param string|null $lastEntity When paginating results, this is the ID of the last entity shown to the user |
| 42 | * (i.e., the last row when going forwards, and the first row when going backwards). |
| 43 | * @param int $direction In which direction to scan, self::DIR_FORWARDS or self::DIR_BACKWARDS |
| 44 | * @return array<array|null> An array of arrays with information about the requested WikiProjects. Elements might be |
| 45 | * null if there is no sitelink for the current wiki (might happen when the sitelink is removed after the WDQS query |
| 46 | * was last run). QIDs are used as array keys, even for null elements. |
| 47 | * @phan-return array<string,array{label:string,description:string,sitelink:string}|null> |
| 48 | * @throws CannotQueryWDQSException |
| 49 | * @throws CannotQueryWikibaseException |
| 50 | */ |
| 51 | public function getWikiProjects( |
| 52 | string $languageCode, |
| 53 | int $limit, |
| 54 | ?string $lastEntity = null, |
| 55 | int $direction = self::DIR_FORWARDS |
| 56 | ): array { |
| 57 | $allIDs = $this->wikiProjectIDLookup->getWikiProjectIDs(); |
| 58 | $lastPos = false; |
| 59 | if ( $lastEntity !== null ) { |
| 60 | $lastPos = array_search( $lastEntity, $allIDs, true ); |
| 61 | } |
| 62 | |
| 63 | if ( $lastPos !== false ) { |
| 64 | if ( $direction === self::DIR_FORWARDS ) { |
| 65 | $wantedIDs = array_slice( $allIDs, $lastPos + 1, $limit ); |
| 66 | } elseif ( $lastPos > $limit ) { |
| 67 | $wantedIDs = array_slice( $allIDs, $lastPos - $limit, $limit ); |
| 68 | } else { |
| 69 | $wantedIDs = array_slice( $allIDs, 0, $lastPos ); |
| 70 | } |
| 71 | } else { |
| 72 | $offset = $direction === self::DIR_FORWARDS ? 0 : -$limit; |
| 73 | $wantedIDs = array_slice( $allIDs, $offset, $limit ); |
| 74 | } |
| 75 | return $this->getDataForEntities( $wantedIDs, $languageCode ); |
| 76 | } |
| 77 | |
| 78 | /** |
| 79 | * @return bool Whether any WikiProjects exist on the current wiki. |
| 80 | * @throws CannotQueryWDQSException |
| 81 | */ |
| 82 | public function hasWikiProjects(): bool { |
| 83 | return $this->wikiProjectIDLookup->getWikiProjectIDs() !== []; |
| 84 | } |
| 85 | |
| 86 | /** |
| 87 | * @param string $lastID Entity ID to check. The caller must verify that this is a valid ID, or an exception |
| 88 | * will be thrown. |
| 89 | * @param int $direction self::DIR_FORWARDS or self::DIR_BACKWARDS |
| 90 | * @return bool Whether any WikiProjects exist after the specified offset in the given direction (for pagination). |
| 91 | */ |
| 92 | public function hasWikiProjectsAfter( string $lastID, int $direction ): bool { |
| 93 | $allIDs = $this->wikiProjectIDLookup->getWikiProjectIDs(); |
| 94 | $offsetKey = array_search( $lastID, $allIDs, true ); |
| 95 | if ( $offsetKey === false ) { |
| 96 | throw new InvalidArgumentException( "Entity $lastID not found." ); |
| 97 | } |
| 98 | return $direction === self::DIR_FORWARDS |
| 99 | ? $offsetKey < array_key_last( $allIDs ) |
| 100 | : $offsetKey > 0; |
| 101 | } |
| 102 | |
| 103 | /** |
| 104 | * @param string $entityID |
| 105 | * @return bool Whether the given ID corresponds to a known entity. |
| 106 | * @throws CannotQueryWikiProjectsException |
| 107 | */ |
| 108 | public function isKnownEntity( string $entityID ): bool { |
| 109 | return in_array( $entityID, $this->wikiProjectIDLookup->getWikiProjectIDs(), true ); |
| 110 | } |
| 111 | |
| 112 | /** |
| 113 | * @param array $entityIDs |
| 114 | * @param string $languageCode |
| 115 | * @return array<array|null> |
| 116 | * @phan-return array<string,array{label:string,description:string,sitelink:string}|null> |
| 117 | * @throws CannotQueryWikibaseException |
| 118 | */ |
| 119 | private function getDataForEntities( array $entityIDs, string $languageCode ): array { |
| 120 | $entitiesHash = sha1( implode( ',', $entityIDs ) ); |
| 121 | return $this->cache->getWithSetCallback( |
| 122 | // Can be cached globally, since entity IDs are unique. |
| 123 | $this->cache->makeGlobalKey( 'WikimediaCampaignEvents-WikiProjects', $languageCode, $entitiesHash ), |
| 124 | WANObjectCache::TTL_HOUR, |
| 125 | fn () => $this->computeDataForEntities( $entityIDs, $languageCode ) |
| 126 | ); |
| 127 | } |
| 128 | |
| 129 | /** |
| 130 | * @param string[] $entityIDs |
| 131 | * @param string $languageCode |
| 132 | * @return array<array|null> |
| 133 | * @phan-return array<string,array{label:string,description:string,sitelink:string}|null> |
| 134 | * @throws CannotQueryWikibaseException |
| 135 | */ |
| 136 | private function computeDataForEntities( array $entityIDs, string $languageCode ): array { |
| 137 | $entities = $this->queryWikidataAPI( $entityIDs, $languageCode ); |
| 138 | $wikiProjects = []; |
| 139 | foreach ( $entities as $id => $entity ) { |
| 140 | $siteLink = $this->buildEntitySiteLink( $entity ); |
| 141 | $wikiProjectData = null; |
| 142 | if ( $siteLink ) { |
| 143 | $wikiProjectData = [ |
| 144 | 'label' => $entity['labels'][$languageCode]['value'] ?? '', |
| 145 | 'description' => $entity['descriptions'][$languageCode]['value'] ?? '', |
| 146 | 'sitelink' => $siteLink, |
| 147 | ]; |
| 148 | } |
| 149 | $wikiProjects[$id] = $wikiProjectData; |
| 150 | } |
| 151 | return $wikiProjects; |
| 152 | } |
| 153 | |
| 154 | /** |
| 155 | * @param string[] $entityIDs |
| 156 | * @param string $languageCode |
| 157 | * @return array[] |
| 158 | * @phan-return array<string,array{labels:array,descriptions:array}> |
| 159 | * @throws CannotQueryWikibaseException |
| 160 | */ |
| 161 | private function queryWikidataAPI( array $entityIDs, string $languageCode ): array { |
| 162 | $batches = array_chunk( $entityIDs, 50 ); |
| 163 | $entities = []; |
| 164 | foreach ( $batches as $batch ) { |
| 165 | $batchResponse = $this->queryWikidataAPIBatch( $batch, $languageCode ); |
| 166 | $entities = array_merge( $entities, $batchResponse['entities'] ); |
| 167 | } |
| 168 | return $entities; |
| 169 | } |
| 170 | |
| 171 | /** |
| 172 | * @param string[] $entityIDs |
| 173 | * @param string $languageCode |
| 174 | * @return array |
| 175 | * @throws CannotQueryWikibaseException |
| 176 | */ |
| 177 | private function queryWikidataAPIBatch( array $entityIDs, string $languageCode ): array { |
| 178 | // 'claims' to be added for more data. |
| 179 | $props = [ 'labels', 'descriptions', 'sitelinks/urls' ]; |
| 180 | $params = [ |
| 181 | 'action' => 'wbgetentities', |
| 182 | 'format' => 'json', |
| 183 | 'ids' => implode( '|', $entityIDs ), |
| 184 | 'props' => implode( '|', $props ), |
| 185 | 'languages' => $languageCode, |
| 186 | 'languagefallback' => true, |
| 187 | 'formatversion' => 2, |
| 188 | ]; |
| 189 | $url = 'https://www.wikidata.org/w/api.php' . '?' . http_build_query( $params ); |
| 190 | $options = [ |
| 191 | 'method' => 'GET' |
| 192 | ]; |
| 193 | $req = $this->httpRequestFactory->create( $url, $options, __METHOD__ ); |
| 194 | |
| 195 | $status = $req->execute(); |
| 196 | if ( !$status->isGood() ) { |
| 197 | throw new CannotQueryWikibaseException( "Bad status from WD API: $status" ); |
| 198 | } |
| 199 | |
| 200 | try { |
| 201 | $parsedResponse = json_decode( $req->getContent(), true, 512, JSON_THROW_ON_ERROR ); |
| 202 | } catch ( JsonException $e ) { |
| 203 | throw new CannotQueryWikibaseException( "Invalid JSON from WD API", 0, $e ); |
| 204 | } |
| 205 | |
| 206 | return $parsedResponse; |
| 207 | } |
| 208 | |
| 209 | /** |
| 210 | * @param array $entity |
| 211 | * @return string|null |
| 212 | */ |
| 213 | private function buildEntitySiteLink( array $entity ): ?string { |
| 214 | $siteId = WikiMap::getCurrentWikiId(); |
| 215 | return array_key_exists( $siteId, $entity['sitelinks'] ) ? $entity['sitelinks'][$siteId]['url'] : null; |
| 216 | } |
| 217 | |
| 218 | public static function invertDirection( int $direction ): int { |
| 219 | return $direction === self::DIR_FORWARDS ? self::DIR_BACKWARDS : self::DIR_FORWARDS; |
| 220 | } |
| 221 | } |