Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
5.04% |
7 / 139 |
|
7.14% |
1 / 14 |
CRAP | |
0.00% |
0 / 1 |
| Hooks | |
5.04% |
7 / 139 |
|
7.14% |
1 / 14 |
2839.45 | |
0.00% |
0 / 1 |
| __construct | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| onParserFirstCallInit | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
2 | |||
| onArticleDeleteComplete | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
2 | |||
| onLinksUpdateComplete | |
0.00% |
0 / 11 |
|
0.00% |
0 / 1 |
30 | |||
| getCoordinatesIfFile | |
0.00% |
0 / 17 |
|
0.00% |
0 / 1 |
72 | |||
| doLinksUpdate | |
0.00% |
0 / 39 |
|
0.00% |
0 / 1 |
210 | |||
| onFileUpload | |
0.00% |
0 / 11 |
|
0.00% |
0 / 1 |
6 | |||
| onOutputPageParserOutput | |
0.00% |
0 / 11 |
|
0.00% |
0 / 1 |
56 | |||
| onSearchIndexFields | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
20 | |||
| onSearchDataForIndex | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| onSearchDataForIndex2 | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| doSearchDataForIndex | |
0.00% |
0 / 17 |
|
0.00% |
0 / 1 |
56 | |||
| coordToElastic | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
2 | |||
| createQueryGeoSearchBackend | |
100.00% |
7 / 7 |
|
100.00% |
1 / 1 |
4 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace GeoData; |
| 4 | |
| 5 | use ApiQuery; |
| 6 | use CirrusSearch\CirrusSearch; |
| 7 | use Content; |
| 8 | use ContentHandler; |
| 9 | use File; |
| 10 | use GeoData\Api\QueryGeoSearch; |
| 11 | use GeoData\Api\QueryGeoSearchDb; |
| 12 | use GeoData\Api\QueryGeoSearchElastic; |
| 13 | use GeoData\Search\CoordinatesIndexField; |
| 14 | use ManualLogEntry; |
| 15 | use MediaWiki\Config\Config; |
| 16 | use MediaWiki\Content\Hook\SearchDataForIndexHook; |
| 17 | use MediaWiki\Deferred\DeferredUpdates; |
| 18 | use MediaWiki\Deferred\LinksUpdate\LinksUpdate; |
| 19 | use MediaWiki\Hook\FileUploadHook; |
| 20 | use MediaWiki\Hook\LinksUpdateCompleteHook; |
| 21 | use MediaWiki\Hook\ParserFirstCallInitHook; |
| 22 | use MediaWiki\Linker\LinkTarget; |
| 23 | use MediaWiki\MainConfigNames; |
| 24 | use MediaWiki\MediaWikiServices; |
| 25 | use MediaWiki\Output\Hook\OutputPageParserOutputHook; |
| 26 | use MediaWiki\Page\Hook\ArticleDeleteCompleteHook; |
| 27 | use MediaWiki\Parser\Parser; |
| 28 | use MediaWiki\Parser\ParserOutput; |
| 29 | use MediaWiki\Revision\RevisionRecord; |
| 30 | use MediaWiki\User\User; |
| 31 | use SearchEngine; |
| 32 | use WikiPage; |
| 33 | |
| 34 | /** |
| 35 | * Hook handlers |
| 36 | * @todo: tests |
| 37 | */ |
| 38 | class Hooks implements |
| 39 | SearchDataForIndexHook, |
| 40 | OutputPageParserOutputHook, |
| 41 | ParserFirstCallInitHook, |
| 42 | ArticleDeleteCompleteHook, |
| 43 | LinksUpdateCompleteHook, |
| 44 | FileUploadHook |
| 45 | { |
| 46 | |
| 47 | private Config $config; |
| 48 | |
| 49 | public function __construct( Config $config ) { |
| 50 | $this->config = $config; |
| 51 | } |
| 52 | |
| 53 | /** |
| 54 | * ParserFirstCallInit hook handler |
| 55 | * @see https://www.mediawiki.org/wiki/Manual:Hooks/ParserFirstCallInit |
| 56 | * |
| 57 | * @param Parser $parser |
| 58 | */ |
| 59 | public function onParserFirstCallInit( $parser ) { |
| 60 | $parser->setFunctionHook( 'coordinates', |
| 61 | [ new CoordinatesParserFunction(), 'coordinates' ], |
| 62 | Parser::SFH_OBJECT_ARGS |
| 63 | ); |
| 64 | } |
| 65 | |
| 66 | /** |
| 67 | * ArticleDeleteComplete hook handler |
| 68 | * @see https://www.mediawiki.org/wiki/Manual:Hooks/ArticleDeleteComplete |
| 69 | * |
| 70 | * @param WikiPage $article |
| 71 | * @param User $user |
| 72 | * @param string $reason |
| 73 | * @param int $id |
| 74 | * @param Content|null $content |
| 75 | * @param ManualLogEntry $logEntry |
| 76 | * @param int $archivedRevisionCount |
| 77 | */ |
| 78 | public function onArticleDeleteComplete( $article, $user, $reason, $id, |
| 79 | $content, $logEntry, $archivedRevisionCount |
| 80 | ) { |
| 81 | $dbw = MediaWikiServices::getInstance()->getConnectionProvider()->getPrimaryDatabase(); |
| 82 | $dbw->newDeleteQueryBuilder() |
| 83 | ->deleteFrom( 'geo_tags' ) |
| 84 | ->where( [ 'gt_page_id' => $id ] ) |
| 85 | ->caller( __METHOD__ ) |
| 86 | ->execute(); |
| 87 | } |
| 88 | |
| 89 | /** |
| 90 | * LinksUpdateComplete hook handler |
| 91 | * @see https://www.mediawiki.org/wiki/Manual:Hooks/LinksUpdateComplete |
| 92 | * |
| 93 | * @param LinksUpdate $linksUpdate |
| 94 | * @param int|null $ticket |
| 95 | */ |
| 96 | public function onLinksUpdateComplete( $linksUpdate, $ticket ) { |
| 97 | $out = $linksUpdate->getParserOutput(); |
| 98 | $data = []; |
| 99 | $coordFromMetadata = self::getCoordinatesIfFile( $linksUpdate->getTitle() ); |
| 100 | $coordsOutput = CoordinatesOutput::getFromParserOutput( $out ); |
| 101 | if ( $coordsOutput ) { |
| 102 | // Use coordinates from file metadata unless overridden on description page |
| 103 | if ( $coordFromMetadata && !$coordsOutput->hasPrimary() ) { |
| 104 | $coordsOutput->addPrimary( $coordFromMetadata ); |
| 105 | } |
| 106 | $data = $coordsOutput->getAll(); |
| 107 | } elseif ( $coordFromMetadata ) { |
| 108 | $data[] = $coordFromMetadata; |
| 109 | } |
| 110 | self::doLinksUpdate( $data, $linksUpdate->getPageId(), $ticket ); |
| 111 | } |
| 112 | |
| 113 | private static function getCoordinatesIfFile( LinkTarget $title ): ?Coord { |
| 114 | if ( !$title->inNamespace( NS_FILE ) ) { |
| 115 | return null; |
| 116 | } |
| 117 | $file = MediaWikiServices::getInstance()->getRepoGroup()->getLocalRepo() |
| 118 | ->findFile( $title, [ 'ignoreRedirect' => true ] ); |
| 119 | if ( !$file ) { |
| 120 | return null; |
| 121 | } |
| 122 | $metadata = $file->getMetadataItems( [ 'GPSLatitude', 'GPSLongitude' ] ); |
| 123 | if ( isset( $metadata['GPSLatitude'] ) && isset( $metadata['GPSLongitude'] ) ) { |
| 124 | $lat = $metadata['GPSLatitude']; |
| 125 | $lon = $metadata['GPSLongitude']; |
| 126 | $globe = new Globe(); |
| 127 | // T165800: Skip files with meaningless 0, 0 coordinates |
| 128 | if ( ( $lat || $lon ) && |
| 129 | $globe->coordinatesAreValid( $lat, $lon ) |
| 130 | ) { |
| 131 | $coord = new Coord( $lat, $lon, $globe->getName() ); |
| 132 | $coord->primary = true; |
| 133 | return $coord; |
| 134 | } |
| 135 | } |
| 136 | return null; |
| 137 | } |
| 138 | |
| 139 | /** |
| 140 | * @param Coord[] $coords |
| 141 | * @param int $pageId |
| 142 | * @param int|null $ticket |
| 143 | * @throws \Wikimedia\Rdbms\DBUnexpectedError |
| 144 | */ |
| 145 | private static function doLinksUpdate( array $coords, $pageId, $ticket ) { |
| 146 | $services = MediaWikiServices::getInstance(); |
| 147 | $config = $services->getMainConfig(); |
| 148 | $indexGranularity = $config->get( 'GeoDataBackend' ) === 'db' ? |
| 149 | $config->get( 'GeoDataIndexGranularity' ) : null; |
| 150 | |
| 151 | $add = []; |
| 152 | $delete = []; |
| 153 | $primary = ( isset( $coords[0] ) && $coords[0]->primary ) ? $coords[0] : null; |
| 154 | foreach ( GeoData::getAllCoordinates( $pageId, [], DB_PRIMARY ) as $old ) { |
| 155 | $delete[$old->id] = $old; |
| 156 | } |
| 157 | foreach ( $coords as $new ) { |
| 158 | if ( !$new->primary && $new->equalsTo( $primary ) ) { |
| 159 | // Don't save secondary coordinates pointing to the same place as the primary one |
| 160 | continue; |
| 161 | } |
| 162 | $match = false; |
| 163 | foreach ( $delete as $id => $old ) { |
| 164 | if ( $new->fullyEqualsTo( $old ) ) { |
| 165 | unset( $delete[$id] ); |
| 166 | $match = true; |
| 167 | break; |
| 168 | } |
| 169 | } |
| 170 | if ( !$match ) { |
| 171 | $add[] = $new->getRow( $pageId, $indexGranularity ); |
| 172 | } |
| 173 | } |
| 174 | |
| 175 | $dbw = $services->getConnectionProvider()->getPrimaryDatabase(); |
| 176 | $lbFactory = $services->getDBLoadBalancerFactory(); |
| 177 | $ticket = $ticket ?: $lbFactory->getEmptyTransactionTicket( __METHOD__ ); |
| 178 | $batchSize = $config->get( MainConfigNames::UpdateRowsPerQuery ); |
| 179 | |
| 180 | $deleteIds = array_keys( $delete ); |
| 181 | foreach ( array_chunk( $deleteIds, $batchSize ) as $deleteIdBatch ) { |
| 182 | $dbw->newDeleteQueryBuilder() |
| 183 | ->deleteFrom( 'geo_tags' ) |
| 184 | ->where( [ 'gt_id' => $deleteIdBatch ] ) |
| 185 | ->caller( __METHOD__ ) |
| 186 | ->execute(); |
| 187 | $lbFactory->commitAndWaitForReplication( __METHOD__, $ticket ); |
| 188 | } |
| 189 | |
| 190 | foreach ( array_chunk( $add, $batchSize ) as $addBatch ) { |
| 191 | $dbw->newInsertQueryBuilder() |
| 192 | ->insertInto( 'geo_tags' ) |
| 193 | ->rows( $addBatch ) |
| 194 | ->caller( __METHOD__ ) |
| 195 | ->execute(); |
| 196 | $lbFactory->commitAndWaitForReplication( __METHOD__, $ticket ); |
| 197 | } |
| 198 | } |
| 199 | |
| 200 | /** |
| 201 | * FileUpload hook handler |
| 202 | * @see https://www.mediawiki.org/wiki/Manual:Hooks/FileUpload |
| 203 | * |
| 204 | * @param File $file |
| 205 | * @param bool $reupload |
| 206 | * @param bool $hasDescription |
| 207 | */ |
| 208 | public function onFileUpload( $file, $reupload, $hasDescription ) { |
| 209 | $wp = MediaWikiServices::getInstance()->getWikiPageFactory()->newFromTitle( $file->getTitle() ); |
| 210 | $po = $wp->makeParserOptions( 'canonical' ); |
| 211 | $pout = $wp->getParserOutput( $po ); |
| 212 | if ( !$pout ) { |
| 213 | wfDebugLog( 'mobile', |
| 214 | __METHOD__ . "(): no parser output returned for file {$file->getName()}" |
| 215 | ); |
| 216 | } else { |
| 217 | // Make sure this has outer transaction scope (though the hook fires |
| 218 | // in a deferred AutoCommitUdpate update, so it should be safe anyway). |
| 219 | $lu = new LinksUpdate( $file->getTitle(), $pout ); |
| 220 | DeferredUpdates::addCallableUpdate( function () use ( $lu ) { |
| 221 | $this->onLinksUpdateComplete( $lu, null ); |
| 222 | } ); |
| 223 | } |
| 224 | } |
| 225 | |
| 226 | /** |
| 227 | * @inheritDoc |
| 228 | */ |
| 229 | public function onOutputPageParserOutput( $out, $po ): void { |
| 230 | $geoDataInJS = $this->config->get( 'GeoDataInJS' ); |
| 231 | |
| 232 | if ( $geoDataInJS && CoordinatesOutput::getFromParserOutput( $po ) ) { |
| 233 | $coord = CoordinatesOutput::getFromParserOutput( $po )->getPrimary(); |
| 234 | if ( !$coord ) { |
| 235 | return; |
| 236 | } |
| 237 | $result = []; |
| 238 | foreach ( $geoDataInJS as $param ) { |
| 239 | if ( isset( $coord->$param ) ) { |
| 240 | $result[$param] = $coord->$param; |
| 241 | } |
| 242 | } |
| 243 | if ( $result ) { |
| 244 | $out->addJsConfigVars( 'wgCoordinates', $result ); |
| 245 | } |
| 246 | } |
| 247 | } |
| 248 | |
| 249 | /** |
| 250 | * Search index fields hook handler |
| 251 | * Adds our stuff to CirrusSearch/Elasticsearch schema |
| 252 | * |
| 253 | * @param array &$fields |
| 254 | * @param SearchEngine $engine |
| 255 | */ |
| 256 | public function onSearchIndexFields( array &$fields, SearchEngine $engine ) { |
| 257 | $useCirrus = $this->config->get( 'GeoDataUseCirrusSearch' ); |
| 258 | $backend = $this->config->get( 'GeoDataBackend' ); |
| 259 | if ( !$useCirrus && $backend !== 'elastic' ) { |
| 260 | return; |
| 261 | } |
| 262 | if ( $engine instanceof CirrusSearch ) { |
| 263 | /** |
| 264 | * @var CirrusSearch $engine |
| 265 | */ |
| 266 | $fields['coordinates'] = CoordinatesIndexField::build( |
| 267 | 'coordinates', $engine->getConfig(), $engine ); |
| 268 | } else { |
| 269 | // Unsupported SearchEngine or explicitly disabled by config |
| 270 | } |
| 271 | } |
| 272 | |
| 273 | /** |
| 274 | * @inheritDoc |
| 275 | */ |
| 276 | public function onSearchDataForIndex( |
| 277 | &$fields, |
| 278 | $handler, |
| 279 | $page, |
| 280 | $output, |
| 281 | $engine |
| 282 | ) { |
| 283 | self::doSearchDataForIndex( $fields, $output, $page ); |
| 284 | } |
| 285 | |
| 286 | /** |
| 287 | * SearchDataForIndex hook handler |
| 288 | * |
| 289 | * @param array &$fields |
| 290 | * @param ContentHandler $handler |
| 291 | * @param WikiPage $page |
| 292 | * @param ParserOutput $output |
| 293 | * @param SearchEngine $engine |
| 294 | * @param RevisionRecord $revision |
| 295 | */ |
| 296 | public function onSearchDataForIndex2( |
| 297 | array &$fields, |
| 298 | ContentHandler $handler, |
| 299 | WikiPage $page, |
| 300 | ParserOutput $output, |
| 301 | SearchEngine $engine, |
| 302 | RevisionRecord $revision |
| 303 | ) { |
| 304 | self::doSearchDataForIndex( $fields, $output, $page ); |
| 305 | } |
| 306 | |
| 307 | /** |
| 308 | * Attach coordinates to the index document |
| 309 | * |
| 310 | * @param array &$fields |
| 311 | * @param ParserOutput $parserOutput |
| 312 | * @param WikiPage $page |
| 313 | * @return void |
| 314 | */ |
| 315 | private function doSearchDataForIndex( array &$fields, ParserOutput $parserOutput, WikiPage $page ): void { |
| 316 | $useCirrus = $this->config->get( 'GeoDataUseCirrusSearch' ); |
| 317 | $backend = $this->config->get( 'GeoDataBackend' ); |
| 318 | |
| 319 | if ( ( $useCirrus || $backend == 'elastic' ) ) { |
| 320 | $coordsOutput = CoordinatesOutput::getFromParserOutput( $parserOutput ); |
| 321 | $allCoords = $coordsOutput !== null ? $coordsOutput->getAll() : []; |
| 322 | $coords = []; |
| 323 | |
| 324 | /** @var Coord $coord */ |
| 325 | foreach ( $allCoords as $coord ) { |
| 326 | if ( $coord->globe !== Globe::EARTH ) { |
| 327 | continue; |
| 328 | } |
| 329 | if ( !$coord->isValid() ) { |
| 330 | wfDebugLog( 'CirrusSearchChangeFailed', |
| 331 | "Invalid coordinates [{$coord->lat}, {$coord->lon}] on page " |
| 332 | . $page->getTitle()->getPrefixedText() |
| 333 | ); |
| 334 | continue; |
| 335 | } |
| 336 | $coords[] = self::coordToElastic( $coord ); |
| 337 | } |
| 338 | $fields['coordinates'] = $coords; |
| 339 | } |
| 340 | } |
| 341 | |
| 342 | /** |
| 343 | * Transforms coordinates into an array for insertion onto Elasticsearch |
| 344 | * |
| 345 | * @param Coord $coord |
| 346 | * @return array |
| 347 | */ |
| 348 | public static function coordToElastic( Coord $coord ) { |
| 349 | $result = $coord->getAsArray(); |
| 350 | $result['coord'] = [ 'lat' => $coord->lat, 'lon' => $coord->lon ]; |
| 351 | unset( $result['id'] ); |
| 352 | unset( $result['lat'] ); |
| 353 | unset( $result['lon'] ); |
| 354 | |
| 355 | return $result; |
| 356 | } |
| 357 | |
| 358 | /** |
| 359 | * @param ApiQuery $query |
| 360 | * @param string $moduleName |
| 361 | * @return QueryGeoSearch |
| 362 | */ |
| 363 | public static function createQueryGeoSearchBackend( ApiQuery $query, $moduleName ): QueryGeoSearch { |
| 364 | $geoDataBackend = $query->getConfig()->get( 'GeoDataBackend' ); |
| 365 | |
| 366 | switch ( strtolower( $geoDataBackend ) ) { |
| 367 | case 'db': |
| 368 | return new QueryGeoSearchDb( $query, $moduleName ); |
| 369 | case 'elastic': |
| 370 | return new QueryGeoSearchElastic( $query, $moduleName ); |
| 371 | default: |
| 372 | throw new \RuntimeException( 'GeoDataBackend data backend cannot be empty' ); |
| 373 | } |
| 374 | } |
| 375 | } |