Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 128 |
|
0.00% |
0 / 13 |
CRAP | |
0.00% |
0 / 1 |
| Hooks | |
0.00% |
0 / 128 |
|
0.00% |
0 / 13 |
2550 | |
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 / 16 |
|
0.00% |
0 / 1 |
42 | |||
| doLinksUpdate | |
0.00% |
0 / 36 |
|
0.00% |
0 / 1 |
210 | |||
| onFileUpload | |
0.00% |
0 / 11 |
|
0.00% |
0 / 1 |
6 | |||
| onOutputPageParserOutput | |
0.00% |
0 / 12 |
|
0.00% |
0 / 1 |
42 | |||
| onSearchIndexFields | |
0.00% |
0 / 6 |
|
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 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace GeoData; |
| 4 | |
| 5 | use CirrusSearch\CirrusSearch; |
| 6 | use GeoData\Search\CoordinatesIndexField; |
| 7 | use MediaWiki\Config\Config; |
| 8 | use MediaWiki\Content\Content; |
| 9 | use MediaWiki\Content\ContentHandler; |
| 10 | use MediaWiki\Content\Hook\SearchDataForIndexHook; |
| 11 | use MediaWiki\Deferred\DeferredUpdates; |
| 12 | use MediaWiki\Deferred\Hook\LinksUpdateCompleteHook; |
| 13 | use MediaWiki\Deferred\LinksUpdate\LinksUpdate; |
| 14 | use MediaWiki\FileRepo\File\File; |
| 15 | use MediaWiki\FileRepo\Hook\FileUploadHook; |
| 16 | use MediaWiki\FileRepo\RepoGroup; |
| 17 | use MediaWiki\Linker\LinkTarget; |
| 18 | use MediaWiki\Logging\ManualLogEntry; |
| 19 | use MediaWiki\MainConfigNames; |
| 20 | use MediaWiki\Output\Hook\OutputPageParserOutputHook; |
| 21 | use MediaWiki\Page\Hook\ArticleDeleteCompleteHook; |
| 22 | use MediaWiki\Page\WikiPage; |
| 23 | use MediaWiki\Page\WikiPageFactory; |
| 24 | use MediaWiki\Parser\Hook\ParserFirstCallInitHook; |
| 25 | use MediaWiki\Parser\Parser; |
| 26 | use MediaWiki\Parser\ParserOutput; |
| 27 | use MediaWiki\Revision\RevisionRecord; |
| 28 | use MediaWiki\Search\SearchEngine; |
| 29 | use MediaWiki\User\User; |
| 30 | use Wikimedia\Rdbms\IConnectionProvider; |
| 31 | use Wikimedia\Rdbms\LBFactory; |
| 32 | |
| 33 | /** |
| 34 | * Hook handlers |
| 35 | * @todo tests |
| 36 | */ |
| 37 | class Hooks implements |
| 38 | SearchDataForIndexHook, |
| 39 | OutputPageParserOutputHook, |
| 40 | ParserFirstCallInitHook, |
| 41 | ArticleDeleteCompleteHook, |
| 42 | LinksUpdateCompleteHook, |
| 43 | FileUploadHook |
| 44 | { |
| 45 | |
| 46 | public function __construct( |
| 47 | private readonly Config $config, |
| 48 | private readonly IConnectionProvider $connectionProvider, |
| 49 | private readonly LBFactory $lbFactory, |
| 50 | private readonly RepoGroup $repoGroup, |
| 51 | private readonly WikiPageFactory $wikiPageFactory, |
| 52 | ) { |
| 53 | } |
| 54 | |
| 55 | /** |
| 56 | * ParserFirstCallInit hook handler |
| 57 | * @see https://www.mediawiki.org/wiki/Manual:Hooks/ParserFirstCallInit |
| 58 | * |
| 59 | * @param Parser $parser |
| 60 | */ |
| 61 | public function onParserFirstCallInit( $parser ) { |
| 62 | $parser->setFunctionHook( 'coordinates', |
| 63 | ( new CoordinatesParserFunction( $this->config ) )->coordinates( ... ), |
| 64 | Parser::SFH_OBJECT_ARGS |
| 65 | ); |
| 66 | } |
| 67 | |
| 68 | /** |
| 69 | * ArticleDeleteComplete hook handler |
| 70 | * |
| 71 | * @see https://www.mediawiki.org/wiki/Manual:Hooks/ArticleDeleteComplete |
| 72 | * |
| 73 | * @param WikiPage $wikiPage |
| 74 | * @param User $user |
| 75 | * @param string $reason |
| 76 | * @param int $id |
| 77 | * @param Content|null $content |
| 78 | * @param ManualLogEntry $logEntry |
| 79 | * @param int $archivedRevisionCount |
| 80 | */ |
| 81 | public function onArticleDeleteComplete( |
| 82 | $wikiPage, $user, $reason, $id, |
| 83 | $content, $logEntry, $archivedRevisionCount |
| 84 | ) { |
| 85 | $dbw = $this->connectionProvider->getPrimaryDatabase(); |
| 86 | $dbw->newDeleteQueryBuilder() |
| 87 | ->deleteFrom( 'geo_tags' ) |
| 88 | ->where( [ 'gt_page_id' => $id ] ) |
| 89 | ->caller( __METHOD__ ) |
| 90 | ->execute(); |
| 91 | } |
| 92 | |
| 93 | /** |
| 94 | * LinksUpdateComplete hook handler |
| 95 | * @see https://www.mediawiki.org/wiki/Manual:Hooks/LinksUpdateComplete |
| 96 | * |
| 97 | * @param LinksUpdate $linksUpdate |
| 98 | * @param int|null $ticket |
| 99 | */ |
| 100 | public function onLinksUpdateComplete( $linksUpdate, $ticket ) { |
| 101 | $out = $linksUpdate->getParserOutput(); |
| 102 | $data = []; |
| 103 | $coordFromMetadata = $this->getCoordinatesIfFile( $linksUpdate->getTitle() ); |
| 104 | $coordsOutput = CoordinatesOutput::getFromParserOutput( $out ); |
| 105 | if ( $coordsOutput ) { |
| 106 | // Use coordinates from file metadata unless overridden on description page |
| 107 | if ( $coordFromMetadata && !$coordsOutput->hasPrimary() ) { |
| 108 | $coordsOutput->addPrimary( $coordFromMetadata ); |
| 109 | } |
| 110 | $data = $coordsOutput->getAll(); |
| 111 | } elseif ( $coordFromMetadata ) { |
| 112 | $data[] = $coordFromMetadata; |
| 113 | } |
| 114 | $this->doLinksUpdate( $data, $linksUpdate->getPageId(), $ticket ); |
| 115 | } |
| 116 | |
| 117 | private function getCoordinatesIfFile( LinkTarget $title ): ?Coord { |
| 118 | if ( !$title->inNamespace( NS_FILE ) ) { |
| 119 | return null; |
| 120 | } |
| 121 | $file = $this->repoGroup->getLocalRepo() |
| 122 | ->findFile( $title, [ 'ignoreRedirect' => true ] ); |
| 123 | if ( !$file ) { |
| 124 | return null; |
| 125 | } |
| 126 | $metadata = $file->getMetadataItems( [ 'GPSLatitude', 'GPSLongitude' ] ); |
| 127 | // T165800: Skip files with meaningless 0, 0 coordinates |
| 128 | if ( !empty( $metadata['GPSLatitude'] ) || !empty( $metadata['GPSLongitude'] ) ) { |
| 129 | $lat = $metadata['GPSLatitude'] ?? 0; |
| 130 | $lon = $metadata['GPSLongitude'] ?? 0; |
| 131 | // We assume GPS exist only on Earth |
| 132 | $globe = new Globe( Globe::EARTH ); |
| 133 | if ( $globe->coordinatesAreValid( $lat, $lon ) ) { |
| 134 | $coord = new Coord( $lat, $lon, $globe ); |
| 135 | $coord->primary = true; |
| 136 | return $coord; |
| 137 | } |
| 138 | } |
| 139 | return null; |
| 140 | } |
| 141 | |
| 142 | /** |
| 143 | * @param Coord[] $coords |
| 144 | * @param int $pageId |
| 145 | * @param int|null $ticket |
| 146 | * @throws \Wikimedia\Rdbms\DBUnexpectedError |
| 147 | */ |
| 148 | private function doLinksUpdate( array $coords, $pageId, $ticket ) { |
| 149 | $indexGranularity = $this->config->get( 'GeoDataBackend' ) === 'db' ? |
| 150 | $this->config->get( 'GeoDataIndexGranularity' ) : null; |
| 151 | |
| 152 | $add = []; |
| 153 | $delete = []; |
| 154 | $primary = isset( $coords[0] ) && $coords[0]->primary ? $coords[0] : null; |
| 155 | foreach ( GeoData::getAllCoordinates( $pageId, [], DB_PRIMARY ) as $old ) { |
| 156 | $delete[$old->id] = $old; |
| 157 | } |
| 158 | foreach ( $coords as $new ) { |
| 159 | if ( !$new->primary && $new->equalsTo( $primary ) ) { |
| 160 | // Don't save secondary coordinates pointing to the same place as the primary one |
| 161 | continue; |
| 162 | } |
| 163 | $match = false; |
| 164 | foreach ( $delete as $id => $old ) { |
| 165 | if ( $new->fullyEqualsTo( $old ) ) { |
| 166 | unset( $delete[$id] ); |
| 167 | $match = true; |
| 168 | break; |
| 169 | } |
| 170 | } |
| 171 | if ( !$match ) { |
| 172 | $add[] = $new->getRow( $pageId, $indexGranularity ); |
| 173 | } |
| 174 | } |
| 175 | |
| 176 | $dbw = $this->connectionProvider->getPrimaryDatabase(); |
| 177 | $ticket = $ticket ?: $this->lbFactory->getEmptyTransactionTicket( __METHOD__ ); |
| 178 | $batchSize = $this->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 | $this->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 | $this->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 = $this->wikiPageFactory->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( $outputPage, $parserOutput ): void { |
| 230 | $geoDataInJS = $this->config->get( 'GeoDataInJS' ); |
| 231 | if ( !$geoDataInJS ) { |
| 232 | return; |
| 233 | } |
| 234 | |
| 235 | $coord = CoordinatesOutput::getFromParserOutput( $parserOutput )?->getPrimary(); |
| 236 | if ( !$coord ) { |
| 237 | return; |
| 238 | } |
| 239 | |
| 240 | $result = []; |
| 241 | foreach ( $geoDataInJS as $param ) { |
| 242 | if ( isset( $coord->$param ) ) { |
| 243 | $result[$param] = $coord->$param; |
| 244 | } |
| 245 | } |
| 246 | if ( $result ) { |
| 247 | $outputPage->addJsConfigVars( 'wgCoordinates', $result ); |
| 248 | } |
| 249 | } |
| 250 | |
| 251 | /** |
| 252 | * Search index fields hook handler |
| 253 | * Adds our stuff to CirrusSearch/Elasticsearch schema |
| 254 | * |
| 255 | * @param array &$fields |
| 256 | * @param SearchEngine $engine |
| 257 | */ |
| 258 | public function onSearchIndexFields( array &$fields, SearchEngine $engine ) { |
| 259 | if ( !$this->config->get( 'GeoDataUseCirrusSearch' ) && |
| 260 | $this->config->get( 'GeoDataBackend' ) !== 'elastic' |
| 261 | ) { |
| 262 | return; |
| 263 | } |
| 264 | |
| 265 | if ( $engine instanceof CirrusSearch ) { |
| 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 $page ): void { |
| 316 | if ( !$this->config->get( 'GeoDataUseCirrusSearch' ) && |
| 317 | $this->config->get( 'GeoDataBackend' ) !== 'elastic' |
| 318 | ) { |
| 319 | return; |
| 320 | } |
| 321 | |
| 322 | $coordsOutput = CoordinatesOutput::getFromParserOutput( $parserOutput ); |
| 323 | $allCoords = $coordsOutput ? $coordsOutput->getAll() : []; |
| 324 | $coords = []; |
| 325 | |
| 326 | foreach ( $allCoords as $coord ) { |
| 327 | if ( !$coord->sameGlobe( Globe::EARTH ) ) { |
| 328 | continue; |
| 329 | } |
| 330 | if ( !$coord->isValid() ) { |
| 331 | wfDebugLog( 'CirrusSearchChangeFailed', |
| 332 | "Invalid coordinates [{$coord->lat}, {$coord->lon}] on page " |
| 333 | . $page->getTitle()->getPrefixedText() |
| 334 | ); |
| 335 | continue; |
| 336 | } |
| 337 | $coords[] = $this->coordToElastic( $coord ); |
| 338 | } |
| 339 | $fields['coordinates'] = $coords; |
| 340 | } |
| 341 | |
| 342 | /** |
| 343 | * Transforms coordinates into an array for insertion onto Elasticsearch |
| 344 | */ |
| 345 | private function coordToElastic( Coord $coord ): array { |
| 346 | $result = $coord->getAsArray(); |
| 347 | $result['coord'] = [ 'lat' => $coord->lat, 'lon' => $coord->lon ]; |
| 348 | unset( $result['id'] ); |
| 349 | unset( $result['lat'] ); |
| 350 | unset( $result['lon'] ); |
| 351 | |
| 352 | return $result; |
| 353 | } |
| 354 | |
| 355 | } |