Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
96.24% |
179 / 186 |
|
90.00% |
9 / 10 |
CRAP | |
0.00% |
0 / 1 |
| ImportExistingFilesToScanTable | |
99.44% |
179 / 180 |
|
90.00% |
9 / 10 |
32 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
31 / 31 |
|
100.00% |
1 / 1 |
1 | |||
| getUpdateKey | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| doDBUpdates | |
97.92% |
47 / 48 |
|
0.00% |
0 / 1 |
7 | |||
| generateDBUpdatesReturnValue | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
3 | |||
| getTablesToProcess | |
100.00% |
9 / 9 |
|
100.00% |
1 / 1 |
4 | |||
| getEstimatedNumberOfBatchesForTable | |
100.00% |
19 / 19 |
|
100.00% |
1 / 1 |
4 | |||
| getTemporaryBatchSize | |
100.00% |
14 / 14 |
|
100.00% |
1 / 1 |
2 | |||
| getFileSelectQueryBuilder | |
100.00% |
17 / 17 |
|
100.00% |
1 / 1 |
3 | |||
| getRowsForBatch | |
100.00% |
24 / 24 |
|
100.00% |
1 / 1 |
3 | |||
| performBatch | |
100.00% |
12 / 12 |
|
100.00% |
1 / 1 |
4 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace MediaWiki\Extension\MediaModeration\Maintenance; |
| 4 | |
| 5 | use LoggedUpdateMaintenance; |
| 6 | use MediaWiki\Extension\MediaModeration\Services\MediaModerationDatabaseLookup; |
| 7 | use MediaWiki\Extension\MediaModeration\Services\MediaModerationFileFactory; |
| 8 | use MediaWiki\Extension\MediaModeration\Services\MediaModerationFileLookup; |
| 9 | use MediaWiki\Extension\MediaModeration\Services\MediaModerationFileProcessor; |
| 10 | use MediaWiki\FileRepo\File\FileSelectQueryBuilder; |
| 11 | use Wikimedia\Rdbms\IReadableDatabase; |
| 12 | use Wikimedia\Rdbms\SelectQueryBuilder; |
| 13 | |
| 14 | $IP = getenv( 'MW_INSTALL_PATH' ); |
| 15 | if ( $IP === false ) { |
| 16 | $IP = __DIR__ . '/../../..'; |
| 17 | } |
| 18 | require_once "$IP/maintenance/Maintenance.php"; |
| 19 | |
| 20 | class ImportExistingFilesToScanTable extends LoggedUpdateMaintenance { |
| 21 | |
| 22 | /** @var string[] The DB tables that have images to be imported to mediamoderation_scan. */ |
| 23 | public const TABLES_TO_IMPORT_FROM = [ |
| 24 | 'image', |
| 25 | 'filearchive', |
| 26 | 'oldimage', |
| 27 | ]; |
| 28 | |
| 29 | private IReadableDatabase $dbr; |
| 30 | private MediaModerationFileProcessor $mediaModerationFileProcessor; |
| 31 | private MediaModerationDatabaseLookup $mediaModerationDatabaseLookup; |
| 32 | private MediaModerationFileFactory $mediaModerationFileFactory; |
| 33 | private MediaModerationFileLookup $mediaModerationFileLookup; |
| 34 | |
| 35 | public function __construct() { |
| 36 | parent::__construct(); |
| 37 | $this->requireExtension( 'MediaModeration' ); |
| 38 | $this->addDescription( 'Populates the mediamoderation_scan table with existing images from the wiki.' ); |
| 39 | $this->addOption( |
| 40 | 'sleep', |
| 41 | 'Sleep time (in seconds) between every batch. Default: 1', |
| 42 | false, |
| 43 | true |
| 44 | ); |
| 45 | $this->addOption( |
| 46 | 'start-timestamp', |
| 47 | 'The timestamp which to start importing files from. Default is for no timestamp start point ' . |
| 48 | '(which means importing all images)', |
| 49 | false, |
| 50 | true |
| 51 | ); |
| 52 | $this->addOption( |
| 53 | 'table', |
| 54 | 'Allows specifying which table(s) files should be imported from. Default is all supported tables.', |
| 55 | false, |
| 56 | true, |
| 57 | false, |
| 58 | true |
| 59 | ); |
| 60 | $this->addOption( |
| 61 | 'mark-complete', |
| 62 | 'Allows controlling whether this script should be considered completely run for the purposes ' . |
| 63 | 'of the updatelog. If provided the script will be marked as complete. ' . |
| 64 | "Default is to consider the script completely run if the 'table' and 'start-timestamp' options were left " . |
| 65 | 'as the default and the script does not error out.', |
| 66 | ); |
| 67 | } |
| 68 | |
| 69 | /** @inheritDoc */ |
| 70 | protected function getUpdateKey() { |
| 71 | return __CLASS__; |
| 72 | } |
| 73 | |
| 74 | /** @inheritDoc */ |
| 75 | protected function doDBUpdates() { |
| 76 | $services = $this->getServiceContainer(); |
| 77 | $loadBalancerFactory = $services->getDBLoadBalancerFactory(); |
| 78 | $this->dbr = $loadBalancerFactory->getReplicaDatabase(); |
| 79 | $this->mediaModerationFileFactory = $services->get( 'MediaModerationFileFactory' ); |
| 80 | $this->mediaModerationFileProcessor = $services->get( 'MediaModerationFileProcessor' ); |
| 81 | $this->mediaModerationDatabaseLookup = $services->get( 'MediaModerationDatabaseLookup' ); |
| 82 | $this->mediaModerationFileLookup = $services->get( 'MediaModerationFileLookup' ); |
| 83 | |
| 84 | // Get the list of tables to import images from. |
| 85 | $tablesToProcess = $this->getTablesToProcess(); |
| 86 | if ( $tablesToProcess === false ) { |
| 87 | // If the value is false, then return false |
| 88 | // as this indicates the script did not run. |
| 89 | return false; |
| 90 | } |
| 91 | |
| 92 | foreach ( $tablesToProcess as $table ) { |
| 93 | $batchSize = $this->getBatchSize() ?? 200; |
| 94 | $this->output( "Now importing rows from the table '$table' in batches of $batchSize.\n" ); |
| 95 | $previousBatchFinalTimestamp = $this->getOption( 'start-timestamp', '' ); |
| 96 | if ( $previousBatchFinalTimestamp ) { |
| 97 | $this->output( |
| 98 | "Starting from timestamp $previousBatchFinalTimestamp and importing files with a " . |
| 99 | "greater timestamp.\n" |
| 100 | ); |
| 101 | } |
| 102 | $expectedNumberOfBatches = $this->getEstimatedNumberOfBatchesForTable( |
| 103 | $table, $previousBatchFinalTimestamp |
| 104 | ); |
| 105 | $batchNo = 1; |
| 106 | do { |
| 107 | $outputString = "Batch $batchNo of ~$expectedNumberOfBatches"; |
| 108 | if ( $previousBatchFinalTimestamp ) { |
| 109 | $outputString .= " with rows starting at timestamp $previousBatchFinalTimestamp"; |
| 110 | } |
| 111 | $this->output( $outputString . ".\n" ); |
| 112 | [ |
| 113 | $filesLeft, |
| 114 | $previousBatchFinalTimestamp |
| 115 | ] = $this->performBatch( $table, $previousBatchFinalTimestamp ); |
| 116 | sleep( intval( $this->getOption( 'sleep', 1 ) ) ); |
| 117 | $loadBalancerFactory->waitForReplication(); |
| 118 | $batchNo += 1; |
| 119 | } while ( $filesLeft ); |
| 120 | } |
| 121 | $returnValue = $this->generateDBUpdatesReturnValue(); |
| 122 | if ( $this->hasOption( 'force' ) ) { |
| 123 | // If the script was run with the force option, then don't |
| 124 | // print out about how the script has been or has not been |
| 125 | // marked as completed. |
| 126 | return $returnValue; |
| 127 | } |
| 128 | if ( $returnValue ) { |
| 129 | $this->output( "Script marked as completed (added to updatelog).\n" ); |
| 130 | } else { |
| 131 | $this->output( |
| 132 | 'Script not marked as completed (not added to updatelog). The script was marked as not complete ' . |
| 133 | "because not all the images on the wiki were processed in this run of the script.\n" . |
| 134 | 'To mark the script as complete and not have it run again through update.php, make sure to run the ' . |
| 135 | "script again with the 'mark-complete' option specified. You should only do this once you are sure " . |
| 136 | "that all the images on the wiki have been imported.\n" |
| 137 | ); |
| 138 | } |
| 139 | return $returnValue; |
| 140 | } |
| 141 | |
| 142 | /** |
| 143 | * Return the value that should be used as the return value |
| 144 | * of ::doDBUpdates. This value depends on the options |
| 145 | * passed to the script. |
| 146 | * |
| 147 | * If true is returned, it should only be done if the script |
| 148 | * either has imported all images or the caller of the maintenance |
| 149 | * script has specifically intended for the script to marked as |
| 150 | * complete. |
| 151 | * |
| 152 | * @return bool |
| 153 | */ |
| 154 | protected function generateDBUpdatesReturnValue(): bool { |
| 155 | // Return true if mark-complete is specified, or if both: |
| 156 | // * start-timestamp is not specified or an empty string, and |
| 157 | // * table is not specified or includes all tables listed in self::TABLES_TO_IMPORT_FROM. |
| 158 | return $this->hasOption( 'mark-complete' ) || |
| 159 | ( |
| 160 | $this->getOption( 'start-timestamp', '' ) === '' && |
| 161 | $this->getOption( 'table', self::TABLES_TO_IMPORT_FROM ) == self::TABLES_TO_IMPORT_FROM |
| 162 | ); |
| 163 | } |
| 164 | |
| 165 | /** |
| 166 | * Processes the user supplied list of tables to process, |
| 167 | * with the default being all supported tables. |
| 168 | * |
| 169 | * Prints an error if the supplied arguments are invalid. |
| 170 | * |
| 171 | * @return false|array The list of tables to process, or false if the list was not valid. |
| 172 | */ |
| 173 | protected function getTablesToProcess() { |
| 174 | $tablesToProcess = $this->getOption( 'table', self::TABLES_TO_IMPORT_FROM ); |
| 175 | if ( !count( $tablesToProcess ) ) { |
| 176 | $this->error( "The array of tables to have images imported from cannot be empty.\n" ); |
| 177 | return false; |
| 178 | } |
| 179 | foreach ( $tablesToProcess as $table ) { |
| 180 | if ( !in_array( $table, self::TABLES_TO_IMPORT_FROM ) ) { |
| 181 | $this->error( "The table option value '$table' is not a valid table to import images from.\n" ); |
| 182 | return false; |
| 183 | } |
| 184 | } |
| 185 | return $tablesToProcess; |
| 186 | } |
| 187 | |
| 188 | /** |
| 189 | * Gets the expected number of batches needed to process a table. |
| 190 | * This is used just for visual display and the actual number of batches |
| 191 | * may be higher or lower. |
| 192 | * |
| 193 | * @param string $table |
| 194 | * @param string $startTimestamp The timestamp that the processing will start from. |
| 195 | * @return int |
| 196 | */ |
| 197 | protected function getEstimatedNumberOfBatchesForTable( string $table, string $startTimestamp ): int { |
| 198 | // Get the row count for the $table. |
| 199 | $queryBuilder = $this->dbr->newSelectQueryBuilder() |
| 200 | ->select( 'COUNT(*)' ) |
| 201 | ->from( $table ); |
| 202 | if ( $startTimestamp ) { |
| 203 | $queryBuilder->where( $this->dbr->expr( |
| 204 | $this->mediaModerationFileLookup->getTimestampFieldForTable( $table ), |
| 205 | '>=', |
| 206 | $this->dbr->timestamp( $startTimestamp ) |
| 207 | ) ); |
| 208 | } |
| 209 | $rowCountInTable = $queryBuilder |
| 210 | ->caller( __METHOD__ ) |
| 211 | ->fetchField(); |
| 212 | // If the row count is zero, then one batch will be performed. |
| 213 | if ( !$rowCountInTable ) { |
| 214 | return 1; |
| 215 | } |
| 216 | // The expected batch count is the number of rows in the table |
| 217 | // divided by the batch size. This may be higher than the actual |
| 218 | // batch count, as it may be temporarily increased to prevent |
| 219 | // infinite loops. |
| 220 | $batchSize = $this->getBatchSize() ?? 200; |
| 221 | $expectedBatchesCount = ceil( $rowCountInTable / $batchSize ); |
| 222 | if ( $rowCountInTable % $batchSize === 0 ) { |
| 223 | // If the batch size divides the row count without a remainder, then |
| 224 | // the expected batch count needs to be increased by one as one |
| 225 | // more batch will be performed at the end with no rows found. |
| 226 | $expectedBatchesCount += 1; |
| 227 | } |
| 228 | return $expectedBatchesCount; |
| 229 | } |
| 230 | |
| 231 | /** |
| 232 | * Gets the temporary batch size for use by ::processBatch if it was provided |
| 233 | * with $shouldRaiseBatchSize as the boolean 'true'. This is one more than |
| 234 | * the number of files with the $previousBatchFinalTimestamp. |
| 235 | * |
| 236 | * @param FileSelectQueryBuilder $fileSelectQueryBuilder The cloned FileSelectQueryBuilder that is being |
| 237 | * built in ::getSelectFileQueryBuilder. This needs to be cloned to avoid issues with this |
| 238 | * method modifying the query builder. |
| 239 | * @param string $timestampField |
| 240 | * @param string $previousBatchFinalTimestamp |
| 241 | * @return int |
| 242 | */ |
| 243 | protected function getTemporaryBatchSize( |
| 244 | FileSelectQueryBuilder $fileSelectQueryBuilder, string $timestampField, string $previousBatchFinalTimestamp |
| 245 | ): int { |
| 246 | $filesWithTheCutoffTimestamp = (int)$fileSelectQueryBuilder |
| 247 | ->clearFields() |
| 248 | ->field( 'COUNT(*)' ) |
| 249 | ->where( $this->dbr->expr( $timestampField, '=', $this->dbr->timestamp( $previousBatchFinalTimestamp ) ) ) |
| 250 | ->caller( __METHOD__ ) |
| 251 | ->fetchField(); |
| 252 | // Sanity check that the new batch size would actually be larger (otherwise |
| 253 | // leave the batch size as is as it will be fine). |
| 254 | if ( $filesWithTheCutoffTimestamp >= ( $this->getBatchSize() ?? 200 ) ) { |
| 255 | $batchSize = $filesWithTheCutoffTimestamp + 1; |
| 256 | $this->output( |
| 257 | "Temporarily raised the batch size to $batchSize due to files with the same upload timestamp. " . |
| 258 | "This is done to prevent an infinite loop. Consider raising the batch size to avoid this.\n" |
| 259 | ); |
| 260 | return $batchSize; |
| 261 | } |
| 262 | return $this->getBatchSize() ?? 200; |
| 263 | } |
| 264 | |
| 265 | /** |
| 266 | * Gets the appropriate FileSelectQueryBuilder for the $table and |
| 267 | * applies the WHERE conditions, ORDER BY and LIMIT. |
| 268 | * |
| 269 | * @param string $table The table name currently being processed. |
| 270 | * @param string $previousBatchFinalTimestamp The timestamp which the last batch stopped at. This |
| 271 | * is used to filter for files with this timestamp or a newer timestamp. |
| 272 | * @param bool $shouldRaiseBatchSize Used to indicate that the previous batch ended and started on |
| 273 | * the same timestamp, so this batch should reattempt that timestamp |
| 274 | * but with a temporarily raised batch size to account for this. |
| 275 | * @return FileSelectQueryBuilder |
| 276 | */ |
| 277 | protected function getFileSelectQueryBuilder( |
| 278 | string $table, string $previousBatchFinalTimestamp, bool $shouldRaiseBatchSize |
| 279 | ): FileSelectQueryBuilder { |
| 280 | // Get the appropriate FileSelectQueryBuilder using MediaModerationDatabaseLookup::getFileSelectQueryBuilder |
| 281 | $fileSelectQueryBuilder = $this->mediaModerationFileLookup->getFileSelectQueryBuilder( $table ); |
| 282 | $timestampField = $this->mediaModerationFileLookup->getTimestampFieldForTable( $table ); |
| 283 | $batchSize = $this->getBatchSize() ?? 200; |
| 284 | if ( $shouldRaiseBatchSize ) { |
| 285 | // If the previous batch started and ended on the same timestamp, |
| 286 | // then temporarily raise the batch count to 1 more than the number |
| 287 | // of files with this timestamp to avoid an infinite loop. |
| 288 | $batchSize = $this->getTemporaryBatchSize( |
| 289 | clone $fileSelectQueryBuilder, |
| 290 | $timestampField, |
| 291 | $previousBatchFinalTimestamp |
| 292 | ); |
| 293 | } |
| 294 | // If the timestamp is not empty, filter for entries with a greater timestamp |
| 295 | // than the cutoff timestamp. |
| 296 | if ( $previousBatchFinalTimestamp ) { |
| 297 | $fileSelectQueryBuilder |
| 298 | ->where( $this->dbr->expr( |
| 299 | $timestampField, '>=', $this->dbr->timestamp( $previousBatchFinalTimestamp ) ) ); |
| 300 | } |
| 301 | // Order by the timestamp (oldest to newest) and set the limit as the batch size. |
| 302 | $fileSelectQueryBuilder |
| 303 | ->orderBy( $timestampField, SelectQueryBuilder::SORT_ASC ) |
| 304 | ->limit( $batchSize ); |
| 305 | return $fileSelectQueryBuilder; |
| 306 | } |
| 307 | |
| 308 | /** |
| 309 | * Gets the rows for a batch along with the timestamp for the last file in the batch. |
| 310 | * |
| 311 | * @param string $table |
| 312 | * @param string $previousBatchFinalTimestamp |
| 313 | * @return array The rows for the batch, the timestamp for the last file in the results list, and the |
| 314 | * LIMIT used for the batch. |
| 315 | */ |
| 316 | protected function getRowsForBatch( string $table, string $previousBatchFinalTimestamp ): array { |
| 317 | // Get the FileSelectQueryBuilder with everything but the caller specified. |
| 318 | $fileSelectQueryBuilder = $this->getFileSelectQueryBuilder( |
| 319 | $table, $previousBatchFinalTimestamp, false |
| 320 | ); |
| 321 | // Specify the caller and then get the rows from the DB. |
| 322 | $rows = $fileSelectQueryBuilder |
| 323 | ->caller( __METHOD__ ) |
| 324 | ->fetchResultSet(); |
| 325 | $lastFileTimestamp = $previousBatchFinalTimestamp; |
| 326 | // Check whether the last file in this batch has the same timestamp as in |
| 327 | // $previousBatchFinalTimestamp. If so, then increase the batch size to |
| 328 | // prevent an infinite loop which would be caused by processing the same |
| 329 | // files over and over again with that timestamp. |
| 330 | if ( $rows->numRows() ) { |
| 331 | $rows->seek( $rows->numRows() - 1 ); |
| 332 | $lastFileObject = $this->mediaModerationFileFactory->getFileObjectForRow( $rows->fetchObject(), $table ); |
| 333 | if ( $previousBatchFinalTimestamp === $lastFileObject->getTimestamp() ) { |
| 334 | // Temporarily raise the batch size for the next batch as the |
| 335 | // last timestamp in this batch is the same as the last timestamp |
| 336 | // for the last batch. |
| 337 | $fileSelectQueryBuilder = $this->getFileSelectQueryBuilder( |
| 338 | $table, $previousBatchFinalTimestamp, true |
| 339 | ); |
| 340 | $rows = $fileSelectQueryBuilder |
| 341 | ->caller( __METHOD__ ) |
| 342 | ->fetchResultSet(); |
| 343 | $rows->seek( $rows->numRows() - 1 ); |
| 344 | $lastFileObject = $this->mediaModerationFileFactory->getFileObjectForRow( |
| 345 | $rows->fetchObject(), $table |
| 346 | ); |
| 347 | } |
| 348 | // Store the timestamp for the last file, and return it to the caller later in this method. |
| 349 | $lastFileTimestamp = $lastFileObject->getTimestamp(); |
| 350 | // Reset the position of the pointer for the caller to be able to use a foreach loop on $rows. |
| 351 | $rows->rewind(); |
| 352 | } |
| 353 | return [ $rows, $lastFileTimestamp, $fileSelectQueryBuilder->getQueryInfo()['options']['LIMIT'] ]; |
| 354 | } |
| 355 | |
| 356 | /** |
| 357 | * Perform a batch of imports to the mediamoderation_scan table from the $table |
| 358 | * starting at the $lastFileTimestamp going towards newer files. |
| 359 | * |
| 360 | * @param string $table |
| 361 | * @param string $previousBatchFinalTimestamp |
| 362 | * @return array First value is whether another batch should be run, second value is the new value of |
| 363 | * $previousBatchFinalTimestamp, and third value is the new value of $shouldRaiseBatchSize |
| 364 | */ |
| 365 | protected function performBatch( string $table, string $previousBatchFinalTimestamp ): array { |
| 366 | [ $rows, $lastFileTimestamp, $batchSizeUsedForBatch ] = $this->getRowsForBatch( |
| 367 | $table, $previousBatchFinalTimestamp |
| 368 | ); |
| 369 | foreach ( $rows as $row ) { |
| 370 | // Get the File or ArchivedFile object for this $row. |
| 371 | $fileObject = $this->mediaModerationFileFactory->getFileObjectForRow( $row, $table ); |
| 372 | // Exclude any file that has a SHA-1 value that is false or empty. |
| 373 | // This can happen in some filearchive rows where the image no |
| 374 | // longer exists. |
| 375 | // Also check if the SHA-1 exists in the scan table using a replica DB |
| 376 | // before attempting to insert the file to reduce the number of |
| 377 | // unnecessary reads on the primary DB. |
| 378 | if ( |
| 379 | $fileObject->getSha1() && |
| 380 | !$this->mediaModerationDatabaseLookup->fileExistsInScanTable( $fileObject ) |
| 381 | ) { |
| 382 | $this->mediaModerationFileProcessor->insertFile( $fileObject ); |
| 383 | } |
| 384 | } |
| 385 | // Return false as the first item of the array if the number of rows processed |
| 386 | // was less than the batch size. This will happen when there are no more images |
| 387 | // to process. |
| 388 | return [ |
| 389 | $rows->count() >= $batchSizeUsedForBatch, |
| 390 | $lastFileTimestamp |
| 391 | ]; |
| 392 | } |
| 393 | } |
| 394 | |
| 395 | $maintClass = ImportExistingFilesToScanTable::class; |
| 396 | require_once RUN_MAINTENANCE_IF_MAIN; |