Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
96.15% |
175 / 182 |
|
40.00% |
2 / 5 |
CRAP | |
0.00% |
0 / 1 |
| BackfillInterwikiRightsLog | |
96.15% |
175 / 182 |
|
40.00% |
2 / 5 |
32 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
6 / 6 |
|
100.00% |
1 / 1 |
1 | |||
| execute | |
96.32% |
131 / 136 |
|
0.00% |
0 / 1 |
20 | |||
| getTargetUserName | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| getUpToDateUserName | |
90.91% |
10 / 11 |
|
0.00% |
0 / 1 |
5.02 | |||
| getRenames | |
96.30% |
26 / 27 |
|
0.00% |
0 / 1 |
5 | |||
| 1 | <?php |
| 2 | /** |
| 3 | * @license GPL-2.0-or-later |
| 4 | * |
| 5 | * @file |
| 6 | * @ingroup Maintenance |
| 7 | */ |
| 8 | |
| 9 | declare( strict_types = 1 ); |
| 10 | |
| 11 | use MediaWiki\Logger\LoggerFactory; |
| 12 | use MediaWiki\Logging\DatabaseLogEntry; |
| 13 | use MediaWiki\Logging\LogEntry; |
| 14 | use MediaWiki\Logging\ManualLogEntry; |
| 15 | use MediaWiki\MainConfigNames; |
| 16 | use MediaWiki\Maintenance\Maintenance; |
| 17 | use MediaWiki\Title\Title; |
| 18 | use MediaWiki\User\UserIdentityValue; |
| 19 | use MediaWiki\WikiMap\WikiMap; |
| 20 | use Wikimedia\Rdbms\IExpression; |
| 21 | use Wikimedia\Rdbms\LikeValue; |
| 22 | use Wikimedia\Rdbms\SelectQueryBuilder; |
| 23 | use Wikimedia\Timestamp\ConvertibleTimestamp; |
| 24 | use Wikimedia\Timestamp\TimestampFormat as TS; |
| 25 | |
| 26 | // @codeCoverageIgnoreStart |
| 27 | require_once __DIR__ . '/Maintenance.php'; |
| 28 | // @codeCoverageIgnoreEnd |
| 29 | |
| 30 | /** |
| 31 | * Maintenance script to copy interwiki rights changes from log on the remote wiki to the current wiki |
| 32 | * |
| 33 | * @ingroup Maintenance |
| 34 | */ |
| 35 | class BackfillInterwikiRightsLog extends Maintenance { |
| 36 | private string $interwikiDelimiter; |
| 37 | |
| 38 | public function __construct() { |
| 39 | parent::__construct(); |
| 40 | $this->addDescription( 'Backfill interwiki rights log from the specified wiki' ); |
| 41 | $this->addArg( 'before', 'Only interwiki rights logs before this timestamp will be processed' ); |
| 42 | $this->addOption( 'remote-wiki', 'The wiki to read logs from', true, true ); |
| 43 | $this->addOption( 'dry-run', 'Perform a dry run, copy nothing' ); |
| 44 | $this->setBatchSize( 200 ); |
| 45 | } |
| 46 | |
| 47 | public function execute() { |
| 48 | $dryRun = $this->hasOption( 'dry-run' ); |
| 49 | $sourceWiki = $this->getOption( 'remote-wiki' ); |
| 50 | $cutoffTimestamp = ConvertibleTimestamp::convert( TS::MW, $this->getArg( 0 ) ); |
| 51 | |
| 52 | $currentWiki = WikiMap::getCurrentWikiId(); |
| 53 | if ( $sourceWiki === $currentWiki ) { |
| 54 | $this->output( "Source wiki must be different from the current wiki.\n" ); |
| 55 | return; |
| 56 | } |
| 57 | |
| 58 | $sourceDb = $this->getReplicaDB( $sourceWiki ); |
| 59 | $this->interwikiDelimiter = $this->getConfig()->get( MainConfigNames::UserrightsInterwikiDelimiter ); |
| 60 | $titlePattern = new LikeValue( $sourceDb->anyString(), $this->interwikiDelimiter . $currentWiki ); |
| 61 | |
| 62 | if ( $dryRun ) { |
| 63 | $this->output( "DRY RUN: No changes will be made\n" ); |
| 64 | } |
| 65 | |
| 66 | $minTimestamp = $sourceDb->newSelectQueryBuilder() |
| 67 | ->select( 'log_timestamp' ) |
| 68 | ->from( 'logging' ) |
| 69 | ->where( [ |
| 70 | 'log_type' => 'rights', |
| 71 | 'log_action' => 'rights', |
| 72 | ] ) |
| 73 | ->orderBy( 'log_timestamp', SelectQueryBuilder::SORT_ASC ) |
| 74 | ->caller( __METHOD__ ) |
| 75 | ->fetchField(); |
| 76 | |
| 77 | if ( $minTimestamp === false ) { |
| 78 | $this->output( "No source data found, exiting\n" ); |
| 79 | return; |
| 80 | } |
| 81 | |
| 82 | $lastLogId = 0; |
| 83 | $lastTimestamp = $minTimestamp; |
| 84 | $count = 0; |
| 85 | $skipped = 0; |
| 86 | $minInsertedId = null; |
| 87 | $maxInsertedId = null; |
| 88 | while ( true ) { |
| 89 | $rows = DatabaseLogEntry::newSelectQueryBuilder( $sourceDb ) |
| 90 | ->where( [ |
| 91 | 'log_type' => 'rights', |
| 92 | 'log_action' => 'rights', |
| 93 | $sourceDb->expr( 'log_title', IExpression::LIKE, $titlePattern ), |
| 94 | $sourceDb->expr( 'log_timestamp', '<', $sourceDb->timestamp( $cutoffTimestamp ) ), |
| 95 | ] ) |
| 96 | ->where( |
| 97 | $sourceDb->buildComparison( '>', [ |
| 98 | 'log_timestamp' => $sourceDb->timestamp( $lastTimestamp ), |
| 99 | 'log_id' => $lastLogId, |
| 100 | ] ) |
| 101 | ) |
| 102 | ->orderBy( [ 'log_timestamp', 'log_id' ], SelectQueryBuilder::SORT_ASC ) |
| 103 | ->limit( $this->getBatchSize() ) |
| 104 | ->caller( __METHOD__ ) |
| 105 | ->fetchResultSet(); |
| 106 | |
| 107 | if ( $rows->numRows() === 0 ) { |
| 108 | break; |
| 109 | } |
| 110 | $this->output( "Processing batch of {$rows->numRows()} log entries...\n" ); |
| 111 | |
| 112 | $this->beginTransactionRound( __METHOD__ ); |
| 113 | |
| 114 | $originalEntries = []; |
| 115 | $targetUserNames = []; |
| 116 | foreach ( $rows as $row ) { |
| 117 | $entry = DatabaseLogEntry::newFromRow( $row ); |
| 118 | $originalEntries[] = $entry; |
| 119 | $targetUserNames[] = $this->getTargetUserName( $entry ); |
| 120 | } |
| 121 | |
| 122 | $renames = $this->getRenames( $targetUserNames ); |
| 123 | |
| 124 | $logsToInsert = []; |
| 125 | // For deduplication query |
| 126 | $timestampsPresent = []; |
| 127 | foreach ( $originalEntries as $originalEntry ) { |
| 128 | $lastLogId = $originalEntry->getId(); |
| 129 | $lastTimestamp = $originalEntry->getTimestamp(); |
| 130 | |
| 131 | $targetName = $this->getTargetUserName( $originalEntry ); |
| 132 | $targetNewName = $this->getUpToDateUserName( $targetName, $originalEntry->getTimestamp(), $renames ); |
| 133 | if ( $targetNewName !== $targetName ) { |
| 134 | $this->output( "Renaming $targetName to $targetNewName in entry $lastLogId\n" ); |
| 135 | } |
| 136 | $targetName = $targetNewName; |
| 137 | $localTarget = Title::newFromText( $targetName, $originalEntry->getTarget()->getNamespace() ); |
| 138 | |
| 139 | $params = $originalEntry->getParameters(); |
| 140 | if ( $originalEntry->isLegacy() ) { |
| 141 | // We must ensure that the inserted log entry is in the current form, so that we don't create |
| 142 | // a yet another params schema |
| 143 | $legacyParams = $originalEntry->getParameters(); |
| 144 | if ( count( $legacyParams ) > 1 ) { |
| 145 | $oldGroups = $legacyParams[0] === '' ? [] : |
| 146 | array_map( 'trim', explode( ',', $legacyParams[0] ) ); |
| 147 | $newGroups = $legacyParams[1] === '' ? [] : |
| 148 | array_map( 'trim', explode( ',', $legacyParams[1] ) ); |
| 149 | $params = [ |
| 150 | '4::oldgroups' => $oldGroups, |
| 151 | '5::newgroups' => $newGroups, |
| 152 | ]; |
| 153 | } |
| 154 | } |
| 155 | |
| 156 | $performerName = $originalEntry->getPerformerIdentity()->getName(); |
| 157 | $performer = UserIdentityValue::newExternal( $sourceWiki, $performerName ); |
| 158 | |
| 159 | $logEntry = new ManualLogEntry( 'rights', 'rights' ); |
| 160 | $logEntry->setTimestamp( $originalEntry->getTimestamp() ); |
| 161 | $logEntry->setPerformer( $performer ); |
| 162 | $logEntry->setTarget( $localTarget ); |
| 163 | $logEntry->setComment( $originalEntry->getComment() ); |
| 164 | $logEntry->setParameters( $params ); |
| 165 | $logEntry->setDeleted( $originalEntry->getDeleted() ); |
| 166 | $logsToInsert[] = $logEntry; |
| 167 | $timestampsPresent[] = $logEntry->getTimestamp(); |
| 168 | } |
| 169 | |
| 170 | $existingRows = DatabaseLogEntry::newSelectQueryBuilder( $this->getReplicaDB() ) |
| 171 | ->where( [ |
| 172 | 'log_type' => 'rights', |
| 173 | 'log_action' => 'rights', |
| 174 | 'log_timestamp' => array_map( |
| 175 | $this->getReplicaDB()->timestamp( ... ), |
| 176 | $timestampsPresent |
| 177 | ), |
| 178 | ] ) |
| 179 | ->caller( __METHOD__ ) |
| 180 | ->fetchResultSet(); |
| 181 | |
| 182 | // keyed by timestamp => array of target users |
| 183 | $existingChanges = []; |
| 184 | foreach ( $existingRows as $row ) { |
| 185 | $entry = DatabaseLogEntry::newFromRow( $row ); |
| 186 | $existingChanges[ $entry->getTimestamp() ][] = $entry->getTarget()->getText(); |
| 187 | } |
| 188 | |
| 189 | foreach ( $logsToInsert as $logEntry ) { |
| 190 | // If the target user's rights were already changed at the same timestamp, skip so that we don't |
| 191 | // duplicate entries. This leaves room to false positives, where the user's rights are changed by |
| 192 | // different users at the same time. It's unlikely and we accept this risk here |
| 193 | if ( |
| 194 | isset( $existingChanges[ $logEntry->getTimestamp() ] ) |
| 195 | && in_array( $logEntry->getTarget()->getText(), $existingChanges[ $logEntry->getTimestamp() ] ) |
| 196 | ) { |
| 197 | $skipped++; |
| 198 | continue; |
| 199 | } |
| 200 | |
| 201 | if ( !$dryRun ) { |
| 202 | $id = $logEntry->insert(); |
| 203 | |
| 204 | if ( $minInsertedId === null ) { |
| 205 | $minInsertedId = $id; |
| 206 | } |
| 207 | $maxInsertedId = $id; |
| 208 | } |
| 209 | $count++; |
| 210 | } |
| 211 | |
| 212 | $this->commitTransactionRound( __METHOD__ ); |
| 213 | } |
| 214 | |
| 215 | $this->output( "Skipped $skipped log entries.\n" ); |
| 216 | if ( $dryRun ) { |
| 217 | $this->output( "Would insert $count log entries.\n" ); |
| 218 | } else { |
| 219 | LoggerFactory::getInstance( 'logentry' )->info( |
| 220 | 'Backfilled {count} interwiki rights log entries from {sourceWiki}.', |
| 221 | [ |
| 222 | 'count' => $count, |
| 223 | 'sourceWiki' => $sourceWiki, |
| 224 | 'minInsertedId' => $minInsertedId, |
| 225 | 'maxInsertedId' => $maxInsertedId, |
| 226 | ] |
| 227 | ); |
| 228 | |
| 229 | $minInsertedId ??= '(null)'; |
| 230 | $maxInsertedId ??= '(null)'; |
| 231 | $this->output( "Inserted $count log entries, with ids between $minInsertedId and $maxInsertedId.\n" ); |
| 232 | } |
| 233 | } |
| 234 | |
| 235 | private function getTargetUserName( LogEntry $logEntry ): string { |
| 236 | $originalTargetText = $logEntry->getTarget()->getText(); |
| 237 | return explode( $this->interwikiDelimiter, $originalTargetText )[0]; |
| 238 | } |
| 239 | |
| 240 | private function getUpToDateUserName( string $originalName, string $timestamp, array $renames ): string { |
| 241 | while ( array_key_exists( $originalName, $renames ) ) { |
| 242 | $renameFound = false; |
| 243 | foreach ( $renames[$originalName] as $renameTimestamp => $newName ) { |
| 244 | if ( $renameTimestamp > $timestamp ) { |
| 245 | $originalName = $newName; |
| 246 | $timestamp = $renameTimestamp; |
| 247 | $renameFound = true; |
| 248 | break; |
| 249 | } |
| 250 | } |
| 251 | if ( !$renameFound ) { |
| 252 | break; |
| 253 | } |
| 254 | } |
| 255 | return $originalName; |
| 256 | } |
| 257 | |
| 258 | /** |
| 259 | * Search for renames affecting the specified users |
| 260 | * @param list<string> $originalUserNames The users to resolve the renames for |
| 261 | * @return array<string,list<array{0:string,1:string}>> For each original name, a list of new names, keyed by |
| 262 | * and ordered by the rename timestamp |
| 263 | */ |
| 264 | private function getRenames( array $originalUserNames ): array { |
| 265 | $renames = []; |
| 266 | $dbr = $this->getReplicaDB(); |
| 267 | |
| 268 | // Convert usernames to the title form (with underscores). Use space form only in output |
| 269 | $originalUserNames = array_map( static fn ( $name ) => strtr( $name, ' ', '_' ), $originalUserNames ); |
| 270 | |
| 271 | while ( $originalUserNames ) { |
| 272 | $originalUserNames = array_unique( $originalUserNames ); |
| 273 | $batch = array_splice( $originalUserNames, 0, 100 ); |
| 274 | $renameLogs = DatabaseLogEntry::newSelectQueryBuilder( $dbr ) |
| 275 | ->where( [ |
| 276 | 'log_namespace' => NS_USER, |
| 277 | 'log_title' => $batch, |
| 278 | 'log_type' => 'renameuser', |
| 279 | ] ) |
| 280 | ->orderBy( 'log_timestamp' ) |
| 281 | ->caller( __METHOD__ ) |
| 282 | ->fetchResultSet(); |
| 283 | |
| 284 | foreach ( $renameLogs as $renameLog ) { |
| 285 | $log = DatabaseLogEntry::newFromRow( $renameLog ); |
| 286 | |
| 287 | $oldName = $log->getTarget()->getDBkey(); |
| 288 | $timestamp = $log->getTimestamp(); |
| 289 | $params = $log->getParameters(); |
| 290 | $newName = strtr( $params['5::newuser'] ?? $params[0] ?? '', ' ', '_' ); |
| 291 | |
| 292 | if ( $newName === '' ) { |
| 293 | // Invalid log entry, ignore |
| 294 | continue; |
| 295 | } |
| 296 | |
| 297 | $renames[$oldName][$timestamp] = strtr( $newName, '_', ' ' ); |
| 298 | if ( !array_key_exists( $newName, $renames ) ) { |
| 299 | // Follow up on the next renames affecting the same user |
| 300 | $originalUserNames[] = $newName; |
| 301 | } |
| 302 | } |
| 303 | } |
| 304 | |
| 305 | return $renames; |
| 306 | } |
| 307 | } |
| 308 | |
| 309 | // @codeCoverageIgnoreStart |
| 310 | $maintClass = BackfillInterwikiRightsLog::class; |
| 311 | require_once RUN_MAINTENANCE_IF_MAIN; |
| 312 | // @codeCoverageIgnoreEnd |