Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
100.00% |
108 / 108 |
|
100.00% |
7 / 7 |
CRAP | |
100.00% |
1 / 1 |
| FindMissingActors | |
100.00% |
108 / 108 |
|
100.00% |
7 / 7 |
23 | |
100.00% |
1 / 1 |
| __construct | |
100.00% |
17 / 17 |
|
100.00% |
1 / 1 |
1 | |||
| getTables | |
100.00% |
11 / 11 |
|
100.00% |
1 / 1 |
1 | |||
| getTableInfo | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| getNewActorId | |
100.00% |
17 / 17 |
|
100.00% |
1 / 1 |
6 | |||
| execute | |
100.00% |
23 / 23 |
|
100.00% |
1 / 1 |
7 | |||
| findBadActors | |
100.00% |
25 / 25 |
|
100.00% |
1 / 1 |
6 | |||
| overwriteActorIDs | |
100.00% |
13 / 13 |
|
100.00% |
1 / 1 |
1 | |||
| 1 | <?php |
| 2 | /** |
| 3 | * @license GPL-2.0-or-later |
| 4 | * @file |
| 5 | * @ingroup Maintenance |
| 6 | */ |
| 7 | |
| 8 | use MediaWiki\Exception\CannotCreateActorException; |
| 9 | use MediaWiki\Maintenance\Maintenance; |
| 10 | use MediaWiki\User\ActorNormalization; |
| 11 | use MediaWiki\User\UserFactory; |
| 12 | use MediaWiki\User\UserNameUtils; |
| 13 | |
| 14 | // @codeCoverageIgnoreStart |
| 15 | require_once __DIR__ . '/Maintenance.php'; |
| 16 | // @codeCoverageIgnoreEnd |
| 17 | |
| 18 | /** |
| 19 | * Maintenance script for finding and replacing invalid actor IDs, see T261325 and T307738. |
| 20 | * |
| 21 | * @ingroup Maintenance |
| 22 | */ |
| 23 | class FindMissingActors extends Maintenance { |
| 24 | |
| 25 | private UserFactory $userFactory; |
| 26 | private UserNameUtils $userNameUtils; |
| 27 | private ActorNormalization $actorNormalization; |
| 28 | |
| 29 | public function __construct() { |
| 30 | parent::__construct(); |
| 31 | |
| 32 | $this->addDescription( 'Find and fix invalid actor IDs.' ); |
| 33 | $this->addOption( 'field', 'The name of a database field to process', |
| 34 | true, true ); |
| 35 | $this->addOption( 'type', 'Which type of invalid actors to find or fix, ' |
| 36 | . 'missing or broken (with empty actor_name which can\'t be associated ' |
| 37 | . 'with an existing user).', |
| 38 | false, true ); |
| 39 | $this->addOption( 'skip', 'A comma-separated list of actor IDs to skip.', |
| 40 | false, true ); |
| 41 | $this->addOption( 'overwrite-with', 'Replace invalid actors with this user. ' |
| 42 | . 'Typically, this would be "Unknown user", but it could be any reserved ' |
| 43 | . 'system user (per $wgReservedUsernames) or locally registered user. ' |
| 44 | . 'If not given, invalid actors will only be listed, not fixed. ' |
| 45 | . 'You will be prompted for confirmation before data is written. ', |
| 46 | false, true ); |
| 47 | |
| 48 | $this->setBatchSize( 1000 ); |
| 49 | } |
| 50 | |
| 51 | /** |
| 52 | * @return array |
| 53 | */ |
| 54 | private function getTables() { |
| 55 | return [ |
| 56 | 'ar_actor' => [ 'archive', 'ar_actor', 'ar_id' ], |
| 57 | 'img_actor' => [ 'image', 'img_actor', 'img_name' ], |
| 58 | 'oi_actor' => [ 'oldimage', 'oi_actor', 'oi_archive_name' ], // no index on oi_archive_name! |
| 59 | 'fa_actor' => [ 'filearchive', 'fa_actor', 'fa_id' ], |
| 60 | 'fr_actor' => [ 'filerevision', 'fr_actor', 'fr_id' ], |
| 61 | 'rc_actor' => [ 'recentchanges', 'rc_actor', 'rc_id' ], |
| 62 | 'log_actor' => [ 'logging', 'log_actor', 'log_id' ], |
| 63 | 'rev_actor' => [ 'revision', 'rev_actor', 'rev_id' ], |
| 64 | 'bl_by_actor' => [ 'block', 'bl_by_actor', 'bl_id' ], // no index on bl_by_actor! |
| 65 | ]; |
| 66 | } |
| 67 | |
| 68 | /** |
| 69 | * @param string $field |
| 70 | * @return array|null |
| 71 | */ |
| 72 | private function getTableInfo( $field ) { |
| 73 | $tables = $this->getTables(); |
| 74 | return $tables[$field] ?? null; |
| 75 | } |
| 76 | |
| 77 | /** |
| 78 | * Returns the actor ID of the user specified with the --overwrite-with option, |
| 79 | * or null if --overwrite-with is not set. |
| 80 | * |
| 81 | * Existing users and reserved system users are supported. |
| 82 | * If the user does not have an actor ID yet, one will be assigned. |
| 83 | * |
| 84 | * @return int|null |
| 85 | */ |
| 86 | private function getNewActorId() { |
| 87 | $name = $this->getOption( 'overwrite-with' ); |
| 88 | |
| 89 | if ( $name === null ) { |
| 90 | return null; |
| 91 | } |
| 92 | |
| 93 | $user = $this->userFactory->newFromName( $name ); |
| 94 | |
| 95 | if ( !$user ) { |
| 96 | $this->fatalError( "Not a valid user name: '$name'" ); |
| 97 | } |
| 98 | |
| 99 | if ( $user->isRegistered() ) { |
| 100 | $this->output( "Using existing user: '$user'\n" ); |
| 101 | } elseif ( !$this->userNameUtils->isUsable( $user->getName() ) ) { |
| 102 | $this->output( "Using system user: '{$user->getName()}'\n" ); |
| 103 | } else { |
| 104 | $this->fatalError( "Unknown user: '{$user->getName()}'" ); |
| 105 | } |
| 106 | |
| 107 | $dbw = $this->getPrimaryDB(); |
| 108 | |
| 109 | try { |
| 110 | $actorId = $this->actorNormalization->acquireActorId( $user, $dbw ); |
| 111 | } catch ( CannotCreateActorException ) { |
| 112 | $this->fatalError( "Failed to acquire an actor ID for user '$user'" ); |
| 113 | } |
| 114 | |
| 115 | $this->output( "Replacement actor ID is $actorId.\n" ); |
| 116 | return $actorId; |
| 117 | } |
| 118 | |
| 119 | public function execute() { |
| 120 | $services = $this->getServiceContainer(); |
| 121 | $this->userFactory = $services->getUserFactory(); |
| 122 | $this->userNameUtils = $services->getUserNameUtils(); |
| 123 | $this->actorNormalization = $services->getActorNormalization(); |
| 124 | |
| 125 | $field = $this->getOption( 'field' ); |
| 126 | if ( !$this->getTableInfo( $field ) ) { |
| 127 | $this->fatalError( "Unknown field: $field.\n" ); |
| 128 | } |
| 129 | |
| 130 | $type = $this->getOption( 'type', 'missing' ); |
| 131 | if ( $type !== 'missing' && $type !== 'broken' ) { |
| 132 | $this->fatalError( "Unknown type: $type.\n" ); |
| 133 | } |
| 134 | |
| 135 | $skip = $this->parseIntList( $this->getOption( 'skip', '' ) ); |
| 136 | $overwrite = $this->getNewActorId(); |
| 137 | |
| 138 | $bad = $this->findBadActors( $field, $type, $skip ); |
| 139 | |
| 140 | if ( $bad && $overwrite ) { |
| 141 | $this->output( "\n" ); |
| 142 | $this->output( "Do you want to OVERWRITE the listed actor IDs?\n" ); |
| 143 | $this->output( "Information about the invalid IDs will be lost!\n" ); |
| 144 | $this->output( "\n" ); |
| 145 | $confirm = static::readconsole( 'Type "yes" to continue: ' ); |
| 146 | |
| 147 | if ( $confirm === 'yes' ) { |
| 148 | $this->overwriteActorIDs( $field, array_keys( $bad ), $overwrite ); |
| 149 | } else { |
| 150 | $this->fatalError( 'Aborted.' ); |
| 151 | } |
| 152 | } |
| 153 | |
| 154 | $this->output( "Done.\n" ); |
| 155 | } |
| 156 | |
| 157 | /** |
| 158 | * Find rows that have bad actor IDs. |
| 159 | * |
| 160 | * @param string $field the database field in which to detect bad actor IDs. |
| 161 | * @param string $type type of bad actors, missing or broken. |
| 162 | * @param int[] $skip bad actor IDs not to replace. |
| 163 | * |
| 164 | * @return array a list of row IDs, identifying rows in which the actor ID needs to be replaced. |
| 165 | */ |
| 166 | private function findBadActors( $field, $type, $skip ) { |
| 167 | [ $table, $actorField, $idField ] = $this->getTableInfo( $field ); |
| 168 | $this->output( "Finding invalid actor IDs in $table.$actorField...\n" ); |
| 169 | |
| 170 | $dbr = $this->getServiceContainer()->getDBLoadBalancer()->getConnection( DB_REPLICA, 'vslow' ); |
| 171 | |
| 172 | /* |
| 173 | We are building an SQL query like this one here, performing a left join |
| 174 | to detect rows in $table that lack a matching row in the actor table. |
| 175 | |
| 176 | In this example, $field is 'log_actor', so $table is 'logging', |
| 177 | $actorField is 'log_actor', and $idField is 'log_id'. |
| 178 | Further, $skip is [ 1, 2, 3, 4 ] and the batch size is 1000. |
| 179 | |
| 180 | SELECT log_id |
| 181 | FROM logging |
| 182 | LEFT JOIN actor ON log_actor = actor_id |
| 183 | WHERE actor_id IS NULL |
| 184 | AND log_actor NOT IN (1, 2, 3, 4) |
| 185 | LIMIT 1000; |
| 186 | */ |
| 187 | |
| 188 | $queryBuilder = $dbr->newSelectQueryBuilder() |
| 189 | ->select( [ $actorField, $idField ] ) |
| 190 | ->from( $table ) |
| 191 | ->leftJoin( 'actor', null, [ "$actorField = actor_id" ] ) |
| 192 | ->where( $type == 'missing' ? [ 'actor_id' => null ] : [ 'actor_name' => '' ] ) |
| 193 | ->limit( $this->getBatchSize() ); |
| 194 | |
| 195 | if ( $skip ) { |
| 196 | $queryBuilder->andWhere( $dbr->expr( $actorField, '!=', $skip ) ); |
| 197 | } |
| 198 | |
| 199 | $res = $queryBuilder->caller( __METHOD__ )->fetchResultSet(); |
| 200 | $count = $res->numRows(); |
| 201 | |
| 202 | $bad = []; |
| 203 | |
| 204 | if ( $count ) { |
| 205 | $this->output( "\t\tID\tACTOR\n" ); |
| 206 | } |
| 207 | |
| 208 | foreach ( $res as $row ) { |
| 209 | $id = $row->$idField; |
| 210 | $actor = (int)( $row->$actorField ); |
| 211 | |
| 212 | $bad[$id] = $actor; |
| 213 | $this->output( "\t\t$id\t$actor\n" ); |
| 214 | } |
| 215 | |
| 216 | $this->output( "\tFound $count invalid actor IDs.\n" ); |
| 217 | |
| 218 | if ( $count >= $this->getBatchSize() ) { |
| 219 | $this->output( "\tBatch size reached, run again after fixing the current batch.\n" ); |
| 220 | } |
| 221 | |
| 222 | return $bad; |
| 223 | } |
| 224 | |
| 225 | /** |
| 226 | * Overwrite the actor ID in a given set of rows. |
| 227 | * |
| 228 | * @param string $field the database field in which to replace IDs. |
| 229 | * @param array $ids The row IDs of the rows in which the actor ID should be replaced |
| 230 | * @param int $overwrite The actor ID to write to the rows identified by $ids. |
| 231 | * |
| 232 | * @return int |
| 233 | */ |
| 234 | private function overwriteActorIDs( $field, array $ids, int $overwrite ) { |
| 235 | [ $table, $actorField, $idField ] = $this->getTableInfo( $field ); |
| 236 | |
| 237 | $count = count( $ids ); |
| 238 | $this->output( "OVERWRITING $count actor IDs in $table.$actorField with $overwrite...\n" ); |
| 239 | |
| 240 | $dbw = $this->getPrimaryDB(); |
| 241 | |
| 242 | $dbw->newUpdateQueryBuilder() |
| 243 | ->update( $table ) |
| 244 | ->set( [ $actorField => $overwrite ] ) |
| 245 | ->where( [ $idField => $ids ] ) |
| 246 | ->caller( __METHOD__ )->execute(); |
| 247 | |
| 248 | $count = $dbw->affectedRows(); |
| 249 | |
| 250 | $this->waitForReplication(); |
| 251 | $this->output( "\tUpdated $count rows.\n" ); |
| 252 | |
| 253 | return $count; |
| 254 | } |
| 255 | |
| 256 | } |
| 257 | |
| 258 | // @codeCoverageIgnoreStart |
| 259 | $maintClass = FindMissingActors::class; |
| 260 | require_once RUN_MAINTENANCE_IF_MAIN; |
| 261 | // @codeCoverageIgnoreEnd |