Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
100.00% |
115 / 115 |
|
100.00% |
3 / 3 |
CRAP | |
100.00% |
1 / 1 |
| RemoveUnusedAccounts | |
100.00% |
115 / 115 |
|
100.00% |
3 / 3 |
18 | |
100.00% |
1 / 1 |
| __construct | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
1 | |||
| execute | |
100.00% |
85 / 85 |
|
100.00% |
1 / 1 |
13 | |||
| isInactiveAccount | |
100.00% |
26 / 26 |
|
100.00% |
1 / 1 |
4 | |||
| 1 | <?php |
| 2 | /** |
| 3 | * Remove unused user accounts from the database |
| 4 | * An unused account is one which has made no edits |
| 5 | * |
| 6 | * @license GPL-2.0-or-later |
| 7 | * @file |
| 8 | * @ingroup Maintenance |
| 9 | * @author Rob Church <robchur@gmail.com> |
| 10 | */ |
| 11 | |
| 12 | use MediaWiki\Maintenance\Maintenance; |
| 13 | use Wikimedia\Timestamp\TimestampFormat as TS; |
| 14 | |
| 15 | // @codeCoverageIgnoreStart |
| 16 | require_once __DIR__ . '/Maintenance.php'; |
| 17 | // @codeCoverageIgnoreEnd |
| 18 | |
| 19 | /** |
| 20 | * Maintenance script that removes unused user accounts from the database. |
| 21 | * |
| 22 | * @ingroup Maintenance |
| 23 | */ |
| 24 | class RemoveUnusedAccounts extends Maintenance { |
| 25 | public function __construct() { |
| 26 | parent::__construct(); |
| 27 | $this->addOption( 'delete', 'Actually delete the account' ); |
| 28 | $this->addOption( 'ignore-groups', 'List of comma-separated groups to exclude', false, true ); |
| 29 | $this->addOption( 'ignore-touched', 'Skip accounts touched in last N days', false, true ); |
| 30 | } |
| 31 | |
| 32 | public function execute() { |
| 33 | $services = $this->getServiceContainer(); |
| 34 | $userFactory = $services->getUserFactory(); |
| 35 | $userGroupManager = $services->getUserGroupManager(); |
| 36 | $this->output( "Remove unused accounts\n\n" ); |
| 37 | |
| 38 | # Do an initial scan for inactive accounts and report the result |
| 39 | $this->output( "Checking for unused user accounts...\n" ); |
| 40 | $delUser = []; |
| 41 | $delActor = []; |
| 42 | $dbr = $this->getReplicaDB(); |
| 43 | $res = $dbr->newSelectQueryBuilder() |
| 44 | ->select( [ 'user_id', 'user_name', 'user_touched', 'actor_id' ] ) |
| 45 | ->from( 'user' ) |
| 46 | ->leftJoin( 'actor', null, 'user_id = actor_user' ) |
| 47 | ->caller( __METHOD__ )->fetchResultSet(); |
| 48 | if ( $this->hasOption( 'ignore-groups' ) ) { |
| 49 | $excludedGroups = explode( ',', $this->getOption( 'ignore-groups' ) ); |
| 50 | } else { |
| 51 | $excludedGroups = []; |
| 52 | } |
| 53 | $touched = $this->getOption( 'ignore-touched', "1" ); |
| 54 | if ( !ctype_digit( $touched ) ) { |
| 55 | $this->fatalError( "Please put a valid positive integer on the --ignore-touched parameter." ); |
| 56 | } |
| 57 | $touchedSeconds = 86400 * $touched; |
| 58 | foreach ( $res as $row ) { |
| 59 | # Check the account, but ignore it if it's within a $excludedGroups |
| 60 | # group or if it's touched within the $touchedSeconds seconds. |
| 61 | $instance = $userFactory->newFromId( $row->user_id ); |
| 62 | if ( count( |
| 63 | array_intersect( $userGroupManager->getUserEffectiveGroups( $instance ), $excludedGroups ) ) == 0 |
| 64 | && $this->isInactiveAccount( $row->actor_id ?? null, true ) |
| 65 | && wfTimestamp( TS::UNIX, $row->user_touched ) < wfTimestamp( TS::UNIX, time() - $touchedSeconds |
| 66 | ) |
| 67 | ) { |
| 68 | # Inactive; print out the name and flag it |
| 69 | $delUser[] = $row->user_id; |
| 70 | if ( isset( $row->actor_id ) && $row->actor_id ) { |
| 71 | $delActor[] = $row->actor_id; |
| 72 | } |
| 73 | $this->output( $row->user_name . "\n" ); |
| 74 | } |
| 75 | } |
| 76 | $count = count( $delUser ); |
| 77 | $this->output( "...found {$count}.\n" ); |
| 78 | |
| 79 | # If required, go back and delete each marked account |
| 80 | if ( $count > 0 && $this->hasOption( 'delete' ) ) { |
| 81 | $this->output( "\nDeleting unused accounts..." ); |
| 82 | $dbw = $this->getPrimaryDB(); |
| 83 | $dbw->newDeleteQueryBuilder() |
| 84 | ->deleteFrom( 'user' ) |
| 85 | ->where( [ 'user_id' => $delUser ] ) |
| 86 | ->caller( __METHOD__ )->execute(); |
| 87 | # Keep actor rows referenced from block |
| 88 | $keep = $dbw->newSelectQueryBuilder() |
| 89 | ->select( 'bl_by_actor' ) |
| 90 | ->from( 'block' ) |
| 91 | ->where( [ 'bl_by_actor' => $delActor ] ) |
| 92 | ->caller( __METHOD__ )->fetchFieldValues(); |
| 93 | $del = array_diff( $delActor, $keep ); |
| 94 | if ( $del ) { |
| 95 | $dbw->newDeleteQueryBuilder() |
| 96 | ->deleteFrom( 'actor' ) |
| 97 | ->where( [ 'actor_id' => $del ] ) |
| 98 | ->caller( __METHOD__ )->execute(); |
| 99 | } |
| 100 | $dbw->newDeleteQueryBuilder() |
| 101 | ->deleteFrom( 'user_groups' ) |
| 102 | ->where( [ 'ug_user' => $delUser ] ) |
| 103 | ->caller( __METHOD__ )->execute(); |
| 104 | $dbw->newDeleteQueryBuilder() |
| 105 | ->deleteFrom( 'user_former_groups' ) |
| 106 | ->where( [ 'ufg_user' => $delUser ] ) |
| 107 | ->caller( __METHOD__ )->execute(); |
| 108 | $dbw->newDeleteQueryBuilder() |
| 109 | ->deleteFrom( 'user_properties' ) |
| 110 | ->where( [ 'up_user' => $delUser ] ) |
| 111 | ->caller( __METHOD__ )->execute(); |
| 112 | $dbw->newDeleteQueryBuilder() |
| 113 | ->deleteFrom( 'logging' ) |
| 114 | ->where( [ 'log_actor' => $delActor ] ) |
| 115 | ->caller( __METHOD__ )->execute(); |
| 116 | $dbw->newDeleteQueryBuilder() |
| 117 | ->deleteFrom( 'recentchanges' ) |
| 118 | ->where( [ 'rc_actor' => $delActor ] ) |
| 119 | ->caller( __METHOD__ )->execute(); |
| 120 | $this->output( "done.\n" ); |
| 121 | # Update the site_stats.ss_users field |
| 122 | $users = $dbw->newSelectQueryBuilder() |
| 123 | ->select( 'COUNT(*)' ) |
| 124 | ->from( 'user' ) |
| 125 | ->caller( __METHOD__ )->fetchField(); |
| 126 | $dbw->newUpdateQueryBuilder() |
| 127 | ->update( 'site_stats' ) |
| 128 | ->set( [ 'ss_users' => $users ] ) |
| 129 | ->where( [ 'ss_row_id' => 1 ] ) |
| 130 | ->caller( __METHOD__ ) |
| 131 | ->execute(); |
| 132 | } elseif ( $count > 0 ) { |
| 133 | $this->output( "\nRun the script again with --delete to remove them from the database.\n" ); |
| 134 | } |
| 135 | $this->output( "\n" ); |
| 136 | } |
| 137 | |
| 138 | /** |
| 139 | * Could the specified user account be deemed inactive? |
| 140 | * (No edits, no deleted edits, no log entries, no current/old uploads) |
| 141 | * |
| 142 | * @param int|null $actor User's actor ID |
| 143 | * @param bool $primary Perform checking on the primary DB |
| 144 | * @return bool |
| 145 | */ |
| 146 | private function isInactiveAccount( $actor, $primary = false ) { |
| 147 | if ( $actor === null ) { |
| 148 | // There's no longer a way for a user to be active in any of |
| 149 | // these tables without having an actor ID. The only way to link |
| 150 | // to a user row is via an actor row. |
| 151 | return true; |
| 152 | } |
| 153 | |
| 154 | $dbo = $primary ? $this->getPrimaryDB() : $this->getReplicaDB(); |
| 155 | $checks = [ |
| 156 | 'archive' => 'ar', |
| 157 | 'image' => 'img', |
| 158 | 'oldimage' => 'oi', |
| 159 | 'filearchive' => 'fa', |
| 160 | 'revision' => 'rev', |
| 161 | ]; |
| 162 | $count = 0; |
| 163 | |
| 164 | $this->beginTransactionRound( __METHOD__ ); |
| 165 | foreach ( $checks as $table => $prefix ) { |
| 166 | $count += (int)$dbo->newSelectQueryBuilder() |
| 167 | ->select( 'COUNT(*)' ) |
| 168 | ->from( $table ) |
| 169 | ->where( [ "{$prefix}_actor" => $actor ] ) |
| 170 | ->caller( __METHOD__ ) |
| 171 | ->fetchField(); |
| 172 | } |
| 173 | |
| 174 | $count += (int)$dbo->newSelectQueryBuilder() |
| 175 | ->select( 'COUNT(*)' ) |
| 176 | ->from( 'logging' ) |
| 177 | ->where( [ 'log_actor' => $actor, $dbo->expr( 'log_type', '!=', 'newusers' ) ] ) |
| 178 | ->caller( __METHOD__ )->fetchField(); |
| 179 | |
| 180 | $this->commitTransactionRound( __METHOD__ ); |
| 181 | |
| 182 | return $count == 0; |
| 183 | } |
| 184 | } |
| 185 | |
| 186 | // @codeCoverageIgnoreStart |
| 187 | $maintClass = RemoveUnusedAccounts::class; |
| 188 | require_once RUN_MAINTENANCE_IF_MAIN; |
| 189 | // @codeCoverageIgnoreEnd |