Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 111 |
|
0.00% |
0 / 4 |
CRAP | |
0.00% |
0 / 1 |
| PopulateGlobalEditCount | |
0.00% |
0 / 105 |
|
0.00% |
0 / 4 |
380 | |
0.00% |
0 / 1 |
| __construct | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
2 | |||
| init | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| execute | |
0.00% |
0 / 90 |
|
0.00% |
0 / 1 |
210 | |||
| showProgress | |
0.00% |
0 / 9 |
|
0.00% |
0 / 1 |
12 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace MediaWiki\Extension\CentralAuth\Maintenance; |
| 4 | |
| 5 | use Maintenance; |
| 6 | use MediaWiki\Extension\CentralAuth\CentralAuthDatabaseManager; |
| 7 | use MediaWiki\Extension\CentralAuth\CentralAuthServices; |
| 8 | use MediaWiki\MediaWikiServices; |
| 9 | use RuntimeException; |
| 10 | |
| 11 | $IP = getenv( 'MW_INSTALL_PATH' ); |
| 12 | if ( $IP === false ) { |
| 13 | $IP = __DIR__ . '/../../..'; |
| 14 | } |
| 15 | require_once "$IP/maintenance/Maintenance.php"; |
| 16 | |
| 17 | class PopulateGlobalEditCount extends Maintenance { |
| 18 | private const READ_BATCH_SIZE = 1000; |
| 19 | |
| 20 | /** @var CentralAuthDatabaseManager */ |
| 21 | private $databaseManager; |
| 22 | |
| 23 | /** @var int */ |
| 24 | private $lastReportTime = 0; |
| 25 | |
| 26 | public function __construct() { |
| 27 | parent::__construct(); |
| 28 | $this->requireExtension( 'CentralAuth' ); |
| 29 | |
| 30 | // Batch size for write queries |
| 31 | $this->setBatchSize( 100 ); |
| 32 | $this->addOption( 'start', 'gu_id value to start at', false, true ); |
| 33 | } |
| 34 | |
| 35 | private function init() { |
| 36 | $services = MediaWikiServices::getInstance(); |
| 37 | $this->databaseManager = CentralAuthServices::getDatabaseManager( $services ); |
| 38 | } |
| 39 | |
| 40 | public function execute() { |
| 41 | $this->init(); |
| 42 | $dbcr = $this->databaseManager->getCentralReplicaDB(); |
| 43 | $dbcw = $this->databaseManager->getCentralPrimaryDB(); |
| 44 | $lastId = (int)$dbcr->newSelectQueryBuilder() |
| 45 | ->select( 'MAX(gu_id)' ) |
| 46 | ->from( 'globaluser' ) |
| 47 | ->caller( __METHOD__ ) |
| 48 | ->fetchField(); |
| 49 | |
| 50 | $numGlobalAccounts = 0; |
| 51 | $numUpdated = 0; |
| 52 | |
| 53 | $start = (int)$this->getOption( 'start', 0 ); |
| 54 | for ( $batchStartId = $start; $batchStartId < $lastId; $batchStartId += self::READ_BATCH_SIZE ) { |
| 55 | $this->showProgress( $batchStartId, $lastId ); |
| 56 | $batchEndId = $batchStartId + self::READ_BATCH_SIZE - 1; |
| 57 | $res = $dbcr->newSelectQueryBuilder() |
| 58 | ->select( [ |
| 59 | 'lu_global_id', |
| 60 | 'lu_local_id', |
| 61 | 'lu_wiki', |
| 62 | 'gec_count' |
| 63 | ] ) |
| 64 | // The globaluser table is just needed for batch ordering |
| 65 | ->from( 'globaluser' ) |
| 66 | ->join( 'localuser', null, [ 'lu_name=gu_name' ] ) |
| 67 | ->leftJoin( 'global_edit_count', null, [ 'gu_id=gec_user' ] ) |
| 68 | ->where( [ |
| 69 | "gu_id BETWEEN $batchStartId AND $batchEndId", |
| 70 | "lu_global_id <> 0", |
| 71 | ] ) |
| 72 | ->orderBy( [ 'gu_id', 'lu_wiki' ] ) |
| 73 | ->caller( __METHOD__ ) |
| 74 | ->fetchResultSet(); |
| 75 | |
| 76 | /** @var int[][] $localIds All local IDs by wiki */ |
| 77 | $localIds = []; |
| 78 | /** @var int[][] $globalIds The global ID for each wiki/local ID */ |
| 79 | $globalIds = []; |
| 80 | /** @var int[] $originalCounts The gec_count for each global ID */ |
| 81 | $originalCounts = []; |
| 82 | /** @var int[] $countsByGlobalId The final computed counts */ |
| 83 | $countsByGlobalId = []; |
| 84 | |
| 85 | // Read the global result into arrays |
| 86 | foreach ( $res as $row ) { |
| 87 | $localIds[$row->lu_wiki][] = (int)$row->lu_local_id; |
| 88 | if ( !isset( $originalCounts[(int)$row->lu_global_id] ) ) { |
| 89 | $count = $row->gec_count === null ? null : (int)$row->gec_count; |
| 90 | $originalCounts[(int)$row->lu_global_id] = $count; |
| 91 | } |
| 92 | $globalIds[$row->lu_wiki][(int)$row->lu_local_id] = (int)$row->lu_global_id; |
| 93 | $countsByGlobalId[(int)$row->lu_global_id] = 0; |
| 94 | } |
| 95 | |
| 96 | $numGlobalAccounts += count( $originalCounts ); |
| 97 | |
| 98 | // Get the edit counts on each wiki using a batch query |
| 99 | foreach ( $localIds as $wiki => $ids ) { |
| 100 | $dblr = $this->databaseManager->getLocalDB( DB_REPLICA, $wiki ); |
| 101 | $res = $dblr->newSelectQueryBuilder() |
| 102 | ->select( [ |
| 103 | 'user_id', |
| 104 | 'user_editcount' |
| 105 | ] ) |
| 106 | ->from( 'user' ) |
| 107 | ->where( [ 'user_id' => $ids ] ) |
| 108 | ->caller( __METHOD__ ) |
| 109 | ->fetchResultSet(); |
| 110 | |
| 111 | foreach ( $res as $row ) { |
| 112 | if ( !isset( $globalIds[$wiki][$row->user_id] ) ) { |
| 113 | throw new RuntimeException( "lost user ID {$row->user_id}" ); |
| 114 | } |
| 115 | $countsByGlobalId[$globalIds[$wiki][$row->user_id]] += (int)$row->user_editcount; |
| 116 | } |
| 117 | } |
| 118 | |
| 119 | // Find users with a count mismatch |
| 120 | $updates = []; |
| 121 | $inserts = []; |
| 122 | foreach ( $countsByGlobalId as $id => $count ) { |
| 123 | if ( $originalCounts[$id] !== $count ) { |
| 124 | if ( $originalCounts[$id] === null ) { |
| 125 | $inserts[] = [ |
| 126 | 'gec_user' => $id, |
| 127 | 'gec_count' => $count |
| 128 | ]; |
| 129 | } else { |
| 130 | $updates[$id] = $count; |
| 131 | } |
| 132 | } |
| 133 | } |
| 134 | |
| 135 | // Do the writes in small batches |
| 136 | foreach ( array_chunk( $inserts, $this->getBatchSize() ) as $insertBatch ) { |
| 137 | $this->beginTransaction( $dbcw, __METHOD__ ); |
| 138 | $dbcw->newInsertQueryBuilder() |
| 139 | ->insertInto( 'global_edit_count' ) |
| 140 | ->rows( $insertBatch ) |
| 141 | ->caller( __METHOD__ ) |
| 142 | ->execute(); |
| 143 | $this->commitTransaction( $dbcw, __METHOD__ ); |
| 144 | } |
| 145 | |
| 146 | foreach ( array_chunk( $updates, $this->getBatchSize(), true ) as $updateBatch ) { |
| 147 | $this->beginTransaction( $dbcw, __METHOD__ ); |
| 148 | foreach ( $updateBatch as $id => $count ) { |
| 149 | $dbcw->newUpdateQueryBuilder() |
| 150 | ->update( 'global_edit_count' ) |
| 151 | ->set( [ 'gec_count' => $count ] ) |
| 152 | ->where( [ 'gec_user' => $id ] ) |
| 153 | ->caller( __METHOD__ ) |
| 154 | ->execute(); |
| 155 | } |
| 156 | $this->commitTransaction( $dbcw, __METHOD__ ); |
| 157 | } |
| 158 | |
| 159 | $numUpdated += count( $updates ) + count( $inserts ); |
| 160 | } |
| 161 | |
| 162 | $this->showProgress( $lastId, $lastId ); |
| 163 | $this->output( "Complete. Updated $numUpdated of $numGlobalAccounts edit counts.\n" ); |
| 164 | } |
| 165 | |
| 166 | /** |
| 167 | * @param int $position |
| 168 | * @param int $end |
| 169 | */ |
| 170 | private function showProgress( $position, $end ) { |
| 171 | $now = time(); |
| 172 | if ( !$this->isQuiet() && $now - $this->lastReportTime >= 10 ) { |
| 173 | printf( |
| 174 | "... %d / %d (%-.2f%%)\n", |
| 175 | $position, |
| 176 | $end, |
| 177 | $position / $end * 100 |
| 178 | ); |
| 179 | $this->lastReportTime = $now; |
| 180 | } |
| 181 | } |
| 182 | } |
| 183 | |
| 184 | $maintClass = PopulateGlobalEditCount::class; |
| 185 | require RUN_MAINTENANCE_IF_MAIN; |