MediaWiki  REL1_22
populateImageSha1.php
Go to the documentation of this file.
00001 <?php
00024 require_once __DIR__ . '/Maintenance.php';
00025 
00031 class PopulateImageSha1 extends LoggedUpdateMaintenance {
00032     public function __construct() {
00033         parent::__construct();
00034         $this->mDescription = "Populate the img_sha1 field";
00035         $this->addOption( 'force', "Recalculate sha1 for rows that already have a value" );
00036         $this->addOption( 'method', "Use 'pipe' to pipe to mysql command line,\n" .
00037             "\t\tdefault uses Database class", false, true );
00038         $this->addOption( 'file', 'Fix for a specific file, without File: namespace prefixed', false, true );
00039     }
00040 
00041     protected function getUpdateKey() {
00042         return 'populate img_sha1';
00043     }
00044 
00045     protected function updateSkippedMessage() {
00046         return 'img_sha1 column of image table already populated.';
00047     }
00048 
00049     public function execute() {
00050         if ( $this->getOption( 'file' ) ) {
00051             $this->doDBUpdates(); // skip update log checks/saves
00052         } else {
00053             parent::execute();
00054         }
00055     }
00056 
00057     public function doDBUpdates() {
00058         $method = $this->getOption( 'method', 'normal' );
00059         $file = $this->getOption( 'file', '' );
00060         $force = $this->getOption( 'force' );
00061         $isRegen = ( $force || $file != '' ); // forced recalculation?
00062 
00063         $t = -microtime( true );
00064         $dbw = wfGetDB( DB_MASTER );
00065         if ( $file != '' ) {
00066             $res = $dbw->select(
00067                 'image',
00068                 array( 'img_name' ),
00069                 array( 'img_name' => $file ),
00070                 __METHOD__
00071             );
00072             if ( !$res ) {
00073                 $this->error( "No such file: $file", true );
00074                 return false;
00075             }
00076             $this->output( "Populating img_sha1 field for specified files\n" );
00077         } else {
00078             if ( $force ) {
00079                 $conds = array();
00080                 $this->output( "Populating and recalculating img_sha1 field\n" );
00081             } else {
00082                 $conds = array( 'img_sha1' => '' );
00083                 $this->output( "Populating img_sha1 field\n" );
00084             }
00085             $res = $dbw->select( 'image', array( 'img_name' ), $conds, __METHOD__ );
00086         }
00087 
00088         $imageTable = $dbw->tableName( 'image' );
00089         $oldImageTable = $dbw->tableName( 'oldimage' );
00090 
00091         if ( $method == 'pipe' ) {
00092             // Opening a pipe allows the SHA-1 operation to be done in parallel
00093             // with the database write operation, because the writes are queued
00094             // in the pipe buffer. This can improve performance by up to a
00095             // factor of 2.
00096             global $wgDBuser, $wgDBserver, $wgDBpassword, $wgDBname;
00097             $cmd = 'mysql -u' . wfEscapeShellArg( $wgDBuser ) .
00098                 ' -h' . wfEscapeShellArg( $wgDBserver ) .
00099                 ' -p' . wfEscapeShellArg( $wgDBpassword, $wgDBname );
00100             $this->output( "Using pipe method\n" );
00101             $pipe = popen( $cmd, 'w' );
00102         }
00103 
00104         $numRows = $res->numRows();
00105         $i = 0;
00106         foreach ( $res as $row ) {
00107             if ( $i % $this->mBatchSize == 0 ) {
00108                 $this->output( sprintf(
00109                     "Done %d of %d, %5.3f%%  \r", $i, $numRows, $i / $numRows * 100 ) );
00110                 wfWaitForSlaves();
00111             }
00112             $file = wfLocalFile( $row->img_name );
00113             if ( !$file ) {
00114                 continue;
00115             }
00116             // Upgrade the current file version...
00117             $sha1 = $file->getRepo()->getFileSha1( $file->getPath() );
00118             if ( strval( $sha1 ) !== '' ) { // file on disk and hashed properly
00119                 if ( $isRegen && $file->getSha1() !== $sha1 ) {
00120                     // The population was probably done already. If the old SHA1
00121                     // does not match, then both fix the SHA1 and the metadata.
00122                     $file->upgradeRow();
00123                 } else {
00124                     $sql = "UPDATE $imageTable SET img_sha1=" . $dbw->addQuotes( $sha1 ) .
00125                         " WHERE img_name=" . $dbw->addQuotes( $file->getName() );
00126                     if ( $method == 'pipe' ) {
00127                         fwrite( $pipe, "$sql;\n" );
00128                     } else {
00129                         $dbw->query( $sql, __METHOD__ );
00130                     }
00131                 }
00132             }
00133             // Upgrade the old file versions...
00134             foreach ( $file->getHistory() as $oldFile ) {
00135                 $sha1 = $oldFile->getRepo()->getFileSha1( $oldFile->getPath() );
00136                 if ( strval( $sha1 ) !== '' ) { // file on disk and hashed properly
00137                     if ( $isRegen && $oldFile->getSha1() !== $sha1 ) {
00138                         // The population was probably done already. If the old SHA1
00139                         // does not match, then both fix the SHA1 and the metadata.
00140                         $oldFile->upgradeRow();
00141                     } else {
00142                         $sql = "UPDATE $oldImageTable SET oi_sha1=" . $dbw->addQuotes( $sha1 ) .
00143                             " WHERE (oi_name=" . $dbw->addQuotes( $oldFile->getName() ) . " AND" .
00144                             " oi_archive_name=" . $dbw->addQuotes( $oldFile->getArchiveName() ) . ")";
00145                         if ( $method == 'pipe' ) {
00146                             fwrite( $pipe, "$sql;\n" );
00147                         } else {
00148                             $dbw->query( $sql, __METHOD__ );
00149                         }
00150                     }
00151                 }
00152             }
00153             $i++;
00154         }
00155         if ( $method == 'pipe' ) {
00156             fflush( $pipe );
00157             pclose( $pipe );
00158         }
00159         $t += microtime( true );
00160         $this->output( sprintf( "\nDone %d files in %.1f seconds\n", $numRows, $t ) );
00161 
00162         return !$file; // we only updated *some* files, don't log
00163     }
00164 }
00165 
00166 $maintClass = "PopulateImageSha1";
00167 require_once RUN_MAINTENANCE_IF_MAIN;