maintenance/rebuildImages.php

   1 <?php
   2 /**
   3  * Script to update image metadata records
   4  *
   5  * Usage: php rebuildImages.php [--missing] [--dry-run]
   6  * Options:
   7  *   --missing  Crawl the uploads dir for images without records, and
   8  *              add them only.
   9  *
  10  * Copyright © 2005 Brion Vibber <brion@pobox.com>
  11  * http://www.mediawiki.org/
  12  *
  13  * This program is free software; you can redistribute it and/or modify
  14  * it under the terms of the GNU General Public License as published by
  15  * the Free Software Foundation; either version 2 of the License, or
  16  * (at your option) any later version.
  17  *
  18  * This program is distributed in the hope that it will be useful,
  19  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  20  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  21  * GNU General Public License for more details.
  22  *
  23  * You should have received a copy of the GNU General Public License along
  24  * with this program; if not, write to the Free Software Foundation, Inc.,
  25  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  26  * http://www.gnu.org/copyleft/gpl.html
  27  *
  28  * @file
  29  * @author Brion Vibber <brion at pobox.com>
  30  * @ingroup maintenance
  31  */
  32
  33 require_once( dirname( __FILE__ ) . '/Maintenance.php' );
  34
  35 class ImageBuilder extends Maintenance {
  36         function __construct() {
  37                 parent::__construct();
  38
  39                 $this->mDescription = 'Script to update image metadata records';
  40
  41                 $this->addOption( 'missing', 'Check for files without associated database record' );
  42                 $this->addOption( 'dry-run', 'Only report, don\'t update the database' );
  43         }
  44
  45         public function execute() {
  46                 $this->dbw = wfGetDB( DB_MASTER );
  47                 $this->maxLag = 10; # if slaves are lagged more than 10 secs, wait
  48                 $this->dryrun = $this->hasOption( 'dry-run' );
  49                 if ( $this->dryrun ) {
  50                         $GLOBALS['wgReadOnly'] = 'Dry run mode, image upgrades are suppressed';
  51                 }
  52
  53                 if ( $this->hasOption( 'missing' ) ) {
  54                         $this->crawlMissing();
  55                 } else {
  56                         $this->build();
  57                 }
  58         }
  59
  60         function getRepo() {
  61                 if ( !isset( $this->repo ) ) {
  62                         $this->repo = RepoGroup::singleton()->getLocalRepo();
  63                 }
  64                 return $this->repo;
  65         }
  66
  67         function build() {
  68                 $this->buildImage();
  69                 $this->buildOldImage();
  70         }
  71
  72         function init( $count, $table ) {
  73                 $this->processed = 0;
  74                 $this->updated = 0;
  75                 $this->count = $count;
  76                 $this->startTime = wfTime();
  77                 $this->table = $table;
  78         }
  79
  80         function progress( $updated ) {
  81                 $this->updated += $updated;
  82                 $this->processed++;
  83                 if ( $this->processed % 100 != 0 ) {
  84                         return;
  85                 }
  86                 $portion = $this->processed / $this->count;
  87                 $updateRate = $this->updated / $this->processed;
  88
  89                 $now = wfTime();
  90                 $delta = $now - $this->startTime;
  91                 $estimatedTotalTime = $delta / $portion;
  92                 $eta = $this->startTime + $estimatedTotalTime;
  93                 $rate = $this->processed / $delta;
  94
  95                 $this->output( sprintf( "%s: %6.2f%% done on %s; ETA %s [%d/%d] %.2f/sec <%.2f%% updated>\n",
  96                         wfTimestamp( TS_DB, intval( $now ) ),
  97                         $portion * 100.0,
  98                         $this->table,
  99                         wfTimestamp( TS_DB, intval( $eta ) ),
 100                         $this->processed,
 101                         $this->count,
 102                         $rate,
 103                         $updateRate * 100.0 ) );
 104                 flush();
 105         }
 106
 107         function buildTable( $table, $key, $callback ) {
 108                 $count = $this->dbw->selectField( $table, 'count(*)', '', __METHOD__ );
 109                 $this->init( $count, $table );
 110                 $this->output( "Processing $table...\n" );
 111
 112                 $result = wfGetDB( DB_SLAVE )->select( $table, '*', array(), __METHOD__ );
 113
 114                 foreach ( $result as $row ) {
 115                         $update = call_user_func( $callback, $row, null );
 116                         if ( $update ) {
 117                                 $this->progress( 1 );
 118                         } else {
 119                                 $this->progress( 0 );
 120                         }
 121                 }
 122                 $this->output( "Finished $table... $this->updated of $this->processed rows updated\n" );
 123         }
 124
 125         function buildImage() {
 126                 $callback = array( $this, 'imageCallback' );
 127                 $this->buildTable( 'image', 'img_name', $callback );
 128         }
 129
 130         function imageCallback( $row, $copy ) {
 131                 // Create a File object from the row
 132                 // This will also upgrade it
 133                 $file = $this->getRepo()->newFileFromRow( $row );
 134                 return $file->getUpgraded();
 135         }
 136
 137         function buildOldImage() {
 138                 $this->buildTable( 'oldimage', 'oi_archive_name',
 139                         array( $this, 'oldimageCallback' ) );
 140         }
 141
 142         function oldimageCallback( $row, $copy ) {
 143                 // Create a File object from the row
 144                 // This will also upgrade it
 145                 if ( $row->oi_archive_name == '' ) {
 146                         $this->output( "Empty oi_archive_name for oi_name={$row->oi_name}\n" );
 147                         return false;
 148                 }
 149                 $file = $this->getRepo()->newFileFromRow( $row );
 150                 return $file->getUpgraded();
 151         }
 152
 153         function crawlMissing() {
 154                 $repo = RepoGroup::singleton()->getLocalRepo();
 155                 $repo->enumFilesInFS( array( $this, 'checkMissingImage' ) );
 156         }
 157
 158         function checkMissingImage( $fullpath ) {
 159                 $filename = wfBaseName( $fullpath );
 160                 if ( is_dir( $fullpath ) ) {
 161                         return;
 162                 }
 163                 if ( is_link( $fullpath ) ) {
 164                         $this->output( "skipping symlink at $fullpath\n" );
 165                         return;
 166                 }
 167                 $row = $this->dbw->selectRow( 'image',
 168                         array( 'img_name' ),
 169                         array( 'img_name' => $filename ),
 170                         __METHOD__ );
 171
 172                 if ( $row ) {
 173                         // already known, move on
 174                         return;
 175                 } else {
 176                         $this->addMissingImage( $filename, $fullpath );
 177                 }
 178         }
 179
 180         function addMissingImage( $filename, $fullpath ) {
 181                 $timestamp = $this->dbw->timestamp( filemtime( $fullpath ) );
 182
 183                 global $wgContLang;
 184                 $altname = $wgContLang->checkTitleEncoding( $filename );
 185                 if ( $altname != $filename ) {
 186                         if ( $this->dryrun ) {
 187                                 $filename = $altname;
 188                                 $this->output( "Estimating transcoding... $altname\n" );
 189                         } else {
 190                                 $filename = $this->renameFile( $filename );
 191                         }
 192                 }
 193
 194                 if ( $filename == '' ) {
 195                         $this->output( "Empty filename for $fullpath\n" );
 196                         return;
 197                 }
 198                 if ( !$this->dryrun ) {
 199                         $file = wfLocalFile( $filename );
 200                         if ( !$file->recordUpload( '', '(recovered file, missing upload log entry)', '', '', '',
 201                                 false, $timestamp ) )
 202                         {
 203                                 $this->output( "Error uploading file $fullpath\n" );
 204                                 return;
 205                         }
 206                 }
 207                 $this->output( $fullpath . "\n" );
 208         }
 209 }
 210
 211 $maintClass = 'ImageBuilder';
 212 require( RUN_MAINTENANCE_IF_MAIN );