maintenance/cleanupImages.php

   1 <?php
   2 /*
   3  * Script to clean up broken, unparseable upload filenames.
   4  *
   5  * Usage: php cleanupImages.php [--fix]
   6  * Options:
   7  *   --fix  Actually clean up titles; otherwise just checks for them
   8  *
   9  * Copyright (C) 2005-2006 Brion Vibber <brion@pobox.com>
  10  * http://www.mediawiki.org/
  11  *
  12  * This program is free software; you can redistribute it and/or modify
  13  * it under the terms of the GNU General Public License as published by
  14  * the Free Software Foundation; either version 2 of the License, or
  15  * (at your option) any later version.
  16  *
  17  * This program is distributed in the hope that it will be useful,
  18  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  19  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  20  * GNU General Public License for more details.
  21  *
  22  * You should have received a copy of the GNU General Public License along
  23  * with this program; if not, write to the Free Software Foundation, Inc.,
  24  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  25  * http://www.gnu.org/copyleft/gpl.html
  26  *
  27  * @author Brion Vibber <brion at pobox.com>
  28  * @addtogroup maintenance
  29  */
  30
  31 require_once( 'commandLine.inc' );
  32 require_once( 'cleanupTable.inc' );
  33
  34 class ImageCleanup extends TableCleanup {
  35         function __construct( $dryrun = false ) {
  36                 parent::__construct( 'image', $dryrun );
  37         }
  38
  39         function processPage( $row ) {
  40                 global $wgContLang;
  41
  42                 $source = $row->img_name;
  43                 if( $source == '' ) {
  44                         // Ye olde empty rows. Just kill them.
  45                         $this->killRow( $source );
  46                         return $this->progress( 1 );
  47                 }
  48
  49                 $cleaned = $source;
  50
  51                 // About half of old bad image names have percent-codes
  52                 $cleaned = rawurldecode( $cleaned );
  53
  54                 // Some are old latin-1
  55                 $cleaned = $wgContLang->checkTitleEncoding( $cleaned );
  56
  57                 // Many of remainder look like non-normalized unicode
  58                 $cleaned = UtfNormal::cleanUp( $cleaned );
  59
  60                 $title = Title::makeTitleSafe( NS_IMAGE, $cleaned );
  61
  62                 if( is_null( $title ) ) {
  63                         $this->log( "page $source ($cleaned) is illegal." );
  64                         $safe = $this->buildSafeTitle( $cleaned );
  65                         $this->pokeFile( $source, $safe );
  66                         return $this->progress( 1 );
  67                 }
  68
  69                 if( $title->getDbKey() !== $source ) {
  70                         $munged = $title->getDbKey();
  71                         $this->log( "page $source ($munged) doesn't match self." );
  72                         $this->pokeFile( $source, $munged );
  73                         return $this->progress( 1 );
  74                 }
  75
  76                 $this->progress( 0 );
  77         }
  78
  79         function killRow( $name ) {
  80                 if( $this->dryrun ) {
  81                         $this->log( "DRY RUN: would delete bogus row '$name'" );
  82                 } else {
  83                         $this->log( "deleting bogus row '$name'" );
  84                         $db = wfGetDB( DB_MASTER );
  85                         $db->delete( 'image',
  86                                 array( 'img_name' => $name ),
  87                                 __METHOD__ );
  88                 }
  89         }
  90
  91         function filePath( $name ) {
  92                 if ( !isset( $this->repo ) ) {
  93                         $this->repo = RepoGroup::singleton()->getLocalRepo();
  94                 }
  95                 return $this->repo->getRootDirectory() . '/' . $this->repo->getHashPath( $name ) . $name;
  96         }
  97
  98         function pokeFile( $orig, $new ) {
  99                 $path = $this->filePath( $orig );
 100                 if( !file_exists( $path ) ) {
 101                         $this->log( "missing file: $path" );
 102                         return $this->killRow( $orig );
 103                 }
 104
 105                 $db = wfGetDB( DB_MASTER );
 106                 $version = 0;
 107                 $final = $new;
 108
 109                 while( $db->selectField( 'image', 'img_name',
 110                         array( 'img_name' => $final ), __METHOD__ ) ) {
 111                         $this->log( "Rename conflicts with '$final'..." );
 112                         $version++;
 113                         $final = $this->appendTitle( $new, "_$version" );
 114                 }
 115
 116                 $finalPath = $this->filePath( $final );
 117
 118                 if( $this->dryrun ) {
 119                         $this->log( "DRY RUN: would rename $path to $finalPath" );
 120                 } else {
 121                         $this->log( "renaming $path to $finalPath" );
 122                         $db->begin();
 123                         $db->update( 'image',
 124                                 array( 'img_name' => $final ),
 125                                 array( 'img_name' => $orig ),
 126                                 __METHOD__ );
 127                         $dir = dirname( $finalPath );
 128                         if( !file_exists( $dir ) ) {
 129                                 if( !mkdir( $dir, 0777, true ) ) {
 130                                         $this->log( "RENAME FAILED, COULD NOT CREATE $dir" );
 131                                         $db->rollback();
 132                                         return;
 133                                 }
 134                         }
 135                         if( rename( $path, $finalPath ) ) {
 136                                 $db->commit();
 137                         } else {
 138                                 $this->log( "RENAME FAILED" );
 139                                 $db->rollback();
 140                         }
 141                 }
 142         }
 143
 144         function appendTitle( $name, $suffix ) {
 145                 return preg_replace( '/^(.*)(\..*?)$/',
 146                         "\\1$suffix\\2", $name );
 147         }
 148
 149         function buildSafeTitle( $name ) {
 150                 global $wgLegalTitleChars;
 151                 $x = preg_replace_callback(
 152                         "/([^$wgLegalTitleChars])/",
 153                         array( $this, 'hexChar' ),
 154                         $name );
 155
 156                 $test = Title::makeTitleSafe( NS_IMAGE, $x );
 157                 if( is_null( $test ) || $test->getDbKey() !== $x ) {
 158                         $this->log( "Unable to generate safe title from '$name', got '$x'" );
 159                         return false;
 160                 }
 161
 162                 return $x;
 163         }
 164 }
 165
 166 $wgUser->setName( 'Conversion script' );
 167 $caps = new ImageCleanup( !isset( $options['fix'] ) );
 168 $caps->cleanup();
 169
 170