<?php
-/*
- * Script to clean up broken, unparseable upload filenames.
+/**
+ * Clean up broken, unparseable upload filenames.
*
- * Usage: php cleanupImages.php [--fix]
- * Options:
- * --fix Actually clean up titles; otherwise just checks for them
- *
- * Copyright (C) 2005-2006 Brion Vibber <brion@pobox.com>
- * http://www.mediawiki.org/
+ * Copyright © 2005-2006 Brion Vibber <brion@pobox.com>
+ * https://www.mediawiki.org/
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
* http://www.gnu.org/copyleft/gpl.html
*
+ * @file
* @author Brion Vibber <brion at pobox.com>
- * @addtogroup maintenance
+ * @ingroup Maintenance
*/
-require_once( 'commandLine.inc' );
-require_once( 'cleanupTable.inc' );
+require_once __DIR__ . '/cleanupTable.inc';
+/**
+ * Maintenance script to clean up broken, unparseable upload filenames.
+ *
+ * @ingroup Maintenance
+ */
class ImageCleanup extends TableCleanup {
- function __construct( $dryrun = false ) {
- parent::__construct( 'image', $dryrun );
+ protected $defaultParams = [
+ 'table' => 'image',
+ 'conds' => [],
+ 'index' => 'img_name',
+ 'callback' => 'processRow',
+ ];
+
+ public function __construct() {
+ parent::__construct();
+ $this->addDescription( 'Script to clean up broken, unparseable upload filenames' );
}
- function processPage( $row ) {
+ protected function processRow( $row ) {
global $wgContLang;
$source = $row->img_name;
- if( $source == '' ) {
+ if ( $source == '' ) {
// Ye olde empty rows. Just kill them.
$this->killRow( $source );
+
return $this->progress( 1 );
}
-
+
$cleaned = $source;
-
+
// About half of old bad image names have percent-codes
$cleaned = rawurldecode( $cleaned );
-
+
+ // We also have some HTML entities there
+ $cleaned = Sanitizer::decodeCharReferences( $cleaned );
+
// Some are old latin-1
$cleaned = $wgContLang->checkTitleEncoding( $cleaned );
-
+
// Many of remainder look like non-normalized unicode
- $cleaned = UtfNormal::cleanUp( $cleaned );
-
- $title = Title::makeTitleSafe( NS_IMAGE, $cleaned );
-
- if( is_null( $title ) ) {
- $this->log( "page $source ($cleaned) is illegal." );
+ $cleaned = $wgContLang->normalize( $cleaned );
+
+ $title = Title::makeTitleSafe( NS_FILE, $cleaned );
+
+ if ( is_null( $title ) ) {
+ $this->output( "page $source ($cleaned) is illegal.\n" );
$safe = $this->buildSafeTitle( $cleaned );
+ if ( $safe === false ) {
+ return $this->progress( 0 );
+ }
$this->pokeFile( $source, $safe );
+
return $this->progress( 1 );
}
- if( $title->getDbKey() !== $source ) {
- $munged = $title->getDbKey();
- $this->log( "page $source ($munged) doesn't match self." );
+ if ( $title->getDBkey() !== $source ) {
+ $munged = $title->getDBkey();
+ $this->output( "page $source ($munged) doesn't match self.\n" );
$this->pokeFile( $source, $munged );
+
return $this->progress( 1 );
}
- $this->progress( 0 );
+ return $this->progress( 0 );
}
-
- function killRow( $name ) {
- if( $this->dryrun ) {
- $this->log( "DRY RUN: would delete bogus row '$name'" );
+
+ /**
+ * @param string $name
+ */
+ private function killRow( $name ) {
+ if ( $this->dryrun ) {
+ $this->output( "DRY RUN: would delete bogus row '$name'\n" );
} else {
- $this->log( "deleting bogus row '$name'" );
- $db = wfGetDB( DB_MASTER );
+ $this->output( "deleting bogus row '$name'\n" );
+ $db = $this->getDB( DB_MASTER );
$db->delete( 'image',
- array( 'img_name' => $name ),
+ [ 'img_name' => $name ],
__METHOD__ );
}
}
-
- function filePath( $name ) {
+
+ private function filePath( $name ) {
if ( !isset( $this->repo ) ) {
$this->repo = RepoGroup::singleton()->getLocalRepo();
}
+
return $this->repo->getRootDirectory() . '/' . $this->repo->getHashPath( $name ) . $name;
}
-
- function pokeFile( $orig, $new ) {
+
+ private function imageExists( $name, $db ) {
+ return $db->selectField( 'image', '1', [ 'img_name' => $name ], __METHOD__ );
+ }
+
+ private function pageExists( $name, $db ) {
+ return $db->selectField(
+ 'page',
+ '1',
+ [ 'page_namespace' => NS_FILE, 'page_title' => $name ],
+ __METHOD__
+ );
+ }
+
+ private function pokeFile( $orig, $new ) {
$path = $this->filePath( $orig );
- if( !file_exists( $path ) ) {
- $this->log( "missing file: $path" );
- return $this->killRow( $orig );
+ if ( !file_exists( $path ) ) {
+ $this->output( "missing file: $path\n" );
+ $this->killRow( $orig );
+
+ return;
}
-
- $db = wfGetDB( DB_MASTER );
+
+ $db = $this->getDB( DB_MASTER );
+
+ /*
+ * To prevent key collisions in the update() statements below,
+ * if the target title exists in the image table, or if both the
+ * original and target titles exist in the page table, append
+ * increasing version numbers until the target title exists in
+ * neither. (See also T18916.)
+ */
$version = 0;
$final = $new;
-
- while( $db->selectField( 'image', 'img_name',
- array( 'img_name' => $final ), __METHOD__ ) ) {
- $this->log( "Rename conflicts with '$final'..." );
+ $conflict = ( $this->imageExists( $final, $db ) ||
+ ( $this->pageExists( $orig, $db ) && $this->pageExists( $final, $db ) ) );
+
+ while ( $conflict ) {
+ $this->output( "Rename conflicts with '$final'...\n" );
$version++;
$final = $this->appendTitle( $new, "_$version" );
+ $conflict = ( $this->imageExists( $final, $db ) || $this->pageExists( $final, $db ) );
}
-
+
$finalPath = $this->filePath( $final );
-
- if( $this->dryrun ) {
- $this->log( "DRY RUN: would rename $path to $finalPath" );
+
+ if ( $this->dryrun ) {
+ $this->output( "DRY RUN: would rename $path to $finalPath\n" );
} else {
- $this->log( "renaming $path to $finalPath" );
- $db->begin();
+ $this->output( "renaming $path to $finalPath\n" );
+ // @todo FIXME: Should this use File::move()?
+ $this->beginTransaction( $db, __METHOD__ );
$db->update( 'image',
- array( 'img_name' => $final ),
- array( 'img_name' => $orig ),
+ [ 'img_name' => $final ],
+ [ 'img_name' => $orig ],
+ __METHOD__ );
+ $db->update( 'oldimage',
+ [ 'oi_name' => $final ],
+ [ 'oi_name' => $orig ],
+ __METHOD__ );
+ $db->update( 'page',
+ [ 'page_title' => $final ],
+ [ 'page_title' => $orig, 'page_namespace' => NS_FILE ],
__METHOD__ );
$dir = dirname( $finalPath );
- if( !file_exists( $dir ) ) {
- if( !mkdir( $dir, 0777, true ) ) {
- $this->log( "RENAME FAILED, COULD NOT CREATE $dir" );
- $db->rollback();
+ if ( !file_exists( $dir ) ) {
+ if ( !wfMkdirParents( $dir, null, __METHOD__ ) ) {
+ $this->output( "RENAME FAILED, COULD NOT CREATE $dir" );
+ $this->rollbackTransaction( $db, __METHOD__ );
+
return;
}
}
- if( rename( $path, $finalPath ) ) {
- $db->commit();
+ if ( rename( $path, $finalPath ) ) {
+ $this->commitTransaction( $db, __METHOD__ );
} else {
- $this->log( "RENAME FAILED" );
- $db->rollback();
+ $this->error( "RENAME FAILED" );
+ $this->rollbackTransaction( $db, __METHOD__ );
}
}
}
-
- function appendTitle( $name, $suffix ) {
+
+ private function appendTitle( $name, $suffix ) {
return preg_replace( '/^(.*)(\..*?)$/',
"\\1$suffix\\2", $name );
}
-
- function buildSafeTitle( $name ) {
- global $wgLegalTitleChars;
+
+ private function buildSafeTitle( $name ) {
$x = preg_replace_callback(
- "/([^$wgLegalTitleChars])/",
- array( $this, 'hexChar' ),
+ '/([^' . Title::legalChars() . ']|~)/',
+ [ $this, 'hexChar' ],
$name );
-
- $test = Title::makeTitleSafe( NS_IMAGE, $x );
- if( is_null( $test ) || $test->getDbKey() !== $x ) {
- $this->log( "Unable to generate safe title from '$name', got '$x'" );
+
+ $test = Title::makeTitleSafe( NS_FILE, $x );
+ if ( is_null( $test ) || $test->getDBkey() !== $x ) {
+ $this->error( "Unable to generate safe title from '$name', got '$x'" );
+
return false;
}
-
+
return $x;
}
}
-$wgUser->setName( 'Conversion script' );
-$caps = new ImageCleanup( !isset( $options['fix'] ) );
-$caps->cleanup();
-
-
+$maintClass = "ImageCleanup";
+require_once RUN_MAINTENANCE_IF_MAIN;
scripts.mit.edu / autoinstallsdev / mediawiki.git / blobdiff