function __construct() {
parent::__construct();
- $this->mDescription = "Find unregistered files in the 'public' repo zone.";
+ $this->addDescription( "Find unregistered files in the 'public' repo zone." );
$this->addOption( 'subdir',
'Only scan files in this subdirectory (e.g. "a/a0")', false, true );
$this->addOption( 'verbose', "Mention file paths checked" );
$this->output( "Scanning files under $directory:\n" );
}
- $list = $repo->getBackend()->getFileList( array( 'dir' => $directory ) );
+ $list = $repo->getBackend()->getFileList( [ 'dir' => $directory ] );
if ( $list === null ) {
$this->error( "Could not get file listing.", 1 );
}
- $nameBatch = array();
+ $pathBatch = [];
foreach ( $list as $path ) {
if ( preg_match( '#^(thumb|deleted)/#', $path ) ) {
continue; // handle ugly nested containers on stock installs
}
- $nameBatch[] = basename( $path );
- if ( count( $nameBatch ) >= $this->mBatchSize ) {
- $this->checkFiles( $repo, $nameBatch, $verbose );
- $nameBatch = array();
+ $pathBatch[] = $path;
+ if ( count( $pathBatch ) >= $this->mBatchSize ) {
+ $this->checkFiles( $repo, $pathBatch, $verbose );
+ $pathBatch = [];
}
}
- $this->checkFiles( $repo, $nameBatch, $verbose );
+ $this->checkFiles( $repo, $pathBatch, $verbose );
}
- protected function checkFiles( LocalRepo $repo, array $names, $verbose ) {
- if ( !count( $names ) ) {
+ protected function checkFiles( LocalRepo $repo, array $paths, $verbose ) {
+ if ( !count( $paths ) ) {
return;
}
$dbr = $repo->getSlaveDB();
- $imgIN = array();
- $oiWheres = array();
- foreach ( $names as $name ) {
- if ( strpos( $name, '!' ) !== false ) {
+ $curNames = [];
+ $oldNames = [];
+ $imgIN = [];
+ $oiWheres = [];
+ foreach ( $paths as $path ) {
+ $name = basename( $path );
+ if ( preg_match( '#^archive/#', $path ) ) {
if ( $verbose ) {
$this->output( "Checking old file $name\n" );
}
- list( , $base ) = explode( '!', $name ); // <TS_MW>!<img_name>
+ $oldNames[] = $name;
+ list( , $base ) = explode( '!', $name, 2 ); // <TS_MW>!<img_name>
$oiWheres[] = $dbr->makeList(
- array( 'oi_name' => $base, 'oi_archive_name' => $name ),
+ [ 'oi_name' => $base, 'oi_archive_name' => $name ],
LIST_AND
);
} else {
$this->output( "Checking current file $name\n" );
}
+ $curNames[] = $name;
$imgIN[] = $name;
}
}
$res = $dbr->query(
$dbr->unionQueries(
- array(
+ [
$dbr->selectSQLText(
'image',
- array( 'name' => 'img_name' ),
- array( 'img_name' => $imgIN )
+ [ 'name' => 'img_name', 'old' => 0 ],
+ $imgIN ? [ 'img_name' => $imgIN ] : '1=0'
),
$dbr->selectSQLText(
'oldimage',
- array( 'name' => 'oi_archive_name' ),
- $dbr->makeList( $oiWheres, LIST_OR )
+ [ 'name' => 'oi_archive_name', 'old' => 1 ],
+ $oiWheres ? $dbr->makeList( $oiWheres, LIST_OR ) : '1=0'
)
- ),
+ ],
true // UNION ALL (performance)
),
__METHOD__
);
- $namesFound = array();
+ $curNamesFound = [];
+ $oldNamesFound = [];
foreach ( $res as $row ) {
- $namesFound[] = $row->name;
+ if ( $row->old ) {
+ $oldNamesFound[] = $row->name;
+ } else {
+ $curNamesFound[] = $row->name;
+ }
}
- $namesOrphans = array_diff( $names, $namesFound );
- foreach ( $namesOrphans as $name ) {
+ foreach ( array_diff( $curNames, $curNamesFound ) as $name ) {
+ $file = $repo->newFile( $name );
// Print name and public URL to ease recovery
- if ( strpos( $name, '!' ) !== false ) {
- list( , $base ) = explode( '!', $name ); // <TS_MW>!<img_name>
- $file = $repo->newFromArchiveName( Title::makeTitle( NS_FILE, $base ), $name );
+ if ( $file ) {
+ $this->output( $name . "\n" . $file->getCanonicalUrl() . "\n\n" );
} else {
- $file = $repo->newFile( $name );
+ $this->error( "Cannot get URL for bad file title '$name'" );
}
- $this->output( $name . "\n" . $file->getUrl() . "\n\n" );
+ }
+
+ foreach ( array_diff( $oldNames, $oldNamesFound ) as $name ) {
+ list( , $base ) = explode( '!', $name, 2 ); // <TS_MW>!<img_name>
+ $file = $repo->newFromArchiveName( Title::makeTitle( NS_FILE, $base ), $name );
+ // Print name and public URL to ease recovery
+ $this->output( $name . "\n" . $file->getCanonicalUrl() . "\n\n" );
}
}
}