MediaWiki  master
findOrphanedFiles.php
Go to the documentation of this file.
1 <?php
21 require_once __DIR__ . '/Maintenance.php';
22 
24  public function __construct() {
25  parent::__construct();
26 
27  $this->addDescription( "Find unregistered files in the 'public' repo zone." );
28  $this->addOption( 'subdir',
29  'Only scan files in this subdirectory (e.g. "a/a0")', false, true );
30  $this->addOption( 'verbose', "Mention file paths checked" );
31  $this->setBatchSize( 500 );
32  }
33 
34  public function execute() {
35  $subdir = $this->getOption( 'subdir', '' );
36  $verbose = $this->hasOption( 'verbose' );
37 
38  $repo = RepoGroup::singleton()->getLocalRepo();
39  if ( $repo->hasSha1Storage() ) {
40  $this->fatalError( "Local repo uses SHA-1 file storage names; aborting." );
41  }
42 
43  $directory = $repo->getZonePath( 'public' );
44  if ( $subdir != '' ) {
45  $directory .= "/$subdir/";
46  }
47 
48  if ( $verbose ) {
49  $this->output( "Scanning files under $directory:\n" );
50  }
51 
52  $list = $repo->getBackend()->getFileList( [ 'dir' => $directory ] );
53  if ( $list === null ) {
54  $this->fatalError( "Could not get file listing." );
55  }
56 
57  $pathBatch = [];
58  foreach ( $list as $path ) {
59  if ( preg_match( '#^(thumb|deleted)/#', $path ) ) {
60  continue; // handle ugly nested containers on stock installs
61  }
62 
63  $pathBatch[] = $path;
64  if ( count( $pathBatch ) >= $this->getBatchSize() ) {
65  $this->checkFiles( $repo, $pathBatch, $verbose );
66  $pathBatch = [];
67  }
68  }
69  $this->checkFiles( $repo, $pathBatch, $verbose );
70  }
71 
72  protected function checkFiles( LocalRepo $repo, array $paths, $verbose ) {
73  if ( !count( $paths ) ) {
74  return;
75  }
76 
77  $dbr = $repo->getReplicaDB();
78 
79  $curNames = [];
80  $oldNames = [];
81  $imgIN = [];
82  $oiWheres = [];
83  foreach ( $paths as $path ) {
84  $name = basename( $path );
85  if ( preg_match( '#^archive/#', $path ) ) {
86  if ( $verbose ) {
87  $this->output( "Checking old file $name\n" );
88  }
89 
90  $oldNames[] = $name;
91  list( , $base ) = explode( '!', $name, 2 ); // <TS_MW>!<img_name>
92  $oiWheres[] = $dbr->makeList(
93  [ 'oi_name' => $base, 'oi_archive_name' => $name ],
94  LIST_AND
95  );
96  } else {
97  if ( $verbose ) {
98  $this->output( "Checking current file $name\n" );
99  }
100 
101  $curNames[] = $name;
102  $imgIN[] = $name;
103  }
104  }
105 
106  $res = $dbr->query(
107  $dbr->unionQueries(
108  [
109  $dbr->selectSQLText(
110  'image',
111  [ 'name' => 'img_name', 'old' => 0 ],
112  $imgIN ? [ 'img_name' => $imgIN ] : '1=0'
113  ),
114  $dbr->selectSQLText(
115  'oldimage',
116  [ 'name' => 'oi_archive_name', 'old' => 1 ],
117  $oiWheres ? $dbr->makeList( $oiWheres, LIST_OR ) : '1=0'
118  )
119  ],
120  $dbr::UNION_ALL
121  ),
122  __METHOD__
123  );
124 
125  $curNamesFound = [];
126  $oldNamesFound = [];
127  foreach ( $res as $row ) {
128  if ( $row->old ) {
129  $oldNamesFound[] = $row->name;
130  } else {
131  $curNamesFound[] = $row->name;
132  }
133  }
134 
135  foreach ( array_diff( $curNames, $curNamesFound ) as $name ) {
136  $file = $repo->newFile( $name );
137  // Print name and public URL to ease recovery
138  if ( $file ) {
139  $this->output( $name . "\n" . $file->getCanonicalUrl() . "\n\n" );
140  } else {
141  $this->error( "Cannot get URL for bad file title '$name'" );
142  }
143  }
144 
145  foreach ( array_diff( $oldNames, $oldNamesFound ) as $name ) {
146  list( , $base ) = explode( '!', $name, 2 ); // <TS_MW>!<img_name>
147  $file = $repo->newFromArchiveName( Title::makeTitle( NS_FILE, $base ), $name );
148  // Print name and public URL to ease recovery
149  $this->output( $name . "\n" . $file->getCanonicalUrl() . "\n\n" );
150  }
151  }
152 }
153 
154 $maintClass = FindOrphanedFiles::class;
155 require_once RUN_MAINTENANCE_IF_MAIN;
RUN_MAINTENANCE_IF_MAIN
const RUN_MAINTENANCE_IF_MAIN
Definition: Maintenance.php:39
FindOrphanedFiles
Definition: findOrphanedFiles.php:23
RepoGroup\singleton
static singleton()
Definition: RepoGroup.php:60
Maintenance\fatalError
fatalError( $msg, $exitCode=1)
Output a message and terminate the current script.
Definition: Maintenance.php:508
Maintenance\addDescription
addDescription( $text)
Set the description text.
Definition: Maintenance.php:352
LocalRepo\getReplicaDB
getReplicaDB()
Get a connection to the replica DB.
Definition: LocalRepo.php:469
NS_FILE
const NS_FILE
Definition: Defines.php:66
$file
if(PHP_SAPI !='cli-server') if(!isset( $_SERVER['SCRIPT_FILENAME'])) $file
Item class for a filearchive table row.
Definition: router.php:42
FindOrphanedFiles\execute
execute()
Do the actual work.
Definition: findOrphanedFiles.php:34
Maintenance
Abstract maintenance class for quickly writing and churning out maintenance scripts with minimal effo...
Definition: Maintenance.php:86
$res
$res
Definition: testCompression.php:54
$base
$base
Definition: generateLocalAutoload.php:11
LIST_AND
const LIST_AND
Definition: Defines.php:39
$dbr
$dbr
Definition: testCompression.php:52
FindOrphanedFiles\checkFiles
checkFiles(LocalRepo $repo, array $paths, $verbose)
Definition: findOrphanedFiles.php:72
LIST_OR
const LIST_OR
Definition: Defines.php:42
FileRepo\newFile
newFile( $title, $time=false)
Create a new File object from the local repository.
Definition: FileRepo.php:414
Maintenance\addOption
addOption( $name, $description, $required=false, $withArg=false, $shortName=false, $multiOccurrence=false)
Add a parameter to the script.
Definition: Maintenance.php:271
Title\makeTitle
static makeTitle( $ns, $title, $fragment='', $interwiki='')
Create a new Title from a namespace index and a DB key.
Definition: Title.php:584
LocalRepo\newFromArchiveName
newFromArchiveName( $title, $archiveName)
Definition: LocalRepo.php:86
$maintClass
$maintClass
Definition: findOrphanedFiles.php:154
Maintenance\getOption
getOption( $name, $default=null)
Get an option, or return the default.
Definition: Maintenance.php:306
$path
$path
Definition: NoLocalSettings.php:25
Maintenance\getBatchSize
getBatchSize()
Returns batch size.
Definition: Maintenance.php:390
Maintenance\error
error( $err, $die=0)
Throw an error to the user.
Definition: Maintenance.php:485
Maintenance\output
output( $out, $channel=null)
Throw some output to the user.
Definition: Maintenance.php:457
Maintenance\hasOption
hasOption( $name)
Checks to see if a particular option exists.
Definition: Maintenance.php:292
LocalRepo
A repository that stores files in the local filesystem and registers them in the wiki's own database.
Definition: LocalRepo.php:37
Maintenance\setBatchSize
setBatchSize( $s=0)
Set the batch size.
Definition: Maintenance.php:398
FindOrphanedFiles\__construct
__construct()
Default constructor.
Definition: findOrphanedFiles.php:24