MediaWiki  master
cleanupUploadStash.php
Go to the documentation of this file.
1 <?php
28 require_once __DIR__ . '/Maintenance.php';
29 
37 
38  public function __construct() {
39  parent::__construct();
40  $this->addDescription( 'Clean up abandoned files in temporary uploaded file stash' );
41  $this->setBatchSize( 50 );
42  }
43 
44  public function execute() {
45  global $wgUploadStashMaxAge;
46 
47  $repo = RepoGroup::singleton()->getLocalRepo();
48  $tempRepo = $repo->getTempRepo();
49 
50  $dbr = $repo->getReplicaDB();
51 
52  // how far back should this look for files to delete?
53  $cutoff = time() - $wgUploadStashMaxAge;
54 
55  $this->output( "Getting list of files to clean up...\n" );
56  $res = $dbr->select(
57  'uploadstash',
58  'us_key',
59  'us_timestamp < ' . $dbr->addQuotes( $dbr->timestamp( $cutoff ) ),
60  __METHOD__
61  );
62 
63  // Delete all registered stash files...
64  if ( $res->numRows() == 0 ) {
65  $this->output( "No stashed files to cleanup according to the DB.\n" );
66  } else {
67  // finish the read before starting writes.
68  $keys = [];
69  foreach ( $res as $row ) {
70  array_push( $keys, $row->us_key );
71  }
72 
73  $this->output( 'Removing ' . count( $keys ) . " file(s)...\n" );
74  // this could be done some other, more direct/efficient way, but using
75  // UploadStash's own methods means it's less likely to fall accidentally
76  // out-of-date someday
77  $stash = new UploadStash( $repo );
78 
79  $i = 0;
80  foreach ( $keys as $key ) {
81  $i++;
82  try {
83  $stash->getFile( $key, true );
84  $stash->removeFileNoAuth( $key );
85  } catch ( UploadStashException $ex ) {
86  $type = get_class( $ex );
87  $this->output( "Failed removing stashed upload with key: $key ($type)\n" );
88  }
89  if ( $i % 100 == 0 ) {
91  $this->output( "$i\n" );
92  }
93  }
94  $this->output( "$i done\n" );
95  }
96 
97  // Delete all the corresponding thumbnails...
98  $dir = $tempRepo->getZonePath( 'thumb' );
99  $iterator = $tempRepo->getBackend()->getFileList( [ 'dir' => $dir, 'adviseStat' => 1 ] );
100  $this->output( "Deleting old thumbnails...\n" );
101  $i = 0;
102  $batch = []; // operation batch
103  foreach ( $iterator as $file ) {
104  if ( wfTimestamp( TS_UNIX, $tempRepo->getFileTimestamp( "$dir/$file" ) ) < $cutoff ) {
105  $batch[] = [ 'op' => 'delete', 'src' => "$dir/$file" ];
106  if ( count( $batch ) >= $this->getBatchSize() ) {
107  $this->doOperations( $tempRepo, $batch );
108  $i += count( $batch );
109  $batch = [];
110  $this->output( "$i\n" );
111  }
112  }
113  }
114  if ( count( $batch ) ) {
115  $this->doOperations( $tempRepo, $batch );
116  $i += count( $batch );
117  }
118  $this->output( "$i done\n" );
119 
120  // Apparently lots of stash files are not registered in the DB...
121  $dir = $tempRepo->getZonePath( 'public' );
122  $iterator = $tempRepo->getBackend()->getFileList( [ 'dir' => $dir, 'adviseStat' => 1 ] );
123  $this->output( "Deleting orphaned temp files...\n" );
124  if ( strpos( $dir, '/local-temp' ) === false ) { // sanity check
125  $this->fatalError( "Temp repo is not using the temp container." );
126  }
127  $i = 0;
128  $batch = []; // operation batch
129  foreach ( $iterator as $file ) {
130  if ( wfTimestamp( TS_UNIX, $tempRepo->getFileTimestamp( "$dir/$file" ) ) < $cutoff ) {
131  $batch[] = [ 'op' => 'delete', 'src' => "$dir/$file" ];
132  if ( count( $batch ) >= $this->getBatchSize() ) {
133  $this->doOperations( $tempRepo, $batch );
134  $i += count( $batch );
135  $batch = [];
136  $this->output( "$i\n" );
137  }
138  }
139  }
140  if ( count( $batch ) ) {
141  $this->doOperations( $tempRepo, $batch );
142  $i += count( $batch );
143  }
144  $this->output( "$i done\n" );
145  }
146 
147  protected function doOperations( FileRepo $tempRepo, array $ops ) {
148  $status = $tempRepo->getBackend()->doQuickOperations( $ops );
149  if ( !$status->isOK() ) {
150  // @phan-suppress-next-line PhanUndeclaredMethod
151  $this->error( print_r( $status->getErrorsArray(), true ) );
152  }
153  }
154 }
155 
156 $maintClass = CleanupUploadStash::class;
157 require_once RUN_MAINTENANCE_IF_MAIN;
if(PHP_SAPI !='cli-server') if(!isset( $_SERVER['SCRIPT_FILENAME'])) $file
Item class for a filearchive table row.
Definition: router.php:42
const RUN_MAINTENANCE_IF_MAIN
Definition: Maintenance.php:39
$wgUploadStashMaxAge
The maximum age of temporary (incomplete) uploaded files.
error( $err, $die=0)
Throw an error to the user.
Maintenance script to remove old or broken uploads from temporary uploaded file storage and clean up ...
Abstract maintenance class for quickly writing and churning out maintenance scripts with minimal effo...
Definition: Maintenance.php:86
setBatchSize( $s=0)
Set the batch size.
getBackend()
Get the file backend instance.
Definition: FileRepo.php:225
wfTimestamp( $outputtype=TS_UNIX, $ts=0)
Get a timestamp string in one of various formats.
static singleton()
Definition: RepoGroup.php:60
wfWaitForSlaves( $ifWritesSince=null, $wiki=false, $cluster=false, $timeout=null)
Waits for the replica DBs to catch up to the master position.
addDescription( $text)
Set the description text.
output( $out, $channel=null)
Throw some output to the user.
UploadStash is intended to accomplish a few things:
Definition: UploadStash.php:53
getBatchSize()
Returns batch size.
Base class for file repositories.
Definition: FileRepo.php:39
fatalError( $msg, $exitCode=1)
Output a message and terminate the current script.
doOperations(FileRepo $tempRepo, array $ops)
return true
Definition: router.php:92