MediaWiki  master
populateFilearchiveSha1.php
Go to the documentation of this file.
1 <?php
24 require_once __DIR__ . '/Maintenance.php';
25 
27 
35  public function __construct() {
36  parent::__construct();
37  $this->addDescription( 'Populate the fa_sha1 field from fa_storage_key' );
38  }
39 
40  protected function getUpdateKey() {
41  return 'populate fa_sha1';
42  }
43 
44  protected function updateSkippedMessage() {
45  return 'fa_sha1 column of filearchive table already populated.';
46  }
47 
48  public function doDBUpdates() {
49  $startTime = microtime( true );
50  $dbw = $this->getDB( DB_PRIMARY );
51  $table = 'filearchive';
52  $conds = [ 'fa_sha1' => '', 'fa_storage_key IS NOT NULL' ];
53 
54  if ( !$dbw->fieldExists( $table, 'fa_sha1', __METHOD__ ) ) {
55  $this->output( "fa_sha1 column does not exist\n\n", true );
56 
57  return false;
58  }
59 
60  $this->output( "Populating fa_sha1 field from fa_storage_key\n" );
61  $endId = $dbw->selectField( $table, 'MAX(fa_id)', '', __METHOD__ );
62 
63  $batchSize = $this->getBatchSize();
64  $done = 0;
65  $lbFactory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
66 
67  do {
68  $res = $dbw->select(
69  $table,
70  [ 'fa_id', 'fa_storage_key' ],
71  $conds,
72  __METHOD__,
73  [ 'LIMIT' => $batchSize ]
74  );
75 
76  $i = 0;
77  foreach ( $res as $row ) {
78  if ( $row->fa_storage_key == '' ) {
79  // Revision was missing pre-deletion
80  continue;
81  }
82  $sha1 = LocalRepo::getHashFromKey( $row->fa_storage_key );
83  $dbw->update( $table,
84  [ 'fa_sha1' => $sha1 ],
85  [ 'fa_id' => $row->fa_id ],
86  __METHOD__
87  );
88  $lastId = $row->fa_id;
89  $i++;
90  }
91 
92  $done += $i;
93  if ( $i !== $batchSize ) {
94  break;
95  }
96 
97  // print status and let replica DBs catch up
98  $this->output( sprintf(
99  // @phan-suppress-next-line PhanPossiblyUndeclaredVariable $lastId is set for non-empty $res
100  "id %d done (up to %d), %5.3f%% \r", $lastId, $endId, $lastId / $endId * 100 ) );
101  $lbFactory->waitForReplication();
102  } while ( true );
103 
104  $processingTime = microtime( true ) - $startTime;
105  $this->output( sprintf( "\nDone %d files in %.1f seconds\n", $done, $processingTime ) );
106 
107  // we only updated *some* files, don't log
108  return true;
109  }
110 }
111 
112 $maintClass = PopulateFilearchiveSha1::class;
113 require_once RUN_MAINTENANCE_IF_MAIN;
static getHashFromKey( $key)
Gets the SHA1 hash from a storage key.
Definition: LocalRepo.php:223
Class for scripts that perform database maintenance and want to log the update in updatelog so we can...
getDB( $db, $groups=[], $dbDomain=false)
Returns a database to be used by current maintenance script.
output( $out, $channel=null)
Throw some output to the user.
getBatchSize()
Returns batch size.
addDescription( $text)
Set the description text.
Service locator for MediaWiki core services.
Maintenance script to populate the fa_sha1 field.
__construct()
Default constructor.
updateSkippedMessage()
Message to show that the update was done already and was just skipped.
doDBUpdates()
Do the actual work.
getUpdateKey()
Get the update key name to go in the update log table.
const DB_PRIMARY
Definition: defines.php:28