MediaWiki  master
populateParentId.php
Go to the documentation of this file.
1 <?php
26 require_once __DIR__ . '/Maintenance.php';
27 
29 
37  public function __construct() {
38  parent::__construct();
39  $this->addDescription( 'Populates rev_parent_id' );
40  }
41 
42  protected function getUpdateKey() {
43  return 'populate rev_parent_id';
44  }
45 
46  protected function updateSkippedMessage() {
47  return 'rev_parent_id column of revision table already populated.';
48  }
49 
50  protected function doDBUpdates() {
51  $batchSize = $this->getBatchSize();
52  $db = $this->getDB( DB_MASTER );
53  if ( !$db->tableExists( 'revision' ) ) {
54  $this->error( "revision table does not exist" );
55 
56  return false;
57  }
58  $this->output( "Populating rev_parent_id column\n" );
59  $start = $db->selectField( 'revision', 'MIN(rev_id)', '', __FUNCTION__ );
60  $end = $db->selectField( 'revision', 'MAX(rev_id)', '', __FUNCTION__ );
61  if ( $start === null || $end === null ) {
62  $this->output( "...revision table seems to be empty, nothing to do.\n" );
63 
64  return true;
65  }
66  # Do remaining chunk
67  $blockStart = intval( $start );
68  $blockEnd = intval( $start ) + $batchSize - 1;
69  $count = 0;
70  $changed = 0;
71  $lbFactory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
72  while ( $blockStart <= $end ) {
73  $this->output( "...doing rev_id from $blockStart to $blockEnd\n" );
74  $cond = "rev_id BETWEEN $blockStart AND $blockEnd";
75  $res = $db->select( 'revision',
76  [ 'rev_id', 'rev_page', 'rev_timestamp', 'rev_parent_id' ],
77  [ $cond, 'rev_parent_id' => null ], __METHOD__ );
78  # Go through and update rev_parent_id from these rows.
79  # Assume that the previous revision of the title was
80  # the original previous revision of the title when the
81  # edit was made...
82  foreach ( $res as $row ) {
83  # First, check rows with the same timestamp other than this one
84  # with a smaller rev ID. The highest ID "wins". This avoids loops
85  # as timestamp can only decrease and never loops with IDs (from parent to parent)
86  $previousID = $db->selectField( 'revision', 'rev_id',
87  [ 'rev_page' => $row->rev_page, 'rev_timestamp' => $row->rev_timestamp,
88  "rev_id < " . intval( $row->rev_id ) ],
89  __METHOD__,
90  [ 'ORDER BY' => 'rev_id DESC' ] );
91  # If there are none, check the highest ID with a lower timestamp
92  if ( !$previousID ) {
93  # Get the highest older timestamp
94  $lastTimestamp = $db->selectField(
95  'revision',
96  'rev_timestamp',
97  [
98  'rev_page' => $row->rev_page,
99  "rev_timestamp < " . $db->addQuotes( $row->rev_timestamp )
100  ],
101  __METHOD__,
102  [ 'ORDER BY' => 'rev_timestamp DESC' ]
103  );
104  # If there is one, let the highest rev ID win
105  if ( $lastTimestamp ) {
106  $previousID = $db->selectField( 'revision', 'rev_id',
107  [ 'rev_page' => $row->rev_page, 'rev_timestamp' => $lastTimestamp ],
108  __METHOD__,
109  [ 'ORDER BY' => 'rev_id DESC' ] );
110  }
111  }
112  $previousID = intval( $previousID );
113  if ( $previousID != $row->rev_parent_id ) {
114  $changed++;
115  }
116  # Update the row...
117  $db->update( 'revision',
118  [ 'rev_parent_id' => $previousID ],
119  [ 'rev_id' => $row->rev_id ],
120  __METHOD__ );
121  $count++;
122  }
123  $blockStart += $batchSize;
124  $blockEnd += $batchSize;
125  $lbFactory->waitForReplication();
126  }
127  $this->output( "rev_parent_id population complete ... {$count} rows [{$changed} changed]\n" );
128 
129  return true;
130  }
131 }
132 
133 $maintClass = PopulateParentId::class;
134 require_once RUN_MAINTENANCE_IF_MAIN;
RUN_MAINTENANCE_IF_MAIN
const RUN_MAINTENANCE_IF_MAIN
Definition: Maintenance.php:38
MediaWiki\MediaWikiServices
MediaWikiServices is the service locator for the application scope of MediaWiki.
Definition: MediaWikiServices.php:146
Maintenance\addDescription
addDescription( $text)
Set the description text.
Definition: Maintenance.php:323
$maintClass
$maintClass
Definition: populateParentId.php:133
PopulateParentId\__construct
__construct()
Default constructor.
Definition: populateParentId.php:37
PopulateParentId\getUpdateKey
getUpdateKey()
Get the update key name to go in the update log table.
Definition: populateParentId.php:42
$res
$res
Definition: testCompression.php:57
LoggedUpdateMaintenance
Class for scripts that perform database maintenance and want to log the update in updatelog so we can...
Definition: LoggedUpdateMaintenance.php:26
PopulateParentId\updateSkippedMessage
updateSkippedMessage()
Message to show that the update was done already and was just skipped.
Definition: populateParentId.php:46
DB_MASTER
const DB_MASTER
Definition: defines.php:26
PopulateParentId\doDBUpdates
doDBUpdates()
Do the actual work.
Definition: populateParentId.php:50
PopulateParentId
Maintenance script that makes the required database updates for rev_parent_id to be of any use.
Definition: populateParentId.php:36
Maintenance\getDB
getDB( $db, $groups=[], $dbDomain=false)
Returns a database to be used by current maintenance script.
Definition: Maintenance.php:1350
Maintenance\getBatchSize
getBatchSize()
Returns batch size.
Definition: Maintenance.php:362
Maintenance\error
error( $err, $die=0)
Throw an error to the user.
Definition: Maintenance.php:457
Maintenance\output
output( $out, $channel=null)
Throw some output to the user.
Definition: Maintenance.php:429