maintenance/populateParentId.php

   1 <?php
   2 /*
   3  * Makes the required database updates for rev_parent_id
   4  * to be of any use. It can be used for some simple tracking
   5  * and to find new page edits by users.
   6  */
   7
   8 require_once( "Maintenance.php" );
   9
  10 class PopulateParentId extends Maintenance {
  11
  12         // Batch size
  13         const BATCH_SIZE = 200;
  14
  15         public function __construct() {
  16                 parent::__construct();
  17                 $this->mDescription = "Populates rev_parent_id";
  18         }
  19
  20         public function execute() {
  21                 $db = wfGetDB( DB_MASTER );
  22                 if ( !$db->tableExists( 'revision' ) ) {
  23                         $this->error( "revision table does not exist\n", true );
  24                 }
  25                 $this->output( "Populating rev_parent_id column\n" );
  26                 $start = $db->selectField( 'revision', 'MIN(rev_id)', false, __FUNCTION__ );
  27                 $end = $db->selectField( 'revision', 'MAX(rev_id)', false, __FUNCTION__ );
  28                 if( is_null( $start ) || is_null( $end ) ){
  29                         $this->output( "...revision table seems to be empty.\n" );
  30                         $db->insert( 'updatelog',
  31                                 array( 'ul_key' => 'populate rev_parent_id' ),
  32                                 __FUNCTION__,
  33                                 'IGNORE' );
  34                         return;
  35                 }
  36                 # Do remaining chunk
  37                 $end += self::BATCH_SIZE - 1;
  38                 $blockStart = intval( $start );
  39                 $blockEnd = intval( $start ) + self::BATCH_SIZE - 1;
  40                 $count = 0;
  41                 $changed = 0;
  42                 while( $blockEnd <= $end ) {
  43                         $this->output( "...doing rev_id from $blockStart to $blockEnd\n" );
  44                         $cond = "rev_id BETWEEN $blockStart AND $blockEnd";
  45                         $res = $db->select( 'revision',
  46                                 array('rev_id','rev_page','rev_timestamp','rev_parent_id'),
  47                                 $cond, __FUNCTION__ );
  48                         # Go through and update rev_parent_id from these rows.
  49                         # Assume that the previous revision of the title was
  50                         # the original previous revision of the title when the
  51                         # edit was made...
  52                         foreach( $res as $row ) {
  53                                 # First, check rows with the same timestamp other than this one
  54                                 # with a smaller rev ID. The highest ID "wins". This avoids loops
  55                                 # as timestamp can only decrease and never loops with IDs (from parent to parent)
  56                                 $previousID = $db->selectField( 'revision', 'rev_id',
  57                                         array( 'rev_page' => $row->rev_page, 'rev_timestamp' => $row->rev_timestamp,
  58                                                 "rev_id < " . intval( $row->rev_id ) ),
  59                                         __FUNCTION__,
  60                                         array( 'ORDER BY' => 'rev_id DESC' ) );
  61                                 # If there are none, check the the highest ID with a lower timestamp
  62                                 if( !$previousID ) {
  63                                         # Get the highest older timestamp
  64                                         $lastTimestamp = $db->selectField( 'revision', 'rev_timestamp',
  65                                                 array( 'rev_page' => $row->rev_page, "rev_timestamp < " . $db->addQuotes( $row->rev_timestamp ) ),
  66                                                 __FUNCTION__,
  67                                                 array( 'ORDER BY' => 'rev_timestamp DESC' ) );
  68                                         # If there is one, let the highest rev ID win
  69                                         if( $lastTimestamp ) {
  70                                                 $previousID = $db->selectField( 'revision', 'rev_id',
  71                                                         array( 'rev_page' => $row->rev_page, 'rev_timestamp' => $lastTimestamp ),
  72                                                         __FUNCTION__,
  73                                                         array( 'ORDER BY' => 'rev_id DESC' ) );
  74                                         }
  75                                 }
  76                                 $previousID = intval($previousID);
  77                                 if( $previousID != $row->rev_parent_id )
  78                                         $changed++;
  79                                 # Update the row...
  80                                 $db->update( 'revision',
  81                                         array( 'rev_parent_id' => $previousID ),
  82                                         array( 'rev_id' => $row->rev_id ),
  83                                         __FUNCTION__ );
  84                                 $count++;
  85                         }
  86                         $blockStart += self::BATCH_SIZE - 1;
  87                         $blockEnd += self::BATCH_SIZE - 1;
  88                         wfWaitForSlaves( 5 );
  89                 }
  90                 $logged = $db->insert( 'updatelog',
  91                         array( 'ul_key' => 'populate rev_parent_id' ),
  92                         __FUNCTION__,
  93                         'IGNORE' );
  94                 if( $logged ) {
  95                         $this->output( "rev_parent_id population complete ... {$count} rows [{$changed} changed]\n" );
  96                         return true;
  97                 } else {
  98                         $this->output( "Could not insert rev_parent_id population row.\n" );
  99                         return false;
 100                 }
 101         }
 102 }
 103
 104 $maintClass = "PopulateParentId";
 105 require_once( DO_MAINTENANCE );