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  * This program is free software; you can redistribute it and/or modify
   8  * it under the terms of the GNU General Public License as published by
   9  * the Free Software Foundation; either version 2 of the License, or
  10  * (at your option) any later version.
  11  *
  12  * This program is distributed in the hope that it will be useful,
  13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  15  * GNU General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU General Public License along
  18  * with this program; if not, write to the Free Software Foundation, Inc.,
  19  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  20  * http://www.gnu.org/copyleft/gpl.html
  21  *
  22  * @ingroup Maintenance
  23  */
  24
  25 require_once( dirname(__FILE__) . '/Maintenance.php' );
  26
  27 class PopulateParentId extends Maintenance {
  28         public function __construct() {
  29                 parent::__construct();
  30                 $this->mDescription = "Populates rev_parent_id";
  31                 $this->setBatchSize( 200 );
  32         }
  33
  34         public function execute() {
  35                 $db = wfGetDB( DB_MASTER );
  36                 if ( !$db->tableExists( 'revision' ) ) {
  37                         $this->error( "revision table does not exist", true );
  38                 }
  39                 $this->output( "Populating rev_parent_id column\n" );
  40                 $start = $db->selectField( 'revision', 'MIN(rev_id)', false, __FUNCTION__ );
  41                 $end = $db->selectField( 'revision', 'MAX(rev_id)', false, __FUNCTION__ );
  42                 if( is_null( $start ) || is_null( $end ) ){
  43                         $this->output( "...revision table seems to be empty.\n" );
  44                         $db->insert( 'updatelog',
  45                                 array( 'ul_key' => 'populate rev_parent_id' ),
  46                                 __METHOD__,
  47                                 'IGNORE' );
  48                         return;
  49                 }
  50                 # Do remaining chunk
  51                 $end += $this->mBatchSize - 1;
  52                 $blockStart = intval( $start );
  53                 $blockEnd = intval( $start ) + $this->mBatchSize - 1;
  54                 $count = 0;
  55                 $changed = 0;
  56                 while( $blockEnd <= $end ) {
  57                         $this->output( "...doing rev_id from $blockStart to $blockEnd\n" );
  58                         $cond = "rev_id BETWEEN $blockStart AND $blockEnd";
  59                         $res = $db->select( 'revision',
  60                                 array('rev_id','rev_page','rev_timestamp','rev_parent_id'),
  61                                 $cond, __METHOD__ );
  62                         # Go through and update rev_parent_id from these rows.
  63                         # Assume that the previous revision of the title was
  64                         # the original previous revision of the title when the
  65                         # edit was made...
  66                         foreach( $res as $row ) {
  67                                 # First, check rows with the same timestamp other than this one
  68                                 # with a smaller rev ID. The highest ID "wins". This avoids loops
  69                                 # as timestamp can only decrease and never loops with IDs (from parent to parent)
  70                                 $previousID = $db->selectField( 'revision', 'rev_id',
  71                                         array( 'rev_page' => $row->rev_page, 'rev_timestamp' => $row->rev_timestamp,
  72                                                 "rev_id < " . intval( $row->rev_id ) ),
  73                                         __METHOD__,
  74                                         array( 'ORDER BY' => 'rev_id DESC' ) );
  75                                 # If there are none, check the the highest ID with a lower timestamp
  76                                 if( !$previousID ) {
  77                                         # Get the highest older timestamp
  78                                         $lastTimestamp = $db->selectField( 'revision', 'rev_timestamp',
  79                                                 array( 'rev_page' => $row->rev_page, "rev_timestamp < " . $db->addQuotes( $row->rev_timestamp ) ),
  80                                                 __METHOD__,
  81                                                 array( 'ORDER BY' => 'rev_timestamp DESC' ) );
  82                                         # If there is one, let the highest rev ID win
  83                                         if( $lastTimestamp ) {
  84                                                 $previousID = $db->selectField( 'revision', 'rev_id',
  85                                                         array( 'rev_page' => $row->rev_page, 'rev_timestamp' => $lastTimestamp ),
  86                                                         __METHOD__,
  87                                                         array( 'ORDER BY' => 'rev_id DESC' ) );
  88                                         }
  89                                 }
  90                                 $previousID = intval($previousID);
  91                                 if( $previousID != $row->rev_parent_id )
  92                                         $changed++;
  93                                 # Update the row...
  94                                 $db->update( 'revision',
  95                                         array( 'rev_parent_id' => $previousID ),
  96                                         array( 'rev_id' => $row->rev_id ),
  97                                         __METHOD__ );
  98                                 $count++;
  99                         }
 100                         $blockStart += $this->mBatchSize - 1;
 101                         $blockEnd += $this->mBatchSize - 1;
 102                         wfWaitForSlaves( 5 );
 103                 }
 104                 $logged = $db->insert( 'updatelog',
 105                         array( 'ul_key' => 'populate rev_parent_id' ),
 106                         __METHOD__,
 107                         'IGNORE' );
 108                 if( $logged ) {
 109                         $this->output( "rev_parent_id population complete ... {$count} rows [{$changed} changed]\n" );
 110                         return true;
 111                 } else {
 112                         $this->output( "Could not insert rev_parent_id population row.\n" );
 113                         return false;
 114                 }
 115         }
 116 }
 117
 118 $maintClass = "PopulateParentId";
 119 require_once( DO_MAINTENANCE );