maintenance/populateParentId.inc

   1 <?php
   2
   3 define( 'BATCH_SIZE', 200 );
   4
   5 function populate_rev_parent_id( $db ) {
   6         wfOut( "Populating rev_parent_id column\n" );
   7         $start = $db->selectField( 'revision', 'MIN(rev_id)', false, __FUNCTION__ );
   8         $end = $db->selectField( 'revision', 'MAX(rev_id)', false, __FUNCTION__ );
   9         if( is_null( $start ) || is_null( $end ) ){
  10                 wfOut( "...revision table seems to be empty.\n" );
  11                 $db->insert( 'updatelog',
  12                         array( 'ul_key' => 'populate rev_parent_id' ),
  13                         __FUNCTION__,
  14                         'IGNORE' );
  15                 return;
  16         }
  17         # Do remaining chunk
  18         $end += BATCH_SIZE - 1;
  19         $blockStart = intval( $start );
  20         $blockEnd = intval( $start ) + BATCH_SIZE - 1;
  21         $count = 0;
  22         $changed = 0;
  23         while( $blockEnd <= $end ) {
  24                 wfOut( "...doing rev_id from $blockStart to $blockEnd\n" );
  25                 $cond = "rev_id BETWEEN $blockStart AND $blockEnd";
  26                 $res = $db->select( 'revision',
  27                         array('rev_id','rev_page','rev_timestamp','rev_parent_id'),
  28                         $cond, __FUNCTION__ );
  29                 # Go through and update rev_parent_id from these rows.
  30                 # Assume that the previous revision of the title was
  31                 # the original previous revision of the title when the
  32                 # edit was made...
  33                 foreach( $res as $row ) {
  34                         # First, check rows with the same timestamp other than this one
  35                         # with a smaller rev ID. The highest ID "wins". This avoids loops
  36                         # as timestamp can only decrease and never loops with IDs (from parent to parent)
  37                         $previousID = $db->selectField( 'revision', 'rev_id',
  38                                 array( 'rev_page' => $row->rev_page, 'rev_timestamp' => $row->rev_timestamp,
  39                                         "rev_id < " . intval( $row->rev_id ) ),
  40                                 __FUNCTION__,
  41                                 array( 'ORDER BY' => 'rev_id DESC' ) );
  42                         # If there are none, check the the highest ID with a lower timestamp
  43                         if( !$previousID ) {
  44                                 # Get the highest older timestamp
  45                                 $lastTimestamp = $db->selectField( 'revision', 'rev_timestamp',
  46                                         array( 'rev_page' => $row->rev_page, "rev_timestamp < " . $db->addQuotes( $row->rev_timestamp ) ),
  47                                         __FUNCTION__,
  48                                         array( 'ORDER BY' => 'rev_timestamp DESC' ) );
  49                                 # If there is one, let the highest rev ID win
  50                                 if( $lastTimestamp ) {
  51                                         $previousID = $db->selectField( 'revision', 'rev_id',
  52                                                 array( 'rev_page' => $row->rev_page, 'rev_timestamp' => $lastTimestamp ),
  53                                                 __FUNCTION__,
  54                                                 array( 'ORDER BY' => 'rev_id DESC' ) );
  55                                 }
  56                         }
  57                         $previousID = intval($previousID);
  58                         if( $previousID != $row->rev_parent_id )
  59                                 $changed++;
  60                         # Update the row...
  61                         $db->update( 'revision',
  62                                 array( 'rev_parent_id' => $previousID ),
  63                                 array( 'rev_id' => $row->rev_id ),
  64                                 __FUNCTION__ );
  65                         $count++;
  66                 }
  67                 $blockStart += BATCH_SIZE - 1;
  68                 $blockEnd += BATCH_SIZE - 1;
  69                 wfWaitForSlaves( 5 );
  70         }
  71         $logged = $db->insert( 'updatelog',
  72                 array( 'ul_key' => 'populate rev_parent_id' ),
  73                 __FUNCTION__,
  74                 'IGNORE' );
  75         if( $logged ) {
  76                 wfOut( "rev_parent_id population complete ... {$count} rows [{$changed} changed]\n" );
  77                 return true;
  78         } else {
  79                 wfOut( "Could not insert rev_parent_id population row.\n" );
  80                 return false;
  81         }
  82 }
  83