maintenance/populateParentId.inc

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