maintenance/populateParentId.php

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