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         if( is_null( $start ) || is_null( $end ) ){
  26                 echo "...revision table seems to be empty.\n";
  27                 $db->insert( 'updatelog',
  28                         array( 'ul_key' => 'populate rev_parent_id' ),
  29                         __FUNCTION__,
  30                         'IGNORE' );
  31                 return;
  32         }
  33         # Do remaining chunk
  34         $end += BATCH_SIZE - 1;
  35         $blockStart = $start;
  36         $blockEnd = $start + BATCH_SIZE - 1;
  37         $count = 0;
  38         $changed = 0;
  39         while( $blockEnd <= $end ) {
  40                 echo "...doing rev_id from $blockStart to $blockEnd\n";
  41                 $cond = "rev_id BETWEEN $blockStart AND $blockEnd";
  42                 $res = $db->select( 'revision',
  43                         array('rev_id','rev_page','rev_timestamp','rev_parent_id'),
  44                         $cond, __FUNCTION__ );
  45                 # Go through and update rev_parent_id from these rows.
  46                 # Assume that the previous revision of the title was
  47                 # the original previous revision of the title when the
  48                 # edit was made...
  49                 foreach( $res as $row ) {
  50                         # First, check rows with the same timestamp other than this one
  51                         # with a smaller rev ID. The highest ID "wins". This avoids loops
  52                         # as timestamp can only decrease and never loops with IDs (from parent to parent)
  53                         $previousID = $db->selectField( 'revision', 'rev_id',
  54                                 array( 'rev_page' => $row->rev_page, 'rev_timestamp' => $row->rev_timestamp,
  55                                         "rev_id < {$row->rev_id}" ),
  56                                 __FUNCTION__,
  57                                 array( 'ORDER BY' => 'rev_id DESC' ) );
  58                         # If there are none, check the the highest ID with a lower timestamp
  59                         if( !$previousID ) {
  60                                 # Get the highest older timestamp
  61                                 $lastTimestamp = $db->selectField( 'revision', 'rev_timestamp',
  62                                         array( 'rev_page' => $row->rev_page, "rev_timestamp < '{$row->rev_timestamp}'" ),
  63                                         __FUNCTION__,
  64                                         array( 'ORDER BY' => 'rev_timestamp DESC' ) );
  65                                 # If there is one, let the highest rev ID win
  66                                 if( $lastTimestamp ) {
  67                                         $previousID = $db->selectField( 'revision', 'rev_id',
  68                                                 array( 'rev_page' => $row->rev_page, 'rev_timestamp' => $lastTimestamp ),
  69                                                 __FUNCTION__,
  70                                                 array( 'ORDER BY' => 'rev_id DESC' ) );
  71                                 }
  72                         }
  73                         $previousID = intval($previousID);
  74                         if( $previousID != $row->rev_parent_id )
  75                                 $changed++;
  76                         # Update the row...
  77                         $db->update( 'revision',
  78                                 array( 'rev_parent_id' => $previousID ),
  79                                 array( 'rev_id' => $row->rev_id ),
  80                                 __FUNCTION__ );
  81                         $count++;
  82                 }
  83                 $blockStart += BATCH_SIZE - 1;
  84                 $blockEnd += BATCH_SIZE - 1;
  85                 wfWaitForSlaves( 5 );
  86         }
  87         $logged = $db->insert( 'updatelog',
  88                 array( 'ul_key' => 'populate rev_parent_id' ),
  89                 __FUNCTION__,
  90                 'IGNORE' );
  91         if( $logged ) {
  92                 echo "rev_parent_id population complete ... {$count} rows [{$changed} changed]\n";
  93                 return true;
  94         } else {
  95                 echo "Could not insert rev_parent_id population row.\n";
  96                 return false;
  97         }
  98 }
  99
 100