maintenance/archives/upgradeLogging.php

   1 <?php
   2 /**
   3  * Replication-safe online upgrade script for log_id/log_deleted
   4  *
   5  * @file
   6  * @ingroup MaintenanceArchive
   7  */
   8
   9 require( dirname(__FILE__).'/../commandLine.inc' );
  10
  11 class UpdateLogging {
  12         var $dbw;
  13         var $batchSize = 1000;
  14         var $minTs = false;
  15
  16         function execute() {
  17                 $this->dbw = wfGetDB( DB_MASTER );
  18                 $logging = $this->dbw->tableName( 'logging' );
  19                 $logging_1_10 = $this->dbw->tableName( 'logging_1_10' );
  20                 $logging_pre_1_10 = $this->dbw->tableName( 'logging_pre_1_10' );
  21
  22                 if ( $this->dbw->tableExists( 'logging_pre_1_10' ) && !$this->dbw->tableExists( 'logging' ) ) {
  23                         # Fix previous aborted run
  24                         echo "Cleaning up from previous aborted run\n";
  25                         $this->dbw->query( "RENAME TABLE $logging_pre_1_10 TO $logging", __METHOD__ );
  26                 }
  27
  28                 if ( $this->dbw->tableExists( 'logging_pre_1_10' ) ) {
  29                         echo "This script has already been run to completion\n";
  30                         return;
  31                 }
  32
  33                 # Create the target table
  34                 if ( !$this->dbw->tableExists( 'logging_1_10' ) ) {
  35                         global $wgDBTableOptions;
  36
  37                         $sql = <<<EOT
  38 CREATE TABLE $logging_1_10 (
  39   -- Log ID, for referring to this specific log entry, probably for deletion and such.
  40   log_id int unsigned NOT NULL auto_increment,
  41
  42   -- Symbolic keys for the general log type and the action type
  43   -- within the log. The output format will be controlled by the
  44   -- action field, but only the type controls categorization.
  45   log_type varbinary(10) NOT NULL default '',
  46   log_action varbinary(10) NOT NULL default '',
  47
  48   -- Timestamp. Duh.
  49   log_timestamp binary(14) NOT NULL default '19700101000000',
  50
  51   -- The user who performed this action; key to user_id
  52   log_user int unsigned NOT NULL default 0,
  53
  54   -- Key to the page affected. Where a user is the target,
  55   -- this will point to the user page.
  56   log_namespace int NOT NULL default 0,
  57   log_title varchar(255) binary NOT NULL default '',
  58
  59   -- Freeform text. Interpreted as edit history comments.
  60   log_comment varchar(255) NOT NULL default '',
  61
  62   -- LF separated list of miscellaneous parameters
  63   log_params blob NOT NULL,
  64
  65   -- rev_deleted for logs
  66   log_deleted tinyint unsigned NOT NULL default '0',
  67
  68   PRIMARY KEY log_id (log_id),
  69   KEY type_time (log_type, log_timestamp),
  70   KEY user_time (log_user, log_timestamp),
  71   KEY page_time (log_namespace, log_title, log_timestamp),
  72   KEY times (log_timestamp)
  73
  74 ) $wgDBTableOptions
  75 EOT;
  76                         echo "Creating table logging_1_10\n";
  77                         $this->dbw->query( $sql, __METHOD__ );
  78                 }
  79
  80                 # Synchronise the tables
  81                 echo "Doing initial sync...\n";
  82                 $this->sync( 'logging', 'logging_1_10' );
  83                 echo "Sync done\n\n";
  84
  85                 # Rename the old table away
  86                 echo "Renaming the old table to $logging_pre_1_10\n";
  87                 $this->dbw->query( "RENAME TABLE $logging TO $logging_pre_1_10", __METHOD__ );
  88
  89                 # Copy remaining old rows
  90                 # Done before the new table is active so that $copyPos is accurate
  91                 echo "Doing final sync...\n";
  92                 $this->sync( 'logging_pre_1_10', 'logging_1_10' );
  93
  94                 # Move the new table in
  95                 echo "Moving the new table in...\n";
  96                 $this->dbw->query( "RENAME TABLE $logging_1_10 TO $logging", __METHOD__ );
  97                 echo "Finished.\n";
  98         }
  99
 100         /**
 101          * Copy all rows from $srcTable to $dstTable
 102          */
 103         function sync( $srcTable, $dstTable ) {
 104                 $batchSize = 1000;
 105                 $minTs = $this->dbw->selectField( $srcTable, 'MIN(log_timestamp)', false, __METHOD__ );
 106                 $minTsUnix = wfTimestamp( TS_UNIX, $minTs );
 107                 $numRowsCopied = 0;
 108
 109                 while ( true ) {
 110                         $maxTs = $this->dbw->selectField( $srcTable, 'MAX(log_timestamp)', false, __METHOD__ );
 111                         $copyPos = $this->dbw->selectField( $dstTable, 'MAX(log_timestamp)', false, __METHOD__ );
 112                         $maxTsUnix = wfTimestamp( TS_UNIX, $maxTs );
 113                         $copyPosUnix = wfTimestamp( TS_UNIX, $copyPos );
 114
 115                         if ( $copyPos === null ) {
 116                                 $percent = 0;
 117                         } else {
 118                                 $percent = ( $copyPosUnix - $minTsUnix ) / ( $maxTsUnix - $minTsUnix ) * 100;
 119                         }
 120                         printf( "%s  %.2f%%\n", $copyPos, $percent );
 121
 122                         # Handle all entries with timestamp equal to $copyPos
 123                         if ( $copyPos !== null ) {
 124                                 $numRowsCopied += $this->copyExactMatch( $srcTable, $dstTable, $copyPos );
 125                         }
 126
 127                         # Now copy a batch of rows
 128                         if ( $copyPos === null ) {
 129                                 $conds = false;
 130                         } else {
 131                                 $conds = array( 'log_timestamp > ' . $this->dbw->addQuotes( $copyPos ) );
 132                         }
 133                         $srcRes = $this->dbw->select( $srcTable, '*', $conds, __METHOD__,
 134                                 array( 'LIMIT' => $batchSize, 'ORDER BY' => 'log_timestamp' ) );
 135
 136                         if ( ! $srcRes->numRows() ) {
 137                                 # All done
 138                                 break;
 139                         }
 140
 141                         $batch = array();
 142                         foreach ( $srcRes as $srcRow ) {
 143                                 $batch[] = (array)$srcRow;
 144                         }
 145                         $this->dbw->insert( $dstTable, $batch, __METHOD__ );
 146                         $numRowsCopied += count( $batch );
 147
 148                         wfWaitForSlaves( 5 );
 149                 }
 150                 echo "Copied $numRowsCopied rows\n";
 151         }
 152
 153         function copyExactMatch( $srcTable, $dstTable, $copyPos ) {
 154                 $numRowsCopied = 0;
 155                 $srcRes = $this->dbw->select( $srcTable, '*', array( 'log_timestamp' => $copyPos ), __METHOD__ );
 156                 $dstRes = $this->dbw->select( $dstTable, '*', array( 'log_timestamp' => $copyPos ), __METHOD__ );
 157
 158                 if ( $srcRes->numRows() ) {
 159                         $srcRow = $srcRes->fetchObject();
 160                         $srcFields = array_keys( (array)$srcRow );
 161                         $srcRes->seek( 0 );
 162                         $dstRowsSeen = array();
 163
 164                         # Make a hashtable of rows that already exist in the destination
 165                         foreach ( $dstRes as $dstRow ) {
 166                                 $reducedDstRow = array();
 167                                 foreach ( $srcFields as $field ) {
 168                                         $reducedDstRow[$field] = $dstRow->$field;
 169                                 }
 170                                 $hash = md5( serialize( $reducedDstRow ) );
 171                                 $dstRowsSeen[$hash] = true;
 172                         }
 173
 174                         # Copy all the source rows that aren't already in the destination
 175                         foreach ( $srcRes as $srcRow ) {
 176                                 $hash = md5( serialize( (array)$srcRow ) );
 177                                 if ( !isset( $dstRowsSeen[$hash] ) ) {
 178                                         $this->dbw->insert( $dstTable, (array)$srcRow, __METHOD__ );
 179                                         $numRowsCopied++;
 180                                 }
 181                         }
 182                 }
 183                 return $numRowsCopied;
 184         }
 185 }
 186
 187 $ul = new UpdateLogging;
 188 $ul->execute();
 189