maintenance/archives/upgradeLogging.php

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