maintenance/convertUtf8.php

   1 <?php
   2 /**
   3  * @package MediaWiki
   4  * @subpackage Maintenance
   5  */
   6
   7 die("This file is not complete; it's checked in so I don't forget it.");
   8
   9 /*
  10 UTF-8 conversion of DOOOOOOOM
  11
  12 1. Lock the wiki
  13 2. Make a convertlist of all pages
  14 3. Enable CONVERTLOCK mode and switch to UTF-8
  15 4. As quick as possible, convert the cur, images, *links, user, etc tables. Clear cache tables.
  16 5. Unlock the wiki. Attempts to access pages on the convertlist will be trapped to read-only.
  17 6. Go through the list, fixing up old revisions. Remove pages from the convertlist.
  18 */
  19
  20
  21 class UtfUpdater {
  22         function UtfUpdater() {
  23                 $this->db =& wfGetDB( DB_MASTER );
  24         }
  25
  26         function toUtf8( $string ) {
  27                 if( function_exists( 'iconv' ) ) {
  28                         # There are likely to be Windows code page 1252 chars in there.
  29                         # Convert them to the proper UTF-8 chars if possible.
  30                         return iconv( 'CP1252', 'UTF-8', $string );
  31                 } else {
  32                         # Will work from plain iso 8859-1 and may corrupt these chars
  33                         return utf8_encode( $string );
  34                 }
  35         }
  36
  37         function clearTable( $table ) {
  38                 print "Clearing $table...\n";
  39                 $tableName = $this->db->tableName( $table );
  40                 $this->db->query( 'TRUNCATE $tableName' );
  41         }
  42
  43         /**
  44          * @param string $table Table to be converted
  45          * @param string $key Primary key, to identify fields in the UPDATE. If NULL, all fields will be used to match.
  46          * @param array $fields List of all fields to grab and convert. If null, will assume you want the $key, and will ask for DISTINCT.
  47          * @param array $timestamp A field which should be updated to the current timestamp on changed records.
  48          * @access private
  49          */
  50         function convertTable( $table, $key, $fields = null, $timestamp = null ) {
  51                 $fname = 'UtfUpdater::convertTable';
  52                 if( $fields ) {
  53                         $distinct = '';
  54                 } else {
  55                         # If working on one key only, there will be multiple rows.
  56                         # Use DISTINCT to return only one and save us some trouble.
  57                         $fields = array( $key );
  58                         $distinct = 'DISTINCT';
  59                 }
  60                 $condition = '';
  61                 foreach( $fields as $field ) {
  62                         if( $condition ) $condition .= ' OR ';
  63                         $condition .= "$field RLIKE '[\x80-\xff]'";
  64                 }
  65                 $res = $this->db->selectArray(
  66                         $table,
  67                         array_merge( $fields, array( $key ) ),
  68                         $condition,
  69                         $fname,
  70                         $distinct );
  71                 print "Converting " . $this->db->numResults( $res ) . " rows from $table:\n";
  72                 $n = 0;
  73                 while( $s = $this->db->fetchObject( $res ) ) {
  74                         $set = array();
  75                         foreach( $fields as $field ) {
  76                                 $set[] = $this->toUtf8( $s->$field );
  77                         }
  78                         if( $timestamp ) {
  79                                 $set[$timestamp] = $this->db->timestamp();
  80                         }
  81                         if( $key ) {
  82                                 $keyCond = array( $key, $s->$key );
  83                         } else {
  84                                 $keyCond = array();
  85                                 foreach( $fields as $field ) {
  86                                         $keyCond[$field] = $s->$field;
  87                                 }
  88                         }
  89                         $this->db->updateArray(
  90                                 $table,
  91                                 $set,
  92                                 $keyCond,
  93                                 $fname );
  94                         if( ++$n % 100 == 0 ) echo "$n\n";
  95                 }
  96                 echo "$n done.\n";
  97                 $this->db->freeResult( $res );
  98         }
  99
 100         function lockTables( $tables ) {
 101                 $query = '';
 102                 foreach( $tables as $table ) {
 103                         $tableName = $this->db->tableName( $table );
 104                         if( $query ) $query .= ', ';
 105                         $query .= '$tableName WRITE';
 106                 }
 107                 $this->db->query( 'LOCK TABLES ' . $query );
 108         }
 109
 110         function updateAll() {
 111                 $this->lockTables( array(
 112                         'linkscc', 'objectcache', 'searchindex', 'querycache',
 113                         'ipblocks', 'user', 'page', 'revision', 'recentchanges',
 114                         'brokenlinks', 'categorylinks', 'imagelinks', 'watchlist',
 115                         'image', 'oldimage', 'archive' ) );
 116
 117                 # These are safe to clear out:
 118                 $this->clearTable( 'linkscc' );
 119                 $this->clearTable( 'objectcache' );
 120
 121                 # These need to be rebuild if used:
 122                 $this->clearTable( 'searchindex' );
 123                 $this->clearTable( 'querycache' );
 124
 125                 # And convert the rest...
 126                 $this->convertTable( 'ipblocks', 'ipb_id', array( 'ipb_reason' ) );
 127                 $this->convertTable( 'user', 'user_id',
 128                         array( 'user_name', 'user_real_name', 'user_options' ),
 129                         'user_touched' );
 130                 $this->convertTable( 'page', 'page_id',
 131                         array( 'page_title' ), 'page_touched' );
 132                 $this->convertTable( 'revision', 'rev_id',
 133                         array( 'rev_user_text', 'rev_comment' ) );
 134
 135                 $this->convertTable( 'recentchanges', 'rc_id',
 136                         array( 'rc_user_text', 'rc_title', 'rc_comment' ) );
 137
 138                 $this->convertTable( 'brokenlinks', 'bl_to' );
 139                 $this->convertTable( 'categorylinks', 'cl_to' );
 140                 $this->convertTable( 'imagelinks', 'il_to' );
 141                 $this->convertTable( 'watchlist', 'wl_title' );
 142
 143                 # FIXME: We'll also need to change the files.
 144                 $this->convertTable( 'image', 'img_name',
 145                         array( 'img_name', 'img_description', 'img_user_text' ) );
 146                 $this->convertTable( 'oldimage', 'archive_name',
 147                         array( 'oi_name', 'oi_archive_name', 'oi_description', 'oi_user_text' ) );
 148
 149                 # Don't change the ar_text entries; use $wgLegacyEncoding to read them at runtime
 150                 $this->convertTable( 'archive', null,
 151                         array( 'ar_title', 'ar_comment', 'ar_user_text' ) );
 152                 echo "Not converting text table: be sure to set \$wgLegacyEncoding!\n";
 153
 154                 $this->db->query( 'UNLOCK TABLES' );
 155         }
 156
 157 }
 158
 159 ?>