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.
  16  * Clear cache tables.
  17  * 5. Unlock the wiki. Attempts to access pages on the convertlist will be
  18  * trapped to read-only.
  19  * 6. Go through the list, fixing up old revisions. Remove pages from the
  20  * convertlist.
  21  */
  22
  23 /**
  24  * @package MediaWiki
  25  * @subpackage Maintenance
  26  */
  27 class UtfUpdater {
  28         /** Constructor, set the database */
  29         function UtfUpdater() {
  30                 $this->db =& wfGetDB( DB_MASTER );
  31         }
  32
  33         /**
  34          * @param string $string A string to be converted to UTF-8
  35          */
  36         function toUtf8( $string ) {
  37                 if( function_exists( 'iconv' ) ) {
  38                         # There are likely to be Windows code page 1252 chars in there.
  39                         # Convert them to the proper UTF-8 chars if possible.
  40                         return iconv( 'CP1252', 'UTF-8', $string );
  41                 } else {
  42                         # Will work from plain iso 8859-1 and may corrupt these chars
  43                         return utf8_encode( $string );
  44                 }
  45         }
  46
  47         /**
  48          * Truncate a table.
  49          * @param string $table The table name to be truncated
  50          */
  51         function clearTable( $table ) {
  52                 print "Clearing $table...\n";
  53                 $tableName = $this->db->tableName( $table );
  54                 $this->db->query( 'TRUNCATE $tableName' );
  55         }
  56
  57         /**
  58          * @param string $table Table to be converted
  59          * @param string $key Primary key, to identify fields in the UPDATE. If NULL, all fields will be used to match.
  60          * @param array $fields List of all fields to grab and convert. If null, will assume you want the $key, and will ask for DISTINCT.
  61          * @param array $timestamp A field which should be updated to the current timestamp on changed records.
  62          * @access private
  63          */
  64         function convertTable( $table, $key, $fields = null, $timestamp = null ) {
  65                 $fname = 'UtfUpdater::convertTable';
  66                 if( $fields ) {
  67                         $distinct = '';
  68                 } else {
  69                         # If working on one key only, there will be multiple rows.
  70                         # Use DISTINCT to return only one and save us some trouble.
  71                         $fields = array( $key );
  72                         $distinct = 'DISTINCT';
  73                 }
  74                 $condition = '';
  75                 foreach( $fields as $field ) {
  76                         if( $condition ) $condition .= ' OR ';
  77                         $condition .= "$field RLIKE '[\x80-\xff]'";
  78                 }
  79                 $res = $this->db->selectArray(
  80                         $table,
  81                         array_merge( $fields, array( $key ) ),
  82                         $condition,
  83                         $fname,
  84                         $distinct );
  85                 print "Converting " . $this->db->numResults( $res ) . " rows from $table:\n";
  86                 $n = 0;
  87                 while( $s = $this->db->fetchObject( $res ) ) {
  88                         $set = array();
  89                         foreach( $fields as $field ) {
  90                                 $set[] = $this->toUtf8( $s->$field );
  91                         }
  92                         if( $timestamp ) {
  93                                 $set[$timestamp] = $this->db->timestamp();
  94                         }
  95                         if( $key ) {
  96                                 $keyCond = array( $key, $s->$key );
  97                         } else {
  98                                 $keyCond = array();
  99                                 foreach( $fields as $field ) {
 100                                         $keyCond[$field] = $s->$field;
 101                                 }
 102                         }
 103                         $this->db->updateArray(
 104                                 $table,
 105                                 $set,
 106                                 $keyCond,
 107                                 $fname );
 108                         if( ++$n % 100 == 0 ) echo "$n\n";
 109                 }
 110                 echo "$n done.\n";
 111                 $this->db->freeResult( $res );
 112         }
 113
 114         /**
 115          * Lock tables.
 116          * @param array $tables An array of table to be locked.
 117          */
 118         function lockTables( $tables ) {
 119                 $query = '';
 120                 foreach( $tables as $table ) {
 121                         $tableName = $this->db->tableName( $table );
 122                         if( $query ) $query .= ', ';
 123                         $query .= '$tableName WRITE';
 124                 }
 125                 $this->db->query( 'LOCK TABLES ' . $query );
 126         }
 127
 128         /**
 129          * @todo document
 130          */
 131         function updateAll() {
 132                 $this->lockTables( array(
 133                         'linkscc', 'objectcache', 'searchindex', 'querycache',
 134                         'ipblocks', 'user', 'page', 'revision', 'recentchanges',
 135                         'brokenlinks', 'categorylinks', 'imagelinks', 'watchlist',
 136                         'image', 'oldimage', 'archive' ) );
 137
 138                 # These are safe to clear out:
 139                 $this->clearTable( 'linkscc' );
 140                 $this->clearTable( 'objectcache' );
 141
 142                 # These need to be rebuild if used:
 143                 $this->clearTable( 'searchindex' );
 144                 $this->clearTable( 'querycache' );
 145
 146                 # And convert the rest...
 147                 $this->convertTable( 'ipblocks', 'ipb_id', array( 'ipb_reason' ) );
 148                 $this->convertTable( 'user', 'user_id',
 149                         array( 'user_name', 'user_real_name', 'user_options' ),
 150                         'user_touched' );
 151                 $this->convertTable( 'page', 'page_id',
 152                         array( 'page_title' ), 'page_touched' );
 153                 $this->convertTable( 'revision', 'rev_id',
 154                         array( 'rev_user_text', 'rev_comment' ) );
 155
 156                 $this->convertTable( 'recentchanges', 'rc_id',
 157                         array( 'rc_user_text', 'rc_title', 'rc_comment' ) );
 158
 159                 $this->convertTable( 'pagelinks', 'pl_title' );
 160                 $this->convertTable( 'categorylinks', 'cl_to' );
 161                 $this->convertTable( 'imagelinks', 'il_to' );
 162                 $this->convertTable( 'watchlist', 'wl_title' );
 163
 164                 # FIXME: We'll also need to change the files.
 165                 $this->convertTable( 'image', 'img_name',
 166                         array( 'img_name', 'img_description', 'img_user_text' ) );
 167                 $this->convertTable( 'oldimage', 'archive_name',
 168                         array( 'oi_name', 'oi_archive_name', 'oi_description', 'oi_user_text' ) );
 169
 170                 # Don't change the ar_text entries; use $wgLegacyEncoding to read them at runtime
 171                 $this->convertTable( 'archive', null,
 172                         array( 'ar_title', 'ar_comment', 'ar_user_text' ) );
 173                 echo "Not converting text table: be sure to set \$wgLegacyEncoding!\n";
 174
 175                 $this->db->query( 'UNLOCK TABLES' );
 176         }
 177
 178 }
 179 ?>