maintenance/upgrade1_5.php

   1 <?php
   2
   3 // Alternate 1.4 -> 1.5 schema upgrade
   4 // This does only the main tables + UTF-8
   5 // and is designed to allow upgrades to interleave
   6 // with other updates on the replication stream so
   7 // that large wikis can be upgraded without disrupting
   8 // other services.
   9 //
  10 // Note: this script DOES NOT apply every update, nor
  11 // will it probably handle much older versions, etc.
  12 // Run this, FOLLOWED BY update.php, for upgrading
  13 // from 1.4.5 release to 1.5.
  14
  15 $options = array( 'step' );
  16
  17 require_once( 'commandLine.inc' );
  18 require_once( 'cleanupDupes.inc' );
  19 require_once( 'userDupes.inc' );
  20 require_once( 'updaters.inc' );
  21
  22 $upgrade = new FiveUpgrade();
  23 $step = isset( $options['step'] ) ? $options['step'] : null;
  24 $upgrade->upgrade( $step );
  25
  26 class FiveUpgrade {
  27         function FiveUpgrade() {
  28                 global $wgDatabase;
  29                 $this->conversionTables = $this->prepareWindows1252();
  30                 $this->dbw =& $this->newConnection();
  31                 $this->dbr =& $this->newConnection();
  32                 $this->dbr->bufferResults( false );
  33                 $this->cleanupSwaps = array();
  34         }
  35
  36         function doing( $step ) {
  37                 return is_null( $this->step ) || $step == $this->step;
  38         }
  39
  40         function upgrade( $step ) {
  41                 $this->step = $step;
  42                 if( $this->doing( 'page' ) )
  43                         $this->upgradePage();
  44                 if( $this->doing( 'links' ) )
  45                         $this->upgradeLinks();
  46                 if( $this->doing( 'user' ) )
  47                         $this->upgradeUser();
  48                 if( $this->doing( 'image' ) )
  49                         $this->upgradeImage();
  50                 if( $this->doing( 'oldimage' ) )
  51                         $this->upgradeOldImage();
  52                 if( $this->doing( 'watchlist' ) )
  53                         $this->upgradeWatchlist();
  54                 if( $this->doing( 'logging' ) )
  55                         $this->upgradeLogging();
  56
  57                 if( $this->doing( 'cleanup' ) )
  58                         $this->upgradeCleanup();
  59         }
  60
  61
  62         /**
  63          * Open a second connection to the master server, with buffering off.
  64          * This will let us stream large datasets in and write in chunks on the
  65          * other end.
  66          * @return Database
  67          * @access private
  68          */
  69         function &newConnection() {
  70                 global $wgDBadminuser, $wgDBadminpassword;
  71                 global $wgDBserver, $wgDBname;
  72                 $db =& new Database( $wgDBserver, $wgDBadminuser, $wgDBadminpassword, $wgDBname );
  73                 return $db;
  74         }
  75
  76         /**
  77          * Prepare a conversion array for converting Windows Code Page 1252 to
  78          * UTF-8. This should provide proper conversion of text that was miscoded
  79          * as Windows-1252 by naughty user-agents, and doesn't rely on an outside
  80          * iconv library.
  81          *
  82          * @return array
  83          * @access private
  84          */
  85         function prepareWindows1252() {
  86                 # Mappings from:
  87                 # http://www.unicode.org/Public/MAPPINGS/VENDORS/MICSFT/WINDOWS/CP1252.TXT
  88                 static $cp1252 = array(
  89                         0x80 => 0x20AC, #EURO SIGN
  90                         0x81 => UNICODE_REPLACEMENT,
  91                         0x82 => 0x201A, #SINGLE LOW-9 QUOTATION MARK
  92                         0x83 => 0x0192, #LATIN SMALL LETTER F WITH HOOK
  93                         0x84 => 0x201E, #DOUBLE LOW-9 QUOTATION MARK
  94                         0x85 => 0x2026, #HORIZONTAL ELLIPSIS
  95                         0x86 => 0x2020, #DAGGER
  96                         0x87 => 0x2021, #DOUBLE DAGGER
  97                         0x88 => 0x02C6, #MODIFIER LETTER CIRCUMFLEX ACCENT
  98                         0x89 => 0x2030, #PER MILLE SIGN
  99                         0x8A => 0x0160, #LATIN CAPITAL LETTER S WITH CARON
 100                         0x8B => 0x2039, #SINGLE LEFT-POINTING ANGLE QUOTATION MARK
 101                         0x8C => 0x0152, #LATIN CAPITAL LIGATURE OE
 102                         0x8D => UNICODE_REPLACEMENT,
 103                         0x8E => 0x017D, #LATIN CAPITAL LETTER Z WITH CARON
 104                         0x8F => UNICODE_REPLACEMENT,
 105                         0x90 => UNICODE_REPLACEMENT,
 106                         0x91 => 0x2018, #LEFT SINGLE QUOTATION MARK
 107                         0x92 => 0x2019, #RIGHT SINGLE QUOTATION MARK
 108                         0x93 => 0x201C, #LEFT DOUBLE QUOTATION MARK
 109                         0x94 => 0x201D, #RIGHT DOUBLE QUOTATION MARK
 110                         0x95 => 0x2022, #BULLET
 111                         0x96 => 0x2013, #EN DASH
 112                         0x97 => 0x2014, #EM DASH
 113                         0x98 => 0x02DC, #SMALL TILDE
 114                         0x99 => 0x2122, #TRADE MARK SIGN
 115                         0x9A => 0x0161, #LATIN SMALL LETTER S WITH CARON
 116                         0x9B => 0x203A, #SINGLE RIGHT-POINTING ANGLE QUOTATION MARK
 117                         0x9C => 0x0153, #LATIN SMALL LIGATURE OE
 118                         0x9D => UNICODE_REPLACEMENT,
 119                         0x9E => 0x017E, #LATIN SMALL LETTER Z WITH CARON
 120                         0x9F => 0x0178, #LATIN CAPITAL LETTER Y WITH DIAERESIS
 121                         );
 122                 $pairs = array();
 123                 for( $i = 0; $i < 0x100; $i++ ) {
 124                         $unicode = isset( $cp1252[$i] ) ? $cp1252[$i] : $i;
 125                         $pairs[chr( $i )] = codepointToUtf8( $unicode );
 126                 }
 127                 return $pairs;
 128         }
 129
 130         /**
 131          * Convert from 8-bit Windows-1252 to UTF-8 if necessary.
 132          * @param string $text
 133          * @return string
 134          * @access private
 135          */
 136         function conv( $text ) {
 137                 global $wgUseLatin1;
 138                 return is_null( $text )
 139                         ? null
 140                         : ( $wgUseLatin1
 141                                 ? strtr( $text, $this->conversionTables )
 142                                 : $text );
 143         }
 144
 145         /**
 146          * Dump timestamp and message to output
 147          * @param string $message
 148          * @access private
 149          */
 150         function log( $message ) {
 151                 echo wfTimestamp( TS_DB ) . ': ' . $message . "\n";
 152                 flush();
 153         }
 154
 155         /**
 156          * Initialize the chunked-insert system.
 157          * Rows will be inserted in chunks of the given number, rather
 158          * than in a giant INSERT...SELECT query, to keep the serialized
 159          * MySQL database replication from getting hung up. This way other
 160          * things can be going on during conversion without waiting for
 161          * slaves to catch up as badly.
 162          *
 163          * @param int $chunksize Number of rows to insert at once
 164          * @param int $final Total expected number of rows / id of last row,
 165          *                   used for progress reports.
 166          * @param string $table to insert on
 167          * @param string $fname function name to report in SQL
 168          * @access private
 169          */
 170         function setChunkScale( $chunksize, $final, $table, $fname ) {
 171                 $this->chunkSize  = $chunksize;
 172                 $this->chunkFinal = $final;
 173                 $this->chunkCount = 0;
 174                 $this->chunkStartTime = wfTime();
 175                 $this->chunkOptions = array();
 176                 $this->chunkTable = $table;
 177                 $this->chunkFunction = $fname;
 178         }
 179
 180         /**
 181          * Chunked inserts: perform an insert if we've reached the chunk limit.
 182          * Prints a progress report with estimated completion time.
 183          * @param array &$chunk -- This will be emptied if an insert is done.
 184          * @param int $key A key identifier to use in progress estimation in
 185          *                 place of the number of rows inserted. Use this if
 186          *                 you provided a max key number instead of a count
 187          *                 as the final chunk number in setChunkScale()
 188          * @access private
 189          */
 190         function addChunk( &$chunk, $key = null ) {
 191                 if( count( $chunk ) >= $this->chunkSize ) {
 192                         $this->insertChunk( $chunk );
 193
 194                         $this->chunkCount += count( $chunk );
 195                         $now = wfTime();
 196                         $delta = $now - $this->chunkStartTime;
 197                         $rate = $this->chunkCount / $delta;
 198
 199                         if( is_null( $key ) ) {
 200                                 $completed = $this->chunkCount;
 201                         } else {
 202                                 $completed = $key;
 203                         }
 204                         $portion = $completed / $this->chunkFinal;
 205
 206                         $estimatedTotalTime = $delta / $portion;
 207                         $eta = $this->chunkStartTime + $estimatedTotalTime;
 208
 209                         printf( "%s: %6.2f%% done on %s; ETA %s [%d/%d] %.2f/sec\n",
 210                                 wfTimestamp( TS_DB, intval( $now ) ),
 211                                 $portion * 100.0,
 212                                 $this->chunkTable,
 213                                 wfTimestamp( TS_DB, intval( $eta ) ),
 214                                 $completed,
 215                                 $this->chunkFinal,
 216                                 $rate );
 217                         flush();
 218
 219                         $chunk = array();
 220                 }
 221         }
 222
 223         /**
 224          * Chunked inserts: perform an insert unconditionally, at the end, and log.
 225          * @param array &$chunk -- This will be emptied if an insert is done.
 226          * @access private
 227          */
 228         function lastChunk( &$chunk ) {
 229                 $n = count( $chunk );
 230                 if( $n > 0 ) {
 231                         $this->insertChunk( $chunk );
 232                 }
 233                 $this->log( "100.00% done on $this->chunkTable (last chunk $n rows)." );
 234         }
 235
 236         /**
 237          * Chunked inserts: perform an insert.
 238          * @param array &$chunk -- This will be emptied if an insert is done.
 239          * @access private
 240          */
 241         function insertChunk( &$chunk ) {
 242                 $this->dbw->insert( $this->chunkTable, $chunk, $this->chunkFunction, $this->chunkOptions );
 243         }
 244
 245
 246         function upgradePage() {
 247                 $fname = "FiveUpgrade::upgradePage";
 248                 $chunksize = 100;
 249
 250
 251                 $this->log( "Checking cur table for unique title index and applying if necessary" );
 252                 checkDupes( true );
 253
 254                 $this->log( "...converting from cur/old to page/revision/text DB structure." );
 255
 256                 extract( $this->dbw->tableNames( 'cur', 'old', 'page', 'revision', 'text' ) );
 257
 258                 $this->log( "Creating page and revision tables..." );
 259                 $this->dbw->query("CREATE TABLE $page (
 260                         page_id int(8) unsigned NOT NULL auto_increment,
 261                         page_namespace int NOT NULL,
 262                         page_title varchar(255) binary NOT NULL,
 263                         page_restrictions tinyblob NOT NULL default '',
 264                         page_counter bigint(20) unsigned NOT NULL default '0',
 265                         page_is_redirect tinyint(1) unsigned NOT NULL default '0',
 266                         page_is_new tinyint(1) unsigned NOT NULL default '0',
 267                         page_random real unsigned NOT NULL,
 268                         page_touched char(14) binary NOT NULL default '',
 269                         page_latest int(8) unsigned NOT NULL,
 270                         page_len int(8) unsigned NOT NULL,
 271
 272                         PRIMARY KEY page_id (page_id),
 273                         UNIQUE INDEX name_title (page_namespace,page_title),
 274                         INDEX (page_random),
 275                         INDEX (page_len)
 276                         ) TYPE=InnoDB", $fname );
 277                 $this->dbw->query("CREATE TABLE $revision (
 278                         rev_id int(8) unsigned NOT NULL auto_increment,
 279                         rev_page int(8) unsigned NOT NULL,
 280                         rev_comment tinyblob NOT NULL default '',
 281                         rev_user int(5) unsigned NOT NULL default '0',
 282                         rev_user_text varchar(255) binary NOT NULL default '',
 283                         rev_timestamp char(14) binary NOT NULL default '',
 284                         rev_minor_edit tinyint(1) unsigned NOT NULL default '0',
 285                         rev_deleted tinyint(1) unsigned NOT NULL default '0',
 286
 287                         PRIMARY KEY rev_page_id (rev_page, rev_id),
 288                         UNIQUE INDEX rev_id (rev_id),
 289                         INDEX rev_timestamp (rev_timestamp),
 290                         INDEX page_timestamp (rev_page,rev_timestamp),
 291                         INDEX user_timestamp (rev_user,rev_timestamp),
 292                         INDEX usertext_timestamp (rev_user_text,rev_timestamp)
 293                         ) TYPE=InnoDB", $fname );
 294
 295                 $maxold = $this->dbw->selectField( 'old', 'max(old_id)', '', $fname );
 296                 $this->log( "Last old record is {$maxold}" );
 297
 298                 global $wgLegacySchemaConversion;
 299                 if( $wgLegacySchemaConversion ) {
 300                         // Create HistoryBlobCurStub entries.
 301                         // Text will be pulled from the leftover 'cur' table at runtime.
 302                         echo "......Moving metadata from cur; using blob references to text in cur table.\n";
 303                         $cur_text = "concat('O:18:\"historyblobcurstub\":1:{s:6:\"mCurId\";i:',cur_id,';}')";
 304                         $cur_flags = "'object'";
 305                 } else {
 306                         // Copy all cur text in immediately: this may take longer but avoids
 307                         // having to keep an extra table around.
 308                         echo "......Moving text from cur.\n";
 309                         $cur_text = 'cur_text';
 310                         $cur_flags = "''";
 311                 }
 312
 313                 $maxcur = $this->dbw->selectField( 'cur', 'max(cur_id)', '', $fname );
 314                 $this->log( "Last cur entry is $maxcur" );
 315
 316                 /**
 317                  * Copy placeholder records for each page's current version into old
 318                  * Don't do any conversion here; text records are converted at runtime
 319                  * based on the flags (and may be originally binary!) while the meta
 320                  * fields will be converted in the old -> rev and cur -> page steps.
 321                  */
 322                 $this->setChunkScale( $chunksize, $maxcur, 'old', $fname );
 323                 $result = $this->dbr->query(
 324                         "SELECT cur_id, cur_namespace, cur_title, $cur_text AS text, cur_comment,
 325                         cur_user, cur_user_text, cur_timestamp, cur_minor_edit, $cur_flags AS flags
 326                         FROM $cur
 327                         ORDER BY cur_id", $fname );
 328                 $add = array();
 329                 while( $row = $this->dbr->fetchObject( $result ) ) {
 330                         $add[] = array(
 331                                 'old_namespace'  => $row->cur_namespace,
 332                                 'old_title'      => $row->cur_title,
 333                                 'old_text'       => $row->text,
 334                                 'old_comment'    => $row->cur_comment,
 335                                 'old_user'       => $row->cur_user,
 336                                 'old_user_text'  => $row->cur_user_text,
 337                                 'old_timestamp'  => $row->cur_timestamp,
 338                                 'old_minor_edit' => $row->cur_minor_edit,
 339                                 'old_flags'      => $row->flags );
 340                         $this->addChunk( $add, $row->cur_id );
 341                 }
 342                 $this->lastChunk( $add );
 343                 $this->dbr->freeResult( $result );
 344
 345                 /**
 346                  * Copy revision metadata from old into revision.
 347                  * We'll also do UTF-8 conversion of usernames and comments.
 348                  */
 349                 #$newmaxold = $this->dbw->selectField( 'old', 'max(old_id)', '', $fname );
 350                 #$this->setChunkScale( $chunksize, $newmaxold, 'revision', $fname );
 351                 $countold = $this->dbw->selectField( 'old', 'count(old_id)', '', $fname );
 352                 $this->setChunkScale( $chunksize, $countold, 'revision', $fname );
 353
 354                 $this->log( "......Setting up revision table." );
 355                 $result = $this->dbr->query(
 356                         "SELECT old_id, cur_id, old_comment, old_user, old_user_text,
 357                         old_timestamp, old_minor_edit
 358                         FROM $old,$cur WHERE old_namespace=cur_namespace AND old_title=cur_title",
 359                         $fname );
 360
 361                 $add = array();
 362                 while( $row = $this->dbr->fetchObject( $result ) ) {
 363                         $add[] = array(
 364                                 'rev_id'         =>              $row->old_id,
 365                                 'rev_page'       =>              $row->cur_id,
 366                                 'rev_comment'    => $this->conv( $row->old_comment ),
 367                                 'rev_user'       =>              $row->old_user,
 368                                 'rev_user_text'  => $this->conv( $row->old_user_text ),
 369                                 'rev_timestamp'  =>              $row->old_timestamp,
 370                                 'rev_minor_edit' =>              $row->old_minor_edit );
 371                         $this->addChunk( $add );
 372                 }
 373                 $this->lastChunk( $add );
 374                 $this->dbr->freeResult( $result );
 375
 376
 377                 /**
 378                  * Copy page metadata from cur into page.
 379                  * We'll also do UTF-8 conversion of titles.
 380                  */
 381                 $this->log( "......Setting up page table." );
 382                 $this->setChunkScale( $chunksize, $maxcur, 'page', $fname );
 383                 $result = $this->dbr->query( "
 384                         SELECT cur_id, cur_namespace, cur_title, cur_restrictions, cur_counter, cur_is_redirect, cur_is_new,
 385                                 cur_random, cur_touched, rev_id, LENGTH(cur_text) AS len
 386                         FROM $cur,$revision
 387                         WHERE cur_id=rev_page AND rev_timestamp=cur_timestamp AND rev_id > {$maxold}
 388                         ORDER BY cur_id", $fname );
 389                 $add = array();
 390                 while( $row = $this->dbr->fetchObject( $result ) ) {
 391                         $add[] = array(
 392                                 'page_id'           =>              $row->cur_id,
 393                                 'page_namespace'    =>              $row->cur_namespace,
 394                                 'page_title'        => $this->conv( $row->cur_title ),
 395                                 'page_restrictions' =>              $row->cur_restrictions,
 396                                 'page_counter'      =>              $row->cur_counter,
 397                                 'page_is_redirect'  =>              $row->cur_is_redirect,
 398                                 'page_is_new'       =>              $row->cur_is_new,
 399                                 'page_random'       =>              $row->cur_random,
 400                                 'page_touched'      =>              $this->dbw->timestamp(),
 401                                 'page_latest'       =>              $row->rev_id,
 402                                 'page_len'          =>              $row->len );
 403                         $this->addChunk( $add, $row->cur_id );
 404                 }
 405                 $this->lastChunk( $add );
 406                 $this->dbr->freeResult( $result );
 407
 408                 $this->log( "...done with cur/old -> page/revision." );
 409         }
 410
 411         function upgradeLinks() {
 412                 $fname = 'FiveUpgrade::upgradeLinks';
 413                 $chunksize = 200;
 414                 extract( $this->dbw->tableNames( 'links', 'brokenlinks', 'pagelinks', 'page' ) );
 415
 416                 $this->log( 'Creating pagelinks table...' );
 417                 $this->dbw->query( "
 418 CREATE TABLE $pagelinks (
 419   -- Key to the page_id of the page containing the link.
 420   pl_from int(8) unsigned NOT NULL default '0',
 421
 422   -- Key to page_namespace/page_title of the target page.
 423   -- The target page may or may not exist, and due to renames
 424   -- and deletions may refer to different page records as time
 425   -- goes by.
 426   pl_namespace int NOT NULL default '0',
 427   pl_title varchar(255) binary NOT NULL default '',
 428
 429   UNIQUE KEY pl_from(pl_from,pl_namespace,pl_title),
 430   KEY (pl_namespace,pl_title)
 431
 432 ) TYPE=InnoDB" );
 433
 434                 $this->log( 'Importing live links -> pagelinks' );
 435                 $nlinks = $this->dbw->selectField( 'links', 'count(*)', '', $fname );
 436                 if( $nlinks ) {
 437                         $this->setChunkScale( $chunksize, $nlinks, 'pagelinks', $fname );
 438                         $result = $this->dbr->query( "
 439                           SELECT l_from,page_namespace,page_title
 440                                 FROM $links, $page
 441                                 WHERE l_to=page_id", $fname );
 442                         $add = array();
 443                         while( $row = $this->dbr->fetchObject( $result ) ) {
 444                                 $add[] = array(
 445                                         'pl_from'      => $row->l_from,
 446                                         'pl_namespace' => $row->page_namespace,
 447                                         'pl_title'     => $row->page_title );
 448                                 $this->addChunk( $add );
 449                         }
 450                         $this->lastChunk( $add );
 451                 } else {
 452                         $this->log( 'no links!' );
 453                 }
 454
 455                 $this->log( 'Importing brokenlinks -> pagelinks' );
 456                 $nbrokenlinks = $this->dbw->selectField( 'brokenlinks', 'count(*)', '', $fname );
 457                 if( $nbrokenlinks ) {
 458                         $this->setChunkScale( $chunksize, $nbrokenlinks, 'pagelinks', $fname );
 459                         $this->chunkOptions = array( 'IGNORE' );
 460                         $result = $this->dbr->query(
 461                                 "SELECT bl_from, bl_to FROM $brokenlinks",
 462                                 $fname );
 463                         $add = array();
 464                         while( $row = $this->dbr->fetchObject( $result ) ) {
 465                                 $pagename = $this->conv( $row->bl_to );
 466                                 $title = Title::newFromText( $pagename );
 467                                 if( is_null( $title ) ) {
 468                                         $this->log( "** invalid brokenlink: $row->bl_from -> '$pagename' (converted from '$row->bl_to')" );
 469                                 } else {
 470                                         $add[] = array(
 471                                                 'pl_from'      => $row->bl_from,
 472                                                 'pl_namespace' => $title->getNamespace(),
 473                                                 'pl_title'     => $title->getDBkey() );
 474                                         $this->addChunk( $add );
 475                                 }
 476                         }
 477                         $this->lastChunk( $add );
 478                 } else {
 479                         $this->log( 'no brokenlinks!' );
 480                 }
 481
 482                 $this->log( 'Done with links.' );
 483         }
 484
 485         function upgradeUser() {
 486                 $fname = 'FiveUpgrade::upgradeUser';
 487                 $chunksize = 100;
 488                 $preauth = 0;
 489
 490                 // Apply unique index, if necessary:
 491                 $duper = new UserDupes( $this->dbw );
 492                 if( $duper->hasUniqueIndex() ) {
 493                         $this->log( "Already have unique user_name index." );
 494                 } else {
 495                         $this->log( "Clearing user duplicates..." );
 496                         if( !$duper->clearDupes() ) {
 497                                 $this->log( "WARNING: Duplicate user accounts, may explode!" );
 498                         }
 499                 }
 500
 501                 /** Convert encoding on options, etc */
 502                 extract( $this->dbw->tableNames( 'user', 'user_temp', 'user_old' ) );
 503
 504                 $this->log( 'Migrating user table to user_temp...' );
 505                 $this->dbw->query( "CREATE TABLE $user_temp (
 506   user_id int(5) unsigned NOT NULL auto_increment,
 507   user_name varchar(255) binary NOT NULL default '',
 508   user_real_name varchar(255) binary NOT NULL default '',
 509   user_password tinyblob NOT NULL default '',
 510   user_newpassword tinyblob NOT NULL default '',
 511   user_email tinytext NOT NULL default '',
 512   user_options blob NOT NULL default '',
 513   user_touched char(14) binary NOT NULL default '',
 514   user_token char(32) binary NOT NULL default '',
 515   user_email_authenticated CHAR(14) BINARY,
 516   user_email_token CHAR(32) BINARY,
 517   user_email_token_expires CHAR(14) BINARY,
 518
 519   PRIMARY KEY user_id (user_id),
 520   UNIQUE INDEX user_name (user_name),
 521   INDEX (user_email_token)
 522
 523 ) TYPE=InnoDB", $fname );
 524
 525                 // Fix encoding for Latin-1 upgrades, and add some fields.
 526                 $numusers = $this->dbw->selectField( 'user', 'count(*)', '', $fname );
 527                 $this->setChunkScale( $chunksize, $numusers, 'user_temp', $fname );
 528                 $result = $this->dbr->select( 'user',
 529                         array(
 530                                 'user_id',
 531                                 'user_name',
 532                                 'user_real_name',
 533                                 'user_password',
 534                                 'user_newpassword',
 535                                 'user_email',
 536                                 'user_options',
 537                                 'user_touched',
 538                                 'user_token' ),
 539                         '',
 540                         $fname );
 541                 $add = array();
 542                 while( $row = $this->dbr->fetchObject( $result ) ) {
 543                         $now = $this->dbw->timestamp();
 544                         $add[] = array(
 545                                 'user_id'                  =>              $row->user_id,
 546                                 'user_name'                => $this->conv( $row->user_name ),
 547                                 'user_real_name'           => $this->conv( $row->user_real_name ),
 548                                 'user_password'            =>              $row->user_password,
 549                                 'user_newpassword'         =>              $row->user_newpassword,
 550                                 'user_email'               => $this->conv( $row->user_email ),
 551                                 'user_options'             => $this->conv( $row->user_options ),
 552                                 'user_touched'             =>              $now,
 553                                 'user_token'               =>              $row->user_token,
 554                                 'user_email_authenticated' =>              $preauth ? $now : null,
 555                                 'user_email_token'         =>              null,
 556                                 'user_email_token_expires' =>              null );
 557                         $this->addChunk( $add );
 558                 }
 559                 $this->lastChunk( $add );
 560                 $this->dbr->freeResult( $result );
 561                 $this->cleanupSwaps[] = 'user';
 562         }
 563
 564         function upgradeImage() {
 565                 $fname = 'FiveUpgrade::upgradeImage';
 566                 $chunksize = 100;
 567
 568                 extract( $this->dbw->tableNames( 'image', 'image_temp', 'image_old' ) );
 569                 $this->log( 'Creating temporary image_temp to merge into...' );
 570                 $this->dbw->query( <<<END
 571 CREATE TABLE $image_temp (
 572   img_name varchar(255) binary NOT NULL default '',
 573   img_size int(8) unsigned NOT NULL default '0',
 574   img_width int(5)  NOT NULL default '0',
 575   img_height int(5)  NOT NULL default '0',
 576   img_metadata mediumblob NOT NULL,
 577   img_bits int(3)  NOT NULL default '0',
 578   img_media_type ENUM("UNKNOWN", "BITMAP", "DRAWING", "AUDIO", "VIDEO", "MULTIMEDIA", "OFFICE", "TEXT", "EXECUTABLE", "ARCHIVE") default NULL,
 579   img_major_mime ENUM("unknown", "application", "audio", "image", "text", "video", "message", "model", "multipart") NOT NULL default "unknown",
 580   img_minor_mime varchar(32) NOT NULL default "unknown",
 581   img_description tinyblob NOT NULL default '',
 582   img_user int(5) unsigned NOT NULL default '0',
 583   img_user_text varchar(255) binary NOT NULL default '',
 584   img_timestamp char(14) binary NOT NULL default '',
 585
 586   PRIMARY KEY img_name (img_name),
 587   INDEX img_size (img_size),
 588   INDEX img_timestamp (img_timestamp)
 589 ) TYPE=InnoDB
 590 END
 591                 , $fname);
 592
 593                 $numimages = $this->dbw->selectField( 'image', 'count(*)', '', $fname );
 594                 $result = $this->dbr->select( 'image',
 595                         array(
 596                                 'img_name',
 597                                 'img_size',
 598                                 'img_description',
 599                                 'img_user',
 600                                 'img_user_text',
 601                                 'img_timestamp' ),
 602                         '',
 603                         $fname );
 604                 $add = array();
 605                 $this->setChunkScale( $chunksize, $numimages, 'image_temp', $fname );
 606                 while( $row = $this->dbr->fetchObject( $result ) ) {
 607                         // Fill in the new image info fields
 608                         $info = $this->imageInfo( $row->img_name );
 609
 610                         // Update and convert encoding
 611                         $add[] = array(
 612                                 'img_name'        => $this->conv( $row->img_name ),
 613                                 'img_size'        =>              $row->img_size,
 614                                 'img_width'       =>              $info['width'],
 615                                 'img_height'      =>              $info['height'],
 616                                 'img_metadata'    =>              "", // loaded on-demand
 617                                 'img_bits'        =>              $info['bits'],
 618                                 'img_media_type'  =>              $info['media'],
 619                                 'img_major_mime'  =>              $info['major'],
 620                                 'img_minor_mime'  =>              $info['minor'],
 621                                 'img_description' => $this->conv( $row->img_description ),
 622                                 'img_user'        =>              $row->img_user,
 623                                 'img_user_text'   => $this->conv( $row->img_user_text ),
 624                                 'img_timestamp'   =>              $row->img_timestamp );
 625
 626                         // If doing UTF8 conversion the file must be renamed
 627                         $this->renameFile( $row->img_name, 'wfImageDir' );
 628                 }
 629                 $this->lastChunk( $add );
 630
 631                 $this->log( 'done with image table.' );
 632                 $this->cleanupSwaps[] = 'image';
 633         }
 634
 635         function imageInfo( $name, $subdirCallback='wfImageDir', $basename = null ) {
 636                 if( is_null( $basename ) ) $basename = $name;
 637                 $dir = call_user_func( $subdirCallback, $basename );
 638                 $filename = $dir . '/' . $name;
 639                 $info = array(
 640                         'width'  => 0,
 641                         'height' => 0,
 642                         'bits'   => 0,
 643                         'media'  => '',
 644                         'major'  => '',
 645                         'minor'  => '' );
 646
 647                 $magic =& wfGetMimeMagic();
 648                 $mime = $magic->guessMimeType( $filename, true );
 649                 list( $info['major'], $info['minor'] ) = explode( '/', $mime );
 650
 651                 $info['media'] = $magic->getMediaType( $filename, $mime );
 652
 653                 # Height and width
 654                 $gis = false;
 655                 if( $mime == 'image/svg' ) {
 656                         $gis = wfGetSVGsize( $this->imagePath );
 657                 } elseif( $magic->isPHPImageType( $mime ) ) {
 658                         $gis = getimagesize( $filename );
 659                 } else {
 660                         $this->log( "Surprising mime type: $mime" );
 661                 }
 662                 if( $gis ) {
 663                         $info['width' ] = $gis[0];
 664                         $info['height'] = $gis[1];
 665                 }
 666                 if( isset( $gis['bits'] ) ) {
 667                         $info['bits'] = $gis['bits'];
 668                 }
 669
 670                 return $info;
 671         }
 672
 673
 674         /**
 675          * Truncate a table.
 676          * @param string $table The table name to be truncated
 677          */
 678         function clearTable( $table ) {
 679                 print "Clearing $table...\n";
 680                 $tableName = $this->db->tableName( $table );
 681                 $this->db->query( 'TRUNCATE $tableName' );
 682         }
 683
 684         /**
 685          * Rename a given image or archived image file to the converted filename,
 686          * leaving a symlink for URL compatibility.
 687          *
 688          * @param string $oldname pre-conversion filename
 689          * @param string $basename pre-conversion base filename for dir hashing, if an archive
 690          * @access private
 691          */
 692         function renameFile( $oldname, $subdirCallback='wfImageDir', $basename=null ) {
 693                 $newname = $this->conv( $oldname );
 694                 if( $newname == $oldname ) {
 695                         // No need to rename; another field triggered this row.
 696                         return;
 697                 }
 698
 699                 if( is_null( $basename ) ) $basename = $oldname;
 700                 $ubasename = $this->conv( $basename );
 701                 $oldpath = call_user_func( $subdirCallback, $basename ) . '/' . $oldname;
 702                 $newpath = call_user_func( $subdirCallback, $ubasename ) . '/' . $newname;
 703
 704                 $this->log( "$oldpath -> $newpath" );
 705                 if( rename( $oldpath, $newpath ) ) {
 706                         $relpath = $this->relativize( $newpath, dirname( $oldpath ) );
 707                         if( !symlink( $relpath, $oldpath ) ) {
 708                                 $this->log( "... symlink failed!" );
 709                         }
 710                 } else {
 711                         $this->log( "... rename failed!" );
 712                 }
 713         }
 714
 715         /**
 716          * Generate a relative path name to the given file.
 717          * Assumes Unix-style paths, separators, and semantics.
 718          *
 719          * @param string $path Absolute destination path including target filename
 720          * @param string $from Absolute source path, directory only
 721          * @return string
 722          * @access private
 723          * @static
 724          */
 725         function relativize( $path, $from ) {
 726                 $pieces  = explode( '/', dirname( $path ) );
 727                 $against = explode( '/', $from );
 728
 729                 // Trim off common prefix
 730                 while( count( $pieces ) && count( $against )
 731                         && $pieces[0] == $against[0] ) {
 732                         array_shift( $pieces );
 733                         array_shift( $against );
 734                 }
 735
 736                 // relative dots to bump us to the parent
 737                 while( count( $against ) ) {
 738                         array_unshift( $pieces, '..' );
 739                         array_shift( $against );
 740                 }
 741
 742                 array_push( $pieces, basename( $path ) );
 743
 744                 return implode( '/', $pieces );
 745         }
 746
 747         function upgradeOldImage() {
 748                 $fname = 'FiveUpgrade::upgradeOldImage';
 749                 $chunksize = 100;
 750
 751                 extract( $this->dbw->tableNames( 'oldimage', 'oldimage_temp', 'oldimage_old' ) );
 752                 $this->log( 'Creating temporary oldimage_temp to merge into...' );
 753                 $this->dbw->query( <<<END
 754 CREATE TABLE $oldimage_temp (
 755   -- Base filename: key to image.img_name
 756   oi_name varchar(255) binary NOT NULL default '',
 757
 758   -- Filename of the archived file.
 759   -- This is generally a timestamp and '!' prepended to the base name.
 760   oi_archive_name varchar(255) binary NOT NULL default '',
 761
 762   -- Other fields as in image...
 763   oi_size int(8) unsigned NOT NULL default 0,
 764   oi_width int(5) NOT NULL default 0,
 765   oi_height int(5) NOT NULL default 0,
 766   oi_bits int(3) NOT NULL default 0,
 767   oi_description tinyblob NOT NULL default '',
 768   oi_user int(5) unsigned NOT NULL default '0',
 769   oi_user_text varchar(255) binary NOT NULL default '',
 770   oi_timestamp char(14) binary NOT NULL default '',
 771
 772   INDEX oi_name (oi_name(10))
 773
 774 ) TYPE=InnoDB;
 775 END
 776                 , $fname);
 777
 778                 $numimages = $this->dbw->selectField( 'oldimage', 'count(*)', '', $fname );
 779                 $result = $this->dbr->select( 'oldimage',
 780                         array(
 781                                 'oi_name',
 782                                 'oi_archive_name',
 783                                 'oi_size',
 784                                 'oi_description',
 785                                 'oi_user',
 786                                 'oi_user_text',
 787                                 'oi_timestamp' ),
 788                         '',
 789                         $fname );
 790                 $add = array();
 791                 $this->setChunkScale( $chunksize, $numimages, 'oldimage_temp', $fname );
 792                 while( $row = $this->dbr->fetchObject( $result ) ) {
 793                         // Fill in the new image info fields
 794                         $info = $this->imageInfo( $row->oi_archive_name, 'wfImageArchiveDir', $row->oi_name );
 795
 796                         // Update and convert encoding
 797                         $add[] = array(
 798                                 'oi_name'         => $this->conv( $row->oi_name ),
 799                                 'oi_archive_name' => $this->conv( $row->oi_archive_name ),
 800                                 'oi_size'         =>              $row->oi_size,
 801                                 'oi_width'        =>              $info['width'],
 802                                 'oi_height'       =>              $info['height'],
 803                                 'oi_bits'         =>              $info['bits'],
 804                                 'oi_description'  => $this->conv( $row->oi_description ),
 805                                 'oi_user'         =>              $row->oi_user,
 806                                 'oi_user_text'    => $this->conv( $row->oi_user_text ),
 807                                 'oi_timestamp'    =>              $row->oi_timestamp );
 808
 809                         // If doing UTF8 conversion the file must be renamed
 810                         $this->renameFile( $row->oi_archive_name, 'wfImageArchiveDir', $row->oi_name );
 811                 }
 812                 $this->lastChunk( $add );
 813
 814                 $this->log( 'done with oldimage table.' );
 815                 $this->cleanupSwaps[] = 'oldimage';
 816         }
 817
 818
 819         function upgradeWatchlist() {
 820                 $fname = 'FiveUpgrade::upgradeWatchlist';
 821                 $chunksize = 100;
 822
 823                 extract( $this->dbw->tableNames( 'watchlist', 'watchlist_temp' ) );
 824
 825                 $this->log( 'Migrating watchlist table to watchlist_temp...' );
 826                 $this->dbw->query(
 827 "CREATE TABLE $watchlist_temp (
 828   -- Key to user_id
 829   wl_user int(5) unsigned NOT NULL,
 830
 831   -- Key to page_namespace/page_title
 832   -- Note that users may watch patches which do not exist yet,
 833   -- or existed in the past but have been deleted.
 834   wl_namespace int NOT NULL default '0',
 835   wl_title varchar(255) binary NOT NULL default '',
 836
 837   -- Timestamp when user was last sent a notification e-mail;
 838   -- cleared when the user visits the page.
 839   -- FIXME: add proper null support etc
 840   wl_notificationtimestamp varchar(14) binary NOT NULL default '0',
 841
 842   UNIQUE KEY (wl_user, wl_namespace, wl_title),
 843   KEY namespace_title (wl_namespace,wl_title)
 844
 845 ) TYPE=InnoDB;", $fname );
 846
 847                 // Fix encoding for Latin-1 upgrades, add some fields,
 848                 // and double article to article+talk pairs
 849                 $numwatched = $this->dbw->selectField( 'watchlist', 'count(*)', '', $fname );
 850
 851                 $this->setChunkScale( $chunksize, $numwatched * 2, 'watchlist_temp', $fname );
 852                 $result = $this->dbr->select( 'watchlist',
 853                         array(
 854                                 'wl_user',
 855                                 'wl_namespace',
 856                                 'wl_title' ),
 857                         '',
 858                         $fname );
 859
 860                 $add = array();
 861                 while( $row = $this->dbr->fetchObject( $result ) ) {
 862                         $now = $this->dbw->timestamp();
 863                         $add[] = array(
 864                                 'wl_user'      =>                        $row->wl_user,
 865                                 'wl_namespace' => Namespace::getSubject( $row->wl_namespace ),
 866                                 'wl_title'     =>           $this->conv( $row->wl_title ),
 867                                 'wl_notificationtimestamp' =>            '0' );
 868                         $this->addChunk( $add );
 869
 870                         $add[] = array(
 871                                 'wl_user'      =>                        $row->wl_user,
 872                                 'wl_namespace' =>    Namespace::getTalk( $row->wl_namespace ),
 873                                 'wl_title'     =>           $this->conv( $row->wl_title ),
 874                                 'wl_notificationtimestamp' =>            '0' );
 875                         $this->addChunk( $add );
 876                 }
 877                 $this->lastChunk( $add );
 878                 $this->dbr->freeResult( $result );
 879
 880                 $this->log( 'Done converting watchlist.' );
 881                 $this->cleanupSwaps[] = 'watchlist';
 882         }
 883
 884         function upgradeLogging() {
 885                 global $wgUseLatin1;
 886                 if( !$wgUseLatin1 ) {
 887                         $this->log( 'Not latin1; no change to logging table.' );
 888                         return;
 889                 }
 890
 891                 $fname = 'FiveUpgrade::upgradeLogging';
 892                 $chunksize = 100;
 893
 894                 extract( $this->dbw->tableNames( 'logging', 'logging_temp' ) );
 895
 896                 $this->log( 'Migrating logging table to logging_temp...' );
 897                 $this->dbw->query(
 898 "CREATE TABLE $logging_temp (
 899   -- Symbolic keys for the general log type and the action type
 900   -- within the log. The output format will be controlled by the
 901   -- action field, but only the type controls categorization.
 902   log_type char(10) NOT NULL default '',
 903   log_action char(10) NOT NULL default '',
 904
 905   -- Timestamp. Duh.
 906   log_timestamp char(14) NOT NULL default '19700101000000',
 907
 908   -- The user who performed this action; key to user_id
 909   log_user int unsigned NOT NULL default 0,
 910
 911   -- Key to the page affected. Where a user is the target,
 912   -- this will point to the user page.
 913   log_namespace int NOT NULL default 0,
 914   log_title varchar(255) binary NOT NULL default '',
 915
 916   -- Freeform text. Interpreted as edit history comments.
 917   log_comment varchar(255) NOT NULL default '',
 918
 919   -- LF separated list of miscellaneous parameters
 920   log_params blob NOT NULL default '',
 921
 922   KEY type_time (log_type, log_timestamp),
 923   KEY user_time (log_user, log_timestamp),
 924   KEY page_time (log_namespace, log_title, log_timestamp)
 925
 926 ) TYPE=InnoDB", $fname );
 927
 928                 $numlogged = $this->dbw->selectField( 'logging', 'count(*)', '', $fname );
 929                 $this->setChunkScale( $chunksize, $numlogged, 'logging_temp', $fname );
 930
 931                 $result = $this->dbr->select( 'logging',
 932                         array(
 933                                 'log_type',
 934                                 'log_action',
 935                                 'log_timestamp',
 936                                 'log_user',
 937                                 'log_namespace',
 938                                 'log_title',
 939                                 'log_comment',
 940                                 'log_params' ),
 941                         '',
 942                         $fname );
 943
 944                 $add = array();
 945                 while( $row = $this->dbr->fetchObject( $result ) ) {
 946                         $now = $this->dbw->timestamp();
 947                         $add[] = array(
 948                                 'log_type'      =>              $row->log_type,
 949                                 'log_action'    =>              $row->log_type,
 950                                 'log_timestamp' =>              $row->log_timestamp,
 951                                 'log_user'      =>              $row->log_user,
 952                                 'log_namespace' =>              $row->log_namespace,
 953                                 'log_title'     => $this->conv( $row->log_title ),
 954                                 'log_comment'   => $this->conv( $row->log_comment ),
 955                                 'log_params'    => $this->conv( $row->log_params ) );
 956                         $this->addChunk( $add );
 957                 }
 958                 $this->lastChunk( $add );
 959                 $this->dbr->freeResult( $result );
 960
 961                 $this->log( 'Done converting logging.' );
 962                 $this->cleanupSwaps[] = 'logging';
 963         }
 964
 965         /**
 966          * Rename all our temporary tables into final place.
 967          * We've left things in place so a read-only wiki can continue running
 968          * on the old code during all this.
 969          */
 970         function upgradeCleanup() {
 971                 $this->renameTable( 'old', 'text' );
 972
 973                 foreach( $this->cleanupSwaps as $table ) {
 974                         $this->swap( $table );
 975                 }
 976         }
 977
 978         function renameTable( $from, $to ) {
 979                 $this->log( 'Renaming $from to $to...' );
 980
 981                 $fromtable = $this->dbw->tableName( $from );
 982                 $totable   = $this->dbw->tableName( $to );
 983                 $this->dbw->query( "ALTER TABLE $fromtable RENAME TO $totable" );
 984         }
 985
 986         function swap( $base ) {
 987                 $this->renameTable( $base, "{$base}_old" );
 988                 $this->renameTable( "{$base}_temp", $base );
 989         }
 990
 991 }
 992
 993 ?>