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                 if( $this->doing( 'archive' ) )
  57                         $this->upgradeArchive();
  58
  59                 if( $this->doing( 'cleanup' ) )
  60                         $this->upgradeCleanup();
  61         }
  62
  63
  64         /**
  65          * Open a second connection to the master server, with buffering off.
  66          * This will let us stream large datasets in and write in chunks on the
  67          * other end.
  68          * @return Database
  69          * @access private
  70          */
  71         function &newConnection() {
  72                 global $wgDBadminuser, $wgDBadminpassword;
  73                 global $wgDBserver, $wgDBname;
  74                 $db =& new Database( $wgDBserver, $wgDBadminuser, $wgDBadminpassword, $wgDBname );
  75                 return $db;
  76         }
  77
  78         /**
  79          * Prepare a conversion array for converting Windows Code Page 1252 to
  80          * UTF-8. This should provide proper conversion of text that was miscoded
  81          * as Windows-1252 by naughty user-agents, and doesn't rely on an outside
  82          * iconv library.
  83          *
  84          * @return array
  85          * @access private
  86          */
  87         function prepareWindows1252() {
  88                 # Mappings from:
  89                 # http://www.unicode.org/Public/MAPPINGS/VENDORS/MICSFT/WINDOWS/CP1252.TXT
  90                 static $cp1252 = array(
  91                         0x80 => 0x20AC, #EURO SIGN
  92                         0x81 => UNICODE_REPLACEMENT,
  93                         0x82 => 0x201A, #SINGLE LOW-9 QUOTATION MARK
  94                         0x83 => 0x0192, #LATIN SMALL LETTER F WITH HOOK
  95                         0x84 => 0x201E, #DOUBLE LOW-9 QUOTATION MARK
  96                         0x85 => 0x2026, #HORIZONTAL ELLIPSIS
  97                         0x86 => 0x2020, #DAGGER
  98                         0x87 => 0x2021, #DOUBLE DAGGER
  99                         0x88 => 0x02C6, #MODIFIER LETTER CIRCUMFLEX ACCENT
 100                         0x89 => 0x2030, #PER MILLE SIGN
 101                         0x8A => 0x0160, #LATIN CAPITAL LETTER S WITH CARON
 102                         0x8B => 0x2039, #SINGLE LEFT-POINTING ANGLE QUOTATION MARK
 103                         0x8C => 0x0152, #LATIN CAPITAL LIGATURE OE
 104                         0x8D => UNICODE_REPLACEMENT,
 105                         0x8E => 0x017D, #LATIN CAPITAL LETTER Z WITH CARON
 106                         0x8F => UNICODE_REPLACEMENT,
 107                         0x90 => UNICODE_REPLACEMENT,
 108                         0x91 => 0x2018, #LEFT SINGLE QUOTATION MARK
 109                         0x92 => 0x2019, #RIGHT SINGLE QUOTATION MARK
 110                         0x93 => 0x201C, #LEFT DOUBLE QUOTATION MARK
 111                         0x94 => 0x201D, #RIGHT DOUBLE QUOTATION MARK
 112                         0x95 => 0x2022, #BULLET
 113                         0x96 => 0x2013, #EN DASH
 114                         0x97 => 0x2014, #EM DASH
 115                         0x98 => 0x02DC, #SMALL TILDE
 116                         0x99 => 0x2122, #TRADE MARK SIGN
 117                         0x9A => 0x0161, #LATIN SMALL LETTER S WITH CARON
 118                         0x9B => 0x203A, #SINGLE RIGHT-POINTING ANGLE QUOTATION MARK
 119                         0x9C => 0x0153, #LATIN SMALL LIGATURE OE
 120                         0x9D => UNICODE_REPLACEMENT,
 121                         0x9E => 0x017E, #LATIN SMALL LETTER Z WITH CARON
 122                         0x9F => 0x0178, #LATIN CAPITAL LETTER Y WITH DIAERESIS
 123                         );
 124                 $pairs = array();
 125                 for( $i = 0; $i < 0x100; $i++ ) {
 126                         $unicode = isset( $cp1252[$i] ) ? $cp1252[$i] : $i;
 127                         $pairs[chr( $i )] = codepointToUtf8( $unicode );
 128                 }
 129                 return $pairs;
 130         }
 131
 132         /**
 133          * Convert from 8-bit Windows-1252 to UTF-8 if necessary.
 134          * @param string $text
 135          * @return string
 136          * @access private
 137          */
 138         function conv( $text ) {
 139                 global $wgUseLatin1;
 140                 return is_null( $text )
 141                         ? null
 142                         : ( $wgUseLatin1
 143                                 ? strtr( $text, $this->conversionTables )
 144                                 : $text );
 145         }
 146
 147         /**
 148          * Dump timestamp and message to output
 149          * @param string $message
 150          * @access private
 151          */
 152         function log( $message ) {
 153                 echo wfTimestamp( TS_DB ) . ': ' . $message . "\n";
 154                 flush();
 155         }
 156
 157         /**
 158          * Initialize the chunked-insert system.
 159          * Rows will be inserted in chunks of the given number, rather
 160          * than in a giant INSERT...SELECT query, to keep the serialized
 161          * MySQL database replication from getting hung up. This way other
 162          * things can be going on during conversion without waiting for
 163          * slaves to catch up as badly.
 164          *
 165          * @param int $chunksize Number of rows to insert at once
 166          * @param int $final Total expected number of rows / id of last row,
 167          *                   used for progress reports.
 168          * @param string $table to insert on
 169          * @param string $fname function name to report in SQL
 170          * @access private
 171          */
 172         function setChunkScale( $chunksize, $final, $table, $fname ) {
 173                 $this->chunkSize  = $chunksize;
 174                 $this->chunkFinal = $final;
 175                 $this->chunkCount = 0;
 176                 $this->chunkStartTime = wfTime();
 177                 $this->chunkOptions = array();
 178                 $this->chunkTable = $table;
 179                 $this->chunkFunction = $fname;
 180         }
 181
 182         /**
 183          * Chunked inserts: perform an insert if we've reached the chunk limit.
 184          * Prints a progress report with estimated completion time.
 185          * @param array &$chunk -- This will be emptied if an insert is done.
 186          * @param int $key A key identifier to use in progress estimation in
 187          *                 place of the number of rows inserted. Use this if
 188          *                 you provided a max key number instead of a count
 189          *                 as the final chunk number in setChunkScale()
 190          * @access private
 191          */
 192         function addChunk( &$chunk, $key = null ) {
 193                 if( count( $chunk ) >= $this->chunkSize ) {
 194                         $this->insertChunk( $chunk );
 195
 196                         $this->chunkCount += count( $chunk );
 197                         $now = wfTime();
 198                         $delta = $now - $this->chunkStartTime;
 199                         $rate = $this->chunkCount / $delta;
 200
 201                         if( is_null( $key ) ) {
 202                                 $completed = $this->chunkCount;
 203                         } else {
 204                                 $completed = $key;
 205                         }
 206                         $portion = $completed / $this->chunkFinal;
 207
 208                         $estimatedTotalTime = $delta / $portion;
 209                         $eta = $this->chunkStartTime + $estimatedTotalTime;
 210
 211                         printf( "%s: %6.2f%% done on %s; ETA %s [%d/%d] %.2f/sec\n",
 212                                 wfTimestamp( TS_DB, intval( $now ) ),
 213                                 $portion * 100.0,
 214                                 $this->chunkTable,
 215                                 wfTimestamp( TS_DB, intval( $eta ) ),
 216                                 $completed,
 217                                 $this->chunkFinal,
 218                                 $rate );
 219                         flush();
 220
 221                         $chunk = array();
 222                 }
 223         }
 224
 225         /**
 226          * Chunked inserts: perform an insert unconditionally, at the end, and log.
 227          * @param array &$chunk -- This will be emptied if an insert is done.
 228          * @access private
 229          */
 230         function lastChunk( &$chunk ) {
 231                 $n = count( $chunk );
 232                 if( $n > 0 ) {
 233                         $this->insertChunk( $chunk );
 234                 }
 235                 $this->log( "100.00% done on $this->chunkTable (last chunk $n rows)." );
 236         }
 237
 238         /**
 239          * Chunked inserts: perform an insert.
 240          * @param array &$chunk -- This will be emptied if an insert is done.
 241          * @access private
 242          */
 243         function insertChunk( &$chunk ) {
 244                 $this->dbw->insert( $this->chunkTable, $chunk, $this->chunkFunction, $this->chunkOptions );
 245         }
 246
 247
 248         function upgradePage() {
 249                 $fname = "FiveUpgrade::upgradePage";
 250                 $chunksize = 100;
 251
 252
 253                 $this->log( "Checking cur table for unique title index and applying if necessary" );
 254                 checkDupes( true );
 255
 256                 $this->log( "...converting from cur/old to page/revision/text DB structure." );
 257
 258                 extract( $this->dbw->tableNames( 'cur', 'old', 'page', 'revision', 'text' ) );
 259
 260                 $this->log( "Creating page and revision tables..." );
 261                 $this->dbw->query("CREATE TABLE $page (
 262                         page_id int(8) unsigned NOT NULL auto_increment,
 263                         page_namespace int NOT NULL,
 264                         page_title varchar(255) binary NOT NULL,
 265                         page_restrictions tinyblob NOT NULL default '',
 266                         page_counter bigint(20) unsigned NOT NULL default '0',
 267                         page_is_redirect tinyint(1) unsigned NOT NULL default '0',
 268                         page_is_new tinyint(1) unsigned NOT NULL default '0',
 269                         page_random real unsigned NOT NULL,
 270                         page_touched char(14) binary NOT NULL default '',
 271                         page_latest int(8) unsigned NOT NULL,
 272                         page_len int(8) unsigned NOT NULL,
 273
 274                         PRIMARY KEY page_id (page_id),
 275                         UNIQUE INDEX name_title (page_namespace,page_title),
 276                         INDEX (page_random),
 277                         INDEX (page_len)
 278                         ) TYPE=InnoDB", $fname );
 279                 $this->dbw->query("CREATE TABLE $revision (
 280                         rev_id int(8) unsigned NOT NULL auto_increment,
 281                         rev_page int(8) unsigned NOT NULL,
 282                         rev_comment tinyblob NOT NULL default '',
 283                         rev_user int(5) unsigned NOT NULL default '0',
 284                         rev_user_text varchar(255) binary NOT NULL default '',
 285                         rev_timestamp char(14) binary NOT NULL default '',
 286                         rev_minor_edit tinyint(1) unsigned NOT NULL default '0',
 287                         rev_deleted tinyint(1) unsigned NOT NULL default '0',
 288
 289                         PRIMARY KEY rev_page_id (rev_page, rev_id),
 290                         UNIQUE INDEX rev_id (rev_id),
 291                         INDEX rev_timestamp (rev_timestamp),
 292                         INDEX page_timestamp (rev_page,rev_timestamp),
 293                         INDEX user_timestamp (rev_user,rev_timestamp),
 294                         INDEX usertext_timestamp (rev_user_text,rev_timestamp)
 295                         ) TYPE=InnoDB", $fname );
 296
 297                 $maxold = $this->dbw->selectField( 'old', 'max(old_id)', '', $fname );
 298                 $this->log( "Last old record is {$maxold}" );
 299
 300                 global $wgLegacySchemaConversion;
 301                 if( $wgLegacySchemaConversion ) {
 302                         // Create HistoryBlobCurStub entries.
 303                         // Text will be pulled from the leftover 'cur' table at runtime.
 304                         echo "......Moving metadata from cur; using blob references to text in cur table.\n";
 305                         $cur_text = "concat('O:18:\"historyblobcurstub\":1:{s:6:\"mCurId\";i:',cur_id,';}')";
 306                         $cur_flags = "'object'";
 307                 } else {
 308                         // Copy all cur text in immediately: this may take longer but avoids
 309                         // having to keep an extra table around.
 310                         echo "......Moving text from cur.\n";
 311                         $cur_text = 'cur_text';
 312                         $cur_flags = "''";
 313                 }
 314
 315                 $maxcur = $this->dbw->selectField( 'cur', 'max(cur_id)', '', $fname );
 316                 $this->log( "Last cur entry is $maxcur" );
 317
 318                 /**
 319                  * Copy placeholder records for each page's current version into old
 320                  * Don't do any conversion here; text records are converted at runtime
 321                  * based on the flags (and may be originally binary!) while the meta
 322                  * fields will be converted in the old -> rev and cur -> page steps.
 323                  */
 324                 $this->setChunkScale( $chunksize, $maxcur, 'old', $fname );
 325                 $result = $this->dbr->query(
 326                         "SELECT cur_id, cur_namespace, cur_title, $cur_text AS text, cur_comment,
 327                         cur_user, cur_user_text, cur_timestamp, cur_minor_edit, $cur_flags AS flags
 328                         FROM $cur
 329                         ORDER BY cur_id", $fname );
 330                 $add = array();
 331                 while( $row = $this->dbr->fetchObject( $result ) ) {
 332                         $add[] = array(
 333                                 'old_namespace'  => $row->cur_namespace,
 334                                 'old_title'      => $row->cur_title,
 335                                 'old_text'       => $row->text,
 336                                 'old_comment'    => $row->cur_comment,
 337                                 'old_user'       => $row->cur_user,
 338                                 'old_user_text'  => $row->cur_user_text,
 339                                 'old_timestamp'  => $row->cur_timestamp,
 340                                 'old_minor_edit' => $row->cur_minor_edit,
 341                                 'old_flags'      => $row->flags );
 342                         $this->addChunk( $add, $row->cur_id );
 343                 }
 344                 $this->lastChunk( $add );
 345                 $this->dbr->freeResult( $result );
 346
 347                 /**
 348                  * Copy revision metadata from old into revision.
 349                  * We'll also do UTF-8 conversion of usernames and comments.
 350                  */
 351                 #$newmaxold = $this->dbw->selectField( 'old', 'max(old_id)', '', $fname );
 352                 #$this->setChunkScale( $chunksize, $newmaxold, 'revision', $fname );
 353                 $countold = $this->dbw->selectField( 'old', 'count(old_id)', '', $fname );
 354                 $this->setChunkScale( $chunksize, $countold, 'revision', $fname );
 355
 356                 $this->log( "......Setting up revision table." );
 357                 $result = $this->dbr->query(
 358                         "SELECT old_id, cur_id, old_comment, old_user, old_user_text,
 359                         old_timestamp, old_minor_edit
 360                         FROM $old,$cur WHERE old_namespace=cur_namespace AND old_title=cur_title",
 361                         $fname );
 362
 363                 $add = array();
 364                 while( $row = $this->dbr->fetchObject( $result ) ) {
 365                         $add[] = array(
 366                                 'rev_id'         =>              $row->old_id,
 367                                 'rev_page'       =>              $row->cur_id,
 368                                 'rev_comment'    => $this->conv( $row->old_comment ),
 369                                 'rev_user'       =>              $row->old_user,
 370                                 'rev_user_text'  => $this->conv( $row->old_user_text ),
 371                                 'rev_timestamp'  =>              $row->old_timestamp,
 372                                 'rev_minor_edit' =>              $row->old_minor_edit );
 373                         $this->addChunk( $add );
 374                 }
 375                 $this->lastChunk( $add );
 376                 $this->dbr->freeResult( $result );
 377
 378
 379                 /**
 380                  * Copy page metadata from cur into page.
 381                  * We'll also do UTF-8 conversion of titles.
 382                  */
 383                 $this->log( "......Setting up page table." );
 384                 $this->setChunkScale( $chunksize, $maxcur, 'page', $fname );
 385                 $result = $this->dbr->query( "
 386                         SELECT cur_id, cur_namespace, cur_title, cur_restrictions, cur_counter, cur_is_redirect, cur_is_new,
 387                                 cur_random, cur_touched, rev_id, LENGTH(cur_text) AS len
 388                         FROM $cur,$revision
 389                         WHERE cur_id=rev_page AND rev_timestamp=cur_timestamp AND rev_id > {$maxold}
 390                         ORDER BY cur_id", $fname );
 391                 $add = array();
 392                 while( $row = $this->dbr->fetchObject( $result ) ) {
 393                         $add[] = array(
 394                                 'page_id'           =>              $row->cur_id,
 395                                 'page_namespace'    =>              $row->cur_namespace,
 396                                 'page_title'        => $this->conv( $row->cur_title ),
 397                                 'page_restrictions' =>              $row->cur_restrictions,
 398                                 'page_counter'      =>              $row->cur_counter,
 399                                 'page_is_redirect'  =>              $row->cur_is_redirect,
 400                                 'page_is_new'       =>              $row->cur_is_new,
 401                                 'page_random'       =>              $row->cur_random,
 402                                 'page_touched'      =>              $this->dbw->timestamp(),
 403                                 'page_latest'       =>              $row->rev_id,
 404                                 'page_len'          =>              $row->len );
 405                         $this->addChunk( $add, $row->cur_id );
 406                 }
 407                 $this->lastChunk( $add );
 408                 $this->dbr->freeResult( $result );
 409
 410                 $this->log( "...done with cur/old -> page/revision." );
 411         }
 412
 413         function upgradeLinks() {
 414                 $fname = 'FiveUpgrade::upgradeLinks';
 415                 $chunksize = 200;
 416                 extract( $this->dbw->tableNames( 'links', 'brokenlinks', 'pagelinks', 'page' ) );
 417
 418                 $this->log( 'Creating pagelinks table...' );
 419                 $this->dbw->query( "
 420 CREATE TABLE $pagelinks (
 421   -- Key to the page_id of the page containing the link.
 422   pl_from int(8) unsigned NOT NULL default '0',
 423
 424   -- Key to page_namespace/page_title of the target page.
 425   -- The target page may or may not exist, and due to renames
 426   -- and deletions may refer to different page records as time
 427   -- goes by.
 428   pl_namespace int NOT NULL default '0',
 429   pl_title varchar(255) binary NOT NULL default '',
 430
 431   UNIQUE KEY pl_from(pl_from,pl_namespace,pl_title),
 432   KEY (pl_namespace,pl_title)
 433
 434 ) TYPE=InnoDB" );
 435
 436                 $this->log( 'Importing live links -> pagelinks' );
 437                 $nlinks = $this->dbw->selectField( 'links', 'count(*)', '', $fname );
 438                 if( $nlinks ) {
 439                         $this->setChunkScale( $chunksize, $nlinks, 'pagelinks', $fname );
 440                         $result = $this->dbr->query( "
 441                           SELECT l_from,page_namespace,page_title
 442                                 FROM $links, $page
 443                                 WHERE l_to=page_id", $fname );
 444                         $add = array();
 445                         while( $row = $this->dbr->fetchObject( $result ) ) {
 446                                 $add[] = array(
 447                                         'pl_from'      => $row->l_from,
 448                                         'pl_namespace' => $row->page_namespace,
 449                                         'pl_title'     => $row->page_title );
 450                                 $this->addChunk( $add );
 451                         }
 452                         $this->lastChunk( $add );
 453                 } else {
 454                         $this->log( 'no links!' );
 455                 }
 456
 457                 $this->log( 'Importing brokenlinks -> pagelinks' );
 458                 $nbrokenlinks = $this->dbw->selectField( 'brokenlinks', 'count(*)', '', $fname );
 459                 if( $nbrokenlinks ) {
 460                         $this->setChunkScale( $chunksize, $nbrokenlinks, 'pagelinks', $fname );
 461                         $this->chunkOptions = array( 'IGNORE' );
 462                         $result = $this->dbr->query(
 463                                 "SELECT bl_from, bl_to FROM $brokenlinks",
 464                                 $fname );
 465                         $add = array();
 466                         while( $row = $this->dbr->fetchObject( $result ) ) {
 467                                 $pagename = $this->conv( $row->bl_to );
 468                                 $title = Title::newFromText( $pagename );
 469                                 if( is_null( $title ) ) {
 470                                         $this->log( "** invalid brokenlink: $row->bl_from -> '$pagename' (converted from '$row->bl_to')" );
 471                                 } else {
 472                                         $add[] = array(
 473                                                 'pl_from'      => $row->bl_from,
 474                                                 'pl_namespace' => $title->getNamespace(),
 475                                                 'pl_title'     => $title->getDBkey() );
 476                                         $this->addChunk( $add );
 477                                 }
 478                         }
 479                         $this->lastChunk( $add );
 480                 } else {
 481                         $this->log( 'no brokenlinks!' );
 482                 }
 483
 484                 $this->log( 'Done with links.' );
 485         }
 486
 487         function upgradeUser() {
 488                 $fname = 'FiveUpgrade::upgradeUser';
 489                 $chunksize = 100;
 490                 $preauth = 0;
 491
 492                 // Apply unique index, if necessary:
 493                 $duper = new UserDupes( $this->dbw );
 494                 if( $duper->hasUniqueIndex() ) {
 495                         $this->log( "Already have unique user_name index." );
 496                 } else {
 497                         $this->log( "Clearing user duplicates..." );
 498                         if( !$duper->clearDupes() ) {
 499                                 $this->log( "WARNING: Duplicate user accounts, may explode!" );
 500                         }
 501                 }
 502
 503                 /** Convert encoding on options, etc */
 504                 extract( $this->dbw->tableNames( 'user', 'user_temp', 'user_old' ) );
 505
 506                 $this->log( 'Migrating user table to user_temp...' );
 507                 $this->dbw->query( "CREATE TABLE $user_temp (
 508   user_id int(5) unsigned NOT NULL auto_increment,
 509   user_name varchar(255) binary NOT NULL default '',
 510   user_real_name varchar(255) binary NOT NULL default '',
 511   user_password tinyblob NOT NULL default '',
 512   user_newpassword tinyblob NOT NULL default '',
 513   user_email tinytext NOT NULL default '',
 514   user_options blob NOT NULL default '',
 515   user_touched char(14) binary NOT NULL default '',
 516   user_token char(32) binary NOT NULL default '',
 517   user_email_authenticated CHAR(14) BINARY,
 518   user_email_token CHAR(32) BINARY,
 519   user_email_token_expires CHAR(14) BINARY,
 520
 521   PRIMARY KEY user_id (user_id),
 522   UNIQUE INDEX user_name (user_name),
 523   INDEX (user_email_token)
 524
 525 ) TYPE=InnoDB", $fname );
 526
 527                 // Fix encoding for Latin-1 upgrades, and add some fields.
 528                 $numusers = $this->dbw->selectField( 'user', 'count(*)', '', $fname );
 529                 $this->setChunkScale( $chunksize, $numusers, 'user_temp', $fname );
 530                 $result = $this->dbr->select( 'user',
 531                         array(
 532                                 'user_id',
 533                                 'user_name',
 534                                 'user_real_name',
 535                                 'user_password',
 536                                 'user_newpassword',
 537                                 'user_email',
 538                                 'user_options',
 539                                 'user_touched',
 540                                 'user_token' ),
 541                         '',
 542                         $fname );
 543                 $add = array();
 544                 while( $row = $this->dbr->fetchObject( $result ) ) {
 545                         $now = $this->dbw->timestamp();
 546                         $add[] = array(
 547                                 'user_id'                  =>              $row->user_id,
 548                                 'user_name'                => $this->conv( $row->user_name ),
 549                                 'user_real_name'           => $this->conv( $row->user_real_name ),
 550                                 'user_password'            =>              $row->user_password,
 551                                 'user_newpassword'         =>              $row->user_newpassword,
 552                                 'user_email'               => $this->conv( $row->user_email ),
 553                                 'user_options'             => $this->conv( $row->user_options ),
 554                                 'user_touched'             =>              $now,
 555                                 'user_token'               =>              $row->user_token,
 556                                 'user_email_authenticated' =>              $preauth ? $now : null,
 557                                 'user_email_token'         =>              null,
 558                                 'user_email_token_expires' =>              null );
 559                         $this->addChunk( $add );
 560                 }
 561                 $this->lastChunk( $add );
 562                 $this->dbr->freeResult( $result );
 563                 $this->cleanupSwaps[] = 'user';
 564         }
 565
 566         function upgradeImage() {
 567                 $fname = 'FiveUpgrade::upgradeImage';
 568                 $chunksize = 100;
 569
 570                 extract( $this->dbw->tableNames( 'image', 'image_temp', 'image_old' ) );
 571                 $this->log( 'Creating temporary image_temp to merge into...' );
 572                 $this->dbw->query( <<<END
 573 CREATE TABLE $image_temp (
 574   img_name varchar(255) binary NOT NULL default '',
 575   img_size int(8) unsigned NOT NULL default '0',
 576   img_width int(5)  NOT NULL default '0',
 577   img_height int(5)  NOT NULL default '0',
 578   img_metadata mediumblob NOT NULL,
 579   img_bits int(3)  NOT NULL default '0',
 580   img_media_type ENUM("UNKNOWN", "BITMAP", "DRAWING", "AUDIO", "VIDEO", "MULTIMEDIA", "OFFICE", "TEXT", "EXECUTABLE", "ARCHIVE") default NULL,
 581   img_major_mime ENUM("unknown", "application", "audio", "image", "text", "video", "message", "model", "multipart") NOT NULL default "unknown",
 582   img_minor_mime varchar(32) NOT NULL default "unknown",
 583   img_description tinyblob NOT NULL default '',
 584   img_user int(5) unsigned NOT NULL default '0',
 585   img_user_text varchar(255) binary NOT NULL default '',
 586   img_timestamp char(14) binary NOT NULL default '',
 587
 588   PRIMARY KEY img_name (img_name),
 589   INDEX img_size (img_size),
 590   INDEX img_timestamp (img_timestamp)
 591 ) TYPE=InnoDB
 592 END
 593                 , $fname);
 594
 595                 $numimages = $this->dbw->selectField( 'image', 'count(*)', '', $fname );
 596                 $result = $this->dbr->select( 'image',
 597                         array(
 598                                 'img_name',
 599                                 'img_size',
 600                                 'img_description',
 601                                 'img_user',
 602                                 'img_user_text',
 603                                 'img_timestamp' ),
 604                         '',
 605                         $fname );
 606                 $add = array();
 607                 $this->setChunkScale( $chunksize, $numimages, 'image_temp', $fname );
 608                 while( $row = $this->dbr->fetchObject( $result ) ) {
 609                         // Fill in the new image info fields
 610                         $info = $this->imageInfo( $row->img_name );
 611
 612                         // Update and convert encoding
 613                         $add[] = array(
 614                                 'img_name'        => $this->conv( $row->img_name ),
 615                                 'img_size'        =>              $row->img_size,
 616                                 'img_width'       =>              $info['width'],
 617                                 'img_height'      =>              $info['height'],
 618                                 'img_metadata'    =>              "", // loaded on-demand
 619                                 'img_bits'        =>              $info['bits'],
 620                                 'img_media_type'  =>              $info['media'],
 621                                 'img_major_mime'  =>              $info['major'],
 622                                 'img_minor_mime'  =>              $info['minor'],
 623                                 'img_description' => $this->conv( $row->img_description ),
 624                                 'img_user'        =>              $row->img_user,
 625                                 'img_user_text'   => $this->conv( $row->img_user_text ),
 626                                 'img_timestamp'   =>              $row->img_timestamp );
 627
 628                         // If doing UTF8 conversion the file must be renamed
 629                         $this->renameFile( $row->img_name, 'wfImageDir' );
 630                 }
 631                 $this->lastChunk( $add );
 632
 633                 $this->log( 'done with image table.' );
 634                 $this->cleanupSwaps[] = 'image';
 635         }
 636
 637         function imageInfo( $name, $subdirCallback='wfImageDir', $basename = null ) {
 638                 if( is_null( $basename ) ) $basename = $name;
 639                 $dir = call_user_func( $subdirCallback, $basename );
 640                 $filename = $dir . '/' . $name;
 641                 $info = array(
 642                         'width'  => 0,
 643                         'height' => 0,
 644                         'bits'   => 0,
 645                         'media'  => '',
 646                         'major'  => '',
 647                         'minor'  => '' );
 648
 649                 $magic =& wfGetMimeMagic();
 650                 $mime = $magic->guessMimeType( $filename, true );
 651                 list( $info['major'], $info['minor'] ) = explode( '/', $mime );
 652
 653                 $info['media'] = $magic->getMediaType( $filename, $mime );
 654
 655                 # Height and width
 656                 $gis = false;
 657                 if( $mime == 'image/svg' ) {
 658                         $gis = wfGetSVGsize( $this->imagePath );
 659                 } elseif( $magic->isPHPImageType( $mime ) ) {
 660                         $gis = getimagesize( $filename );
 661                 } else {
 662                         $this->log( "Surprising mime type: $mime" );
 663                 }
 664                 if( $gis ) {
 665                         $info['width' ] = $gis[0];
 666                         $info['height'] = $gis[1];
 667                 }
 668                 if( isset( $gis['bits'] ) ) {
 669                         $info['bits'] = $gis['bits'];
 670                 }
 671
 672                 return $info;
 673         }
 674
 675
 676         /**
 677          * Truncate a table.
 678          * @param string $table The table name to be truncated
 679          */
 680         function clearTable( $table ) {
 681                 print "Clearing $table...\n";
 682                 $tableName = $this->db->tableName( $table );
 683                 $this->db->query( 'TRUNCATE $tableName' );
 684         }
 685
 686         /**
 687          * Rename a given image or archived image file to the converted filename,
 688          * leaving a symlink for URL compatibility.
 689          *
 690          * @param string $oldname pre-conversion filename
 691          * @param string $basename pre-conversion base filename for dir hashing, if an archive
 692          * @access private
 693          */
 694         function renameFile( $oldname, $subdirCallback='wfImageDir', $basename=null ) {
 695                 $newname = $this->conv( $oldname );
 696                 if( $newname == $oldname ) {
 697                         // No need to rename; another field triggered this row.
 698                         return;
 699                 }
 700
 701                 if( is_null( $basename ) ) $basename = $oldname;
 702                 $ubasename = $this->conv( $basename );
 703                 $oldpath = call_user_func( $subdirCallback, $basename ) . '/' . $oldname;
 704                 $newpath = call_user_func( $subdirCallback, $ubasename ) . '/' . $newname;
 705
 706                 $this->log( "$oldpath -> $newpath" );
 707                 if( rename( $oldpath, $newpath ) ) {
 708                         $relpath = $this->relativize( $newpath, dirname( $oldpath ) );
 709                         if( !symlink( $relpath, $oldpath ) ) {
 710                                 $this->log( "... symlink failed!" );
 711                         }
 712                 } else {
 713                         $this->log( "... rename failed!" );
 714                 }
 715         }
 716
 717         /**
 718          * Generate a relative path name to the given file.
 719          * Assumes Unix-style paths, separators, and semantics.
 720          *
 721          * @param string $path Absolute destination path including target filename
 722          * @param string $from Absolute source path, directory only
 723          * @return string
 724          * @access private
 725          * @static
 726          */
 727         function relativize( $path, $from ) {
 728                 $pieces  = explode( '/', dirname( $path ) );
 729                 $against = explode( '/', $from );
 730
 731                 // Trim off common prefix
 732                 while( count( $pieces ) && count( $against )
 733                         && $pieces[0] == $against[0] ) {
 734                         array_shift( $pieces );
 735                         array_shift( $against );
 736                 }
 737
 738                 // relative dots to bump us to the parent
 739                 while( count( $against ) ) {
 740                         array_unshift( $pieces, '..' );
 741                         array_shift( $against );
 742                 }
 743
 744                 array_push( $pieces, basename( $path ) );
 745
 746                 return implode( '/', $pieces );
 747         }
 748
 749         function upgradeOldImage() {
 750                 $fname = 'FiveUpgrade::upgradeOldImage';
 751                 $chunksize = 100;
 752
 753                 extract( $this->dbw->tableNames( 'oldimage', 'oldimage_temp', 'oldimage_old' ) );
 754                 $this->log( 'Creating temporary oldimage_temp to merge into...' );
 755                 $this->dbw->query( <<<END
 756 CREATE TABLE $oldimage_temp (
 757   -- Base filename: key to image.img_name
 758   oi_name varchar(255) binary NOT NULL default '',
 759
 760   -- Filename of the archived file.
 761   -- This is generally a timestamp and '!' prepended to the base name.
 762   oi_archive_name varchar(255) binary NOT NULL default '',
 763
 764   -- Other fields as in image...
 765   oi_size int(8) unsigned NOT NULL default 0,
 766   oi_width int(5) NOT NULL default 0,
 767   oi_height int(5) NOT NULL default 0,
 768   oi_bits int(3) NOT NULL default 0,
 769   oi_description tinyblob NOT NULL default '',
 770   oi_user int(5) unsigned NOT NULL default '0',
 771   oi_user_text varchar(255) binary NOT NULL default '',
 772   oi_timestamp char(14) binary NOT NULL default '',
 773
 774   INDEX oi_name (oi_name(10))
 775
 776 ) TYPE=InnoDB;
 777 END
 778                 , $fname);
 779
 780                 $numimages = $this->dbw->selectField( 'oldimage', 'count(*)', '', $fname );
 781                 $result = $this->dbr->select( 'oldimage',
 782                         array(
 783                                 'oi_name',
 784                                 'oi_archive_name',
 785                                 'oi_size',
 786                                 'oi_description',
 787                                 'oi_user',
 788                                 'oi_user_text',
 789                                 'oi_timestamp' ),
 790                         '',
 791                         $fname );
 792                 $add = array();
 793                 $this->setChunkScale( $chunksize, $numimages, 'oldimage_temp', $fname );
 794                 while( $row = $this->dbr->fetchObject( $result ) ) {
 795                         // Fill in the new image info fields
 796                         $info = $this->imageInfo( $row->oi_archive_name, 'wfImageArchiveDir', $row->oi_name );
 797
 798                         // Update and convert encoding
 799                         $add[] = array(
 800                                 'oi_name'         => $this->conv( $row->oi_name ),
 801                                 'oi_archive_name' => $this->conv( $row->oi_archive_name ),
 802                                 'oi_size'         =>              $row->oi_size,
 803                                 'oi_width'        =>              $info['width'],
 804                                 'oi_height'       =>              $info['height'],
 805                                 'oi_bits'         =>              $info['bits'],
 806                                 'oi_description'  => $this->conv( $row->oi_description ),
 807                                 'oi_user'         =>              $row->oi_user,
 808                                 'oi_user_text'    => $this->conv( $row->oi_user_text ),
 809                                 'oi_timestamp'    =>              $row->oi_timestamp );
 810
 811                         // If doing UTF8 conversion the file must be renamed
 812                         $this->renameFile( $row->oi_archive_name, 'wfImageArchiveDir', $row->oi_name );
 813                 }
 814                 $this->lastChunk( $add );
 815
 816                 $this->log( 'done with oldimage table.' );
 817                 $this->cleanupSwaps[] = 'oldimage';
 818         }
 819
 820
 821         function upgradeWatchlist() {
 822                 $fname = 'FiveUpgrade::upgradeWatchlist';
 823                 $chunksize = 100;
 824
 825                 extract( $this->dbw->tableNames( 'watchlist', 'watchlist_temp' ) );
 826
 827                 $this->log( 'Migrating watchlist table to watchlist_temp...' );
 828                 $this->dbw->query(
 829 "CREATE TABLE $watchlist_temp (
 830   -- Key to user_id
 831   wl_user int(5) unsigned NOT NULL,
 832
 833   -- Key to page_namespace/page_title
 834   -- Note that users may watch patches which do not exist yet,
 835   -- or existed in the past but have been deleted.
 836   wl_namespace int NOT NULL default '0',
 837   wl_title varchar(255) binary NOT NULL default '',
 838
 839   -- Timestamp when user was last sent a notification e-mail;
 840   -- cleared when the user visits the page.
 841   -- FIXME: add proper null support etc
 842   wl_notificationtimestamp varchar(14) binary NOT NULL default '0',
 843
 844   UNIQUE KEY (wl_user, wl_namespace, wl_title),
 845   KEY namespace_title (wl_namespace,wl_title)
 846
 847 ) TYPE=InnoDB;", $fname );
 848
 849                 // Fix encoding for Latin-1 upgrades, add some fields,
 850                 // and double article to article+talk pairs
 851                 $numwatched = $this->dbw->selectField( 'watchlist', 'count(*)', '', $fname );
 852
 853                 $this->setChunkScale( $chunksize, $numwatched * 2, 'watchlist_temp', $fname );
 854                 $result = $this->dbr->select( 'watchlist',
 855                         array(
 856                                 'wl_user',
 857                                 'wl_namespace',
 858                                 'wl_title' ),
 859                         '',
 860                         $fname );
 861
 862                 $add = array();
 863                 while( $row = $this->dbr->fetchObject( $result ) ) {
 864                         $now = $this->dbw->timestamp();
 865                         $add[] = array(
 866                                 'wl_user'      =>                        $row->wl_user,
 867                                 'wl_namespace' => Namespace::getSubject( $row->wl_namespace ),
 868                                 'wl_title'     =>           $this->conv( $row->wl_title ),
 869                                 'wl_notificationtimestamp' =>            '0' );
 870                         $this->addChunk( $add );
 871
 872                         $add[] = array(
 873                                 'wl_user'      =>                        $row->wl_user,
 874                                 'wl_namespace' =>    Namespace::getTalk( $row->wl_namespace ),
 875                                 'wl_title'     =>           $this->conv( $row->wl_title ),
 876                                 'wl_notificationtimestamp' =>            '0' );
 877                         $this->addChunk( $add );
 878                 }
 879                 $this->lastChunk( $add );
 880                 $this->dbr->freeResult( $result );
 881
 882                 $this->log( 'Done converting watchlist.' );
 883                 $this->cleanupSwaps[] = 'watchlist';
 884         }
 885
 886         function upgradeLogging() {
 887                 global $wgUseLatin1;
 888                 if( !$wgUseLatin1 ) {
 889                         $this->log( 'Not latin1; no change to logging table.' );
 890                         return;
 891                 }
 892
 893                 $fname = 'FiveUpgrade::upgradeLogging';
 894                 $chunksize = 100;
 895
 896                 extract( $this->dbw->tableNames( 'logging', 'logging_temp' ) );
 897
 898                 $this->log( 'Migrating logging table to logging_temp...' );
 899                 $this->dbw->query(
 900 "CREATE TABLE $logging_temp (
 901   -- Symbolic keys for the general log type and the action type
 902   -- within the log. The output format will be controlled by the
 903   -- action field, but only the type controls categorization.
 904   log_type char(10) NOT NULL default '',
 905   log_action char(10) NOT NULL default '',
 906
 907   -- Timestamp. Duh.
 908   log_timestamp char(14) NOT NULL default '19700101000000',
 909
 910   -- The user who performed this action; key to user_id
 911   log_user int unsigned NOT NULL default 0,
 912
 913   -- Key to the page affected. Where a user is the target,
 914   -- this will point to the user page.
 915   log_namespace int NOT NULL default 0,
 916   log_title varchar(255) binary NOT NULL default '',
 917
 918   -- Freeform text. Interpreted as edit history comments.
 919   log_comment varchar(255) NOT NULL default '',
 920
 921   -- LF separated list of miscellaneous parameters
 922   log_params blob NOT NULL default '',
 923
 924   KEY type_time (log_type, log_timestamp),
 925   KEY user_time (log_user, log_timestamp),
 926   KEY page_time (log_namespace, log_title, log_timestamp)
 927
 928 ) TYPE=InnoDB", $fname );
 929
 930                 $numlogged = $this->dbw->selectField( 'logging', 'count(*)', '', $fname );
 931                 $this->setChunkScale( $chunksize, $numlogged, 'logging_temp', $fname );
 932
 933                 $result = $this->dbr->select( 'logging',
 934                         array(
 935                                 'log_type',
 936                                 'log_action',
 937                                 'log_timestamp',
 938                                 'log_user',
 939                                 'log_namespace',
 940                                 'log_title',
 941                                 'log_comment',
 942                                 'log_params' ),
 943                         '',
 944                         $fname );
 945
 946                 $add = array();
 947                 while( $row = $this->dbr->fetchObject( $result ) ) {
 948                         $now = $this->dbw->timestamp();
 949                         $add[] = array(
 950                                 'log_type'      =>              $row->log_type,
 951                                 'log_action'    =>              $row->log_type,
 952                                 'log_timestamp' =>              $row->log_timestamp,
 953                                 'log_user'      =>              $row->log_user,
 954                                 'log_namespace' =>              $row->log_namespace,
 955                                 'log_title'     => $this->conv( $row->log_title ),
 956                                 'log_comment'   => $this->conv( $row->log_comment ),
 957                                 'log_params'    => $this->conv( $row->log_params ) );
 958                         $this->addChunk( $add );
 959                 }
 960                 $this->lastChunk( $add );
 961                 $this->dbr->freeResult( $result );
 962
 963                 $this->log( 'Done converting logging.' );
 964                 $this->cleanupSwaps[] = 'logging';
 965         }
 966
 967         function upgradeArchive() {
 968                 $fname = 'FiveUpgrade::upgradeArchive';
 969                 $chunksize = 100;
 970
 971                 extract( $this->dbw->tableNames( 'archive', 'archive_temp' ) );
 972
 973                 $this->log( 'Migrating archive table to archive_temp...' );
 974                 $this->dbw->query(
 975 "CREATE TABLE $archive_temp (
 976   ar_namespace int NOT NULL default '0',
 977   ar_title varchar(255) binary NOT NULL default '',
 978
 979   -- Newly deleted pages will not store text in this table,
 980   -- but will reference the separately existing text rows.
 981   -- This field is retained for backwards compatibility,
 982   -- so old archived pages will remain accessible after
 983   -- upgrading from 1.4 to 1.5.
 984   ar_text mediumblob NOT NULL default '',
 985
 986   -- Basic revision stuff...
 987   ar_comment tinyblob NOT NULL default '',
 988   ar_user int(5) unsigned NOT NULL default '0',
 989   ar_user_text varchar(255) binary NOT NULL,
 990   ar_timestamp char(14) binary NOT NULL default '',
 991   ar_minor_edit tinyint(1) NOT NULL default '0',
 992
 993   -- See ar_text note.
 994   ar_flags tinyblob NOT NULL default '',
 995
 996   -- When revisions are deleted, their unique rev_id is stored
 997   -- here so it can be retained after undeletion. This is necessary
 998   -- to retain permalinks to given revisions after accidental delete
 999   -- cycles or messy operations like history merges.
1000   --
1001   -- Old entries from 1.4 will be NULL here, and a new rev_id will
1002   -- be created on undeletion for those revisions.
1003   ar_rev_id int(8) unsigned,
1004
1005   -- For newly deleted revisions, this is the text.old_id key to the
1006   -- actual stored text. To avoid breaking the block-compression scheme
1007   -- and otherwise making storage changes harder, the actual text is
1008   -- *not* deleted from the text table, merely hidden by removal of the
1009   -- page and revision entries.
1010   --
1011   -- Old entries deleted under 1.2-1.4 will have NULL here, and their
1012   -- ar_text and ar_flags fields will be used to create a new text
1013   -- row upon undeletion.
1014   ar_text_id int(8) unsigned,
1015
1016   KEY name_title_timestamp (ar_namespace,ar_title,ar_timestamp)
1017
1018 ) TYPE=InnoDB", $fname );
1019
1020                 $numarchived = $this->dbw->selectField( 'archive', 'count(*)', '', $fname );
1021                 $this->setChunkScale( $chunksize, $numarchived, 'archive_temp', $fname );
1022
1023                 $result = $this->dbr->select( 'archive',
1024                         array(
1025                                 'ar_namespace',
1026                                 'ar_title',
1027                                 'ar_text',
1028                                 'ar_comment',
1029                                 'ar_user',
1030                                 'ar_user_text',
1031                                 'ar_timestamp',
1032                                 'ar_minor_edit',
1033                                 'ar_flags' ),
1034                         '',
1035                         $fname );
1036
1037                 $add = array();
1038                 while( $row = $this->dbr->fetchObject( $result ) ) {
1039                         $now = $this->dbw->timestamp();
1040                         $add[] = array(
1041                                 'ar_namespace'  =>              $row->ar_namespace,
1042                                 'ar_title'      => $this->conv( $row->ar_title ),
1043                                 'ar_text'       =>              $row->ar_text,
1044                                 'ar_comment'    => $this->conv( $row->ar_comment ),
1045                                 'ar_user'       =>              $row->ar_user,
1046                                 'ar_user_text'  => $this->conv( $row->ar_user_text ),
1047                                 'ar_timestamp'  =>              $row->ar_user_text,
1048                                 'ar_minor_edit' =>              $row->ar_minor_edit,
1049                                 'ar_flags'      =>              $row->ar_flags,
1050                                 'ar_rev_id'     => null,
1051                                 'ar_text_id'    => null );
1052                         $this->addChunk( $add );
1053                 }
1054                 $this->lastChunk( $add );
1055                 $this->dbr->freeResult( $result );
1056
1057                 $this->log( 'Done converting archive.' );
1058                 $this->cleanupSwaps[] = 'archive';
1059         }
1060
1061         /**
1062          * Rename all our temporary tables into final place.
1063          * We've left things in place so a read-only wiki can continue running
1064          * on the old code during all this.
1065          */
1066         function upgradeCleanup() {
1067                 $this->renameTable( 'old', 'text' );
1068
1069                 foreach( $this->cleanupSwaps as $table ) {
1070                         $this->swap( $table );
1071                 }
1072         }
1073
1074         function renameTable( $from, $to ) {
1075                 $this->log( 'Renaming $from to $to...' );
1076
1077                 $fromtable = $this->dbw->tableName( $from );
1078                 $totable   = $this->dbw->tableName( $to );
1079                 $this->dbw->query( "ALTER TABLE $fromtable RENAME TO $totable" );
1080         }
1081
1082         function swap( $base ) {
1083                 $this->renameTable( $base, "{$base}_old" );
1084                 $this->renameTable( "{$base}_temp", $base );
1085         }
1086
1087 }
1088
1089 ?>