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