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 define( 'MW_UPGRADE_COPY',     false );
  23 define( 'MW_UPGRADE_ENCODE',   true  );
  24 define( 'MW_UPGRADE_NULL',     null  );
  25 define( 'MW_UPGRADE_CALLBACK', null  ); // for self-documentation only
  26
  27 class FiveUpgrade {
  28         function FiveUpgrade() {
  29                 global $wgDatabase;
  30                 $this->conversionTables = $this->prepareWindows1252();
  31                 $this->dbw =& $this->newConnection();
  32                 $this->dbr =& $this->newConnection();
  33                 $this->dbr->bufferResults( false );
  34                 $this->cleanupSwaps = array();
  35
  36                 $this->emailAuth = false; # don't preauthenticate emails
  37         }
  38
  39         function doing( $step ) {
  40                 return is_null( $this->step ) || $step == $this->step;
  41         }
  42
  43         function upgrade( $step ) {
  44                 $this->step = $step;
  45                 if( $this->doing( 'page' ) )
  46                         $this->upgradePage();
  47                 if( $this->doing( 'links' ) )
  48                         $this->upgradeLinks();
  49                 if( $this->doing( 'user' ) )
  50                         $this->upgradeUser();
  51                 if( $this->doing( 'image' ) )
  52                         $this->upgradeImage();
  53                 if( $this->doing( 'oldimage' ) )
  54                         $this->upgradeOldImage();
  55                 if( $this->doing( 'watchlist' ) )
  56                         $this->upgradeWatchlist();
  57                 if( $this->doing( 'logging' ) )
  58                         $this->upgradeLogging();
  59                 if( $this->doing( 'archive' ) )
  60                         $this->upgradeArchive();
  61                 if( $this->doing( 'imagelinks' ) )
  62                         $this->upgradeImagelinks();
  63                 if( $this->doing( 'categorylinks' ) )
  64                         $this->upgradeCategorylinks();
  65
  66                 if( $this->doing( 'cleanup' ) )
  67                         $this->upgradeCleanup();
  68         }
  69
  70
  71         /**
  72          * Open a second connection to the master server, with buffering off.
  73          * This will let us stream large datasets in and write in chunks on the
  74          * other end.
  75          * @return Database
  76          * @access private
  77          */
  78         function &newConnection() {
  79                 global $wgDBadminuser, $wgDBadminpassword;
  80                 global $wgDBserver, $wgDBname;
  81                 $db =& new Database( $wgDBserver, $wgDBadminuser, $wgDBadminpassword, $wgDBname );
  82                 return $db;
  83         }
  84
  85         /**
  86          * Prepare a conversion array for converting Windows Code Page 1252 to
  87          * UTF-8. This should provide proper conversion of text that was miscoded
  88          * as Windows-1252 by naughty user-agents, and doesn't rely on an outside
  89          * iconv library.
  90          *
  91          * @return array
  92          * @access private
  93          */
  94         function prepareWindows1252() {
  95                 # Mappings from:
  96                 # http://www.unicode.org/Public/MAPPINGS/VENDORS/MICSFT/WINDOWS/CP1252.TXT
  97                 static $cp1252 = array(
  98                         0x80 => 0x20AC, #EURO SIGN
  99                         0x81 => UNICODE_REPLACEMENT,
 100                         0x82 => 0x201A, #SINGLE LOW-9 QUOTATION MARK
 101                         0x83 => 0x0192, #LATIN SMALL LETTER F WITH HOOK
 102                         0x84 => 0x201E, #DOUBLE LOW-9 QUOTATION MARK
 103                         0x85 => 0x2026, #HORIZONTAL ELLIPSIS
 104                         0x86 => 0x2020, #DAGGER
 105                         0x87 => 0x2021, #DOUBLE DAGGER
 106                         0x88 => 0x02C6, #MODIFIER LETTER CIRCUMFLEX ACCENT
 107                         0x89 => 0x2030, #PER MILLE SIGN
 108                         0x8A => 0x0160, #LATIN CAPITAL LETTER S WITH CARON
 109                         0x8B => 0x2039, #SINGLE LEFT-POINTING ANGLE QUOTATION MARK
 110                         0x8C => 0x0152, #LATIN CAPITAL LIGATURE OE
 111                         0x8D => UNICODE_REPLACEMENT,
 112                         0x8E => 0x017D, #LATIN CAPITAL LETTER Z WITH CARON
 113                         0x8F => UNICODE_REPLACEMENT,
 114                         0x90 => UNICODE_REPLACEMENT,
 115                         0x91 => 0x2018, #LEFT SINGLE QUOTATION MARK
 116                         0x92 => 0x2019, #RIGHT SINGLE QUOTATION MARK
 117                         0x93 => 0x201C, #LEFT DOUBLE QUOTATION MARK
 118                         0x94 => 0x201D, #RIGHT DOUBLE QUOTATION MARK
 119                         0x95 => 0x2022, #BULLET
 120                         0x96 => 0x2013, #EN DASH
 121                         0x97 => 0x2014, #EM DASH
 122                         0x98 => 0x02DC, #SMALL TILDE
 123                         0x99 => 0x2122, #TRADE MARK SIGN
 124                         0x9A => 0x0161, #LATIN SMALL LETTER S WITH CARON
 125                         0x9B => 0x203A, #SINGLE RIGHT-POINTING ANGLE QUOTATION MARK
 126                         0x9C => 0x0153, #LATIN SMALL LIGATURE OE
 127                         0x9D => UNICODE_REPLACEMENT,
 128                         0x9E => 0x017E, #LATIN SMALL LETTER Z WITH CARON
 129                         0x9F => 0x0178, #LATIN CAPITAL LETTER Y WITH DIAERESIS
 130                         );
 131                 $pairs = array();
 132                 for( $i = 0; $i < 0x100; $i++ ) {
 133                         $unicode = isset( $cp1252[$i] ) ? $cp1252[$i] : $i;
 134                         $pairs[chr( $i )] = codepointToUtf8( $unicode );
 135                 }
 136                 return $pairs;
 137         }
 138
 139         /**
 140          * Convert from 8-bit Windows-1252 to UTF-8 if necessary.
 141          * @param string $text
 142          * @return string
 143          * @access private
 144          */
 145         function conv( $text ) {
 146                 global $wgUseLatin1;
 147                 return is_null( $text )
 148                         ? null
 149                         : ( $wgUseLatin1
 150                                 ? strtr( $text, $this->conversionTables )
 151                                 : $text );
 152         }
 153
 154         /**
 155          * Dump timestamp and message to output
 156          * @param string $message
 157          * @access private
 158          */
 159         function log( $message ) {
 160                 echo wfTimestamp( TS_DB ) . ': ' . $message . "\n";
 161                 flush();
 162         }
 163
 164         /**
 165          * Initialize the chunked-insert system.
 166          * Rows will be inserted in chunks of the given number, rather
 167          * than in a giant INSERT...SELECT query, to keep the serialized
 168          * MySQL database replication from getting hung up. This way other
 169          * things can be going on during conversion without waiting for
 170          * slaves to catch up as badly.
 171          *
 172          * @param int $chunksize Number of rows to insert at once
 173          * @param int $final Total expected number of rows / id of last row,
 174          *                   used for progress reports.
 175          * @param string $table to insert on
 176          * @param string $fname function name to report in SQL
 177          * @access private
 178          */
 179         function setChunkScale( $chunksize, $final, $table, $fname ) {
 180                 $this->chunkSize  = $chunksize;
 181                 $this->chunkFinal = $final;
 182                 $this->chunkCount = 0;
 183                 $this->chunkStartTime = wfTime();
 184                 $this->chunkOptions = array();
 185                 $this->chunkTable = $table;
 186                 $this->chunkFunction = $fname;
 187         }
 188
 189         /**
 190          * Chunked inserts: perform an insert if we've reached the chunk limit.
 191          * Prints a progress report with estimated completion time.
 192          * @param array &$chunk -- This will be emptied if an insert is done.
 193          * @param int $key A key identifier to use in progress estimation in
 194          *                 place of the number of rows inserted. Use this if
 195          *                 you provided a max key number instead of a count
 196          *                 as the final chunk number in setChunkScale()
 197          * @access private
 198          */
 199         function addChunk( &$chunk, $key = null ) {
 200                 if( count( $chunk ) >= $this->chunkSize ) {
 201                         $this->insertChunk( $chunk );
 202
 203                         $this->chunkCount += count( $chunk );
 204                         $now = wfTime();
 205                         $delta = $now - $this->chunkStartTime;
 206                         $rate = $this->chunkCount / $delta;
 207
 208                         if( is_null( $key ) ) {
 209                                 $completed = $this->chunkCount;
 210                         } else {
 211                                 $completed = $key;
 212                         }
 213                         $portion = $completed / $this->chunkFinal;
 214
 215                         $estimatedTotalTime = $delta / $portion;
 216                         $eta = $this->chunkStartTime + $estimatedTotalTime;
 217
 218                         printf( "%s: %6.2f%% done on %s; ETA %s [%d/%d] %.2f/sec\n",
 219                                 wfTimestamp( TS_DB, intval( $now ) ),
 220                                 $portion * 100.0,
 221                                 $this->chunkTable,
 222                                 wfTimestamp( TS_DB, intval( $eta ) ),
 223                                 $completed,
 224                                 $this->chunkFinal,
 225                                 $rate );
 226                         flush();
 227
 228                         $chunk = array();
 229                 }
 230         }
 231
 232         /**
 233          * Chunked inserts: perform an insert unconditionally, at the end, and log.
 234          * @param array &$chunk -- This will be emptied if an insert is done.
 235          * @access private
 236          */
 237         function lastChunk( &$chunk ) {
 238                 $n = count( $chunk );
 239                 if( $n > 0 ) {
 240                         $this->insertChunk( $chunk );
 241                 }
 242                 $this->log( "100.00% done on $this->chunkTable (last chunk $n rows)." );
 243         }
 244
 245         /**
 246          * Chunked inserts: perform an insert.
 247          * @param array &$chunk -- This will be emptied if an insert is done.
 248          * @access private
 249          */
 250         function insertChunk( &$chunk ) {
 251                 $this->dbw->insert( $this->chunkTable, $chunk, $this->chunkFunction, $this->chunkOptions );
 252         }
 253
 254
 255         /**
 256          * Copy and transcode a table to table_temp.
 257          * @param string $name Base name of the source table
 258          * @param string $tabledef CREATE TABLE definition, w/ $1 for the name
 259          * @param array $fields set of destination fields to these constants:
 260          *              MW_UPGRADE_COPY   - straight copy
 261          *              MW_UPGRADE_ENCODE - for old Latin1 wikis, conv to UTF-8
 262          *              MW_UPGRADE_NULL   - just put NULL
 263          * @param callable $callback An optional callback to modify the data
 264          *                           or perform other processing. Func should be
 265          *                           ( object $row, array $copy ) and return $copy
 266          * @access private
 267          */
 268         function copyTable( $name, $tabledef, $fields, $callback = null ) {
 269                 $fname = 'FiveUpgrade::copyTable';
 270
 271                 $name_temp = $name . '_temp';
 272                 $this->log( "Migrating $name table to $name_temp..." );
 273
 274                 $table      = $this->dbw->tableName( $name );
 275                 $table_temp = $this->dbw->tableName( $name_temp );
 276
 277                 // Create temporary table; we're going to copy everything in there,
 278                 // then at the end rename the final tables into place.
 279                 $def = str_replace( '$1', $table_temp, $tabledef );
 280                 $this->dbw->query( $def, $fname );
 281
 282                 $numRecords = $this->dbw->selectField( $name, 'COUNT(*)', '', $fname );
 283                 $this->setChunkScale( 100, $numRecords, $name_temp, $fname );
 284
 285                 // Pull all records from the second, streaming database connection.
 286                 $sourceFields = array_keys( array_filter( $fields,
 287                         create_function( '$x', 'return $x !== MW_UPGRADE_NULL;' ) ) );
 288                 $result = $this->dbr->select( $name,
 289                         $sourceFields,
 290                         '',
 291                         $fname );
 292
 293                 $add = array();
 294                 while( $row = $this->dbr->fetchObject( $result ) ) {
 295                         $copy = array();
 296                         foreach( $fields as $field => $source ) {
 297                                 if( $source === MW_UPGRADE_COPY ) {
 298                                         $copy[$field] = $row->$field;
 299                                 } elseif( $source === MW_UPGRADE_ENCODE ) {
 300                                         $copy[$field] = $this->conv( $row->$field );
 301                                 } elseif( $source === MW_UPGRADE_NULL ) {
 302                                         $copy[$field] = null;
 303                                 } else {
 304                                         $this->log( "Unknown field copy type: $field => $source" );
 305                                 }
 306                         }
 307                         if( is_callable( $callback ) ) {
 308                                 $copy = call_user_func( $callback, $row, $copy );
 309                         }
 310                         $add[] = $copy;
 311                         $this->addChunk( $add );
 312                 }
 313                 $this->lastChunk( $add );
 314                 $this->dbr->freeResult( $result );
 315
 316                 $this->log( "Done converting $name." );
 317                 $this->cleanupSwaps[] = $name;
 318         }
 319
 320         function upgradePage() {
 321                 $fname = "FiveUpgrade::upgradePage";
 322                 $chunksize = 100;
 323
 324
 325                 $this->log( "Checking cur table for unique title index and applying if necessary" );
 326                 checkDupes( true );
 327
 328                 $this->log( "...converting from cur/old to page/revision/text DB structure." );
 329
 330                 extract( $this->dbw->tableNames( 'cur', 'old', 'page', 'revision', 'text' ) );
 331
 332                 $this->log( "Creating page and revision tables..." );
 333                 $this->dbw->query("CREATE TABLE $page (
 334                         page_id int(8) unsigned NOT NULL auto_increment,
 335                         page_namespace int NOT NULL,
 336                         page_title varchar(255) binary NOT NULL,
 337                         page_restrictions tinyblob NOT NULL default '',
 338                         page_counter bigint(20) unsigned NOT NULL default '0',
 339                         page_is_redirect tinyint(1) unsigned NOT NULL default '0',
 340                         page_is_new tinyint(1) unsigned NOT NULL default '0',
 341                         page_random real unsigned NOT NULL,
 342                         page_touched char(14) binary NOT NULL default '',
 343                         page_latest int(8) unsigned NOT NULL,
 344                         page_len int(8) unsigned NOT NULL,
 345
 346                         PRIMARY KEY page_id (page_id),
 347                         UNIQUE INDEX name_title (page_namespace,page_title),
 348                         INDEX (page_random),
 349                         INDEX (page_len)
 350                         ) TYPE=InnoDB", $fname );
 351                 $this->dbw->query("CREATE TABLE $revision (
 352                         rev_id int(8) unsigned NOT NULL auto_increment,
 353                         rev_page int(8) unsigned NOT NULL,
 354                         rev_comment tinyblob NOT NULL default '',
 355                         rev_user int(5) unsigned NOT NULL default '0',
 356                         rev_user_text varchar(255) binary NOT NULL default '',
 357                         rev_timestamp char(14) binary NOT NULL default '',
 358                         rev_minor_edit tinyint(1) unsigned NOT NULL default '0',
 359                         rev_deleted tinyint(1) unsigned NOT NULL default '0',
 360
 361                         PRIMARY KEY rev_page_id (rev_page, rev_id),
 362                         UNIQUE INDEX rev_id (rev_id),
 363                         INDEX rev_timestamp (rev_timestamp),
 364                         INDEX page_timestamp (rev_page,rev_timestamp),
 365                         INDEX user_timestamp (rev_user,rev_timestamp),
 366                         INDEX usertext_timestamp (rev_user_text,rev_timestamp)
 367                         ) TYPE=InnoDB", $fname );
 368
 369                 $maxold = $this->dbw->selectField( 'old', 'max(old_id)', '', $fname );
 370                 $this->log( "Last old record is {$maxold}" );
 371
 372                 global $wgLegacySchemaConversion;
 373                 if( $wgLegacySchemaConversion ) {
 374                         // Create HistoryBlobCurStub entries.
 375                         // Text will be pulled from the leftover 'cur' table at runtime.
 376                         echo "......Moving metadata from cur; using blob references to text in cur table.\n";
 377                         $cur_text = "concat('O:18:\"historyblobcurstub\":1:{s:6:\"mCurId\";i:',cur_id,';}')";
 378                         $cur_flags = "'object'";
 379                 } else {
 380                         // Copy all cur text in immediately: this may take longer but avoids
 381                         // having to keep an extra table around.
 382                         echo "......Moving text from cur.\n";
 383                         $cur_text = 'cur_text';
 384                         $cur_flags = "''";
 385                 }
 386
 387                 $maxcur = $this->dbw->selectField( 'cur', 'max(cur_id)', '', $fname );
 388                 $this->log( "Last cur entry is $maxcur" );
 389
 390                 /**
 391                  * Copy placeholder records for each page's current version into old
 392                  * Don't do any conversion here; text records are converted at runtime
 393                  * based on the flags (and may be originally binary!) while the meta
 394                  * fields will be converted in the old -> rev and cur -> page steps.
 395                  */
 396                 $this->setChunkScale( $chunksize, $maxcur, 'old', $fname );
 397                 $result = $this->dbr->query(
 398                         "SELECT cur_id, cur_namespace, cur_title, $cur_text AS text, cur_comment,
 399                         cur_user, cur_user_text, cur_timestamp, cur_minor_edit, $cur_flags AS flags
 400                         FROM $cur
 401                         ORDER BY cur_id", $fname );
 402                 $add = array();
 403                 while( $row = $this->dbr->fetchObject( $result ) ) {
 404                         $add[] = array(
 405                                 'old_namespace'  => $row->cur_namespace,
 406                                 'old_title'      => $row->cur_title,
 407                                 'old_text'       => $row->text,
 408                                 'old_comment'    => $row->cur_comment,
 409                                 'old_user'       => $row->cur_user,
 410                                 'old_user_text'  => $row->cur_user_text,
 411                                 'old_timestamp'  => $row->cur_timestamp,
 412                                 'old_minor_edit' => $row->cur_minor_edit,
 413                                 'old_flags'      => $row->flags );
 414                         $this->addChunk( $add, $row->cur_id );
 415                 }
 416                 $this->lastChunk( $add );
 417                 $this->dbr->freeResult( $result );
 418
 419                 /**
 420                  * Copy revision metadata from old into revision.
 421                  * We'll also do UTF-8 conversion of usernames and comments.
 422                  */
 423                 #$newmaxold = $this->dbw->selectField( 'old', 'max(old_id)', '', $fname );
 424                 #$this->setChunkScale( $chunksize, $newmaxold, 'revision', $fname );
 425                 $countold = $this->dbw->selectField( 'old', 'count(old_id)', '', $fname );
 426                 $this->setChunkScale( $chunksize, $countold, 'revision', $fname );
 427
 428                 $this->log( "......Setting up revision table." );
 429                 $result = $this->dbr->query(
 430                         "SELECT old_id, cur_id, old_comment, old_user, old_user_text,
 431                         old_timestamp, old_minor_edit
 432                         FROM $old,$cur WHERE old_namespace=cur_namespace AND old_title=cur_title",
 433                         $fname );
 434
 435                 $add = array();
 436                 while( $row = $this->dbr->fetchObject( $result ) ) {
 437                         $add[] = array(
 438                                 'rev_id'         =>              $row->old_id,
 439                                 'rev_page'       =>              $row->cur_id,
 440                                 'rev_comment'    => $this->conv( $row->old_comment ),
 441                                 'rev_user'       =>              $row->old_user,
 442                                 'rev_user_text'  => $this->conv( $row->old_user_text ),
 443                                 'rev_timestamp'  =>              $row->old_timestamp,
 444                                 'rev_minor_edit' =>              $row->old_minor_edit );
 445                         $this->addChunk( $add );
 446                 }
 447                 $this->lastChunk( $add );
 448                 $this->dbr->freeResult( $result );
 449
 450
 451                 /**
 452                  * Copy page metadata from cur into page.
 453                  * We'll also do UTF-8 conversion of titles.
 454                  */
 455                 $this->log( "......Setting up page table." );
 456                 $this->setChunkScale( $chunksize, $maxcur, 'page', $fname );
 457                 $result = $this->dbr->query( "
 458                         SELECT cur_id, cur_namespace, cur_title, cur_restrictions, cur_counter, cur_is_redirect, cur_is_new,
 459                                 cur_random, cur_touched, rev_id, LENGTH(cur_text) AS len
 460                         FROM $cur,$revision
 461                         WHERE cur_id=rev_page AND rev_timestamp=cur_timestamp AND rev_id > {$maxold}
 462                         ORDER BY cur_id", $fname );
 463                 $add = array();
 464                 while( $row = $this->dbr->fetchObject( $result ) ) {
 465                         $add[] = array(
 466                                 'page_id'           =>              $row->cur_id,
 467                                 'page_namespace'    =>              $row->cur_namespace,
 468                                 'page_title'        => $this->conv( $row->cur_title ),
 469                                 'page_restrictions' =>              $row->cur_restrictions,
 470                                 'page_counter'      =>              $row->cur_counter,
 471                                 'page_is_redirect'  =>              $row->cur_is_redirect,
 472                                 'page_is_new'       =>              $row->cur_is_new,
 473                                 'page_random'       =>              $row->cur_random,
 474                                 'page_touched'      =>              $this->dbw->timestamp(),
 475                                 'page_latest'       =>              $row->rev_id,
 476                                 'page_len'          =>              $row->len );
 477                         $this->addChunk( $add, $row->cur_id );
 478                 }
 479                 $this->lastChunk( $add );
 480                 $this->dbr->freeResult( $result );
 481
 482                 $this->log( "...done with cur/old -> page/revision." );
 483         }
 484
 485         function upgradeLinks() {
 486                 $fname = 'FiveUpgrade::upgradeLinks';
 487                 $chunksize = 200;
 488                 extract( $this->dbw->tableNames( 'links', 'brokenlinks', 'pagelinks', 'page' ) );
 489
 490                 $this->log( 'Creating pagelinks table...' );
 491                 $this->dbw->query( "
 492 CREATE TABLE $pagelinks (
 493   -- Key to the page_id of the page containing the link.
 494   pl_from int(8) unsigned NOT NULL default '0',
 495
 496   -- Key to page_namespace/page_title of the target page.
 497   -- The target page may or may not exist, and due to renames
 498   -- and deletions may refer to different page records as time
 499   -- goes by.
 500   pl_namespace int NOT NULL default '0',
 501   pl_title varchar(255) binary NOT NULL default '',
 502
 503   UNIQUE KEY pl_from(pl_from,pl_namespace,pl_title),
 504   KEY (pl_namespace,pl_title)
 505
 506 ) TYPE=InnoDB" );
 507
 508                 $this->log( 'Importing live links -> pagelinks' );
 509                 $nlinks = $this->dbw->selectField( 'links', 'count(*)', '', $fname );
 510                 if( $nlinks ) {
 511                         $this->setChunkScale( $chunksize, $nlinks, 'pagelinks', $fname );
 512                         $result = $this->dbr->query( "
 513                           SELECT l_from,page_namespace,page_title
 514                                 FROM $links, $page
 515                                 WHERE l_to=page_id", $fname );
 516                         $add = array();
 517                         while( $row = $this->dbr->fetchObject( $result ) ) {
 518                                 $add[] = array(
 519                                         'pl_from'      => $row->l_from,
 520                                         'pl_namespace' => $row->page_namespace,
 521                                         'pl_title'     => $row->page_title );
 522                                 $this->addChunk( $add );
 523                         }
 524                         $this->lastChunk( $add );
 525                 } else {
 526                         $this->log( 'no links!' );
 527                 }
 528
 529                 $this->log( 'Importing brokenlinks -> pagelinks' );
 530                 $nbrokenlinks = $this->dbw->selectField( 'brokenlinks', 'count(*)', '', $fname );
 531                 if( $nbrokenlinks ) {
 532                         $this->setChunkScale( $chunksize, $nbrokenlinks, 'pagelinks', $fname );
 533                         $this->chunkOptions = array( 'IGNORE' );
 534                         $result = $this->dbr->query(
 535                                 "SELECT bl_from, bl_to FROM $brokenlinks",
 536                                 $fname );
 537                         $add = array();
 538                         while( $row = $this->dbr->fetchObject( $result ) ) {
 539                                 $pagename = $this->conv( $row->bl_to );
 540                                 $title = Title::newFromText( $pagename );
 541                                 if( is_null( $title ) ) {
 542                                         $this->log( "** invalid brokenlink: $row->bl_from -> '$pagename' (converted from '$row->bl_to')" );
 543                                 } else {
 544                                         $add[] = array(
 545                                                 'pl_from'      => $row->bl_from,
 546                                                 'pl_namespace' => $title->getNamespace(),
 547                                                 'pl_title'     => $title->getDBkey() );
 548                                         $this->addChunk( $add );
 549                                 }
 550                         }
 551                         $this->lastChunk( $add );
 552                 } else {
 553                         $this->log( 'no brokenlinks!' );
 554                 }
 555
 556                 $this->log( 'Done with links.' );
 557         }
 558
 559         function upgradeUser() {
 560                 // Apply unique index, if necessary:
 561                 $duper = new UserDupes( $this->dbw );
 562                 if( $duper->hasUniqueIndex() ) {
 563                         $this->log( "Already have unique user_name index." );
 564                 } else {
 565                         $this->log( "Clearing user duplicates..." );
 566                         if( !$duper->clearDupes() ) {
 567                                 $this->log( "WARNING: Duplicate user accounts, may explode!" );
 568                         }
 569                 }
 570
 571                 $tabledef = <<<END
 572 CREATE TABLE $1 (
 573   user_id int(5) unsigned NOT NULL auto_increment,
 574   user_name varchar(255) binary NOT NULL default '',
 575   user_real_name varchar(255) binary NOT NULL default '',
 576   user_password tinyblob NOT NULL default '',
 577   user_newpassword tinyblob NOT NULL default '',
 578   user_email tinytext NOT NULL default '',
 579   user_options blob NOT NULL default '',
 580   user_touched char(14) binary NOT NULL default '',
 581   user_token char(32) binary NOT NULL default '',
 582   user_email_authenticated CHAR(14) BINARY,
 583   user_email_token CHAR(32) BINARY,
 584   user_email_token_expires CHAR(14) BINARY,
 585
 586   PRIMARY KEY user_id (user_id),
 587   UNIQUE INDEX user_name (user_name),
 588   INDEX (user_email_token)
 589
 590 ) TYPE=InnoDB
 591 END;
 592                 $fields = array(
 593                         'user_id'                  => MW_UPGRADE_COPY,
 594                         'user_name'                => MW_UPGRADE_ENCODE,
 595                         'user_real_name'           => MW_UPGRADE_ENCODE,
 596                         'user_password'            => MW_UPGRADE_COPY,
 597                         'user_newpassword'         => MW_UPGRADE_COPY,
 598                         'user_email'               => MW_UPGRADE_ENCODE,
 599                         'user_options'             => MW_UPGRADE_ENCODE,
 600                         'user_touched'             => MW_UPGRADE_CALLBACK,
 601                         'user_token'               => MW_UPGRADE_COPY,
 602                         'user_email_authenticated' => MW_UPGRADE_CALLBACK,
 603                         'user_email_token'         => MW_UPGRADE_NULL,
 604                         'user_email_token_expires' => MW_UPGRADE_NULL );
 605                 $this->copyTable( 'user', $tabledef, $fields,
 606                         array( &$this, 'userCallback' ) );
 607         }
 608
 609         function userCallback( $row, $copy ) {
 610                 $now = $this->dbw->timestamp();
 611                 $copy['user_touched'] = $now;
 612                 $copy['user_email_authenticated'] = $this->emailAuth ? $now : null;
 613                 return $copy;
 614         }
 615
 616         function upgradeImage() {
 617                 $tabledef = <<<END
 618 CREATE TABLE $1 (
 619   img_name varchar(255) binary NOT NULL default '',
 620   img_size int(8) unsigned NOT NULL default '0',
 621   img_width int(5)  NOT NULL default '0',
 622   img_height int(5)  NOT NULL default '0',
 623   img_metadata mediumblob NOT NULL,
 624   img_bits int(3)  NOT NULL default '0',
 625   img_media_type ENUM("UNKNOWN", "BITMAP", "DRAWING", "AUDIO", "VIDEO", "MULTIMEDIA", "OFFICE", "TEXT", "EXECUTABLE", "ARCHIVE") default NULL,
 626   img_major_mime ENUM("unknown", "application", "audio", "image", "text", "video", "message", "model", "multipart") NOT NULL default "unknown",
 627   img_minor_mime varchar(32) NOT NULL default "unknown",
 628   img_description tinyblob NOT NULL default '',
 629   img_user int(5) unsigned NOT NULL default '0',
 630   img_user_text varchar(255) binary NOT NULL default '',
 631   img_timestamp char(14) binary NOT NULL default '',
 632
 633   PRIMARY KEY img_name (img_name),
 634   INDEX img_size (img_size),
 635   INDEX img_timestamp (img_timestamp)
 636 ) TYPE=InnoDB
 637 END;
 638                 $fields = array(
 639                         'img_name'        => MW_UPGRADE_ENCODE,
 640                         'img_size'        => MW_UPGRADE_COPY,
 641                         'img_width'       => MW_UPGRADE_CALLBACK,
 642                         'img_height'      => MW_UPGRADE_CALLBACK,
 643                         'img_metadata'    => MW_UPGRADE_CALLBACK,
 644                         'img_bits'        => MW_UPGRADE_CALLBACK,
 645                         'img_media_type'  => MW_UPGRADE_CALLBACK,
 646                         'img_major_mime'  => MW_UPGRADE_CALLBACK,
 647                         'img_minor_mime'  => MW_UPGRADE_CALLBACK,
 648                         'img_description' => MW_UPGRADE_ENCODE,
 649                         'img_user'        => MW_UPGRADE_COPY,
 650                         'img_user_text'   => MW_UPGRADE_ENCODE,
 651                         'img_timestamp'   => MW_UPGRADE_COPY );
 652                 $this->copyTable( 'image', $tabledef, $fields,
 653                         array( &$this, 'imageCallback' ) );
 654         }
 655
 656         function imageCallback( $row, $copy ) {
 657                 // Fill in the new image info fields
 658                 $info = $this->imageInfo( $row->img_name );
 659
 660                 $copy['img_width'     ] = $info['width'];
 661                 $copy['img_height'    ] = $info['height'];
 662                 $copy['img_metadata'  ] = ""; // loaded on-demand
 663                 $copy['img_bits'      ] = $info['bits'];
 664                 $copy['img_media_type'] = $info['media'];
 665                 $copy['img_major_mime'] = $info['major'];
 666                 $copy['img_minor_mime'] = $info['minor'];
 667
 668                 // If doing UTF8 conversion the file must be renamed
 669                 $this->renameFile( $row->img_name, 'wfImageDir' );
 670
 671                 return $copy;
 672         }
 673
 674         function imageInfo( $name, $subdirCallback='wfImageDir', $basename = null ) {
 675                 if( is_null( $basename ) ) $basename = $name;
 676                 $dir = call_user_func( $subdirCallback, $basename );
 677                 $filename = $dir . '/' . $name;
 678                 $info = array(
 679                         'width'  => 0,
 680                         'height' => 0,
 681                         'bits'   => 0,
 682                         'media'  => '',
 683                         'major'  => '',
 684                         'minor'  => '' );
 685
 686                 $magic =& wfGetMimeMagic();
 687                 $mime = $magic->guessMimeType( $filename, true );
 688                 list( $info['major'], $info['minor'] ) = explode( '/', $mime );
 689
 690                 $info['media'] = $magic->getMediaType( $filename, $mime );
 691
 692                 # Height and width
 693                 $gis = false;
 694                 if( $mime == 'image/svg' ) {
 695                         $gis = wfGetSVGsize( $this->imagePath );
 696                 } elseif( $magic->isPHPImageType( $mime ) ) {
 697                         $gis = getimagesize( $filename );
 698                 } else {
 699                         $this->log( "Surprising mime type: $mime" );
 700                 }
 701                 if( $gis ) {
 702                         $info['width' ] = $gis[0];
 703                         $info['height'] = $gis[1];
 704                 }
 705                 if( isset( $gis['bits'] ) ) {
 706                         $info['bits'] = $gis['bits'];
 707                 }
 708
 709                 return $info;
 710         }
 711
 712
 713         /**
 714          * Truncate a table.
 715          * @param string $table The table name to be truncated
 716          */
 717         function clearTable( $table ) {
 718                 print "Clearing $table...\n";
 719                 $tableName = $this->db->tableName( $table );
 720                 $this->db->query( 'TRUNCATE $tableName' );
 721         }
 722
 723         /**
 724          * Rename a given image or archived image file to the converted filename,
 725          * leaving a symlink for URL compatibility.
 726          *
 727          * @param string $oldname pre-conversion filename
 728          * @param string $basename pre-conversion base filename for dir hashing, if an archive
 729          * @access private
 730          */
 731         function renameFile( $oldname, $subdirCallback='wfImageDir', $basename=null ) {
 732                 $newname = $this->conv( $oldname );
 733                 if( $newname == $oldname ) {
 734                         // No need to rename; another field triggered this row.
 735                         return;
 736                 }
 737
 738                 if( is_null( $basename ) ) $basename = $oldname;
 739                 $ubasename = $this->conv( $basename );
 740                 $oldpath = call_user_func( $subdirCallback, $basename ) . '/' . $oldname;
 741                 $newpath = call_user_func( $subdirCallback, $ubasename ) . '/' . $newname;
 742
 743                 $this->log( "$oldpath -> $newpath" );
 744                 if( rename( $oldpath, $newpath ) ) {
 745                         $relpath = $this->relativize( $newpath, dirname( $oldpath ) );
 746                         if( !symlink( $relpath, $oldpath ) ) {
 747                                 $this->log( "... symlink failed!" );
 748                         }
 749                 } else {
 750                         $this->log( "... rename failed!" );
 751                 }
 752         }
 753
 754         /**
 755          * Generate a relative path name to the given file.
 756          * Assumes Unix-style paths, separators, and semantics.
 757          *
 758          * @param string $path Absolute destination path including target filename
 759          * @param string $from Absolute source path, directory only
 760          * @return string
 761          * @access private
 762          * @static
 763          */
 764         function relativize( $path, $from ) {
 765                 $pieces  = explode( '/', dirname( $path ) );
 766                 $against = explode( '/', $from );
 767
 768                 // Trim off common prefix
 769                 while( count( $pieces ) && count( $against )
 770                         && $pieces[0] == $against[0] ) {
 771                         array_shift( $pieces );
 772                         array_shift( $against );
 773                 }
 774
 775                 // relative dots to bump us to the parent
 776                 while( count( $against ) ) {
 777                         array_unshift( $pieces, '..' );
 778                         array_shift( $against );
 779                 }
 780
 781                 array_push( $pieces, basename( $path ) );
 782
 783                 return implode( '/', $pieces );
 784         }
 785
 786         function upgradeOldImage() {
 787                 $tabledef = <<<END
 788 CREATE TABLE $1 (
 789   -- Base filename: key to image.img_name
 790   oi_name varchar(255) binary NOT NULL default '',
 791
 792   -- Filename of the archived file.
 793   -- This is generally a timestamp and '!' prepended to the base name.
 794   oi_archive_name varchar(255) binary NOT NULL default '',
 795
 796   -- Other fields as in image...
 797   oi_size int(8) unsigned NOT NULL default 0,
 798   oi_width int(5) NOT NULL default 0,
 799   oi_height int(5) NOT NULL default 0,
 800   oi_bits int(3) NOT NULL default 0,
 801   oi_description tinyblob NOT NULL default '',
 802   oi_user int(5) unsigned NOT NULL default '0',
 803   oi_user_text varchar(255) binary NOT NULL default '',
 804   oi_timestamp char(14) binary NOT NULL default '',
 805
 806   INDEX oi_name (oi_name(10))
 807
 808 ) TYPE=InnoDB;
 809 END;
 810                 $fields = array(
 811                         'oi_name'         => MW_UPGRADE_ENCODE,
 812                         'oi_archive_name' => MW_UPGRADE_ENCODE,
 813                         'oi_size'         => MW_UPGRADE_COPY,
 814                         'oi_width'        => MW_UPGRADE_CALLBACK,
 815                         'oi_height'       => MW_UPGRADE_CALLBACK,
 816                         'oi_bits'         => MW_UPGRADE_CALLBACK,
 817                         'oi_description'  => MW_UPGRADE_ENCODE,
 818                         'oi_user'         => MW_UPGRADE_COPY,
 819                         'oi_user_text'    => MW_UPGRADE_ENCODE,
 820                         'oi_timestamp'    => MW_UPGRADE_COPY );
 821                 $this->copyTable( 'oldimage', $tabledef, $fields,
 822                         array( &$this, 'oldimageCallback' ) );
 823         }
 824
 825         function oldimageCallback( $row, $copy ) {
 826                 // Fill in the new image info fields
 827                 $info = $this->imageInfo( $row->oi_archive_name, 'wfImageArchiveDir', $row->oi_name );
 828                 $copy['oi_width' ] = $info['width' ];
 829                 $copy['oi_height'] = $info['height'];
 830                 $copy['oi_bits'  ] = $info['bits'  ];
 831
 832                 // If doing UTF8 conversion the file must be renamed
 833                 $this->renameFile( $row->oi_archive_name, 'wfImageArchiveDir', $row->oi_name );
 834
 835                 return $copy;
 836         }
 837
 838
 839         function upgradeWatchlist() {
 840                 $fname = 'FiveUpgrade::upgradeWatchlist';
 841                 $chunksize = 100;
 842
 843                 extract( $this->dbw->tableNames( 'watchlist', 'watchlist_temp' ) );
 844
 845                 $this->log( 'Migrating watchlist table to watchlist_temp...' );
 846                 $this->dbw->query(
 847 "CREATE TABLE $watchlist_temp (
 848   -- Key to user_id
 849   wl_user int(5) unsigned NOT NULL,
 850
 851   -- Key to page_namespace/page_title
 852   -- Note that users may watch patches which do not exist yet,
 853   -- or existed in the past but have been deleted.
 854   wl_namespace int NOT NULL default '0',
 855   wl_title varchar(255) binary NOT NULL default '',
 856
 857   -- Timestamp when user was last sent a notification e-mail;
 858   -- cleared when the user visits the page.
 859   -- FIXME: add proper null support etc
 860   wl_notificationtimestamp varchar(14) binary NOT NULL default '0',
 861
 862   UNIQUE KEY (wl_user, wl_namespace, wl_title),
 863   KEY namespace_title (wl_namespace,wl_title)
 864
 865 ) TYPE=InnoDB;", $fname );
 866
 867                 // Fix encoding for Latin-1 upgrades, add some fields,
 868                 // and double article to article+talk pairs
 869                 $numwatched = $this->dbw->selectField( 'watchlist', 'count(*)', '', $fname );
 870
 871                 $this->setChunkScale( $chunksize, $numwatched * 2, 'watchlist_temp', $fname );
 872                 $result = $this->dbr->select( 'watchlist',
 873                         array(
 874                                 'wl_user',
 875                                 'wl_namespace',
 876                                 'wl_title' ),
 877                         '',
 878                         $fname );
 879
 880                 $add = array();
 881                 while( $row = $this->dbr->fetchObject( $result ) ) {
 882                         $now = $this->dbw->timestamp();
 883                         $add[] = array(
 884                                 'wl_user'      =>                        $row->wl_user,
 885                                 'wl_namespace' => Namespace::getSubject( $row->wl_namespace ),
 886                                 'wl_title'     =>           $this->conv( $row->wl_title ),
 887                                 'wl_notificationtimestamp' =>            '0' );
 888                         $this->addChunk( $add );
 889
 890                         $add[] = array(
 891                                 'wl_user'      =>                        $row->wl_user,
 892                                 'wl_namespace' =>    Namespace::getTalk( $row->wl_namespace ),
 893                                 'wl_title'     =>           $this->conv( $row->wl_title ),
 894                                 'wl_notificationtimestamp' =>            '0' );
 895                         $this->addChunk( $add );
 896                 }
 897                 $this->lastChunk( $add );
 898                 $this->dbr->freeResult( $result );
 899
 900                 $this->log( 'Done converting watchlist.' );
 901                 $this->cleanupSwaps[] = 'watchlist';
 902         }
 903
 904         function upgradeLogging() {
 905                 $tabledef = <<<END
 906 CREATE TABLE $1 (
 907   -- Symbolic keys for the general log type and the action type
 908   -- within the log. The output format will be controlled by the
 909   -- action field, but only the type controls categorization.
 910   log_type char(10) NOT NULL default '',
 911   log_action char(10) NOT NULL default '',
 912
 913   -- Timestamp. Duh.
 914   log_timestamp char(14) NOT NULL default '19700101000000',
 915
 916   -- The user who performed this action; key to user_id
 917   log_user int unsigned NOT NULL default 0,
 918
 919   -- Key to the page affected. Where a user is the target,
 920   -- this will point to the user page.
 921   log_namespace int NOT NULL default 0,
 922   log_title varchar(255) binary NOT NULL default '',
 923
 924   -- Freeform text. Interpreted as edit history comments.
 925   log_comment varchar(255) NOT NULL default '',
 926
 927   -- LF separated list of miscellaneous parameters
 928   log_params blob NOT NULL default '',
 929
 930   KEY type_time (log_type, log_timestamp),
 931   KEY user_time (log_user, log_timestamp),
 932   KEY page_time (log_namespace, log_title, log_timestamp)
 933
 934 ) TYPE=InnoDB
 935 END;
 936                 $fields = array(
 937                         'log_type'      => MW_UPGRADE_COPY,
 938                         'log_action'    => MW_UPGRADE_COPY,
 939                         'log_timestamp' => MW_UPGRADE_COPY,
 940                         'log_user'      => MW_UPGRADE_COPY,
 941                         'log_namespace' => MW_UPGRADE_COPY,
 942                         'log_title'     => MW_UPGRADE_ENCODE,
 943                         'log_comment'   => MW_UPGRADE_ENCODE,
 944                         'log_params'    => MW_UPGRADE_ENCODE );
 945                 $this->copyTable( 'archive', $tabledef, $fields );
 946         }
 947
 948         function upgradeArchive() {
 949                 $tabledef = <<<END
 950 CREATE TABLE $1 (
 951   ar_namespace int NOT NULL default '0',
 952   ar_title varchar(255) binary NOT NULL default '',
 953   ar_text mediumblob NOT NULL default '',
 954
 955   ar_comment tinyblob NOT NULL default '',
 956   ar_user int(5) unsigned NOT NULL default '0',
 957   ar_user_text varchar(255) binary NOT NULL,
 958   ar_timestamp char(14) binary NOT NULL default '',
 959   ar_minor_edit tinyint(1) NOT NULL default '0',
 960
 961   ar_flags tinyblob NOT NULL default '',
 962
 963   ar_rev_id int(8) unsigned,
 964   ar_text_id int(8) unsigned,
 965
 966   KEY name_title_timestamp (ar_namespace,ar_title,ar_timestamp)
 967
 968 ) TYPE=InnoDB
 969 END;
 970                 $fields = array(
 971                         'ar_namespace'  => MW_UPGRADE_COPY,
 972                         'ar_title'      => MW_UPGRADE_ENCODE,
 973                         'ar_text'       => MW_UPGRADE_COPY,
 974                         'ar_comment'    => MW_UPGRADE_ENCODE,
 975                         'ar_user'       => MW_UPGRADE_COPY,
 976                         'ar_user_text'  => MW_UPGRADE_ENCODE,
 977                         'ar_timestamp'  => MW_UPGRADE_COPY,
 978                         'ar_minor_edit' => MW_UPGRADE_COPY,
 979                         'ar_flags'      => MW_UPGRADE_COPY,
 980                         'ar_rev_id'     => MW_UPGRADE_NULL,
 981                         'ar_text_id'    => MW_UPGRADE_NULL );
 982                 $this->copyTable( 'archive', $tabledef, $fields );
 983         }
 984
 985         function upgradeImagelinks() {
 986                 global $wgUseLatin1;
 987                 if( $wgUseLatin1 ) {
 988                         $tabledef = <<<END
 989 CREATE TABLE $1 (
 990   -- Key to page_id of the page containing the image / media link.
 991   il_from int(8) unsigned NOT NULL default '0',
 992
 993   -- Filename of target image.
 994   -- This is also the page_title of the file's description page;
 995   -- all such pages are in namespace 6 (NS_IMAGE).
 996   il_to varchar(255) binary NOT NULL default '',
 997
 998   UNIQUE KEY il_from(il_from,il_to),
 999   KEY (il_to)
1000
1001 ) TYPE=InnoDB
1002 END;
1003                         $fields = array(
1004                                 'il_from' => MW_UPGRADE_COPY,
1005                                 'il_to'   => MW_UPGRADE_ENCODE );
1006                         $this->copyTable( 'imagelinks', $tabledef, $fields );
1007                 }
1008         }
1009
1010         function upgradeCategorylinks() {
1011                 global $wgUseLatin1;
1012                 if( $wgUseLatin1 ) {
1013                         $tabledef = <<<END
1014 CREATE TABLE $1 (
1015   cl_from int(8) unsigned NOT NULL default '0',
1016   cl_to varchar(255) binary NOT NULL default '',
1017   cl_sortkey varchar(86) binary NOT NULL default '',
1018   cl_timestamp timestamp NOT NULL,
1019
1020   UNIQUE KEY cl_from(cl_from,cl_to),
1021   KEY cl_sortkey(cl_to,cl_sortkey),
1022   KEY cl_timestamp(cl_to,cl_timestamp)
1023 ) TYPE=InnoDB
1024 END;
1025                         $fields = array(
1026                                 'cl_from'      => MW_UPGRADE_COPY,
1027                                 'cl_to'        => MW_UPGRADE_ENCODE,
1028                                 'cl_sortkey'   => MW_UPGRADE_ENCODE,
1029                                 'cl_timestamp' => MW_UPGRADE_COPY );
1030                         $this->copyTable( 'categorylinks', $tabledef, $fields );
1031                 }
1032         }
1033
1034         /**
1035          * Rename all our temporary tables into final place.
1036          * We've left things in place so a read-only wiki can continue running
1037          * on the old code during all this.
1038          */
1039         function upgradeCleanup() {
1040                 $this->renameTable( 'old', 'text' );
1041
1042                 foreach( $this->cleanupSwaps as $table ) {
1043                         $this->swap( $table );
1044                 }
1045         }
1046
1047         function renameTable( $from, $to ) {
1048                 $this->log( 'Renaming $from to $to...' );
1049
1050                 $fromtable = $this->dbw->tableName( $from );
1051                 $totable   = $this->dbw->tableName( $to );
1052                 $this->dbw->query( "ALTER TABLE $fromtable RENAME TO $totable" );
1053         }
1054
1055         function swap( $base ) {
1056                 $this->renameTable( $base, "{$base}_old" );
1057                 $this->renameTable( "{$base}_temp", $base );
1058         }
1059
1060 }
1061
1062 $upgrade = new FiveUpgrade();
1063 $step = isset( $options['step'] ) ? $options['step'] : null;
1064 $upgrade->upgrade( $step );
1065
1066 ?>