7 require_once( 'updaters.inc' );
9 define( 'MW_UPGRADE_COPY', false );
10 define( 'MW_UPGRADE_ENCODE', true );
11 define( 'MW_UPGRADE_NULL', null );
12 define( 'MW_UPGRADE_CALLBACK', null ); // for self-documentation only
15 * @ingroup Maintenance
18 function FiveUpgrade() {
19 $this->conversionTables = $this->prepareWindows1252();
21 $this->loadBalancers = array();
22 $this->dbw = wfGetDB( DB_MASTER );
23 $this->dbr = $this->streamConnection();
25 $this->cleanupSwaps = array();
26 $this->emailAuth = false; # don't preauthenticate emails
27 $this->maxLag = 10; # if slaves are lagged more than 10 secs, wait
30 function doing( $step ) {
31 return is_null( $this->step ) || $step == $this->step;
34 function upgrade( $step ) {
51 foreach ( $tables as $table ) {
52 if ( $this->doing( $table ) ) {
53 $method = 'upgrade' . ucfirst( $table );
58 if ( $this->doing( 'cleanup' ) ) {
59 $this->upgradeCleanup();
65 * Open a connection to the master server with the admin rights.
69 function newConnection() {
70 $lb = wfGetLBFactory()->newMainLB();
71 $db = $lb->getConnection( DB_MASTER );
73 $this->loadBalancers[] = $lb;
78 * Commit transactions and close the connections when we're done...
81 foreach ( $this->loadBalancers as $lb ) {
82 $lb->commitMasterChanges();
88 * Open a second connection to the master server, with buffering off.
89 * This will let us stream large datasets in and write in chunks on the
94 function streamConnection() {
98 $db = $this->newConnection();
99 $db->bufferResults( false );
100 if ( $wgDBtype == 'mysql' ) {
101 $db->query( "SET net_read_timeout=$timeout" );
102 $db->query( "SET net_write_timeout=$timeout" );
108 * Prepare a conversion array for converting Windows Code Page 1252 to
109 * UTF-8. This should provide proper conversion of text that was miscoded
110 * as Windows-1252 by naughty user-agents, and doesn't rely on an outside
116 function prepareWindows1252() {
118 # http://www.unicode.org/Public/MAPPINGS/VENDORS/MICSFT/WINDOWS/CP1252.TXT
119 static $cp1252 = array(
120 0x80 => 0x20AC, # EURO SIGN
121 0x81 => 0xFFFD, # REPLACEMENT CHARACTER (no mapping)
122 0x82 => 0x201A, # SINGLE LOW-9 QUOTATION MARK
123 0x83 => 0x0192, # LATIN SMALL LETTER F WITH HOOK
124 0x84 => 0x201E, # DOUBLE LOW-9 QUOTATION MARK
125 0x85 => 0x2026, # HORIZONTAL ELLIPSIS
126 0x86 => 0x2020, # DAGGER
127 0x87 => 0x2021, # DOUBLE DAGGER
128 0x88 => 0x02C6, # MODIFIER LETTER CIRCUMFLEX ACCENT
129 0x89 => 0x2030, # PER MILLE SIGN
130 0x8A => 0x0160, # LATIN CAPITAL LETTER S WITH CARON
131 0x8B => 0x2039, # SINGLE LEFT-POINTING ANGLE QUOTATION MARK
132 0x8C => 0x0152, # LATIN CAPITAL LIGATURE OE
133 0x8D => 0xFFFD, # REPLACEMENT CHARACTER (no mapping)
134 0x8E => 0x017D, # LATIN CAPITAL LETTER Z WITH CARON
135 0x8F => 0xFFFD, # REPLACEMENT CHARACTER (no mapping)
136 0x90 => 0xFFFD, # REPLACEMENT CHARACTER (no mapping)
137 0x91 => 0x2018, # LEFT SINGLE QUOTATION MARK
138 0x92 => 0x2019, # RIGHT SINGLE QUOTATION MARK
139 0x93 => 0x201C, # LEFT DOUBLE QUOTATION MARK
140 0x94 => 0x201D, # RIGHT DOUBLE QUOTATION MARK
141 0x95 => 0x2022, # BULLET
142 0x96 => 0x2013, # EN DASH
143 0x97 => 0x2014, # EM DASH
144 0x98 => 0x02DC, # SMALL TILDE
145 0x99 => 0x2122, # TRADE MARK SIGN
146 0x9A => 0x0161, # LATIN SMALL LETTER S WITH CARON
147 0x9B => 0x203A, # SINGLE RIGHT-POINTING ANGLE QUOTATION MARK
148 0x9C => 0x0153, # LATIN SMALL LIGATURE OE
149 0x9D => 0xFFFD, # REPLACEMENT CHARACTER (no mapping)
150 0x9E => 0x017E, # LATIN SMALL LETTER Z WITH CARON
151 0x9F => 0x0178, # LATIN CAPITAL LETTER Y WITH DIAERESIS
154 for ( $i = 0; $i < 0x100; $i++ ) {
155 $unicode = isset( $cp1252[$i] ) ? $cp1252[$i] : $i;
156 $pairs[chr( $i )] = codepointToUtf8( $unicode );
162 * Convert from 8-bit Windows-1252 to UTF-8 if necessary.
163 * @param string $text
167 function conv( $text ) {
169 return is_null( $text )
172 ? strtr( $text, $this->conversionTables )
177 * Dump timestamp and message to output
178 * @param string $message
181 function log( $message ) {
182 echo wfWikiID() . ' ' . wfTimestamp( TS_DB ) . ': ' . $message . "\n";
187 * Initialize the chunked-insert system.
188 * Rows will be inserted in chunks of the given number, rather
189 * than in a giant INSERT...SELECT query, to keep the serialized
190 * MySQL database replication from getting hung up. This way other
191 * things can be going on during conversion without waiting for
192 * slaves to catch up as badly.
194 * @param int $chunksize Number of rows to insert at once
195 * @param int $final Total expected number of rows / id of last row,
196 * used for progress reports.
197 * @param string $table to insert on
198 * @param string $fname function name to report in SQL
201 function setChunkScale( $chunksize, $final, $table, $fname ) {
202 $this->chunkSize = $chunksize;
203 $this->chunkFinal = $final;
204 $this->chunkCount = 0;
205 $this->chunkStartTime = wfTime();
206 $this->chunkOptions = array( 'IGNORE' );
207 $this->chunkTable = $table;
208 $this->chunkFunction = $fname;
212 * Chunked inserts: perform an insert if we've reached the chunk limit.
213 * Prints a progress report with estimated completion time.
214 * @param array &$chunk -- This will be emptied if an insert is done.
215 * @param int $key A key identifier to use in progress estimation in
216 * place of the number of rows inserted. Use this if
217 * you provided a max key number instead of a count
218 * as the final chunk number in setChunkScale()
221 function addChunk( &$chunk, $key = null ) {
222 if ( count( $chunk ) >= $this->chunkSize ) {
223 $this->insertChunk( $chunk );
225 $this->chunkCount += count( $chunk );
227 $delta = $now - $this->chunkStartTime;
228 $rate = $this->chunkCount / $delta;
230 if ( is_null( $key ) ) {
231 $completed = $this->chunkCount;
235 $portion = $completed / $this->chunkFinal;
237 $estimatedTotalTime = $delta / $portion;
238 $eta = $this->chunkStartTime + $estimatedTotalTime;
240 printf( "%s: %6.2f%% done on %s; ETA %s [%d/%d] %.2f/sec\n",
241 wfTimestamp( TS_DB, intval( $now ) ),
244 wfTimestamp( TS_DB, intval( $eta ) ),
255 * Chunked inserts: perform an insert unconditionally, at the end, and log.
256 * @param array &$chunk -- This will be emptied if an insert is done.
259 function lastChunk( &$chunk ) {
260 $n = count( $chunk );
262 $this->insertChunk( $chunk );
264 $this->log( "100.00% done on $this->chunkTable (last chunk $n rows)." );
268 * Chunked inserts: perform an insert.
269 * @param array &$chunk -- This will be emptied if an insert is done.
272 function insertChunk( &$chunk ) {
273 // Give slaves a chance to catch up
274 wfWaitForSlaves( $this->maxLag );
275 $this->dbw->insert( $this->chunkTable, $chunk, $this->chunkFunction, $this->chunkOptions );
280 * Copy and transcode a table to table_temp.
281 * @param string $name Base name of the source table
282 * @param string $tabledef CREATE TABLE definition, w/ $1 for the name
283 * @param array $fields set of destination fields to these constants:
284 * MW_UPGRADE_COPY - straight copy
285 * MW_UPGRADE_ENCODE - for old Latin1 wikis, conv to UTF-8
286 * MW_UPGRADE_NULL - just put NULL
287 * @param callable $callback An optional callback to modify the data
288 * or perform other processing. Func should be
289 * ( object $row, array $copy ) and return $copy
292 function copyTable( $name, $tabledef, $fields, $callback = null ) {
293 $fname = 'FiveUpgrade::copyTable';
295 $name_temp = $name . '_temp';
296 $this->log( "Migrating $name table to $name_temp..." );
298 $table_temp = $this->dbw->tableName( $name_temp );
300 // Create temporary table; we're going to copy everything in there,
301 // then at the end rename the final tables into place.
302 $def = str_replace( '$1', $table_temp, $tabledef );
303 $this->dbw->query( $def, $fname );
305 $numRecords = $this->dbw->selectField( $name, 'COUNT(*)', '', $fname );
306 $this->setChunkScale( 100, $numRecords, $name_temp, $fname );
308 // Pull all records from the second, streaming database connection.
309 $sourceFields = array_keys( array_filter( $fields,
310 create_function( '$x', 'return $x !== MW_UPGRADE_NULL;' ) ) );
311 $result = $this->dbr->select( $name,
317 while ( $row = $this->dbr->fetchObject( $result ) ) {
319 foreach ( $fields as $field => $source ) {
320 if ( $source === MW_UPGRADE_COPY ) {
321 $copy[$field] = $row->$field;
322 } elseif ( $source === MW_UPGRADE_ENCODE ) {
323 $copy[$field] = $this->conv( $row->$field );
324 } elseif ( $source === MW_UPGRADE_NULL ) {
325 $copy[$field] = null;
327 $this->log( "Unknown field copy type: $field => $source" );
330 if ( is_callable( $callback ) ) {
331 $copy = call_user_func( $callback, $row, $copy );
334 $this->addChunk( $add );
336 $this->lastChunk( $add );
338 $this->log( "Done converting $name." );
339 $this->cleanupSwaps[] = $name;
342 function upgradePage() {
343 $fname = "FiveUpgrade::upgradePage";
346 if ( $this->dbw->tableExists( 'page' ) ) {
347 $this->log( 'Page table already exists; aborting.' );
351 $this->log( "Checking cur table for unique title index and applying if necessary" );
354 $this->log( "...converting from cur/old to page/revision/text DB structure." );
356 list ( $cur, $old, $page, $revision, $text ) = $this->dbw->tableNamesN( 'cur', 'old', 'page', 'revision', 'text' );
358 $this->log( "Creating page and revision tables..." );
359 $this->dbw->query( "CREATE TABLE $page (
360 page_id int(8) unsigned NOT NULL auto_increment,
361 page_namespace int NOT NULL,
362 page_title varchar(255) binary NOT NULL,
363 page_restrictions tinyblob NOT NULL default '',
364 page_counter bigint(20) unsigned NOT NULL default '0',
365 page_is_redirect tinyint(1) unsigned NOT NULL default '0',
366 page_is_new tinyint(1) unsigned NOT NULL default '0',
367 page_random real unsigned NOT NULL,
368 page_touched char(14) binary NOT NULL default '',
369 page_latest int(8) unsigned NOT NULL,
370 page_len int(8) unsigned NOT NULL,
372 PRIMARY KEY page_id (page_id),
373 UNIQUE INDEX name_title (page_namespace,page_title),
376 ) TYPE=InnoDB", $fname );
377 $this->dbw->query( "CREATE TABLE $revision (
378 rev_id int(8) unsigned NOT NULL auto_increment,
379 rev_page int(8) unsigned NOT NULL,
380 rev_text_id int(8) unsigned NOT NULL,
381 rev_comment tinyblob NOT NULL default '',
382 rev_user int(5) unsigned NOT NULL default '0',
383 rev_user_text varchar(255) binary NOT NULL default '',
384 rev_timestamp char(14) binary NOT NULL default '',
385 rev_minor_edit tinyint(1) unsigned NOT NULL default '0',
386 rev_deleted tinyint(1) unsigned NOT NULL default '0',
388 PRIMARY KEY rev_page_id (rev_page, rev_id),
389 UNIQUE INDEX rev_id (rev_id),
390 INDEX rev_timestamp (rev_timestamp),
391 INDEX page_timestamp (rev_page,rev_timestamp),
392 INDEX user_timestamp (rev_user,rev_timestamp),
393 INDEX usertext_timestamp (rev_user_text,rev_timestamp)
394 ) TYPE=InnoDB", $fname );
396 $maxold = intval( $this->dbw->selectField( 'old', 'max(old_id)', '', $fname ) );
397 $this->log( "Last old record is {$maxold}" );
399 global $wgLegacySchemaConversion;
400 if ( $wgLegacySchemaConversion ) {
401 // Create HistoryBlobCurStub entries.
402 // Text will be pulled from the leftover 'cur' table at runtime.
403 echo "......Moving metadata from cur; using blob references to text in cur table.\n";
404 $cur_text = "concat('O:18:\"historyblobcurstub\":1:{s:6:\"mCurId\";i:',cur_id,';}')";
405 $cur_flags = "'object'";
407 // Copy all cur text in immediately: this may take longer but avoids
408 // having to keep an extra table around.
409 echo "......Moving text from cur.\n";
410 $cur_text = 'cur_text';
414 $maxcur = $this->dbw->selectField( 'cur', 'max(cur_id)', '', $fname );
415 $this->log( "Last cur entry is $maxcur" );
418 * Copy placeholder records for each page's current version into old
419 * Don't do any conversion here; text records are converted at runtime
420 * based on the flags (and may be originally binary!) while the meta
421 * fields will be converted in the old -> rev and cur -> page steps.
423 $this->setChunkScale( $chunksize, $maxcur, 'old', $fname );
424 $result = $this->dbr->query(
425 "SELECT cur_id, cur_namespace, cur_title, $cur_text AS text, cur_comment,
426 cur_user, cur_user_text, cur_timestamp, cur_minor_edit, $cur_flags AS flags
428 ORDER BY cur_id", $fname );
430 while ( $row = $this->dbr->fetchObject( $result ) ) {
432 'old_namespace' => $row->cur_namespace,
433 'old_title' => $row->cur_title,
434 'old_text' => $row->text,
435 'old_comment' => $row->cur_comment,
436 'old_user' => $row->cur_user,
437 'old_user_text' => $row->cur_user_text,
438 'old_timestamp' => $row->cur_timestamp,
439 'old_minor_edit' => $row->cur_minor_edit,
440 'old_flags' => $row->flags );
441 $this->addChunk( $add, $row->cur_id );
443 $this->lastChunk( $add );
446 * Copy revision metadata from old into revision.
447 * We'll also do UTF-8 conversion of usernames and comments.
449 # $newmaxold = $this->dbw->selectField( 'old', 'max(old_id)', '', $fname );
450 # $this->setChunkScale( $chunksize, $newmaxold, 'revision', $fname );
451 # $countold = $this->dbw->selectField( 'old', 'count(old_id)', '', $fname );
452 $countold = $this->dbw->selectField( 'old', 'max(old_id)', '', $fname );
453 $this->setChunkScale( $chunksize, $countold, 'revision', $fname );
455 $this->log( "......Setting up revision table." );
456 $result = $this->dbr->query(
457 "SELECT old_id, cur_id, old_comment, old_user, old_user_text,
458 old_timestamp, old_minor_edit
459 FROM $old,$cur WHERE old_namespace=cur_namespace AND old_title=cur_title",
463 while ( $row = $this->dbr->fetchObject( $result ) ) {
465 'rev_id' => $row->old_id,
466 'rev_page' => $row->cur_id,
467 'rev_text_id' => $row->old_id,
468 'rev_comment' => $this->conv( $row->old_comment ),
469 'rev_user' => $row->old_user,
470 'rev_user_text' => $this->conv( $row->old_user_text ),
471 'rev_timestamp' => $row->old_timestamp,
472 'rev_minor_edit' => $row->old_minor_edit );
473 $this->addChunk( $add );
475 $this->lastChunk( $add );
479 * Copy page metadata from cur into page.
480 * We'll also do UTF-8 conversion of titles.
482 $this->log( "......Setting up page table." );
483 $this->setChunkScale( $chunksize, $maxcur, 'page', $fname );
484 $result = $this->dbr->query( "
485 SELECT cur_id, cur_namespace, cur_title, cur_restrictions, cur_counter, cur_is_redirect, cur_is_new,
486 cur_random, cur_touched, rev_id, LENGTH(cur_text) AS len
488 WHERE cur_id=rev_page AND rev_timestamp=cur_timestamp AND rev_id > {$maxold}
489 ORDER BY cur_id", $fname );
491 while ( $row = $this->dbr->fetchObject( $result ) ) {
493 'page_id' => $row->cur_id,
494 'page_namespace' => $row->cur_namespace,
495 'page_title' => $this->conv( $row->cur_title ),
496 'page_restrictions' => $row->cur_restrictions,
497 'page_counter' => $row->cur_counter,
498 'page_is_redirect' => $row->cur_is_redirect,
499 'page_is_new' => $row->cur_is_new,
500 'page_random' => $row->cur_random,
501 'page_touched' => $this->dbw->timestamp(),
502 'page_latest' => $row->rev_id,
503 'page_len' => $row->len );
504 # $this->addChunk( $add, $row->cur_id );
505 $this->addChunk( $add );
507 $this->lastChunk( $add );
509 $this->log( "...done with cur/old -> page/revision." );
512 function upgradeLinks() {
513 $fname = 'FiveUpgrade::upgradeLinks';
515 list ( $links, $brokenlinks, $pagelinks, $cur ) = $this->dbw->tableNamesN( 'links', 'brokenlinks', 'pagelinks', 'cur' );
517 $this->log( 'Checking for interwiki table change in case of bogus items...' );
518 if ( $this->dbw->fieldExists( 'interwiki', 'iw_trans' ) ) {
519 $this->log( 'interwiki has iw_trans.' );
522 $this->log( 'adding iw_trans...' );
523 $this->dbw->sourceFile( $IP . '/maintenance/archives/patch-interwiki-trans.sql' );
524 $this->log( 'added iw_trans.' );
527 $this->log( 'Creating pagelinks table...' );
529 CREATE TABLE $pagelinks (
530 -- Key to the page_id of the page containing the link.
531 pl_from int(8) unsigned NOT NULL default '0',
533 -- Key to page_namespace/page_title of the target page.
534 -- The target page may or may not exist, and due to renames
535 -- and deletions may refer to different page records as time
537 pl_namespace int NOT NULL default '0',
538 pl_title varchar(255) binary NOT NULL default '',
540 UNIQUE KEY pl_from(pl_from,pl_namespace,pl_title),
541 KEY (pl_namespace,pl_title)
545 $this->log( 'Importing live links -> pagelinks' );
546 $nlinks = $this->dbw->selectField( 'links', 'count(*)', '', $fname );
548 $this->setChunkScale( $chunksize, $nlinks, 'pagelinks', $fname );
549 $result = $this->dbr->query( "
550 SELECT l_from,cur_namespace,cur_title
552 WHERE l_to=cur_id", $fname );
554 while ( $row = $this->dbr->fetchObject( $result ) ) {
556 'pl_from' => $row->l_from,
557 'pl_namespace' => $row->cur_namespace,
558 'pl_title' => $this->conv( $row->cur_title ) );
559 $this->addChunk( $add );
561 $this->lastChunk( $add );
563 $this->log( 'no links!' );
566 $this->log( 'Importing brokenlinks -> pagelinks' );
567 $nbrokenlinks = $this->dbw->selectField( 'brokenlinks', 'count(*)', '', $fname );
568 if ( $nbrokenlinks ) {
569 $this->setChunkScale( $chunksize, $nbrokenlinks, 'pagelinks', $fname );
570 $result = $this->dbr->query(
571 "SELECT bl_from, bl_to FROM $brokenlinks",
574 while ( $row = $this->dbr->fetchObject( $result ) ) {
575 $pagename = $this->conv( $row->bl_to );
576 $title = Title::newFromText( $pagename );
577 if ( is_null( $title ) ) {
578 $this->log( "** invalid brokenlink: $row->bl_from -> '$pagename' (converted from '$row->bl_to')" );
581 'pl_from' => $row->bl_from,
582 'pl_namespace' => $title->getNamespace(),
583 'pl_title' => $title->getDBkey() );
584 $this->addChunk( $add );
587 $this->lastChunk( $add );
589 $this->log( 'no brokenlinks!' );
592 $this->log( 'Done with links.' );
595 function upgradeUser() {
596 // Apply unique index, if necessary:
597 $duper = new UserDupes( $this->dbw );
598 if ( $duper->hasUniqueIndex() ) {
599 $this->log( "Already have unique user_name index." );
601 $this->log( "Clearing user duplicates..." );
602 if ( !$duper->clearDupes() ) {
603 $this->log( "WARNING: Duplicate user accounts, may explode!" );
609 user_id int(5) unsigned NOT NULL auto_increment,
610 user_name varchar(255) binary NOT NULL default '',
611 user_real_name varchar(255) binary NOT NULL default '',
612 user_password tinyblob NOT NULL default '',
613 user_newpassword tinyblob NOT NULL default '',
614 user_email tinytext NOT NULL default '',
615 user_options blob NOT NULL default '',
616 user_touched char(14) binary NOT NULL default '',
617 user_token char(32) binary NOT NULL default '',
618 user_email_authenticated CHAR(14) BINARY,
619 user_email_token CHAR(32) BINARY,
620 user_email_token_expires CHAR(14) BINARY,
622 PRIMARY KEY user_id (user_id),
623 UNIQUE INDEX user_name (user_name),
624 INDEX (user_email_token)
629 'user_id' => MW_UPGRADE_COPY,
630 'user_name' => MW_UPGRADE_ENCODE,
631 'user_real_name' => MW_UPGRADE_ENCODE,
632 'user_password' => MW_UPGRADE_COPY,
633 'user_newpassword' => MW_UPGRADE_COPY,
634 'user_email' => MW_UPGRADE_ENCODE,
635 'user_options' => MW_UPGRADE_ENCODE,
636 'user_touched' => MW_UPGRADE_CALLBACK,
637 'user_token' => MW_UPGRADE_COPY,
638 'user_email_authenticated' => MW_UPGRADE_CALLBACK,
639 'user_email_token' => MW_UPGRADE_NULL,
640 'user_email_token_expires' => MW_UPGRADE_NULL );
641 $this->copyTable( 'user', $tabledef, $fields,
642 array( &$this, 'userCallback' ) );
645 function userCallback( $row, $copy ) {
646 $now = $this->dbw->timestamp();
647 $copy['user_touched'] = $now;
648 $copy['user_email_authenticated'] = $this->emailAuth ? $now : null;
652 function upgradeImage() {
655 img_name varchar(255) binary NOT NULL default '',
656 img_size int(8) unsigned NOT NULL default '0',
657 img_width int(5) NOT NULL default '0',
658 img_height int(5) NOT NULL default '0',
659 img_metadata mediumblob NOT NULL,
660 img_bits int(3) NOT NULL default '0',
661 img_media_type ENUM("UNKNOWN", "BITMAP", "DRAWING", "AUDIO", "VIDEO", "MULTIMEDIA", "OFFICE", "TEXT", "EXECUTABLE", "ARCHIVE") default NULL,
662 img_major_mime ENUM("unknown", "application", "audio", "image", "text", "video", "message", "model", "multipart") NOT NULL default "unknown",
663 img_minor_mime varchar(32) NOT NULL default "unknown",
664 img_description tinyblob NOT NULL default '',
665 img_user int(5) unsigned NOT NULL default '0',
666 img_user_text varchar(255) binary NOT NULL default '',
667 img_timestamp char(14) binary NOT NULL default '',
669 PRIMARY KEY img_name (img_name),
670 INDEX img_size (img_size),
671 INDEX img_timestamp (img_timestamp)
675 'img_name' => MW_UPGRADE_ENCODE,
676 'img_size' => MW_UPGRADE_COPY,
677 'img_width' => MW_UPGRADE_CALLBACK,
678 'img_height' => MW_UPGRADE_CALLBACK,
679 'img_metadata' => MW_UPGRADE_CALLBACK,
680 'img_bits' => MW_UPGRADE_CALLBACK,
681 'img_media_type' => MW_UPGRADE_CALLBACK,
682 'img_major_mime' => MW_UPGRADE_CALLBACK,
683 'img_minor_mime' => MW_UPGRADE_CALLBACK,
684 'img_description' => MW_UPGRADE_ENCODE,
685 'img_user' => MW_UPGRADE_COPY,
686 'img_user_text' => MW_UPGRADE_ENCODE,
687 'img_timestamp' => MW_UPGRADE_COPY );
688 $this->copyTable( 'image', $tabledef, $fields,
689 array( &$this, 'imageCallback' ) );
692 function imageCallback( $row, $copy ) {
694 if ( !isset( $options['noimage'] ) ) {
695 // Fill in the new image info fields
696 $info = $this->imageInfo( $row->img_name );
698 $copy['img_width' ] = $info['width'];
699 $copy['img_height' ] = $info['height'];
700 $copy['img_metadata' ] = ""; // loaded on-demand
701 $copy['img_bits' ] = $info['bits'];
702 $copy['img_media_type'] = $info['media'];
703 $copy['img_major_mime'] = $info['major'];
704 $copy['img_minor_mime'] = $info['minor'];
707 // If doing UTF8 conversion the file must be renamed
708 $this->renameFile( $row->img_name, 'wfImageDir' );
713 function imageInfo( $filename ) {
722 $magic = MimeMagic::singleton();
723 $mime = $magic->guessMimeType( $filename, true );
724 list( $info['major'], $info['minor'] ) = explode( '/', $mime );
726 $info['media'] = $magic->getMediaType( $filename, $mime );
728 $image = UnregisteredLocalFile::newFromPath( $filename, $mime );
730 $info['width'] = $image->getWidth();
731 $info['height'] = $image->getHeight();
733 $gis = $image->getImageSize( $filename );
734 if ( isset( $gis['bits'] ) ) {
735 $info['bits'] = $gis['bits'];
744 * @param string $table The table name to be truncated
746 function clearTable( $table ) {
747 print "Clearing $table...\n";
748 $tableName = $this->db->tableName( $table );
749 $this->db->query( "TRUNCATE $tableName" );
753 * Rename a given image or archived image file to the converted filename,
754 * leaving a symlink for URL compatibility.
756 * @param string $oldname pre-conversion filename
757 * @param string $basename pre-conversion base filename for dir hashing, if an archive
760 function renameFile( $oldname, $subdirCallback = 'wfImageDir', $basename = null ) {
761 $newname = $this->conv( $oldname );
762 if ( $newname == $oldname ) {
763 // No need to rename; another field triggered this row.
767 if ( is_null( $basename ) ) $basename = $oldname;
768 $ubasename = $this->conv( $basename );
769 $oldpath = call_user_func( $subdirCallback, $basename ) . '/' . $oldname;
770 $newpath = call_user_func( $subdirCallback, $ubasename ) . '/' . $newname;
772 $this->log( "$oldpath -> $newpath" );
773 if ( rename( $oldpath, $newpath ) ) {
774 $relpath = wfRelativePath( $newpath, dirname( $oldpath ) );
775 if ( !symlink( $relpath, $oldpath ) ) {
776 $this->log( "... symlink failed!" );
780 $this->log( "... rename failed!" );
785 function upgradeOldImage() {
788 -- Base filename: key to image.img_name
789 oi_name varchar(255) binary NOT NULL default '',
791 -- Filename of the archived file.
792 -- This is generally a timestamp and '!' prepended to the base name.
793 oi_archive_name varchar(255) binary NOT NULL default '',
795 -- Other fields as in image...
796 oi_size int(8) unsigned NOT NULL default 0,
797 oi_width int(5) NOT NULL default 0,
798 oi_height int(5) NOT NULL default 0,
799 oi_bits int(3) NOT NULL default 0,
800 oi_description tinyblob NOT NULL default '',
801 oi_user int(5) unsigned NOT NULL default '0',
802 oi_user_text varchar(255) binary NOT NULL default '',
803 oi_timestamp char(14) binary NOT NULL default '',
805 INDEX oi_name (oi_name(10))
810 'oi_name' => MW_UPGRADE_ENCODE,
811 'oi_archive_name' => MW_UPGRADE_ENCODE,
812 'oi_size' => MW_UPGRADE_COPY,
813 'oi_width' => MW_UPGRADE_CALLBACK,
814 'oi_height' => MW_UPGRADE_CALLBACK,
815 'oi_bits' => MW_UPGRADE_CALLBACK,
816 'oi_description' => MW_UPGRADE_ENCODE,
817 'oi_user' => MW_UPGRADE_COPY,
818 'oi_user_text' => MW_UPGRADE_ENCODE,
819 'oi_timestamp' => MW_UPGRADE_COPY );
820 $this->copyTable( 'oldimage', $tabledef, $fields,
821 array( &$this, 'oldimageCallback' ) );
824 function oldimageCallback( $row, $copy ) {
826 if ( !isset( $options['noimage'] ) ) {
827 // Fill in the new image info fields
828 $info = $this->imageInfo( $row->oi_archive_name, 'wfImageArchiveDir', $row->oi_name );
829 $copy['oi_width' ] = $info['width' ];
830 $copy['oi_height'] = $info['height'];
831 $copy['oi_bits' ] = $info['bits' ];
834 // If doing UTF8 conversion the file must be renamed
835 $this->renameFile( $row->oi_archive_name, 'wfImageArchiveDir', $row->oi_name );
841 function upgradeWatchlist() {
842 $fname = 'FiveUpgrade::upgradeWatchlist';
845 list ( $watchlist, $watchlist_temp ) = $this->dbw->tableNamesN( 'watchlist', 'watchlist_temp' );
847 $this->log( 'Migrating watchlist table to watchlist_temp...' );
849 "CREATE TABLE $watchlist_temp (
851 wl_user int(5) unsigned NOT NULL,
853 -- Key to page_namespace/page_title
854 -- Note that users may watch patches which do not exist yet,
855 -- or existed in the past but have been deleted.
856 wl_namespace int NOT NULL default '0',
857 wl_title varchar(255) binary NOT NULL default '',
859 -- Timestamp when user was last sent a notification e-mail;
860 -- cleared when the user visits the page.
861 -- FIXME: add proper null support etc
862 wl_notificationtimestamp varchar(14) binary NOT NULL default '0',
864 UNIQUE KEY (wl_user, wl_namespace, wl_title),
865 KEY namespace_title (wl_namespace,wl_title)
867 ) TYPE=InnoDB;", $fname );
869 // Fix encoding for Latin-1 upgrades, add some fields,
870 // and double article to article+talk pairs
871 $numwatched = $this->dbw->selectField( 'watchlist', 'count(*)', '', $fname );
873 $this->setChunkScale( $chunksize, $numwatched * 2, 'watchlist_temp', $fname );
874 $result = $this->dbr->select( 'watchlist',
883 while ( $row = $this->dbr->fetchObject( $result ) ) {
885 'wl_user' => $row->wl_user,
886 'wl_namespace' => MWNamespace::getSubject( $row->wl_namespace ),
887 'wl_title' => $this->conv( $row->wl_title ),
888 'wl_notificationtimestamp' => '0' );
889 $this->addChunk( $add );
892 'wl_user' => $row->wl_user,
893 'wl_namespace' => MWNamespace::getTalk( $row->wl_namespace ),
894 'wl_title' => $this->conv( $row->wl_title ),
895 'wl_notificationtimestamp' => '0' );
896 $this->addChunk( $add );
898 $this->lastChunk( $add );
900 $this->log( 'Done converting watchlist.' );
901 $this->cleanupSwaps[] = 'watchlist';
904 function upgradeLogging() {
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 '',
914 log_timestamp char(14) NOT NULL default '19700101000000',
916 -- The user who performed this action; key to user_id
917 log_user int unsigned NOT NULL default 0,
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 '',
924 -- Freeform text. Interpreted as edit history comments.
925 log_comment varchar(255) NOT NULL default '',
927 -- LF separated list of miscellaneous parameters
928 log_params blob NOT NULL default '',
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)
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( 'logging', $tabledef, $fields );
948 function upgradeArchive() {
951 ar_namespace int NOT NULL default '0',
952 ar_title varchar(255) binary NOT NULL default '',
953 ar_text mediumblob NOT NULL default '',
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',
961 ar_flags tinyblob NOT NULL default '',
963 ar_rev_id int(8) unsigned,
964 ar_text_id int(8) unsigned,
966 KEY name_title_timestamp (ar_namespace,ar_title,ar_timestamp)
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 );
985 function upgradeImagelinks() {
987 if ( $wgUseLatin1 ) {
990 -- Key to page_id of the page containing the image / media link.
991 il_from int(8) unsigned NOT NULL default '0',
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_FILE).
996 il_to varchar(255) binary NOT NULL default '',
998 UNIQUE KEY il_from(il_from,il_to),
1004 'il_from' => MW_UPGRADE_COPY,
1005 'il_to' => MW_UPGRADE_ENCODE );
1006 $this->copyTable( 'imagelinks', $tabledef, $fields );
1010 function upgradeCategorylinks() {
1011 global $wgUseLatin1;
1012 if ( $wgUseLatin1 ) {
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,
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)
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 );
1034 function upgradeIpblocks() {
1035 global $wgUseLatin1;
1036 if ( $wgUseLatin1 ) {
1039 ipb_id int(8) NOT NULL auto_increment,
1040 ipb_address varchar(40) binary NOT NULL default '',
1041 ipb_user int(8) unsigned NOT NULL default '0',
1042 ipb_by int(8) unsigned NOT NULL default '0',
1043 ipb_reason tinyblob NOT NULL default '',
1044 ipb_timestamp char(14) binary NOT NULL default '',
1045 ipb_auto tinyint(1) NOT NULL default '0',
1046 ipb_expiry char(14) binary NOT NULL default '',
1048 PRIMARY KEY ipb_id (ipb_id),
1049 INDEX ipb_address (ipb_address),
1050 INDEX ipb_user (ipb_user)
1055 'ipb_id' => MW_UPGRADE_COPY,
1056 'ipb_address' => MW_UPGRADE_COPY,
1057 'ipb_user' => MW_UPGRADE_COPY,
1058 'ipb_by' => MW_UPGRADE_COPY,
1059 'ipb_reason' => MW_UPGRADE_ENCODE,
1060 'ipb_timestamp' => MW_UPGRADE_COPY,
1061 'ipb_auto' => MW_UPGRADE_COPY,
1062 'ipb_expiry' => MW_UPGRADE_COPY );
1063 $this->copyTable( 'ipblocks', $tabledef, $fields );
1067 function upgradeRecentchanges() {
1068 // There's a format change in the namespace field
1071 rc_id int(8) NOT NULL auto_increment,
1072 rc_timestamp varchar(14) binary NOT NULL default '',
1073 rc_cur_time varchar(14) binary NOT NULL default '',
1075 rc_user int(10) unsigned NOT NULL default '0',
1076 rc_user_text varchar(255) binary NOT NULL default '',
1078 rc_namespace int NOT NULL default '0',
1079 rc_title varchar(255) binary NOT NULL default '',
1081 rc_comment varchar(255) binary NOT NULL default '',
1082 rc_minor tinyint(3) unsigned NOT NULL default '0',
1084 rc_bot tinyint(3) unsigned NOT NULL default '0',
1085 rc_new tinyint(3) unsigned NOT NULL default '0',
1087 rc_cur_id int(10) unsigned NOT NULL default '0',
1088 rc_this_oldid int(10) unsigned NOT NULL default '0',
1089 rc_last_oldid int(10) unsigned NOT NULL default '0',
1091 rc_type tinyint(3) unsigned NOT NULL default '0',
1092 rc_moved_to_ns tinyint(3) unsigned NOT NULL default '0',
1093 rc_moved_to_title varchar(255) binary NOT NULL default '',
1095 rc_patrolled tinyint(3) unsigned NOT NULL default '0',
1097 rc_ip char(15) NOT NULL default '',
1099 PRIMARY KEY rc_id (rc_id),
1100 INDEX rc_timestamp (rc_timestamp),
1101 INDEX rc_namespace_title (rc_namespace, rc_title),
1102 INDEX rc_cur_id (rc_cur_id),
1103 INDEX new_name_timestamp(rc_new,rc_namespace,rc_timestamp),
1109 'rc_id' => MW_UPGRADE_COPY,
1110 'rc_timestamp' => MW_UPGRADE_COPY,
1111 'rc_cur_time' => MW_UPGRADE_COPY,
1112 'rc_user' => MW_UPGRADE_COPY,
1113 'rc_user_text' => MW_UPGRADE_ENCODE,
1114 'rc_namespace' => MW_UPGRADE_COPY,
1115 'rc_title' => MW_UPGRADE_ENCODE,
1116 'rc_comment' => MW_UPGRADE_ENCODE,
1117 'rc_minor' => MW_UPGRADE_COPY,
1118 'rc_bot' => MW_UPGRADE_COPY,
1119 'rc_new' => MW_UPGRADE_COPY,
1120 'rc_cur_id' => MW_UPGRADE_COPY,
1121 'rc_this_oldid' => MW_UPGRADE_COPY,
1122 'rc_last_oldid' => MW_UPGRADE_COPY,
1123 'rc_type' => MW_UPGRADE_COPY,
1124 'rc_moved_to_ns' => MW_UPGRADE_COPY,
1125 'rc_moved_to_title' => MW_UPGRADE_ENCODE,
1126 'rc_patrolled' => MW_UPGRADE_COPY,
1127 'rc_ip' => MW_UPGRADE_COPY );
1128 $this->copyTable( 'recentchanges', $tabledef, $fields );
1131 function upgradeQuerycache() {
1132 // There's a format change in the namespace field
1135 -- A key name, generally the base name of of the special page.
1136 qc_type char(32) NOT NULL,
1138 -- Some sort of stored value. Sizes, counts...
1139 qc_value int(5) unsigned NOT NULL default '0',
1141 -- Target namespace+title
1142 qc_namespace int NOT NULL default '0',
1143 qc_title char(255) binary NOT NULL default '',
1145 KEY (qc_type,qc_value)
1150 'qc_type' => MW_UPGRADE_COPY,
1151 'qc_value' => MW_UPGRADE_COPY,
1152 'qc_namespace' => MW_UPGRADE_COPY,
1153 'qc_title' => MW_UPGRADE_ENCODE );
1154 $this->copyTable( 'querycache', $tabledef, $fields );
1158 * Check for duplicate rows in "cur" table and move duplicates entries in
1161 * This was in cleanupDupes.inc before.
1163 function checkDupes() {
1164 $dbw = wfGetDB( DB_MASTER );
1165 if ( $dbw->indexExists( 'cur', 'name_title' ) &&
1166 $dbw->indexUnique( 'cur', 'name_title' ) ) {
1167 echo wfWikiID() . ": cur table has the current unique index; no duplicate entries.\n";
1169 } elseif ( $dbw->indexExists( 'cur', 'name_title_dup_prevention' ) ) {
1170 echo wfWikiID() . ": cur table has a temporary name_title_dup_prevention unique index; no duplicate entries.\n";
1174 echo wfWikiID() . ": cur table has the old non-unique index and may have duplicate entries.\n";
1176 $dbw = wfGetDB( DB_MASTER );
1177 $cur = $dbw->tableName( 'cur' );
1178 $old = $dbw->tableName( 'old' );
1179 $dbw->query( "LOCK TABLES $cur WRITE, $old WRITE" );
1180 echo "Checking for duplicate cur table entries... (this may take a while on a large wiki)\n";
1181 $res = $dbw->query( <<<END
1182 SELECT cur_namespace,cur_title,count(*) as c,min(cur_id) as id
1184 GROUP BY cur_namespace,cur_title
1188 $n = $dbw->numRows( $res );
1189 echo "Found $n titles with duplicate entries.\n";
1191 echo "Correcting...\n";
1192 while ( $row = $dbw->fetchObject( $res ) ) {
1193 $ns = intval( $row->cur_namespace );
1194 $title = $dbw->addQuotes( $row->cur_title );
1196 # Get the first responding ID; that'll be the one we keep.
1197 $id = $dbw->selectField( 'cur', 'cur_id', array(
1198 'cur_namespace' => $row->cur_namespace,
1199 'cur_title' => $row->cur_title ) );
1201 echo "$ns:$row->cur_title (canonical ID $id)\n";
1202 if ( $id != $row->id ) {
1203 echo " ** minimum ID $row->id; ";
1204 $timeMin = $dbw->selectField( 'cur', 'cur_timestamp', array(
1205 'cur_id' => $row->id ) );
1206 $timeFirst = $dbw->selectField( 'cur', 'cur_timestamp', array(
1207 'cur_id' => $id ) );
1208 if ( $timeMin == $timeFirst ) {
1209 echo "timestamps match at $timeFirst; ok\n";
1211 echo "timestamps don't match! min: $timeMin, first: $timeFirst; ";
1212 if ( $timeMin > $timeFirst ) {
1214 echo "keeping minimum: $id\n";
1216 echo "keeping first: $id\n";
1224 (old_namespace, old_title, old_text,
1225 old_comment, old_user, old_user_text,
1226 old_timestamp, old_minor_edit, old_flags,
1228 SELECT cur_namespace, cur_title, cur_text,
1229 cur_comment, cur_user, cur_user_text,
1230 cur_timestamp, cur_minor_edit, '',
1233 WHERE cur_namespace=$ns
1234 AND cur_title=$title
1241 WHERE cur_namespace=$ns
1242 AND cur_title=$title
1248 $dbw->query( 'UNLOCK TABLES' );
1253 * Rename all our temporary tables into final place.
1254 * We've left things in place so a read-only wiki can continue running
1255 * on the old code during all this.
1257 function upgradeCleanup() {
1258 $this->renameTable( 'old', 'text' );
1260 foreach ( $this->cleanupSwaps as $table ) {
1261 $this->swap( $table );
1265 function renameTable( $from, $to ) {
1266 $this->log( "Renaming $from to $to..." );
1268 $fromtable = $this->dbw->tableName( $from );
1269 $totable = $this->dbw->tableName( $to );
1270 $this->dbw->query( "ALTER TABLE $fromtable RENAME TO $totable" );
1273 function swap( $base ) {
1274 $this->renameTable( $base, "{$base}_old" );
1275 $this->renameTable( "{$base}_temp", $base );