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 require_once( 'commandLine.inc' );
  11 require_once( 'cleanupDupes.inc' );
  12
  13 $upgrade = new FiveUpgrade();
  14 $upgrade->upgrade();
  15
  16 class FiveUpgrade {
  17         function FiveUpgrade() {
  18                 global $wgDatabase;
  19                 $this->conversionTables = $this->prepareWindows1252();
  20                 $this->dbw =& $this->newConnection();
  21                 $this->dbr =& $this->newConnection();
  22                 $this->dbr->bufferResults( false );
  23         }
  24
  25         function upgrade() {
  26                 $this->upgradePage();
  27                 $this->upgradeLinks();
  28         }
  29
  30
  31         /**
  32          * Open a second connection to the master server, with buffering off.
  33          * This will let us stream large datasets in and write in chunks on the
  34          * other end.
  35          * @return Database
  36          * @access private
  37          */
  38         function &newConnection() {
  39                 global $wgDBadminuser, $wgDBadminpassword;
  40                 global $wgDBserver, $wgDBname;
  41                 $db =& new Database( $wgDBserver, $wgDBadminuser, $wgDBadminpassword, $wgDBname );
  42                 return $db;
  43         }
  44
  45         /**
  46          * Prepare a conversion array for converting Windows Code Page 1252 to
  47          * UTF-8. This should provide proper conversion of text that was miscoded
  48          * as Windows-1252 by naughty user-agents, and doesn't rely on an outside
  49          * iconv library.
  50          *
  51          * @return array
  52          * @access private
  53          */
  54         function prepareWindows1252() {
  55                 # Mappings from:
  56                 # http://www.unicode.org/Public/MAPPINGS/VENDORS/MICSFT/WINDOWS/CP1252.TXT
  57                 static $cp1252 = array(
  58                         0x80 => 0x20AC, #EURO SIGN
  59                         0x81 => UNICODE_REPLACEMENT,
  60                         0x82 => 0x201A, #SINGLE LOW-9 QUOTATION MARK
  61                         0x83 => 0x0192, #LATIN SMALL LETTER F WITH HOOK
  62                         0x84 => 0x201E, #DOUBLE LOW-9 QUOTATION MARK
  63                         0x85 => 0x2026, #HORIZONTAL ELLIPSIS
  64                         0x86 => 0x2020, #DAGGER
  65                         0x87 => 0x2021, #DOUBLE DAGGER
  66                         0x88 => 0x02C6, #MODIFIER LETTER CIRCUMFLEX ACCENT
  67                         0x89 => 0x2030, #PER MILLE SIGN
  68                         0x8A => 0x0160, #LATIN CAPITAL LETTER S WITH CARON
  69                         0x8B => 0x2039, #SINGLE LEFT-POINTING ANGLE QUOTATION MARK
  70                         0x8C => 0x0152, #LATIN CAPITAL LIGATURE OE
  71                         0x8D => UNICODE_REPLACEMENT,
  72                         0x8E => 0x017D, #LATIN CAPITAL LETTER Z WITH CARON
  73                         0x8F => UNICODE_REPLACEMENT,
  74                         0x90 => UNICODE_REPLACEMENT,
  75                         0x91 => 0x2018, #LEFT SINGLE QUOTATION MARK
  76                         0x92 => 0x2019, #RIGHT SINGLE QUOTATION MARK
  77                         0x93 => 0x201C, #LEFT DOUBLE QUOTATION MARK
  78                         0x94 => 0x201D, #RIGHT DOUBLE QUOTATION MARK
  79                         0x95 => 0x2022, #BULLET
  80                         0x96 => 0x2013, #EN DASH
  81                         0x97 => 0x2014, #EM DASH
  82                         0x98 => 0x02DC, #SMALL TILDE
  83                         0x99 => 0x2122, #TRADE MARK SIGN
  84                         0x9A => 0x0161, #LATIN SMALL LETTER S WITH CARON
  85                         0x9B => 0x203A, #SINGLE RIGHT-POINTING ANGLE QUOTATION MARK
  86                         0x9C => 0x0153, #LATIN SMALL LIGATURE OE
  87                         0x9D => UNICODE_REPLACEMENT,
  88                         0x9E => 0x017E, #LATIN SMALL LETTER Z WITH CARON
  89                         0x9F => 0x0178, #LATIN CAPITAL LETTER Y WITH DIAERESIS
  90                         );
  91                 $pairs = array();
  92                 for( $i = 0; $i < 0x100; $i++ ) {
  93                         $unicode = isset( $cp1252[$i] ) ? $cp1252[$i] : $i;
  94                         $pairs[chr( $i )] = codepointToUtf8( $unicode );
  95                 }
  96                 return $pairs;
  97         }
  98
  99         /**
 100          * Convert from 8-bit Windows-1252 to UTF-8 if necessary.
 101          * @param string $text
 102          * @return string
 103          * @access private
 104          */
 105         function conv( $text ) {
 106                 global $wgUseLatin1;
 107                 if( $wgUseLatin1 ) {
 108                         return strtr( $text, $this->conversionTables );
 109                 } else {
 110                         return $text;
 111                 }
 112         }
 113
 114         /**
 115          * Dump timestamp and message to output
 116          * @param string $message
 117          * @access private
 118          */
 119         function log( $message ) {
 120                 echo wfTimestamp( TS_DB ) . ': ' . $message . "\n";
 121                 flush();
 122         }
 123
 124         /**
 125          * Initialize the chunked-insert system.
 126          * Rows will be inserted in chunks of the given number, rather
 127          * than in a giant INSERT...SELECT query, to keep the serialized
 128          * MySQL database replication from getting hung up. This way other
 129          * things can be going on during conversion without waiting for
 130          * slaves to catch up as badly.
 131          *
 132          * @param int $chunksize Number of rows to insert at once
 133          * @param int $final Total expected number of rows / id of last row,
 134          *                   used for progress reports.
 135          * @access private
 136          */
 137         function setChunkScale( $chunksize, $final ) {
 138                 $this->chunkSize  = $chunksize;
 139                 $this->chunkFinal = $final;
 140                 $this->chunkCount = 0;
 141                 $this->chunkStartTime = wfTime();
 142                 $this->chunkOptions = array();
 143         }
 144
 145         /**
 146          * Chunked inserts: perform an insert if we've reached the chunk limit.
 147          * Prints a progress report with estimated completion time.
 148          * @param string $table
 149          * @param array &$chunk -- This will be emptied if an insert is done.
 150          * @param string $fname function name to report in SQL
 151          * @param int $key A key identifier to use in progress estimation in
 152          *                 place of the number of rows inserted. Use this if
 153          *                 you provided a max key number instead of a count
 154          *                 as the final chunk number in setChunkScale()
 155          * @access private
 156          */
 157         function addChunk( $table, &$chunk, $fname, $key = null ) {
 158                 if( count( $chunk ) >= $this->chunkSize ) {
 159                         $this->insertChunk( $table, $chunk, $fname );
 160
 161                         $this->chunkCount += count( $chunk );
 162                         $now = wfTime();
 163                         $delta = $now - $this->chunkStartTime;
 164                         $rate = $this->chunkCount / $delta;
 165
 166                         if( is_null( $key ) ) {
 167                                 $completed = $this->chunkCount;
 168                         } else {
 169                                 $completed = $key;
 170                         }
 171                         $portion = $completed / $this->chunkFinal;
 172
 173                         $estimatedTotalTime = $delta / $portion;
 174                         $eta = $this->chunkStartTime + $estimatedTotalTime;
 175
 176                         printf( "%s: %6.2f%% done on %s; ETA %s [%d/%d] %.2f/sec\n",
 177                                 wfTimestamp( TS_DB, intval( $now ) ),
 178                                 $portion * 100.0,
 179                                 $table,
 180                                 wfTimestamp( TS_DB, intval( $eta ) ),
 181                                 $completed,
 182                                 $this->chunkFinal,
 183                                 $rate );
 184                         flush();
 185
 186                         $chunk = array();
 187                 }
 188         }
 189
 190         /**
 191          * Chunked inserts: perform an insert unconditionally, at the end, and log.
 192          * @param string $table
 193          * @param array &$chunk -- This will be emptied if an insert is done.
 194          * @param string $fname function name to report in SQL
 195          * @access private
 196          */
 197         function lastChunk( $table, &$chunk, $fname ) {
 198                 $n = count( $chunk );
 199                 if( $n > 0 ) {
 200                         $this->insertChunk( $table, $chunk, $fname );
 201                 }
 202                 $this->log( "100.00% done on $table (last chunk $n rows)." );
 203         }
 204
 205         /**
 206          * Chunked inserts: perform an insert.
 207          * @param string $table
 208          * @param array &$chunk -- This will be emptied if an insert is done.
 209          * @param string $fname function name to report in SQL
 210          * @access private
 211          */
 212         function insertChunk( $table, &$chunk, $fname ) {
 213                 $this->dbw->insert( $table, $chunk, $fname, $this->chunkOptions );
 214         }
 215
 216         function upgradePage() {
 217                 $fname = "FiveUpgrade::upgradePage";
 218                 $chunksize = 500;
 219
 220
 221                 $this->log( "Checking cur table for unique title index and applying if necessary" );
 222                 checkDupes( true );
 223
 224                 $this->log( "...converting from cur/old to page/revision/text DB structure." );
 225
 226                 extract( $this->dbw->tableNames( 'cur', 'old', 'page', 'revision', 'text' ) );
 227
 228                 $this->log( "Creating page and revision tables..." );
 229                 $this->dbw->query("CREATE TABLE $page (
 230                         page_id int(8) unsigned NOT NULL auto_increment,
 231                         page_namespace int NOT NULL,
 232                         page_title varchar(255) binary NOT NULL,
 233                         page_restrictions tinyblob NOT NULL default '',
 234                         page_counter bigint(20) unsigned NOT NULL default '0',
 235                         page_is_redirect tinyint(1) unsigned NOT NULL default '0',
 236                         page_is_new tinyint(1) unsigned NOT NULL default '0',
 237                         page_random real unsigned NOT NULL,
 238                         page_touched char(14) binary NOT NULL default '',
 239                         page_latest int(8) unsigned NOT NULL,
 240                         page_len int(8) unsigned NOT NULL,
 241
 242                         PRIMARY KEY page_id (page_id),
 243                         UNIQUE INDEX name_title (page_namespace,page_title),
 244                         INDEX (page_random),
 245                         INDEX (page_len)
 246                         ) TYPE=InnoDB", $fname );
 247                 $this->dbw->query("CREATE TABLE $revision (
 248                         rev_id int(8) unsigned NOT NULL auto_increment,
 249                         rev_page int(8) unsigned NOT NULL,
 250                         rev_comment tinyblob NOT NULL default '',
 251                         rev_user int(5) unsigned NOT NULL default '0',
 252                         rev_user_text varchar(255) binary NOT NULL default '',
 253                         rev_timestamp char(14) binary NOT NULL default '',
 254                         rev_minor_edit tinyint(1) unsigned NOT NULL default '0',
 255                         rev_deleted tinyint(1) unsigned NOT NULL default '0',
 256
 257                         PRIMARY KEY rev_page_id (rev_page, rev_id),
 258                         UNIQUE INDEX rev_id (rev_id),
 259                         INDEX rev_timestamp (rev_timestamp),
 260                         INDEX page_timestamp (rev_page,rev_timestamp),
 261                         INDEX user_timestamp (rev_user,rev_timestamp),
 262                         INDEX usertext_timestamp (rev_user_text,rev_timestamp)
 263                         ) TYPE=InnoDB", $fname );
 264
 265                 $maxold = $this->dbw->selectField( 'old', 'max(old_id)', '', $fname );
 266                 $this->log( "Last old record is {$maxold}" );
 267
 268                 global $wgLegacySchemaConversion;
 269                 if( $wgLegacySchemaConversion ) {
 270                         // Create HistoryBlobCurStub entries.
 271                         // Text will be pulled from the leftover 'cur' table at runtime.
 272                         echo "......Moving metadata from cur; using blob references to text in cur table.\n";
 273                         $cur_text = "concat('O:18:\"historyblobcurstub\":1:{s:6:\"mCurId\";i:',cur_id,';}')";
 274                         $cur_flags = "'object'";
 275                 } else {
 276                         // Copy all cur text in immediately: this may take longer but avoids
 277                         // having to keep an extra table around.
 278                         echo "......Moving text from cur.\n";
 279                         $cur_text = 'cur_text';
 280                         $cur_flags = "''";
 281                 }
 282
 283                 $maxcur = $this->dbw->selectField( 'cur', 'max(cur_id)', '', $fname );
 284                 $this->log( "Last cur entry is $maxcur" );
 285
 286                 /**
 287                  * Copy placeholder records for each page's current version into old
 288                  * Don't do any conversion here; text records are converted at runtime
 289                  * based on the flags (and may be originally binary!) while the meta
 290                  * fields will be converted in the old -> rev and cur -> page steps.
 291                  */
 292                 $this->setChunkScale( $chunksize, $maxcur );
 293                 $result = $this->dbr->query(
 294                         "SELECT cur_id, cur_namespace, cur_title, $cur_text AS text, cur_comment,
 295                         cur_user, cur_user_text, cur_timestamp, cur_minor_edit, $cur_flags AS flags
 296                         FROM $cur
 297                         ORDER BY cur_id", $fname );
 298                 $add = array();
 299                 while( $row = $this->dbr->fetchObject( $result ) ) {
 300                         $add[] = array(
 301                                 'old_namespace'  => $row->cur_namespace,
 302                                 'old_title'      => $row->cur_title,
 303                                 'old_text'       => $row->text,
 304                                 'old_comment'    => $row->cur_comment,
 305                                 'old_user'       => $row->cur_user,
 306                                 'old_user_text'  => $row->cur_user_text,
 307                                 'old_timestamp'  => $row->cur_timestamp,
 308                                 'old_minor_edit' => $row->cur_minor_edit,
 309                                 'old_flags'      => $row->flags );
 310                         $this->addChunk( 'old', $add, $fname, $row->cur_id );
 311                 }
 312                 $this->lastChunk( 'old', $add, $fname );
 313                 $this->dbr->freeResult( $result );
 314
 315                 /**
 316                  * Copy revision metadata from old into revision.
 317                  * We'll also do UTF-8 conversion of usernames and comments.
 318                  */
 319                 #$newmaxold = $this->dbw->selectField( 'old', 'max(old_id)', '', $fname );
 320                 #$this->setChunkScale( $chunksize, $newmaxold );
 321                 $countold = $this->dbw->selectField( 'old', 'count(old_id)', '', $fname );
 322                 $this->setChunkScale( $chunksize, $countold );
 323
 324                 $this->log( "......Setting up revision table." );
 325                 $result = $this->dbr->query(
 326                         "SELECT old_id, cur_id, old_comment, old_user, old_user_text,
 327                         old_timestamp, old_minor_edit
 328                         FROM $old,$cur WHERE old_namespace=cur_namespace AND old_title=cur_title",
 329                         $fname );
 330
 331                 $add = array();
 332                 while( $row = $this->dbr->fetchObject( $result ) ) {
 333                         $add[] = array(
 334                                 'rev_id'         =>              $row->old_id,
 335                                 'rev_page'       =>              $row->cur_id,
 336                                 'rev_comment'    => $this->conv( $row->old_comment ),
 337                                 'rev_user'       =>              $row->old_user,
 338                                 'rev_user_text'  => $this->conv( $row->old_user_text ),
 339                                 'rev_timestamp'  =>              $row->old_timestamp,
 340                                 'rev_minor_edit' =>              $row->old_minor_edit );
 341                         $this->addChunk( 'revision', $add, $fname );
 342                 }
 343                 $this->lastChunk( 'revision', $add, $fname );
 344                 $this->dbr->freeResult( $result );
 345
 346
 347                 /**
 348                  * Copy revision metadata from cur into page.
 349                  * We'll also do UTF-8 conversion of titles.
 350                  */
 351                 $this->log( "......Setting up page table." );
 352                 $this->setChunkScale( $chunksize, $maxcur );
 353                 $result = $this->dbr->query( "
 354                         SELECT cur_id, cur_namespace, cur_title, cur_restrictions, cur_counter, cur_is_redirect, cur_is_new,
 355                                 cur_random, cur_touched, rev_id, LENGTH(cur_text) AS len
 356                         FROM $cur,$revision
 357                         WHERE cur_id=rev_page AND rev_timestamp=cur_timestamp AND rev_id > {$maxold}
 358                         ORDER BY cur_id", $fname );
 359                 $add = array();
 360                 while( $row = $this->dbr->fetchObject( $result ) ) {
 361                         $add[] = array(
 362                                 'page_id'           =>              $row->cur_id,
 363                                 'page_namespace'    =>              $row->cur_namespace,
 364                                 'page_title'        => $this->conv( $row->cur_title ),
 365                                 'page_restrictions' =>              $row->cur_restrictions,
 366                                 'page_counter'      =>              $row->cur_counter,
 367                                 'page_is_redirect'  =>              $row->cur_is_redirect,
 368                                 'page_is_new'       =>              $row->cur_is_new,
 369                                 'page_random'       =>              $row->cur_random,
 370                                 'page_touched'      =>              $this->dbw->timestamp(),
 371                                 'page_latest'       =>              $row->rev_id,
 372                                 'page_len'          =>              $row->len );
 373                         $this->addChunk( 'page', $add, $fname, $row->cur_id );
 374                 }
 375                 $this->lastChunk( 'page', $add, $fname );
 376                 $this->dbr->freeResult( $result );
 377
 378                 $this->log( "......Renaming old." );
 379                 $this->dbw->query( "ALTER TABLE $old RENAME TO $text", $fname );
 380
 381                 $this->log( "...done." );
 382         }
 383
 384         function upgradeLinks() {
 385                 $fname = 'FiveUpgrade::upgradeLinks';
 386                 $chunksize = 1000;
 387                 extract( $this->dbw->tableNames( 'links', 'brokenlinks', 'pagelinks', 'page' ) );
 388
 389                 $this->log( 'Creating pagelinks table...' );
 390                 $this->dbw->query( "
 391 CREATE TABLE $pagelinks (
 392   -- Key to the page_id of the page containing the link.
 393   pl_from int(8) unsigned NOT NULL default '0',
 394
 395   -- Key to page_namespace/page_title of the target page.
 396   -- The target page may or may not exist, and due to renames
 397   -- and deletions may refer to different page records as time
 398   -- goes by.
 399   pl_namespace int NOT NULL default '0',
 400   pl_title varchar(255) binary NOT NULL default '',
 401
 402   UNIQUE KEY pl_from(pl_from,pl_namespace,pl_title),
 403   KEY (pl_namespace,pl_title)
 404
 405 ) TYPE=InnoDB" );
 406
 407                 $this->log( 'Importing live links -> pagelinks' );
 408                 $nlinks = $this->dbw->selectField( 'links', 'count(*)', '', $fname );
 409                 if( $nlinks ) {
 410                         $this->setChunkScale( $chunksize, $nlinks );
 411                         $result = $this->dbr->query( "
 412                           SELECT l_from,page_namespace,page_title
 413                                 FROM $links, $page
 414                                 WHERE l_to=page_id", $fname );
 415                         $add = array();
 416                         while( $row = $this->dbr->fetchObject( $result ) ) {
 417                                 $add[] = array(
 418                                         'pl_from'      => $row->l_from,
 419                                         'pl_namespace' => $row->page_namespace,
 420                                         'pl_title'     => $row->page_title );
 421                                 $this->addChunk( 'pagelinks', $add, $fname );
 422                         }
 423                         $this->lastChunk( 'pagelinks', $add, $fname );
 424                 } else {
 425                         $this->log( 'no links!' );
 426                 }
 427
 428                 $this->log( 'Importing brokenlinks -> pagelinks' );
 429                 $nbrokenlinks = $this->dbw->selectField( 'brokenlinks', 'count(*)', '', $fname );
 430                 if( $nbrokenlinks ) {
 431                         $this->setChunkScale( $chunksize, $nbrokenlinks );
 432                         $this->chunkOptions = array( 'IGNORE' );
 433                         $result = $this->dbr->query(
 434                                 "SELECT bl_from, bl_to FROM $brokenlinks",
 435                                 $fname );
 436                         $add = array();
 437                         while( $row = $this->dbr->fetchObject( $result ) ) {
 438                                 $pagename = $this->conv( $row->bl_to );
 439                                 $title = Title::newFromText( $pagename );
 440                                 if( is_null( $title ) ) {
 441                                         $this->log( "** invalid brokenlink: $row->bl_from -> '$pagename' (converted from '$row->bl_to')" );
 442                                 } else {
 443                                         $add[] = array(
 444                                                 'pl_from'      => $row->bl_from,
 445                                                 'pl_namespace' => $title->getNamespace(),
 446                                                 'pl_title'     => $title->getDBkey() );
 447                                         $this->addChunk( 'pagelinks', $add, $fname );
 448                                 }
 449                         }
 450                         $this->lastChunk( 'pagelinks', $add, $fname );
 451                 } else {
 452                         $this->log( 'no brokenlinks!' );
 453                 }
 454
 455                 $this->log( 'Done with links.' );
 456         }
 457 }
 458
 459 ?>