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