maintenance/backupTextPass.inc

   1 <?php
   2 /**
   3  * BackupDumper that postprocesses XML dumps from dumpBackup.php to add page text
   4  *
   5  * Copyright (C) 2005 Brion Vibber <brion@pobox.com>
   6  * http://www.mediawiki.org/
   7  *
   8  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 2 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with this program; if not, write to the Free Software Foundation, Inc.,
  20  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  21  * http://www.gnu.org/copyleft/gpl.html
  22  *
  23  * @file
  24  * @ingroup Maintenance
  25  */
  26
  27
  28 /**
  29  * @ingroup Maintenance
  30  */
  31 class TextPassDumper extends BackupDumper {
  32         var $prefetch = null;
  33         var $input = "php://stdin";
  34         var $history = WikiExporter::FULL;
  35         var $fetchCount = 0;
  36         var $prefetchCount = 0;
  37         var $prefetchCountLast = 0;
  38         var $fetchCountLast = 0;
  39
  40         var $maxFailures = 5;
  41         var $maxConsecutiveFailedTextRetrievals = 200;
  42         var $failureTimeout = 5; // Seconds to sleep after db failure
  43
  44         var $php = "php";
  45         var $spawn = false;
  46         var $spawnProc = false;
  47         var $spawnWrite = false;
  48         var $spawnRead = false;
  49         var $spawnErr = false;
  50
  51         var $xmlwriterobj = false;
  52
  53         // when we spend more than maxTimeAllowed seconds on this run, we continue
  54         // processing until we write out the next complete page, then save output file(s),
  55         // rename it/them and open new one(s)
  56         var $maxTimeAllowed = 0;  // 0 = no limit
  57         var $timeExceeded = false;
  58         var $firstPageWritten = false;
  59         var $lastPageWritten = false;
  60         var $checkpointJustWritten = false;
  61         var $checkpointFiles = array();
  62
  63         /**
  64          * @var DatabaseBase
  65          */
  66         protected $db;
  67
  68
  69         /**
  70          * Drop the database connection $this->db and try to get a new one.
  71          *
  72          * This function tries to get a /different/ connection if this is
  73          * possible. Hence, (if this is possible) it switches to a different
  74          * failover upon each call.
  75          *
  76          * This function resets $this->lb and closes all connections on it.
  77          *
  78          * @throws MWException
  79          */
  80         function rotateDb() {
  81                 // Cleaning up old connections
  82                 if ( isset( $this->lb ) ) {
  83                         $this->lb->closeAll();
  84                         unset( $this->lb );
  85                 }
  86
  87                 if ( isset( $this->db ) && $this->db->isOpen() ) {
  88                         throw new MWException( 'DB is set and has not been closed by the Load Balancer' );
  89                 }
  90
  91                 unset( $this->db );
  92
  93                 // Trying to set up new connection.
  94                 // We do /not/ retry upon failure, but delegate to encapsulating logic, to avoid
  95                 // individually retrying at different layers of code.
  96
  97                 // 1. The LoadBalancer.
  98                 try {
  99                         $this->lb = wfGetLBFactory()->newMainLB();
 100                 } catch ( Exception $e ) {
 101                         throw new MWException( __METHOD__ . " rotating DB failed to obtain new load balancer (" . $e->getMessage() . ")" );
 102                 }
 103
 104
 105                 // 2. The Connection, through the load balancer.
 106                 try {
 107                         $this->db = $this->lb->getConnection( DB_SLAVE, 'backup' );
 108                 } catch ( Exception $e ) {
 109                         throw new MWException( __METHOD__ . " rotating DB failed to obtain new database (" . $e->getMessage() . ")" );
 110                 }
 111         }
 112
 113
 114         function initProgress( $history ) {
 115                 parent::initProgress();
 116                 $this->timeOfCheckpoint = $this->startTime;
 117         }
 118
 119         function dump( $history, $text = WikiExporter::TEXT ) {
 120                 // This shouldn't happen if on console... ;)
 121                 header( 'Content-type: text/html; charset=UTF-8' );
 122
 123                 // Notice messages will foul up your XML output even if they're
 124                 // relatively harmless.
 125                 if ( ini_get( 'display_errors' ) )
 126                         ini_set( 'display_errors', 'stderr' );
 127
 128                 $this->initProgress( $this->history );
 129
 130                 // We are trying to get an initial database connection to avoid that the
 131                 // first try of this request's first call to getText fails. However, if
 132                 // obtaining a good DB connection fails it's not a serious issue, as
 133                 // getText does retry upon failure and can start without having a working
 134                 // DB connection.
 135                 try {
 136                         $this->rotateDb();
 137                 } catch ( Exception $e ) {
 138                         // We do not even count this as failure. Just let eventual
 139                         // watchdogs know.
 140                         $this->progress( "Getting initial DB connection failed (" .
 141                                 $e->getMessage() . ")" );
 142                 }
 143
 144                 $this->egress = new ExportProgressFilter( $this->sink, $this );
 145
 146                 // it would be nice to do it in the constructor, oh well. need egress set
 147                 $this->finalOptionCheck();
 148
 149                 // we only want this so we know how to close a stream :-P
 150                 $this->xmlwriterobj = new XmlDumpWriter();
 151
 152                 $input = fopen( $this->input, "rt" );
 153                 $result = $this->readDump( $input );
 154
 155                 if ( WikiError::isError( $result ) ) {
 156                         throw new MWException( $result->getMessage() );
 157                 }
 158
 159                 if ( $this->spawnProc ) {
 160                         $this->closeSpawn();
 161                 }
 162
 163                 $this->report( true );
 164         }
 165
 166         function processOption( $opt, $val, $param ) {
 167                 global $IP;
 168                 $url = $this->processFileOpt( $val, $param );
 169
 170                 switch( $opt ) {
 171                 case 'prefetch':
 172                         require_once "$IP/maintenance/backupPrefetch.inc";
 173                         $this->prefetch = new BaseDump( $url );
 174                         break;
 175                 case 'stub':
 176                         $this->input = $url;
 177                         break;
 178                 case 'maxtime':
 179                         $this->maxTimeAllowed = intval( $val ) * 60;
 180                         break;
 181                 case 'checkpointfile':
 182                         $this->checkpointFiles[] = $val;
 183                         break;
 184                 case 'current':
 185                         $this->history = WikiExporter::CURRENT;
 186                         break;
 187                 case 'full':
 188                         $this->history = WikiExporter::FULL;
 189                         break;
 190                 case 'spawn':
 191                         $this->spawn = true;
 192                         if ( $val ) {
 193                                 $this->php = $val;
 194                         }
 195                         break;
 196                 }
 197         }
 198
 199         function processFileOpt( $val, $param ) {
 200                 $fileURIs = explode( ';', $param );
 201                 foreach ( $fileURIs as $URI ) {
 202                         switch( $val ) {
 203                                 case "file":
 204                                         $newURI = $URI;
 205                                         break;
 206                                 case "gzip":
 207                                         $newURI = "compress.zlib://$URI";
 208                                         break;
 209                                 case "bzip2":
 210                                         $newURI = "compress.bzip2://$URI";
 211                                         break;
 212                                 case "7zip":
 213                                         $newURI = "mediawiki.compress.7z://$URI";
 214                                         break;
 215                                 default:
 216                                         $newURI = $URI;
 217                         }
 218                         $newFileURIs[] = $newURI;
 219                 }
 220                 $val = implode( ';', $newFileURIs );
 221                 return $val;
 222         }
 223
 224         /**
 225          * Overridden to include prefetch ratio if enabled.
 226          */
 227         function showReport() {
 228                 if ( !$this->prefetch ) {
 229                         parent::showReport();
 230                         return;
 231                 }
 232
 233                 if ( $this->reporting ) {
 234                         $now = wfTimestamp( TS_DB );
 235                         $nowts = wfTime();
 236                         $deltaAll = wfTime() - $this->startTime;
 237                         $deltaPart = wfTime() - $this->lastTime;
 238                         $this->pageCountPart = $this->pageCount - $this->pageCountLast;
 239                         $this->revCountPart = $this->revCount - $this->revCountLast;
 240
 241                         if ( $deltaAll ) {
 242                                 $portion = $this->revCount / $this->maxCount;
 243                                 $eta = $this->startTime + $deltaAll / $portion;
 244                                 $etats = wfTimestamp( TS_DB, intval( $eta ) );
 245                                 if ( $this->fetchCount ) {
 246                                         $fetchRate = 100.0 * $this->prefetchCount / $this->fetchCount;
 247                                 } else {
 248                                         $fetchRate = '-';
 249                                 }
 250                                 $pageRate = $this->pageCount / $deltaAll;
 251                                 $revRate = $this->revCount / $deltaAll;
 252                         } else {
 253                                 $pageRate = '-';
 254                                 $revRate = '-';
 255                                 $etats = '-';
 256                                 $fetchRate = '-';
 257                         }
 258                         if ( $deltaPart ) {
 259                                 if ( $this->fetchCountLast ) {
 260                                         $fetchRatePart = 100.0 * $this->prefetchCountLast / $this->fetchCountLast;
 261                                 } else {
 262                                         $fetchRatePart = '-';
 263                                 }
 264                                 $pageRatePart = $this->pageCountPart / $deltaPart;
 265                                 $revRatePart = $this->revCountPart / $deltaPart;
 266
 267                         } else {
 268                                 $fetchRatePart = '-';
 269                                 $pageRatePart = '-';
 270                                 $revRatePart = '-';
 271                         }
 272                         $this->progress( sprintf( "%s: %s (ID %d) %d pages (%0.1f|%0.1f/sec all|curr), %d revs (%0.1f|%0.1f/sec all|curr), %0.1f%%|%0.1f%% prefetched (all|curr), ETA %s [max %d]",
 273                                         $now, wfWikiID(), $this->ID, $this->pageCount, $pageRate, $pageRatePart, $this->revCount, $revRate, $revRatePart, $fetchRate, $fetchRatePart, $etats, $this->maxCount ) );
 274                         $this->lastTime = $nowts;
 275                         $this->revCountLast = $this->revCount;
 276                         $this->prefetchCountLast = $this->prefetchCount;
 277                         $this->fetchCountLast = $this->fetchCount;
 278                 }
 279         }
 280
 281         function setTimeExceeded() {
 282                 $this->timeExceeded = True;
 283         }
 284
 285         function checkIfTimeExceeded() {
 286                 if ( $this->maxTimeAllowed &&  ( $this->lastTime - $this->timeOfCheckpoint  > $this->maxTimeAllowed ) ) {
 287                         return true;
 288                 }
 289                 return false;
 290         }
 291
 292         function finalOptionCheck() {
 293                 if ( ( $this->checkpointFiles && ! $this->maxTimeAllowed ) ||
 294                         ( $this->maxTimeAllowed && !$this->checkpointFiles ) ) {
 295                         throw new MWException( "Options checkpointfile and maxtime must be specified together.\n" );
 296                 }
 297                 foreach ( $this->checkpointFiles as $checkpointFile ) {
 298                         $count = substr_count ( $checkpointFile, "%s" );
 299                         if ( $count != 2 ) {
 300                                 throw new MWException( "Option checkpointfile must contain two '%s' for substitution of first and last pageids, count is $count instead, file is $checkpointFile.\n" );
 301                         }
 302                 }
 303
 304                 if ( $this->checkpointFiles ) {
 305                         $filenameList = (array)$this->egress->getFilenames();
 306                         if ( count( $filenameList ) != count( $this->checkpointFiles ) ) {
 307                                 throw new MWException( "One checkpointfile must be specified for each output option, if maxtime is used.\n" );
 308                         }
 309                 }
 310         }
 311
 312         function readDump( $input ) {
 313                 $this->buffer = "";
 314                 $this->openElement = false;
 315                 $this->atStart = true;
 316                 $this->state = "";
 317                 $this->lastName = "";
 318                 $this->thisPage = 0;
 319                 $this->thisRev = 0;
 320
 321                 $parser = xml_parser_create( "UTF-8" );
 322                 xml_parser_set_option( $parser, XML_OPTION_CASE_FOLDING, false );
 323
 324                 xml_set_element_handler( $parser, array( &$this, 'startElement' ), array( &$this, 'endElement' ) );
 325                 xml_set_character_data_handler( $parser, array( &$this, 'characterData' ) );
 326
 327                 $offset = 0; // for context extraction on error reporting
 328                 $bufferSize = 512 * 1024;
 329                 do {
 330                         if ( $this->checkIfTimeExceeded() ) {
 331                                 $this->setTimeExceeded();
 332                         }
 333                         $chunk = fread( $input, $bufferSize );
 334                         if ( !xml_parse( $parser, $chunk, feof( $input ) ) ) {
 335                                 wfDebug( "TextDumpPass::readDump encountered XML parsing error\n" );
 336                                 return new WikiXmlError( $parser, 'XML import parse failure', $chunk, $offset );
 337                         }
 338                         $offset += strlen( $chunk );
 339                 } while ( $chunk !== false && !feof( $input ) );
 340                 if ( $this->maxTimeAllowed ) {
 341                         $filenameList = (array)$this->egress->getFilenames();
 342                         // we wrote some stuff after last checkpoint that needs renamed
 343                         if ( file_exists( $filenameList[0] ) ) {
 344                                 $newFilenames = array();
 345                                 # we might have just written the header and footer and had no
 346                                 # pages or revisions written... perhaps they were all deleted
 347                                 # there's no pageID 0 so we use that. the caller is responsible
 348                                 # for deciding what to do with a file containing only the
 349                                 # siteinfo information and the mw tags.
 350                                 if ( ! $this->firstPageWritten ) {
 351                                         $firstPageID = str_pad( 0, 9, "0", STR_PAD_LEFT );
 352                                         $lastPageID = str_pad( 0, 9, "0", STR_PAD_LEFT );
 353                                 }
 354                                 else {
 355                                         $firstPageID = str_pad( $this->firstPageWritten, 9, "0", STR_PAD_LEFT );
 356                                         $lastPageID = str_pad( $this->lastPageWritten, 9, "0", STR_PAD_LEFT );
 357                                 }
 358                                 for ( $i = 0; $i < count( $filenameList ); $i++ ) {
 359                                         $checkpointNameFilledIn = sprintf( $this->checkpointFiles[$i], $firstPageID, $lastPageID );
 360                                         $fileinfo = pathinfo( $filenameList[$i] );
 361                                         $newFilenames[] = $fileinfo['dirname'] . '/' . $checkpointNameFilledIn;
 362                                 }
 363                                 $this->egress->closeAndRename( $newFilenames );
 364                         }
 365                 }
 366                 xml_parser_free( $parser );
 367
 368                 return true;
 369         }
 370
 371         /**
 372          * Tries to get the revision text for a revision id.
 373          *
 374          * Upon errors, retries (Up to $this->maxFailures tries each call).
 375          * If still no good revision get could be found even after this retrying, "" is returned.
 376          * If no good revision text could be returned for
 377          * $this->maxConsecutiveFailedTextRetrievals consecutive calls to getText, MWException
 378          * is thrown.
 379          *
 380          * @param $id string The revision id to get the text for
 381          *
 382          * @return string The revision text for $id, or ""
 383          * @throws MWException
 384          */
 385         function getText( $id ) {
 386                 $prefetchNotTried = true; // Whether or not we already tried to get the text via prefetch.
 387                 $text = false; // The candidate for a good text. false if no proper value.
 388                 $failures = 0; // The number of times, this invocation of getText already failed.
 389
 390                 static $consecutiveFailedTextRetrievals = 0; // The number of times getText failed without
 391                                                              // yielding a good text in between.
 392
 393                 $this->fetchCount++;
 394
 395                 // To allow to simply return on success and do not have to worry about book keeping,
 396                 // we assume, this fetch works (possible after some retries). Nevertheless, we koop
 397                 // the old value, so we can restore it, if problems occur (See after the while loop).
 398                 $oldConsecutiveFailedTextRetrievals = $consecutiveFailedTextRetrievals;
 399                 $consecutiveFailedTextRetrievals = 0;
 400
 401                 while ( $failures < $this->maxFailures ) {
 402
 403                         // As soon as we found a good text for the $id, we will return immediately.
 404                         // Hence, if we make it past the try catch block, we know that we did not
 405                         // find a good text.
 406
 407                         try {
 408                                 // Step 1: Get some text (or reuse from previous iteratuon if checking
 409                                 //         for plausibility failed)
 410
 411                                 // Trying to get prefetch, if it has not been tried before
 412                                 if ( $text === false && isset( $this->prefetch ) && $prefetchNotTried ) {
 413                                         $prefetchNotTried = false;
 414                                         $tryIsPrefetch = true;
 415                                         $text = $this->prefetch->prefetch( $this->thisPage, $this->thisRev );
 416                                         if ( $text === null ) {
 417                                                 $text = false;
 418                                         }
 419                                 }
 420
 421                                 if ( $text === false ) {
 422                                         // Fallback to asking the database
 423                                         $tryIsPrefetch = false;
 424                                         if ( $this->spawn ) {
 425                                                 $text = $this->getTextSpawned( $id );
 426                                         } else {
 427                                                 $text = $this->getTextDb( $id );
 428                                         }
 429                                 }
 430
 431                                 if ( $text === false ) {
 432                                         throw new MWException( "Generic error while obtaining text for id " . $id );
 433                                 }
 434
 435                                 // We received a good candidate for the text of $id via some method
 436
 437                                 // Step 2: Checking for plausibility and return the text if it is
 438                                 //         plausible
 439                                 $revID = intval( $this->thisRev );
 440                                 if ( ! isset( $this->db ) ) {
 441                                         throw new MWException( "No database available" );
 442                                 }
 443                                 $revLength = $this->db->selectField( 'revision', 'rev_len', array( 'rev_id' => $revID ) );
 444                                 if ( strlen( $text ) == $revLength ) {
 445                                         if ( $tryIsPrefetch ) {
 446                                                 $this->prefetchCount++;
 447                                         }
 448                                         return $text;
 449                                 }
 450
 451                                 $text = false;
 452                                 throw new MWException( "Received text is unplausible for id " . $id );
 453
 454                         } catch ( Exception $e ) {
 455                                 $msg = "getting/checking text " . $id . " failed (" . $e->getMessage() . ")";
 456                                 if ( $failures + 1 < $this->maxFailures ) {
 457                                         $msg .= " (Will retry " . ( $this->maxFailures - $failures - 1 ) . " more times)";
 458                                 }
 459                                 $this->progress( $msg );
 460                         }
 461
 462                         // Something went wrong; we did not a text that was plausible :(
 463                         $failures++;
 464
 465
 466                         // After backing off for some time, we try to reboot the whole process as
 467                         // much as possible to not carry over failures from one part to the other
 468                         // parts
 469                         sleep( $this->failureTimeout );
 470                         try {
 471                                 $this->rotateDb();
 472                                 if ( $this->spawn ) {
 473                                         $this->closeSpawn();
 474                                         $this->openSpawn();
 475                                 }
 476                         } catch ( Exception $e ) {
 477                                 $this->progress( "Rebooting getText infrastructure failed (" . $e->getMessage() . ")" .
 478                                         " Trying to continue anyways" );
 479                         }
 480                 }
 481
 482                 // Retirieving a good text for $id failed (at least) maxFailures times.
 483                 // We abort for this $id.
 484
 485                 // Restoring the consecutive failures, and maybe aborting, if the dump
 486                 // is too broken.
 487                 $consecutiveFailedTextRetrievals = $oldConsecutiveFailedTextRetrievals + 1;
 488                 if ( $consecutiveFailedTextRetrievals > $this->maxConsecutiveFailedTextRetrievals ) {
 489                         throw new MWException( "Graceful storage failure" );
 490                 }
 491
 492                 return "";
 493         }
 494
 495
 496         /**
 497          * May throw a database error if, say, the server dies during query.
 498          * @param $id
 499          * @return bool|string
 500          * @throws MWException
 501          */
 502         private function getTextDb( $id ) {
 503                 global $wgContLang;
 504                 if ( ! isset( $this->db ) ) {
 505                         throw new MWException( __METHOD__ . "No database available" );
 506                 }
 507                 $row = $this->db->selectRow( 'text',
 508                         array( 'old_text', 'old_flags' ),
 509                         array( 'old_id' => $id ),
 510                         __METHOD__ );
 511                 $text = Revision::getRevisionText( $row );
 512                 if ( $text === false ) {
 513                         return false;
 514                 }
 515                 $stripped = str_replace( "\r", "", $text );
 516                 $normalized = $wgContLang->normalize( $stripped );
 517                 return $normalized;
 518         }
 519
 520         private function getTextSpawned( $id ) {
 521                 wfSuppressWarnings();
 522                 if ( !$this->spawnProc ) {
 523                         // First time?
 524                         $this->openSpawn();
 525                 }
 526                 $text = $this->getTextSpawnedOnce( $id );
 527                 wfRestoreWarnings();
 528                 return $text;
 529         }
 530
 531         function openSpawn() {
 532                 global $IP;
 533
 534                 if ( file_exists( "$IP/../multiversion/MWScript.php" ) ) {
 535                         $cmd = implode( " ",
 536                                 array_map( 'wfEscapeShellArg',
 537                                         array(
 538                                                 $this->php,
 539                                                 "$IP/../multiversion/MWScript.php",
 540                                                 "fetchText.php",
 541                                                 '--wiki', wfWikiID() ) ) );
 542                 }
 543                 else {
 544                         $cmd = implode( " ",
 545                                 array_map( 'wfEscapeShellArg',
 546                                         array(
 547                                                 $this->php,
 548                                                 "$IP/maintenance/fetchText.php",
 549                                                 '--wiki', wfWikiID() ) ) );
 550                 }
 551                 $spec = array(
 552                         0 => array( "pipe", "r" ),
 553                         1 => array( "pipe", "w" ),
 554                         2 => array( "file", "/dev/null", "a" ) );
 555                 $pipes = array();
 556
 557                 $this->progress( "Spawning database subprocess: $cmd" );
 558                 $this->spawnProc = proc_open( $cmd, $spec, $pipes );
 559                 if ( !$this->spawnProc ) {
 560                         // shit
 561                         $this->progress( "Subprocess spawn failed." );
 562                         return false;
 563                 }
 564                 list(
 565                         $this->spawnWrite, // -> stdin
 566                         $this->spawnRead,  // <- stdout
 567                 ) = $pipes;
 568
 569                 return true;
 570         }
 571
 572         private function closeSpawn() {
 573                 wfSuppressWarnings();
 574                 if ( $this->spawnRead )
 575                         fclose( $this->spawnRead );
 576                 $this->spawnRead = false;
 577                 if ( $this->spawnWrite )
 578                         fclose( $this->spawnWrite );
 579                 $this->spawnWrite = false;
 580                 if ( $this->spawnErr )
 581                         fclose( $this->spawnErr );
 582                 $this->spawnErr = false;
 583                 if ( $this->spawnProc )
 584                         pclose( $this->spawnProc );
 585                 $this->spawnProc = false;
 586                 wfRestoreWarnings();
 587         }
 588
 589         private function getTextSpawnedOnce( $id ) {
 590                 global $wgContLang;
 591
 592                 $ok = fwrite( $this->spawnWrite, "$id\n" );
 593                 // $this->progress( ">> $id" );
 594                 if ( !$ok ) return false;
 595
 596                 $ok = fflush( $this->spawnWrite );
 597                 // $this->progress( ">> [flush]" );
 598                 if ( !$ok ) return false;
 599
 600                 // check that the text id they are sending is the one we asked for
 601                 // this avoids out of sync revision text errors we have encountered in the past
 602                 $newId = fgets( $this->spawnRead );
 603                 if ( $newId === false ) {
 604                         return false;
 605                 }
 606                 if ( $id != intval( $newId ) ) {
 607                         return false;
 608                 }
 609
 610                 $len = fgets( $this->spawnRead );
 611                 // $this->progress( "<< " . trim( $len ) );
 612                 if ( $len === false ) return false;
 613
 614                 $nbytes = intval( $len );
 615                 // actual error, not zero-length text
 616                 if ( $nbytes < 0 ) return false;
 617
 618                 $text = "";
 619
 620                 // Subprocess may not send everything at once, we have to loop.
 621                 while ( $nbytes > strlen( $text ) ) {
 622                         $buffer = fread( $this->spawnRead, $nbytes - strlen( $text ) );
 623                         if ( $buffer === false ) break;
 624                         $text .= $buffer;
 625                 }
 626
 627                 $gotbytes = strlen( $text );
 628                 if ( $gotbytes != $nbytes ) {
 629                         $this->progress( "Expected $nbytes bytes from database subprocess, got $gotbytes " );
 630                         return false;
 631                 }
 632
 633                 // Do normalization in the dump thread...
 634                 $stripped = str_replace( "\r", "", $text );
 635                 $normalized = $wgContLang->normalize( $stripped );
 636                 return $normalized;
 637         }
 638
 639         function startElement( $parser, $name, $attribs ) {
 640                 $this->checkpointJustWritten = false;
 641
 642                 $this->clearOpenElement( null );
 643                 $this->lastName = $name;
 644
 645                 if ( $name == 'revision' ) {
 646                         $this->state = $name;
 647                         $this->egress->writeOpenPage( null, $this->buffer );
 648                         $this->buffer = "";
 649                 } elseif ( $name == 'page' ) {
 650                         $this->state = $name;
 651                         if ( $this->atStart ) {
 652                                 $this->egress->writeOpenStream( $this->buffer );
 653                                 $this->buffer = "";
 654                                 $this->atStart = false;
 655                         }
 656                 }
 657
 658                 if ( $name == "text" && isset( $attribs['id'] ) ) {
 659                         $text = $this->getText( $attribs['id'] );
 660                         $this->openElement = array( $name, array( 'xml:space' => 'preserve' ) );
 661                         if ( strlen( $text ) > 0 ) {
 662                                 $this->characterData( $parser, $text );
 663                         }
 664                 } else {
 665                         $this->openElement = array( $name, $attribs );
 666                 }
 667         }
 668
 669         function endElement( $parser, $name ) {
 670                 $this->checkpointJustWritten = false;
 671
 672                 if ( $this->openElement ) {
 673                         $this->clearOpenElement( "" );
 674                 } else {
 675                         $this->buffer .= "</$name>";
 676                 }
 677
 678                 if ( $name == 'revision' ) {
 679                         $this->egress->writeRevision( null, $this->buffer );
 680                         $this->buffer = "";
 681                         $this->thisRev = "";
 682                 } elseif ( $name == 'page' ) {
 683                         if ( ! $this->firstPageWritten ) {
 684                                 $this->firstPageWritten = trim( $this->thisPage );
 685                         }
 686                         $this->lastPageWritten = trim( $this->thisPage );
 687                         if ( $this->timeExceeded ) {
 688                                 $this->egress->writeClosePage( $this->buffer );
 689                                 // nasty hack, we can't just write the chardata after the
 690                                 // page tag, it will include leading blanks from the next line
 691                                 $this->egress->sink->write( "\n" );
 692
 693                                 $this->buffer = $this->xmlwriterobj->closeStream();
 694                                 $this->egress->writeCloseStream( $this->buffer );
 695
 696                                 $this->buffer = "";
 697                                 $this->thisPage = "";
 698                                 // this could be more than one file if we had more than one output arg
 699
 700                                 $filenameList = (array)$this->egress->getFilenames();
 701                                 $newFilenames = array();
 702                                 $firstPageID = str_pad( $this->firstPageWritten, 9, "0", STR_PAD_LEFT );
 703                                 $lastPageID = str_pad( $this->lastPageWritten, 9, "0", STR_PAD_LEFT );
 704                                 for ( $i = 0; $i < count( $filenameList ); $i++ ) {
 705                                         $checkpointNameFilledIn = sprintf( $this->checkpointFiles[$i], $firstPageID, $lastPageID );
 706                                         $fileinfo = pathinfo( $filenameList[$i] );
 707                                         $newFilenames[] = $fileinfo['dirname'] . '/' . $checkpointNameFilledIn;
 708                                 }
 709                                 $this->egress->closeRenameAndReopen( $newFilenames );
 710                                 $this->buffer = $this->xmlwriterobj->openStream();
 711                                 $this->timeExceeded = false;
 712                                 $this->timeOfCheckpoint = $this->lastTime;
 713                                 $this->firstPageWritten = false;
 714                                 $this->checkpointJustWritten = true;
 715                         }
 716                         else {
 717                                 $this->egress->writeClosePage( $this->buffer );
 718                                 $this->buffer = "";
 719                                 $this->thisPage = "";
 720                         }
 721
 722                 } elseif ( $name == 'mediawiki' ) {
 723                         $this->egress->writeCloseStream( $this->buffer );
 724                         $this->buffer = "";
 725                 }
 726         }
 727
 728         function characterData( $parser, $data ) {
 729                 $this->clearOpenElement( null );
 730                 if ( $this->lastName == "id" ) {
 731                         if ( $this->state == "revision" ) {
 732                                 $this->thisRev .= $data;
 733                         } elseif ( $this->state == "page" ) {
 734                                 $this->thisPage .= $data;
 735                         }
 736                 }
 737                 // have to skip the newline left over from closepagetag line of
 738                 // end of checkpoint files. nasty hack!!
 739                 if ( $this->checkpointJustWritten ) {
 740                         if ( $data[0] == "\n" ) {
 741                                 $data = substr( $data, 1 );
 742                         }
 743                         $this->checkpointJustWritten = false;
 744                 }
 745                 $this->buffer .= htmlspecialchars( $data );
 746         }
 747
 748         function clearOpenElement( $style ) {
 749                 if ( $this->openElement ) {
 750                         $this->buffer .= Xml::element( $this->openElement[0], $this->openElement[1], $style );
 751                         $this->openElement = false;
 752                 }
 753         }
 754 }