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 require_once( dirname( __FILE__ ) . '/backup.inc' );
  28
  29 /**
  30  * @ingroup Maintenance
  31  */
  32 class TextPassDumper extends BackupDumper {
  33         var $prefetch = null;
  34         var $input = "php://stdin";
  35         var $history = WikiExporter::FULL;
  36         var $fetchCount = 0;
  37         var $prefetchCount = 0;
  38         var $prefetchCountLast = 0;
  39         var $fetchCountLast = 0;
  40
  41         var $maxFailures = 5;
  42         var $maxConsecutiveFailedTextRetrievals = 200;
  43         var $failureTimeout = 5; // Seconds to sleep after db failure
  44
  45         var $php = "php";
  46         var $spawn = false;
  47
  48         /**
  49          * @var bool|resource
  50          */
  51         var $spawnProc = false;
  52
  53         /**
  54          * @var bool|resource
  55          */
  56         var $spawnWrite = false;
  57
  58         /**
  59          * @var bool|resource
  60          */
  61         var $spawnRead = false;
  62
  63         /**
  64          * @var bool|resource
  65          */
  66         var $spawnErr = false;
  67
  68         var $xmlwriterobj = false;
  69
  70         // when we spend more than maxTimeAllowed seconds on this run, we continue
  71         // processing until we write out the next complete page, then save output file(s),
  72         // rename it/them and open new one(s)
  73         var $maxTimeAllowed = 0;  // 0 = no limit
  74         var $timeExceeded = false;
  75         var $firstPageWritten = false;
  76         var $lastPageWritten = false;
  77         var $checkpointJustWritten = false;
  78         var $checkpointFiles = array();
  79
  80         /**
  81          * @var DatabaseBase
  82          */
  83         protected $db;
  84
  85
  86         /**
  87          * Drop the database connection $this->db and try to get a new one.
  88          *
  89          * This function tries to get a /different/ connection if this is
  90          * possible. Hence, (if this is possible) it switches to a different
  91          * failover upon each call.
  92          *
  93          * This function resets $this->lb and closes all connections on it.
  94          *
  95          * @throws MWException
  96          */
  97         function rotateDb() {
  98                 // Cleaning up old connections
  99                 if ( isset( $this->lb ) ) {
 100                         $this->lb->closeAll();
 101                         unset( $this->lb );
 102                 }
 103
 104                 if ( $this->forcedDb !== null ) {
 105                         $this->db = $this->forcedDb;
 106                         return;
 107                 }
 108
 109                 if ( isset( $this->db ) && $this->db->isOpen() ) {
 110                         throw new MWException( 'DB is set and has not been closed by the Load Balancer' );
 111                 }
 112
 113                 unset( $this->db );
 114
 115                 // Trying to set up new connection.
 116                 // We do /not/ retry upon failure, but delegate to encapsulating logic, to avoid
 117                 // individually retrying at different layers of code.
 118
 119                 // 1. The LoadBalancer.
 120                 try {
 121                         $this->lb = wfGetLBFactory()->newMainLB();
 122                 } catch ( Exception $e ) {
 123                         throw new MWException( __METHOD__ . " rotating DB failed to obtain new load balancer (" . $e->getMessage() . ")" );
 124                 }
 125
 126
 127                 // 2. The Connection, through the load balancer.
 128                 try {
 129                         $this->db = $this->lb->getConnection( DB_SLAVE, 'backup' );
 130                 } catch ( Exception $e ) {
 131                         throw new MWException( __METHOD__ . " rotating DB failed to obtain new database (" . $e->getMessage() . ")" );
 132                 }
 133         }
 134
 135
 136         function initProgress( $history = WikiExporter::FULL ) {
 137                 parent::initProgress();
 138                 $this->timeOfCheckpoint = $this->startTime;
 139         }
 140
 141         function dump( $history, $text = WikiExporter::TEXT ) {
 142                 // Notice messages will foul up your XML output even if they're
 143                 // relatively harmless.
 144                 if ( ini_get( 'display_errors' ) )
 145                         ini_set( 'display_errors', 'stderr' );
 146
 147                 $this->initProgress( $this->history );
 148
 149                 // We are trying to get an initial database connection to avoid that the
 150                 // first try of this request's first call to getText fails. However, if
 151                 // obtaining a good DB connection fails it's not a serious issue, as
 152                 // getText does retry upon failure and can start without having a working
 153                 // DB connection.
 154                 try {
 155                         $this->rotateDb();
 156                 } catch ( Exception $e ) {
 157                         // We do not even count this as failure. Just let eventual
 158                         // watchdogs know.
 159                         $this->progress( "Getting initial DB connection failed (" .
 160                                 $e->getMessage() . ")" );
 161                 }
 162
 163                 $this->egress = new ExportProgressFilter( $this->sink, $this );
 164
 165                 // it would be nice to do it in the constructor, oh well. need egress set
 166                 $this->finalOptionCheck();
 167
 168                 // we only want this so we know how to close a stream :-P
 169                 $this->xmlwriterobj = new XmlDumpWriter();
 170
 171                 $input = fopen( $this->input, "rt" );
 172                 $result = $this->readDump( $input );
 173
 174                 if ( WikiError::isError( $result ) ) {
 175                         throw new MWException( $result->getMessage() );
 176                 }
 177
 178                 if ( $this->spawnProc ) {
 179                         $this->closeSpawn();
 180                 }
 181
 182                 $this->report( true );
 183         }
 184
 185         function processOption( $opt, $val, $param ) {
 186                 global $IP;
 187                 $url = $this->processFileOpt( $val, $param );
 188
 189                 switch( $opt ) {
 190                 case 'prefetch':
 191                         require_once "$IP/maintenance/backupPrefetch.inc";
 192                         $this->prefetch = new BaseDump( $url );
 193                         break;
 194                 case 'stub':
 195                         $this->input = $url;
 196                         break;
 197                 case 'maxtime':
 198                         $this->maxTimeAllowed = intval( $val ) * 60;
 199                         break;
 200                 case 'checkpointfile':
 201                         $this->checkpointFiles[] = $val;
 202                         break;
 203                 case 'current':
 204                         $this->history = WikiExporter::CURRENT;
 205                         break;
 206                 case 'full':
 207                         $this->history = WikiExporter::FULL;
 208                         break;
 209                 case 'spawn':
 210                         $this->spawn = true;
 211                         if ( $val ) {
 212                                 $this->php = $val;
 213                         }
 214                         break;
 215                 }
 216         }
 217
 218         function processFileOpt( $val, $param ) {
 219                 $fileURIs = explode( ';', $param );
 220                 foreach ( $fileURIs as $URI ) {
 221                         switch( $val ) {
 222                                 case "file":
 223                                         $newURI = $URI;
 224                                         break;
 225                                 case "gzip":
 226                                         $newURI = "compress.zlib://$URI";
 227                                         break;
 228                                 case "bzip2":
 229                                         $newURI = "compress.bzip2://$URI";
 230                                         break;
 231                                 case "7zip":
 232                                         $newURI = "mediawiki.compress.7z://$URI";
 233                                         break;
 234                                 default:
 235                                         $newURI = $URI;
 236                         }
 237                         $newFileURIs[] = $newURI;
 238                 }
 239                 $val = implode( ';', $newFileURIs );
 240                 return $val;
 241         }
 242
 243         /**
 244          * Overridden to include prefetch ratio if enabled.
 245          */
 246         function showReport() {
 247                 if ( !$this->prefetch ) {
 248                         parent::showReport();
 249                         return;
 250                 }
 251
 252                 if ( $this->reporting ) {
 253                         $now = wfTimestamp( TS_DB );
 254                         $nowts = wfTime();
 255                         $deltaAll = wfTime() - $this->startTime;
 256                         $deltaPart = wfTime() - $this->lastTime;
 257                         $this->pageCountPart = $this->pageCount - $this->pageCountLast;
 258                         $this->revCountPart = $this->revCount - $this->revCountLast;
 259
 260                         if ( $deltaAll ) {
 261                                 $portion = $this->revCount / $this->maxCount;
 262                                 $eta = $this->startTime + $deltaAll / $portion;
 263                                 $etats = wfTimestamp( TS_DB, intval( $eta ) );
 264                                 if ( $this->fetchCount ) {
 265                                         $fetchRate = 100.0 * $this->prefetchCount / $this->fetchCount;
 266                                 } else {
 267                                         $fetchRate = '-';
 268                                 }
 269                                 $pageRate = $this->pageCount / $deltaAll;
 270                                 $revRate = $this->revCount / $deltaAll;
 271                         } else {
 272                                 $pageRate = '-';
 273                                 $revRate = '-';
 274                                 $etats = '-';
 275                                 $fetchRate = '-';
 276                         }
 277                         if ( $deltaPart ) {
 278                                 if ( $this->fetchCountLast ) {
 279                                         $fetchRatePart = 100.0 * $this->prefetchCountLast / $this->fetchCountLast;
 280                                 } else {
 281                                         $fetchRatePart = '-';
 282                                 }
 283                                 $pageRatePart = $this->pageCountPart / $deltaPart;
 284                                 $revRatePart = $this->revCountPart / $deltaPart;
 285
 286                         } else {
 287                                 $fetchRatePart = '-';
 288                                 $pageRatePart = '-';
 289                                 $revRatePart = '-';
 290                         }
 291                         $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]",
 292                                         $now, wfWikiID(), $this->ID, $this->pageCount, $pageRate, $pageRatePart, $this->revCount, $revRate, $revRatePart, $fetchRate, $fetchRatePart, $etats, $this->maxCount ) );
 293                         $this->lastTime = $nowts;
 294                         $this->revCountLast = $this->revCount;
 295                         $this->prefetchCountLast = $this->prefetchCount;
 296                         $this->fetchCountLast = $this->fetchCount;
 297                 }
 298         }
 299
 300         function setTimeExceeded() {
 301                 $this->timeExceeded = True;
 302         }
 303
 304         function checkIfTimeExceeded() {
 305                 if ( $this->maxTimeAllowed &&  ( $this->lastTime - $this->timeOfCheckpoint  > $this->maxTimeAllowed ) ) {
 306                         return true;
 307                 }
 308                 return false;
 309         }
 310
 311         function finalOptionCheck() {
 312                 if ( ( $this->checkpointFiles && ! $this->maxTimeAllowed ) ||
 313                         ( $this->maxTimeAllowed && !$this->checkpointFiles ) ) {
 314                         throw new MWException( "Options checkpointfile and maxtime must be specified together.\n" );
 315                 }
 316                 foreach ( $this->checkpointFiles as $checkpointFile ) {
 317                         $count = substr_count ( $checkpointFile, "%s" );
 318                         if ( $count != 2 ) {
 319                                 throw new MWException( "Option checkpointfile must contain two '%s' for substitution of first and last pageids, count is $count instead, file is $checkpointFile.\n" );
 320                         }
 321                 }
 322
 323                 if ( $this->checkpointFiles ) {
 324                         $filenameList = (array)$this->egress->getFilenames();
 325                         if ( count( $filenameList ) != count( $this->checkpointFiles ) ) {
 326                                 throw new MWException( "One checkpointfile must be specified for each output option, if maxtime is used.\n" );
 327                         }
 328                 }
 329         }
 330
 331         function readDump( $input ) {
 332                 $this->buffer = "";
 333                 $this->openElement = false;
 334                 $this->atStart = true;
 335                 $this->state = "";
 336                 $this->lastName = "";
 337                 $this->thisPage = 0;
 338                 $this->thisRev = 0;
 339
 340                 $parser = xml_parser_create( "UTF-8" );
 341                 xml_parser_set_option( $parser, XML_OPTION_CASE_FOLDING, false );
 342
 343                 xml_set_element_handler( $parser, array( &$this, 'startElement' ), array( &$this, 'endElement' ) );
 344                 xml_set_character_data_handler( $parser, array( &$this, 'characterData' ) );
 345
 346                 $offset = 0; // for context extraction on error reporting
 347                 $bufferSize = 512 * 1024;
 348                 do {
 349                         if ( $this->checkIfTimeExceeded() ) {
 350                                 $this->setTimeExceeded();
 351                         }
 352                         $chunk = fread( $input, $bufferSize );
 353                         if ( !xml_parse( $parser, $chunk, feof( $input ) ) ) {
 354                                 wfDebug( "TextDumpPass::readDump encountered XML parsing error\n" );
 355                                 return new WikiXmlError( $parser, 'XML import parse failure', $chunk, $offset );
 356                         }
 357                         $offset += strlen( $chunk );
 358                 } while ( $chunk !== false && !feof( $input ) );
 359                 if ( $this->maxTimeAllowed ) {
 360                         $filenameList = (array)$this->egress->getFilenames();
 361                         // we wrote some stuff after last checkpoint that needs renamed
 362                         if ( file_exists( $filenameList[0] ) ) {
 363                                 $newFilenames = array();
 364                                 # we might have just written the header and footer and had no
 365                                 # pages or revisions written... perhaps they were all deleted
 366                                 # there's no pageID 0 so we use that. the caller is responsible
 367                                 # for deciding what to do with a file containing only the
 368                                 # siteinfo information and the mw tags.
 369                                 if ( ! $this->firstPageWritten ) {
 370                                         $firstPageID = str_pad( 0, 9, "0", STR_PAD_LEFT );
 371                                         $lastPageID = str_pad( 0, 9, "0", STR_PAD_LEFT );
 372                                 }
 373                                 else {
 374                                         $firstPageID = str_pad( $this->firstPageWritten, 9, "0", STR_PAD_LEFT );
 375                                         $lastPageID = str_pad( $this->lastPageWritten, 9, "0", STR_PAD_LEFT );
 376                                 }
 377                                 for ( $i = 0; $i < count( $filenameList ); $i++ ) {
 378                                         $checkpointNameFilledIn = sprintf( $this->checkpointFiles[$i], $firstPageID, $lastPageID );
 379                                         $fileinfo = pathinfo( $filenameList[$i] );
 380                                         $newFilenames[] = $fileinfo['dirname'] . '/' . $checkpointNameFilledIn;
 381                                 }
 382                                 $this->egress->closeAndRename( $newFilenames );
 383                         }
 384                 }
 385                 xml_parser_free( $parser );
 386
 387                 return true;
 388         }
 389
 390         /**
 391          * Tries to get the revision text for a revision id.
 392          *
 393          * Upon errors, retries (Up to $this->maxFailures tries each call).
 394          * If still no good revision get could be found even after this retrying, "" is returned.
 395          * If no good revision text could be returned for
 396          * $this->maxConsecutiveFailedTextRetrievals consecutive calls to getText, MWException
 397          * is thrown.
 398          *
 399          * @param $id string The revision id to get the text for
 400          *
 401          * @return string The revision text for $id, or ""
 402          * @throws MWException
 403          */
 404         function getText( $id ) {
 405                 $prefetchNotTried = true; // Whether or not we already tried to get the text via prefetch.
 406                 $text = false; // The candidate for a good text. false if no proper value.
 407                 $failures = 0; // The number of times, this invocation of getText already failed.
 408
 409                 static $consecutiveFailedTextRetrievals = 0; // The number of times getText failed without
 410                                                              // yielding a good text in between.
 411
 412                 $this->fetchCount++;
 413
 414                 // To allow to simply return on success and do not have to worry about book keeping,
 415                 // we assume, this fetch works (possible after some retries). Nevertheless, we koop
 416                 // the old value, so we can restore it, if problems occur (See after the while loop).
 417                 $oldConsecutiveFailedTextRetrievals = $consecutiveFailedTextRetrievals;
 418                 $consecutiveFailedTextRetrievals = 0;
 419
 420                 while ( $failures < $this->maxFailures ) {
 421
 422                         // As soon as we found a good text for the $id, we will return immediately.
 423                         // Hence, if we make it past the try catch block, we know that we did not
 424                         // find a good text.
 425
 426                         try {
 427                                 // Step 1: Get some text (or reuse from previous iteratuon if checking
 428                                 //         for plausibility failed)
 429
 430                                 // Trying to get prefetch, if it has not been tried before
 431                                 if ( $text === false && isset( $this->prefetch ) && $prefetchNotTried ) {
 432                                         $prefetchNotTried = false;
 433                                         $tryIsPrefetch = true;
 434                                         $text = $this->prefetch->prefetch( intval( $this->thisPage ),
 435                                                 intval( $this->thisRev ) );
 436                                         if ( $text === null ) {
 437                                                 $text = false;
 438                                         }
 439                                 }
 440
 441                                 if ( $text === false ) {
 442                                         // Fallback to asking the database
 443                                         $tryIsPrefetch = false;
 444                                         if ( $this->spawn ) {
 445                                                 $text = $this->getTextSpawned( $id );
 446                                         } else {
 447                                                 $text = $this->getTextDb( $id );
 448                                         }
 449
 450                                         // No more checks for texts from DB for now.
 451                                         // If we received something that is not false,
 452                                         // We treat it as good text, regardless of whether it actually is or is not
 453                                         if ( $text !== false ) {
 454                                                 return $text;
 455                                         }
 456                                 }
 457
 458                                 if ( $text === false ) {
 459                                         throw new MWException( "Generic error while obtaining text for id " . $id );
 460                                 }
 461
 462                                 // We received a good candidate for the text of $id via some method
 463
 464                                 // Step 2: Checking for plausibility and return the text if it is
 465                                 //         plausible
 466                                 $revID = intval( $this->thisRev );
 467                                 if ( ! isset( $this->db ) ) {
 468                                         throw new MWException( "No database available" );
 469                                 }
 470                                 $revLength = $this->db->selectField( 'revision', 'rev_len', array( 'rev_id' => $revID ) );
 471                                 if ( strlen( $text ) == $revLength ) {
 472                                         if ( $tryIsPrefetch ) {
 473                                                 $this->prefetchCount++;
 474                                         }
 475                                         return $text;
 476                                 }
 477
 478                                 $text = false;
 479                                 throw new MWException( "Received text is unplausible for id " . $id );
 480
 481                         } catch ( Exception $e ) {
 482                                 $msg = "getting/checking text " . $id . " failed (" . $e->getMessage() . ")";
 483                                 if ( $failures + 1 < $this->maxFailures ) {
 484                                         $msg .= " (Will retry " . ( $this->maxFailures - $failures - 1 ) . " more times)";
 485                                 }
 486                                 $this->progress( $msg );
 487                         }
 488
 489                         // Something went wrong; we did not a text that was plausible :(
 490                         $failures++;
 491
 492                         // A failure in a prefetch hit does not warrant resetting db connection etc.
 493                         if ( ! $tryIsPrefetch ) {
 494                                 // After backing off for some time, we try to reboot the whole process as
 495                                 // much as possible to not carry over failures from one part to the other
 496                                 // parts
 497                                 sleep( $this->failureTimeout );
 498                                 try {
 499                                         $this->rotateDb();
 500                                         if ( $this->spawn ) {
 501                                                 $this->closeSpawn();
 502                                                 $this->openSpawn();
 503                                         }
 504                                 } catch ( Exception $e ) {
 505                                         $this->progress( "Rebooting getText infrastructure failed (" . $e->getMessage() . ")" .
 506                                                 " Trying to continue anyways" );
 507                                 }
 508                         }
 509                 }
 510
 511                 // Retirieving a good text for $id failed (at least) maxFailures times.
 512                 // We abort for this $id.
 513
 514                 // Restoring the consecutive failures, and maybe aborting, if the dump
 515                 // is too broken.
 516                 $consecutiveFailedTextRetrievals = $oldConsecutiveFailedTextRetrievals + 1;
 517                 if ( $consecutiveFailedTextRetrievals > $this->maxConsecutiveFailedTextRetrievals ) {
 518                         throw new MWException( "Graceful storage failure" );
 519                 }
 520
 521                 return "";
 522         }
 523
 524
 525         /**
 526          * May throw a database error if, say, the server dies during query.
 527          * @param $id
 528          * @return bool|string
 529          * @throws MWException
 530          */
 531         private function getTextDb( $id ) {
 532                 global $wgContLang;
 533                 if ( ! isset( $this->db ) ) {
 534                         throw new MWException( __METHOD__ . "No database available" );
 535                 }
 536                 $row = $this->db->selectRow( 'text',
 537                         array( 'old_text', 'old_flags' ),
 538                         array( 'old_id' => $id ),
 539                         __METHOD__ );
 540                 $text = Revision::getRevisionText( $row );
 541                 if ( $text === false ) {
 542                         return false;
 543                 }
 544                 $stripped = str_replace( "\r", "", $text );
 545                 $normalized = $wgContLang->normalize( $stripped );
 546                 return $normalized;
 547         }
 548
 549         private function getTextSpawned( $id ) {
 550                 wfSuppressWarnings();
 551                 if ( !$this->spawnProc ) {
 552                         // First time?
 553                         $this->openSpawn();
 554                 }
 555                 $text = $this->getTextSpawnedOnce( $id );
 556                 wfRestoreWarnings();
 557                 return $text;
 558         }
 559
 560         function openSpawn() {
 561                 global $IP;
 562
 563                 if ( file_exists( "$IP/../multiversion/MWScript.php" ) ) {
 564                         $cmd = implode( " ",
 565                                 array_map( 'wfEscapeShellArg',
 566                                         array(
 567                                                 $this->php,
 568                                                 "$IP/../multiversion/MWScript.php",
 569                                                 "fetchText.php",
 570                                                 '--wiki', wfWikiID() ) ) );
 571                 }
 572                 else {
 573                         $cmd = implode( " ",
 574                                 array_map( 'wfEscapeShellArg',
 575                                         array(
 576                                                 $this->php,
 577                                                 "$IP/maintenance/fetchText.php",
 578                                                 '--wiki', wfWikiID() ) ) );
 579                 }
 580                 $spec = array(
 581                         0 => array( "pipe", "r" ),
 582                         1 => array( "pipe", "w" ),
 583                         2 => array( "file", "/dev/null", "a" ) );
 584                 $pipes = array();
 585
 586                 $this->progress( "Spawning database subprocess: $cmd" );
 587                 $this->spawnProc = proc_open( $cmd, $spec, $pipes );
 588                 if ( !$this->spawnProc ) {
 589                         // shit
 590                         $this->progress( "Subprocess spawn failed." );
 591                         return false;
 592                 }
 593                 list(
 594                         $this->spawnWrite, // -> stdin
 595                         $this->spawnRead,  // <- stdout
 596                 ) = $pipes;
 597
 598                 return true;
 599         }
 600
 601         private function closeSpawn() {
 602                 wfSuppressWarnings();
 603                 if ( $this->spawnRead )
 604                         fclose( $this->spawnRead );
 605                 $this->spawnRead = false;
 606                 if ( $this->spawnWrite )
 607                         fclose( $this->spawnWrite );
 608                 $this->spawnWrite = false;
 609                 if ( $this->spawnErr )
 610                         fclose( $this->spawnErr );
 611                 $this->spawnErr = false;
 612                 if ( $this->spawnProc )
 613                         pclose( $this->spawnProc );
 614                 $this->spawnProc = false;
 615                 wfRestoreWarnings();
 616         }
 617
 618         private function getTextSpawnedOnce( $id ) {
 619                 global $wgContLang;
 620
 621                 $ok = fwrite( $this->spawnWrite, "$id\n" );
 622                 // $this->progress( ">> $id" );
 623                 if ( !$ok ) return false;
 624
 625                 $ok = fflush( $this->spawnWrite );
 626                 // $this->progress( ">> [flush]" );
 627                 if ( !$ok ) return false;
 628
 629                 // check that the text id they are sending is the one we asked for
 630                 // this avoids out of sync revision text errors we have encountered in the past
 631                 $newId = fgets( $this->spawnRead );
 632                 if ( $newId === false ) {
 633                         return false;
 634                 }
 635                 if ( $id != intval( $newId ) ) {
 636                         return false;
 637                 }
 638
 639                 $len = fgets( $this->spawnRead );
 640                 // $this->progress( "<< " . trim( $len ) );
 641                 if ( $len === false ) return false;
 642
 643                 $nbytes = intval( $len );
 644                 // actual error, not zero-length text
 645                 if ( $nbytes < 0 ) return false;
 646
 647                 $text = "";
 648
 649                 // Subprocess may not send everything at once, we have to loop.
 650                 while ( $nbytes > strlen( $text ) ) {
 651                         $buffer = fread( $this->spawnRead, $nbytes - strlen( $text ) );
 652                         if ( $buffer === false ) break;
 653                         $text .= $buffer;
 654                 }
 655
 656                 $gotbytes = strlen( $text );
 657                 if ( $gotbytes != $nbytes ) {
 658                         $this->progress( "Expected $nbytes bytes from database subprocess, got $gotbytes " );
 659                         return false;
 660                 }
 661
 662                 // Do normalization in the dump thread...
 663                 $stripped = str_replace( "\r", "", $text );
 664                 $normalized = $wgContLang->normalize( $stripped );
 665                 return $normalized;
 666         }
 667
 668         function startElement( $parser, $name, $attribs ) {
 669                 $this->checkpointJustWritten = false;
 670
 671                 $this->clearOpenElement( null );
 672                 $this->lastName = $name;
 673
 674                 if ( $name == 'revision' ) {
 675                         $this->state = $name;
 676                         $this->egress->writeOpenPage( null, $this->buffer );
 677                         $this->buffer = "";
 678                 } elseif ( $name == 'page' ) {
 679                         $this->state = $name;
 680                         if ( $this->atStart ) {
 681                                 $this->egress->writeOpenStream( $this->buffer );
 682                                 $this->buffer = "";
 683                                 $this->atStart = false;
 684                         }
 685                 }
 686
 687                 if ( $name == "text" && isset( $attribs['id'] ) ) {
 688                         $text = $this->getText( $attribs['id'] );
 689                         $this->openElement = array( $name, array( 'xml:space' => 'preserve' ) );
 690                         if ( strlen( $text ) > 0 ) {
 691                                 $this->characterData( $parser, $text );
 692                         }
 693                 } else {
 694                         $this->openElement = array( $name, $attribs );
 695                 }
 696         }
 697
 698         function endElement( $parser, $name ) {
 699                 $this->checkpointJustWritten = false;
 700
 701                 if ( $this->openElement ) {
 702                         $this->clearOpenElement( "" );
 703                 } else {
 704                         $this->buffer .= "</$name>";
 705                 }
 706
 707                 if ( $name == 'revision' ) {
 708                         $this->egress->writeRevision( null, $this->buffer );
 709                         $this->buffer = "";
 710                         $this->thisRev = "";
 711                 } elseif ( $name == 'page' ) {
 712                         if ( ! $this->firstPageWritten ) {
 713                                 $this->firstPageWritten = trim( $this->thisPage );
 714                         }
 715                         $this->lastPageWritten = trim( $this->thisPage );
 716                         if ( $this->timeExceeded ) {
 717                                 $this->egress->writeClosePage( $this->buffer );
 718                                 // nasty hack, we can't just write the chardata after the
 719                                 // page tag, it will include leading blanks from the next line
 720                                 $this->egress->sink->write( "\n" );
 721
 722                                 $this->buffer = $this->xmlwriterobj->closeStream();
 723                                 $this->egress->writeCloseStream( $this->buffer );
 724
 725                                 $this->buffer = "";
 726                                 $this->thisPage = "";
 727                                 // this could be more than one file if we had more than one output arg
 728
 729                                 $filenameList = (array)$this->egress->getFilenames();
 730                                 $newFilenames = array();
 731                                 $firstPageID = str_pad( $this->firstPageWritten, 9, "0", STR_PAD_LEFT );
 732                                 $lastPageID = str_pad( $this->lastPageWritten, 9, "0", STR_PAD_LEFT );
 733                                 for ( $i = 0; $i < count( $filenameList ); $i++ ) {
 734                                         $checkpointNameFilledIn = sprintf( $this->checkpointFiles[$i], $firstPageID, $lastPageID );
 735                                         $fileinfo = pathinfo( $filenameList[$i] );
 736                                         $newFilenames[] = $fileinfo['dirname'] . '/' . $checkpointNameFilledIn;
 737                                 }
 738                                 $this->egress->closeRenameAndReopen( $newFilenames );
 739                                 $this->buffer = $this->xmlwriterobj->openStream();
 740                                 $this->timeExceeded = false;
 741                                 $this->timeOfCheckpoint = $this->lastTime;
 742                                 $this->firstPageWritten = false;
 743                                 $this->checkpointJustWritten = true;
 744                         }
 745                         else {
 746                                 $this->egress->writeClosePage( $this->buffer );
 747                                 $this->buffer = "";
 748                                 $this->thisPage = "";
 749                         }
 750
 751                 } elseif ( $name == 'mediawiki' ) {
 752                         $this->egress->writeCloseStream( $this->buffer );
 753                         $this->buffer = "";
 754                 }
 755         }
 756
 757         function characterData( $parser, $data ) {
 758                 $this->clearOpenElement( null );
 759                 if ( $this->lastName == "id" ) {
 760                         if ( $this->state == "revision" ) {
 761                                 $this->thisRev .= $data;
 762                         } elseif ( $this->state == "page" ) {
 763                                 $this->thisPage .= $data;
 764                         }
 765                 }
 766                 // have to skip the newline left over from closepagetag line of
 767                 // end of checkpoint files. nasty hack!!
 768                 if ( $this->checkpointJustWritten ) {
 769                         if ( $data[0] == "\n" ) {
 770                                 $data = substr( $data, 1 );
 771                         }
 772                         $this->checkpointJustWritten = false;
 773                 }
 774                 $this->buffer .= htmlspecialchars( $data );
 775         }
 776
 777         function clearOpenElement( $style ) {
 778                 if ( $this->openElement ) {
 779                         $this->buffer .= Xml::element( $this->openElement[0], $this->openElement[1], $style );
 780                         $this->openElement = false;
 781                 }
 782         }
 783 }