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