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  * https://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 __DIR__ . '/backup.inc';
  28
  29 /**
  30  * @ingroup Maintenance
  31  */
  32 class TextPassDumper extends BackupDumper {
  33         public $prefetch = null;
  34         protected $input = "php://stdin";
  35         protected $history = WikiExporter::FULL;
  36         protected $fetchCount = 0;
  37         protected $prefetchCount = 0;
  38         protected $prefetchCountLast = 0;
  39         protected $fetchCountLast = 0;
  40
  41         protected $maxFailures = 5;
  42         protected $maxConsecutiveFailedTextRetrievals = 200;
  43         protected $failureTimeout = 5; // Seconds to sleep after db failure
  44
  45         protected $php = "php";
  46         protected $spawn = false;
  47
  48         /**
  49          * @var bool|resource
  50          */
  51         protected $spawnProc = false;
  52
  53         /**
  54          * @var bool|resource
  55          */
  56         protected $spawnWrite = false;
  57
  58         /**
  59          * @var bool|resource
  60          */
  61         protected $spawnRead = false;
  62
  63         /**
  64          * @var bool|resource
  65          */
  66         protected $spawnErr = false;
  67
  68         protected $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         protected $maxTimeAllowed = 0;  // 0 = no limit
  74         protected $timeExceeded = false;
  75         protected $firstPageWritten = false;
  76         protected $lastPageWritten = false;
  77         protected $checkpointJustWritten = false;
  78         protected $checkpointFiles = array();
  79
  80         /**
  81          * @var DatabaseBase
  82          */
  83         protected $db;
  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__
 123                                 . " rotating DB failed to obtain new load balancer (" . $e->getMessage() . ")" );
 124                 }
 125
 126                 // 2. The Connection, through the load balancer.
 127                 try {
 128                         $this->db = $this->lb->getConnection( DB_SLAVE, 'dump' );
 129                 } catch ( Exception $e ) {
 130                         throw new MWException( __METHOD__
 131                                 . " rotating DB failed to obtain new database (" . $e->getMessage() . ")" );
 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
 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                 $this->readDump( $input );
 173
 174                 if ( $this->spawnProc ) {
 175                         $this->closeSpawn();
 176                 }
 177
 178                 $this->report( true );
 179         }
 180
 181         function processOption( $opt, $val, $param ) {
 182                 global $IP;
 183                 $url = $this->processFileOpt( $val, $param );
 184
 185                 switch ( $opt ) {
 186                 case 'prefetch':
 187                         require_once "$IP/maintenance/backupPrefetch.inc";
 188                         $this->prefetch = new BaseDump( $url );
 189                         break;
 190                 case 'stub':
 191                         $this->input = $url;
 192                         break;
 193                 case 'maxtime':
 194                         $this->maxTimeAllowed = intval( $val ) * 60;
 195                         break;
 196                 case 'checkpointfile':
 197                         $this->checkpointFiles[] = $val;
 198                         break;
 199                 case 'current':
 200                         $this->history = WikiExporter::CURRENT;
 201                         break;
 202                 case 'full':
 203                         $this->history = WikiExporter::FULL;
 204                         break;
 205                 case 'spawn':
 206                         $this->spawn = true;
 207                         if ( $val ) {
 208                                 $this->php = $val;
 209                         }
 210                         break;
 211                 }
 212         }
 213
 214         function processFileOpt( $val, $param ) {
 215                 $fileURIs = explode( ';', $param );
 216                 foreach ( $fileURIs as $URI ) {
 217                         switch ( $val ) {
 218                                 case "file":
 219                                         $newURI = $URI;
 220                                         break;
 221                                 case "gzip":
 222                                         $newURI = "compress.zlib://$URI";
 223                                         break;
 224                                 case "bzip2":
 225                                         $newURI = "compress.bzip2://$URI";
 226                                         break;
 227                                 case "7zip":
 228                                         $newURI = "mediawiki.compress.7z://$URI";
 229                                         break;
 230                                 default:
 231                                         $newURI = $URI;
 232                         }
 233                         $newFileURIs[] = $newURI;
 234                 }
 235                 $val = implode( ';', $newFileURIs );
 236                 return $val;
 237         }
 238
 239         /**
 240          * Overridden to include prefetch ratio if enabled.
 241          */
 242         function showReport() {
 243                 if ( !$this->prefetch ) {
 244                         parent::showReport();
 245                         return;
 246                 }
 247
 248                 if ( $this->reporting ) {
 249                         $now = wfTimestamp( TS_DB );
 250                         $nowts = microtime( true );
 251                         $deltaAll = $nowts - $this->startTime;
 252                         $deltaPart = $nowts - $this->lastTime;
 253                         $this->pageCountPart = $this->pageCount - $this->pageCountLast;
 254                         $this->revCountPart = $this->revCount - $this->revCountLast;
 255
 256                         if ( $deltaAll ) {
 257                                 $portion = $this->revCount / $this->maxCount;
 258                                 $eta = $this->startTime + $deltaAll / $portion;
 259                                 $etats = wfTimestamp( TS_DB, intval( $eta ) );
 260                                 if ( $this->fetchCount ) {
 261                                         $fetchRate = 100.0 * $this->prefetchCount / $this->fetchCount;
 262                                 } else {
 263                                         $fetchRate = '-';
 264                                 }
 265                                 $pageRate = $this->pageCount / $deltaAll;
 266                                 $revRate = $this->revCount / $deltaAll;
 267                         } else {
 268                                 $pageRate = '-';
 269                                 $revRate = '-';
 270                                 $etats = '-';
 271                                 $fetchRate = '-';
 272                         }
 273                         if ( $deltaPart ) {
 274                                 if ( $this->fetchCountLast ) {
 275                                         $fetchRatePart = 100.0 * $this->prefetchCountLast / $this->fetchCountLast;
 276                                 } else {
 277                                         $fetchRatePart = '-';
 278                                 }
 279                                 $pageRatePart = $this->pageCountPart / $deltaPart;
 280                                 $revRatePart = $this->revCountPart / $deltaPart;
 281
 282                         } else {
 283                                 $fetchRatePart = '-';
 284                                 $pageRatePart = '-';
 285                                 $revRatePart = '-';
 286                         }
 287                         $this->progress( sprintf(
 288                                 "%s: %s (ID %d) %d pages (%0.1f|%0.1f/sec all|curr), "
 289                                         . "%d revs (%0.1f|%0.1f/sec all|curr), %0.1f%%|%0.1f%% "
 290                                         . "prefetched (all|curr), ETA %s [max %d]",
 291                                 $now, wfWikiID(), $this->ID, $this->pageCount, $pageRate,
 292                                 $pageRatePart, $this->revCount, $revRate, $revRatePart,
 293                                 $fetchRate, $fetchRatePart, $etats, $this->maxCount
 294                         ) );
 295                         $this->lastTime = $nowts;
 296                         $this->revCountLast = $this->revCount;
 297                         $this->prefetchCountLast = $this->prefetchCount;
 298                         $this->fetchCountLast = $this->fetchCount;
 299                 }
 300         }
 301
 302         function setTimeExceeded() {
 303                 $this->timeExceeded = true;
 304         }
 305
 306         function checkIfTimeExceeded() {
 307                 if ( $this->maxTimeAllowed
 308                         && ( $this->lastTime - $this->timeOfCheckpoint > $this->maxTimeAllowed )
 309                 ) {
 310                         return true;
 311                 }
 312
 313                 return false;
 314         }
 315
 316         function finalOptionCheck() {
 317                 if ( ( $this->checkpointFiles && ! $this->maxTimeAllowed ) ||
 318                         ( $this->maxTimeAllowed && !$this->checkpointFiles ) ) {
 319                         throw new MWException( "Options checkpointfile and maxtime must be specified together.\n" );
 320                 }
 321                 foreach ( $this->checkpointFiles as $checkpointFile ) {
 322                         $count = substr_count ( $checkpointFile, "%s" );
 323                         if ( $count != 2 ) {
 324                                 throw new MWException( "Option checkpointfile must contain two '%s' "
 325                                         . "for substitution of first and last pageids, count is $count instead, "
 326                                         . "file is $checkpointFile.\n" );
 327                         }
 328                 }
 329
 330                 if ( $this->checkpointFiles ) {
 331                         $filenameList = (array)$this->egress->getFilenames();
 332                         if ( count( $filenameList ) != count( $this->checkpointFiles ) ) {
 333                                 throw new MWException( "One checkpointfile must be specified "
 334                                         . "for each output option, if maxtime is used.\n" );
 335                         }
 336                 }
 337         }
 338
 339         /**
 340          * @throws MWException Failure to parse XML input
 341          * @return bool
 342          */
 343         function readDump( $input ) {
 344                 $this->buffer = "";
 345                 $this->openElement = false;
 346                 $this->atStart = true;
 347                 $this->state = "";
 348                 $this->lastName = "";
 349                 $this->thisPage = 0;
 350                 $this->thisRev = 0;
 351
 352                 $parser = xml_parser_create( "UTF-8" );
 353                 xml_parser_set_option( $parser, XML_OPTION_CASE_FOLDING, false );
 354
 355                 xml_set_element_handler(
 356                         $parser,
 357                         array( &$this, 'startElement' ),
 358                         array( &$this, 'endElement' )
 359                 );
 360                 xml_set_character_data_handler( $parser, array( &$this, 'characterData' ) );
 361
 362                 $offset = 0; // for context extraction on error reporting
 363                 $bufferSize = 512 * 1024;
 364                 do {
 365                         if ( $this->checkIfTimeExceeded() ) {
 366                                 $this->setTimeExceeded();
 367                         }
 368                         $chunk = fread( $input, $bufferSize );
 369                         if ( !xml_parse( $parser, $chunk, feof( $input ) ) ) {
 370                                 wfDebug( "TextDumpPass::readDump encountered XML parsing error\n" );
 371
 372                                 $byte = xml_get_current_byte_index( $parser );
 373                                 $msg = wfMessage( 'xml-error-string',
 374                                         'XML import parse failure',
 375                                         xml_get_current_line_number( $parser ),
 376                                         xml_get_current_column_number( $parser ),
 377                                         $byte . ( is_null( $chunk ) ? null : ( '; "' . substr( $chunk, $byte -$offset, 16 ) . '"' ) ),
 378                                         xml_error_string( xml_get_error_code( $parser ) ) )->escaped();
 379
 380                                 xml_parser_free( $parser );
 381
 382                                 throw new MWException( $msg );
 383                         }
 384                         $offset += strlen( $chunk );
 385                 } while ( $chunk !== false && !feof( $input ) );
 386                 if ( $this->maxTimeAllowed ) {
 387                         $filenameList = (array)$this->egress->getFilenames();
 388                         // we wrote some stuff after last checkpoint that needs renamed
 389                         if ( file_exists( $filenameList[0] ) ) {
 390                                 $newFilenames = array();
 391                                 # we might have just written the header and footer and had no
 392                                 # pages or revisions written... perhaps they were all deleted
 393                                 # there's no pageID 0 so we use that. the caller is responsible
 394                                 # for deciding what to do with a file containing only the
 395                                 # siteinfo information and the mw tags.
 396                                 if ( ! $this->firstPageWritten ) {
 397                                         $firstPageID = str_pad( 0, 9, "0", STR_PAD_LEFT );
 398                                         $lastPageID = str_pad( 0, 9, "0", STR_PAD_LEFT );
 399                                 } else {
 400                                         $firstPageID = str_pad( $this->firstPageWritten, 9, "0", STR_PAD_LEFT );
 401                                         $lastPageID = str_pad( $this->lastPageWritten, 9, "0", STR_PAD_LEFT );
 402                                 }
 403
 404                                 $filenameCount = count( $filenameList );
 405                                 for ( $i = 0; $i < $filenameCount; $i++ ) {
 406                                         $checkpointNameFilledIn = sprintf( $this->checkpointFiles[$i], $firstPageID, $lastPageID );
 407                                         $fileinfo = pathinfo( $filenameList[$i] );
 408                                         $newFilenames[] = $fileinfo['dirname'] . '/' . $checkpointNameFilledIn;
 409                                 }
 410                                 $this->egress->closeAndRename( $newFilenames );
 411                         }
 412                 }
 413                 xml_parser_free( $parser );
 414
 415                 return true;
 416         }
 417
 418         /**
 419          * Tries to get the revision text for a revision id.
 420          *
 421          * Upon errors, retries (Up to $this->maxFailures tries each call).
 422          * If still no good revision get could be found even after this retrying, "" is returned.
 423          * If no good revision text could be returned for
 424          * $this->maxConsecutiveFailedTextRetrievals consecutive calls to getText, MWException
 425          * is thrown.
 426          *
 427          * @param string $id The revision id to get the text for
 428          *
 429          * @return string The revision text for $id, or ""
 430          * @throws MWException
 431          */
 432         function getText( $id ) {
 433                 global $wgContentHandlerUseDB;
 434
 435                 $prefetchNotTried = true; // Whether or not we already tried to get the text via prefetch.
 436                 $text = false; // The candidate for a good text. false if no proper value.
 437                 $failures = 0; // The number of times, this invocation of getText already failed.
 438
 439                 // The number of times getText failed without yielding a good text in between.
 440                 static $consecutiveFailedTextRetrievals = 0;
 441
 442                 $this->fetchCount++;
 443
 444                 // To allow to simply return on success and do not have to worry about book keeping,
 445                 // we assume, this fetch works (possible after some retries). Nevertheless, we koop
 446                 // the old value, so we can restore it, if problems occur (See after the while loop).
 447                 $oldConsecutiveFailedTextRetrievals = $consecutiveFailedTextRetrievals;
 448                 $consecutiveFailedTextRetrievals = 0;
 449
 450                 while ( $failures < $this->maxFailures ) {
 451
 452                         // As soon as we found a good text for the $id, we will return immediately.
 453                         // Hence, if we make it past the try catch block, we know that we did not
 454                         // find a good text.
 455
 456                         try {
 457                                 // Step 1: Get some text (or reuse from previous iteratuon if checking
 458                                 //         for plausibility failed)
 459
 460                                 // Trying to get prefetch, if it has not been tried before
 461                                 if ( $text === false && isset( $this->prefetch ) && $prefetchNotTried ) {
 462                                         $prefetchNotTried = false;
 463                                         $tryIsPrefetch = true;
 464                                         $text = $this->prefetch->prefetch( intval( $this->thisPage ),
 465                                                 intval( $this->thisRev ) );
 466                                         if ( $text === null ) {
 467                                                 $text = false;
 468                                         }
 469                                 }
 470
 471                                 if ( $text === false ) {
 472                                         // Fallback to asking the database
 473                                         $tryIsPrefetch = false;
 474                                         if ( $this->spawn ) {
 475                                                 $text = $this->getTextSpawned( $id );
 476                                         } else {
 477                                                 $text = $this->getTextDb( $id );
 478                                         }
 479
 480                                         // No more checks for texts from DB for now.
 481                                         // If we received something that is not false,
 482                                         // We treat it as good text, regardless of whether it actually is or is not
 483                                         if ( $text !== false ) {
 484                                                 return $text;
 485                                         }
 486                                 }
 487
 488                                 if ( $text === false ) {
 489                                         throw new MWException( "Generic error while obtaining text for id " . $id );
 490                                 }
 491
 492                                 // We received a good candidate for the text of $id via some method
 493
 494                                 // Step 2: Checking for plausibility and return the text if it is
 495                                 //         plausible
 496                                 $revID = intval( $this->thisRev );
 497                                 if ( ! isset( $this->db ) ) {
 498                                         throw new MWException( "No database available" );
 499                                 }
 500
 501                                 $revLength = strlen( $text );
 502                                 if ( $wgContentHandlerUseDB ) {
 503                                         $row = $this->db->selectRow(
 504                                                 'revision',
 505                                                 array( 'rev_len', 'rev_content_model' ),
 506                                                 array( 'rev_id' => $revID ),
 507                                                 __METHOD__
 508                                         );
 509                                         if ( $row ) {
 510                                                 // only check the length for the wikitext content handler,
 511                                                 // it's a wasted (and failed) check otherwise
 512                                                 if ( $row->rev_content_model == CONTENT_MODEL_WIKITEXT ) {
 513                                                         $revLength = $row->rev_len;
 514                                                 }
 515                                         }
 516
 517                                 }
 518                                 else {
 519                                         $revLength = $this->db->selectField( 'revision', 'rev_len', array( 'rev_id' => $revID ) );
 520                                 }
 521
 522                                 if ( strlen( $text ) == $revLength ) {
 523                                         if ( $tryIsPrefetch ) {
 524                                                 $this->prefetchCount++;
 525                                         }
 526                                         return $text;
 527                                 }
 528
 529                                 $text = false;
 530                                 throw new MWException( "Received text is unplausible for id " . $id );
 531
 532                         } catch ( Exception $e ) {
 533                                 $msg = "getting/checking text " . $id . " failed (" . $e->getMessage() . ")";
 534                                 if ( $failures + 1 < $this->maxFailures ) {
 535                                         $msg .= " (Will retry " . ( $this->maxFailures - $failures - 1 ) . " more times)";
 536                                 }
 537                                 $this->progress( $msg );
 538                         }
 539
 540                         // Something went wrong; we did not a text that was plausible :(
 541                         $failures++;
 542
 543                         // A failure in a prefetch hit does not warrant resetting db connection etc.
 544                         if ( ! $tryIsPrefetch ) {
 545                                 // After backing off for some time, we try to reboot the whole process as
 546                                 // much as possible to not carry over failures from one part to the other
 547                                 // parts
 548                                 sleep( $this->failureTimeout );
 549                                 try {
 550                                         $this->rotateDb();
 551                                         if ( $this->spawn ) {
 552                                                 $this->closeSpawn();
 553                                                 $this->openSpawn();
 554                                         }
 555                                 } catch ( Exception $e ) {
 556                                         $this->progress( "Rebooting getText infrastructure failed (" . $e->getMessage() . ")" .
 557                                                 " Trying to continue anyways" );
 558                                 }
 559                         }
 560                 }
 561
 562                 // Retirieving a good text for $id failed (at least) maxFailures times.
 563                 // We abort for this $id.
 564
 565                 // Restoring the consecutive failures, and maybe aborting, if the dump
 566                 // is too broken.
 567                 $consecutiveFailedTextRetrievals = $oldConsecutiveFailedTextRetrievals + 1;
 568                 if ( $consecutiveFailedTextRetrievals > $this->maxConsecutiveFailedTextRetrievals ) {
 569                         throw new MWException( "Graceful storage failure" );
 570                 }
 571
 572                 return "";
 573         }
 574
 575         /**
 576          * May throw a database error if, say, the server dies during query.
 577          * @param int $id
 578          * @return bool|string
 579          * @throws MWException
 580          */
 581         private function getTextDb( $id ) {
 582                 global $wgContLang;
 583                 if ( ! isset( $this->db ) ) {
 584                         throw new MWException( __METHOD__ . "No database available" );
 585                 }
 586                 $row = $this->db->selectRow( 'text',
 587                         array( 'old_text', 'old_flags' ),
 588                         array( 'old_id' => $id ),
 589                         __METHOD__ );
 590                 $text = Revision::getRevisionText( $row );
 591                 if ( $text === false ) {
 592                         return false;
 593                 }
 594                 $stripped = str_replace( "\r", "", $text );
 595                 $normalized = $wgContLang->normalize( $stripped );
 596                 return $normalized;
 597         }
 598
 599         private function getTextSpawned( $id ) {
 600                 wfSuppressWarnings();
 601                 if ( !$this->spawnProc ) {
 602                         // First time?
 603                         $this->openSpawn();
 604                 }
 605                 $text = $this->getTextSpawnedOnce( $id );
 606                 wfRestoreWarnings();
 607                 return $text;
 608         }
 609
 610         function openSpawn() {
 611                 global $IP;
 612
 613                 if ( file_exists( "$IP/../multiversion/MWScript.php" ) ) {
 614                         $cmd = implode( " ",
 615                                 array_map( 'wfEscapeShellArg',
 616                                         array(
 617                                                 $this->php,
 618                                                 "$IP/../multiversion/MWScript.php",
 619                                                 "fetchText.php",
 620                                                 '--wiki', wfWikiID() ) ) );
 621                 }
 622                 else {
 623                         $cmd = implode( " ",
 624                                 array_map( 'wfEscapeShellArg',
 625                                         array(
 626                                                 $this->php,
 627                                                 "$IP/maintenance/fetchText.php",
 628                                                 '--wiki', wfWikiID() ) ) );
 629                 }
 630                 $spec = array(
 631                         0 => array( "pipe", "r" ),
 632                         1 => array( "pipe", "w" ),
 633                         2 => array( "file", "/dev/null", "a" ) );
 634                 $pipes = array();
 635
 636                 $this->progress( "Spawning database subprocess: $cmd" );
 637                 $this->spawnProc = proc_open( $cmd, $spec, $pipes );
 638                 if ( !$this->spawnProc ) {
 639                         // shit
 640                         $this->progress( "Subprocess spawn failed." );
 641                         return false;
 642                 }
 643                 list(
 644                         $this->spawnWrite, // -> stdin
 645                         $this->spawnRead,  // <- stdout
 646                 ) = $pipes;
 647
 648                 return true;
 649         }
 650
 651         private function closeSpawn() {
 652                 wfSuppressWarnings();
 653                 if ( $this->spawnRead ) {
 654                         fclose( $this->spawnRead );
 655                 }
 656                 $this->spawnRead = false;
 657                 if ( $this->spawnWrite ) {
 658                         fclose( $this->spawnWrite );
 659                 }
 660                 $this->spawnWrite = false;
 661                 if ( $this->spawnErr ) {
 662                         fclose( $this->spawnErr );
 663                 }
 664                 $this->spawnErr = false;
 665                 if ( $this->spawnProc ) {
 666                         pclose( $this->spawnProc );
 667                 }
 668                 $this->spawnProc = false;
 669                 wfRestoreWarnings();
 670         }
 671
 672         private function getTextSpawnedOnce( $id ) {
 673                 global $wgContLang;
 674
 675                 $ok = fwrite( $this->spawnWrite, "$id\n" );
 676                 // $this->progress( ">> $id" );
 677                 if ( !$ok ) {
 678                         return false;
 679                 }
 680
 681                 $ok = fflush( $this->spawnWrite );
 682                 // $this->progress( ">> [flush]" );
 683                 if ( !$ok ) {
 684                         return false;
 685                 }
 686
 687                 // check that the text id they are sending is the one we asked for
 688                 // this avoids out of sync revision text errors we have encountered in the past
 689                 $newId = fgets( $this->spawnRead );
 690                 if ( $newId === false ) {
 691                         return false;
 692                 }
 693                 if ( $id != intval( $newId ) ) {
 694                         return false;
 695                 }
 696
 697                 $len = fgets( $this->spawnRead );
 698                 // $this->progress( "<< " . trim( $len ) );
 699                 if ( $len === false ) {
 700                         return false;
 701                 }
 702
 703                 $nbytes = intval( $len );
 704                 // actual error, not zero-length text
 705                 if ( $nbytes < 0 ) {
 706                         return false;
 707                 }
 708
 709                 $text = "";
 710
 711                 // Subprocess may not send everything at once, we have to loop.
 712                 while ( $nbytes > strlen( $text ) ) {
 713                         $buffer = fread( $this->spawnRead, $nbytes - strlen( $text ) );
 714                         if ( $buffer === false ) {
 715                                 break;
 716                         }
 717                         $text .= $buffer;
 718                 }
 719
 720                 $gotbytes = strlen( $text );
 721                 if ( $gotbytes != $nbytes ) {
 722                         $this->progress( "Expected $nbytes bytes from database subprocess, got $gotbytes " );
 723                         return false;
 724                 }
 725
 726                 // Do normalization in the dump thread...
 727                 $stripped = str_replace( "\r", "", $text );
 728                 $normalized = $wgContLang->normalize( $stripped );
 729                 return $normalized;
 730         }
 731
 732         function startElement( $parser, $name, $attribs ) {
 733                 $this->checkpointJustWritten = false;
 734
 735                 $this->clearOpenElement( null );
 736                 $this->lastName = $name;
 737
 738                 if ( $name == 'revision' ) {
 739                         $this->state = $name;
 740                         $this->egress->writeOpenPage( null, $this->buffer );
 741                         $this->buffer = "";
 742                 } elseif ( $name == 'page' ) {
 743                         $this->state = $name;
 744                         if ( $this->atStart ) {
 745                                 $this->egress->writeOpenStream( $this->buffer );
 746                                 $this->buffer = "";
 747                                 $this->atStart = false;
 748                         }
 749                 }
 750
 751                 if ( $name == "text" && isset( $attribs['id'] ) ) {
 752                         $text = $this->getText( $attribs['id'] );
 753                         $this->openElement = array( $name, array( 'xml:space' => 'preserve' ) );
 754                         if ( strlen( $text ) > 0 ) {
 755                                 $this->characterData( $parser, $text );
 756                         }
 757                 } else {
 758                         $this->openElement = array( $name, $attribs );
 759                 }
 760         }
 761
 762         function endElement( $parser, $name ) {
 763                 $this->checkpointJustWritten = false;
 764
 765                 if ( $this->openElement ) {
 766                         $this->clearOpenElement( "" );
 767                 } else {
 768                         $this->buffer .= "</$name>";
 769                 }
 770
 771                 if ( $name == 'revision' ) {
 772                         $this->egress->writeRevision( null, $this->buffer );
 773                         $this->buffer = "";
 774                         $this->thisRev = "";
 775                 } elseif ( $name == 'page' ) {
 776                         if ( ! $this->firstPageWritten ) {
 777                                 $this->firstPageWritten = trim( $this->thisPage );
 778                         }
 779                         $this->lastPageWritten = trim( $this->thisPage );
 780                         if ( $this->timeExceeded ) {
 781                                 $this->egress->writeClosePage( $this->buffer );
 782                                 // nasty hack, we can't just write the chardata after the
 783                                 // page tag, it will include leading blanks from the next line
 784                                 $this->egress->sink->write( "\n" );
 785
 786                                 $this->buffer = $this->xmlwriterobj->closeStream();
 787                                 $this->egress->writeCloseStream( $this->buffer );
 788
 789                                 $this->buffer = "";
 790                                 $this->thisPage = "";
 791                                 // this could be more than one file if we had more than one output arg
 792
 793                                 $filenameList = (array)$this->egress->getFilenames();
 794                                 $newFilenames = array();
 795                                 $firstPageID = str_pad( $this->firstPageWritten, 9, "0", STR_PAD_LEFT );
 796                                 $lastPageID = str_pad( $this->lastPageWritten, 9, "0", STR_PAD_LEFT );
 797                                 $filenamesCount = count( $filenameList );
 798                                 for ( $i = 0; $i < $filenamesCount; $i++ ) {
 799                                         $checkpointNameFilledIn = sprintf( $this->checkpointFiles[$i], $firstPageID, $lastPageID );
 800                                         $fileinfo = pathinfo( $filenameList[$i] );
 801                                         $newFilenames[] = $fileinfo['dirname'] . '/' . $checkpointNameFilledIn;
 802                                 }
 803                                 $this->egress->closeRenameAndReopen( $newFilenames );
 804                                 $this->buffer = $this->xmlwriterobj->openStream();
 805                                 $this->timeExceeded = false;
 806                                 $this->timeOfCheckpoint = $this->lastTime;
 807                                 $this->firstPageWritten = false;
 808                                 $this->checkpointJustWritten = true;
 809                         } else {
 810                                 $this->egress->writeClosePage( $this->buffer );
 811                                 $this->buffer = "";
 812                                 $this->thisPage = "";
 813                         }
 814
 815                 } elseif ( $name == 'mediawiki' ) {
 816                         $this->egress->writeCloseStream( $this->buffer );
 817                         $this->buffer = "";
 818                 }
 819         }
 820
 821         function characterData( $parser, $data ) {
 822                 $this->clearOpenElement( null );
 823                 if ( $this->lastName == "id" ) {
 824                         if ( $this->state == "revision" ) {
 825                                 $this->thisRev .= $data;
 826                         } elseif ( $this->state == "page" ) {
 827                                 $this->thisPage .= $data;
 828                         }
 829                 }
 830                 // have to skip the newline left over from closepagetag line of
 831                 // end of checkpoint files. nasty hack!!
 832                 if ( $this->checkpointJustWritten ) {
 833                         if ( $data[0] == "\n" ) {
 834                                 $data = substr( $data, 1 );
 835                         }
 836                         $this->checkpointJustWritten = false;
 837                 }
 838                 $this->buffer .= htmlspecialchars( $data );
 839         }
 840
 841         function clearOpenElement( $style ) {
 842                 if ( $this->openElement ) {
 843                         $this->buffer .= Xml::element( $this->openElement[0], $this->openElement[1], $style );
 844                         $this->openElement = false;
 845                 }
 846         }
 847 }