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 __DIR__ . '/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
 148                 $this->initProgress( $this->history );
 149
 150                 // We are trying to get an initial database connection to avoid that the
 151                 // first try of this request's first call to getText fails. However, if
 152                 // obtaining a good DB connection fails it's not a serious issue, as
 153                 // getText does retry upon failure and can start without having a working
 154                 // DB connection.
 155                 try {
 156                         $this->rotateDb();
 157                 } catch ( Exception $e ) {
 158                         // We do not even count this as failure. Just let eventual
 159                         // watchdogs know.
 160                         $this->progress( "Getting initial DB connection failed (" .
 161                                 $e->getMessage() . ")" );
 162                 }
 163
 164                 $this->egress = new ExportProgressFilter( $this->sink, $this );
 165
 166                 // it would be nice to do it in the constructor, oh well. need egress set
 167                 $this->finalOptionCheck();
 168
 169                 // we only want this so we know how to close a stream :-P
 170                 $this->xmlwriterobj = new XmlDumpWriter();
 171
 172                 $input = fopen( $this->input, "rt" );
 173                 $this->readDump( $input );
 174
 175                 if ( $this->spawnProc ) {
 176                         $this->closeSpawn();
 177                 }
 178
 179                 $this->report( true );
 180         }
 181
 182         function processOption( $opt, $val, $param ) {
 183                 global $IP;
 184                 $url = $this->processFileOpt( $val, $param );
 185
 186                 switch ( $opt ) {
 187                 case 'prefetch':
 188                         require_once "$IP/maintenance/backupPrefetch.inc";
 189                         $this->prefetch = new BaseDump( $url );
 190                         break;
 191                 case 'stub':
 192                         $this->input = $url;
 193                         break;
 194                 case 'maxtime':
 195                         $this->maxTimeAllowed = intval( $val ) * 60;
 196                         break;
 197                 case 'checkpointfile':
 198                         $this->checkpointFiles[] = $val;
 199                         break;
 200                 case 'current':
 201                         $this->history = WikiExporter::CURRENT;
 202                         break;
 203                 case 'full':
 204                         $this->history = WikiExporter::FULL;
 205                         break;
 206                 case 'spawn':
 207                         $this->spawn = true;
 208                         if ( $val ) {
 209                                 $this->php = $val;
 210                         }
 211                         break;
 212                 }
 213         }
 214
 215         function processFileOpt( $val, $param ) {
 216                 $fileURIs = explode( ';', $param );
 217                 foreach ( $fileURIs as $URI ) {
 218                         switch ( $val ) {
 219                                 case "file":
 220                                         $newURI = $URI;
 221                                         break;
 222                                 case "gzip":
 223                                         $newURI = "compress.zlib://$URI";
 224                                         break;
 225                                 case "bzip2":
 226                                         $newURI = "compress.bzip2://$URI";
 227                                         break;
 228                                 case "7zip":
 229                                         $newURI = "mediawiki.compress.7z://$URI";
 230                                         break;
 231                                 default:
 232                                         $newURI = $URI;
 233                         }
 234                         $newFileURIs[] = $newURI;
 235                 }
 236                 $val = implode( ';', $newFileURIs );
 237                 return $val;
 238         }
 239
 240         /**
 241          * Overridden to include prefetch ratio if enabled.
 242          */
 243         function showReport() {
 244                 if ( !$this->prefetch ) {
 245                         parent::showReport();
 246                         return;
 247                 }
 248
 249                 if ( $this->reporting ) {
 250                         $now = wfTimestamp( TS_DB );
 251                         $nowts = microtime( true );
 252                         $deltaAll = $nowts - $this->startTime;
 253                         $deltaPart = $nowts - $this->lastTime;
 254                         $this->pageCountPart = $this->pageCount - $this->pageCountLast;
 255                         $this->revCountPart = $this->revCount - $this->revCountLast;
 256
 257                         if ( $deltaAll ) {
 258                                 $portion = $this->revCount / $this->maxCount;
 259                                 $eta = $this->startTime + $deltaAll / $portion;
 260                                 $etats = wfTimestamp( TS_DB, intval( $eta ) );
 261                                 if ( $this->fetchCount ) {
 262                                         $fetchRate = 100.0 * $this->prefetchCount / $this->fetchCount;
 263                                 } else {
 264                                         $fetchRate = '-';
 265                                 }
 266                                 $pageRate = $this->pageCount / $deltaAll;
 267                                 $revRate = $this->revCount / $deltaAll;
 268                         } else {
 269                                 $pageRate = '-';
 270                                 $revRate = '-';
 271                                 $etats = '-';
 272                                 $fetchRate = '-';
 273                         }
 274                         if ( $deltaPart ) {
 275                                 if ( $this->fetchCountLast ) {
 276                                         $fetchRatePart = 100.0 * $this->prefetchCountLast / $this->fetchCountLast;
 277                                 } else {
 278                                         $fetchRatePart = '-';
 279                                 }
 280                                 $pageRatePart = $this->pageCountPart / $deltaPart;
 281                                 $revRatePart = $this->revCountPart / $deltaPart;
 282
 283                         } else {
 284                                 $fetchRatePart = '-';
 285                                 $pageRatePart = '-';
 286                                 $revRatePart = '-';
 287                         }
 288                         $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]",
 289                                         $now, wfWikiID(), $this->ID, $this->pageCount, $pageRate, $pageRatePart, $this->revCount, $revRate, $revRatePart, $fetchRate, $fetchRatePart, $etats, $this->maxCount ) );
 290                         $this->lastTime = $nowts;
 291                         $this->revCountLast = $this->revCount;
 292                         $this->prefetchCountLast = $this->prefetchCount;
 293                         $this->fetchCountLast = $this->fetchCount;
 294                 }
 295         }
 296
 297         function setTimeExceeded() {
 298                 $this->timeExceeded = true;
 299         }
 300
 301         function checkIfTimeExceeded() {
 302                 if ( $this->maxTimeAllowed && ( $this->lastTime - $this->timeOfCheckpoint > $this->maxTimeAllowed ) ) {
 303                         return true;
 304                 }
 305                 return false;
 306         }
 307
 308         function finalOptionCheck() {
 309                 if ( ( $this->checkpointFiles && ! $this->maxTimeAllowed ) ||
 310                         ( $this->maxTimeAllowed && !$this->checkpointFiles ) ) {
 311                         throw new MWException( "Options checkpointfile and maxtime must be specified together.\n" );
 312                 }
 313                 foreach ( $this->checkpointFiles as $checkpointFile ) {
 314                         $count = substr_count ( $checkpointFile, "%s" );
 315                         if ( $count != 2 ) {
 316                                 throw new MWException( "Option checkpointfile must contain two '%s' for substitution of first and last pageids, count is $count instead, file is $checkpointFile.\n" );
 317                         }
 318                 }
 319
 320                 if ( $this->checkpointFiles ) {
 321                         $filenameList = (array)$this->egress->getFilenames();
 322                         if ( count( $filenameList ) != count( $this->checkpointFiles ) ) {
 323                                 throw new MWException( "One checkpointfile must be specified for each output option, if maxtime is used.\n" );
 324                         }
 325                 }
 326         }
 327
 328         /**
 329          * @throws MWException Failure to parse XML input
 330          * @return true
 331          */
 332         function readDump( $input ) {
 333                 $this->buffer = "";
 334                 $this->openElement = false;
 335                 $this->atStart = true;
 336                 $this->state = "";
 337                 $this->lastName = "";
 338                 $this->thisPage = 0;
 339                 $this->thisRev = 0;
 340
 341                 $parser = xml_parser_create( "UTF-8" );
 342                 xml_parser_set_option( $parser, XML_OPTION_CASE_FOLDING, false );
 343
 344                 xml_set_element_handler( $parser, array( &$this, 'startElement' ), array( &$this, 'endElement' ) );
 345                 xml_set_character_data_handler( $parser, array( &$this, 'characterData' ) );
 346
 347                 $offset = 0; // for context extraction on error reporting
 348                 $bufferSize = 512 * 1024;
 349                 do {
 350                         if ( $this->checkIfTimeExceeded() ) {
 351                                 $this->setTimeExceeded();
 352                         }
 353                         $chunk = fread( $input, $bufferSize );
 354                         if ( !xml_parse( $parser, $chunk, feof( $input ) ) ) {
 355                                 wfDebug( "TextDumpPass::readDump encountered XML parsing error\n" );
 356
 357                                 $byte = xml_get_current_byte_index( $parser );
 358                                 $msg = wfMessage( 'xml-error-string',
 359                                         'XML import parse failure',
 360                                         xml_get_current_line_number( $parser ),
 361                                         xml_get_current_column_number( $parser ),
 362                                         $byte . ( is_null( $chunk ) ? null : ( '; "' . substr( $chunk, $byte -$offset, 16 ) . '"' ) ),
 363                                         xml_error_string( xml_get_error_code( $parser ) ) )->escaped();
 364
 365                                 xml_parser_free( $parser );
 366
 367                                 throw new MWException( $msg );
 368                         }
 369                         $offset += strlen( $chunk );
 370                 } while ( $chunk !== false && !feof( $input ) );
 371                 if ( $this->maxTimeAllowed ) {
 372                         $filenameList = (array)$this->egress->getFilenames();
 373                         // we wrote some stuff after last checkpoint that needs renamed
 374                         if ( file_exists( $filenameList[0] ) ) {
 375                                 $newFilenames = array();
 376                                 # we might have just written the header and footer and had no
 377                                 # pages or revisions written... perhaps they were all deleted
 378                                 # there's no pageID 0 so we use that. the caller is responsible
 379                                 # for deciding what to do with a file containing only the
 380                                 # siteinfo information and the mw tags.
 381                                 if ( ! $this->firstPageWritten ) {
 382                                         $firstPageID = str_pad( 0, 9, "0", STR_PAD_LEFT );
 383                                         $lastPageID = str_pad( 0, 9, "0", STR_PAD_LEFT );
 384                                 }
 385                                 else {
 386                                         $firstPageID = str_pad( $this->firstPageWritten, 9, "0", STR_PAD_LEFT );
 387                                         $lastPageID = str_pad( $this->lastPageWritten, 9, "0", STR_PAD_LEFT );
 388                                 }
 389                                 for ( $i = 0; $i < count( $filenameList ); $i++ ) {
 390                                         $checkpointNameFilledIn = sprintf( $this->checkpointFiles[$i], $firstPageID, $lastPageID );
 391                                         $fileinfo = pathinfo( $filenameList[$i] );
 392                                         $newFilenames[] = $fileinfo['dirname'] . '/' . $checkpointNameFilledIn;
 393                                 }
 394                                 $this->egress->closeAndRename( $newFilenames );
 395                         }
 396                 }
 397                 xml_parser_free( $parser );
 398
 399                 return true;
 400         }
 401
 402         /**
 403          * Tries to get the revision text for a revision id.
 404          *
 405          * Upon errors, retries (Up to $this->maxFailures tries each call).
 406          * If still no good revision get could be found even after this retrying, "" is returned.
 407          * If no good revision text could be returned for
 408          * $this->maxConsecutiveFailedTextRetrievals consecutive calls to getText, MWException
 409          * is thrown.
 410          *
 411          * @param $id string The revision id to get the text for
 412          *
 413          * @return string The revision text for $id, or ""
 414          * @throws MWException
 415          */
 416         function getText( $id ) {
 417                 $prefetchNotTried = true; // Whether or not we already tried to get the text via prefetch.
 418                 $text = false; // The candidate for a good text. false if no proper value.
 419                 $failures = 0; // The number of times, this invocation of getText already failed.
 420
 421                 static $consecutiveFailedTextRetrievals = 0; // The number of times getText failed without
 422                                                              // yielding a good text in between.
 423
 424                 $this->fetchCount++;
 425
 426                 // To allow to simply return on success and do not have to worry about book keeping,
 427                 // we assume, this fetch works (possible after some retries). Nevertheless, we koop
 428                 // the old value, so we can restore it, if problems occur (See after the while loop).
 429                 $oldConsecutiveFailedTextRetrievals = $consecutiveFailedTextRetrievals;
 430                 $consecutiveFailedTextRetrievals = 0;
 431
 432                 while ( $failures < $this->maxFailures ) {
 433
 434                         // As soon as we found a good text for the $id, we will return immediately.
 435                         // Hence, if we make it past the try catch block, we know that we did not
 436                         // find a good text.
 437
 438                         try {
 439                                 // Step 1: Get some text (or reuse from previous iteratuon if checking
 440                                 //         for plausibility failed)
 441
 442                                 // Trying to get prefetch, if it has not been tried before
 443                                 if ( $text === false && isset( $this->prefetch ) && $prefetchNotTried ) {
 444                                         $prefetchNotTried = false;
 445                                         $tryIsPrefetch = true;
 446                                         $text = $this->prefetch->prefetch( intval( $this->thisPage ),
 447                                                 intval( $this->thisRev ) );
 448                                         if ( $text === null ) {
 449                                                 $text = false;
 450                                         }
 451                                 }
 452
 453                                 if ( $text === false ) {
 454                                         // Fallback to asking the database
 455                                         $tryIsPrefetch = false;
 456                                         if ( $this->spawn ) {
 457                                                 $text = $this->getTextSpawned( $id );
 458                                         } else {
 459                                                 $text = $this->getTextDb( $id );
 460                                         }
 461
 462                                         // No more checks for texts from DB for now.
 463                                         // If we received something that is not false,
 464                                         // We treat it as good text, regardless of whether it actually is or is not
 465                                         if ( $text !== false ) {
 466                                                 return $text;
 467                                         }
 468                                 }
 469
 470                                 if ( $text === false ) {
 471                                         throw new MWException( "Generic error while obtaining text for id " . $id );
 472                                 }
 473
 474                                 // We received a good candidate for the text of $id via some method
 475
 476                                 // Step 2: Checking for plausibility and return the text if it is
 477                                 //         plausible
 478                                 $revID = intval( $this->thisRev );
 479                                 if ( ! isset( $this->db ) ) {
 480                                         throw new MWException( "No database available" );
 481                                 }
 482                                 $revLength = $this->db->selectField( 'revision', 'rev_len', array( 'rev_id' => $revID ) );
 483                                 if ( strlen( $text ) == $revLength ) {
 484                                         if ( $tryIsPrefetch ) {
 485                                                 $this->prefetchCount++;
 486                                         }
 487                                         return $text;
 488                                 }
 489
 490                                 $text = false;
 491                                 throw new MWException( "Received text is unplausible for id " . $id );
 492
 493                         } catch ( Exception $e ) {
 494                                 $msg = "getting/checking text " . $id . " failed (" . $e->getMessage() . ")";
 495                                 if ( $failures + 1 < $this->maxFailures ) {
 496                                         $msg .= " (Will retry " . ( $this->maxFailures - $failures - 1 ) . " more times)";
 497                                 }
 498                                 $this->progress( $msg );
 499                         }
 500
 501                         // Something went wrong; we did not a text that was plausible :(
 502                         $failures++;
 503
 504                         // A failure in a prefetch hit does not warrant resetting db connection etc.
 505                         if ( ! $tryIsPrefetch ) {
 506                                 // After backing off for some time, we try to reboot the whole process as
 507                                 // much as possible to not carry over failures from one part to the other
 508                                 // parts
 509                                 sleep( $this->failureTimeout );
 510                                 try {
 511                                         $this->rotateDb();
 512                                         if ( $this->spawn ) {
 513                                                 $this->closeSpawn();
 514                                                 $this->openSpawn();
 515                                         }
 516                                 } catch ( Exception $e ) {
 517                                         $this->progress( "Rebooting getText infrastructure failed (" . $e->getMessage() . ")" .
 518                                                 " Trying to continue anyways" );
 519                                 }
 520                         }
 521                 }
 522
 523                 // Retirieving a good text for $id failed (at least) maxFailures times.
 524                 // We abort for this $id.
 525
 526                 // Restoring the consecutive failures, and maybe aborting, if the dump
 527                 // is too broken.
 528                 $consecutiveFailedTextRetrievals = $oldConsecutiveFailedTextRetrievals + 1;
 529                 if ( $consecutiveFailedTextRetrievals > $this->maxConsecutiveFailedTextRetrievals ) {
 530                         throw new MWException( "Graceful storage failure" );
 531                 }
 532
 533                 return "";
 534         }
 535
 536
 537         /**
 538          * May throw a database error if, say, the server dies during query.
 539          * @param $id
 540          * @return bool|string
 541          * @throws MWException
 542          */
 543         private function getTextDb( $id ) {
 544                 global $wgContLang;
 545                 if ( ! isset( $this->db ) ) {
 546                         throw new MWException( __METHOD__ . "No database available" );
 547                 }
 548                 $row = $this->db->selectRow( 'text',
 549                         array( 'old_text', 'old_flags' ),
 550                         array( 'old_id' => $id ),
 551                         __METHOD__ );
 552                 $text = Revision::getRevisionText( $row );
 553                 if ( $text === false ) {
 554                         return false;
 555                 }
 556                 $stripped = str_replace( "\r", "", $text );
 557                 $normalized = $wgContLang->normalize( $stripped );
 558                 return $normalized;
 559         }
 560
 561         private function getTextSpawned( $id ) {
 562                 wfSuppressWarnings();
 563                 if ( !$this->spawnProc ) {
 564                         // First time?
 565                         $this->openSpawn();
 566                 }
 567                 $text = $this->getTextSpawnedOnce( $id );
 568                 wfRestoreWarnings();
 569                 return $text;
 570         }
 571
 572         function openSpawn() {
 573                 global $IP;
 574
 575                 if ( file_exists( "$IP/../multiversion/MWScript.php" ) ) {
 576                         $cmd = implode( " ",
 577                                 array_map( 'wfEscapeShellArg',
 578                                         array(
 579                                                 $this->php,
 580                                                 "$IP/../multiversion/MWScript.php",
 581                                                 "fetchText.php",
 582                                                 '--wiki', wfWikiID() ) ) );
 583                 }
 584                 else {
 585                         $cmd = implode( " ",
 586                                 array_map( 'wfEscapeShellArg',
 587                                         array(
 588                                                 $this->php,
 589                                                 "$IP/maintenance/fetchText.php",
 590                                                 '--wiki', wfWikiID() ) ) );
 591                 }
 592                 $spec = array(
 593                         0 => array( "pipe", "r" ),
 594                         1 => array( "pipe", "w" ),
 595                         2 => array( "file", "/dev/null", "a" ) );
 596                 $pipes = array();
 597
 598                 $this->progress( "Spawning database subprocess: $cmd" );
 599                 $this->spawnProc = proc_open( $cmd, $spec, $pipes );
 600                 if ( !$this->spawnProc ) {
 601                         // shit
 602                         $this->progress( "Subprocess spawn failed." );
 603                         return false;
 604                 }
 605                 list(
 606                         $this->spawnWrite, // -> stdin
 607                         $this->spawnRead,  // <- stdout
 608                 ) = $pipes;
 609
 610                 return true;
 611         }
 612
 613         private function closeSpawn() {
 614                 wfSuppressWarnings();
 615                 if ( $this->spawnRead ) {
 616                         fclose( $this->spawnRead );
 617                 }
 618                 $this->spawnRead = false;
 619                 if ( $this->spawnWrite ) {
 620                         fclose( $this->spawnWrite );
 621                 }
 622                 $this->spawnWrite = false;
 623                 if ( $this->spawnErr ) {
 624                         fclose( $this->spawnErr );
 625                 }
 626                 $this->spawnErr = false;
 627                 if ( $this->spawnProc ) {
 628                         pclose( $this->spawnProc );
 629                 }
 630                 $this->spawnProc = false;
 631                 wfRestoreWarnings();
 632         }
 633
 634         private function getTextSpawnedOnce( $id ) {
 635                 global $wgContLang;
 636
 637                 $ok = fwrite( $this->spawnWrite, "$id\n" );
 638                 // $this->progress( ">> $id" );
 639                 if ( !$ok ) {
 640                         return false;
 641                 }
 642
 643                 $ok = fflush( $this->spawnWrite );
 644                 // $this->progress( ">> [flush]" );
 645                 if ( !$ok ) {
 646                         return false;
 647                 }
 648
 649                 // check that the text id they are sending is the one we asked for
 650                 // this avoids out of sync revision text errors we have encountered in the past
 651                 $newId = fgets( $this->spawnRead );
 652                 if ( $newId === false ) {
 653                         return false;
 654                 }
 655                 if ( $id != intval( $newId ) ) {
 656                         return false;
 657                 }
 658
 659                 $len = fgets( $this->spawnRead );
 660                 // $this->progress( "<< " . trim( $len ) );
 661                 if ( $len === false ) {
 662                         return false;
 663                 }
 664
 665                 $nbytes = intval( $len );
 666                 // actual error, not zero-length text
 667                 if ( $nbytes < 0 ) {
 668                         return false;
 669                 }
 670
 671                 $text = "";
 672
 673                 // Subprocess may not send everything at once, we have to loop.
 674                 while ( $nbytes > strlen( $text ) ) {
 675                         $buffer = fread( $this->spawnRead, $nbytes - strlen( $text ) );
 676                         if ( $buffer === false ) {
 677                                 break;
 678                         }
 679                         $text .= $buffer;
 680                 }
 681
 682                 $gotbytes = strlen( $text );
 683                 if ( $gotbytes != $nbytes ) {
 684                         $this->progress( "Expected $nbytes bytes from database subprocess, got $gotbytes " );
 685                         return false;
 686                 }
 687
 688                 // Do normalization in the dump thread...
 689                 $stripped = str_replace( "\r", "", $text );
 690                 $normalized = $wgContLang->normalize( $stripped );
 691                 return $normalized;
 692         }
 693
 694         function startElement( $parser, $name, $attribs ) {
 695                 $this->checkpointJustWritten = false;
 696
 697                 $this->clearOpenElement( null );
 698                 $this->lastName = $name;
 699
 700                 if ( $name == 'revision' ) {
 701                         $this->state = $name;
 702                         $this->egress->writeOpenPage( null, $this->buffer );
 703                         $this->buffer = "";
 704                 } elseif ( $name == 'page' ) {
 705                         $this->state = $name;
 706                         if ( $this->atStart ) {
 707                                 $this->egress->writeOpenStream( $this->buffer );
 708                                 $this->buffer = "";
 709                                 $this->atStart = false;
 710                         }
 711                 }
 712
 713                 if ( $name == "text" && isset( $attribs['id'] ) ) {
 714                         $text = $this->getText( $attribs['id'] );
 715                         $this->openElement = array( $name, array( 'xml:space' => 'preserve' ) );
 716                         if ( strlen( $text ) > 0 ) {
 717                                 $this->characterData( $parser, $text );
 718                         }
 719                 } else {
 720                         $this->openElement = array( $name, $attribs );
 721                 }
 722         }
 723
 724         function endElement( $parser, $name ) {
 725                 $this->checkpointJustWritten = false;
 726
 727                 if ( $this->openElement ) {
 728                         $this->clearOpenElement( "" );
 729                 } else {
 730                         $this->buffer .= "</$name>";
 731                 }
 732
 733                 if ( $name == 'revision' ) {
 734                         $this->egress->writeRevision( null, $this->buffer );
 735                         $this->buffer = "";
 736                         $this->thisRev = "";
 737                 } elseif ( $name == 'page' ) {
 738                         if ( ! $this->firstPageWritten ) {
 739                                 $this->firstPageWritten = trim( $this->thisPage );
 740                         }
 741                         $this->lastPageWritten = trim( $this->thisPage );
 742                         if ( $this->timeExceeded ) {
 743                                 $this->egress->writeClosePage( $this->buffer );
 744                                 // nasty hack, we can't just write the chardata after the
 745                                 // page tag, it will include leading blanks from the next line
 746                                 $this->egress->sink->write( "\n" );
 747
 748                                 $this->buffer = $this->xmlwriterobj->closeStream();
 749                                 $this->egress->writeCloseStream( $this->buffer );
 750
 751                                 $this->buffer = "";
 752                                 $this->thisPage = "";
 753                                 // this could be more than one file if we had more than one output arg
 754
 755                                 $filenameList = (array)$this->egress->getFilenames();
 756                                 $newFilenames = array();
 757                                 $firstPageID = str_pad( $this->firstPageWritten, 9, "0", STR_PAD_LEFT );
 758                                 $lastPageID = str_pad( $this->lastPageWritten, 9, "0", STR_PAD_LEFT );
 759                                 for ( $i = 0; $i < count( $filenameList ); $i++ ) {
 760                                         $checkpointNameFilledIn = sprintf( $this->checkpointFiles[$i], $firstPageID, $lastPageID );
 761                                         $fileinfo = pathinfo( $filenameList[$i] );
 762                                         $newFilenames[] = $fileinfo['dirname'] . '/' . $checkpointNameFilledIn;
 763                                 }
 764                                 $this->egress->closeRenameAndReopen( $newFilenames );
 765                                 $this->buffer = $this->xmlwriterobj->openStream();
 766                                 $this->timeExceeded = false;
 767                                 $this->timeOfCheckpoint = $this->lastTime;
 768                                 $this->firstPageWritten = false;
 769                                 $this->checkpointJustWritten = true;
 770                         }
 771                         else {
 772                                 $this->egress->writeClosePage( $this->buffer );
 773                                 $this->buffer = "";
 774                                 $this->thisPage = "";
 775                         }
 776
 777                 } elseif ( $name == 'mediawiki' ) {
 778                         $this->egress->writeCloseStream( $this->buffer );
 779                         $this->buffer = "";
 780                 }
 781         }
 782
 783         function characterData( $parser, $data ) {
 784                 $this->clearOpenElement( null );
 785                 if ( $this->lastName == "id" ) {
 786                         if ( $this->state == "revision" ) {
 787                                 $this->thisRev .= $data;
 788                         } elseif ( $this->state == "page" ) {
 789                                 $this->thisPage .= $data;
 790                         }
 791                 }
 792                 // have to skip the newline left over from closepagetag line of
 793                 // end of checkpoint files. nasty hack!!
 794                 if ( $this->checkpointJustWritten ) {
 795                         if ( $data[0] == "\n" ) {
 796                                 $data = substr( $data, 1 );
 797                         }
 798                         $this->checkpointJustWritten = false;
 799                 }
 800                 $this->buffer .= htmlspecialchars( $data );
 801         }
 802
 803         function clearOpenElement( $style ) {
 804                 if ( $this->openElement ) {
 805                         $this->buffer .= Xml::element( $this->openElement[0], $this->openElement[1], $style );
 806                         $this->openElement = false;
 807                 }
 808         }
 809 }