maintenance/dumpTextPass.php

   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 require_once __DIR__ . '/../includes/export/WikiExporter.php';
  29
  30 /**
  31  * @ingroup Maintenance
  32  */
  33 class TextPassDumper extends BackupDumper {
  34         public $prefetch = null;
  35
  36         // when we spend more than maxTimeAllowed seconds on this run, we continue
  37         // processing until we write out the next complete page, then save output file(s),
  38         // rename it/them and open new one(s)
  39         public $maxTimeAllowed = 0; // 0 = no limit
  40
  41         protected $input = "php://stdin";
  42         protected $history = WikiExporter::FULL;
  43         protected $fetchCount = 0;
  44         protected $prefetchCount = 0;
  45         protected $prefetchCountLast = 0;
  46         protected $fetchCountLast = 0;
  47
  48         protected $maxFailures = 5;
  49         protected $maxConsecutiveFailedTextRetrievals = 200;
  50         protected $failureTimeout = 5; // Seconds to sleep after db failure
  51
  52         protected $bufferSize = 524288; // In bytes. Maximum size to read from the stub in on go.
  53
  54         protected $php = "php";
  55         protected $spawn = false;
  56
  57         /**
  58          * @var bool|resource
  59          */
  60         protected $spawnProc = false;
  61
  62         /**
  63          * @var bool|resource
  64          */
  65         protected $spawnWrite = false;
  66
  67         /**
  68          * @var bool|resource
  69          */
  70         protected $spawnRead = false;
  71
  72         /**
  73          * @var bool|resource
  74          */
  75         protected $spawnErr = false;
  76
  77         protected $xmlwriterobj = false;
  78
  79         protected $timeExceeded = false;
  80         protected $firstPageWritten = false;
  81         protected $lastPageWritten = false;
  82         protected $checkpointJustWritten = false;
  83         protected $checkpointFiles = array();
  84
  85         /**
  86          * @var DatabaseBase
  87          */
  88         protected $db;
  89
  90         /**
  91          * @param array $args For backward compatibility
  92          */
  93         function __construct( $args = null ) {
  94                 parent::__construct();
  95
  96                 $this->mDescription = <<<TEXT
  97 This script postprocesses XML dumps from dumpBackup.php to add
  98 page text which was stubbed out (using --stub).
  99
 100 XML input is accepted on stdin.
 101 XML output is sent to stdout; progress reports are sent to stderr.
 102 TEXT;
 103                 $this->stderr = fopen( "php://stderr", "wt" );
 104
 105                 $this->addOption( 'stub', 'To load a compressed stub dump instead of stdin. ' .
 106                         'Specify as --stub=<type>:<file>.', false, true );
 107                 $this->addOption( 'prefetch', 'Use a prior dump file as a text source, to savepressure on the ' .
 108                         'database. (Requires the XMLReader extension). Specify as --prefetch=<type>:<file>',
 109                         false, true );
 110                 $this->addOption( 'maxtime', 'Write out checkpoint file after this many minutes (writing' .
 111                         'out complete page, closing xml file properly, and opening new one' .
 112                         'with header).  This option requires the checkpointfile option.', false, true );
 113                 $this->addOption( 'checkpointfile', 'Use this string for checkpoint filenames,substituting ' .
 114                         'first pageid written for the first %s (required) and the last pageid written for the ' .
 115                         'second %s if it exists.', false, true, false, true ); // This can be specified multiple times
 116                 $this->addOption( 'quiet', 'Don\'t dump status reports to stderr.' );
 117                 $this->addOption( 'current', 'Base ETA on number of pages in database instead of all revisions' );
 118                 $this->addOption( 'spawn', 'Spawn a subprocess for loading text records' );
 119                 $this->addOption( 'buffersize', 'Buffer size in bytes to use for reading the stub. ' .
 120                         '(Default: 512KB, Minimum: 4KB)', false, true );
 121
 122                 if ( $args ) {
 123                         $this->loadWithArgv( $args );
 124                         $this->processOptions();
 125                 }
 126         }
 127
 128         function execute() {
 129                 $this->processOptions();
 130                 $this->dump( true );
 131         }
 132
 133         function processOptions() {
 134                 global $IP;
 135
 136                 parent::processOptions();
 137
 138                 if ( $this->hasOption( 'buffersize' ) ) {
 139                         $this->bufferSize = max( intval( $this->getOption( 'buffersize' ) ), 4 * 1024 );
 140                 }
 141
 142                 if ( $this->hasOption( 'prefetch' ) ) {
 143                         require_once "$IP/maintenance/backupPrefetch.inc";
 144                         $url = $this->processFileOpt( $this->getOption( 'prefetch' ) );
 145                         $this->prefetch = new BaseDump( $url );
 146                 }
 147
 148                 if ( $this->hasOption( 'stub' ) ) {
 149                         $this->input = $this->processFileOpt( $this->getOption( 'stub' ) );
 150                 }
 151
 152                 if ( $this->hasOption( 'maxtime' ) ) {
 153                         $this->maxTimeAllowed = intval( $this->getOption( 'maxtime' ) ) * 60;
 154                 }
 155
 156                 if ( $this->hasOption( 'checkpointfile' ) ) {
 157                         $this->checkpointFiles = $this->getOption( 'checkpointfile' );
 158                 }
 159
 160                 if ( $this->hasOption( 'current' ) ) {
 161                         $this->history = WikiExporter::CURRENT;
 162                 }
 163
 164                 if ( $this->hasOption( 'full' ) ) {
 165                         $this->history = WikiExporter::FULL;
 166                 }
 167
 168                 if ( $this->hasOption( 'spawn' ) ) {
 169                         $this->spawn = true;
 170                         $val = $this->getOption( 'spawn' );
 171                         if ( $val !== 1 ) {
 172                                 $this->php = $val;
 173                         }
 174                 }
 175         }
 176
 177         /**
 178          * Drop the database connection $this->db and try to get a new one.
 179          *
 180          * This function tries to get a /different/ connection if this is
 181          * possible. Hence, (if this is possible) it switches to a different
 182          * failover upon each call.
 183          *
 184          * This function resets $this->lb and closes all connections on it.
 185          *
 186          * @throws MWException
 187          */
 188         function rotateDb() {
 189                 // Cleaning up old connections
 190                 if ( isset( $this->lb ) ) {
 191                         $this->lb->closeAll();
 192                         unset( $this->lb );
 193                 }
 194
 195                 if ( $this->forcedDb !== null ) {
 196                         $this->db = $this->forcedDb;
 197
 198                         return;
 199                 }
 200
 201                 if ( isset( $this->db ) && $this->db->isOpen() ) {
 202                         throw new MWException( 'DB is set and has not been closed by the Load Balancer' );
 203                 }
 204
 205                 unset( $this->db );
 206
 207                 // Trying to set up new connection.
 208                 // We do /not/ retry upon failure, but delegate to encapsulating logic, to avoid
 209                 // individually retrying at different layers of code.
 210
 211                 // 1. The LoadBalancer.
 212                 try {
 213                         $this->lb = wfGetLBFactory()->newMainLB();
 214                 } catch ( Exception $e ) {
 215                         throw new MWException( __METHOD__
 216                                 . " rotating DB failed to obtain new load balancer (" . $e->getMessage() . ")" );
 217                 }
 218
 219                 // 2. The Connection, through the load balancer.
 220                 try {
 221                         $this->db = $this->lb->getConnection( DB_SLAVE, 'dump' );
 222                 } catch ( Exception $e ) {
 223                         throw new MWException( __METHOD__
 224                                 . " rotating DB failed to obtain new database (" . $e->getMessage() . ")" );
 225                 }
 226         }
 227
 228         function initProgress( $history = WikiExporter::FULL ) {
 229                 parent::initProgress();
 230                 $this->timeOfCheckpoint = $this->startTime;
 231         }
 232
 233         function dump( $history, $text = WikiExporter::TEXT ) {
 234                 // Notice messages will foul up your XML output even if they're
 235                 // relatively harmless.
 236                 if ( ini_get( 'display_errors' ) ) {
 237                         ini_set( 'display_errors', 'stderr' );
 238                 }
 239
 240                 $this->initProgress( $this->history );
 241
 242                 // We are trying to get an initial database connection to avoid that the
 243                 // first try of this request's first call to getText fails. However, if
 244                 // obtaining a good DB connection fails it's not a serious issue, as
 245                 // getText does retry upon failure and can start without having a working
 246                 // DB connection.
 247                 try {
 248                         $this->rotateDb();
 249                 } catch ( Exception $e ) {
 250                         // We do not even count this as failure. Just let eventual
 251                         // watchdogs know.
 252                         $this->progress( "Getting initial DB connection failed (" .
 253                                 $e->getMessage() . ")" );
 254                 }
 255
 256                 $this->egress = new ExportProgressFilter( $this->sink, $this );
 257
 258                 // it would be nice to do it in the constructor, oh well. need egress set
 259                 $this->finalOptionCheck();
 260
 261                 // we only want this so we know how to close a stream :-P
 262                 $this->xmlwriterobj = new XmlDumpWriter();
 263
 264                 $input = fopen( $this->input, "rt" );
 265                 $this->readDump( $input );
 266
 267                 if ( $this->spawnProc ) {
 268                         $this->closeSpawn();
 269                 }
 270
 271                 $this->report( true );
 272         }
 273
 274         function processFileOpt( $opt ) {
 275                 $split = explode( ':', $opt, 2 );
 276                 $val = $split[0];
 277                 $param = '';
 278                 if ( count( $split ) === 2 ) {
 279                         $param = $split[1];
 280                 }
 281                 $fileURIs = explode( ';', $param );
 282                 foreach ( $fileURIs as $URI ) {
 283                         switch ( $val ) {
 284                                 case "file":
 285                                         $newURI = $URI;
 286                                         break;
 287                                 case "gzip":
 288                                         $newURI = "compress.zlib://$URI";
 289                                         break;
 290                                 case "bzip2":
 291                                         $newURI = "compress.bzip2://$URI";
 292                                         break;
 293                                 case "7zip":
 294                                         $newURI = "mediawiki.compress.7z://$URI";
 295                                         break;
 296                                 default:
 297                                         $newURI = $URI;
 298                         }
 299                         $newFileURIs[] = $newURI;
 300                 }
 301                 $val = implode( ';', $newFileURIs );
 302
 303                 return $val;
 304         }
 305
 306         /**
 307          * Overridden to include prefetch ratio if enabled.
 308          */
 309         function showReport() {
 310                 if ( !$this->prefetch ) {
 311                         parent::showReport();
 312
 313                         return;
 314                 }
 315
 316                 if ( $this->reporting ) {
 317                         $now = wfTimestamp( TS_DB );
 318                         $nowts = microtime( true );
 319                         $deltaAll = $nowts - $this->startTime;
 320                         $deltaPart = $nowts - $this->lastTime;
 321                         $this->pageCountPart = $this->pageCount - $this->pageCountLast;
 322                         $this->revCountPart = $this->revCount - $this->revCountLast;
 323
 324                         if ( $deltaAll ) {
 325                                 $portion = $this->revCount / $this->maxCount;
 326                                 $eta = $this->startTime + $deltaAll / $portion;
 327                                 $etats = wfTimestamp( TS_DB, intval( $eta ) );
 328                                 if ( $this->fetchCount ) {
 329                                         $fetchRate = 100.0 * $this->prefetchCount / $this->fetchCount;
 330                                 } else {
 331                                         $fetchRate = '-';
 332                                 }
 333                                 $pageRate = $this->pageCount / $deltaAll;
 334                                 $revRate = $this->revCount / $deltaAll;
 335                         } else {
 336                                 $pageRate = '-';
 337                                 $revRate = '-';
 338                                 $etats = '-';
 339                                 $fetchRate = '-';
 340                         }
 341                         if ( $deltaPart ) {
 342                                 if ( $this->fetchCountLast ) {
 343                                         $fetchRatePart = 100.0 * $this->prefetchCountLast / $this->fetchCountLast;
 344                                 } else {
 345                                         $fetchRatePart = '-';
 346                                 }
 347                                 $pageRatePart = $this->pageCountPart / $deltaPart;
 348                                 $revRatePart = $this->revCountPart / $deltaPart;
 349                         } else {
 350                                 $fetchRatePart = '-';
 351                                 $pageRatePart = '-';
 352                                 $revRatePart = '-';
 353                         }
 354                         $this->progress( sprintf(
 355                                 "%s: %s (ID %d) %d pages (%0.1f|%0.1f/sec all|curr), "
 356                                         . "%d revs (%0.1f|%0.1f/sec all|curr), %0.1f%%|%0.1f%% "
 357                                         . "prefetched (all|curr), ETA %s [max %d]",
 358                                 $now, wfWikiID(), $this->ID, $this->pageCount, $pageRate,
 359                                 $pageRatePart, $this->revCount, $revRate, $revRatePart,
 360                                 $fetchRate, $fetchRatePart, $etats, $this->maxCount
 361                         ) );
 362                         $this->lastTime = $nowts;
 363                         $this->revCountLast = $this->revCount;
 364                         $this->prefetchCountLast = $this->prefetchCount;
 365                         $this->fetchCountLast = $this->fetchCount;
 366                 }
 367         }
 368
 369         function setTimeExceeded() {
 370                 $this->timeExceeded = true;
 371         }
 372
 373         function checkIfTimeExceeded() {
 374                 if ( $this->maxTimeAllowed
 375                         && ( $this->lastTime - $this->timeOfCheckpoint > $this->maxTimeAllowed )
 376                 ) {
 377                         return true;
 378                 }
 379
 380                 return false;
 381         }
 382
 383         function finalOptionCheck() {
 384                 if ( ( $this->checkpointFiles && !$this->maxTimeAllowed )
 385                         || ( $this->maxTimeAllowed && !$this->checkpointFiles )
 386                 ) {
 387                         throw new MWException( "Options checkpointfile and maxtime must be specified together.\n" );
 388                 }
 389                 foreach ( $this->checkpointFiles as $checkpointFile ) {
 390                         $count = substr_count( $checkpointFile, "%s" );
 391                         if ( $count != 2 ) {
 392                                 throw new MWException( "Option checkpointfile must contain two '%s' "
 393                                         . "for substitution of first and last pageids, count is $count instead, "
 394                                         . "file is $checkpointFile.\n" );
 395                         }
 396                 }
 397
 398                 if ( $this->checkpointFiles ) {
 399                         $filenameList = (array)$this->egress->getFilenames();
 400                         if ( count( $filenameList ) != count( $this->checkpointFiles ) ) {
 401                                 throw new MWException( "One checkpointfile must be specified "
 402                                         . "for each output option, if maxtime is used.\n" );
 403                         }
 404                 }
 405         }
 406
 407         /**
 408          * @throws MWException Failure to parse XML input
 409          * @param string $input
 410          * @return bool
 411          */
 412         function readDump( $input ) {
 413                 $this->buffer = "";
 414                 $this->openElement = false;
 415                 $this->atStart = true;
 416                 $this->state = "";
 417                 $this->lastName = "";
 418                 $this->thisPage = 0;
 419                 $this->thisRev = 0;
 420                 $this->thisRevModel = null;
 421                 $this->thisRevFormat = null;
 422
 423                 $parser = xml_parser_create( "UTF-8" );
 424                 xml_parser_set_option( $parser, XML_OPTION_CASE_FOLDING, false );
 425
 426                 xml_set_element_handler(
 427                         $parser,
 428                         array( &$this, 'startElement' ),
 429                         array( &$this, 'endElement' )
 430                 );
 431                 xml_set_character_data_handler( $parser, array( &$this, 'characterData' ) );
 432
 433                 $offset = 0; // for context extraction on error reporting
 434                 do {
 435                         if ( $this->checkIfTimeExceeded() ) {
 436                                 $this->setTimeExceeded();
 437                         }
 438                         $chunk = fread( $input, $this->bufferSize );
 439                         if ( !xml_parse( $parser, $chunk, feof( $input ) ) ) {
 440                                 wfDebug( "TextDumpPass::readDump encountered XML parsing error\n" );
 441
 442                                 $byte = xml_get_current_byte_index( $parser );
 443                                 $msg = wfMessage( 'xml-error-string',
 444                                         'XML import parse failure',
 445                                         xml_get_current_line_number( $parser ),
 446                                         xml_get_current_column_number( $parser ),
 447                                         $byte . ( is_null( $chunk ) ? null : ( '; "' . substr( $chunk, $byte - $offset, 16 ) . '"' ) ),
 448                                         xml_error_string( xml_get_error_code( $parser ) ) )->escaped();
 449
 450                                 xml_parser_free( $parser );
 451
 452                                 throw new MWException( $msg );
 453                         }
 454                         $offset += strlen( $chunk );
 455                 } while ( $chunk !== false && !feof( $input ) );
 456                 if ( $this->maxTimeAllowed ) {
 457                         $filenameList = (array)$this->egress->getFilenames();
 458                         // we wrote some stuff after last checkpoint that needs renamed
 459                         if ( file_exists( $filenameList[0] ) ) {
 460                                 $newFilenames = array();
 461                                 # we might have just written the header and footer and had no
 462                                 # pages or revisions written... perhaps they were all deleted
 463                                 # there's no pageID 0 so we use that. the caller is responsible
 464                                 # for deciding what to do with a file containing only the
 465                                 # siteinfo information and the mw tags.
 466                                 if ( !$this->firstPageWritten ) {
 467                                         $firstPageID = str_pad( 0, 9, "0", STR_PAD_LEFT );
 468                                         $lastPageID = str_pad( 0, 9, "0", STR_PAD_LEFT );
 469                                 } else {
 470                                         $firstPageID = str_pad( $this->firstPageWritten, 9, "0", STR_PAD_LEFT );
 471                                         $lastPageID = str_pad( $this->lastPageWritten, 9, "0", STR_PAD_LEFT );
 472                                 }
 473
 474                                 $filenameCount = count( $filenameList );
 475                                 for ( $i = 0; $i < $filenameCount; $i++ ) {
 476                                         $checkpointNameFilledIn = sprintf( $this->checkpointFiles[$i], $firstPageID, $lastPageID );
 477                                         $fileinfo = pathinfo( $filenameList[$i] );
 478                                         $newFilenames[] = $fileinfo['dirname'] . '/' . $checkpointNameFilledIn;
 479                                 }
 480                                 $this->egress->closeAndRename( $newFilenames );
 481                         }
 482                 }
 483                 xml_parser_free( $parser );
 484
 485                 return true;
 486         }
 487
 488         /**
 489          * Applies applicable export transformations to $text.
 490          *
 491          * @param string $text
 492          * @param string $model
 493          * @param string|null $format
 494          *
 495          * @return string
 496          */
 497         private function exportTransform( $text, $model, $format = null ) {
 498                 try {
 499                         $handler = ContentHandler::getForModelID( $model );
 500                         $text = $handler->exportTransform( $text, $format );
 501                 }
 502                 catch ( MWException $ex ) {
 503                         $this->progress(
 504                                 "Unable to apply export transformation for content model '$model': " .
 505                                 $ex->getMessage()
 506                         );
 507                 }
 508
 509                 return $text;
 510         }
 511
 512         /**
 513          * Tries to get the revision text for a revision id.
 514          * Export transformations are applied if the content model can is given or can be
 515          * determined from the database.
 516          *
 517          * Upon errors, retries (Up to $this->maxFailures tries each call).
 518          * If still no good revision get could be found even after this retrying, "" is returned.
 519          * If no good revision text could be returned for
 520          * $this->maxConsecutiveFailedTextRetrievals consecutive calls to getText, MWException
 521          * is thrown.
 522          *
 523          * @param string $id The revision id to get the text for
 524          * @param string|bool|null $model The content model used to determine
 525          *  applicable export transformations.
 526          *  If $model is null, it will be determined from the database.
 527          * @param string|null $format The content format used when applying export transformations.
 528          *
 529          * @throws MWException
 530          * @return string The revision text for $id, or ""
 531          */
 532         function getText( $id, $model = null, $format = null ) {
 533                 global $wgContentHandlerUseDB;
 534
 535                 $prefetchNotTried = true; // Whether or not we already tried to get the text via prefetch.
 536                 $text = false; // The candidate for a good text. false if no proper value.
 537                 $failures = 0; // The number of times, this invocation of getText already failed.
 538
 539                 // The number of times getText failed without yielding a good text in between.
 540                 static $consecutiveFailedTextRetrievals = 0;
 541
 542                 $this->fetchCount++;
 543
 544                 // To allow to simply return on success and do not have to worry about book keeping,
 545                 // we assume, this fetch works (possible after some retries). Nevertheless, we koop
 546                 // the old value, so we can restore it, if problems occur (See after the while loop).
 547                 $oldConsecutiveFailedTextRetrievals = $consecutiveFailedTextRetrievals;
 548                 $consecutiveFailedTextRetrievals = 0;
 549
 550                 if ( $model === null && $wgContentHandlerUseDB ) {
 551                         $row = $this->db->selectRow(
 552                                 'revision',
 553                                 array( 'rev_content_model', 'rev_content_format' ),
 554                                 array( 'rev_id' => $this->thisRev ),
 555                                 __METHOD__
 556                         );
 557
 558                         if ( $row ) {
 559                                 $model = $row->rev_content_model;
 560                                 $format = $row->rev_content_format;
 561                         }
 562                 }
 563
 564                 if ( $model === null || $model === '' ) {
 565                         $model = false;
 566                 }
 567
 568                 while ( $failures < $this->maxFailures ) {
 569
 570                         // As soon as we found a good text for the $id, we will return immediately.
 571                         // Hence, if we make it past the try catch block, we know that we did not
 572                         // find a good text.
 573
 574                         try {
 575                                 // Step 1: Get some text (or reuse from previous iteratuon if checking
 576                                 //         for plausibility failed)
 577
 578                                 // Trying to get prefetch, if it has not been tried before
 579                                 if ( $text === false && isset( $this->prefetch ) && $prefetchNotTried ) {
 580                                         $prefetchNotTried = false;
 581                                         $tryIsPrefetch = true;
 582                                         $text = $this->prefetch->prefetch( intval( $this->thisPage ),
 583                                                 intval( $this->thisRev ) );
 584
 585                                         if ( $text === null ) {
 586                                                 $text = false;
 587                                         }
 588
 589                                         if ( is_string( $text ) && $model !== false ) {
 590                                                 // Apply export transformation to text coming from an old dump.
 591                                                 // The purpose of this transformation is to convert up from legacy
 592                                                 // formats, which may still be used in the older dump that is used
 593                                                 // for pre-fetching. Applying the transformation again should not
 594                                                 // interfere with content that is already in the correct form.
 595                                                 $text = $this->exportTransform( $text, $model, $format );
 596                                         }
 597                                 }
 598
 599                                 if ( $text === false ) {
 600                                         // Fallback to asking the database
 601                                         $tryIsPrefetch = false;
 602                                         if ( $this->spawn ) {
 603                                                 $text = $this->getTextSpawned( $id );
 604                                         } else {
 605                                                 $text = $this->getTextDb( $id );
 606                                         }
 607
 608                                         if ( $text !== false && $model !== false ) {
 609                                                 // Apply export transformation to text coming from the database.
 610                                                 // Prefetched text should already have transformations applied.
 611                                                 $text = $this->exportTransform( $text, $model, $format );
 612                                         }
 613
 614                                         // No more checks for texts from DB for now.
 615                                         // If we received something that is not false,
 616                                         // We treat it as good text, regardless of whether it actually is or is not
 617                                         if ( $text !== false ) {
 618                                                 return $text;
 619                                         }
 620                                 }
 621
 622                                 if ( $text === false ) {
 623                                         throw new MWException( "Generic error while obtaining text for id " . $id );
 624                                 }
 625
 626                                 // We received a good candidate for the text of $id via some method
 627
 628                                 // Step 2: Checking for plausibility and return the text if it is
 629                                 //         plausible
 630                                 $revID = intval( $this->thisRev );
 631                                 if ( !isset( $this->db ) ) {
 632                                         throw new MWException( "No database available" );
 633                                 }
 634
 635                                 if ( $model !== CONTENT_MODEL_WIKITEXT ) {
 636                                         $revLength = strlen( $text );
 637                                 } else {
 638                                         $revLength = $this->db->selectField( 'revision', 'rev_len', array( 'rev_id' => $revID ) );
 639                                 }
 640
 641                                 if ( strlen( $text ) == $revLength ) {
 642                                         if ( $tryIsPrefetch ) {
 643                                                 $this->prefetchCount++;
 644                                         }
 645
 646                                         return $text;
 647                                 }
 648
 649                                 $text = false;
 650                                 throw new MWException( "Received text is unplausible for id " . $id );
 651                         } catch ( Exception $e ) {
 652                                 $msg = "getting/checking text " . $id . " failed (" . $e->getMessage() . ")";
 653                                 if ( $failures + 1 < $this->maxFailures ) {
 654                                         $msg .= " (Will retry " . ( $this->maxFailures - $failures - 1 ) . " more times)";
 655                                 }
 656                                 $this->progress( $msg );
 657                         }
 658
 659                         // Something went wrong; we did not a text that was plausible :(
 660                         $failures++;
 661
 662                         // A failure in a prefetch hit does not warrant resetting db connection etc.
 663                         if ( !$tryIsPrefetch ) {
 664                                 // After backing off for some time, we try to reboot the whole process as
 665                                 // much as possible to not carry over failures from one part to the other
 666                                 // parts
 667                                 sleep( $this->failureTimeout );
 668                                 try {
 669                                         $this->rotateDb();
 670                                         if ( $this->spawn ) {
 671                                                 $this->closeSpawn();
 672                                                 $this->openSpawn();
 673                                         }
 674                                 } catch ( Exception $e ) {
 675                                         $this->progress( "Rebooting getText infrastructure failed (" . $e->getMessage() . ")" .
 676                                                 " Trying to continue anyways" );
 677                                 }
 678                         }
 679                 }
 680
 681                 // Retirieving a good text for $id failed (at least) maxFailures times.
 682                 // We abort for this $id.
 683
 684                 // Restoring the consecutive failures, and maybe aborting, if the dump
 685                 // is too broken.
 686                 $consecutiveFailedTextRetrievals = $oldConsecutiveFailedTextRetrievals + 1;
 687                 if ( $consecutiveFailedTextRetrievals > $this->maxConsecutiveFailedTextRetrievals ) {
 688                         throw new MWException( "Graceful storage failure" );
 689                 }
 690
 691                 return "";
 692         }
 693
 694         /**
 695          * May throw a database error if, say, the server dies during query.
 696          * @param int $id
 697          * @return bool|string
 698          * @throws MWException
 699          */
 700         private function getTextDb( $id ) {
 701                 global $wgContLang;
 702                 if ( !isset( $this->db ) ) {
 703                         throw new MWException( __METHOD__ . "No database available" );
 704                 }
 705                 $row = $this->db->selectRow( 'text',
 706                         array( 'old_text', 'old_flags' ),
 707                         array( 'old_id' => $id ),
 708                         __METHOD__ );
 709                 $text = Revision::getRevisionText( $row );
 710                 if ( $text === false ) {
 711                         return false;
 712                 }
 713                 $stripped = str_replace( "\r", "", $text );
 714                 $normalized = $wgContLang->normalize( $stripped );
 715
 716                 return $normalized;
 717         }
 718
 719         private function getTextSpawned( $id ) {
 720                 MediaWiki\suppressWarnings();
 721                 if ( !$this->spawnProc ) {
 722                         // First time?
 723                         $this->openSpawn();
 724                 }
 725                 $text = $this->getTextSpawnedOnce( $id );
 726                 MediaWiki\restoreWarnings();
 727
 728                 return $text;
 729         }
 730
 731         function openSpawn() {
 732                 global $IP;
 733
 734                 if ( file_exists( "$IP/../multiversion/MWScript.php" ) ) {
 735                         $cmd = implode( " ",
 736                                 array_map( 'wfEscapeShellArg',
 737                                         array(
 738                                                 $this->php,
 739                                                 "$IP/../multiversion/MWScript.php",
 740                                                 "fetchText.php",
 741                                                 '--wiki', wfWikiID() ) ) );
 742                 } else {
 743                         $cmd = implode( " ",
 744                                 array_map( 'wfEscapeShellArg',
 745                                         array(
 746                                                 $this->php,
 747                                                 "$IP/maintenance/fetchText.php",
 748                                                 '--wiki', wfWikiID() ) ) );
 749                 }
 750                 $spec = array(
 751                         0 => array( "pipe", "r" ),
 752                         1 => array( "pipe", "w" ),
 753                         2 => array( "file", "/dev/null", "a" ) );
 754                 $pipes = array();
 755
 756                 $this->progress( "Spawning database subprocess: $cmd" );
 757                 $this->spawnProc = proc_open( $cmd, $spec, $pipes );
 758                 if ( !$this->spawnProc ) {
 759                         $this->progress( "Subprocess spawn failed." );
 760
 761                         return false;
 762                 }
 763                 list(
 764                         $this->spawnWrite, // -> stdin
 765                         $this->spawnRead, // <- stdout
 766                 ) = $pipes;
 767
 768                 return true;
 769         }
 770
 771         private function closeSpawn() {
 772                 MediaWiki\suppressWarnings();
 773                 if ( $this->spawnRead ) {
 774                         fclose( $this->spawnRead );
 775                 }
 776                 $this->spawnRead = false;
 777                 if ( $this->spawnWrite ) {
 778                         fclose( $this->spawnWrite );
 779                 }
 780                 $this->spawnWrite = false;
 781                 if ( $this->spawnErr ) {
 782                         fclose( $this->spawnErr );
 783                 }
 784                 $this->spawnErr = false;
 785                 if ( $this->spawnProc ) {
 786                         pclose( $this->spawnProc );
 787                 }
 788                 $this->spawnProc = false;
 789                 MediaWiki\restoreWarnings();
 790         }
 791
 792         private function getTextSpawnedOnce( $id ) {
 793                 global $wgContLang;
 794
 795                 $ok = fwrite( $this->spawnWrite, "$id\n" );
 796                 // $this->progress( ">> $id" );
 797                 if ( !$ok ) {
 798                         return false;
 799                 }
 800
 801                 $ok = fflush( $this->spawnWrite );
 802                 // $this->progress( ">> [flush]" );
 803                 if ( !$ok ) {
 804                         return false;
 805                 }
 806
 807                 // check that the text id they are sending is the one we asked for
 808                 // this avoids out of sync revision text errors we have encountered in the past
 809                 $newId = fgets( $this->spawnRead );
 810                 if ( $newId === false ) {
 811                         return false;
 812                 }
 813                 if ( $id != intval( $newId ) ) {
 814                         return false;
 815                 }
 816
 817                 $len = fgets( $this->spawnRead );
 818                 // $this->progress( "<< " . trim( $len ) );
 819                 if ( $len === false ) {
 820                         return false;
 821                 }
 822
 823                 $nbytes = intval( $len );
 824                 // actual error, not zero-length text
 825                 if ( $nbytes < 0 ) {
 826                         return false;
 827                 }
 828
 829                 $text = "";
 830
 831                 // Subprocess may not send everything at once, we have to loop.
 832                 while ( $nbytes > strlen( $text ) ) {
 833                         $buffer = fread( $this->spawnRead, $nbytes - strlen( $text ) );
 834                         if ( $buffer === false ) {
 835                                 break;
 836                         }
 837                         $text .= $buffer;
 838                 }
 839
 840                 $gotbytes = strlen( $text );
 841                 if ( $gotbytes != $nbytes ) {
 842                         $this->progress( "Expected $nbytes bytes from database subprocess, got $gotbytes " );
 843
 844                         return false;
 845                 }
 846
 847                 // Do normalization in the dump thread...
 848                 $stripped = str_replace( "\r", "", $text );
 849                 $normalized = $wgContLang->normalize( $stripped );
 850
 851                 return $normalized;
 852         }
 853
 854         function startElement( $parser, $name, $attribs ) {
 855                 $this->checkpointJustWritten = false;
 856
 857                 $this->clearOpenElement( null );
 858                 $this->lastName = $name;
 859
 860                 if ( $name == 'revision' ) {
 861                         $this->state = $name;
 862                         $this->egress->writeOpenPage( null, $this->buffer );
 863                         $this->buffer = "";
 864                 } elseif ( $name == 'page' ) {
 865                         $this->state = $name;
 866                         if ( $this->atStart ) {
 867                                 $this->egress->writeOpenStream( $this->buffer );
 868                                 $this->buffer = "";
 869                                 $this->atStart = false;
 870                         }
 871                 }
 872
 873                 if ( $name == "text" && isset( $attribs['id'] ) ) {
 874                         $id = $attribs['id'];
 875                         $model = trim( $this->thisRevModel );
 876                         $format = trim( $this->thisRevFormat );
 877
 878                         $model = $model === '' ? null : $model;
 879                         $format = $format === '' ? null : $format;
 880
 881                         $text = $this->getText( $id, $model, $format );
 882                         $this->openElement = array( $name, array( 'xml:space' => 'preserve' ) );
 883                         if ( strlen( $text ) > 0 ) {
 884                                 $this->characterData( $parser, $text );
 885                         }
 886                 } else {
 887                         $this->openElement = array( $name, $attribs );
 888                 }
 889         }
 890
 891         function endElement( $parser, $name ) {
 892                 $this->checkpointJustWritten = false;
 893
 894                 if ( $this->openElement ) {
 895                         $this->clearOpenElement( "" );
 896                 } else {
 897                         $this->buffer .= "</$name>";
 898                 }
 899
 900                 if ( $name == 'revision' ) {
 901                         $this->egress->writeRevision( null, $this->buffer );
 902                         $this->buffer = "";
 903                         $this->thisRev = "";
 904                         $this->thisRevModel = null;
 905                         $this->thisRevFormat = null;
 906                 } elseif ( $name == 'page' ) {
 907                         if ( !$this->firstPageWritten ) {
 908                                 $this->firstPageWritten = trim( $this->thisPage );
 909                         }
 910                         $this->lastPageWritten = trim( $this->thisPage );
 911                         if ( $this->timeExceeded ) {
 912                                 $this->egress->writeClosePage( $this->buffer );
 913                                 // nasty hack, we can't just write the chardata after the
 914                                 // page tag, it will include leading blanks from the next line
 915                                 $this->egress->sink->write( "\n" );
 916
 917                                 $this->buffer = $this->xmlwriterobj->closeStream();
 918                                 $this->egress->writeCloseStream( $this->buffer );
 919
 920                                 $this->buffer = "";
 921                                 $this->thisPage = "";
 922                                 // this could be more than one file if we had more than one output arg
 923
 924                                 $filenameList = (array)$this->egress->getFilenames();
 925                                 $newFilenames = array();
 926                                 $firstPageID = str_pad( $this->firstPageWritten, 9, "0", STR_PAD_LEFT );
 927                                 $lastPageID = str_pad( $this->lastPageWritten, 9, "0", STR_PAD_LEFT );
 928                                 $filenamesCount = count( $filenameList );
 929                                 for ( $i = 0; $i < $filenamesCount; $i++ ) {
 930                                         $checkpointNameFilledIn = sprintf( $this->checkpointFiles[$i], $firstPageID, $lastPageID );
 931                                         $fileinfo = pathinfo( $filenameList[$i] );
 932                                         $newFilenames[] = $fileinfo['dirname'] . '/' . $checkpointNameFilledIn;
 933                                 }
 934                                 $this->egress->closeRenameAndReopen( $newFilenames );
 935                                 $this->buffer = $this->xmlwriterobj->openStream();
 936                                 $this->timeExceeded = false;
 937                                 $this->timeOfCheckpoint = $this->lastTime;
 938                                 $this->firstPageWritten = false;
 939                                 $this->checkpointJustWritten = true;
 940                         } else {
 941                                 $this->egress->writeClosePage( $this->buffer );
 942                                 $this->buffer = "";
 943                                 $this->thisPage = "";
 944                         }
 945                 } elseif ( $name == 'mediawiki' ) {
 946                         $this->egress->writeCloseStream( $this->buffer );
 947                         $this->buffer = "";
 948                 }
 949         }
 950
 951         function characterData( $parser, $data ) {
 952                 $this->clearOpenElement( null );
 953                 if ( $this->lastName == "id" ) {
 954                         if ( $this->state == "revision" ) {
 955                                 $this->thisRev .= $data;
 956                         } elseif ( $this->state == "page" ) {
 957                                 $this->thisPage .= $data;
 958                         }
 959                 } elseif ( $this->lastName == "model" ) {
 960                         $this->thisRevModel .= $data;
 961                 } elseif ( $this->lastName == "format" ) {
 962                         $this->thisRevFormat .= $data;
 963                 }
 964
 965                 // have to skip the newline left over from closepagetag line of
 966                 // end of checkpoint files. nasty hack!!
 967                 if ( $this->checkpointJustWritten ) {
 968                         if ( $data[0] == "\n" ) {
 969                                 $data = substr( $data, 1 );
 970                         }
 971                         $this->checkpointJustWritten = false;
 972                 }
 973                 $this->buffer .= htmlspecialchars( $data );
 974         }
 975
 976         function clearOpenElement( $style ) {
 977                 if ( $this->openElement ) {
 978                         $this->buffer .= Xml::element( $this->openElement[0], $this->openElement[1], $style );
 979                         $this->openElement = false;
 980                 }
 981         }
 982 }
 983
 984 $maintClass = 'TextPassDumper';
 985 require_once RUN_MAINTENANCE_IF_MAIN;