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__ . '/includes/BackupDumper.php';
  28 require_once __DIR__ . '/7zip.inc';
  29 require_once __DIR__ . '/../includes/export/WikiExporter.php';
  30
  31 use MediaWiki\MediaWikiServices;
  32 use MediaWiki\Shell\Shell;
  33 use MediaWiki\Storage\BlobAccessException;
  34 use MediaWiki\Storage\SqlBlobStore;
  35 use Wikimedia\Rdbms\IMaintainableDatabase;
  36
  37 /**
  38  * @ingroup Maintenance
  39  */
  40 class TextPassDumper extends BackupDumper {
  41         /** @var BaseDump */
  42         public $prefetch = null;
  43         /** @var string|bool */
  44         private $thisPage;
  45         /** @var string|bool */
  46         private $thisRev;
  47
  48         // when we spend more than maxTimeAllowed seconds on this run, we continue
  49         // processing until we write out the next complete page, then save output file(s),
  50         // rename it/them and open new one(s)
  51         public $maxTimeAllowed = 0; // 0 = no limit
  52
  53         protected $input = "php://stdin";
  54         protected $history = WikiExporter::FULL;
  55         protected $fetchCount = 0;
  56         protected $prefetchCount = 0;
  57         protected $prefetchCountLast = 0;
  58         protected $fetchCountLast = 0;
  59
  60         protected $maxFailures = 5;
  61         protected $maxConsecutiveFailedTextRetrievals = 200;
  62         protected $failureTimeout = 5; // Seconds to sleep after db failure
  63
  64         protected $bufferSize = 524288; // In bytes. Maximum size to read from the stub in on go.
  65
  66         protected $php = "php";
  67         protected $spawn = false;
  68
  69         /**
  70          * @var bool|resource
  71          */
  72         protected $spawnProc = false;
  73
  74         /**
  75          * @var bool|resource
  76          */
  77         protected $spawnWrite = false;
  78
  79         /**
  80          * @var bool|resource
  81          */
  82         protected $spawnRead = false;
  83
  84         /**
  85          * @var bool|resource
  86          */
  87         protected $spawnErr = false;
  88
  89         /**
  90          * @var bool|XmlDumpWriter
  91          */
  92         protected $xmlwriterobj = false;
  93
  94         protected $timeExceeded = false;
  95         protected $firstPageWritten = false;
  96         protected $lastPageWritten = false;
  97         protected $checkpointJustWritten = false;
  98         protected $checkpointFiles = [];
  99
 100         /**
 101          * @var IMaintainableDatabase
 102          */
 103         protected $db;
 104
 105         /**
 106          * @param array|null $args For backward compatibility
 107          */
 108         function __construct( $args = null ) {
 109                 parent::__construct();
 110
 111                 $this->addDescription( <<<TEXT
 112 This script postprocesses XML dumps from dumpBackup.php to add
 113 page text which was stubbed out (using --stub).
 114
 115 XML input is accepted on stdin.
 116 XML output is sent to stdout; progress reports are sent to stderr.
 117 TEXT
 118                 );
 119                 $this->stderr = fopen( "php://stderr", "wt" );
 120
 121                 $this->addOption( 'stub', 'To load a compressed stub dump instead of stdin. ' .
 122                         'Specify as --stub=<type>:<file>.', false, true );
 123                 $this->addOption( 'prefetch', 'Use a prior dump file as a text source, to savepressure on the ' .
 124                         'database. (Requires the XMLReader extension). Specify as --prefetch=<type>:<file>',
 125                         false, true );
 126                 $this->addOption( 'maxtime', 'Write out checkpoint file after this many minutes (writing' .
 127                         'out complete page, closing xml file properly, and opening new one' .
 128                         'with header).  This option requires the checkpointfile option.', false, true );
 129                 $this->addOption( 'checkpointfile', 'Use this string for checkpoint filenames,substituting ' .
 130                         'first pageid written for the first %s (required) and the last pageid written for the ' .
 131                         'second %s if it exists.', false, true, false, true ); // This can be specified multiple times
 132                 $this->addOption( 'quiet', 'Don\'t dump status reports to stderr.' );
 133                 $this->addOption( 'full', 'Dump all revisions of every page' );
 134                 $this->addOption( 'current', 'Base ETA on number of pages in database instead of all revisions' );
 135                 $this->addOption( 'spawn', 'Spawn a subprocess for loading text records' );
 136                 $this->addOption( 'buffersize', 'Buffer size in bytes to use for reading the stub. ' .
 137                         '(Default: 512KB, Minimum: 4KB)', false, true );
 138
 139                 if ( $args ) {
 140                         $this->loadWithArgv( $args );
 141                         $this->processOptions();
 142                 }
 143         }
 144
 145         /**
 146          * @return SqlBlobStore
 147          */
 148         private function getBlobStore() {
 149                 return MediaWikiServices::getInstance()->getBlobStore();
 150         }
 151
 152         function execute() {
 153                 $this->processOptions();
 154                 $this->dump( true );
 155         }
 156
 157         function processOptions() {
 158                 parent::processOptions();
 159
 160                 if ( $this->hasOption( 'buffersize' ) ) {
 161                         $this->bufferSize = max( intval( $this->getOption( 'buffersize' ) ), 4 * 1024 );
 162                 }
 163
 164                 if ( $this->hasOption( 'prefetch' ) ) {
 165                         $url = $this->processFileOpt( $this->getOption( 'prefetch' ) );
 166                         $this->prefetch = new BaseDump( $url );
 167                 }
 168
 169                 if ( $this->hasOption( 'stub' ) ) {
 170                         $this->input = $this->processFileOpt( $this->getOption( 'stub' ) );
 171                 }
 172
 173                 if ( $this->hasOption( 'maxtime' ) ) {
 174                         $this->maxTimeAllowed = intval( $this->getOption( 'maxtime' ) ) * 60;
 175                 }
 176
 177                 if ( $this->hasOption( 'checkpointfile' ) ) {
 178                         $this->checkpointFiles = $this->getOption( 'checkpointfile' );
 179                 }
 180
 181                 if ( $this->hasOption( 'current' ) ) {
 182                         $this->history = WikiExporter::CURRENT;
 183                 }
 184
 185                 if ( $this->hasOption( 'full' ) ) {
 186                         $this->history = WikiExporter::FULL;
 187                 }
 188
 189                 if ( $this->hasOption( 'spawn' ) ) {
 190                         $this->spawn = true;
 191                         $val = $this->getOption( 'spawn' );
 192                         if ( $val !== 1 ) {
 193                                 $this->php = $val;
 194                         }
 195                 }
 196         }
 197
 198         /**
 199          * Drop the database connection $this->db and try to get a new one.
 200          *
 201          * This function tries to get a /different/ connection if this is
 202          * possible. Hence, (if this is possible) it switches to a different
 203          * failover upon each call.
 204          *
 205          * This function resets $this->lb and closes all connections on it.
 206          *
 207          * @throws MWException
 208          */
 209         function rotateDb() {
 210                 // Cleaning up old connections
 211                 if ( isset( $this->lb ) ) {
 212                         $this->lb->closeAll();
 213                         unset( $this->lb );
 214                 }
 215
 216                 if ( $this->forcedDb !== null ) {
 217                         $this->db = $this->forcedDb;
 218
 219                         return;
 220                 }
 221
 222                 if ( isset( $this->db ) && $this->db->isOpen() ) {
 223                         throw new MWException( 'DB is set and has not been closed by the Load Balancer' );
 224                 }
 225
 226                 unset( $this->db );
 227
 228                 // Trying to set up new connection.
 229                 // We do /not/ retry upon failure, but delegate to encapsulating logic, to avoid
 230                 // individually retrying at different layers of code.
 231
 232                 try {
 233                         $lbFactory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
 234                         $this->lb = $lbFactory->newMainLB();
 235                 } catch ( Exception $e ) {
 236                         throw new MWException( __METHOD__
 237                                 . " rotating DB failed to obtain new load balancer (" . $e->getMessage() . ")" );
 238                 }
 239
 240                 try {
 241                         $this->db = $this->lb->getConnection( DB_REPLICA, 'dump' );
 242                 } catch ( Exception $e ) {
 243                         throw new MWException( __METHOD__
 244                                 . " rotating DB failed to obtain new database (" . $e->getMessage() . ")" );
 245                 }
 246         }
 247
 248         function initProgress( $history = WikiExporter::FULL ) {
 249                 parent::initProgress();
 250                 $this->timeOfCheckpoint = $this->startTime;
 251         }
 252
 253         function dump( $history, $text = WikiExporter::TEXT ) {
 254                 // Notice messages will foul up your XML output even if they're
 255                 // relatively harmless.
 256                 if ( ini_get( 'display_errors' ) ) {
 257                         ini_set( 'display_errors', 'stderr' );
 258                 }
 259
 260                 $this->initProgress( $this->history );
 261
 262                 // We are trying to get an initial database connection to avoid that the
 263                 // first try of this request's first call to getText fails. However, if
 264                 // obtaining a good DB connection fails it's not a serious issue, as
 265                 // getText does retry upon failure and can start without having a working
 266                 // DB connection.
 267                 try {
 268                         $this->rotateDb();
 269                 } catch ( Exception $e ) {
 270                         // We do not even count this as failure. Just let eventual
 271                         // watchdogs know.
 272                         $this->progress( "Getting initial DB connection failed (" .
 273                                 $e->getMessage() . ")" );
 274                 }
 275
 276                 $this->egress = new ExportProgressFilter( $this->sink, $this );
 277
 278                 // it would be nice to do it in the constructor, oh well. need egress set
 279                 $this->finalOptionCheck();
 280
 281                 // we only want this so we know how to close a stream :-P
 282                 $this->xmlwriterobj = new XmlDumpWriter();
 283
 284                 $input = fopen( $this->input, "rt" );
 285                 $this->readDump( $input );
 286
 287                 if ( $this->spawnProc ) {
 288                         $this->closeSpawn();
 289                 }
 290
 291                 $this->report( true );
 292         }
 293
 294         function processFileOpt( $opt ) {
 295                 $split = explode( ':', $opt, 2 );
 296                 $val = $split[0];
 297                 $param = '';
 298                 if ( count( $split ) === 2 ) {
 299                         $param = $split[1];
 300                 }
 301                 $fileURIs = explode( ';', $param );
 302                 foreach ( $fileURIs as $URI ) {
 303                         switch ( $val ) {
 304                                 case "file":
 305                                         $newURI = $URI;
 306                                         break;
 307                                 case "gzip":
 308                                         $newURI = "compress.zlib://$URI";
 309                                         break;
 310                                 case "bzip2":
 311                                         $newURI = "compress.bzip2://$URI";
 312                                         break;
 313                                 case "7zip":
 314                                         $newURI = "mediawiki.compress.7z://$URI";
 315                                         break;
 316                                 default:
 317                                         $newURI = $URI;
 318                         }
 319                         $newFileURIs[] = $newURI;
 320                 }
 321                 $val = implode( ';', $newFileURIs );
 322
 323                 return $val;
 324         }
 325
 326         /**
 327          * Overridden to include prefetch ratio if enabled.
 328          */
 329         function showReport() {
 330                 if ( !$this->prefetch ) {
 331                         parent::showReport();
 332
 333                         return;
 334                 }
 335
 336                 if ( $this->reporting ) {
 337                         $now = wfTimestamp( TS_DB );
 338                         $nowts = microtime( true );
 339                         $deltaAll = $nowts - $this->startTime;
 340                         $deltaPart = $nowts - $this->lastTime;
 341                         $this->pageCountPart = $this->pageCount - $this->pageCountLast;
 342                         $this->revCountPart = $this->revCount - $this->revCountLast;
 343
 344                         if ( $deltaAll ) {
 345                                 $portion = $this->revCount / $this->maxCount;
 346                                 $eta = $this->startTime + $deltaAll / $portion;
 347                                 $etats = wfTimestamp( TS_DB, intval( $eta ) );
 348                                 if ( $this->fetchCount ) {
 349                                         $fetchRate = 100.0 * $this->prefetchCount / $this->fetchCount;
 350                                 } else {
 351                                         $fetchRate = '-';
 352                                 }
 353                                 $pageRate = $this->pageCount / $deltaAll;
 354                                 $revRate = $this->revCount / $deltaAll;
 355                         } else {
 356                                 $pageRate = '-';
 357                                 $revRate = '-';
 358                                 $etats = '-';
 359                                 $fetchRate = '-';
 360                         }
 361                         if ( $deltaPart ) {
 362                                 if ( $this->fetchCountLast ) {
 363                                         $fetchRatePart = 100.0 * $this->prefetchCountLast / $this->fetchCountLast;
 364                                 } else {
 365                                         $fetchRatePart = '-';
 366                                 }
 367                                 $pageRatePart = $this->pageCountPart / $deltaPart;
 368                                 $revRatePart = $this->revCountPart / $deltaPart;
 369                         } else {
 370                                 $fetchRatePart = '-';
 371                                 $pageRatePart = '-';
 372                                 $revRatePart = '-';
 373                         }
 374                         $this->progress( sprintf(
 375                                 "%s: %s (ID %d) %d pages (%0.1f|%0.1f/sec all|curr), "
 376                                         . "%d revs (%0.1f|%0.1f/sec all|curr), %0.1f%%|%0.1f%% "
 377                                         . "prefetched (all|curr), ETA %s [max %d]",
 378                                 $now, wfWikiID(), $this->ID, $this->pageCount, $pageRate,
 379                                 $pageRatePart, $this->revCount, $revRate, $revRatePart,
 380                                 $fetchRate, $fetchRatePart, $etats, $this->maxCount
 381                         ) );
 382                         $this->lastTime = $nowts;
 383                         $this->revCountLast = $this->revCount;
 384                         $this->prefetchCountLast = $this->prefetchCount;
 385                         $this->fetchCountLast = $this->fetchCount;
 386                 }
 387         }
 388
 389         function setTimeExceeded() {
 390                 $this->timeExceeded = true;
 391         }
 392
 393         function checkIfTimeExceeded() {
 394                 if ( $this->maxTimeAllowed
 395                         && ( $this->lastTime - $this->timeOfCheckpoint > $this->maxTimeAllowed )
 396                 ) {
 397                         return true;
 398                 }
 399
 400                 return false;
 401         }
 402
 403         function finalOptionCheck() {
 404                 if ( ( $this->checkpointFiles && !$this->maxTimeAllowed )
 405                         || ( $this->maxTimeAllowed && !$this->checkpointFiles )
 406                 ) {
 407                         throw new MWException( "Options checkpointfile and maxtime must be specified together.\n" );
 408                 }
 409                 foreach ( $this->checkpointFiles as $checkpointFile ) {
 410                         $count = substr_count( $checkpointFile, "%s" );
 411                         if ( $count != 2 ) {
 412                                 throw new MWException( "Option checkpointfile must contain two '%s' "
 413                                         . "for substitution of first and last pageids, count is $count instead, "
 414                                         . "file is $checkpointFile.\n" );
 415                         }
 416                 }
 417
 418                 if ( $this->checkpointFiles ) {
 419                         $filenameList = (array)$this->egress->getFilenames();
 420                         if ( count( $filenameList ) != count( $this->checkpointFiles ) ) {
 421                                 throw new MWException( "One checkpointfile must be specified "
 422                                         . "for each output option, if maxtime is used.\n" );
 423                         }
 424                 }
 425         }
 426
 427         /**
 428          * @throws MWException Failure to parse XML input
 429          * @param string $input
 430          * @return bool
 431          */
 432         function readDump( $input ) {
 433                 $this->buffer = "";
 434                 $this->openElement = false;
 435                 $this->atStart = true;
 436                 $this->state = "";
 437                 $this->lastName = "";
 438                 $this->thisPage = 0;
 439                 $this->thisRev = 0;
 440                 $this->thisRevModel = null;
 441                 $this->thisRevFormat = null;
 442
 443                 $parser = xml_parser_create( "UTF-8" );
 444                 xml_parser_set_option( $parser, XML_OPTION_CASE_FOLDING, false );
 445
 446                 xml_set_element_handler(
 447                         $parser,
 448                         [ $this, 'startElement' ],
 449                         [ $this, 'endElement' ]
 450                 );
 451                 xml_set_character_data_handler( $parser, [ $this, 'characterData' ] );
 452
 453                 $offset = 0; // for context extraction on error reporting
 454                 do {
 455                         if ( $this->checkIfTimeExceeded() ) {
 456                                 $this->setTimeExceeded();
 457                         }
 458                         $chunk = fread( $input, $this->bufferSize );
 459                         if ( !xml_parse( $parser, $chunk, feof( $input ) ) ) {
 460                                 wfDebug( "TextDumpPass::readDump encountered XML parsing error\n" );
 461
 462                                 $byte = xml_get_current_byte_index( $parser );
 463                                 $msg = wfMessage( 'xml-error-string',
 464                                         'XML import parse failure',
 465                                         xml_get_current_line_number( $parser ),
 466                                         xml_get_current_column_number( $parser ),
 467                                         $byte . ( is_null( $chunk ) ? null : ( '; "' . substr( $chunk, $byte - $offset, 16 ) . '"' ) ),
 468                                         xml_error_string( xml_get_error_code( $parser ) ) )->escaped();
 469
 470                                 xml_parser_free( $parser );
 471
 472                                 throw new MWException( $msg );
 473                         }
 474                         $offset += strlen( $chunk );
 475                 } while ( $chunk !== false && !feof( $input ) );
 476                 if ( $this->maxTimeAllowed ) {
 477                         $filenameList = (array)$this->egress->getFilenames();
 478                         // we wrote some stuff after last checkpoint that needs renamed
 479                         if ( file_exists( $filenameList[0] ) ) {
 480                                 $newFilenames = [];
 481                                 # we might have just written the header and footer and had no
 482                                 # pages or revisions written... perhaps they were all deleted
 483                                 # there's no pageID 0 so we use that. the caller is responsible
 484                                 # for deciding what to do with a file containing only the
 485                                 # siteinfo information and the mw tags.
 486                                 if ( !$this->firstPageWritten ) {
 487                                         $firstPageID = str_pad( 0, 9, "0", STR_PAD_LEFT );
 488                                         $lastPageID = str_pad( 0, 9, "0", STR_PAD_LEFT );
 489                                 } else {
 490                                         $firstPageID = str_pad( $this->firstPageWritten, 9, "0", STR_PAD_LEFT );
 491                                         $lastPageID = str_pad( $this->lastPageWritten, 9, "0", STR_PAD_LEFT );
 492                                 }
 493
 494                                 $filenameCount = count( $filenameList );
 495                                 for ( $i = 0; $i < $filenameCount; $i++ ) {
 496                                         $checkpointNameFilledIn = sprintf( $this->checkpointFiles[$i], $firstPageID, $lastPageID );
 497                                         $fileinfo = pathinfo( $filenameList[$i] );
 498                                         $newFilenames[] = $fileinfo['dirname'] . '/' . $checkpointNameFilledIn;
 499                                 }
 500                                 $this->egress->closeAndRename( $newFilenames );
 501                         }
 502                 }
 503                 xml_parser_free( $parser );
 504
 505                 return true;
 506         }
 507
 508         /**
 509          * Applies applicable export transformations to $text.
 510          *
 511          * @param string $text
 512          * @param string $model
 513          * @param string|null $format
 514          *
 515          * @return string
 516          */
 517         private function exportTransform( $text, $model, $format = null ) {
 518                 try {
 519                         $handler = ContentHandler::getForModelID( $model );
 520                         $text = $handler->exportTransform( $text, $format );
 521                 }
 522                 catch ( MWException $ex ) {
 523                         $this->progress(
 524                                 "Unable to apply export transformation for content model '$model': " .
 525                                 $ex->getMessage()
 526                         );
 527                 }
 528
 529                 return $text;
 530         }
 531
 532         /**
 533          * Tries to load revision text.
 534          * Export transformations are applied if the content model is given or can be
 535          * determined from the database.
 536          *
 537          * Upon errors, retries (Up to $this->maxFailures tries each call).
 538          * If still no good revision could be found even after this retrying, "" is returned.
 539          * If no good revision text could be returned for
 540          * $this->maxConsecutiveFailedTextRetrievals consecutive calls to getText, MWException
 541          * is thrown.
 542          *
 543          * @param int|string $id Content address, or text row ID.
 544          * @param string|bool|null $model The content model used to determine
 545          *  applicable export transformations.
 546          *  If $model is null, it will be determined from the database.
 547          * @param string|null $format The content format used when applying export transformations.
 548          *
 549          * @throws MWException
 550          * @return string The revision text for $id, or ""
 551          */
 552         function getText( $id, $model = null, $format = null ) {
 553                 global $wgContentHandlerUseDB;
 554
 555                 $prefetchNotTried = true; // Whether or not we already tried to get the text via prefetch.
 556                 $text = false; // The candidate for a good text. false if no proper value.
 557                 $failures = 0; // The number of times, this invocation of getText already failed.
 558
 559                 // The number of times getText failed without yielding a good text in between.
 560                 static $consecutiveFailedTextRetrievals = 0;
 561
 562                 $this->fetchCount++;
 563
 564                 // To allow to simply return on success and do not have to worry about book keeping,
 565                 // we assume, this fetch works (possible after some retries). Nevertheless, we koop
 566                 // the old value, so we can restore it, if problems occur (See after the while loop).
 567                 $oldConsecutiveFailedTextRetrievals = $consecutiveFailedTextRetrievals;
 568                 $consecutiveFailedTextRetrievals = 0;
 569
 570                 if ( $model === null && $wgContentHandlerUseDB ) {
 571                         // TODO: MCR: use content table
 572                         $row = $this->db->selectRow(
 573                                 'revision',
 574                                 [ 'rev_content_model', 'rev_content_format' ],
 575                                 [ 'rev_id' => $this->thisRev ],
 576                                 __METHOD__
 577                         );
 578
 579                         if ( $row ) {
 580                                 $model = $row->rev_content_model;
 581                                 $format = $row->rev_content_format;
 582                         }
 583                 }
 584
 585                 if ( $model === null || $model === '' ) {
 586                         $model = false;
 587                 }
 588
 589                 while ( $failures < $this->maxFailures ) {
 590                         // As soon as we found a good text for the $id, we will return immediately.
 591                         // Hence, if we make it past the try catch block, we know that we did not
 592                         // find a good text.
 593
 594                         try {
 595                                 // Step 1: Get some text (or reuse from previous iteratuon if checking
 596                                 //         for plausibility failed)
 597
 598                                 // Trying to get prefetch, if it has not been tried before
 599                                 if ( $text === false && isset( $this->prefetch ) && $prefetchNotTried ) {
 600                                         $prefetchNotTried = false;
 601                                         $tryIsPrefetch = true;
 602                                         $text = $this->prefetch->prefetch( (int)$this->thisPage, (int)$this->thisRev );
 603
 604                                         if ( $text === null ) {
 605                                                 $text = false;
 606                                         }
 607
 608                                         if ( is_string( $text ) && $model !== false ) {
 609                                                 // Apply export transformation to text coming from an old dump.
 610                                                 // The purpose of this transformation is to convert up from legacy
 611                                                 // formats, which may still be used in the older dump that is used
 612                                                 // for pre-fetching. Applying the transformation again should not
 613                                                 // interfere with content that is already in the correct form.
 614                                                 $text = $this->exportTransform( $text, $model, $format );
 615                                         }
 616                                 }
 617
 618                                 if ( $text === false ) {
 619                                         // Fallback to asking the database
 620                                         $tryIsPrefetch = false;
 621                                         if ( $this->spawn ) {
 622                                                 $text = $this->getTextSpawned( $id );
 623                                         } else {
 624                                                 $text = $this->getTextDb( $id );
 625                                         }
 626
 627                                         if ( $text !== false && $model !== false ) {
 628                                                 // Apply export transformation to text coming from the database.
 629                                                 // Prefetched text should already have transformations applied.
 630                                                 $text = $this->exportTransform( $text, $model, $format );
 631                                         }
 632
 633                                         // No more checks for texts from DB for now.
 634                                         // If we received something that is not false,
 635                                         // We treat it as good text, regardless of whether it actually is or is not
 636                                         if ( $text !== false ) {
 637                                                 return $text;
 638                                         }
 639                                 }
 640
 641                                 if ( $text === false ) {
 642                                         throw new MWException( "Generic error while obtaining text for id " . $id );
 643                                 }
 644
 645                                 // We received a good candidate for the text of $id via some method
 646
 647                                 // Step 2: Checking for plausibility and return the text if it is
 648                                 //         plausible
 649                                 $revID = intval( $this->thisRev );
 650                                 if ( !isset( $this->db ) ) {
 651                                         throw new MWException( "No database available" );
 652                                 }
 653
 654                                 if ( $model !== CONTENT_MODEL_WIKITEXT ) {
 655                                         $revLength = strlen( $text );
 656                                 } else {
 657                                         $revLength = $this->db->selectField( 'revision', 'rev_len', [ 'rev_id' => $revID ] );
 658                                 }
 659
 660                                 if ( strlen( $text ) == $revLength ) {
 661                                         if ( $tryIsPrefetch ) {
 662                                                 $this->prefetchCount++;
 663                                         }
 664
 665                                         return $text;
 666                                 }
 667
 668                                 $text = false;
 669                                 throw new MWException( "Received text is unplausible for id " . $id );
 670                         } catch ( Exception $e ) {
 671                                 $msg = "getting/checking text " . $id . " failed (" . $e->getMessage() . ")";
 672                                 if ( $failures + 1 < $this->maxFailures ) {
 673                                         $msg .= " (Will retry " . ( $this->maxFailures - $failures - 1 ) . " more times)";
 674                                 }
 675                                 $this->progress( $msg );
 676                         }
 677
 678                         // Something went wrong; we did not a text that was plausible :(
 679                         $failures++;
 680
 681                         // A failure in a prefetch hit does not warrant resetting db connection etc.
 682                         if ( !$tryIsPrefetch ) {
 683                                 // After backing off for some time, we try to reboot the whole process as
 684                                 // much as possible to not carry over failures from one part to the other
 685                                 // parts
 686                                 sleep( $this->failureTimeout );
 687                                 try {
 688                                         $this->rotateDb();
 689                                         if ( $this->spawn ) {
 690                                                 $this->closeSpawn();
 691                                                 $this->openSpawn();
 692                                         }
 693                                 } catch ( Exception $e ) {
 694                                         $this->progress( "Rebooting getText infrastructure failed (" . $e->getMessage() . ")" .
 695                                                 " Trying to continue anyways" );
 696                                 }
 697                         }
 698                 }
 699
 700                 // Retirieving a good text for $id failed (at least) maxFailures times.
 701                 // We abort for this $id.
 702
 703                 // Restoring the consecutive failures, and maybe aborting, if the dump
 704                 // is too broken.
 705                 $consecutiveFailedTextRetrievals = $oldConsecutiveFailedTextRetrievals + 1;
 706                 if ( $consecutiveFailedTextRetrievals > $this->maxConsecutiveFailedTextRetrievals ) {
 707                         throw new MWException( "Graceful storage failure" );
 708                 }
 709
 710                 return "";
 711         }
 712
 713         /**
 714          * Loads the serialized content from storage.
 715          *
 716          * @param int|string $id Content address, or text row ID.
 717          * @return bool|string
 718          */
 719         private function getTextDb( $id ) {
 720                 $store = $this->getBlobStore();
 721                 $address = ( is_int( $id ) || strpos( $id, ':' ) === false )
 722                         ? SqlBlobStore::makeAddressFromTextId( (int)$id )
 723                         : $id;
 724
 725                 try {
 726                         $text = $store->getBlob( $address );
 727
 728                         $stripped = str_replace( "\r", "", $text );
 729                         $normalized = MediaWikiServices::getInstance()->getContentLanguage()
 730                                 ->normalize( $stripped );
 731
 732                         return $normalized;
 733                 } catch ( BlobAccessException $ex ) {
 734                         // XXX: log a warning?
 735                         return false;
 736                 }
 737         }
 738
 739         /**
 740          * @param int|string $address Content address, or text row ID.
 741          * @return bool|string
 742          */
 743         private function getTextSpawned( $address ) {
 744                 Wikimedia\suppressWarnings();
 745                 if ( !$this->spawnProc ) {
 746                         // First time?
 747                         $this->openSpawn();
 748                 }
 749                 $text = $this->getTextSpawnedOnce( $address );
 750                 Wikimedia\restoreWarnings();
 751
 752                 return $text;
 753         }
 754
 755         function openSpawn() {
 756                 global $IP;
 757
 758                 if ( file_exists( "$IP/../multiversion/MWScript.php" ) ) {
 759                         $cmd = implode( " ",
 760                                 array_map( [ Shell::class, 'escape' ],
 761                                         [
 762                                                 $this->php,
 763                                                 "$IP/../multiversion/MWScript.php",
 764                                                 "fetchText.php",
 765                                                 '--wiki', wfWikiID() ] ) );
 766                 } else {
 767                         $cmd = implode( " ",
 768                                 array_map( [ Shell::class, 'escape' ],
 769                                         [
 770                                                 $this->php,
 771                                                 "$IP/maintenance/fetchText.php",
 772                                                 '--wiki', wfWikiID() ] ) );
 773                 }
 774                 $spec = [
 775                         0 => [ "pipe", "r" ],
 776                         1 => [ "pipe", "w" ],
 777                         2 => [ "file", "/dev/null", "a" ] ];
 778                 $pipes = [];
 779
 780                 $this->progress( "Spawning database subprocess: $cmd" );
 781                 $this->spawnProc = proc_open( $cmd, $spec, $pipes );
 782                 if ( !$this->spawnProc ) {
 783                         $this->progress( "Subprocess spawn failed." );
 784
 785                         return false;
 786                 }
 787                 list(
 788                         $this->spawnWrite, // -> stdin
 789                         $this->spawnRead, // <- stdout
 790                 ) = $pipes;
 791
 792                 return true;
 793         }
 794
 795         private function closeSpawn() {
 796                 Wikimedia\suppressWarnings();
 797                 if ( $this->spawnRead ) {
 798                         fclose( $this->spawnRead );
 799                 }
 800                 $this->spawnRead = false;
 801                 if ( $this->spawnWrite ) {
 802                         fclose( $this->spawnWrite );
 803                 }
 804                 $this->spawnWrite = false;
 805                 if ( $this->spawnErr ) {
 806                         fclose( $this->spawnErr );
 807                 }
 808                 $this->spawnErr = false;
 809                 if ( $this->spawnProc ) {
 810                         pclose( $this->spawnProc );
 811                 }
 812                 $this->spawnProc = false;
 813                 Wikimedia\restoreWarnings();
 814         }
 815
 816         /**
 817          * @param int|string $address Content address, or text row ID.
 818          * @return bool|string
 819          */
 820         private function getTextSpawnedOnce( $address ) {
 821                 if ( is_int( $address ) || intval( $address ) ) {
 822                         $address = SqlBlobStore::makeAddressFromTextId( (int)$address );
 823                 }
 824
 825                 $ok = fwrite( $this->spawnWrite, "$address\n" );
 826                 // $this->progress( ">> $id" );
 827                 if ( !$ok ) {
 828                         return false;
 829                 }
 830
 831                 $ok = fflush( $this->spawnWrite );
 832                 // $this->progress( ">> [flush]" );
 833                 if ( !$ok ) {
 834                         return false;
 835                 }
 836
 837                 // check that the text address they are sending is the one we asked for
 838                 // this avoids out of sync revision text errors we have encountered in the past
 839                 $newAddress = fgets( $this->spawnRead );
 840                 if ( $newAddress === false ) {
 841                         return false;
 842                 }
 843                 $newAddress = trim( $newAddress );
 844                 if ( strpos( $newAddress, ':' ) === false ) {
 845                         $newAddress = SqlBlobStore::makeAddressFromTextId( intval( $newAddress ) );
 846                 }
 847
 848                 if ( $newAddress !== $address ) {
 849                         return false;
 850                 }
 851
 852                 $len = fgets( $this->spawnRead );
 853                 // $this->progress( "<< " . trim( $len ) );
 854                 if ( $len === false ) {
 855                         return false;
 856                 }
 857
 858                 $nbytes = intval( $len );
 859                 // actual error, not zero-length text
 860                 if ( $nbytes < 0 ) {
 861                         return false;
 862                 }
 863
 864                 $text = "";
 865
 866                 // Subprocess may not send everything at once, we have to loop.
 867                 while ( $nbytes > strlen( $text ) ) {
 868                         $buffer = fread( $this->spawnRead, $nbytes - strlen( $text ) );
 869                         if ( $buffer === false ) {
 870                                 break;
 871                         }
 872                         $text .= $buffer;
 873                 }
 874
 875                 $gotbytes = strlen( $text );
 876                 if ( $gotbytes != $nbytes ) {
 877                         $this->progress( "Expected $nbytes bytes from database subprocess, got $gotbytes " );
 878
 879                         return false;
 880                 }
 881
 882                 // Do normalization in the dump thread...
 883                 $stripped = str_replace( "\r", "", $text );
 884                 $normalized = MediaWikiServices::getInstance()->getContentLanguage()->
 885                         normalize( $stripped );
 886
 887                 return $normalized;
 888         }
 889
 890         function startElement( $parser, $name, $attribs ) {
 891                 $this->checkpointJustWritten = false;
 892
 893                 $this->clearOpenElement( null );
 894                 $this->lastName = $name;
 895
 896                 if ( $name == 'revision' ) {
 897                         $this->state = $name;
 898                         $this->egress->writeOpenPage( null, $this->buffer );
 899                         $this->buffer = "";
 900                 } elseif ( $name == 'page' ) {
 901                         $this->state = $name;
 902                         if ( $this->atStart ) {
 903                                 $this->egress->writeOpenStream( $this->buffer );
 904                                 $this->buffer = "";
 905                                 $this->atStart = false;
 906                         }
 907                 }
 908
 909                 if ( $name == "text" && isset( $attribs['id'] ) ) {
 910                         $id = $attribs['id'];
 911                         $model = trim( $this->thisRevModel );
 912                         $format = trim( $this->thisRevFormat );
 913
 914                         $model = $model === '' ? null : $model;
 915                         $format = $format === '' ? null : $format;
 916
 917                         $text = $this->getText( $id, $model, $format );
 918                         $this->openElement = [ $name, [ 'xml:space' => 'preserve' ] ];
 919                         if ( strlen( $text ) > 0 ) {
 920                                 $this->characterData( $parser, $text );
 921                         }
 922                 } else {
 923                         $this->openElement = [ $name, $attribs ];
 924                 }
 925         }
 926
 927         function endElement( $parser, $name ) {
 928                 $this->checkpointJustWritten = false;
 929
 930                 if ( $this->openElement ) {
 931                         $this->clearOpenElement( "" );
 932                 } else {
 933                         $this->buffer .= "</$name>";
 934                 }
 935
 936                 if ( $name == 'revision' ) {
 937                         $this->egress->writeRevision( null, $this->buffer );
 938                         $this->buffer = "";
 939                         $this->thisRev = "";
 940                         $this->thisRevModel = null;
 941                         $this->thisRevFormat = null;
 942                 } elseif ( $name == 'page' ) {
 943                         if ( !$this->firstPageWritten ) {
 944                                 $this->firstPageWritten = trim( $this->thisPage );
 945                         }
 946                         $this->lastPageWritten = trim( $this->thisPage );
 947                         if ( $this->timeExceeded ) {
 948                                 $this->egress->writeClosePage( $this->buffer );
 949                                 // nasty hack, we can't just write the chardata after the
 950                                 // page tag, it will include leading blanks from the next line
 951                                 $this->egress->sink->write( "\n" );
 952
 953                                 $this->buffer = $this->xmlwriterobj->closeStream();
 954                                 $this->egress->writeCloseStream( $this->buffer );
 955
 956                                 $this->buffer = "";
 957                                 $this->thisPage = "";
 958                                 // this could be more than one file if we had more than one output arg
 959
 960                                 $filenameList = (array)$this->egress->getFilenames();
 961                                 $newFilenames = [];
 962                                 $firstPageID = str_pad( $this->firstPageWritten, 9, "0", STR_PAD_LEFT );
 963                                 $lastPageID = str_pad( $this->lastPageWritten, 9, "0", STR_PAD_LEFT );
 964                                 $filenamesCount = count( $filenameList );
 965                                 for ( $i = 0; $i < $filenamesCount; $i++ ) {
 966                                         $checkpointNameFilledIn = sprintf( $this->checkpointFiles[$i], $firstPageID, $lastPageID );
 967                                         $fileinfo = pathinfo( $filenameList[$i] );
 968                                         $newFilenames[] = $fileinfo['dirname'] . '/' . $checkpointNameFilledIn;
 969                                 }
 970                                 $this->egress->closeRenameAndReopen( $newFilenames );
 971                                 $this->buffer = $this->xmlwriterobj->openStream();
 972                                 $this->timeExceeded = false;
 973                                 $this->timeOfCheckpoint = $this->lastTime;
 974                                 $this->firstPageWritten = false;
 975                                 $this->checkpointJustWritten = true;
 976                         } else {
 977                                 $this->egress->writeClosePage( $this->buffer );
 978                                 $this->buffer = "";
 979                                 $this->thisPage = "";
 980                         }
 981                 } elseif ( $name == 'mediawiki' ) {
 982                         $this->egress->writeCloseStream( $this->buffer );
 983                         $this->buffer = "";
 984                 }
 985         }
 986
 987         function characterData( $parser, $data ) {
 988                 $this->clearOpenElement( null );
 989                 if ( $this->lastName == "id" ) {
 990                         if ( $this->state == "revision" ) {
 991                                 $this->thisRev .= $data;
 992                         } elseif ( $this->state == "page" ) {
 993                                 $this->thisPage .= $data;
 994                         }
 995                 } elseif ( $this->lastName == "model" ) {
 996                         $this->thisRevModel .= $data;
 997                 } elseif ( $this->lastName == "format" ) {
 998                         $this->thisRevFormat .= $data;
 999                 }
1000
1001                 // have to skip the newline left over from closepagetag line of
1002                 // end of checkpoint files. nasty hack!!
1003                 if ( $this->checkpointJustWritten ) {
1004                         if ( $data[0] == "\n" ) {
1005                                 $data = substr( $data, 1 );
1006                         }
1007                         $this->checkpointJustWritten = false;
1008                 }
1009                 $this->buffer .= htmlspecialchars( $data );
1010         }
1011
1012         function clearOpenElement( $style ) {
1013                 if ( $this->openElement ) {
1014                         $this->buffer .= Xml::element( $this->openElement[0], $this->openElement[1], $style );
1015                         $this->openElement = false;
1016                 }
1017         }
1018 }
1019
1020 $maintClass = TextPassDumper::class;
1021 require_once RUN_MAINTENANCE_IF_MAIN;