maintenance/includes/TextPassDumper.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 Dump
  25  * @ingroup Maintenance
  26  */
  27
  28 require_once __DIR__ . '/BackupDumper.php';
  29 require_once __DIR__ . '/SevenZipStream.php';
  30 require_once __DIR__ . '/../../includes/export/WikiExporter.php';
  31
  32 use MediaWiki\MediaWikiServices;
  33 use MediaWiki\Shell\Shell;
  34 use MediaWiki\Storage\BlobAccessException;
  35 use MediaWiki\Storage\SqlBlobStore;
  36 use Wikimedia\Rdbms\IMaintainableDatabase;
  37
  38 /**
  39  * @ingroup Maintenance
  40  */
  41 class TextPassDumper extends BackupDumper {
  42         /** @var BaseDump */
  43         public $prefetch = null;
  44         /** @var string|bool */
  45         private $thisPage;
  46         /** @var string|bool */
  47         private $thisRev;
  48
  49         // when we spend more than maxTimeAllowed seconds on this run, we continue
  50         // processing until we write out the next complete page, then save output file(s),
  51         // rename it/them and open new one(s)
  52         public $maxTimeAllowed = 0; // 0 = no limit
  53
  54         protected $input = "php://stdin";
  55         protected $history = WikiExporter::FULL;
  56         protected $fetchCount = 0;
  57         protected $prefetchCount = 0;
  58         protected $prefetchCountLast = 0;
  59         protected $fetchCountLast = 0;
  60
  61         protected $maxFailures = 5;
  62         protected $maxConsecutiveFailedTextRetrievals = 200;
  63         protected $failureTimeout = 5; // Seconds to sleep after db failure
  64
  65         protected $bufferSize = 524288; // In bytes. Maximum size to read from the stub in on go.
  66
  67         protected $php = "php";
  68         protected $spawn = false;
  69
  70         /**
  71          * @var bool|resource
  72          */
  73         protected $spawnProc = false;
  74
  75         /**
  76          * @var bool|resource
  77          */
  78         protected $spawnWrite = false;
  79
  80         /**
  81          * @var bool|resource
  82          */
  83         protected $spawnRead = false;
  84
  85         /**
  86          * @var bool|resource
  87          */
  88         protected $spawnErr = false;
  89
  90         /**
  91          * @var bool|XmlDumpWriter
  92          */
  93         protected $xmlwriterobj = false;
  94
  95         protected $timeExceeded = false;
  96         protected $firstPageWritten = false;
  97         protected $lastPageWritten = false;
  98         protected $checkpointJustWritten = false;
  99         protected $checkpointFiles = [];
 100
 101         /**
 102          * @var IMaintainableDatabase
 103          */
 104         protected $db;
 105
 106         /**
 107          * @param array|null $args For backward compatibility
 108          */
 109         function __construct( $args = null ) {
 110                 parent::__construct();
 111
 112                 $this->addDescription( <<<TEXT
 113 This script postprocesses XML dumps from dumpBackup.php to add
 114 page text which was stubbed out (using --stub).
 115
 116 XML input is accepted on stdin.
 117 XML output is sent to stdout; progress reports are sent to stderr.
 118 TEXT
 119                 );
 120                 $this->stderr = fopen( "php://stderr", "wt" );
 121
 122                 $this->addOption( 'stub', 'To load a compressed stub dump instead of stdin. ' .
 123                         'Specify as --stub=<type>:<file>.', false, true );
 124                 $this->addOption( 'prefetch', 'Use a prior dump file as a text source, to savepressure on the ' .
 125                         'database. (Requires the XMLReader extension). Specify as --prefetch=<type>:<file>',
 126                         false, true );
 127                 $this->addOption( 'maxtime', 'Write out checkpoint file after this many minutes (writing' .
 128                         'out complete page, closing xml file properly, and opening new one' .
 129                         'with header).  This option requires the checkpointfile option.', false, true );
 130                 $this->addOption( 'checkpointfile', 'Use this string for checkpoint filenames,substituting ' .
 131                         'first pageid written for the first %s (required) and the last pageid written for the ' .
 132                         'second %s if it exists.', false, true, false, true ); // This can be specified multiple times
 133                 $this->addOption( 'quiet', 'Don\'t dump status reports to stderr.' );
 134                 $this->addOption( 'full', 'Dump all revisions of every page' );
 135                 $this->addOption( 'current', 'Base ETA on number of pages in database instead of all revisions' );
 136                 $this->addOption( 'spawn', 'Spawn a subprocess for loading text records, optionally specify ' .
 137                         'php[,mwscript] paths' );
 138                 $this->addOption( 'buffersize', 'Buffer size in bytes to use for reading the stub. ' .
 139                         '(Default: 512KB, Minimum: 4KB)', false, true );
 140
 141                 if ( $args ) {
 142                         $this->loadWithArgv( $args );
 143                         $this->processOptions();
 144                 }
 145         }
 146
 147         /**
 148          * @return SqlBlobStore
 149          */
 150         private function getBlobStore() {
 151                 return MediaWikiServices::getInstance()->getBlobStore();
 152         }
 153
 154         function execute() {
 155                 $this->processOptions();
 156                 $this->dump( true );
 157         }
 158
 159         function processOptions() {
 160                 parent::processOptions();
 161
 162                 if ( $this->hasOption( 'buffersize' ) ) {
 163                         $this->bufferSize = max( intval( $this->getOption( 'buffersize' ) ), 4 * 1024 );
 164                 }
 165
 166                 if ( $this->hasOption( 'prefetch' ) ) {
 167                         $url = $this->processFileOpt( $this->getOption( 'prefetch' ) );
 168                         $this->prefetch = new BaseDump( $url );
 169                 }
 170
 171                 if ( $this->hasOption( 'stub' ) ) {
 172                         $this->input = $this->processFileOpt( $this->getOption( 'stub' ) );
 173                 }
 174
 175                 if ( $this->hasOption( 'maxtime' ) ) {
 176                         $this->maxTimeAllowed = intval( $this->getOption( 'maxtime' ) ) * 60;
 177                 }
 178
 179                 if ( $this->hasOption( 'checkpointfile' ) ) {
 180                         $this->checkpointFiles = $this->getOption( 'checkpointfile' );
 181                 }
 182
 183                 if ( $this->hasOption( 'current' ) ) {
 184                         $this->history = WikiExporter::CURRENT;
 185                 }
 186
 187                 if ( $this->hasOption( 'full' ) ) {
 188                         $this->history = WikiExporter::FULL;
 189                 }
 190
 191                 if ( $this->hasOption( 'spawn' ) ) {
 192                         $this->spawn = true;
 193                         $val = $this->getOption( 'spawn' );
 194                         if ( $val !== 1 ) {
 195                                 $this->php = explode( ',', $val, 2 );
 196                         }
 197                 }
 198         }
 199
 200         /**
 201          * Drop the database connection $this->db and try to get a new one.
 202          *
 203          * This function tries to get a /different/ connection if this is
 204          * possible. Hence, (if this is possible) it switches to a different
 205          * failover upon each call.
 206          *
 207          * This function resets $this->lb and closes all connections on it.
 208          *
 209          * @throws MWException
 210          */
 211         function rotateDb() {
 212                 // Cleaning up old connections
 213                 if ( isset( $this->lb ) ) {
 214                         $this->lb->closeAll();
 215                         unset( $this->lb );
 216                 }
 217
 218                 if ( $this->forcedDb !== null ) {
 219                         $this->db = $this->forcedDb;
 220
 221                         return;
 222                 }
 223
 224                 if ( isset( $this->db ) && $this->db->isOpen() ) {
 225                         throw new MWException( 'DB is set and has not been closed by the Load Balancer' );
 226                 }
 227
 228                 unset( $this->db );
 229
 230                 // Trying to set up new connection.
 231                 // We do /not/ retry upon failure, but delegate to encapsulating logic, to avoid
 232                 // individually retrying at different layers of code.
 233
 234                 try {
 235                         $lbFactory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
 236                         $this->lb = $lbFactory->newMainLB();
 237                 } catch ( Exception $e ) {
 238                         throw new MWException( __METHOD__
 239                                 . " rotating DB failed to obtain new load balancer (" . $e->getMessage() . ")" );
 240                 }
 241
 242                 try {
 243                         $this->db = $this->lb->getConnection( DB_REPLICA, 'dump' );
 244                 } catch ( Exception $e ) {
 245                         throw new MWException( __METHOD__
 246                                 . " rotating DB failed to obtain new database (" . $e->getMessage() . ")" );
 247                 }
 248         }
 249
 250         function initProgress( $history = WikiExporter::FULL ) {
 251                 parent::initProgress();
 252                 $this->timeOfCheckpoint = $this->startTime;
 253         }
 254
 255         function dump( $history, $text = WikiExporter::TEXT ) {
 256                 // Notice messages will foul up your XML output even if they're
 257                 // relatively harmless.
 258                 if ( ini_get( 'display_errors' ) ) {
 259                         ini_set( 'display_errors', 'stderr' );
 260                 }
 261
 262                 $this->initProgress( $this->history );
 263
 264                 // We are trying to get an initial database connection to avoid that the
 265                 // first try of this request's first call to getText fails. However, if
 266                 // obtaining a good DB connection fails it's not a serious issue, as
 267                 // getText does retry upon failure and can start without having a working
 268                 // DB connection.
 269                 try {
 270                         $this->rotateDb();
 271                 } catch ( Exception $e ) {
 272                         // We do not even count this as failure. Just let eventual
 273                         // watchdogs know.
 274                         $this->progress( "Getting initial DB connection failed (" .
 275                                 $e->getMessage() . ")" );
 276                 }
 277
 278                 $this->egress = new ExportProgressFilter( $this->sink, $this );
 279
 280                 // it would be nice to do it in the constructor, oh well. need egress set
 281                 $this->finalOptionCheck();
 282
 283                 // we only want this so we know how to close a stream :-P
 284                 $this->xmlwriterobj = new XmlDumpWriter( XmlDumpWriter::WRITE_CONTENT, $this->schemaVersion );
 285
 286                 $input = fopen( $this->input, "rt" );
 287                 $this->readDump( $input );
 288
 289                 if ( $this->spawnProc ) {
 290                         $this->closeSpawn();
 291                 }
 292
 293                 $this->report( true );
 294         }
 295
 296         function processFileOpt( $opt ) {
 297                 $split = explode( ':', $opt, 2 );
 298                 $val = $split[0];
 299                 $param = '';
 300                 if ( count( $split ) === 2 ) {
 301                         $param = $split[1];
 302                 }
 303                 $fileURIs = explode( ';', $param );
 304                 foreach ( $fileURIs as $URI ) {
 305                         switch ( $val ) {
 306                                 case "file":
 307                                         $newURI = $URI;
 308                                         break;
 309                                 case "gzip":
 310                                         $newURI = "compress.zlib://$URI";
 311                                         break;
 312                                 case "bzip2":
 313                                         $newURI = "compress.bzip2://$URI";
 314                                         break;
 315                                 case "7zip":
 316                                         $newURI = "mediawiki.compress.7z://$URI";
 317                                         break;
 318                                 default:
 319                                         $newURI = $URI;
 320                         }
 321                         $newFileURIs[] = $newURI;
 322                 }
 323                 $val = implode( ';', $newFileURIs );
 324
 325                 return $val;
 326         }
 327
 328         /**
 329          * Overridden to include prefetch ratio if enabled.
 330          */
 331         function showReport() {
 332                 if ( !$this->prefetch ) {
 333                         parent::showReport();
 334
 335                         return;
 336                 }
 337
 338                 if ( $this->reporting ) {
 339                         $now = wfTimestamp( TS_DB );
 340                         $nowts = microtime( true );
 341                         $deltaAll = $nowts - $this->startTime;
 342                         $deltaPart = $nowts - $this->lastTime;
 343                         $this->pageCountPart = $this->pageCount - $this->pageCountLast;
 344                         $this->revCountPart = $this->revCount - $this->revCountLast;
 345
 346                         if ( $deltaAll ) {
 347                                 $portion = $this->revCount / $this->maxCount;
 348                                 $eta = $this->startTime + $deltaAll / $portion;
 349                                 $etats = wfTimestamp( TS_DB, intval( $eta ) );
 350                                 if ( $this->fetchCount ) {
 351                                         $fetchRate = 100.0 * $this->prefetchCount / $this->fetchCount;
 352                                 } else {
 353                                         $fetchRate = '-';
 354                                 }
 355                                 $pageRate = $this->pageCount / $deltaAll;
 356                                 $revRate = $this->revCount / $deltaAll;
 357                         } else {
 358                                 $pageRate = '-';
 359                                 $revRate = '-';
 360                                 $etats = '-';
 361                                 $fetchRate = '-';
 362                         }
 363                         if ( $deltaPart ) {
 364                                 if ( $this->fetchCountLast ) {
 365                                         $fetchRatePart = 100.0 * $this->prefetchCountLast / $this->fetchCountLast;
 366                                 } else {
 367                                         $fetchRatePart = '-';
 368                                 }
 369                                 $pageRatePart = $this->pageCountPart / $deltaPart;
 370                                 $revRatePart = $this->revCountPart / $deltaPart;
 371                         } else {
 372                                 $fetchRatePart = '-';
 373                                 $pageRatePart = '-';
 374                                 $revRatePart = '-';
 375                         }
 376                         $this->progress( sprintf(
 377                                 "%s: %s (ID %d) %d pages (%0.1f|%0.1f/sec all|curr), "
 378                                         . "%d revs (%0.1f|%0.1f/sec all|curr), %0.1f%%|%0.1f%% "
 379                                         . "prefetched (all|curr), ETA %s [max %d]",
 380                                 $now, wfWikiID(), $this->ID, $this->pageCount, $pageRate,
 381                                 $pageRatePart, $this->revCount, $revRate, $revRatePart,
 382                                 $fetchRate, $fetchRatePart, $etats, $this->maxCount
 383                         ) );
 384                         $this->lastTime = $nowts;
 385                         $this->revCountLast = $this->revCount;
 386                         $this->prefetchCountLast = $this->prefetchCount;
 387                         $this->fetchCountLast = $this->fetchCount;
 388                 }
 389         }
 390
 391         function setTimeExceeded() {
 392                 $this->timeExceeded = true;
 393         }
 394
 395         function checkIfTimeExceeded() {
 396                 if ( $this->maxTimeAllowed
 397                         && ( $this->lastTime - $this->timeOfCheckpoint > $this->maxTimeAllowed )
 398                 ) {
 399                         return true;
 400                 }
 401
 402                 return false;
 403         }
 404
 405         function finalOptionCheck() {
 406                 if ( ( $this->checkpointFiles && !$this->maxTimeAllowed )
 407                         || ( $this->maxTimeAllowed && !$this->checkpointFiles )
 408                 ) {
 409                         throw new MWException( "Options checkpointfile and maxtime must be specified together.\n" );
 410                 }
 411                 foreach ( $this->checkpointFiles as $checkpointFile ) {
 412                         $count = substr_count( $checkpointFile, "%s" );
 413                         if ( $count != 2 ) {
 414                                 throw new MWException( "Option checkpointfile must contain two '%s' "
 415                                         . "for substitution of first and last pageids, count is $count instead, "
 416                                         . "file is $checkpointFile.\n" );
 417                         }
 418                 }
 419
 420                 if ( $this->checkpointFiles ) {
 421                         $filenameList = (array)$this->egress->getFilenames();
 422                         if ( count( $filenameList ) != count( $this->checkpointFiles ) ) {
 423                                 throw new MWException( "One checkpointfile must be specified "
 424                                         . "for each output option, if maxtime is used.\n" );
 425                         }
 426                 }
 427         }
 428
 429         /**
 430          * @throws MWException Failure to parse XML input
 431          * @param string $input
 432          * @return bool
 433          */
 434         function readDump( $input ) {
 435                 $this->buffer = "";
 436                 $this->openElement = false;
 437                 $this->atStart = true;
 438                 $this->state = "";
 439                 $this->lastName = "";
 440                 $this->thisPage = 0;
 441                 $this->thisRev = 0;
 442                 $this->thisRevModel = null;
 443                 $this->thisRevFormat = null;
 444
 445                 $parser = xml_parser_create( "UTF-8" );
 446                 xml_parser_set_option( $parser, XML_OPTION_CASE_FOLDING, false );
 447
 448                 xml_set_element_handler(
 449                         $parser,
 450                         [ $this, 'startElement' ],
 451                         [ $this, 'endElement' ]
 452                 );
 453                 xml_set_character_data_handler( $parser, [ $this, 'characterData' ] );
 454
 455                 $offset = 0; // for context extraction on error reporting
 456                 do {
 457                         if ( $this->checkIfTimeExceeded() ) {
 458                                 $this->setTimeExceeded();
 459                         }
 460                         $chunk = fread( $input, $this->bufferSize );
 461                         if ( !xml_parse( $parser, $chunk, feof( $input ) ) ) {
 462                                 wfDebug( "TextDumpPass::readDump encountered XML parsing error\n" );
 463
 464                                 $byte = xml_get_current_byte_index( $parser );
 465                                 $msg = wfMessage( 'xml-error-string',
 466                                         'XML import parse failure',
 467                                         xml_get_current_line_number( $parser ),
 468                                         xml_get_current_column_number( $parser ),
 469                                         $byte . ( is_null( $chunk ) ? null : ( '; "' . substr( $chunk, $byte - $offset, 16 ) . '"' ) ),
 470                                         xml_error_string( xml_get_error_code( $parser ) ) )->escaped();
 471
 472                                 xml_parser_free( $parser );
 473
 474                                 throw new MWException( $msg );
 475                         }
 476                         $offset += strlen( $chunk );
 477                 } while ( $chunk !== false && !feof( $input ) );
 478                 if ( $this->maxTimeAllowed ) {
 479                         $filenameList = (array)$this->egress->getFilenames();
 480                         // we wrote some stuff after last checkpoint that needs renamed
 481                         if ( file_exists( $filenameList[0] ) ) {
 482                                 $newFilenames = [];
 483                                 # we might have just written the header and footer and had no
 484                                 # pages or revisions written... perhaps they were all deleted
 485                                 # there's no pageID 0 so we use that. the caller is responsible
 486                                 # for deciding what to do with a file containing only the
 487                                 # siteinfo information and the mw tags.
 488                                 if ( !$this->firstPageWritten ) {
 489                                         $firstPageID = str_pad( 0, 9, "0", STR_PAD_LEFT );
 490                                         $lastPageID = str_pad( 0, 9, "0", STR_PAD_LEFT );
 491                                 } else {
 492                                         $firstPageID = str_pad( $this->firstPageWritten, 9, "0", STR_PAD_LEFT );
 493                                         $lastPageID = str_pad( $this->lastPageWritten, 9, "0", STR_PAD_LEFT );
 494                                 }
 495
 496                                 $filenameCount = count( $filenameList );
 497                                 for ( $i = 0; $i < $filenameCount; $i++ ) {
 498                                         $checkpointNameFilledIn = sprintf( $this->checkpointFiles[$i], $firstPageID, $lastPageID );
 499                                         $fileinfo = pathinfo( $filenameList[$i] );
 500                                         $newFilenames[] = $fileinfo['dirname'] . '/' . $checkpointNameFilledIn;
 501                                 }
 502                                 $this->egress->closeAndRename( $newFilenames );
 503                         }
 504                 }
 505                 xml_parser_free( $parser );
 506
 507                 return true;
 508         }
 509
 510         /**
 511          * Applies applicable export transformations to $text.
 512          *
 513          * @param string $text
 514          * @param string $model
 515          * @param string|null $format
 516          *
 517          * @return string
 518          */
 519         private function exportTransform( $text, $model, $format = null ) {
 520                 try {
 521                         $handler = ContentHandler::getForModelID( $model );
 522                         $text = $handler->exportTransform( $text, $format );
 523                 }
 524                 catch ( MWException $ex ) {
 525                         $this->progress(
 526                                 "Unable to apply export transformation for content model '$model': " .
 527                                 $ex->getMessage()
 528                         );
 529                 }
 530
 531                 return $text;
 532         }
 533
 534         /**
 535          * Tries to load revision text.
 536          * Export transformations are applied if the content model is given or can be
 537          * determined from the database.
 538          *
 539          * Upon errors, retries (Up to $this->maxFailures tries each call).
 540          * If still no good revision could be found even after this retrying, "" is returned.
 541          * If no good revision text could be returned for
 542          * $this->maxConsecutiveFailedTextRetrievals consecutive calls to getText, MWException
 543          * is thrown.
 544          *
 545          * @param int|string $id Content address, or text row ID.
 546          * @param string|bool|null $model The content model used to determine
 547          *  applicable export transformations.
 548          *  If $model is null, it will be determined from the database.
 549          * @param string|null $format The content format used when applying export transformations.
 550          *
 551          * @throws MWException
 552          * @return string The revision text for $id, or ""
 553          */
 554         function getText( $id, $model = null, $format = null ) {
 555                 global $wgContentHandlerUseDB;
 556
 557                 $prefetchNotTried = true; // Whether or not we already tried to get the text via prefetch.
 558                 $text = false; // The candidate for a good text. false if no proper value.
 559                 $failures = 0; // The number of times, this invocation of getText already failed.
 560
 561                 // The number of times getText failed without yielding a good text in between.
 562                 static $consecutiveFailedTextRetrievals = 0;
 563
 564                 $this->fetchCount++;
 565
 566                 // To allow to simply return on success and do not have to worry about book keeping,
 567                 // we assume, this fetch works (possible after some retries). Nevertheless, we koop
 568                 // the old value, so we can restore it, if problems occur (See after the while loop).
 569                 $oldConsecutiveFailedTextRetrievals = $consecutiveFailedTextRetrievals;
 570                 $consecutiveFailedTextRetrievals = 0;
 571
 572                 if ( $model === null && $wgContentHandlerUseDB ) {
 573                         // TODO: MCR: use content table
 574                         $row = $this->db->selectRow(
 575                                 'revision',
 576                                 [ 'rev_content_model', 'rev_content_format' ],
 577                                 [ 'rev_id' => $this->thisRev ],
 578                                 __METHOD__
 579                         );
 580
 581                         if ( $row ) {
 582                                 $model = $row->rev_content_model;
 583                                 $format = $row->rev_content_format;
 584                         }
 585                 }
 586
 587                 if ( $model === null || $model === '' ) {
 588                         $model = false;
 589                 }
 590
 591                 while ( $failures < $this->maxFailures ) {
 592                         // As soon as we found a good text for the $id, we will return immediately.
 593                         // Hence, if we make it past the try catch block, we know that we did not
 594                         // find a good text.
 595
 596                         try {
 597                                 // Step 1: Get some text (or reuse from previous iteratuon if checking
 598                                 //         for plausibility failed)
 599
 600                                 // Trying to get prefetch, if it has not been tried before
 601                                 if ( $text === false && isset( $this->prefetch ) && $prefetchNotTried ) {
 602                                         $prefetchNotTried = false;
 603                                         $tryIsPrefetch = true;
 604                                         $text = $this->prefetch->prefetch( (int)$this->thisPage, (int)$this->thisRev );
 605
 606                                         if ( $text === null ) {
 607                                                 $text = false;
 608                                         }
 609
 610                                         if ( is_string( $text ) && $model !== false ) {
 611                                                 // Apply export transformation to text coming from an old dump.
 612                                                 // The purpose of this transformation is to convert up from legacy
 613                                                 // formats, which may still be used in the older dump that is used
 614                                                 // for pre-fetching. Applying the transformation again should not
 615                                                 // interfere with content that is already in the correct form.
 616                                                 $text = $this->exportTransform( $text, $model, $format );
 617                                         }
 618                                 }
 619
 620                                 if ( $text === false ) {
 621                                         // Fallback to asking the database
 622                                         $tryIsPrefetch = false;
 623                                         if ( $this->spawn ) {
 624                                                 $text = $this->getTextSpawned( $id );
 625                                         } else {
 626                                                 $text = $this->getTextDb( $id );
 627                                         }
 628
 629                                         if ( $text !== false && $model !== false ) {
 630                                                 // Apply export transformation to text coming from the database.
 631                                                 // Prefetched text should already have transformations applied.
 632                                                 $text = $this->exportTransform( $text, $model, $format );
 633                                         }
 634
 635                                         // No more checks for texts from DB for now.
 636                                         // If we received something that is not false,
 637                                         // We treat it as good text, regardless of whether it actually is or is not
 638                                         if ( $text !== false ) {
 639                                                 return $text;
 640                                         }
 641                                 }
 642
 643                                 if ( $text === false ) {
 644                                         throw new MWException( "Generic error while obtaining text for id " . $id );
 645                                 }
 646
 647                                 // We received a good candidate for the text of $id via some method
 648
 649                                 // Step 2: Checking for plausibility and return the text if it is
 650                                 //         plausible
 651                                 $revID = intval( $this->thisRev );
 652                                 if ( !isset( $this->db ) ) {
 653                                         throw new MWException( "No database available" );
 654                                 }
 655
 656                                 if ( $model !== CONTENT_MODEL_WIKITEXT ) {
 657                                         $revLength = strlen( $text );
 658                                 } else {
 659                                         $revLength = $this->db->selectField( 'revision', 'rev_len', [ 'rev_id' => $revID ] );
 660                                 }
 661
 662                                 if ( strlen( $text ) == $revLength ) {
 663                                         if ( $tryIsPrefetch ) {
 664                                                 $this->prefetchCount++;
 665                                         }
 666
 667                                         return $text;
 668                                 }
 669
 670                                 $text = false;
 671                                 throw new MWException( "Received text is unplausible for id " . $id );
 672                         } catch ( Exception $e ) {
 673                                 $msg = "getting/checking text " . $id . " failed (" . $e->getMessage() . ")";
 674                                 if ( $failures + 1 < $this->maxFailures ) {
 675                                         $msg .= " (Will retry " . ( $this->maxFailures - $failures - 1 ) . " more times)";
 676                                 }
 677                                 $this->progress( $msg );
 678                         }
 679
 680                         // Something went wrong; we did not a text that was plausible :(
 681                         $failures++;
 682
 683                         // A failure in a prefetch hit does not warrant resetting db connection etc.
 684                         if ( !$tryIsPrefetch ) {
 685                                 // After backing off for some time, we try to reboot the whole process as
 686                                 // much as possible to not carry over failures from one part to the other
 687                                 // parts
 688                                 sleep( $this->failureTimeout );
 689                                 try {
 690                                         $this->rotateDb();
 691                                         if ( $this->spawn ) {
 692                                                 $this->closeSpawn();
 693                                                 $this->openSpawn();
 694                                         }
 695                                 } catch ( Exception $e ) {
 696                                         $this->progress( "Rebooting getText infrastructure failed (" . $e->getMessage() . ")" .
 697                                                 " Trying to continue anyways" );
 698                                 }
 699                         }
 700                 }
 701
 702                 // Retirieving a good text for $id failed (at least) maxFailures times.
 703                 // We abort for this $id.
 704
 705                 // Restoring the consecutive failures, and maybe aborting, if the dump
 706                 // is too broken.
 707                 $consecutiveFailedTextRetrievals = $oldConsecutiveFailedTextRetrievals + 1;
 708                 if ( $consecutiveFailedTextRetrievals > $this->maxConsecutiveFailedTextRetrievals ) {
 709                         throw new MWException( "Graceful storage failure" );
 710                 }
 711
 712                 return "";
 713         }
 714
 715         /**
 716          * Loads the serialized content from storage.
 717          *
 718          * @param int|string $id Content address, or text row ID.
 719          * @return bool|string
 720          */
 721         private function getTextDb( $id ) {
 722                 $store = $this->getBlobStore();
 723                 $address = ( is_int( $id ) || strpos( $id, ':' ) === false )
 724                         ? SqlBlobStore::makeAddressFromTextId( (int)$id )
 725                         : $id;
 726
 727                 try {
 728                         $text = $store->getBlob( $address );
 729
 730                         $stripped = str_replace( "\r", "", $text );
 731                         $normalized = MediaWikiServices::getInstance()->getContentLanguage()
 732                                 ->normalize( $stripped );
 733
 734                         return $normalized;
 735                 } catch ( BlobAccessException $ex ) {
 736                         // XXX: log a warning?
 737                         return false;
 738                 }
 739         }
 740
 741         /**
 742          * @param int|string $address Content address, or text row ID.
 743          * @return bool|string
 744          */
 745         private function getTextSpawned( $address ) {
 746                 Wikimedia\suppressWarnings();
 747                 if ( !$this->spawnProc ) {
 748                         // First time?
 749                         $this->openSpawn();
 750                 }
 751                 $text = $this->getTextSpawnedOnce( $address );
 752                 Wikimedia\restoreWarnings();
 753
 754                 return $text;
 755         }
 756
 757         function openSpawn() {
 758                 global $IP;
 759
 760                 if ( count( $this->php ) == 2 ) {
 761                         $mwscriptpath = $this->php[1];
 762                 } else {
 763                         $mwscriptpath = "$IP/../multiversion/MWScript.php";
 764                 }
 765                 if ( file_exists( $mwscriptpath ) ) {
 766                         $cmd = implode( " ",
 767                                 array_map( [ Shell::class, 'escape' ],
 768                                         [
 769                                                 $this->php[0],
 770                                                 $mwscriptpath,
 771                                                 "fetchText.php",
 772                                                 '--wiki', wfWikiID() ] ) );
 773                 } else {
 774                         $cmd = implode( " ",
 775                                 array_map( [ Shell::class, 'escape' ],
 776                                         [
 777                                                 $this->php[0],
 778                                                 "$IP/maintenance/fetchText.php",
 779                                                 '--wiki', wfWikiID() ] ) );
 780                 }
 781                 $spec = [
 782                         0 => [ "pipe", "r" ],
 783                         1 => [ "pipe", "w" ],
 784                         2 => [ "file", "/dev/null", "a" ] ];
 785                 $pipes = [];
 786
 787                 $this->progress( "Spawning database subprocess: $cmd" );
 788                 $this->spawnProc = proc_open( $cmd, $spec, $pipes );
 789                 if ( !$this->spawnProc ) {
 790                         $this->progress( "Subprocess spawn failed." );
 791
 792                         return false;
 793                 }
 794                 list(
 795                         $this->spawnWrite, // -> stdin
 796                         $this->spawnRead, // <- stdout
 797                 ) = $pipes;
 798
 799                 return true;
 800         }
 801
 802         private function closeSpawn() {
 803                 Wikimedia\suppressWarnings();
 804                 if ( $this->spawnRead ) {
 805                         fclose( $this->spawnRead );
 806                 }
 807                 $this->spawnRead = false;
 808                 if ( $this->spawnWrite ) {
 809                         fclose( $this->spawnWrite );
 810                 }
 811                 $this->spawnWrite = false;
 812                 if ( $this->spawnErr ) {
 813                         fclose( $this->spawnErr );
 814                 }
 815                 $this->spawnErr = false;
 816                 if ( $this->spawnProc ) {
 817                         pclose( $this->spawnProc );
 818                 }
 819                 $this->spawnProc = false;
 820                 Wikimedia\restoreWarnings();
 821         }
 822
 823         /**
 824          * @param int|string $address Content address, or text row ID.
 825          * @return bool|string
 826          */
 827         private function getTextSpawnedOnce( $address ) {
 828                 if ( is_int( $address ) || intval( $address ) ) {
 829                         $address = SqlBlobStore::makeAddressFromTextId( (int)$address );
 830                 }
 831
 832                 $ok = fwrite( $this->spawnWrite, "$address\n" );
 833                 // $this->progress( ">> $id" );
 834                 if ( !$ok ) {
 835                         return false;
 836                 }
 837
 838                 $ok = fflush( $this->spawnWrite );
 839                 // $this->progress( ">> [flush]" );
 840                 if ( !$ok ) {
 841                         return false;
 842                 }
 843
 844                 // check that the text address they are sending is the one we asked for
 845                 // this avoids out of sync revision text errors we have encountered in the past
 846                 $newAddress = fgets( $this->spawnRead );
 847                 if ( $newAddress === false ) {
 848                         return false;
 849                 }
 850                 $newAddress = trim( $newAddress );
 851                 if ( strpos( $newAddress, ':' ) === false ) {
 852                         $newAddress = SqlBlobStore::makeAddressFromTextId( intval( $newAddress ) );
 853                 }
 854
 855                 if ( $newAddress !== $address ) {
 856                         return false;
 857                 }
 858
 859                 $len = fgets( $this->spawnRead );
 860                 // $this->progress( "<< " . trim( $len ) );
 861                 if ( $len === false ) {
 862                         return false;
 863                 }
 864
 865                 $nbytes = intval( $len );
 866                 // actual error, not zero-length text
 867                 if ( $nbytes < 0 ) {
 868                         return false;
 869                 }
 870
 871                 $text = "";
 872
 873                 // Subprocess may not send everything at once, we have to loop.
 874                 while ( $nbytes > strlen( $text ) ) {
 875                         $buffer = fread( $this->spawnRead, $nbytes - strlen( $text ) );
 876                         if ( $buffer === false ) {
 877                                 break;
 878                         }
 879                         $text .= $buffer;
 880                 }
 881
 882                 $gotbytes = strlen( $text );
 883                 if ( $gotbytes != $nbytes ) {
 884                         $this->progress( "Expected $nbytes bytes from database subprocess, got $gotbytes " );
 885
 886                         return false;
 887                 }
 888
 889                 // Do normalization in the dump thread...
 890                 $stripped = str_replace( "\r", "", $text );
 891                 $normalized = MediaWikiServices::getInstance()->getContentLanguage()->
 892                         normalize( $stripped );
 893
 894                 return $normalized;
 895         }
 896
 897         function startElement( $parser, $name, $attribs ) {
 898                 $this->checkpointJustWritten = false;
 899
 900                 $this->clearOpenElement( null );
 901                 $this->lastName = $name;
 902
 903                 if ( $name == 'revision' ) {
 904                         $this->state = $name;
 905                         $this->egress->writeOpenPage( null, $this->buffer );
 906                         $this->buffer = "";
 907                 } elseif ( $name == 'page' ) {
 908                         $this->state = $name;
 909                         if ( $this->atStart ) {
 910                                 $this->egress->writeOpenStream( $this->buffer );
 911                                 $this->buffer = "";
 912                                 $this->atStart = false;
 913                         }
 914                 }
 915
 916                 if ( $name == "text" && isset( $attribs['id'] ) ) {
 917                         $id = $attribs['id'];
 918                         $model = trim( $this->thisRevModel );
 919                         $format = trim( $this->thisRevFormat );
 920
 921                         $model = $model === '' ? null : $model;
 922                         $format = $format === '' ? null : $format;
 923
 924                         $text = $this->getText( $id, $model, $format );
 925                         $this->openElement = [ $name, [ 'xml:space' => 'preserve' ] ];
 926                         if ( strlen( $text ) > 0 ) {
 927                                 $this->characterData( $parser, $text );
 928                         }
 929                 } else {
 930                         $this->openElement = [ $name, $attribs ];
 931                 }
 932         }
 933
 934         function endElement( $parser, $name ) {
 935                 $this->checkpointJustWritten = false;
 936
 937                 if ( $this->openElement ) {
 938                         $this->clearOpenElement( "" );
 939                 } else {
 940                         $this->buffer .= "</$name>";
 941                 }
 942
 943                 if ( $name == 'revision' ) {
 944                         $this->egress->writeRevision( null, $this->buffer );
 945                         $this->buffer = "";
 946                         $this->thisRev = "";
 947                         $this->thisRevModel = null;
 948                         $this->thisRevFormat = null;
 949                 } elseif ( $name == 'page' ) {
 950                         if ( !$this->firstPageWritten ) {
 951                                 $this->firstPageWritten = trim( $this->thisPage );
 952                         }
 953                         $this->lastPageWritten = trim( $this->thisPage );
 954                         if ( $this->timeExceeded ) {
 955                                 $this->egress->writeClosePage( $this->buffer );
 956                                 // nasty hack, we can't just write the chardata after the
 957                                 // page tag, it will include leading blanks from the next line
 958                                 $this->egress->sink->write( "\n" );
 959
 960                                 $this->buffer = $this->xmlwriterobj->closeStream();
 961                                 $this->egress->writeCloseStream( $this->buffer );
 962
 963                                 $this->buffer = "";
 964                                 $this->thisPage = "";
 965                                 // this could be more than one file if we had more than one output arg
 966
 967                                 $filenameList = (array)$this->egress->getFilenames();
 968                                 $newFilenames = [];
 969                                 $firstPageID = str_pad( $this->firstPageWritten, 9, "0", STR_PAD_LEFT );
 970                                 $lastPageID = str_pad( $this->lastPageWritten, 9, "0", STR_PAD_LEFT );
 971                                 $filenamesCount = count( $filenameList );
 972                                 for ( $i = 0; $i < $filenamesCount; $i++ ) {
 973                                         $checkpointNameFilledIn = sprintf( $this->checkpointFiles[$i], $firstPageID, $lastPageID );
 974                                         $fileinfo = pathinfo( $filenameList[$i] );
 975                                         $newFilenames[] = $fileinfo['dirname'] . '/' . $checkpointNameFilledIn;
 976                                 }
 977                                 $this->egress->closeRenameAndReopen( $newFilenames );
 978                                 $this->buffer = $this->xmlwriterobj->openStream();
 979                                 $this->timeExceeded = false;
 980                                 $this->timeOfCheckpoint = $this->lastTime;
 981                                 $this->firstPageWritten = false;
 982                                 $this->checkpointJustWritten = true;
 983                         } else {
 984                                 $this->egress->writeClosePage( $this->buffer );
 985                                 $this->buffer = "";
 986                                 $this->thisPage = "";
 987                         }
 988                 } elseif ( $name == 'mediawiki' ) {
 989                         $this->egress->writeCloseStream( $this->buffer );
 990                         $this->buffer = "";
 991                 }
 992         }
 993
 994         function characterData( $parser, $data ) {
 995                 $this->clearOpenElement( null );
 996                 if ( $this->lastName == "id" ) {
 997                         if ( $this->state == "revision" ) {
 998                                 $this->thisRev .= $data;
 999                         } elseif ( $this->state == "page" ) {
1000                                 $this->thisPage .= $data;
1001                         }
1002                 } elseif ( $this->lastName == "model" ) {
1003                         $this->thisRevModel .= $data;
1004                 } elseif ( $this->lastName == "format" ) {
1005                         $this->thisRevFormat .= $data;
1006                 }
1007
1008                 // have to skip the newline left over from closepagetag line of
1009                 // end of checkpoint files. nasty hack!!
1010                 if ( $this->checkpointJustWritten ) {
1011                         if ( $data[0] == "\n" ) {
1012                                 $data = substr( $data, 1 );
1013                         }
1014                         $this->checkpointJustWritten = false;
1015                 }
1016                 $this->buffer .= htmlspecialchars( $data );
1017         }
1018
1019         function clearOpenElement( $style ) {
1020                 if ( $this->openElement ) {
1021                         $this->buffer .= Xml::element( $this->openElement[0], $this->openElement[1], $style );
1022                         $this->openElement = false;
1023                 }
1024         }
1025 }