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          * @suppress PhanTypeObjectUnsetDeclaredProperty
 211          */
 212         function rotateDb() {
 213                 // Cleaning up old connections
 214                 if ( isset( $this->lb ) ) {
 215                         $this->lb->closeAll();
 216                         unset( $this->lb );
 217                 }
 218
 219                 if ( $this->forcedDb !== null ) {
 220                         $this->db = $this->forcedDb;
 221
 222                         return;
 223                 }
 224
 225                 if ( isset( $this->db ) && $this->db->isOpen() ) {
 226                         throw new MWException( 'DB is set and has not been closed by the Load Balancer' );
 227                 }
 228
 229                 unset( $this->db );
 230
 231                 // Trying to set up new connection.
 232                 // We do /not/ retry upon failure, but delegate to encapsulating logic, to avoid
 233                 // individually retrying at different layers of code.
 234
 235                 try {
 236                         $lbFactory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
 237                         $this->lb = $lbFactory->newMainLB();
 238                 } catch ( Exception $e ) {
 239                         throw new MWException( __METHOD__
 240                                 . " rotating DB failed to obtain new load balancer (" . $e->getMessage() . ")" );
 241                 }
 242
 243                 try {
 244                         $this->db = $this->lb->getMaintenanceConnectionRef( DB_REPLICA, 'dump' );
 245                 } catch ( Exception $e ) {
 246                         throw new MWException( __METHOD__
 247                                 . " rotating DB failed to obtain new database (" . $e->getMessage() . ")" );
 248                 }
 249         }
 250
 251         function initProgress( $history = WikiExporter::FULL ) {
 252                 parent::initProgress();
 253                 $this->timeOfCheckpoint = $this->startTime;
 254         }
 255
 256         function dump( $history, $text = WikiExporter::TEXT ) {
 257                 // Notice messages will foul up your XML output even if they're
 258                 // relatively harmless.
 259                 if ( ini_get( 'display_errors' ) ) {
 260                         ini_set( 'display_errors', 'stderr' );
 261                 }
 262
 263                 $this->initProgress( $this->history );
 264
 265                 // We are trying to get an initial database connection to avoid that the
 266                 // first try of this request's first call to getText fails. However, if
 267                 // obtaining a good DB connection fails it's not a serious issue, as
 268                 // getText does retry upon failure and can start without having a working
 269                 // DB connection.
 270                 try {
 271                         $this->rotateDb();
 272                 } catch ( Exception $e ) {
 273                         // We do not even count this as failure. Just let eventual
 274                         // watchdogs know.
 275                         $this->progress( "Getting initial DB connection failed (" .
 276                                 $e->getMessage() . ")" );
 277                 }
 278
 279                 $this->egress = new ExportProgressFilter( $this->sink, $this );
 280
 281                 // it would be nice to do it in the constructor, oh well. need egress set
 282                 $this->finalOptionCheck();
 283
 284                 // we only want this so we know how to close a stream :-P
 285                 $this->xmlwriterobj = new XmlDumpWriter( XmlDumpWriter::WRITE_CONTENT, $this->schemaVersion );
 286
 287                 $input = fopen( $this->input, "rt" );
 288                 $this->readDump( $input );
 289
 290                 if ( $this->spawnProc ) {
 291                         $this->closeSpawn();
 292                 }
 293
 294                 $this->report( true );
 295         }
 296
 297         function processFileOpt( $opt ) {
 298                 $split = explode( ':', $opt, 2 );
 299                 $val = $split[0];
 300                 $param = '';
 301                 if ( count( $split ) === 2 ) {
 302                         $param = $split[1];
 303                 }
 304                 $fileURIs = explode( ';', $param );
 305                 foreach ( $fileURIs as $URI ) {
 306                         switch ( $val ) {
 307                                 case "file":
 308                                         $newURI = $URI;
 309                                         break;
 310                                 case "gzip":
 311                                         $newURI = "compress.zlib://$URI";
 312                                         break;
 313                                 case "bzip2":
 314                                         $newURI = "compress.bzip2://$URI";
 315                                         break;
 316                                 case "7zip":
 317                                         $newURI = "mediawiki.compress.7z://$URI";
 318                                         break;
 319                                 default:
 320                                         $newURI = $URI;
 321                         }
 322                         $newFileURIs[] = $newURI;
 323                 }
 324                 $val = implode( ';', $newFileURIs );
 325
 326                 return $val;
 327         }
 328
 329         /**
 330          * Overridden to include prefetch ratio if enabled.
 331          */
 332         function showReport() {
 333                 if ( !$this->prefetch ) {
 334                         parent::showReport();
 335
 336                         return;
 337                 }
 338
 339                 if ( $this->reporting ) {
 340                         $now = wfTimestamp( TS_DB );
 341                         $nowts = microtime( true );
 342                         $deltaAll = $nowts - $this->startTime;
 343                         $deltaPart = $nowts - $this->lastTime;
 344                         $this->pageCountPart = $this->pageCount - $this->pageCountLast;
 345                         $this->revCountPart = $this->revCount - $this->revCountLast;
 346
 347                         if ( $deltaAll ) {
 348                                 $portion = $this->revCount / $this->maxCount;
 349                                 $eta = $this->startTime + $deltaAll / $portion;
 350                                 $etats = wfTimestamp( TS_DB, intval( $eta ) );
 351                                 if ( $this->fetchCount ) {
 352                                         $fetchRate = 100.0 * $this->prefetchCount / $this->fetchCount;
 353                                 } else {
 354                                         $fetchRate = '-';
 355                                 }
 356                                 $pageRate = $this->pageCount / $deltaAll;
 357                                 $revRate = $this->revCount / $deltaAll;
 358                         } else {
 359                                 $pageRate = '-';
 360                                 $revRate = '-';
 361                                 $etats = '-';
 362                                 $fetchRate = '-';
 363                         }
 364                         if ( $deltaPart ) {
 365                                 if ( $this->fetchCountLast ) {
 366                                         $fetchRatePart = 100.0 * $this->prefetchCountLast / $this->fetchCountLast;
 367                                 } else {
 368                                         $fetchRatePart = '-';
 369                                 }
 370                                 $pageRatePart = $this->pageCountPart / $deltaPart;
 371                                 $revRatePart = $this->revCountPart / $deltaPart;
 372                         } else {
 373                                 $fetchRatePart = '-';
 374                                 $pageRatePart = '-';
 375                                 $revRatePart = '-';
 376                         }
 377
 378                         $dbDomain = WikiMap::getCurrentWikiDbDomain()->getId();
 379                         $this->progress( sprintf(
 380                                 "%s: %s (ID %d) %d pages (%0.1f|%0.1f/sec all|curr), "
 381                                         . "%d revs (%0.1f|%0.1f/sec all|curr), %0.1f%%|%0.1f%% "
 382                                         . "prefetched (all|curr), ETA %s [max %d]",
 383                                 $now, $dbDomain, $this->ID, $this->pageCount, $pageRate,
 384                                 $pageRatePart, $this->revCount, $revRate, $revRatePart,
 385                                 $fetchRate, $fetchRatePart, $etats, $this->maxCount
 386                         ) );
 387                         $this->lastTime = $nowts;
 388                         $this->revCountLast = $this->revCount;
 389                         $this->prefetchCountLast = $this->prefetchCount;
 390                         $this->fetchCountLast = $this->fetchCount;
 391                 }
 392         }
 393
 394         function setTimeExceeded() {
 395                 $this->timeExceeded = true;
 396         }
 397
 398         function checkIfTimeExceeded() {
 399                 if ( $this->maxTimeAllowed
 400                         && ( $this->lastTime - $this->timeOfCheckpoint > $this->maxTimeAllowed )
 401                 ) {
 402                         return true;
 403                 }
 404
 405                 return false;
 406         }
 407
 408         function finalOptionCheck() {
 409                 if ( ( $this->checkpointFiles && !$this->maxTimeAllowed )
 410                         || ( $this->maxTimeAllowed && !$this->checkpointFiles )
 411                 ) {
 412                         throw new MWException( "Options checkpointfile and maxtime must be specified together.\n" );
 413                 }
 414                 foreach ( $this->checkpointFiles as $checkpointFile ) {
 415                         $count = substr_count( $checkpointFile, "%s" );
 416                         if ( $count != 2 ) {
 417                                 throw new MWException( "Option checkpointfile must contain two '%s' "
 418                                         . "for substitution of first and last pageids, count is $count instead, "
 419                                         . "file is $checkpointFile.\n" );
 420                         }
 421                 }
 422
 423                 if ( $this->checkpointFiles ) {
 424                         $filenameList = (array)$this->egress->getFilenames();
 425                         if ( count( $filenameList ) != count( $this->checkpointFiles ) ) {
 426                                 throw new MWException( "One checkpointfile must be specified "
 427                                         . "for each output option, if maxtime is used.\n" );
 428                         }
 429                 }
 430         }
 431
 432         /**
 433          * @throws MWException Failure to parse XML input
 434          * @param string $input
 435          * @return bool
 436          */
 437         function readDump( $input ) {
 438                 $this->buffer = "";
 439                 $this->openElement = false;
 440                 $this->atStart = true;
 441                 $this->state = "";
 442                 $this->lastName = "";
 443                 $this->thisPage = 0;
 444                 $this->thisRev = 0;
 445                 $this->thisRevModel = null;
 446                 $this->thisRevFormat = null;
 447
 448                 $parser = xml_parser_create( "UTF-8" );
 449                 xml_parser_set_option( $parser, XML_OPTION_CASE_FOLDING, false );
 450
 451                 xml_set_element_handler(
 452                         $parser,
 453                         [ $this, 'startElement' ],
 454                         [ $this, 'endElement' ]
 455                 );
 456                 xml_set_character_data_handler( $parser, [ $this, 'characterData' ] );
 457
 458                 $offset = 0; // for context extraction on error reporting
 459                 do {
 460                         if ( $this->checkIfTimeExceeded() ) {
 461                                 $this->setTimeExceeded();
 462                         }
 463                         $chunk = fread( $input, $this->bufferSize );
 464                         if ( !xml_parse( $parser, $chunk, feof( $input ) ) ) {
 465                                 wfDebug( "TextDumpPass::readDump encountered XML parsing error\n" );
 466
 467                                 $byte = xml_get_current_byte_index( $parser );
 468                                 $msg = wfMessage( 'xml-error-string',
 469                                         'XML import parse failure',
 470                                         xml_get_current_line_number( $parser ),
 471                                         xml_get_current_column_number( $parser ),
 472                                         $byte . ( is_null( $chunk ) ? null : ( '; "' . substr( $chunk, $byte - $offset, 16 ) . '"' ) ),
 473                                         xml_error_string( xml_get_error_code( $parser ) ) )->escaped();
 474
 475                                 xml_parser_free( $parser );
 476
 477                                 throw new MWException( $msg );
 478                         }
 479                         $offset += strlen( $chunk );
 480                 } while ( $chunk !== false && !feof( $input ) );
 481                 if ( $this->maxTimeAllowed ) {
 482                         $filenameList = (array)$this->egress->getFilenames();
 483                         // we wrote some stuff after last checkpoint that needs renamed
 484                         if ( file_exists( $filenameList[0] ) ) {
 485                                 $newFilenames = [];
 486                                 # we might have just written the header and footer and had no
 487                                 # pages or revisions written... perhaps they were all deleted
 488                                 # there's no pageID 0 so we use that. the caller is responsible
 489                                 # for deciding what to do with a file containing only the
 490                                 # siteinfo information and the mw tags.
 491                                 if ( !$this->firstPageWritten ) {
 492                                         $firstPageID = str_pad( 0, 9, "0", STR_PAD_LEFT );
 493                                         $lastPageID = str_pad( 0, 9, "0", STR_PAD_LEFT );
 494                                 } else {
 495                                         $firstPageID = str_pad( $this->firstPageWritten, 9, "0", STR_PAD_LEFT );
 496                                         $lastPageID = str_pad( $this->lastPageWritten, 9, "0", STR_PAD_LEFT );
 497                                 }
 498
 499                                 $filenameCount = count( $filenameList );
 500                                 for ( $i = 0; $i < $filenameCount; $i++ ) {
 501                                         $checkpointNameFilledIn = sprintf( $this->checkpointFiles[$i], $firstPageID, $lastPageID );
 502                                         $fileinfo = pathinfo( $filenameList[$i] );
 503                                         $newFilenames[] = $fileinfo['dirname'] . '/' . $checkpointNameFilledIn;
 504                                 }
 505                                 $this->egress->closeAndRename( $newFilenames );
 506                         }
 507                 }
 508                 xml_parser_free( $parser );
 509
 510                 return true;
 511         }
 512
 513         /**
 514          * Applies applicable export transformations to $text.
 515          *
 516          * @param string $text
 517          * @param string $model
 518          * @param string|null $format
 519          *
 520          * @return string
 521          */
 522         private function exportTransform( $text, $model, $format = null ) {
 523                 try {
 524                         $handler = ContentHandler::getForModelID( $model );
 525                         $text = $handler->exportTransform( $text, $format );
 526                 }
 527                 catch ( MWException $ex ) {
 528                         $this->progress(
 529                                 "Unable to apply export transformation for content model '$model': " .
 530                                 $ex->getMessage()
 531                         );
 532                 }
 533
 534                 return $text;
 535         }
 536
 537         /**
 538          * Tries to load revision text.
 539          * Export transformations are applied if the content model is given or can be
 540          * determined from the database.
 541          *
 542          * Upon errors, retries (Up to $this->maxFailures tries each call).
 543          * If still no good revision could be found even after this retrying, "" is returned.
 544          * If no good revision text could be returned for
 545          * $this->maxConsecutiveFailedTextRetrievals consecutive calls to getText, MWException
 546          * is thrown.
 547          *
 548          * @param int|string $id Content address, or text row ID.
 549          * @param string|bool|null $model The content model used to determine
 550          *  applicable export transformations.
 551          *  If $model is null, it will be determined from the database.
 552          * @param string|null $format The content format used when applying export transformations.
 553          *
 554          * @throws MWException
 555          * @return string The revision text for $id, or ""
 556          */
 557         function getText( $id, $model = null, $format = null ) {
 558                 global $wgContentHandlerUseDB;
 559
 560                 $prefetchNotTried = true; // Whether or not we already tried to get the text via prefetch.
 561                 $text = false; // The candidate for a good text. false if no proper value.
 562                 $failures = 0; // The number of times, this invocation of getText already failed.
 563
 564                 // The number of times getText failed without yielding a good text in between.
 565                 static $consecutiveFailedTextRetrievals = 0;
 566
 567                 $this->fetchCount++;
 568
 569                 // To allow to simply return on success and do not have to worry about book keeping,
 570                 // we assume, this fetch works (possible after some retries). Nevertheless, we koop
 571                 // the old value, so we can restore it, if problems occur (See after the while loop).
 572                 $oldConsecutiveFailedTextRetrievals = $consecutiveFailedTextRetrievals;
 573                 $consecutiveFailedTextRetrievals = 0;
 574
 575                 if ( $model === null && $wgContentHandlerUseDB ) {
 576                         // TODO: MCR: use content table
 577                         $row = $this->db->selectRow(
 578                                 'revision',
 579                                 [ 'rev_content_model', 'rev_content_format' ],
 580                                 [ 'rev_id' => $this->thisRev ],
 581                                 __METHOD__
 582                         );
 583
 584                         if ( $row ) {
 585                                 $model = $row->rev_content_model;
 586                                 $format = $row->rev_content_format;
 587                         }
 588                 }
 589
 590                 if ( $model === null || $model === '' ) {
 591                         $model = false;
 592                 }
 593
 594                 while ( $failures < $this->maxFailures ) {
 595                         // As soon as we found a good text for the $id, we will return immediately.
 596                         // Hence, if we make it past the try catch block, we know that we did not
 597                         // find a good text.
 598
 599                         try {
 600                                 // Step 1: Get some text (or reuse from previous iteratuon if checking
 601                                 //         for plausibility failed)
 602
 603                                 // Trying to get prefetch, if it has not been tried before
 604                                 if ( $text === false && isset( $this->prefetch ) && $prefetchNotTried ) {
 605                                         $prefetchNotTried = false;
 606                                         $tryIsPrefetch = true;
 607                                         $text = $this->prefetch->prefetch( (int)$this->thisPage, (int)$this->thisRev );
 608
 609                                         if ( $text === null ) {
 610                                                 $text = false;
 611                                         }
 612
 613                                         if ( is_string( $text ) && $model !== false ) {
 614                                                 // Apply export transformation to text coming from an old dump.
 615                                                 // The purpose of this transformation is to convert up from legacy
 616                                                 // formats, which may still be used in the older dump that is used
 617                                                 // for pre-fetching. Applying the transformation again should not
 618                                                 // interfere with content that is already in the correct form.
 619                                                 $text = $this->exportTransform( $text, $model, $format );
 620                                         }
 621                                 }
 622
 623                                 if ( $text === false ) {
 624                                         // Fallback to asking the database
 625                                         $tryIsPrefetch = false;
 626                                         if ( $this->spawn ) {
 627                                                 $text = $this->getTextSpawned( $id );
 628                                         } else {
 629                                                 $text = $this->getTextDb( $id );
 630                                         }
 631
 632                                         if ( $text !== false && $model !== false ) {
 633                                                 // Apply export transformation to text coming from the database.
 634                                                 // Prefetched text should already have transformations applied.
 635                                                 $text = $this->exportTransform( $text, $model, $format );
 636                                         }
 637
 638                                         // No more checks for texts from DB for now.
 639                                         // If we received something that is not false,
 640                                         // We treat it as good text, regardless of whether it actually is or is not
 641                                         if ( $text !== false ) {
 642                                                 return $text;
 643                                         }
 644                                 }
 645
 646                                 if ( $text === false ) {
 647                                         throw new MWException( "Generic error while obtaining text for id " . $id );
 648                                 }
 649
 650                                 // We received a good candidate for the text of $id via some method
 651
 652                                 // Step 2: Checking for plausibility and return the text if it is
 653                                 //         plausible
 654                                 $revID = intval( $this->thisRev );
 655                                 if ( !isset( $this->db ) ) {
 656                                         throw new MWException( "No database available" );
 657                                 }
 658
 659                                 if ( $model !== CONTENT_MODEL_WIKITEXT ) {
 660                                         $revLength = strlen( $text );
 661                                 } else {
 662                                         $revLength = $this->db->selectField( 'revision', 'rev_len', [ 'rev_id' => $revID ] );
 663                                 }
 664
 665                                 if ( strlen( $text ) == $revLength ) {
 666                                         if ( $tryIsPrefetch ) {
 667                                                 $this->prefetchCount++;
 668                                         }
 669
 670                                         return $text;
 671                                 }
 672
 673                                 $text = false;
 674                                 throw new MWException( "Received text is unplausible for id " . $id );
 675                         } catch ( Exception $e ) {
 676                                 $msg = "getting/checking text " . $id . " failed (" . $e->getMessage() . ")";
 677                                 if ( $failures + 1 < $this->maxFailures ) {
 678                                         $msg .= " (Will retry " . ( $this->maxFailures - $failures - 1 ) . " more times)";
 679                                 }
 680                                 $this->progress( $msg );
 681                         }
 682
 683                         // Something went wrong; we did not a text that was plausible :(
 684                         $failures++;
 685
 686                         // A failure in a prefetch hit does not warrant resetting db connection etc.
 687                         if ( !$tryIsPrefetch ) {
 688                                 // After backing off for some time, we try to reboot the whole process as
 689                                 // much as possible to not carry over failures from one part to the other
 690                                 // parts
 691                                 sleep( $this->failureTimeout );
 692                                 try {
 693                                         $this->rotateDb();
 694                                         if ( $this->spawn ) {
 695                                                 $this->closeSpawn();
 696                                                 $this->openSpawn();
 697                                         }
 698                                 } catch ( Exception $e ) {
 699                                         $this->progress( "Rebooting getText infrastructure failed (" . $e->getMessage() . ")" .
 700                                                 " Trying to continue anyways" );
 701                                 }
 702                         }
 703                 }
 704
 705                 // Retirieving a good text for $id failed (at least) maxFailures times.
 706                 // We abort for this $id.
 707
 708                 // Restoring the consecutive failures, and maybe aborting, if the dump
 709                 // is too broken.
 710                 $consecutiveFailedTextRetrievals = $oldConsecutiveFailedTextRetrievals + 1;
 711                 if ( $consecutiveFailedTextRetrievals > $this->maxConsecutiveFailedTextRetrievals ) {
 712                         throw new MWException( "Graceful storage failure" );
 713                 }
 714
 715                 return "";
 716         }
 717
 718         /**
 719          * Loads the serialized content from storage.
 720          *
 721          * @param int|string $id Content address, or text row ID.
 722          * @return bool|string
 723          */
 724         private function getTextDb( $id ) {
 725                 $store = $this->getBlobStore();
 726                 $address = ( is_int( $id ) || strpos( $id, ':' ) === false )
 727                         ? SqlBlobStore::makeAddressFromTextId( (int)$id )
 728                         : $id;
 729
 730                 try {
 731                         $text = $store->getBlob( $address );
 732
 733                         $stripped = str_replace( "\r", "", $text );
 734                         $normalized = MediaWikiServices::getInstance()->getContentLanguage()
 735                                 ->normalize( $stripped );
 736
 737                         return $normalized;
 738                 } catch ( BlobAccessException $ex ) {
 739                         // XXX: log a warning?
 740                         return false;
 741                 }
 742         }
 743
 744         /**
 745          * @param int|string $address Content address, or text row ID.
 746          * @return bool|string
 747          */
 748         private function getTextSpawned( $address ) {
 749                 Wikimedia\suppressWarnings();
 750                 if ( !$this->spawnProc ) {
 751                         // First time?
 752                         $this->openSpawn();
 753                 }
 754                 $text = $this->getTextSpawnedOnce( $address );
 755                 Wikimedia\restoreWarnings();
 756
 757                 return $text;
 758         }
 759
 760         function openSpawn() {
 761                 global $IP;
 762
 763                 $wiki = WikiMap::getWikiIdFromDbDomain( WikiMap::getCurrentWikiDbDomain() );
 764                 if ( count( $this->php ) == 2 ) {
 765                         $mwscriptpath = $this->php[1];
 766                 } else {
 767                         $mwscriptpath = "$IP/../multiversion/MWScript.php";
 768                 }
 769                 if ( file_exists( $mwscriptpath ) ) {
 770                         $cmd = implode( " ",
 771                                 array_map( [ Shell::class, 'escape' ],
 772                                         [
 773                                                 $this->php[0],
 774                                                 $mwscriptpath,
 775                                                 "fetchText.php",
 776                                                 '--wiki', $wiki ] ) );
 777                 } else {
 778                         $cmd = implode( " ",
 779                                 array_map( [ Shell::class, 'escape' ],
 780                                         [
 781                                                 $this->php[0],
 782                                                 "$IP/maintenance/fetchText.php",
 783                                                 '--wiki', $wiki ] ) );
 784                 }
 785                 $spec = [
 786                         0 => [ "pipe", "r" ],
 787                         1 => [ "pipe", "w" ],
 788                         2 => [ "file", "/dev/null", "a" ] ];
 789                 $pipes = [];
 790
 791                 $this->progress( "Spawning database subprocess: $cmd" );
 792                 $this->spawnProc = proc_open( $cmd, $spec, $pipes );
 793                 if ( !$this->spawnProc ) {
 794                         $this->progress( "Subprocess spawn failed." );
 795
 796                         return false;
 797                 }
 798                 list(
 799                         $this->spawnWrite, // -> stdin
 800                         $this->spawnRead, // <- stdout
 801                 ) = $pipes;
 802
 803                 return true;
 804         }
 805
 806         private function closeSpawn() {
 807                 Wikimedia\suppressWarnings();
 808                 if ( $this->spawnRead ) {
 809                         fclose( $this->spawnRead );
 810                 }
 811                 $this->spawnRead = false;
 812                 if ( $this->spawnWrite ) {
 813                         fclose( $this->spawnWrite );
 814                 }
 815                 $this->spawnWrite = false;
 816                 if ( $this->spawnErr ) {
 817                         fclose( $this->spawnErr );
 818                 }
 819                 $this->spawnErr = false;
 820                 if ( $this->spawnProc ) {
 821                         pclose( $this->spawnProc );
 822                 }
 823                 $this->spawnProc = false;
 824                 Wikimedia\restoreWarnings();
 825         }
 826
 827         /**
 828          * @param int|string $address Content address, or text row ID.
 829          * @return bool|string
 830          */
 831         private function getTextSpawnedOnce( $address ) {
 832                 if ( is_int( $address ) || intval( $address ) ) {
 833                         $address = SqlBlobStore::makeAddressFromTextId( (int)$address );
 834                 }
 835
 836                 $ok = fwrite( $this->spawnWrite, "$address\n" );
 837                 // $this->progress( ">> $id" );
 838                 if ( !$ok ) {
 839                         return false;
 840                 }
 841
 842                 $ok = fflush( $this->spawnWrite );
 843                 // $this->progress( ">> [flush]" );
 844                 if ( !$ok ) {
 845                         return false;
 846                 }
 847
 848                 // check that the text address they are sending is the one we asked for
 849                 // this avoids out of sync revision text errors we have encountered in the past
 850                 $newAddress = fgets( $this->spawnRead );
 851                 if ( $newAddress === false ) {
 852                         return false;
 853                 }
 854                 $newAddress = trim( $newAddress );
 855                 if ( strpos( $newAddress, ':' ) === false ) {
 856                         $newAddress = SqlBlobStore::makeAddressFromTextId( intval( $newAddress ) );
 857                 }
 858
 859                 if ( $newAddress !== $address ) {
 860                         return false;
 861                 }
 862
 863                 $len = fgets( $this->spawnRead );
 864                 // $this->progress( "<< " . trim( $len ) );
 865                 if ( $len === false ) {
 866                         return false;
 867                 }
 868
 869                 $nbytes = intval( $len );
 870                 // actual error, not zero-length text
 871                 if ( $nbytes < 0 ) {
 872                         return false;
 873                 }
 874
 875                 $text = "";
 876
 877                 // Subprocess may not send everything at once, we have to loop.
 878                 while ( $nbytes > strlen( $text ) ) {
 879                         $buffer = fread( $this->spawnRead, $nbytes - strlen( $text ) );
 880                         if ( $buffer === false ) {
 881                                 break;
 882                         }
 883                         $text .= $buffer;
 884                 }
 885
 886                 $gotbytes = strlen( $text );
 887                 if ( $gotbytes != $nbytes ) {
 888                         $this->progress( "Expected $nbytes bytes from database subprocess, got $gotbytes " );
 889
 890                         return false;
 891                 }
 892
 893                 // Do normalization in the dump thread...
 894                 $stripped = str_replace( "\r", "", $text );
 895                 $normalized = MediaWikiServices::getInstance()->getContentLanguage()->
 896                         normalize( $stripped );
 897
 898                 return $normalized;
 899         }
 900
 901         function startElement( $parser, $name, $attribs ) {
 902                 $this->checkpointJustWritten = false;
 903
 904                 $this->clearOpenElement( null );
 905                 $this->lastName = $name;
 906
 907                 if ( $name == 'revision' ) {
 908                         $this->state = $name;
 909                         $this->egress->writeOpenPage( null, $this->buffer );
 910                         $this->buffer = "";
 911                 } elseif ( $name == 'page' ) {
 912                         $this->state = $name;
 913                         if ( $this->atStart ) {
 914                                 $this->egress->writeOpenStream( $this->buffer );
 915                                 $this->buffer = "";
 916                                 $this->atStart = false;
 917                         }
 918                 }
 919
 920                 if ( $name == "text" && isset( $attribs['id'] ) ) {
 921                         $id = $attribs['id'];
 922                         $model = trim( $this->thisRevModel );
 923                         $format = trim( $this->thisRevFormat );
 924
 925                         $model = $model === '' ? null : $model;
 926                         $format = $format === '' ? null : $format;
 927
 928                         $text = $this->getText( $id, $model, $format );
 929                         $this->openElement = [ $name, [ 'xml:space' => 'preserve' ] ];
 930                         if ( strlen( $text ) > 0 ) {
 931                                 $this->characterData( $parser, $text );
 932                         }
 933                 } else {
 934                         $this->openElement = [ $name, $attribs ];
 935                 }
 936         }
 937
 938         function endElement( $parser, $name ) {
 939                 $this->checkpointJustWritten = false;
 940
 941                 if ( $this->openElement ) {
 942                         $this->clearOpenElement( "" );
 943                 } else {
 944                         $this->buffer .= "</$name>";
 945                 }
 946
 947                 if ( $name == 'revision' ) {
 948                         $this->egress->writeRevision( null, $this->buffer );
 949                         $this->buffer = "";
 950                         $this->thisRev = "";
 951                         $this->thisRevModel = null;
 952                         $this->thisRevFormat = null;
 953                 } elseif ( $name == 'page' ) {
 954                         if ( !$this->firstPageWritten ) {
 955                                 $this->firstPageWritten = trim( $this->thisPage );
 956                         }
 957                         $this->lastPageWritten = trim( $this->thisPage );
 958                         if ( $this->timeExceeded ) {
 959                                 $this->egress->writeClosePage( $this->buffer );
 960                                 // nasty hack, we can't just write the chardata after the
 961                                 // page tag, it will include leading blanks from the next line
 962                                 $this->egress->sink->write( "\n" );
 963
 964                                 $this->buffer = $this->xmlwriterobj->closeStream();
 965                                 $this->egress->writeCloseStream( $this->buffer );
 966
 967                                 $this->buffer = "";
 968                                 $this->thisPage = "";
 969                                 // this could be more than one file if we had more than one output arg
 970
 971                                 $filenameList = (array)$this->egress->getFilenames();
 972                                 $newFilenames = [];
 973                                 $firstPageID = str_pad( $this->firstPageWritten, 9, "0", STR_PAD_LEFT );
 974                                 $lastPageID = str_pad( $this->lastPageWritten, 9, "0", STR_PAD_LEFT );
 975                                 $filenamesCount = count( $filenameList );
 976                                 for ( $i = 0; $i < $filenamesCount; $i++ ) {
 977                                         $checkpointNameFilledIn = sprintf( $this->checkpointFiles[$i], $firstPageID, $lastPageID );
 978                                         $fileinfo = pathinfo( $filenameList[$i] );
 979                                         $newFilenames[] = $fileinfo['dirname'] . '/' . $checkpointNameFilledIn;
 980                                 }
 981                                 $this->egress->closeRenameAndReopen( $newFilenames );
 982                                 $this->buffer = $this->xmlwriterobj->openStream();
 983                                 $this->timeExceeded = false;
 984                                 $this->timeOfCheckpoint = $this->lastTime;
 985                                 $this->firstPageWritten = false;
 986                                 $this->checkpointJustWritten = true;
 987                         } else {
 988                                 $this->egress->writeClosePage( $this->buffer );
 989                                 $this->buffer = "";
 990                                 $this->thisPage = "";
 991                         }
 992                 } elseif ( $name == 'mediawiki' ) {
 993                         $this->egress->writeCloseStream( $this->buffer );
 994                         $this->buffer = "";
 995                 }
 996         }
 997
 998         function characterData( $parser, $data ) {
 999                 $this->clearOpenElement( null );
1000                 if ( $this->lastName == "id" ) {
1001                         if ( $this->state == "revision" ) {
1002                                 $this->thisRev .= $data;
1003                         } elseif ( $this->state == "page" ) {
1004                                 $this->thisPage .= $data;
1005                         }
1006                 } elseif ( $this->lastName == "model" ) {
1007                         $this->thisRevModel .= $data;
1008                 } elseif ( $this->lastName == "format" ) {
1009                         $this->thisRevFormat .= $data;
1010                 }
1011
1012                 // have to skip the newline left over from closepagetag line of
1013                 // end of checkpoint files. nasty hack!!
1014                 if ( $this->checkpointJustWritten ) {
1015                         if ( $data[0] == "\n" ) {
1016                                 $data = substr( $data, 1 );
1017                         }
1018                         $this->checkpointJustWritten = false;
1019                 }
1020                 $this->buffer .= htmlspecialchars( $data );
1021         }
1022
1023         function clearOpenElement( $style ) {
1024                 if ( $this->openElement ) {
1025                         $this->buffer .= Xml::element( $this->openElement[0], $this->openElement[1], $style );
1026                         $this->openElement = false;
1027                 }
1028         }
1029 }