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         /** @var array */
  68         protected $php = [];
  69         protected $spawn = false;
  70
  71         /**
  72          * @var bool|resource
  73          */
  74         protected $spawnProc = false;
  75
  76         /**
  77          * @var resource
  78          */
  79         protected $spawnWrite;
  80
  81         /**
  82          * @var resource
  83          */
  84         protected $spawnRead;
  85
  86         /**
  87          * @var bool|resource
  88          */
  89         protected $spawnErr = false;
  90
  91         /**
  92          * @var bool|XmlDumpWriter
  93          */
  94         protected $xmlwriterobj = false;
  95
  96         protected $timeExceeded = false;
  97         protected $firstPageWritten = false;
  98         protected $lastPageWritten = false;
  99         protected $checkpointJustWritten = false;
 100         protected $checkpointFiles = [];
 101
 102         /**
 103          * @var IMaintainableDatabase
 104          */
 105         protected $db;
 106
 107         /**
 108          * @param array|null $args For backward compatibility
 109          */
 110         function __construct( $args = null ) {
 111                 parent::__construct();
 112
 113                 $this->addDescription( <<<TEXT
 114 This script postprocesses XML dumps from dumpBackup.php to add
 115 page text which was stubbed out (using --stub).
 116
 117 XML input is accepted on stdin.
 118 XML output is sent to stdout; progress reports are sent to stderr.
 119 TEXT
 120                 );
 121                 $this->stderr = fopen( "php://stderr", "wt" );
 122
 123                 $this->addOption( 'stub', 'To load a compressed stub dump instead of stdin. ' .
 124                         'Specify as --stub=<type>:<file>.', false, true );
 125                 $this->addOption( 'prefetch', 'Use a prior dump file as a text source, to savepressure on the ' .
 126                         'database. (Requires the XMLReader extension). Specify as --prefetch=<type>:<file>',
 127                         false, true );
 128                 $this->addOption( 'maxtime', 'Write out checkpoint file after this many minutes (writing' .
 129                         'out complete page, closing xml file properly, and opening new one' .
 130                         'with header).  This option requires the checkpointfile option.', false, true );
 131                 $this->addOption( 'checkpointfile', 'Use this string for checkpoint filenames,substituting ' .
 132                         'first pageid written for the first %s (required) and the last pageid written for the ' .
 133                         'second %s if it exists.', false, true, false, true ); // This can be specified multiple times
 134                 $this->addOption( 'quiet', 'Don\'t dump status reports to stderr.' );
 135                 $this->addOption( 'full', 'Dump all revisions of every page' );
 136                 $this->addOption( 'current', 'Base ETA on number of pages in database instead of all revisions' );
 137                 $this->addOption( 'spawn', 'Spawn a subprocess for loading text records, optionally specify ' .
 138                         'php[,mwscript] paths' );
 139                 $this->addOption( 'buffersize', 'Buffer size in bytes to use for reading the stub. ' .
 140                         '(Default: 512KB, Minimum: 4KB)', false, true );
 141
 142                 if ( $args ) {
 143                         $this->loadWithArgv( $args );
 144                         $this->processOptions();
 145                 }
 146         }
 147
 148         /**
 149          * @return SqlBlobStore
 150          */
 151         private function getBlobStore() {
 152                 return MediaWikiServices::getInstance()->getBlobStore();
 153         }
 154
 155         function execute() {
 156                 $this->processOptions();
 157                 $this->dump( true );
 158         }
 159
 160         function processOptions() {
 161                 parent::processOptions();
 162
 163                 if ( $this->hasOption( 'buffersize' ) ) {
 164                         $this->bufferSize = max( intval( $this->getOption( 'buffersize' ) ), 4 * 1024 );
 165                 }
 166
 167                 if ( $this->hasOption( 'prefetch' ) ) {
 168                         $url = $this->processFileOpt( $this->getOption( 'prefetch' ) );
 169                         $this->prefetch = new BaseDump( $url );
 170                 }
 171
 172                 if ( $this->hasOption( 'stub' ) ) {
 173                         $this->input = $this->processFileOpt( $this->getOption( 'stub' ) );
 174                 }
 175
 176                 if ( $this->hasOption( 'maxtime' ) ) {
 177                         $this->maxTimeAllowed = intval( $this->getOption( 'maxtime' ) ) * 60;
 178                 }
 179
 180                 if ( $this->hasOption( 'checkpointfile' ) ) {
 181                         $this->checkpointFiles = $this->getOption( 'checkpointfile' );
 182                 }
 183
 184                 if ( $this->hasOption( 'current' ) ) {
 185                         $this->history = WikiExporter::CURRENT;
 186                 }
 187
 188                 if ( $this->hasOption( 'full' ) ) {
 189                         $this->history = WikiExporter::FULL;
 190                 }
 191
 192                 if ( $this->hasOption( 'spawn' ) ) {
 193                         $this->spawn = true;
 194                         $val = $this->getOption( 'spawn' );
 195                         if ( $val !== 1 ) {
 196                                 $this->php = explode( ',', $val, 2 );
 197                         }
 198                 }
 199         }
 200
 201         /**
 202          * Drop the database connection $this->db and try to get a new one.
 203          *
 204          * This function tries to get a /different/ connection if this is
 205          * possible. Hence, (if this is possible) it switches to a different
 206          * failover upon each call.
 207          *
 208          * This function resets $this->lb and closes all connections on it.
 209          *
 210          * @throws MWException
 211          * @suppress PhanTypeObjectUnsetDeclaredProperty
 212          */
 213         function rotateDb() {
 214                 // Cleaning up old connections
 215                 if ( isset( $this->lb ) ) {
 216                         $this->lb->closeAll();
 217                         unset( $this->lb );
 218                 }
 219
 220                 if ( $this->forcedDb !== null ) {
 221                         $this->db = $this->forcedDb;
 222
 223                         return;
 224                 }
 225
 226                 if ( isset( $this->db ) && $this->db->isOpen() ) {
 227                         throw new MWException( 'DB is set and has not been closed by the Load Balancer' );
 228                 }
 229
 230                 unset( $this->db );
 231
 232                 // Trying to set up new connection.
 233                 // We do /not/ retry upon failure, but delegate to encapsulating logic, to avoid
 234                 // individually retrying at different layers of code.
 235
 236                 try {
 237                         $lbFactory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
 238                         $this->lb = $lbFactory->newMainLB();
 239                 } catch ( Exception $e ) {
 240                         throw new MWException( __METHOD__
 241                                 . " rotating DB failed to obtain new load balancer (" . $e->getMessage() . ")" );
 242                 }
 243
 244                 try {
 245                         $this->db = $this->lb->getMaintenanceConnectionRef( DB_REPLICA, 'dump' );
 246                 } catch ( Exception $e ) {
 247                         throw new MWException( __METHOD__
 248                                 . " rotating DB failed to obtain new database (" . $e->getMessage() . ")" );
 249                 }
 250         }
 251
 252         function initProgress( $history = WikiExporter::FULL ) {
 253                 parent::initProgress();
 254                 $this->timeOfCheckpoint = $this->startTime;
 255         }
 256
 257         function dump( $history, $text = WikiExporter::TEXT ) {
 258                 // Notice messages will foul up your XML output even if they're
 259                 // relatively harmless.
 260                 if ( ini_get( 'display_errors' ) ) {
 261                         ini_set( 'display_errors', 'stderr' );
 262                 }
 263
 264                 $this->initProgress( $this->history );
 265
 266                 // We are trying to get an initial database connection to avoid that the
 267                 // first try of this request's first call to getText fails. However, if
 268                 // obtaining a good DB connection fails it's not a serious issue, as
 269                 // getText does retry upon failure and can start without having a working
 270                 // DB connection.
 271                 try {
 272                         $this->rotateDb();
 273                 } catch ( Exception $e ) {
 274                         // We do not even count this as failure. Just let eventual
 275                         // watchdogs know.
 276                         $this->progress( "Getting initial DB connection failed (" .
 277                                 $e->getMessage() . ")" );
 278                 }
 279
 280                 $this->egress = new ExportProgressFilter( $this->sink, $this );
 281
 282                 // it would be nice to do it in the constructor, oh well. need egress set
 283                 $this->finalOptionCheck();
 284
 285                 // we only want this so we know how to close a stream :-P
 286                 $this->xmlwriterobj = new XmlDumpWriter( XmlDumpWriter::WRITE_CONTENT, $this->schemaVersion );
 287
 288                 $input = fopen( $this->input, "rt" );
 289                 $this->readDump( $input );
 290
 291                 if ( $this->spawnProc ) {
 292                         $this->closeSpawn();
 293                 }
 294
 295                 $this->report( true );
 296         }
 297
 298         function processFileOpt( $opt ) {
 299                 $split = explode( ':', $opt, 2 );
 300                 $val = $split[0];
 301                 $param = '';
 302                 if ( count( $split ) === 2 ) {
 303                         $param = $split[1];
 304                 }
 305                 $fileURIs = explode( ';', $param );
 306                 foreach ( $fileURIs as $URI ) {
 307                         switch ( $val ) {
 308                                 case "file":
 309                                         $newURI = $URI;
 310                                         break;
 311                                 case "gzip":
 312                                         $newURI = "compress.zlib://$URI";
 313                                         break;
 314                                 case "bzip2":
 315                                         $newURI = "compress.bzip2://$URI";
 316                                         break;
 317                                 case "7zip":
 318                                         $newURI = "mediawiki.compress.7z://$URI";
 319                                         break;
 320                                 default:
 321                                         $newURI = $URI;
 322                         }
 323                         $newFileURIs[] = $newURI;
 324                 }
 325                 $val = implode( ';', $newFileURIs );
 326
 327                 return $val;
 328         }
 329
 330         /**
 331          * Overridden to include prefetch ratio if enabled.
 332          */
 333         function showReport() {
 334                 if ( !$this->prefetch ) {
 335                         parent::showReport();
 336
 337                         return;
 338                 }
 339
 340                 if ( $this->reporting ) {
 341                         $now = wfTimestamp( TS_DB );
 342                         $nowts = microtime( true );
 343                         $deltaAll = $nowts - $this->startTime;
 344                         $deltaPart = $nowts - $this->lastTime;
 345                         $this->pageCountPart = $this->pageCount - $this->pageCountLast;
 346                         $this->revCountPart = $this->revCount - $this->revCountLast;
 347
 348                         if ( $deltaAll ) {
 349                                 $portion = $this->revCount / $this->maxCount;
 350                                 $eta = $this->startTime + $deltaAll / $portion;
 351                                 $etats = wfTimestamp( TS_DB, intval( $eta ) );
 352                                 if ( $this->fetchCount ) {
 353                                         $fetchRate = 100.0 * $this->prefetchCount / $this->fetchCount;
 354                                 } else {
 355                                         $fetchRate = '-';
 356                                 }
 357                                 $pageRate = $this->pageCount / $deltaAll;
 358                                 $revRate = $this->revCount / $deltaAll;
 359                         } else {
 360                                 $pageRate = '-';
 361                                 $revRate = '-';
 362                                 $etats = '-';
 363                                 $fetchRate = '-';
 364                         }
 365                         if ( $deltaPart ) {
 366                                 if ( $this->fetchCountLast ) {
 367                                         $fetchRatePart = 100.0 * $this->prefetchCountLast / $this->fetchCountLast;
 368                                 } else {
 369                                         $fetchRatePart = '-';
 370                                 }
 371                                 $pageRatePart = $this->pageCountPart / $deltaPart;
 372                                 $revRatePart = $this->revCountPart / $deltaPart;
 373                         } else {
 374                                 $fetchRatePart = '-';
 375                                 $pageRatePart = '-';
 376                                 $revRatePart = '-';
 377                         }
 378
 379                         $dbDomain = WikiMap::getCurrentWikiDbDomain()->getId();
 380                         $this->progress( sprintf(
 381                                 "%s: %s (ID %d) %d pages (%0.1f|%0.1f/sec all|curr), "
 382                                         . "%d revs (%0.1f|%0.1f/sec all|curr), %0.1f%%|%0.1f%% "
 383                                         . "prefetched (all|curr), ETA %s [max %d]",
 384                                 $now, $dbDomain, $this->ID, $this->pageCount, $pageRate,
 385                                 $pageRatePart, $this->revCount, $revRate, $revRatePart,
 386                                 $fetchRate, $fetchRatePart, $etats, $this->maxCount
 387                         ) );
 388                         $this->lastTime = $nowts;
 389                         $this->revCountLast = $this->revCount;
 390                         $this->prefetchCountLast = $this->prefetchCount;
 391                         $this->fetchCountLast = $this->fetchCount;
 392                 }
 393         }
 394
 395         function setTimeExceeded() {
 396                 $this->timeExceeded = true;
 397         }
 398
 399         function checkIfTimeExceeded() {
 400                 if ( $this->maxTimeAllowed
 401                         && ( $this->lastTime - $this->timeOfCheckpoint > $this->maxTimeAllowed )
 402                 ) {
 403                         return true;
 404                 }
 405
 406                 return false;
 407         }
 408
 409         function finalOptionCheck() {
 410                 if ( ( $this->checkpointFiles && !$this->maxTimeAllowed )
 411                         || ( $this->maxTimeAllowed && !$this->checkpointFiles )
 412                 ) {
 413                         throw new MWException( "Options checkpointfile and maxtime must be specified together.\n" );
 414                 }
 415                 foreach ( $this->checkpointFiles as $checkpointFile ) {
 416                         $count = substr_count( $checkpointFile, "%s" );
 417                         if ( $count != 2 ) {
 418                                 throw new MWException( "Option checkpointfile must contain two '%s' "
 419                                         . "for substitution of first and last pageids, count is $count instead, "
 420                                         . "file is $checkpointFile.\n" );
 421                         }
 422                 }
 423
 424                 if ( $this->checkpointFiles ) {
 425                         $filenameList = (array)$this->egress->getFilenames();
 426                         if ( count( $filenameList ) != count( $this->checkpointFiles ) ) {
 427                                 throw new MWException( "One checkpointfile must be specified "
 428                                         . "for each output option, if maxtime is used.\n" );
 429                         }
 430                 }
 431         }
 432
 433         /**
 434          * @throws MWException Failure to parse XML input
 435          * @param resource $input
 436          * @return bool
 437          */
 438         function readDump( $input ) {
 439                 $this->buffer = "";
 440                 $this->openElement = false;
 441                 $this->atStart = true;
 442                 $this->state = "";
 443                 $this->lastName = "";
 444                 $this->thisPage = 0;
 445                 $this->thisRev = 0;
 446                 $this->thisRevModel = null;
 447                 $this->thisRevFormat = null;
 448
 449                 $parser = xml_parser_create( "UTF-8" );
 450                 xml_parser_set_option( $parser, XML_OPTION_CASE_FOLDING, false );
 451
 452                 xml_set_element_handler(
 453                         $parser,
 454                         [ $this, 'startElement' ],
 455                         [ $this, 'endElement' ]
 456                 );
 457                 xml_set_character_data_handler( $parser, [ $this, 'characterData' ] );
 458
 459                 $offset = 0; // for context extraction on error reporting
 460                 do {
 461                         if ( $this->checkIfTimeExceeded() ) {
 462                                 $this->setTimeExceeded();
 463                         }
 464                         $chunk = fread( $input, $this->bufferSize );
 465                         if ( !xml_parse( $parser, $chunk, feof( $input ) ) ) {
 466                                 wfDebug( "TextDumpPass::readDump encountered XML parsing error\n" );
 467
 468                                 $byte = xml_get_current_byte_index( $parser );
 469                                 $msg = wfMessage( 'xml-error-string',
 470                                         'XML import parse failure',
 471                                         xml_get_current_line_number( $parser ),
 472                                         xml_get_current_column_number( $parser ),
 473                                         $byte . ( is_null( $chunk ) ? null : ( '; "' . substr( $chunk, $byte - $offset, 16 ) . '"' ) ),
 474                                         xml_error_string( xml_get_error_code( $parser ) ) )->escaped();
 475
 476                                 xml_parser_free( $parser );
 477
 478                                 throw new MWException( $msg );
 479                         }
 480                         $offset += strlen( $chunk );
 481                 } while ( $chunk !== false && !feof( $input ) );
 482                 if ( $this->maxTimeAllowed ) {
 483                         $filenameList = (array)$this->egress->getFilenames();
 484                         // we wrote some stuff after last checkpoint that needs renamed
 485                         if ( file_exists( $filenameList[0] ) ) {
 486                                 $newFilenames = [];
 487                                 # we might have just written the header and footer and had no
 488                                 # pages or revisions written... perhaps they were all deleted
 489                                 # there's no pageID 0 so we use that. the caller is responsible
 490                                 # for deciding what to do with a file containing only the
 491                                 # siteinfo information and the mw tags.
 492                                 if ( !$this->firstPageWritten ) {
 493                                         $firstPageID = str_pad( 0, 9, "0", STR_PAD_LEFT );
 494                                         $lastPageID = str_pad( 0, 9, "0", STR_PAD_LEFT );
 495                                 } else {
 496                                         $firstPageID = str_pad( $this->firstPageWritten, 9, "0", STR_PAD_LEFT );
 497                                         $lastPageID = str_pad( $this->lastPageWritten, 9, "0", STR_PAD_LEFT );
 498                                 }
 499
 500                                 $filenameCount = count( $filenameList );
 501                                 for ( $i = 0; $i < $filenameCount; $i++ ) {
 502                                         $checkpointNameFilledIn = sprintf( $this->checkpointFiles[$i], $firstPageID, $lastPageID );
 503                                         $fileinfo = pathinfo( $filenameList[$i] );
 504                                         $newFilenames[] = $fileinfo['dirname'] . '/' . $checkpointNameFilledIn;
 505                                 }
 506                                 $this->egress->closeAndRename( $newFilenames );
 507                         }
 508                 }
 509                 xml_parser_free( $parser );
 510
 511                 return true;
 512         }
 513
 514         /**
 515          * Applies applicable export transformations to $text.
 516          *
 517          * @param string $text
 518          * @param string $model
 519          * @param string|null $format
 520          *
 521          * @return string
 522          */
 523         private function exportTransform( $text, $model, $format = null ) {
 524                 try {
 525                         $handler = ContentHandler::getForModelID( $model );
 526                         $text = $handler->exportTransform( $text, $format );
 527                 }
 528                 catch ( MWException $ex ) {
 529                         $this->progress(
 530                                 "Unable to apply export transformation for content model '$model': " .
 531                                 $ex->getMessage()
 532                         );
 533                 }
 534
 535                 return $text;
 536         }
 537
 538         /**
 539          * Tries to load revision text.
 540          * Export transformations are applied if the content model is given or can be
 541          * determined from the database.
 542          *
 543          * Upon errors, retries (Up to $this->maxFailures tries each call).
 544          * If still no good revision could be found even after this retrying, "" is returned.
 545          * If no good revision text could be returned for
 546          * $this->maxConsecutiveFailedTextRetrievals consecutive calls to getText, MWException
 547          * is thrown.
 548          *
 549          * @param int|string $id Content address, or text row ID.
 550          * @param string|bool|null $model The content model used to determine
 551          *  applicable export transformations.
 552          *  If $model is null, it will be determined from the database.
 553          * @param string|null $format The content format used when applying export transformations.
 554          *
 555          * @throws MWException
 556          * @return string The revision text for $id, or ""
 557          */
 558         function getText( $id, $model = null, $format = null ) {
 559                 global $wgContentHandlerUseDB;
 560
 561                 $prefetchNotTried = true; // Whether or not we already tried to get the text via prefetch.
 562                 $text = false; // The candidate for a good text. false if no proper value.
 563                 $failures = 0; // The number of times, this invocation of getText already failed.
 564
 565                 // The number of times getText failed without yielding a good text in between.
 566                 static $consecutiveFailedTextRetrievals = 0;
 567
 568                 $this->fetchCount++;
 569
 570                 // To allow to simply return on success and do not have to worry about book keeping,
 571                 // we assume, this fetch works (possible after some retries). Nevertheless, we koop
 572                 // the old value, so we can restore it, if problems occur (See after the while loop).
 573                 $oldConsecutiveFailedTextRetrievals = $consecutiveFailedTextRetrievals;
 574                 $consecutiveFailedTextRetrievals = 0;
 575
 576                 if ( $model === null && $wgContentHandlerUseDB ) {
 577                         // TODO: MCR: use content table
 578                         $row = $this->db->selectRow(
 579                                 'revision',
 580                                 [ 'rev_content_model', 'rev_content_format' ],
 581                                 [ 'rev_id' => $this->thisRev ],
 582                                 __METHOD__
 583                         );
 584
 585                         if ( $row ) {
 586                                 $model = $row->rev_content_model;
 587                                 $format = $row->rev_content_format;
 588                         }
 589                 }
 590
 591                 if ( $model === null || $model === '' ) {
 592                         $model = false;
 593                 }
 594
 595                 while ( $failures < $this->maxFailures ) {
 596                         // As soon as we found a good text for the $id, we will return immediately.
 597                         // Hence, if we make it past the try catch block, we know that we did not
 598                         // find a good text.
 599
 600                         try {
 601                                 // Step 1: Get some text (or reuse from previous iteratuon if checking
 602                                 //         for plausibility failed)
 603
 604                                 // Trying to get prefetch, if it has not been tried before
 605                                 if ( $text === false && isset( $this->prefetch ) && $prefetchNotTried ) {
 606                                         $prefetchNotTried = false;
 607                                         $tryIsPrefetch = true;
 608                                         $text = $this->prefetch->prefetch( (int)$this->thisPage, (int)$this->thisRev );
 609
 610                                         if ( $text === null ) {
 611                                                 $text = false;
 612                                         }
 613
 614                                         if ( is_string( $text ) && $model !== false ) {
 615                                                 // Apply export transformation to text coming from an old dump.
 616                                                 // The purpose of this transformation is to convert up from legacy
 617                                                 // formats, which may still be used in the older dump that is used
 618                                                 // for pre-fetching. Applying the transformation again should not
 619                                                 // interfere with content that is already in the correct form.
 620                                                 $text = $this->exportTransform( $text, $model, $format );
 621                                         }
 622                                 }
 623
 624                                 if ( $text === false ) {
 625                                         // Fallback to asking the database
 626                                         $tryIsPrefetch = false;
 627                                         if ( $this->spawn ) {
 628                                                 $text = $this->getTextSpawned( $id );
 629                                         } else {
 630                                                 $text = $this->getTextDb( $id );
 631                                         }
 632
 633                                         if ( $text !== false && $model !== false ) {
 634                                                 // Apply export transformation to text coming from the database.
 635                                                 // Prefetched text should already have transformations applied.
 636                                                 $text = $this->exportTransform( $text, $model, $format );
 637                                         }
 638
 639                                         // No more checks for texts from DB for now.
 640                                         // If we received something that is not false,
 641                                         // We treat it as good text, regardless of whether it actually is or is not
 642                                         if ( $text !== false ) {
 643                                                 return $text;
 644                                         }
 645                                 }
 646
 647                                 if ( $text === false ) {
 648                                         throw new MWException( "Generic error while obtaining text for id " . $id );
 649                                 }
 650
 651                                 // We received a good candidate for the text of $id via some method
 652
 653                                 // Step 2: Checking for plausibility and return the text if it is
 654                                 //         plausible
 655                                 $revID = intval( $this->thisRev );
 656                                 if ( !isset( $this->db ) ) {
 657                                         throw new MWException( "No database available" );
 658                                 }
 659
 660                                 if ( $model !== CONTENT_MODEL_WIKITEXT ) {
 661                                         $revLength = strlen( $text );
 662                                 } else {
 663                                         $revLength = $this->db->selectField( 'revision', 'rev_len', [ 'rev_id' => $revID ] );
 664                                 }
 665
 666                                 if ( strlen( $text ) == $revLength ) {
 667                                         if ( $tryIsPrefetch ) {
 668                                                 $this->prefetchCount++;
 669                                         }
 670
 671                                         return $text;
 672                                 }
 673
 674                                 $text = false;
 675                                 throw new MWException( "Received text is unplausible for id " . $id );
 676                         } catch ( Exception $e ) {
 677                                 $msg = "getting/checking text " . $id . " failed (" . $e->getMessage() . ")";
 678                                 if ( $failures + 1 < $this->maxFailures ) {
 679                                         $msg .= " (Will retry " . ( $this->maxFailures - $failures - 1 ) . " more times)";
 680                                 }
 681                                 $this->progress( $msg );
 682                         }
 683
 684                         // Something went wrong; we did not a text that was plausible :(
 685                         $failures++;
 686
 687                         // A failure in a prefetch hit does not warrant resetting db connection etc.
 688                         if ( !$tryIsPrefetch ) {
 689                                 // After backing off for some time, we try to reboot the whole process as
 690                                 // much as possible to not carry over failures from one part to the other
 691                                 // parts
 692                                 sleep( $this->failureTimeout );
 693                                 try {
 694                                         $this->rotateDb();
 695                                         if ( $this->spawn ) {
 696                                                 $this->closeSpawn();
 697                                                 $this->openSpawn();
 698                                         }
 699                                 } catch ( Exception $e ) {
 700                                         $this->progress( "Rebooting getText infrastructure failed (" . $e->getMessage() . ")" .
 701                                                 " Trying to continue anyways" );
 702                                 }
 703                         }
 704                 }
 705
 706                 // Retirieving a good text for $id failed (at least) maxFailures times.
 707                 // We abort for this $id.
 708
 709                 // Restoring the consecutive failures, and maybe aborting, if the dump
 710                 // is too broken.
 711                 $consecutiveFailedTextRetrievals = $oldConsecutiveFailedTextRetrievals + 1;
 712                 if ( $consecutiveFailedTextRetrievals > $this->maxConsecutiveFailedTextRetrievals ) {
 713                         throw new MWException( "Graceful storage failure" );
 714                 }
 715
 716                 return "";
 717         }
 718
 719         /**
 720          * Loads the serialized content from storage.
 721          *
 722          * @param int|string $id Content address, or text row ID.
 723          * @return bool|string
 724          */
 725         private function getTextDb( $id ) {
 726                 $store = $this->getBlobStore();
 727                 $address = ( is_int( $id ) || strpos( $id, ':' ) === false )
 728                         ? SqlBlobStore::makeAddressFromTextId( (int)$id )
 729                         : $id;
 730
 731                 try {
 732                         $text = $store->getBlob( $address );
 733
 734                         $stripped = str_replace( "\r", "", $text );
 735                         $normalized = MediaWikiServices::getInstance()->getContentLanguage()
 736                                 ->normalize( $stripped );
 737
 738                         return $normalized;
 739                 } catch ( BlobAccessException $ex ) {
 740                         // XXX: log a warning?
 741                         return false;
 742                 }
 743         }
 744
 745         /**
 746          * @param int|string $address Content address, or text row ID.
 747          * @return bool|string
 748          */
 749         private function getTextSpawned( $address ) {
 750                 Wikimedia\suppressWarnings();
 751                 if ( !$this->spawnProc ) {
 752                         // First time?
 753                         $this->openSpawn();
 754                 }
 755                 $text = $this->getTextSpawnedOnce( $address );
 756                 Wikimedia\restoreWarnings();
 757
 758                 return $text;
 759         }
 760
 761         function openSpawn() {
 762                 global $IP;
 763
 764                 $wiki = WikiMap::getWikiIdFromDbDomain( WikiMap::getCurrentWikiDbDomain() );
 765                 if ( count( $this->php ) == 2 ) {
 766                         $mwscriptpath = $this->php[1];
 767                 } else {
 768                         $mwscriptpath = "$IP/../multiversion/MWScript.php";
 769                 }
 770                 if ( file_exists( $mwscriptpath ) ) {
 771                         $cmd = implode( " ",
 772                                 array_map( [ Shell::class, 'escape' ],
 773                                         [
 774                                                 $this->php[0],
 775                                                 $mwscriptpath,
 776                                                 "fetchText.php",
 777                                                 '--wiki', $wiki ] ) );
 778                 } else {
 779                         $cmd = implode( " ",
 780                                 array_map( [ Shell::class, 'escape' ],
 781                                         [
 782                                                 $this->php[0],
 783                                                 "$IP/maintenance/fetchText.php",
 784                                                 '--wiki', $wiki ] ) );
 785                 }
 786                 $spec = [
 787                         0 => [ "pipe", "r" ],
 788                         1 => [ "pipe", "w" ],
 789                         2 => [ "file", "/dev/null", "a" ] ];
 790                 $pipes = [];
 791
 792                 $this->progress( "Spawning database subprocess: $cmd" );
 793                 $this->spawnProc = proc_open( $cmd, $spec, $pipes );
 794                 if ( !$this->spawnProc ) {
 795                         $this->progress( "Subprocess spawn failed." );
 796
 797                         return false;
 798                 }
 799                 list(
 800                         $this->spawnWrite, // -> stdin
 801                         $this->spawnRead, // <- stdout
 802                 ) = $pipes;
 803
 804                 return true;
 805         }
 806
 807         private function closeSpawn() {
 808                 Wikimedia\suppressWarnings();
 809                 if ( $this->spawnRead ) {
 810                         fclose( $this->spawnRead );
 811                 }
 812                 $this->spawnRead = null;
 813                 if ( $this->spawnWrite ) {
 814                         fclose( $this->spawnWrite );
 815                 }
 816                 $this->spawnWrite = null;
 817                 if ( $this->spawnErr ) {
 818                         fclose( $this->spawnErr );
 819                 }
 820                 $this->spawnErr = false;
 821                 if ( $this->spawnProc ) {
 822                         pclose( $this->spawnProc );
 823                 }
 824                 $this->spawnProc = false;
 825                 Wikimedia\restoreWarnings();
 826         }
 827
 828         /**
 829          * @param int|string $address Content address, or text row ID.
 830          * @return bool|string
 831          */
 832         private function getTextSpawnedOnce( $address ) {
 833                 if ( is_int( $address ) || intval( $address ) ) {
 834                         $address = SqlBlobStore::makeAddressFromTextId( (int)$address );
 835                 }
 836
 837                 $ok = fwrite( $this->spawnWrite, "$address\n" );
 838                 // $this->progress( ">> $id" );
 839                 if ( !$ok ) {
 840                         return false;
 841                 }
 842
 843                 $ok = fflush( $this->spawnWrite );
 844                 // $this->progress( ">> [flush]" );
 845                 if ( !$ok ) {
 846                         return false;
 847                 }
 848
 849                 // check that the text address they are sending is the one we asked for
 850                 // this avoids out of sync revision text errors we have encountered in the past
 851                 $newAddress = fgets( $this->spawnRead );
 852                 if ( $newAddress === false ) {
 853                         return false;
 854                 }
 855                 $newAddress = trim( $newAddress );
 856                 if ( strpos( $newAddress, ':' ) === false ) {
 857                         $newAddress = SqlBlobStore::makeAddressFromTextId( intval( $newAddress ) );
 858                 }
 859
 860                 if ( $newAddress !== $address ) {
 861                         return false;
 862                 }
 863
 864                 $len = fgets( $this->spawnRead );
 865                 // $this->progress( "<< " . trim( $len ) );
 866                 if ( $len === false ) {
 867                         return false;
 868                 }
 869
 870                 $nbytes = intval( $len );
 871                 // actual error, not zero-length text
 872                 if ( $nbytes < 0 ) {
 873                         return false;
 874                 }
 875
 876                 $text = "";
 877
 878                 // Subprocess may not send everything at once, we have to loop.
 879                 while ( $nbytes > strlen( $text ) ) {
 880                         $buffer = fread( $this->spawnRead, $nbytes - strlen( $text ) );
 881                         if ( $buffer === false ) {
 882                                 break;
 883                         }
 884                         $text .= $buffer;
 885                 }
 886
 887                 $gotbytes = strlen( $text );
 888                 if ( $gotbytes != $nbytes ) {
 889                         $this->progress( "Expected $nbytes bytes from database subprocess, got $gotbytes " );
 890
 891                         return false;
 892                 }
 893
 894                 // Do normalization in the dump thread...
 895                 $stripped = str_replace( "\r", "", $text );
 896                 $normalized = MediaWikiServices::getInstance()->getContentLanguage()->
 897                         normalize( $stripped );
 898
 899                 return $normalized;
 900         }
 901
 902         function startElement( $parser, $name, $attribs ) {
 903                 $this->checkpointJustWritten = false;
 904
 905                 $this->clearOpenElement( null );
 906                 $this->lastName = $name;
 907
 908                 if ( $name == 'revision' ) {
 909                         $this->state = $name;
 910                         $this->egress->writeOpenPage( null, $this->buffer );
 911                         $this->buffer = "";
 912                 } elseif ( $name == 'page' ) {
 913                         $this->state = $name;
 914                         if ( $this->atStart ) {
 915                                 $this->egress->writeOpenStream( $this->buffer );
 916                                 $this->buffer = "";
 917                                 $this->atStart = false;
 918                         }
 919                 }
 920
 921                 if ( $name == "text" && isset( $attribs['id'] ) ) {
 922                         $id = $attribs['id'];
 923                         $model = trim( $this->thisRevModel );
 924                         $format = trim( $this->thisRevFormat );
 925
 926                         $model = $model === '' ? null : $model;
 927                         $format = $format === '' ? null : $format;
 928
 929                         $text = $this->getText( $id, $model, $format );
 930                         $this->openElement = [ $name, [ 'xml:space' => 'preserve' ] ];
 931                         if ( strlen( $text ) > 0 ) {
 932                                 $this->characterData( $parser, $text );
 933                         }
 934                 } else {
 935                         $this->openElement = [ $name, $attribs ];
 936                 }
 937         }
 938
 939         function endElement( $parser, $name ) {
 940                 $this->checkpointJustWritten = false;
 941
 942                 if ( $this->openElement ) {
 943                         $this->clearOpenElement( "" );
 944                 } else {
 945                         $this->buffer .= "</$name>";
 946                 }
 947
 948                 if ( $name == 'revision' ) {
 949                         $this->egress->writeRevision( null, $this->buffer );
 950                         $this->buffer = "";
 951                         $this->thisRev = "";
 952                         $this->thisRevModel = null;
 953                         $this->thisRevFormat = null;
 954                 } elseif ( $name == 'page' ) {
 955                         if ( !$this->firstPageWritten ) {
 956                                 $this->firstPageWritten = trim( $this->thisPage );
 957                         }
 958                         $this->lastPageWritten = trim( $this->thisPage );
 959                         if ( $this->timeExceeded ) {
 960                                 $this->egress->writeClosePage( $this->buffer );
 961                                 // nasty hack, we can't just write the chardata after the
 962                                 // page tag, it will include leading blanks from the next line
 963                                 $this->egress->sink->write( "\n" );
 964
 965                                 $this->buffer = $this->xmlwriterobj->closeStream();
 966                                 $this->egress->writeCloseStream( $this->buffer );
 967
 968                                 $this->buffer = "";
 969                                 $this->thisPage = "";
 970                                 // this could be more than one file if we had more than one output arg
 971
 972                                 $filenameList = (array)$this->egress->getFilenames();
 973                                 $newFilenames = [];
 974                                 $firstPageID = str_pad( $this->firstPageWritten, 9, "0", STR_PAD_LEFT );
 975                                 $lastPageID = str_pad( $this->lastPageWritten, 9, "0", STR_PAD_LEFT );
 976                                 $filenamesCount = count( $filenameList );
 977                                 for ( $i = 0; $i < $filenamesCount; $i++ ) {
 978                                         $checkpointNameFilledIn = sprintf( $this->checkpointFiles[$i], $firstPageID, $lastPageID );
 979                                         $fileinfo = pathinfo( $filenameList[$i] );
 980                                         $newFilenames[] = $fileinfo['dirname'] . '/' . $checkpointNameFilledIn;
 981                                 }
 982                                 $this->egress->closeRenameAndReopen( $newFilenames );
 983                                 $this->buffer = $this->xmlwriterobj->openStream();
 984                                 $this->timeExceeded = false;
 985                                 $this->timeOfCheckpoint = $this->lastTime;
 986                                 $this->firstPageWritten = false;
 987                                 $this->checkpointJustWritten = true;
 988                         } else {
 989                                 $this->egress->writeClosePage( $this->buffer );
 990                                 $this->buffer = "";
 991                                 $this->thisPage = "";
 992                         }
 993                 } elseif ( $name == 'mediawiki' ) {
 994                         $this->egress->writeCloseStream( $this->buffer );
 995                         $this->buffer = "";
 996                 }
 997         }
 998
 999         function characterData( $parser, $data ) {
1000                 $this->clearOpenElement( null );
1001                 if ( $this->lastName == "id" ) {
1002                         if ( $this->state == "revision" ) {
1003                                 $this->thisRev .= $data;
1004                         } elseif ( $this->state == "page" ) {
1005                                 $this->thisPage .= $data;
1006                         }
1007                 } elseif ( $this->lastName == "model" ) {
1008                         $this->thisRevModel .= $data;
1009                 } elseif ( $this->lastName == "format" ) {
1010                         $this->thisRevFormat .= $data;
1011                 }
1012
1013                 // have to skip the newline left over from closepagetag line of
1014                 // end of checkpoint files. nasty hack!!
1015                 if ( $this->checkpointJustWritten ) {
1016                         if ( $data[0] == "\n" ) {
1017                                 $data = substr( $data, 1 );
1018                         }
1019                         $this->checkpointJustWritten = false;
1020                 }
1021                 $this->buffer .= htmlspecialchars( $data );
1022         }
1023
1024         function clearOpenElement( $style ) {
1025                 if ( $this->openElement ) {
1026                         $this->buffer .= Xml::element( $this->openElement[0], $this->openElement[1], $style );
1027                         $this->openElement = false;
1028                 }
1029         }
1030 }