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