maintenance/dumpTextPass.php

   1 <?php
   2 /**
   3  * BackupDumper that postprocesses XML dumps from dumpBackup.php to add page text
   4  *
   5  * Copyright (C) 2005 Brion Vibber <brion@pobox.com>
   6  * https://www.mediawiki.org/
   7  *
   8  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 2 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with this program; if not, write to the Free Software Foundation, Inc.,
  20  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  21  * http://www.gnu.org/copyleft/gpl.html
  22  *
  23  * @file
  24  * @ingroup Maintenance
  25  */
  26
  27 require_once __DIR__ . '/includes/BackupDumper.php';
  28 require_once __DIR__ . '/7zip.inc';
  29 require_once __DIR__ . '/../includes/export/WikiExporter.php';
  30
  31 use MediaWiki\MediaWikiServices;
  32 use MediaWiki\Storage\BlobAccessException;
  33 use MediaWiki\Storage\SqlBlobStore;
  34 use Wikimedia\Rdbms\IMaintainableDatabase;
  35
  36 /**
  37  * @ingroup Maintenance
  38  */
  39 class TextPassDumper extends BackupDumper {
  40         /** @var BaseDump */
  41         public $prefetch = null;
  42         /** @var string|bool */
  43         private $thisPage;
  44         /** @var string|bool */
  45         private $thisRev;
  46
  47         // when we spend more than maxTimeAllowed seconds on this run, we continue
  48         // processing until we write out the next complete page, then save output file(s),
  49         // rename it/them and open new one(s)
  50         public $maxTimeAllowed = 0; // 0 = no limit
  51
  52         protected $input = "php://stdin";
  53         protected $history = WikiExporter::FULL;
  54         protected $fetchCount = 0;
  55         protected $prefetchCount = 0;
  56         protected $prefetchCountLast = 0;
  57         protected $fetchCountLast = 0;
  58
  59         protected $maxFailures = 5;
  60         protected $maxConsecutiveFailedTextRetrievals = 200;
  61         protected $failureTimeout = 5; // Seconds to sleep after db failure
  62
  63         protected $bufferSize = 524288; // In bytes. Maximum size to read from the stub in on go.
  64
  65         protected $php = "php";
  66         protected $spawn = false;
  67
  68         /**
  69          * @var bool|resource
  70          */
  71         protected $spawnProc = false;
  72
  73         /**
  74          * @var bool|resource
  75          */
  76         protected $spawnWrite = false;
  77
  78         /**
  79          * @var bool|resource
  80          */
  81         protected $spawnRead = false;
  82
  83         /**
  84          * @var bool|resource
  85          */
  86         protected $spawnErr = false;
  87
  88         /**
  89          * @var bool|XmlDumpWriter
  90          */
  91         protected $xmlwriterobj = false;
  92
  93         protected $timeExceeded = false;
  94         protected $firstPageWritten = false;
  95         protected $lastPageWritten = false;
  96         protected $checkpointJustWritten = false;
  97         protected $checkpointFiles = [];
  98
  99         /**
 100          * @var IMaintainableDatabase
 101          */
 102         protected $db;
 103
 104         /**
 105          * @param array|null $args For backward compatibility
 106          */
 107         function __construct( $args = null ) {
 108                 parent::__construct();
 109
 110                 $this->addDescription( <<<TEXT
 111 This script postprocesses XML dumps from dumpBackup.php to add
 112 page text which was stubbed out (using --stub).
 113
 114 XML input is accepted on stdin.
 115 XML output is sent to stdout; progress reports are sent to stderr.
 116 TEXT
 117                 );
 118                 $this->stderr = fopen( "php://stderr", "wt" );
 119
 120                 $this->addOption( 'stub', 'To load a compressed stub dump instead of stdin. ' .
 121                         'Specify as --stub=<type>:<file>.', false, true );
 122                 $this->addOption( 'prefetch', 'Use a prior dump file as a text source, to savepressure on the ' .
 123                         'database. (Requires the XMLReader extension). Specify as --prefetch=<type>:<file>',
 124                         false, true );
 125                 $this->addOption( 'maxtime', 'Write out checkpoint file after this many minutes (writing' .
 126                         'out complete page, closing xml file properly, and opening new one' .
 127                         'with header).  This option requires the checkpointfile option.', false, true );
 128                 $this->addOption( 'checkpointfile', 'Use this string for checkpoint filenames,substituting ' .
 129                         'first pageid written for the first %s (required) and the last pageid written for the ' .
 130                         'second %s if it exists.', false, true, false, true ); // This can be specified multiple times
 131                 $this->addOption( 'quiet', 'Don\'t dump status reports to stderr.' );
 132                 $this->addOption( 'full', 'Dump all revisions of every page' );
 133                 $this->addOption( 'current', 'Base ETA on number of pages in database instead of all revisions' );
 134                 $this->addOption( 'spawn', 'Spawn a subprocess for loading text records' );
 135                 $this->addOption( 'buffersize', 'Buffer size in bytes to use for reading the stub. ' .
 136                         '(Default: 512KB, Minimum: 4KB)', false, true );
 137
 138                 if ( $args ) {
 139                         $this->loadWithArgv( $args );
 140                         $this->processOptions();
 141                 }
 142         }
 143
 144         /**
 145          * @return SqlBlobStore
 146          */
 147         private function getBlobStore() {
 148                 return MediaWikiServices::getInstance()->getBlobStore();
 149         }
 150
 151         function execute() {
 152                 $this->processOptions();
 153                 $this->dump( true );
 154         }
 155
 156         function processOptions() {
 157                 parent::processOptions();
 158
 159                 if ( $this->hasOption( 'buffersize' ) ) {
 160                         $this->bufferSize = max( intval( $this->getOption( 'buffersize' ) ), 4 * 1024 );
 161                 }
 162
 163                 if ( $this->hasOption( 'prefetch' ) ) {
 164                         $url = $this->processFileOpt( $this->getOption( 'prefetch' ) );
 165                         $this->prefetch = new BaseDump( $url );
 166                 }
 167
 168                 if ( $this->hasOption( 'stub' ) ) {
 169                         $this->input = $this->processFileOpt( $this->getOption( 'stub' ) );
 170                 }
 171
 172                 if ( $this->hasOption( 'maxtime' ) ) {
 173                         $this->maxTimeAllowed = intval( $this->getOption( 'maxtime' ) ) * 60;
 174                 }
 175
 176                 if ( $this->hasOption( 'checkpointfile' ) ) {
 177                         $this->checkpointFiles = $this->getOption( 'checkpointfile' );
 178                 }
 179
 180                 if ( $this->hasOption( 'current' ) ) {
 181                         $this->history = WikiExporter::CURRENT;
 182                 }
 183
 184                 if ( $this->hasOption( 'full' ) ) {
 185                         $this->history = WikiExporter::FULL;
 186                 }
 187
 188                 if ( $this->hasOption( 'spawn' ) ) {
 189                         $this->spawn = true;
 190                         $val = $this->getOption( 'spawn' );
 191                         if ( $val !== 1 ) {
 192                                 $this->php = $val;
 193                         }
 194                 }
 195         }
 196
 197         /**
 198          * Drop the database connection $this->db and try to get a new one.
 199          *
 200          * This function tries to get a /different/ connection if this is
 201          * possible. Hence, (if this is possible) it switches to a different
 202          * failover upon each call.
 203          *
 204          * This function resets $this->lb and closes all connections on it.
 205          *
 206          * @throws MWException
 207          */
 208         function rotateDb() {
 209                 // Cleaning up old connections
 210                 if ( isset( $this->lb ) ) {
 211                         $this->lb->closeAll();
 212                         unset( $this->lb );
 213                 }
 214
 215                 if ( $this->forcedDb !== null ) {
 216                         $this->db = $this->forcedDb;
 217
 218                         return;
 219                 }
 220
 221                 if ( isset( $this->db ) && $this->db->isOpen() ) {
 222                         throw new MWException( 'DB is set and has not been closed by the Load Balancer' );
 223                 }
 224
 225                 unset( $this->db );
 226
 227                 // Trying to set up new connection.
 228                 // We do /not/ retry upon failure, but delegate to encapsulating logic, to avoid
 229                 // individually retrying at different layers of code.
 230
 231                 try {
 232                         $lbFactory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
 233                         $this->lb = $lbFactory->newMainLB();
 234                 } catch ( Exception $e ) {
 235                         throw new MWException( __METHOD__
 236                                 . " rotating DB failed to obtain new load balancer (" . $e->getMessage() . ")" );
 237                 }
 238
 239                 try {
 240                         $this->db = $this->lb->getConnection( DB_REPLICA, 'dump' );
 241                 } catch ( Exception $e ) {
 242                         throw new MWException( __METHOD__
 243                                 . " rotating DB failed to obtain new database (" . $e->getMessage() . ")" );
 244                 }
 245         }
 246
 247         function initProgress( $history = WikiExporter::FULL ) {
 248                 parent::initProgress();
 249                 $this->timeOfCheckpoint = $this->startTime;
 250         }
 251
 252         function dump( $history, $text = WikiExporter::TEXT ) {
 253                 // Notice messages will foul up your XML output even if they're
 254                 // relatively harmless.
 255                 if ( ini_get( 'display_errors' ) ) {
 256                         ini_set( 'display_errors', 'stderr' );
 257                 }
 258
 259                 $this->initProgress( $this->history );
 260
 261                 // We are trying to get an initial database connection to avoid that the
 262                 // first try of this request's first call to getText fails. However, if
 263                 // obtaining a good DB connection fails it's not a serious issue, as
 264                 // getText does retry upon failure and can start without having a working
 265                 // DB connection.
 266                 try {
 267                         $this->rotateDb();
 268                 } catch ( Exception $e ) {
 269                         // We do not even count this as failure. Just let eventual
 270                         // watchdogs know.
 271                         $this->progress( "Getting initial DB connection failed (" .
 272                                 $e->getMessage() . ")" );
 273                 }
 274
 275                 $this->egress = new ExportProgressFilter( $this->sink, $this );
 276
 277                 // it would be nice to do it in the constructor, oh well. need egress set
 278                 $this->finalOptionCheck();
 279
 280                 // we only want this so we know how to close a stream :-P
 281                 $this->xmlwriterobj = new XmlDumpWriter();
 282
 283                 $input = fopen( $this->input, "rt" );
 284                 $this->readDump( $input );
 285
 286                 if ( $this->spawnProc ) {
 287                         $this->closeSpawn();
 288                 }
 289
 290                 $this->report( true );
 291         }
 292
 293         function processFileOpt( $opt ) {
 294                 $split = explode( ':', $opt, 2 );
 295                 $val = $split[0];
 296                 $param = '';
 297                 if ( count( $split ) === 2 ) {
 298                         $param = $split[1];
 299                 }
 300                 $fileURIs = explode( ';', $param );
 301                 foreach ( $fileURIs as $URI ) {
 302                         switch ( $val ) {
 303                                 case "file":
 304                                         $newURI = $URI;
 305                                         break;
 306                                 case "gzip":
 307                                         $newURI = "compress.zlib://$URI";
 308                                         break;
 309                                 case "bzip2":
 310                                         $newURI = "compress.bzip2://$URI";
 311                                         break;
 312                                 case "7zip":
 313                                         $newURI = "mediawiki.compress.7z://$URI";
 314                                         break;
 315                                 default:
 316                                         $newURI = $URI;
 317                         }
 318                         $newFileURIs[] = $newURI;
 319                 }
 320                 $val = implode( ';', $newFileURIs );
 321
 322                 return $val;
 323         }
 324
 325         /**
 326          * Overridden to include prefetch ratio if enabled.
 327          */
 328         function showReport() {
 329                 if ( !$this->prefetch ) {
 330                         parent::showReport();
 331
 332                         return;
 333                 }
 334
 335                 if ( $this->reporting ) {
 336                         $now = wfTimestamp( TS_DB );
 337                         $nowts = microtime( true );
 338                         $deltaAll = $nowts - $this->startTime;
 339                         $deltaPart = $nowts - $this->lastTime;
 340                         $this->pageCountPart = $this->pageCount - $this->pageCountLast;
 341                         $this->revCountPart = $this->revCount - $this->revCountLast;
 342
 343                         if ( $deltaAll ) {
 344                                 $portion = $this->revCount / $this->maxCount;
 345                                 $eta = $this->startTime + $deltaAll / $portion;
 346                                 $etats = wfTimestamp( TS_DB, intval( $eta ) );
 347                                 if ( $this->fetchCount ) {
 348                                         $fetchRate = 100.0 * $this->prefetchCount / $this->fetchCount;
 349                                 } else {
 350                                         $fetchRate = '-';
 351                                 }
 352                                 $pageRate = $this->pageCount / $deltaAll;
 353                                 $revRate = $this->revCount / $deltaAll;
 354                         } else {
 355                                 $pageRate = '-';
 356                                 $revRate = '-';
 357                                 $etats = '-';
 358                                 $fetchRate = '-';
 359                         }
 360                         if ( $deltaPart ) {
 361                                 if ( $this->fetchCountLast ) {
 362                                         $fetchRatePart = 100.0 * $this->prefetchCountLast / $this->fetchCountLast;
 363                                 } else {
 364                                         $fetchRatePart = '-';
 365                                 }
 366                                 $pageRatePart = $this->pageCountPart / $deltaPart;
 367                                 $revRatePart = $this->revCountPart / $deltaPart;
 368                         } else {
 369                                 $fetchRatePart = '-';
 370                                 $pageRatePart = '-';
 371                                 $revRatePart = '-';
 372                         }
 373                         $this->progress( sprintf(
 374                                 "%s: %s (ID %d) %d pages (%0.1f|%0.1f/sec all|curr), "
 375                                         . "%d revs (%0.1f|%0.1f/sec all|curr), %0.1f%%|%0.1f%% "
 376                                         . "prefetched (all|curr), ETA %s [max %d]",
 377                                 $now, wfWikiID(), $this->ID, $this->pageCount, $pageRate,
 378                                 $pageRatePart, $this->revCount, $revRate, $revRatePart,
 379                                 $fetchRate, $fetchRatePart, $etats, $this->maxCount
 380                         ) );
 381                         $this->lastTime = $nowts;
 382                         $this->revCountLast = $this->revCount;
 383                         $this->prefetchCountLast = $this->prefetchCount;
 384                         $this->fetchCountLast = $this->fetchCount;
 385                 }
 386         }
 387
 388         function setTimeExceeded() {
 389                 $this->timeExceeded = true;
 390         }
 391
 392         function checkIfTimeExceeded() {
 393                 if ( $this->maxTimeAllowed
 394                         && ( $this->lastTime - $this->timeOfCheckpoint > $this->maxTimeAllowed )
 395                 ) {
 396                         return true;
 397                 }
 398
 399                 return false;
 400         }
 401
 402         function finalOptionCheck() {
 403                 if ( ( $this->checkpointFiles && !$this->maxTimeAllowed )
 404                         || ( $this->maxTimeAllowed && !$this->checkpointFiles )
 405                 ) {
 406                         throw new MWException( "Options checkpointfile and maxtime must be specified together.\n" );
 407                 }
 408                 foreach ( $this->checkpointFiles as $checkpointFile ) {
 409                         $count = substr_count( $checkpointFile, "%s" );
 410                         if ( $count != 2 ) {
 411                                 throw new MWException( "Option checkpointfile must contain two '%s' "
 412                                         . "for substitution of first and last pageids, count is $count instead, "
 413                                         . "file is $checkpointFile.\n" );
 414                         }
 415                 }
 416
 417                 if ( $this->checkpointFiles ) {
 418                         $filenameList = (array)$this->egress->getFilenames();
 419                         if ( count( $filenameList ) != count( $this->checkpointFiles ) ) {
 420                                 throw new MWException( "One checkpointfile must be specified "
 421                                         . "for each output option, if maxtime is used.\n" );
 422                         }
 423                 }
 424         }
 425
 426         /**
 427          * @throws MWException Failure to parse XML input
 428          * @param string $input
 429          * @return bool
 430          */
 431         function readDump( $input ) {
 432                 $this->buffer = "";
 433                 $this->openElement = false;
 434                 $this->atStart = true;
 435                 $this->state = "";
 436                 $this->lastName = "";
 437                 $this->thisPage = 0;
 438                 $this->thisRev = 0;
 439                 $this->thisRevModel = null;
 440                 $this->thisRevFormat = null;
 441
 442                 $parser = xml_parser_create( "UTF-8" );
 443                 xml_parser_set_option( $parser, XML_OPTION_CASE_FOLDING, false );
 444
 445                 xml_set_element_handler(
 446                         $parser,
 447                         [ $this, 'startElement' ],
 448                         [ $this, 'endElement' ]
 449                 );
 450                 xml_set_character_data_handler( $parser, [ $this, 'characterData' ] );
 451
 452                 $offset = 0; // for context extraction on error reporting
 453                 do {
 454                         if ( $this->checkIfTimeExceeded() ) {
 455                                 $this->setTimeExceeded();
 456                         }
 457                         $chunk = fread( $input, $this->bufferSize );
 458                         if ( !xml_parse( $parser, $chunk, feof( $input ) ) ) {
 459                                 wfDebug( "TextDumpPass::readDump encountered XML parsing error\n" );
 460
 461                                 $byte = xml_get_current_byte_index( $parser );
 462                                 $msg = wfMessage( 'xml-error-string',
 463                                         'XML import parse failure',
 464                                         xml_get_current_line_number( $parser ),
 465                                         xml_get_current_column_number( $parser ),
 466                                         $byte . ( is_null( $chunk ) ? null : ( '; "' . substr( $chunk, $byte - $offset, 16 ) . '"' ) ),
 467                                         xml_error_string( xml_get_error_code( $parser ) ) )->escaped();
 468
 469                                 xml_parser_free( $parser );
 470
 471                                 throw new MWException( $msg );
 472                         }
 473                         $offset += strlen( $chunk );
 474                 } while ( $chunk !== false && !feof( $input ) );
 475                 if ( $this->maxTimeAllowed ) {
 476                         $filenameList = (array)$this->egress->getFilenames();
 477                         // we wrote some stuff after last checkpoint that needs renamed
 478                         if ( file_exists( $filenameList[0] ) ) {
 479                                 $newFilenames = [];
 480                                 # we might have just written the header and footer and had no
 481                                 # pages or revisions written... perhaps they were all deleted
 482                                 # there's no pageID 0 so we use that. the caller is responsible
 483                                 # for deciding what to do with a file containing only the
 484                                 # siteinfo information and the mw tags.
 485                                 if ( !$this->firstPageWritten ) {
 486                                         $firstPageID = str_pad( 0, 9, "0", STR_PAD_LEFT );
 487                                         $lastPageID = str_pad( 0, 9, "0", STR_PAD_LEFT );
 488                                 } else {
 489                                         $firstPageID = str_pad( $this->firstPageWritten, 9, "0", STR_PAD_LEFT );
 490                                         $lastPageID = str_pad( $this->lastPageWritten, 9, "0", STR_PAD_LEFT );
 491                                 }
 492
 493                                 $filenameCount = count( $filenameList );
 494                                 for ( $i = 0; $i < $filenameCount; $i++ ) {
 495                                         $checkpointNameFilledIn = sprintf( $this->checkpointFiles[$i], $firstPageID, $lastPageID );
 496                                         $fileinfo = pathinfo( $filenameList[$i] );
 497                                         $newFilenames[] = $fileinfo['dirname'] . '/' . $checkpointNameFilledIn;
 498                                 }
 499                                 $this->egress->closeAndRename( $newFilenames );
 500                         }
 501                 }
 502                 xml_parser_free( $parser );
 503
 504                 return true;
 505         }
 506
 507         /**
 508          * Applies applicable export transformations to $text.
 509          *
 510          * @param string $text
 511          * @param string $model
 512          * @param string|null $format
 513          *
 514          * @return string
 515          */
 516         private function exportTransform( $text, $model, $format = null ) {
 517                 try {
 518                         $handler = ContentHandler::getForModelID( $model );
 519                         $text = $handler->exportTransform( $text, $format );
 520                 }
 521                 catch ( MWException $ex ) {
 522                         $this->progress(
 523                                 "Unable to apply export transformation for content model '$model': " .
 524                                 $ex->getMessage()
 525                         );
 526                 }
 527
 528                 return $text;
 529         }
 530
 531         /**
 532          * Tries to load revision text.
 533          * Export transformations are applied if the content model is given or can be
 534          * determined from the database.
 535          *
 536          * Upon errors, retries (Up to $this->maxFailures tries each call).
 537          * If still no good revision could be found even after this retrying, "" is returned.
 538          * If no good revision text could be returned for
 539          * $this->maxConsecutiveFailedTextRetrievals consecutive calls to getText, MWException
 540          * is thrown.
 541          *
 542          * @param int|string $id Content address, or text row ID.
 543          * @param string|bool|null $model The content model used to determine
 544          *  applicable export transformations.
 545          *  If $model is null, it will be determined from the database.
 546          * @param string|null $format The content format used when applying export transformations.
 547          *
 548          * @throws MWException
 549          * @return string The revision text for $id, or ""
 550          */
 551         function getText( $id, $model = null, $format = null ) {
 552                 global $wgContentHandlerUseDB;
 553
 554                 $prefetchNotTried = true; // Whether or not we already tried to get the text via prefetch.
 555                 $text = false; // The candidate for a good text. false if no proper value.
 556                 $failures = 0; // The number of times, this invocation of getText already failed.
 557
 558                 // The number of times getText failed without yielding a good text in between.
 559                 static $consecutiveFailedTextRetrievals = 0;
 560
 561                 $this->fetchCount++;
 562
 563                 // To allow to simply return on success and do not have to worry about book keeping,
 564                 // we assume, this fetch works (possible after some retries). Nevertheless, we koop
 565                 // the old value, so we can restore it, if problems occur (See after the while loop).
 566                 $oldConsecutiveFailedTextRetrievals = $consecutiveFailedTextRetrievals;
 567                 $consecutiveFailedTextRetrievals = 0;
 568
 569                 if ( $model === null && $wgContentHandlerUseDB ) {
 570                         // TODO: MCR: use content table
 571                         $row = $this->db->selectRow(
 572                                 'revision',
 573                                 [ 'rev_content_model', 'rev_content_format' ],
 574                                 [ 'rev_id' => $this->thisRev ],
 575                                 __METHOD__
 576                         );
 577
 578                         if ( $row ) {
 579                                 $model = $row->rev_content_model;
 580                                 $format = $row->rev_content_format;
 581                         }
 582                 }
 583
 584                 if ( $model === null || $model === '' ) {
 585                         $model = false;
 586                 }
 587
 588                 while ( $failures < $this->maxFailures ) {
 589                         // As soon as we found a good text for the $id, we will return immediately.
 590                         // Hence, if we make it past the try catch block, we know that we did not
 591                         // find a good text.
 592
 593                         try {
 594                                 // Step 1: Get some text (or reuse from previous iteratuon if checking
 595                                 //         for plausibility failed)
 596
 597                                 // Trying to get prefetch, if it has not been tried before
 598                                 if ( $text === false && isset( $this->prefetch ) && $prefetchNotTried ) {
 599                                         $prefetchNotTried = false;
 600                                         $tryIsPrefetch = true;
 601                                         $text = $this->prefetch->prefetch( (int)$this->thisPage, (int)$this->thisRev );
 602
 603                                         if ( $text === null ) {
 604                                                 $text = false;
 605                                         }
 606
 607                                         if ( is_string( $text ) && $model !== false ) {
 608                                                 // Apply export transformation to text coming from an old dump.
 609                                                 // The purpose of this transformation is to convert up from legacy
 610                                                 // formats, which may still be used in the older dump that is used
 611                                                 // for pre-fetching. Applying the transformation again should not
 612                                                 // interfere with content that is already in the correct form.
 613                                                 $text = $this->exportTransform( $text, $model, $format );
 614                                         }
 615                                 }
 616
 617                                 if ( $text === false ) {
 618                                         // Fallback to asking the database
 619                                         $tryIsPrefetch = false;
 620                                         if ( $this->spawn ) {
 621                                                 $text = $this->getTextSpawned( $id );
 622                                         } else {
 623                                                 $text = $this->getTextDb( $id );
 624                                         }
 625
 626                                         if ( $text !== false && $model !== false ) {
 627                                                 // Apply export transformation to text coming from the database.
 628                                                 // Prefetched text should already have transformations applied.
 629                                                 $text = $this->exportTransform( $text, $model, $format );
 630                                         }
 631
 632                                         // No more checks for texts from DB for now.
 633                                         // If we received something that is not false,
 634                                         // We treat it as good text, regardless of whether it actually is or is not
 635                                         if ( $text !== false ) {
 636                                                 return $text;
 637                                         }
 638                                 }
 639
 640                                 if ( $text === false ) {
 641                                         throw new MWException( "Generic error while obtaining text for id " . $id );
 642                                 }
 643
 644                                 // We received a good candidate for the text of $id via some method
 645
 646                                 // Step 2: Checking for plausibility and return the text if it is
 647                                 //         plausible
 648                                 $revID = intval( $this->thisRev );
 649                                 if ( !isset( $this->db ) ) {
 650                                         throw new MWException( "No database available" );
 651                                 }
 652
 653                                 if ( $model !== CONTENT_MODEL_WIKITEXT ) {
 654                                         $revLength = strlen( $text );
 655                                 } else {
 656                                         $revLength = $this->db->selectField( 'revision', 'rev_len', [ 'rev_id' => $revID ] );
 657                                 }
 658
 659                                 if ( strlen( $text ) == $revLength ) {
 660                                         if ( $tryIsPrefetch ) {
 661                                                 $this->prefetchCount++;
 662                                         }
 663
 664                                         return $text;
 665                                 }
 666
 667                                 $text = false;
 668                                 throw new MWException( "Received text is unplausible for id " . $id );
 669                         } catch ( Exception $e ) {
 670                                 $msg = "getting/checking text " . $id . " failed (" . $e->getMessage() . ")";
 671                                 if ( $failures + 1 < $this->maxFailures ) {
 672                                         $msg .= " (Will retry " . ( $this->maxFailures - $failures - 1 ) . " more times)";
 673                                 }
 674                                 $this->progress( $msg );
 675                         }
 676
 677                         // Something went wrong; we did not a text that was plausible :(
 678                         $failures++;
 679
 680                         // A failure in a prefetch hit does not warrant resetting db connection etc.
 681                         if ( !$tryIsPrefetch ) {
 682                                 // After backing off for some time, we try to reboot the whole process as
 683                                 // much as possible to not carry over failures from one part to the other
 684                                 // parts
 685                                 sleep( $this->failureTimeout );
 686                                 try {
 687                                         $this->rotateDb();
 688                                         if ( $this->spawn ) {
 689                                                 $this->closeSpawn();
 690                                                 $this->openSpawn();
 691                                         }
 692                                 } catch ( Exception $e ) {
 693                                         $this->progress( "Rebooting getText infrastructure failed (" . $e->getMessage() . ")" .
 694                                                 " Trying to continue anyways" );
 695                                 }
 696                         }
 697                 }
 698
 699                 // Retirieving a good text for $id failed (at least) maxFailures times.
 700                 // We abort for this $id.
 701
 702                 // Restoring the consecutive failures, and maybe aborting, if the dump
 703                 // is too broken.
 704                 $consecutiveFailedTextRetrievals = $oldConsecutiveFailedTextRetrievals + 1;
 705                 if ( $consecutiveFailedTextRetrievals > $this->maxConsecutiveFailedTextRetrievals ) {
 706                         throw new MWException( "Graceful storage failure" );
 707                 }
 708
 709                 return "";
 710         }
 711
 712         /**
 713          * Loads the serialized content from storage.
 714          *
 715          * @param int|string $id Content address, or text row ID.
 716          * @return bool|string
 717          */
 718         private function getTextDb( $id ) {
 719                 $store = $this->getBlobStore();
 720                 $address = ( is_int( $id ) || strpos( $id, ':' ) === false )
 721                         ? SqlBlobStore::makeAddressFromTextId( (int)$id )
 722                         : $id;
 723
 724                 try {
 725                         $text = $store->getBlob( $address );
 726
 727                         $stripped = str_replace( "\r", "", $text );
 728                         $normalized = MediaWikiServices::getInstance()->getContentLanguage()
 729                                 ->normalize( $stripped );
 730
 731                         return $normalized;
 732                 } catch ( BlobAccessException $ex ) {
 733                         // XXX: log a warning?
 734                         return false;
 735                 }
 736         }
 737
 738         /**
 739          * @param int|string $address Content address, or text row ID.
 740          * @return bool|string
 741          */
 742         private function getTextSpawned( $address ) {
 743                 Wikimedia\suppressWarnings();
 744                 if ( !$this->spawnProc ) {
 745                         // First time?
 746                         $this->openSpawn();
 747                 }
 748                 $text = $this->getTextSpawnedOnce( $address );
 749                 Wikimedia\restoreWarnings();
 750
 751                 return $text;
 752         }
 753
 754         function openSpawn() {
 755                 global $IP;
 756
 757                 if ( file_exists( "$IP/../multiversion/MWScript.php" ) ) {
 758                         $cmd = implode( " ",
 759                                 array_map( 'wfEscapeShellArg',
 760                                         [
 761                                                 $this->php,
 762                                                 "$IP/../multiversion/MWScript.php",
 763                                                 "fetchText.php",
 764                                                 '--wiki', wfWikiID() ] ) );
 765                 } else {
 766                         $cmd = implode( " ",
 767                                 array_map( 'wfEscapeShellArg',
 768                                         [
 769                                                 $this->php,
 770                                                 "$IP/maintenance/fetchText.php",
 771                                                 '--wiki', wfWikiID() ] ) );
 772                 }
 773                 $spec = [
 774                         0 => [ "pipe", "r" ],
 775                         1 => [ "pipe", "w" ],
 776                         2 => [ "file", "/dev/null", "a" ] ];
 777                 $pipes = [];
 778
 779                 $this->progress( "Spawning database subprocess: $cmd" );
 780                 $this->spawnProc = proc_open( $cmd, $spec, $pipes );
 781                 if ( !$this->spawnProc ) {
 782                         $this->progress( "Subprocess spawn failed." );
 783
 784                         return false;
 785                 }
 786                 list(
 787                         $this->spawnWrite, // -> stdin
 788                         $this->spawnRead, // <- stdout
 789                 ) = $pipes;
 790
 791                 return true;
 792         }
 793
 794         private function closeSpawn() {
 795                 Wikimedia\suppressWarnings();
 796                 if ( $this->spawnRead ) {
 797                         fclose( $this->spawnRead );
 798                 }
 799                 $this->spawnRead = false;
 800                 if ( $this->spawnWrite ) {
 801                         fclose( $this->spawnWrite );
 802                 }
 803                 $this->spawnWrite = false;
 804                 if ( $this->spawnErr ) {
 805                         fclose( $this->spawnErr );
 806                 }
 807                 $this->spawnErr = false;
 808                 if ( $this->spawnProc ) {
 809                         pclose( $this->spawnProc );
 810                 }
 811                 $this->spawnProc = false;
 812                 Wikimedia\restoreWarnings();
 813         }
 814
 815         /**
 816          * @param int|string $address Content address, or text row ID.
 817          * @return bool|string
 818          */
 819         private function getTextSpawnedOnce( $address ) {
 820                 if ( is_int( $address ) || intval( $address ) ) {
 821                         $address = SqlBlobStore::makeAddressFromTextId( (int)$address );
 822                 }
 823
 824                 $ok = fwrite( $this->spawnWrite, "$address\n" );
 825                 // $this->progress( ">> $id" );
 826                 if ( !$ok ) {
 827                         return false;
 828                 }
 829
 830                 $ok = fflush( $this->spawnWrite );
 831                 // $this->progress( ">> [flush]" );
 832                 if ( !$ok ) {
 833                         return false;
 834                 }
 835
 836                 // check that the text address they are sending is the one we asked for
 837                 // this avoids out of sync revision text errors we have encountered in the past
 838                 $newAddress = fgets( $this->spawnRead );
 839                 if ( $newAddress === false ) {
 840                         return false;
 841                 }
 842                 $newAddress = trim( $newAddress );
 843                 if ( strpos( $newAddress, ':' ) === false ) {
 844                         $newAddress = SqlBlobStore::makeAddressFromTextId( intval( $newAddress ) );
 845                 }
 846
 847                 if ( $newAddress !== $address ) {
 848                         return false;
 849                 }
 850
 851                 $len = fgets( $this->spawnRead );
 852                 // $this->progress( "<< " . trim( $len ) );
 853                 if ( $len === false ) {
 854                         return false;
 855                 }
 856
 857                 $nbytes = intval( $len );
 858                 // actual error, not zero-length text
 859                 if ( $nbytes < 0 ) {
 860                         return false;
 861                 }
 862
 863                 $text = "";
 864
 865                 // Subprocess may not send everything at once, we have to loop.
 866                 while ( $nbytes > strlen( $text ) ) {
 867                         $buffer = fread( $this->spawnRead, $nbytes - strlen( $text ) );
 868                         if ( $buffer === false ) {
 869                                 break;
 870                         }
 871                         $text .= $buffer;
 872                 }
 873
 874                 $gotbytes = strlen( $text );
 875                 if ( $gotbytes != $nbytes ) {
 876                         $this->progress( "Expected $nbytes bytes from database subprocess, got $gotbytes " );
 877
 878                         return false;
 879                 }
 880
 881                 // Do normalization in the dump thread...
 882                 $stripped = str_replace( "\r", "", $text );
 883                 $normalized = MediaWikiServices::getInstance()->getContentLanguage()->
 884                         normalize( $stripped );
 885
 886                 return $normalized;
 887         }
 888
 889         function startElement( $parser, $name, $attribs ) {
 890                 $this->checkpointJustWritten = false;
 891
 892                 $this->clearOpenElement( null );
 893                 $this->lastName = $name;
 894
 895                 if ( $name == 'revision' ) {
 896                         $this->state = $name;
 897                         $this->egress->writeOpenPage( null, $this->buffer );
 898                         $this->buffer = "";
 899                 } elseif ( $name == 'page' ) {
 900                         $this->state = $name;
 901                         if ( $this->atStart ) {
 902                                 $this->egress->writeOpenStream( $this->buffer );
 903                                 $this->buffer = "";
 904                                 $this->atStart = false;
 905                         }
 906                 }
 907
 908                 if ( $name == "text" && isset( $attribs['id'] ) ) {
 909                         $id = $attribs['id'];
 910                         $model = trim( $this->thisRevModel );
 911                         $format = trim( $this->thisRevFormat );
 912
 913                         $model = $model === '' ? null : $model;
 914                         $format = $format === '' ? null : $format;
 915
 916                         $text = $this->getText( $id, $model, $format );
 917                         $this->openElement = [ $name, [ 'xml:space' => 'preserve' ] ];
 918                         if ( strlen( $text ) > 0 ) {
 919                                 $this->characterData( $parser, $text );
 920                         }
 921                 } else {
 922                         $this->openElement = [ $name, $attribs ];
 923                 }
 924         }
 925
 926         function endElement( $parser, $name ) {
 927                 $this->checkpointJustWritten = false;
 928
 929                 if ( $this->openElement ) {
 930                         $this->clearOpenElement( "" );
 931                 } else {
 932                         $this->buffer .= "</$name>";
 933                 }
 934
 935                 if ( $name == 'revision' ) {
 936                         $this->egress->writeRevision( null, $this->buffer );
 937                         $this->buffer = "";
 938                         $this->thisRev = "";
 939                         $this->thisRevModel = null;
 940                         $this->thisRevFormat = null;
 941                 } elseif ( $name == 'page' ) {
 942                         if ( !$this->firstPageWritten ) {
 943                                 $this->firstPageWritten = trim( $this->thisPage );
 944                         }
 945                         $this->lastPageWritten = trim( $this->thisPage );
 946                         if ( $this->timeExceeded ) {
 947                                 $this->egress->writeClosePage( $this->buffer );
 948                                 // nasty hack, we can't just write the chardata after the
 949                                 // page tag, it will include leading blanks from the next line
 950                                 $this->egress->sink->write( "\n" );
 951
 952                                 $this->buffer = $this->xmlwriterobj->closeStream();
 953                                 $this->egress->writeCloseStream( $this->buffer );
 954
 955                                 $this->buffer = "";
 956                                 $this->thisPage = "";
 957                                 // this could be more than one file if we had more than one output arg
 958
 959                                 $filenameList = (array)$this->egress->getFilenames();
 960                                 $newFilenames = [];
 961                                 $firstPageID = str_pad( $this->firstPageWritten, 9, "0", STR_PAD_LEFT );
 962                                 $lastPageID = str_pad( $this->lastPageWritten, 9, "0", STR_PAD_LEFT );
 963                                 $filenamesCount = count( $filenameList );
 964                                 for ( $i = 0; $i < $filenamesCount; $i++ ) {
 965                                         $checkpointNameFilledIn = sprintf( $this->checkpointFiles[$i], $firstPageID, $lastPageID );
 966                                         $fileinfo = pathinfo( $filenameList[$i] );
 967                                         $newFilenames[] = $fileinfo['dirname'] . '/' . $checkpointNameFilledIn;
 968                                 }
 969                                 $this->egress->closeRenameAndReopen( $newFilenames );
 970                                 $this->buffer = $this->xmlwriterobj->openStream();
 971                                 $this->timeExceeded = false;
 972                                 $this->timeOfCheckpoint = $this->lastTime;
 973                                 $this->firstPageWritten = false;
 974                                 $this->checkpointJustWritten = true;
 975                         } else {
 976                                 $this->egress->writeClosePage( $this->buffer );
 977                                 $this->buffer = "";
 978                                 $this->thisPage = "";
 979                         }
 980                 } elseif ( $name == 'mediawiki' ) {
 981                         $this->egress->writeCloseStream( $this->buffer );
 982                         $this->buffer = "";
 983                 }
 984         }
 985
 986         function characterData( $parser, $data ) {
 987                 $this->clearOpenElement( null );
 988                 if ( $this->lastName == "id" ) {
 989                         if ( $this->state == "revision" ) {
 990                                 $this->thisRev .= $data;
 991                         } elseif ( $this->state == "page" ) {
 992                                 $this->thisPage .= $data;
 993                         }
 994                 } elseif ( $this->lastName == "model" ) {
 995                         $this->thisRevModel .= $data;
 996                 } elseif ( $this->lastName == "format" ) {
 997                         $this->thisRevFormat .= $data;
 998                 }
 999
1000                 // have to skip the newline left over from closepagetag line of
1001                 // end of checkpoint files. nasty hack!!
1002                 if ( $this->checkpointJustWritten ) {
1003                         if ( $data[0] == "\n" ) {
1004                                 $data = substr( $data, 1 );
1005                         }
1006                         $this->checkpointJustWritten = false;
1007                 }
1008                 $this->buffer .= htmlspecialchars( $data );
1009         }
1010
1011         function clearOpenElement( $style ) {
1012                 if ( $this->openElement ) {
1013                         $this->buffer .= Xml::element( $this->openElement[0], $this->openElement[1], $style );
1014                         $this->openElement = false;
1015                 }
1016         }
1017 }
1018
1019 $maintClass = TextPassDumper::class;
1020 require_once RUN_MAINTENANCE_IF_MAIN;