maintenance/includes/TextPassDumper.php

   1 <?php
   2 /**
   3  * BackupDumper that postprocesses XML dumps from dumpBackup.php to add page text
   4  *
   5  * Copyright (C) 2005 Brion Vibber <brion@pobox.com>
   6  * https://www.mediawiki.org/
   7  *
   8  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 2 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with this program; if not, write to the Free Software Foundation, Inc.,
  20  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  21  * http://www.gnu.org/copyleft/gpl.html
  22  *
  23  * @file
  24  * @ingroup Dump
  25  * @ingroup Maintenance
  26  */
  27
  28 require_once __DIR__ . '/BackupDumper.php';
  29 require_once __DIR__ . '/SevenZipStream.php';
  30 require_once __DIR__ . '/../../includes/export/WikiExporter.php';
  31
  32 use MediaWiki\MediaWikiServices;
  33 use MediaWiki\Shell\Shell;
  34 use MediaWiki\Storage\BlobAccessException;
  35 use MediaWiki\Storage\SqlBlobStore;
  36 use Wikimedia\Rdbms\IMaintainableDatabase;
  37
  38 /**
  39  * @ingroup Maintenance
  40  */
  41 class TextPassDumper extends BackupDumper {
  42         /** @var BaseDump */
  43         public $prefetch = null;
  44         /** @var string|bool */
  45         private $thisPage;
  46         /** @var string|bool */
  47         private $thisRev;
  48
  49         // when we spend more than maxTimeAllowed seconds on this run, we continue
  50         // processing until we write out the next complete page, then save output file(s),
  51         // rename it/them and open new one(s)
  52         public $maxTimeAllowed = 0; // 0 = no limit
  53
  54         protected $input = "php://stdin";
  55         protected $history = WikiExporter::FULL;
  56         protected $fetchCount = 0;
  57         protected $prefetchCount = 0;
  58         protected $prefetchCountLast = 0;
  59         protected $fetchCountLast = 0;
  60
  61         protected $maxFailures = 5;
  62         protected $maxConsecutiveFailedTextRetrievals = 200;
  63         protected $failureTimeout = 5; // Seconds to sleep after db failure
  64
  65         protected $bufferSize = 524288; // In bytes. Maximum size to read from the stub in on go.
  66
  67         /** @var array */
  68         protected $php = [];
  69         protected $spawn = false;
  70
  71         /**
  72          * @var bool|resource
  73          */
  74         protected $spawnProc = false;
  75
  76         /**
  77          * @var resource
  78          */
  79         protected $spawnWrite;
  80
  81         /**
  82          * @var resource
  83          */
  84         protected $spawnRead;
  85
  86         /**
  87          * @var bool|resource
  88          */
  89         protected $spawnErr = false;
  90
  91         /**
  92          * @var bool|XmlDumpWriter
  93          */
  94         protected $xmlwriterobj = false;
  95
  96         protected $timeExceeded = false;
  97         protected $firstPageWritten = false;
  98         protected $lastPageWritten = false;
  99         protected $checkpointJustWritten = false;
 100         /** @var string[] */
 101         protected $checkpointFiles = [];
 102
 103         /**
 104          * @var IMaintainableDatabase
 105          */
 106         protected $db;
 107
 108         /**
 109          * @param array|null $args For backward compatibility
 110          */
 111         function __construct( $args = null ) {
 112                 parent::__construct();
 113
 114                 $this->addDescription( <<<TEXT
 115 This script postprocesses XML dumps from dumpBackup.php to add
 116 page text which was stubbed out (using --stub).
 117
 118 XML input is accepted on stdin.
 119 XML output is sent to stdout; progress reports are sent to stderr.
 120 TEXT
 121                 );
 122                 $this->stderr = fopen( "php://stderr", "wt" );
 123
 124                 $this->addOption( 'stub', 'To load a compressed stub dump instead of stdin. ' .
 125                         'Specify as --stub=<type>:<file>.', false, true );
 126                 $this->addOption( 'prefetch', 'Use a prior dump file as a text source, to savepressure on the ' .
 127                         'database. (Requires the XMLReader extension). Specify as --prefetch=<type>:<file>',
 128                         false, true );
 129                 $this->addOption( 'maxtime', 'Write out checkpoint file after this many minutes (writing' .
 130                         'out complete page, closing xml file properly, and opening new one' .
 131                         'with header).  This option requires the checkpointfile option.', false, true );
 132                 $this->addOption( 'checkpointfile', 'Use this string for checkpoint filenames,substituting ' .
 133                         'first pageid written for the first %s (required) and the last pageid written for the ' .
 134                         'second %s if it exists.', false, true, false, true ); // This can be specified multiple times
 135                 $this->addOption( 'quiet', 'Don\'t dump status reports to stderr.' );
 136                 $this->addOption( 'full', 'Dump all revisions of every page' );
 137                 $this->addOption( 'current', 'Base ETA on number of pages in database instead of all revisions' );
 138                 $this->addOption( 'spawn', 'Spawn a subprocess for loading text records, optionally specify ' .
 139                         'php[,mwscript] paths' );
 140                 $this->addOption( 'buffersize', 'Buffer size in bytes to use for reading the stub. ' .
 141                         '(Default: 512KB, Minimum: 4KB)', false, true );
 142
 143                 if ( $args ) {
 144                         $this->loadWithArgv( $args );
 145                         $this->processOptions();
 146                 }
 147         }
 148
 149         /**
 150          * @return SqlBlobStore
 151          */
 152         private function getBlobStore() {
 153                 return MediaWikiServices::getInstance()->getBlobStore();
 154         }
 155
 156         function execute() {
 157                 $this->processOptions();
 158                 $this->dump( true );
 159         }
 160
 161         function processOptions() {
 162                 parent::processOptions();
 163
 164                 if ( $this->hasOption( 'buffersize' ) ) {
 165                         $this->bufferSize = max( intval( $this->getOption( 'buffersize' ) ), 4 * 1024 );
 166                 }
 167
 168                 if ( $this->hasOption( 'prefetch' ) ) {
 169                         $url = $this->processFileOpt( $this->getOption( 'prefetch' ) );
 170                         $this->prefetch = new BaseDump( $url );
 171                 }
 172
 173                 if ( $this->hasOption( 'stub' ) ) {
 174                         $this->input = $this->processFileOpt( $this->getOption( 'stub' ) );
 175                 }
 176
 177                 if ( $this->hasOption( 'maxtime' ) ) {
 178                         $this->maxTimeAllowed = intval( $this->getOption( 'maxtime' ) ) * 60;
 179                 }
 180
 181                 if ( $this->hasOption( 'checkpointfile' ) ) {
 182                         $this->checkpointFiles = $this->getOption( 'checkpointfile' );
 183                 }
 184
 185                 if ( $this->hasOption( 'current' ) ) {
 186                         $this->history = WikiExporter::CURRENT;
 187                 }
 188
 189                 if ( $this->hasOption( 'full' ) ) {
 190                         $this->history = WikiExporter::FULL;
 191                 }
 192
 193                 if ( $this->hasOption( 'spawn' ) ) {
 194                         $this->spawn = true;
 195                         $val = $this->getOption( 'spawn' );
 196                         if ( $val !== 1 ) {
 197                                 $this->php = explode( ',', $val, 2 );
 198                         }
 199                 }
 200         }
 201
 202         /**
 203          * Drop the database connection $this->db and try to get a new one.
 204          *
 205          * This function tries to get a /different/ connection if this is
 206          * possible. Hence, (if this is possible) it switches to a different
 207          * failover upon each call.
 208          *
 209          * This function resets $this->lb and closes all connections on it.
 210          *
 211          * @throws MWException
 212          * @suppress PhanTypeObjectUnsetDeclaredProperty
 213          */
 214         function rotateDb() {
 215                 // Cleaning up old connections
 216                 if ( isset( $this->lb ) ) {
 217                         $this->lb->closeAll();
 218                         unset( $this->lb );
 219                 }
 220
 221                 if ( $this->forcedDb !== null ) {
 222                         $this->db = $this->forcedDb;
 223
 224                         return;
 225                 }
 226
 227                 if ( isset( $this->db ) && $this->db->isOpen() ) {
 228                         throw new MWException( 'DB is set and has not been closed by the Load Balancer' );
 229                 }
 230
 231                 unset( $this->db );
 232
 233                 // Trying to set up new connection.
 234                 // We do /not/ retry upon failure, but delegate to encapsulating logic, to avoid
 235                 // individually retrying at different layers of code.
 236
 237                 try {
 238                         $lbFactory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
 239                         $this->lb = $lbFactory->newMainLB();
 240                 } catch ( Exception $e ) {
 241                         throw new MWException( __METHOD__
 242                                 . " rotating DB failed to obtain new load balancer (" . $e->getMessage() . ")" );
 243                 }
 244
 245                 try {
 246                         $this->db = $this->lb->getMaintenanceConnectionRef( DB_REPLICA, 'dump' );
 247                 } catch ( Exception $e ) {
 248                         throw new MWException( __METHOD__
 249                                 . " rotating DB failed to obtain new database (" . $e->getMessage() . ")" );
 250                 }
 251         }
 252
 253         function initProgress( $history = WikiExporter::FULL ) {
 254                 parent::initProgress();
 255                 $this->timeOfCheckpoint = $this->startTime;
 256         }
 257
 258         function dump( $history, $text = WikiExporter::TEXT ) {
 259                 // Notice messages will foul up your XML output even if they're
 260                 // relatively harmless.
 261                 if ( ini_get( 'display_errors' ) ) {
 262                         ini_set( 'display_errors', 'stderr' );
 263                 }
 264
 265                 $this->initProgress( $this->history );
 266
 267                 // We are trying to get an initial database connection to avoid that the
 268                 // first try of this request's first call to getText fails. However, if
 269                 // obtaining a good DB connection fails it's not a serious issue, as
 270                 // getText does retry upon failure and can start without having a working
 271                 // DB connection.
 272                 try {
 273                         $this->rotateDb();
 274                 } catch ( Exception $e ) {
 275                         // We do not even count this as failure. Just let eventual
 276                         // watchdogs know.
 277                         $this->progress( "Getting initial DB connection failed (" .
 278                                 $e->getMessage() . ")" );
 279                 }
 280
 281                 $this->egress = new ExportProgressFilter( $this->sink, $this );
 282
 283                 // it would be nice to do it in the constructor, oh well. need egress set
 284                 $this->finalOptionCheck();
 285
 286                 // we only want this so we know how to close a stream :-P
 287                 $this->xmlwriterobj = new XmlDumpWriter( XmlDumpWriter::WRITE_CONTENT, $this->schemaVersion );
 288
 289                 $input = fopen( $this->input, "rt" );
 290                 $this->readDump( $input );
 291
 292                 if ( $this->spawnProc ) {
 293                         $this->closeSpawn();
 294                 }
 295
 296                 $this->report( true );
 297         }
 298
 299         function processFileOpt( $opt ) {
 300                 $split = explode( ':', $opt, 2 );
 301                 $val = $split[0];
 302                 $param = '';
 303                 if ( count( $split ) === 2 ) {
 304                         $param = $split[1];
 305                 }
 306                 $fileURIs = explode( ';', $param );
 307                 $newFileURIs = [];
 308                 foreach ( $fileURIs as $URI ) {
 309                         switch ( $val ) {
 310                                 case "file":
 311                                         $newURI = $URI;
 312                                         break;
 313                                 case "gzip":
 314                                         $newURI = "compress.zlib://$URI";
 315                                         break;
 316                                 case "bzip2":
 317                                         $newURI = "compress.bzip2://$URI";
 318                                         break;
 319                                 case "7zip":
 320                                         $newURI = "mediawiki.compress.7z://$URI";
 321                                         break;
 322                                 default:
 323                                         $newURI = $URI;
 324                         }
 325                         $newFileURIs[] = $newURI;
 326                 }
 327                 $val = implode( ';', $newFileURIs );
 328
 329                 return $val;
 330         }
 331
 332         /**
 333          * Overridden to include prefetch ratio if enabled.
 334          */
 335         function showReport() {
 336                 if ( !$this->prefetch ) {
 337                         parent::showReport();
 338
 339                         return;
 340                 }
 341
 342                 if ( $this->reporting ) {
 343                         $now = wfTimestamp( TS_DB );
 344                         $nowts = microtime( true );
 345                         $deltaAll = $nowts - $this->startTime;
 346                         $deltaPart = $nowts - $this->lastTime;
 347                         $this->pageCountPart = $this->pageCount - $this->pageCountLast;
 348                         $this->revCountPart = $this->revCount - $this->revCountLast;
 349
 350                         if ( $deltaAll ) {
 351                                 $portion = $this->revCount / $this->maxCount;
 352                                 $eta = $this->startTime + $deltaAll / $portion;
 353                                 $etats = wfTimestamp( TS_DB, intval( $eta ) );
 354                                 if ( $this->fetchCount ) {
 355                                         $fetchRate = 100.0 * $this->prefetchCount / $this->fetchCount;
 356                                 } else {
 357                                         $fetchRate = '-';
 358                                 }
 359                                 $pageRate = $this->pageCount / $deltaAll;
 360                                 $revRate = $this->revCount / $deltaAll;
 361                         } else {
 362                                 $pageRate = '-';
 363                                 $revRate = '-';
 364                                 $etats = '-';
 365                                 $fetchRate = '-';
 366                         }
 367                         if ( $deltaPart ) {
 368                                 if ( $this->fetchCountLast ) {
 369                                         $fetchRatePart = 100.0 * $this->prefetchCountLast / $this->fetchCountLast;
 370                                 } else {
 371                                         $fetchRatePart = '-';
 372                                 }
 373                                 $pageRatePart = $this->pageCountPart / $deltaPart;
 374                                 $revRatePart = $this->revCountPart / $deltaPart;
 375                         } else {
 376                                 $fetchRatePart = '-';
 377                                 $pageRatePart = '-';
 378                                 $revRatePart = '-';
 379                         }
 380
 381                         $dbDomain = WikiMap::getCurrentWikiDbDomain()->getId();
 382                         $this->progress( sprintf(
 383                                 "%s: %s (ID %d) %d pages (%0.1f|%0.1f/sec all|curr), "
 384                                         . "%d revs (%0.1f|%0.1f/sec all|curr), %0.1f%%|%0.1f%% "
 385                                         . "prefetched (all|curr), ETA %s [max %d]",
 386                                 $now, $dbDomain, $this->ID, $this->pageCount, $pageRate,
 387                                 $pageRatePart, $this->revCount, $revRate, $revRatePart,
 388                                 $fetchRate, $fetchRatePart, $etats, $this->maxCount
 389                         ) );
 390                         $this->lastTime = $nowts;
 391                         $this->revCountLast = $this->revCount;
 392                         $this->prefetchCountLast = $this->prefetchCount;
 393                         $this->fetchCountLast = $this->fetchCount;
 394                 }
 395         }
 396
 397         function setTimeExceeded() {
 398                 $this->timeExceeded = true;
 399         }
 400
 401         function checkIfTimeExceeded() {
 402                 if ( $this->maxTimeAllowed
 403                         && ( $this->lastTime - $this->timeOfCheckpoint > $this->maxTimeAllowed )
 404                 ) {
 405                         return true;
 406                 }
 407
 408                 return false;
 409         }
 410
 411         function finalOptionCheck() {
 412                 if ( ( $this->checkpointFiles && !$this->maxTimeAllowed )
 413                         || ( $this->maxTimeAllowed && !$this->checkpointFiles )
 414                 ) {
 415                         throw new MWException( "Options checkpointfile and maxtime must be specified together.\n" );
 416                 }
 417                 foreach ( $this->checkpointFiles as $checkpointFile ) {
 418                         $count = substr_count( $checkpointFile, "%s" );
 419                         if ( $count != 2 ) {
 420                                 throw new MWException( "Option checkpointfile must contain two '%s' "
 421                                         . "for substitution of first and last pageids, count is $count instead, "
 422                                         . "file is $checkpointFile.\n" );
 423                         }
 424                 }
 425
 426                 if ( $this->checkpointFiles ) {
 427                         $filenameList = (array)$this->egress->getFilenames();
 428                         if ( count( $filenameList ) != count( $this->checkpointFiles ) ) {
 429                                 throw new MWException( "One checkpointfile must be specified "
 430                                         . "for each output option, if maxtime is used.\n" );
 431                         }
 432                 }
 433         }
 434
 435         /**
 436          * @throws MWException Failure to parse XML input
 437          * @param resource $input
 438          * @return bool
 439          */
 440         function readDump( $input ) {
 441                 $this->buffer = "";
 442                 $this->openElement = false;
 443                 $this->atStart = true;
 444                 $this->state = "";
 445                 $this->lastName = "";
 446                 $this->thisPage = 0;
 447                 $this->thisRev = 0;
 448                 $this->thisRevModel = null;
 449                 $this->thisRevFormat = null;
 450
 451                 $parser = xml_parser_create( "UTF-8" );
 452                 xml_parser_set_option( $parser, XML_OPTION_CASE_FOLDING, false );
 453
 454                 xml_set_element_handler(
 455                         $parser,
 456                         [ $this, 'startElement' ],
 457                         [ $this, 'endElement' ]
 458                 );
 459                 xml_set_character_data_handler( $parser, [ $this, 'characterData' ] );
 460
 461                 $offset = 0; // for context extraction on error reporting
 462                 do {
 463                         if ( $this->checkIfTimeExceeded() ) {
 464                                 $this->setTimeExceeded();
 465                         }
 466                         $chunk = fread( $input, $this->bufferSize );
 467                         if ( !xml_parse( $parser, $chunk, feof( $input ) ) ) {
 468                                 wfDebug( "TextDumpPass::readDump encountered XML parsing error\n" );
 469
 470                                 $byte = xml_get_current_byte_index( $parser );
 471                                 $msg = wfMessage( 'xml-error-string',
 472                                         'XML import parse failure',
 473                                         xml_get_current_line_number( $parser ),
 474                                         xml_get_current_column_number( $parser ),
 475                                         $byte . ( is_null( $chunk ) ? null : ( '; "' . substr( $chunk, $byte - $offset, 16 ) . '"' ) ),
 476                                         xml_error_string( xml_get_error_code( $parser ) ) )->escaped();
 477
 478                                 xml_parser_free( $parser );
 479
 480                                 throw new MWException( $msg );
 481                         }
 482                         $offset += strlen( $chunk );
 483                 } while ( $chunk !== false && !feof( $input ) );
 484                 if ( $this->maxTimeAllowed ) {
 485                         $filenameList = (array)$this->egress->getFilenames();
 486                         // we wrote some stuff after last checkpoint that needs renamed
 487                         if ( file_exists( $filenameList[0] ) ) {
 488                                 $newFilenames = [];
 489                                 # we might have just written the header and footer and had no
 490                                 # pages or revisions written... perhaps they were all deleted
 491                                 # there's no pageID 0 so we use that. the caller is responsible
 492                                 # for deciding what to do with a file containing only the
 493                                 # siteinfo information and the mw tags.
 494                                 if ( !$this->firstPageWritten ) {
 495                                         $firstPageID = str_pad( 0, 9, "0", STR_PAD_LEFT );
 496                                         $lastPageID = str_pad( 0, 9, "0", STR_PAD_LEFT );
 497                                 } else {
 498                                         $firstPageID = str_pad( $this->firstPageWritten, 9, "0", STR_PAD_LEFT );
 499                                         $lastPageID = str_pad( $this->lastPageWritten, 9, "0", STR_PAD_LEFT );
 500                                 }
 501
 502                                 $filenameCount = count( $filenameList );
 503                                 for ( $i = 0; $i < $filenameCount; $i++ ) {
 504                                         $checkpointNameFilledIn = sprintf( $this->checkpointFiles[$i], $firstPageID, $lastPageID );
 505                                         $fileinfo = pathinfo( $filenameList[$i] );
 506                                         $newFilenames[] = $fileinfo['dirname'] . '/' . $checkpointNameFilledIn;
 507                                 }
 508                                 $this->egress->closeAndRename( $newFilenames );
 509                         }
 510                 }
 511                 xml_parser_free( $parser );
 512
 513                 return true;
 514         }
 515
 516         /**
 517          * Applies applicable export transformations to $text.
 518          *
 519          * @param string $text
 520          * @param string $model
 521          * @param string|null $format
 522          *
 523          * @return string
 524          */
 525         private function exportTransform( $text, $model, $format = null ) {
 526                 try {
 527                         $handler = ContentHandler::getForModelID( $model );
 528                         $text = $handler->exportTransform( $text, $format );
 529                 }
 530                 catch ( MWException $ex ) {
 531                         $this->progress(
 532                                 "Unable to apply export transformation for content model '$model': " .
 533                                 $ex->getMessage()
 534                         );
 535                 }
 536
 537                 return $text;
 538         }
 539
 540         /**
 541          * Tries to load revision text.
 542          * Export transformations are applied if the content model is given or can be
 543          * determined from the database.
 544          *
 545          * Upon errors, retries (Up to $this->maxFailures tries each call).
 546          * If still no good revision could be found even after this retrying, "" is returned.
 547          * If no good revision text could be returned for
 548          * $this->maxConsecutiveFailedTextRetrievals consecutive calls to getText, MWException
 549          * is thrown.
 550          *
 551          * @param int|string $id Content address, or text row ID.
 552          * @param string|bool|null $model The content model used to determine
 553          *  applicable export transformations.
 554          *  If $model is null, it will be determined from the database.
 555          * @param string|null $format The content format used when applying export transformations.
 556          *
 557          * @throws MWException
 558          * @return string The revision text for $id, or ""
 559          */
 560         function getText( $id, $model = null, $format = null ) {
 561                 global $wgContentHandlerUseDB;
 562
 563                 $prefetchNotTried = true; // Whether or not we already tried to get the text via prefetch.
 564                 $text = false; // The candidate for a good text. false if no proper value.
 565                 $failures = 0; // The number of times, this invocation of getText already failed.
 566
 567                 // The number of times getText failed without yielding a good text in between.
 568                 static $consecutiveFailedTextRetrievals = 0;
 569
 570                 $this->fetchCount++;
 571
 572                 // To allow to simply return on success and do not have to worry about book keeping,
 573                 // we assume, this fetch works (possible after some retries). Nevertheless, we koop
 574                 // the old value, so we can restore it, if problems occur (See after the while loop).
 575                 $oldConsecutiveFailedTextRetrievals = $consecutiveFailedTextRetrievals;
 576                 $consecutiveFailedTextRetrievals = 0;
 577
 578                 if ( $model === null && $wgContentHandlerUseDB ) {
 579                         // TODO: MCR: use content table
 580                         $row = $this->db->selectRow(
 581                                 'revision',
 582                                 [ 'rev_content_model', 'rev_content_format' ],
 583                                 [ 'rev_id' => $this->thisRev ],
 584                                 __METHOD__
 585                         );
 586
 587                         if ( $row ) {
 588                                 $model = $row->rev_content_model;
 589                                 $format = $row->rev_content_format;
 590                         }
 591                 }
 592
 593                 if ( $model === null || $model === '' ) {
 594                         $model = false;
 595                 }
 596
 597                 while ( $failures < $this->maxFailures ) {
 598                         // As soon as we found a good text for the $id, we will return immediately.
 599                         // Hence, if we make it past the try catch block, we know that we did not
 600                         // find a good text.
 601
 602                         try {
 603                                 // Step 1: Get some text (or reuse from previous iteratuon if checking
 604                                 //         for plausibility failed)
 605
 606                                 // Trying to get prefetch, if it has not been tried before
 607                                 if ( $text === false && isset( $this->prefetch ) && $prefetchNotTried ) {
 608                                         $prefetchNotTried = false;
 609                                         $tryIsPrefetch = true;
 610                                         $text = $this->prefetch->prefetch( (int)$this->thisPage, (int)$this->thisRev );
 611
 612                                         if ( $text === null ) {
 613                                                 $text = false;
 614                                         }
 615
 616                                         if ( is_string( $text ) && $model !== false ) {
 617                                                 // Apply export transformation to text coming from an old dump.
 618                                                 // The purpose of this transformation is to convert up from legacy
 619                                                 // formats, which may still be used in the older dump that is used
 620                                                 // for pre-fetching. Applying the transformation again should not
 621                                                 // interfere with content that is already in the correct form.
 622                                                 $text = $this->exportTransform( $text, $model, $format );
 623                                         }
 624                                 }
 625
 626                                 if ( $text === false ) {
 627                                         // Fallback to asking the database
 628                                         $tryIsPrefetch = false;
 629                                         if ( $this->spawn ) {
 630                                                 $text = $this->getTextSpawned( $id );
 631                                         } else {
 632                                                 $text = $this->getTextDb( $id );
 633                                         }
 634
 635                                         if ( $text !== false && $model !== false ) {
 636                                                 // Apply export transformation to text coming from the database.
 637                                                 // Prefetched text should already have transformations applied.
 638                                                 $text = $this->exportTransform( $text, $model, $format );
 639                                         }
 640
 641                                         // No more checks for texts from DB for now.
 642                                         // If we received something that is not false,
 643                                         // We treat it as good text, regardless of whether it actually is or is not
 644                                         if ( $text !== false ) {
 645                                                 return $text;
 646                                         }
 647                                 }
 648
 649                                 if ( $text === false ) {
 650                                         throw new MWException( "Generic error while obtaining text for id " . $id );
 651                                 }
 652
 653                                 // We received a good candidate for the text of $id via some method
 654
 655                                 // Step 2: Checking for plausibility and return the text if it is
 656                                 //         plausible
 657                                 $revID = intval( $this->thisRev );
 658                                 if ( !isset( $this->db ) ) {
 659                                         throw new MWException( "No database available" );
 660                                 }
 661
 662                                 if ( $model !== CONTENT_MODEL_WIKITEXT ) {
 663                                         $revLength = strlen( $text );
 664                                 } else {
 665                                         $revLength = $this->db->selectField( 'revision', 'rev_len', [ 'rev_id' => $revID ] );
 666                                 }
 667
 668                                 if ( strlen( $text ) == $revLength ) {
 669                                         if ( $tryIsPrefetch ) {
 670                                                 $this->prefetchCount++;
 671                                         }
 672
 673                                         return $text;
 674                                 }
 675
 676                                 $text = false;
 677                                 throw new MWException( "Received text is unplausible for id " . $id );
 678                         } catch ( Exception $e ) {
 679                                 $msg = "getting/checking text " . $id . " failed (" . $e->getMessage() . ")";
 680                                 if ( $failures + 1 < $this->maxFailures ) {
 681                                         $msg .= " (Will retry " . ( $this->maxFailures - $failures - 1 ) . " more times)";
 682                                 }
 683                                 $this->progress( $msg );
 684                         }
 685
 686                         // Something went wrong; we did not a text that was plausible :(
 687                         $failures++;
 688
 689                         // A failure in a prefetch hit does not warrant resetting db connection etc.
 690                         if ( !$tryIsPrefetch ) {
 691                                 // After backing off for some time, we try to reboot the whole process as
 692                                 // much as possible to not carry over failures from one part to the other
 693                                 // parts
 694                                 sleep( $this->failureTimeout );
 695                                 try {
 696                                         $this->rotateDb();
 697                                         if ( $this->spawn ) {
 698                                                 $this->closeSpawn();
 699                                                 $this->openSpawn();
 700                                         }
 701                                 } catch ( Exception $e ) {
 702                                         $this->progress( "Rebooting getText infrastructure failed (" . $e->getMessage() . ")" .
 703                                                 " Trying to continue anyways" );
 704                                 }
 705                         }
 706                 }
 707
 708                 // Retirieving a good text for $id failed (at least) maxFailures times.
 709                 // We abort for this $id.
 710
 711                 // Restoring the consecutive failures, and maybe aborting, if the dump
 712                 // is too broken.
 713                 $consecutiveFailedTextRetrievals = $oldConsecutiveFailedTextRetrievals + 1;
 714                 if ( $consecutiveFailedTextRetrievals > $this->maxConsecutiveFailedTextRetrievals ) {
 715                         throw new MWException( "Graceful storage failure" );
 716                 }
 717
 718                 return "";
 719         }
 720
 721         /**
 722          * Loads the serialized content from storage.
 723          *
 724          * @param int|string $id Content address, or text row ID.
 725          * @return bool|string
 726          */
 727         private function getTextDb( $id ) {
 728                 $store = $this->getBlobStore();
 729                 $address = ( is_int( $id ) || strpos( $id, ':' ) === false )
 730                         ? SqlBlobStore::makeAddressFromTextId( (int)$id )
 731                         : $id;
 732
 733                 try {
 734                         $text = $store->getBlob( $address );
 735
 736                         $stripped = str_replace( "\r", "", $text );
 737                         $normalized = MediaWikiServices::getInstance()->getContentLanguage()
 738                                 ->normalize( $stripped );
 739
 740                         return $normalized;
 741                 } catch ( BlobAccessException $ex ) {
 742                         // XXX: log a warning?
 743                         return false;
 744                 }
 745         }
 746
 747         /**
 748          * @param int|string $address Content address, or text row ID.
 749          * @return bool|string
 750          */
 751         private function getTextSpawned( $address ) {
 752                 Wikimedia\suppressWarnings();
 753                 if ( !$this->spawnProc ) {
 754                         // First time?
 755                         $this->openSpawn();
 756                 }
 757                 $text = $this->getTextSpawnedOnce( $address );
 758                 Wikimedia\restoreWarnings();
 759
 760                 return $text;
 761         }
 762
 763         function openSpawn() {
 764                 global $IP;
 765
 766                 $wiki = WikiMap::getWikiIdFromDbDomain( WikiMap::getCurrentWikiDbDomain() );
 767                 if ( count( $this->php ) == 2 ) {
 768                         $mwscriptpath = $this->php[1];
 769                 } else {
 770                         $mwscriptpath = "$IP/../multiversion/MWScript.php";
 771                 }
 772                 if ( file_exists( $mwscriptpath ) ) {
 773                         $cmd = implode( " ",
 774                                 array_map( [ Shell::class, 'escape' ],
 775                                         [
 776                                                 $this->php[0],
 777                                                 $mwscriptpath,
 778                                                 "fetchText.php",
 779                                                 '--wiki', $wiki ] ) );
 780                 } else {
 781                         $cmd = implode( " ",
 782                                 array_map( [ Shell::class, 'escape' ],
 783                                         [
 784                                                 $this->php[0],
 785                                                 "$IP/maintenance/fetchText.php",
 786                                                 '--wiki', $wiki ] ) );
 787                 }
 788                 $spec = [
 789                         0 => [ "pipe", "r" ],
 790                         1 => [ "pipe", "w" ],
 791                         2 => [ "file", "/dev/null", "a" ] ];
 792                 $pipes = [];
 793
 794                 $this->progress( "Spawning database subprocess: $cmd" );
 795                 $this->spawnProc = proc_open( $cmd, $spec, $pipes );
 796                 if ( !$this->spawnProc ) {
 797                         $this->progress( "Subprocess spawn failed." );
 798
 799                         return false;
 800                 }
 801                 list(
 802                         $this->spawnWrite, // -> stdin
 803                         $this->spawnRead, // <- stdout
 804                 ) = $pipes;
 805
 806                 return true;
 807         }
 808
 809         private function closeSpawn() {
 810                 Wikimedia\suppressWarnings();
 811                 if ( $this->spawnRead ) {
 812                         fclose( $this->spawnRead );
 813                 }
 814                 $this->spawnRead = null;
 815                 if ( $this->spawnWrite ) {
 816                         fclose( $this->spawnWrite );
 817                 }
 818                 $this->spawnWrite = null;
 819                 if ( $this->spawnErr ) {
 820                         fclose( $this->spawnErr );
 821                 }
 822                 $this->spawnErr = false;
 823                 if ( $this->spawnProc ) {
 824                         pclose( $this->spawnProc );
 825                 }
 826                 $this->spawnProc = false;
 827                 Wikimedia\restoreWarnings();
 828         }
 829
 830         /**
 831          * @param int|string $address Content address, or text row ID.
 832          * @return bool|string
 833          */
 834         private function getTextSpawnedOnce( $address ) {
 835                 if ( is_int( $address ) || intval( $address ) ) {
 836                         $address = SqlBlobStore::makeAddressFromTextId( (int)$address );
 837                 }
 838
 839                 $ok = fwrite( $this->spawnWrite, "$address\n" );
 840                 // $this->progress( ">> $id" );
 841                 if ( !$ok ) {
 842                         return false;
 843                 }
 844
 845                 $ok = fflush( $this->spawnWrite );
 846                 // $this->progress( ">> [flush]" );
 847                 if ( !$ok ) {
 848                         return false;
 849                 }
 850
 851                 // check that the text address they are sending is the one we asked for
 852                 // this avoids out of sync revision text errors we have encountered in the past
 853                 $newAddress = fgets( $this->spawnRead );
 854                 if ( $newAddress === false ) {
 855                         return false;
 856                 }
 857                 $newAddress = trim( $newAddress );
 858                 if ( strpos( $newAddress, ':' ) === false ) {
 859                         $newAddress = SqlBlobStore::makeAddressFromTextId( intval( $newAddress ) );
 860                 }
 861
 862                 if ( $newAddress !== $address ) {
 863                         return false;
 864                 }
 865
 866                 $len = fgets( $this->spawnRead );
 867                 // $this->progress( "<< " . trim( $len ) );
 868                 if ( $len === false ) {
 869                         return false;
 870                 }
 871
 872                 $nbytes = intval( $len );
 873                 // actual error, not zero-length text
 874                 if ( $nbytes < 0 ) {
 875                         return false;
 876                 }
 877
 878                 $text = "";
 879
 880                 // Subprocess may not send everything at once, we have to loop.
 881                 while ( $nbytes > strlen( $text ) ) {
 882                         $buffer = fread( $this->spawnRead, $nbytes - strlen( $text ) );
 883                         if ( $buffer === false ) {
 884                                 break;
 885                         }
 886                         $text .= $buffer;
 887                 }
 888
 889                 $gotbytes = strlen( $text );
 890                 if ( $gotbytes != $nbytes ) {
 891                         $this->progress( "Expected $nbytes bytes from database subprocess, got $gotbytes " );
 892
 893                         return false;
 894                 }
 895
 896                 // Do normalization in the dump thread...
 897                 $stripped = str_replace( "\r", "", $text );
 898                 $normalized = MediaWikiServices::getInstance()->getContentLanguage()->
 899                         normalize( $stripped );
 900
 901                 return $normalized;
 902         }
 903
 904         function startElement( $parser, $name, $attribs ) {
 905                 $this->checkpointJustWritten = false;
 906
 907                 $this->clearOpenElement( null );
 908                 $this->lastName = $name;
 909
 910                 if ( $name == 'revision' ) {
 911                         $this->state = $name;
 912                         $this->egress->writeOpenPage( null, $this->buffer );
 913                         $this->buffer = "";
 914                 } elseif ( $name == 'page' ) {
 915                         $this->state = $name;
 916                         if ( $this->atStart ) {
 917                                 $this->egress->writeOpenStream( $this->buffer );
 918                                 $this->buffer = "";
 919                                 $this->atStart = false;
 920                         }
 921                 }
 922
 923                 if ( $name == "text" && isset( $attribs['id'] ) ) {
 924                         $id = $attribs['id'];
 925                         $model = trim( $this->thisRevModel );
 926                         $format = trim( $this->thisRevFormat );
 927
 928                         $model = $model === '' ? null : $model;
 929                         $format = $format === '' ? null : $format;
 930
 931                         $text = $this->getText( $id, $model, $format );
 932                         $this->openElement = [ $name, [ 'xml:space' => 'preserve' ] ];
 933                         if ( strlen( $text ) > 0 ) {
 934                                 $this->characterData( $parser, $text );
 935                         }
 936                 } else {
 937                         $this->openElement = [ $name, $attribs ];
 938                 }
 939         }
 940
 941         function endElement( $parser, $name ) {
 942                 $this->checkpointJustWritten = false;
 943
 944                 if ( $this->openElement ) {
 945                         $this->clearOpenElement( "" );
 946                 } else {
 947                         $this->buffer .= "</$name>";
 948                 }
 949
 950                 if ( $name == 'revision' ) {
 951                         $this->egress->writeRevision( null, $this->buffer );
 952                         $this->buffer = "";
 953                         $this->thisRev = "";
 954                         $this->thisRevModel = null;
 955                         $this->thisRevFormat = null;
 956                 } elseif ( $name == 'page' ) {
 957                         if ( !$this->firstPageWritten ) {
 958                                 $this->firstPageWritten = trim( $this->thisPage );
 959                         }
 960                         $this->lastPageWritten = trim( $this->thisPage );
 961                         if ( $this->timeExceeded ) {
 962                                 $this->egress->writeClosePage( $this->buffer );
 963                                 // nasty hack, we can't just write the chardata after the
 964                                 // page tag, it will include leading blanks from the next line
 965                                 $this->egress->sink->write( "\n" );
 966
 967                                 $this->buffer = $this->xmlwriterobj->closeStream();
 968                                 $this->egress->writeCloseStream( $this->buffer );
 969
 970                                 $this->buffer = "";
 971                                 $this->thisPage = "";
 972                                 // this could be more than one file if we had more than one output arg
 973
 974                                 $filenameList = (array)$this->egress->getFilenames();
 975                                 $newFilenames = [];
 976                                 $firstPageID = str_pad( $this->firstPageWritten, 9, "0", STR_PAD_LEFT );
 977                                 $lastPageID = str_pad( $this->lastPageWritten, 9, "0", STR_PAD_LEFT );
 978                                 $filenamesCount = count( $filenameList );
 979                                 for ( $i = 0; $i < $filenamesCount; $i++ ) {
 980                                         $checkpointNameFilledIn = sprintf( $this->checkpointFiles[$i], $firstPageID, $lastPageID );
 981                                         $fileinfo = pathinfo( $filenameList[$i] );
 982                                         $newFilenames[] = $fileinfo['dirname'] . '/' . $checkpointNameFilledIn;
 983                                 }
 984                                 $this->egress->closeRenameAndReopen( $newFilenames );
 985                                 $this->buffer = $this->xmlwriterobj->openStream();
 986                                 $this->timeExceeded = false;
 987                                 $this->timeOfCheckpoint = $this->lastTime;
 988                                 $this->firstPageWritten = false;
 989                                 $this->checkpointJustWritten = true;
 990                         } else {
 991                                 $this->egress->writeClosePage( $this->buffer );
 992                                 $this->buffer = "";
 993                                 $this->thisPage = "";
 994                         }
 995                 } elseif ( $name == 'mediawiki' ) {
 996                         $this->egress->writeCloseStream( $this->buffer );
 997                         $this->buffer = "";
 998                 }
 999         }
1000
1001         function characterData( $parser, $data ) {
1002                 $this->clearOpenElement( null );
1003                 if ( $this->lastName == "id" ) {
1004                         if ( $this->state == "revision" ) {
1005                                 $this->thisRev .= $data;
1006                         } elseif ( $this->state == "page" ) {
1007                                 $this->thisPage .= $data;
1008                         }
1009                 } elseif ( $this->lastName == "model" ) {
1010                         $this->thisRevModel .= $data;
1011                 } elseif ( $this->lastName == "format" ) {
1012                         $this->thisRevFormat .= $data;
1013                 }
1014
1015                 // have to skip the newline left over from closepagetag line of
1016                 // end of checkpoint files. nasty hack!!
1017                 if ( $this->checkpointJustWritten ) {
1018                         if ( $data[0] == "\n" ) {
1019                                 $data = substr( $data, 1 );
1020                         }
1021                         $this->checkpointJustWritten = false;
1022                 }
1023                 $this->buffer .= htmlspecialchars( $data );
1024         }
1025
1026         function clearOpenElement( $style ) {
1027                 if ( $this->openElement ) {
1028                         $this->buffer .= Xml::element( $this->openElement[0], $this->openElement[1], $style );
1029                         $this->openElement = false;
1030                 }
1031         }
1032 }