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