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