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