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