MediaWiki  master
TextPassDumper.php
Go to the documentation of this file.
1 <?php
28 require_once __DIR__ . '/BackupDumper.php';
29 require_once __DIR__ . '/SevenZipStream.php';
30 require_once __DIR__ . '/../../includes/export/WikiExporter.php';
31 
38 
44  public $prefetch = null;
46  private $thisPage;
48  private $thisRev;
50  private $thisRole = null;
51 
52  // when we spend more than maxTimeAllowed seconds on this run, we continue
53  // processing until we write out the next complete page, then save output file(s),
54  // rename it/them and open new one(s)
55  public $maxTimeAllowed = 0; // 0 = no limit
56 
57  protected $input = "php://stdin";
59  protected $fetchCount = 0;
60  protected $prefetchCount = 0;
61  protected $prefetchCountLast = 0;
62  protected $fetchCountLast = 0;
63 
64  protected $maxFailures = 5;
66  protected $failureTimeout = 5; // Seconds to sleep after db failure
67 
68  protected $bufferSize = 524288; // In bytes. Maximum size to read from the stub in on go.
69 
71  protected $php = [ PHP_BINARY ];
72  protected $spawn = false;
73 
77  protected $spawnProc = false;
78 
82  protected $spawnWrite;
83 
87  protected $spawnRead;
88 
92  protected $spawnErr = false;
93 
97  protected $xmlwriterobj = false;
98 
99  protected $timeExceeded = false;
100  protected $firstPageWritten = false;
101  protected $lastPageWritten = false;
102  protected $checkpointJustWritten = false;
104  protected $checkpointFiles = [];
105 
109  protected $db;
110 
114  public function __construct( $args = null ) {
115  parent::__construct();
116 
117  $this->addDescription( <<<TEXT
118 This script postprocesses XML dumps from dumpBackup.php to add
119 page text which was stubbed out (using --stub).
120 
121 XML input is accepted on stdin.
122 XML output is sent to stdout; progress reports are sent to stderr.
123 TEXT
124  );
125  $this->stderr = fopen( "php://stderr", "wt" );
126 
127  $this->addOption( 'stub', 'To load a compressed stub dump instead of stdin. ' .
128  'Specify as --stub=<type>:<file>.', false, true );
129  $this->addOption( 'prefetch', 'Use a prior dump file as a text source, to savepressure on the ' .
130  'database. (Requires the XMLReader extension). Specify as --prefetch=<type>:<file>',
131  false, true );
132  $this->addOption( 'maxtime', 'Write out checkpoint file after this many minutes (writing' .
133  'out complete page, closing xml file properly, and opening new one' .
134  'with header). This option requires the checkpointfile option.', false, true );
135  $this->addOption( 'checkpointfile', 'Use this string for checkpoint filenames,substituting ' .
136  'first pageid written for the first %s (required) and the last pageid written for the ' .
137  'second %s if it exists.', false, true, false, true ); // This can be specified multiple times
138  $this->addOption( 'quiet', 'Don\'t dump status reports to stderr.' );
139  $this->addOption( 'full', 'Dump all revisions of every page' );
140  $this->addOption( 'current', 'Base ETA on number of pages in database instead of all revisions' );
141  $this->addOption( 'spawn', 'Spawn a subprocess for loading text records, optionally specify ' .
142  'php[,mwscript] paths' );
143  $this->addOption( 'buffersize', 'Buffer size in bytes to use for reading the stub. ' .
144  '(Default: 512KB, Minimum: 4KB)', false, true );
145 
146  if ( $args ) {
147  $this->loadWithArgv( $args );
148  $this->processOptions();
149  }
150  }
151 
155  private function getBlobStore() {
156  return MediaWikiServices::getInstance()->getBlobStore();
157  }
158 
159  public function execute() {
160  $this->processOptions();
161  $this->dump( true );
162  }
163 
164  protected function processOptions() {
165  parent::processOptions();
166 
167  if ( $this->hasOption( 'buffersize' ) ) {
168  $this->bufferSize = max( intval( $this->getOption( 'buffersize' ) ), 4 * 1024 );
169  }
170 
171  if ( $this->hasOption( 'prefetch' ) ) {
172  $url = $this->processFileOpt( $this->getOption( 'prefetch' ) );
173  $this->prefetch = new BaseDump( $url );
174  }
175 
176  if ( $this->hasOption( 'stub' ) ) {
177  $this->input = $this->processFileOpt( $this->getOption( 'stub' ) );
178  }
179 
180  if ( $this->hasOption( 'maxtime' ) ) {
181  $this->maxTimeAllowed = intval( $this->getOption( 'maxtime' ) ) * 60;
182  }
183 
184  if ( $this->hasOption( 'checkpointfile' ) ) {
185  $this->checkpointFiles = $this->getOption( 'checkpointfile' );
186  }
187 
188  if ( $this->hasOption( 'current' ) ) {
189  $this->history = WikiExporter::CURRENT;
190  }
191 
192  if ( $this->hasOption( 'full' ) ) {
193  $this->history = WikiExporter::FULL;
194  }
195 
196  if ( $this->hasOption( 'spawn' ) ) {
197  $this->spawn = true;
198  $val = $this->getOption( 'spawn' );
199  if ( $val !== 1 ) {
200  $this->php = explode( ',', $val, 2 );
201  }
202  }
203  }
204 
217  protected function rotateDb() {
218  // Cleaning up old connections
219  if ( isset( $this->lb ) ) {
220  $this->lb->closeAll( __METHOD__ );
221  unset( $this->lb );
222  }
223 
224  if ( $this->forcedDb !== null ) {
225  $this->db = $this->forcedDb;
226 
227  return;
228  }
229 
230  if ( isset( $this->db ) && $this->db->isOpen() ) {
231  throw new MWException( 'DB is set and has not been closed by the Load Balancer' );
232  }
233 
234  unset( $this->db );
235 
236  // Trying to set up new connection.
237  // We do /not/ retry upon failure, but delegate to encapsulating logic, to avoid
238  // individually retrying at different layers of code.
239 
240  try {
241  $lbFactory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
242  $this->lb = $lbFactory->newMainLB();
243  } catch ( Exception $e ) {
244  throw new MWException( __METHOD__
245  . " rotating DB failed to obtain new load balancer (" . $e->getMessage() . ")" );
246  }
247 
248  try {
249  $this->db = $this->lb->getMaintenanceConnectionRef( DB_REPLICA, 'dump' );
250  } catch ( Exception $e ) {
251  throw new MWException( __METHOD__
252  . " rotating DB failed to obtain new database (" . $e->getMessage() . ")" );
253  }
254  }
255 
256  public function initProgress( $history = WikiExporter::FULL ) {
257  parent::initProgress();
258  $this->timeOfCheckpoint = $this->startTime;
259  }
260 
261  public function dump( $history, $text = WikiExporter::TEXT ) {
262  // Notice messages will foul up your XML output even if they're
263  // relatively harmless.
264  if ( ini_get( 'display_errors' ) ) {
265  ini_set( 'display_errors', 'stderr' );
266  }
267 
268  $this->initProgress( $this->history );
269 
270  // We are trying to get an initial database connection to avoid that the
271  // first try of this request's first call to getText fails. However, if
272  // obtaining a good DB connection fails it's not a serious issue, as
273  // getText does retry upon failure and can start without having a working
274  // DB connection.
275  try {
276  $this->rotateDb();
277  } catch ( Exception $e ) {
278  // We do not even count this as failure. Just let eventual
279  // watchdogs know.
280  $this->progress( "Getting initial DB connection failed (" .
281  $e->getMessage() . ")" );
282  }
283 
284  $this->egress = new ExportProgressFilter( $this->sink, $this );
285 
286  // it would be nice to do it in the constructor, oh well. need egress set
287  $this->finalOptionCheck();
288 
289  // we only want this so we know how to close a stream :-P
290  $this->xmlwriterobj = new XmlDumpWriter( XmlDumpWriter::WRITE_CONTENT, $this->schemaVersion );
291 
292  $input = fopen( $this->input, "rt" );
293  $this->readDump( $input );
294 
295  if ( $this->spawnProc ) {
296  $this->closeSpawn();
297  }
298 
299  $this->report( true );
300  }
301 
302  protected function processFileOpt( $opt ) {
303  $split = explode( ':', $opt, 2 );
304  $val = $split[0];
305  $param = '';
306  if ( count( $split ) === 2 ) {
307  $param = $split[1];
308  }
309  $fileURIs = explode( ';', $param );
310  $newFileURIs = [];
311  foreach ( $fileURIs as $URI ) {
312  switch ( $val ) {
313  case "file":
314  $newURI = $URI;
315  break;
316  case "gzip":
317  $newURI = "compress.zlib://$URI";
318  break;
319  case "bzip2":
320  $newURI = "compress.bzip2://$URI";
321  break;
322  case "7zip":
323  $newURI = "mediawiki.compress.7z://$URI";
324  break;
325  default:
326  $newURI = $URI;
327  }
328  $newFileURIs[] = $newURI;
329  }
330  $val = implode( ';', $newFileURIs );
331 
332  return $val;
333  }
334 
338  public function showReport() {
339  if ( !$this->prefetch ) {
340  parent::showReport();
341 
342  return;
343  }
344 
345  if ( $this->reporting ) {
346  $now = wfTimestamp( TS_DB );
347  $nowts = microtime( true );
348  $deltaAll = $nowts - $this->startTime;
349  $deltaPart = $nowts - $this->lastTime;
350  $this->pageCountPart = $this->pageCount - $this->pageCountLast;
351  $this->revCountPart = $this->revCount - $this->revCountLast;
352 
353  if ( $deltaAll ) {
354  $portion = $this->revCount / $this->maxCount;
355  $eta = $this->startTime + $deltaAll / $portion;
356  $etats = wfTimestamp( TS_DB, intval( $eta ) );
357  if ( $this->fetchCount ) {
358  $fetchRate = 100.0 * $this->prefetchCount / $this->fetchCount;
359  } else {
360  $fetchRate = '-';
361  }
362  $pageRate = $this->pageCount / $deltaAll;
363  $revRate = $this->revCount / $deltaAll;
364  } else {
365  $pageRate = '-';
366  $revRate = '-';
367  $etats = '-';
368  $fetchRate = '-';
369  }
370  if ( $deltaPart ) {
371  if ( $this->fetchCountLast ) {
372  $fetchRatePart = 100.0 * $this->prefetchCountLast / $this->fetchCountLast;
373  } else {
374  $fetchRatePart = '-';
375  }
376  $pageRatePart = $this->pageCountPart / $deltaPart;
377  $revRatePart = $this->revCountPart / $deltaPart;
378  } else {
379  $fetchRatePart = '-';
380  $pageRatePart = '-';
381  $revRatePart = '-';
382  }
383 
384  $dbDomain = WikiMap::getCurrentWikiDbDomain()->getId();
385  $this->progress( sprintf(
386  "%s: %s (ID %d) %d pages (%0.1f|%0.1f/sec all|curr), "
387  . "%d revs (%0.1f|%0.1f/sec all|curr), %0.1f%%|%0.1f%% "
388  . "prefetched (all|curr), ETA %s [max %d]",
389  $now, $dbDomain, $this->ID, $this->pageCount, $pageRate,
390  $pageRatePart, $this->revCount, $revRate, $revRatePart,
391  $fetchRate, $fetchRatePart, $etats, $this->maxCount
392  ) );
393  $this->lastTime = $nowts;
394  $this->revCountLast = $this->revCount;
395  $this->prefetchCountLast = $this->prefetchCount;
396  $this->fetchCountLast = $this->fetchCount;
397  }
398  }
399 
400  private function setTimeExceeded() {
401  $this->timeExceeded = true;
402  }
403 
404  private function checkIfTimeExceeded() {
405  if ( $this->maxTimeAllowed
406  && ( $this->lastTime - $this->timeOfCheckpoint > $this->maxTimeAllowed )
407  ) {
408  return true;
409  }
410 
411  return false;
412  }
413 
414  private function finalOptionCheck() {
415  if ( ( $this->checkpointFiles && !$this->maxTimeAllowed )
416  || ( $this->maxTimeAllowed && !$this->checkpointFiles )
417  ) {
418  throw new MWException( "Options checkpointfile and maxtime must be specified together.\n" );
419  }
420  foreach ( $this->checkpointFiles as $checkpointFile ) {
421  $count = substr_count( $checkpointFile, "%s" );
422  if ( $count != 2 ) {
423  throw new MWException( "Option checkpointfile must contain two '%s' "
424  . "for substitution of first and last pageids, count is $count instead, "
425  . "file is $checkpointFile.\n" );
426  }
427  }
428 
429  if ( $this->checkpointFiles ) {
430  $filenameList = (array)$this->egress->getFilenames();
431  if ( count( $filenameList ) != count( $this->checkpointFiles ) ) {
432  throw new MWException( "One checkpointfile must be specified "
433  . "for each output option, if maxtime is used.\n" );
434  }
435  }
436  }
437 
443  protected function readDump( $input ) {
444  $this->buffer = "";
445  $this->openElement = false;
446  $this->atStart = true;
447  $this->state = "";
448  $this->lastName = "";
449  $this->thisPage = 0;
450  $this->thisRev = 0;
451  $this->thisRole = null;
452  $this->thisRevModel = null;
453  $this->thisRevFormat = null;
454 
455  $parser = xml_parser_create( "UTF-8" );
456  xml_parser_set_option( $parser, XML_OPTION_CASE_FOLDING, false );
457 
458  xml_set_element_handler(
459  $parser,
460  [ $this, 'startElement' ],
461  [ $this, 'endElement' ]
462  );
463  xml_set_character_data_handler( $parser, [ $this, 'characterData' ] );
464 
465  $offset = 0; // for context extraction on error reporting
466  do {
467  if ( $this->checkIfTimeExceeded() ) {
468  $this->setTimeExceeded();
469  }
470  $chunk = fread( $input, $this->bufferSize );
471  if ( !xml_parse( $parser, $chunk, feof( $input ) ) ) {
472  wfDebug( "TextDumpPass::readDump encountered XML parsing error" );
473 
474  $byte = xml_get_current_byte_index( $parser );
475  $msg = wfMessage( 'xml-error-string',
476  'XML import parse failure',
477  xml_get_current_line_number( $parser ),
478  xml_get_current_column_number( $parser ),
479  $byte . ( $chunk === null ? null : ( '; "' . substr( $chunk, $byte - $offset, 16 ) . '"' ) ),
480  xml_error_string( xml_get_error_code( $parser ) ) )->escaped();
481 
482  xml_parser_free( $parser );
483 
484  throw new MWException( $msg );
485  }
486  $offset += strlen( $chunk );
487  } while ( $chunk !== false && !feof( $input ) );
488  if ( $this->maxTimeAllowed ) {
489  $filenameList = (array)$this->egress->getFilenames();
490  // we wrote some stuff after last checkpoint that needs renamed
491  if ( file_exists( $filenameList[0] ) ) {
492  $newFilenames = [];
493  # we might have just written the header and footer and had no
494  # pages or revisions written... perhaps they were all deleted
495  # there's no pageID 0 so we use that. the caller is responsible
496  # for deciding what to do with a file containing only the
497  # siteinfo information and the mw tags.
498  if ( !$this->firstPageWritten ) {
499  $firstPageID = str_pad( 0, 9, "0", STR_PAD_LEFT );
500  $lastPageID = str_pad( 0, 9, "0", STR_PAD_LEFT );
501  } else {
502  $firstPageID = str_pad( $this->firstPageWritten, 9, "0", STR_PAD_LEFT );
503  $lastPageID = str_pad( $this->lastPageWritten, 9, "0", STR_PAD_LEFT );
504  }
505 
506  $filenameCount = count( $filenameList );
507  for ( $i = 0; $i < $filenameCount; $i++ ) {
508  $checkpointNameFilledIn = sprintf( $this->checkpointFiles[$i], $firstPageID, $lastPageID );
509  $fileinfo = pathinfo( $filenameList[$i] );
510  $newFilenames[] = $fileinfo['dirname'] . '/' . $checkpointNameFilledIn;
511  }
512  $this->egress->closeAndRename( $newFilenames );
513  }
514  }
515  xml_parser_free( $parser );
516 
517  return true;
518  }
519 
529  private function exportTransform( $text, $model, $format = null ) {
530  try {
531  $text = MediaWikiServices::getInstance()
532  ->getContentHandlerFactory()
533  ->getContentHandler( $model )
534  ->exportTransform( $text, $format );
535  }
536  catch ( MWException $ex ) {
537  wfWarn( "Unable to apply export transformation for content model '$model': " .
538  $ex->getMessage() );
539 
540  $this->progress(
541  "Unable to apply export transformation for content model '$model': " .
542  $ex->getMessage()
543  );
544  }
545 
546  return $text;
547  }
548 
569  protected function getText( $id, $model = null, $format = null, $expSize = null ) {
570  if ( !$this->isValidTextId( $id ) ) {
571  $msg = "Skipping bad text id " . $id . " of revision " . $this->thisRev;
572  $this->progress( $msg );
573  return '';
574  }
575 
576  $model = $model ?: null;
577  $prefetchNotTried = true; // Whether or not we already tried to get the text via prefetch.
578  $text = false; // The candidate for a good text. false if no proper value.
579  $failures = 0; // The number of times, this invocation of getText already failed.
580 
581  // The number of times getText failed without yielding a good text in between.
582  static $consecutiveFailedTextRetrievals = 0;
583 
584  $this->fetchCount++;
585 
586  // To allow to simply return on success and do not have to worry about book keeping,
587  // we assume, this fetch works (possible after some retries). Nevertheless, we koop
588  // the old value, so we can restore it, if problems occur (See after the while loop).
589  $oldConsecutiveFailedTextRetrievals = $consecutiveFailedTextRetrievals;
590  $consecutiveFailedTextRetrievals = 0;
591 
592  while ( $failures < $this->maxFailures ) {
593  // As soon as we found a good text for the $id, we will return immediately.
594  // Hence, if we make it past the try catch block, we know that we did not
595  // find a good text.
596 
597  try {
598  // Step 1: Get some text (or reuse from previous iteratuon if checking
599  // for plausibility failed)
600 
601  // Trying to get prefetch, if it has not been tried before
602  // @phan-suppress-next-line PhanSuspiciousValueComparisonInLoop
603  if ( $text === false && isset( $this->prefetch ) && $prefetchNotTried ) {
604  $prefetchNotTried = false;
605  $tryIsPrefetch = true;
606  $text = $this->prefetch->prefetch(
607  (int)$this->thisPage,
608  (int)$this->thisRev,
609  trim( $this->thisRole )
610  );
611 
612  if ( $text === null ) {
613  $text = false;
614  }
615 
616  if ( is_string( $text ) && $model !== null ) {
617  // Apply export transformation to text coming from an old dump.
618  // The purpose of this transformation is to convert up from legacy
619  // formats, which may still be used in the older dump that is used
620  // for pre-fetching. Applying the transformation again should not
621  // interfere with content that is already in the correct form.
622  $text = $this->exportTransform( $text, $model, $format );
623  }
624  }
625 
626  if ( $text === false ) {
627  // Fallback to asking the database
628  $tryIsPrefetch = false;
629  if ( $this->spawn ) {
630  $text = $this->getTextSpawned( $id );
631  } else {
632  $text = $this->getTextDb( $id );
633  }
634 
635  if ( $text !== false && $model !== null ) {
636  // Apply export transformation to text coming from the database.
637  // Prefetched text should already have transformations applied.
638  $text = $this->exportTransform( $text, $model, $format );
639  }
640 
641  // No more checks for texts from DB for now.
642  // If we received something that is not false,
643  // We treat it as good text, regardless of whether it actually is or is not
644  if ( $text !== false ) {
645  return $text;
646  }
647  }
648 
649  if ( $text === false ) {
650  throw new MWException( "Generic error while obtaining text for id " . $id );
651  }
652 
653  // We received a good candidate for the text of $id via some method
654 
655  // Step 2: Checking for plausibility and return the text if it is
656  // plausible
657 
658  if ( $expSize === null || strlen( $text ) == $expSize ) {
659  if ( $tryIsPrefetch ) {
660  $this->prefetchCount++;
661  }
662 
663  return $text;
664  }
665 
666  $text = false;
667  throw new MWException( "Received text is unplausible for id " . $id );
668  } catch ( Exception $e ) {
669  $msg = "getting/checking text " . $id . " failed (" . $e->getMessage()
670  . ") for revision " . $this->thisRev;
671  if ( $failures + 1 < $this->maxFailures ) {
672  $msg .= " (Will retry " . ( $this->maxFailures - $failures - 1 ) . " more times)";
673  }
674  $this->progress( $msg );
675  }
676 
677  // Something went wrong; we did not a text that was plausible :(
678  $failures++;
679 
680  // A failure in a prefetch hit does not warrant resetting db connection etc.
681  if ( !$tryIsPrefetch ) {
682  // After backing off for some time, we try to reboot the whole process as
683  // much as possible to not carry over failures from one part to the other
684  // parts
685  sleep( $this->failureTimeout );
686  try {
687  $this->rotateDb();
688  if ( $this->spawn ) {
689  $this->closeSpawn();
690  $this->openSpawn();
691  }
692  } catch ( Exception $e ) {
693  $this->progress( "Rebooting getText infrastructure failed (" . $e->getMessage() . ")" .
694  " Trying to continue anyways" );
695  }
696  }
697  }
698 
699  // Retirieving a good text for $id failed (at least) maxFailures times.
700  // We abort for this $id.
701 
702  // Restoring the consecutive failures, and maybe aborting, if the dump
703  // is too broken.
704  $consecutiveFailedTextRetrievals = $oldConsecutiveFailedTextRetrievals + 1;
705  if ( $consecutiveFailedTextRetrievals > $this->maxConsecutiveFailedTextRetrievals ) {
706  throw new MWException( "Graceful storage failure" );
707  }
708 
709  return "";
710  }
711 
718  private function getTextDb( $id ) {
719  $store = $this->getBlobStore();
720  $address = ( is_int( $id ) || strpos( $id, ':' ) === false )
721  ? SqlBlobStore::makeAddressFromTextId( (int)$id )
722  : $id;
723 
724  try {
725  $text = $store->getBlob( $address );
726 
727  $stripped = str_replace( "\r", "", $text );
728  $normalized = MediaWikiServices::getInstance()->getContentLanguage()
729  ->normalize( $stripped );
730 
731  return $normalized;
732  } catch ( BlobAccessException $ex ) {
733  // XXX: log a warning?
734  return false;
735  }
736  }
737 
742  private function getTextSpawned( $address ) {
743  Wikimedia\suppressWarnings();
744  if ( !$this->spawnProc ) {
745  // First time?
746  $this->openSpawn();
747  }
748  $text = $this->getTextSpawnedOnce( $address );
749  Wikimedia\restoreWarnings();
750 
751  return $text;
752  }
753 
754  protected function openSpawn() {
755  global $IP;
756 
757  $wiki = WikiMap::getCurrentWikiId();
758  if ( count( $this->php ) == 2 ) {
759  $mwscriptpath = $this->php[1];
760  } else {
761  $mwscriptpath = "$IP/../multiversion/MWScript.php";
762  }
763  if ( file_exists( $mwscriptpath ) ) {
764  $cmd = implode( " ",
765  array_map( [ Shell::class, 'escape' ],
766  [
767  $this->php[0],
768  $mwscriptpath,
769  "fetchText.php",
770  '--wiki', $wiki ] ) );
771  } else {
772  $cmd = implode( " ",
773  array_map( [ Shell::class, 'escape' ],
774  [
775  $this->php[0],
776  "$IP/maintenance/fetchText.php",
777  '--wiki', $wiki ] ) );
778  }
779  $spec = [
780  0 => [ "pipe", "r" ],
781  1 => [ "pipe", "w" ],
782  2 => [ "file", "/dev/null", "a" ] ];
783  $pipes = [];
784 
785  $this->progress( "Spawning database subprocess: $cmd" );
786  $this->spawnProc = proc_open( $cmd, $spec, $pipes );
787  if ( !$this->spawnProc ) {
788  $this->progress( "Subprocess spawn failed." );
789 
790  return false;
791  }
792  [
793  $this->spawnWrite, // -> stdin
794  $this->spawnRead, // <- stdout
795  ] = $pipes;
796 
797  return true;
798  }
799 
800  private function closeSpawn() {
801  Wikimedia\suppressWarnings();
802  if ( $this->spawnRead ) {
803  fclose( $this->spawnRead );
804  }
805  $this->spawnRead = null;
806  if ( $this->spawnWrite ) {
807  fclose( $this->spawnWrite );
808  }
809  $this->spawnWrite = null;
810  if ( $this->spawnErr ) {
811  fclose( $this->spawnErr );
812  }
813  $this->spawnErr = false;
814  if ( $this->spawnProc ) {
815  pclose( $this->spawnProc );
816  }
817  $this->spawnProc = false;
818  Wikimedia\restoreWarnings();
819  }
820 
825  private function getTextSpawnedOnce( $address ) {
826  if ( is_int( $address ) || intval( $address ) ) {
827  $address = SqlBlobStore::makeAddressFromTextId( (int)$address );
828  }
829 
830  $ok = fwrite( $this->spawnWrite, "$address\n" );
831  // $this->progress( ">> $id" );
832  if ( !$ok ) {
833  return false;
834  }
835 
836  $ok = fflush( $this->spawnWrite );
837  // $this->progress( ">> [flush]" );
838  if ( !$ok ) {
839  return false;
840  }
841 
842  // check that the text address they are sending is the one we asked for
843  // this avoids out of sync revision text errors we have encountered in the past
844  $newAddress = fgets( $this->spawnRead );
845  if ( $newAddress === false ) {
846  return false;
847  }
848  $newAddress = trim( $newAddress );
849  if ( strpos( $newAddress, ':' ) === false ) {
850  $newAddress = SqlBlobStore::makeAddressFromTextId( intval( $newAddress ) );
851  }
852 
853  if ( $newAddress !== $address ) {
854  return false;
855  }
856 
857  $len = fgets( $this->spawnRead );
858  // $this->progress( "<< " . trim( $len ) );
859  if ( $len === false ) {
860  return false;
861  }
862 
863  $nbytes = intval( $len );
864  // actual error, not zero-length text
865  if ( $nbytes < 0 ) {
866  return false;
867  }
868 
869  $text = "";
870 
871  // Subprocess may not send everything at once, we have to loop.
872  while ( $nbytes > strlen( $text ) ) {
873  $buffer = fread( $this->spawnRead, $nbytes - strlen( $text ) );
874  if ( $buffer === false ) {
875  break;
876  }
877  $text .= $buffer;
878  }
879 
880  $gotbytes = strlen( $text );
881  if ( $gotbytes != $nbytes ) {
882  $this->progress( "Expected $nbytes bytes from database subprocess, got $gotbytes " );
883 
884  return false;
885  }
886 
887  // Do normalization in the dump thread...
888  $stripped = str_replace( "\r", "", $text );
889  $normalized = MediaWikiServices::getInstance()->getContentLanguage()->
890  normalize( $stripped );
891 
892  return $normalized;
893  }
894 
895  protected function startElement( $parser, $name, $attribs ) {
896  $this->checkpointJustWritten = false;
897 
898  $this->clearOpenElement( null );
899  $this->lastName = $name;
900 
901  if ( $name == 'revision' ) {
902  $this->state = $name;
903  $this->egress->writeOpenPage( null, $this->buffer );
904  $this->buffer = "";
905  } elseif ( $name == 'page' ) {
906  $this->state = $name;
907  if ( $this->atStart ) {
908  $this->egress->writeOpenStream( $this->buffer );
909  $this->buffer = "";
910  $this->atStart = false;
911  }
912  } elseif ( $name === 'mediawiki' ) {
913  if ( isset( $attribs['version'] ) ) {
914  if ( $attribs['version'] !== $this->schemaVersion ) {
915  throw new MWException( 'Mismatching schema version. '
916  . 'Use the --schema-version option to set the output schema version to '
917  . 'the version declared by the stub file, namely ' . $attribs['version'] );
918  }
919  }
920  }
921 
922  if ( $name == "text" && ( isset( $attribs['id'] ) || isset( $attribs['location'] ) ) ) {
923  $id = $attribs['location'] ?? $attribs['id'];
924  $model = trim( $this->thisRevModel );
925  $format = trim( $this->thisRevFormat );
926 
927  $model = $model === '' ? null : $model;
928  $format = $format === '' ? null : $format;
929  $expSize = !empty( $attribs['bytes'] ) && $model === CONTENT_MODEL_WIKITEXT
930  ? (int)$attribs['bytes'] : null;
931 
932  $text = $this->getText( $id, $model, $format, $expSize );
933 
934  unset( $attribs['id'] );
935  unset( $attribs['location'] );
936  if ( strlen( $text ) > 0 ) {
937  $attribs['xml:space'] = 'preserve';
938  }
939 
940  $this->openElement = [ $name, $attribs ];
941  if ( strlen( $text ) > 0 ) {
942  $this->characterData( $parser, $text );
943  }
944  } else {
945  $this->openElement = [ $name, $attribs ];
946  }
947  }
948 
949  protected function endElement( $parser, $name ) {
950  $this->checkpointJustWritten = false;
951 
952  if ( $this->openElement ) {
953  $this->clearOpenElement( "" );
954  } else {
955  $this->buffer .= "</$name>";
956  }
957 
958  if ( $name == 'revision' ) {
959  $this->egress->writeRevision( null, $this->buffer );
960  $this->buffer = "";
961  $this->thisRev = "";
962  $this->thisRole = null;
963  $this->thisRevModel = null;
964  $this->thisRevFormat = null;
965  } elseif ( $name == 'page' ) {
966  if ( !$this->firstPageWritten ) {
967  $this->firstPageWritten = trim( $this->thisPage );
968  }
969  $this->lastPageWritten = trim( $this->thisPage );
970  if ( $this->timeExceeded ) {
971  $this->egress->writeClosePage( $this->buffer );
972  // nasty hack, we can't just write the chardata after the
973  // page tag, it will include leading blanks from the next line
974  $this->egress->sink->write( "\n" );
975 
976  $this->buffer = $this->xmlwriterobj->closeStream();
977  $this->egress->writeCloseStream( $this->buffer );
978 
979  $this->buffer = "";
980  $this->thisPage = "";
981  // this could be more than one file if we had more than one output arg
982 
983  $filenameList = (array)$this->egress->getFilenames();
984  $newFilenames = [];
985  $firstPageID = str_pad( $this->firstPageWritten, 9, "0", STR_PAD_LEFT );
986  $lastPageID = str_pad( $this->lastPageWritten, 9, "0", STR_PAD_LEFT );
987  $filenamesCount = count( $filenameList );
988  for ( $i = 0; $i < $filenamesCount; $i++ ) {
989  $checkpointNameFilledIn = sprintf( $this->checkpointFiles[$i], $firstPageID, $lastPageID );
990  $fileinfo = pathinfo( $filenameList[$i] );
991  $newFilenames[] = $fileinfo['dirname'] . '/' . $checkpointNameFilledIn;
992  }
993  $this->egress->closeRenameAndReopen( $newFilenames );
994  $this->buffer = $this->xmlwriterobj->openStream();
995  $this->timeExceeded = false;
996  $this->timeOfCheckpoint = $this->lastTime;
997  $this->firstPageWritten = false;
998  $this->checkpointJustWritten = true;
999  } else {
1000  $this->egress->writeClosePage( $this->buffer );
1001  $this->buffer = "";
1002  $this->thisPage = "";
1003  }
1004  } elseif ( $name == 'mediawiki' ) {
1005  $this->egress->writeCloseStream( $this->buffer );
1006  $this->buffer = "";
1007  }
1008  }
1009 
1010  protected function characterData( $parser, $data ) {
1011  $this->clearOpenElement( null );
1012  if ( $this->lastName == "id" ) {
1013  if ( $this->state == "revision" ) {
1014  $this->thisRev .= $data;
1015  $this->thisRole = SlotRecord::MAIN;
1016  } elseif ( $this->state == "page" ) {
1017  $this->thisPage .= $data;
1018  }
1019  } elseif ( $this->lastName == "model" ) {
1020  $this->thisRevModel .= $data;
1021  } elseif ( $this->lastName == "format" ) {
1022  $this->thisRevFormat .= $data;
1023  } elseif ( $this->lastName == "content" ) {
1024  $this->thisRole = "";
1025  $this->thisRevModel = "";
1026  $this->thisRevFormat = "";
1027  } elseif ( $this->lastName == "role" ) {
1028  $this->thisRole .= $data;
1029  }
1030 
1031  // have to skip the newline left over from closepagetag line of
1032  // end of checkpoint files. nasty hack!!
1033  if ( $this->checkpointJustWritten ) {
1034  if ( $data[0] == "\n" ) {
1035  $data = substr( $data, 1 );
1036  }
1037  $this->checkpointJustWritten = false;
1038  }
1039  $this->buffer .= htmlspecialchars( $data );
1040  }
1041 
1042  protected function clearOpenElement( $style ) {
1043  if ( $this->openElement ) {
1044  $this->buffer .= Xml::element( $this->openElement[0], $this->openElement[1], $style );
1045  $this->openElement = false;
1046  }
1047  }
1048 
1049  private function isValidTextId( $id ) {
1050  if ( preg_match( '/:/', $id ) ) {
1051  return $id !== 'tt:0';
1052  } elseif ( preg_match( '/^\d+$/', $id ) ) {
1053  return intval( $id ) > 0;
1054  }
1055 
1056  return false;
1057  }
1058 
1059 }
TextPassDumper\clearOpenElement
clearOpenElement( $style)
Definition: TextPassDumper.php:1042
TextPassDumper\$bufferSize
$bufferSize
Definition: TextPassDumper.php:68
MediaWiki\Shell\Shell
Executes shell commands.
Definition: Shell.php:44
BaseDump
Readahead helper for making large MediaWiki data dumps; reads in a previous XML dump to sequentially ...
Definition: BaseDump.php:44
TextPassDumper\execute
execute()
Do the actual work.
Definition: TextPassDumper.php:159
MediaWiki\Storage\BlobAccessException
Exception representing a failure to access a data blob.
Definition: BlobAccessException.php:33
BackupDumper\revCount
revCount()
Definition: BackupDumper.php:405
WikiMap\getCurrentWikiDbDomain
static getCurrentWikiDbDomain()
Definition: WikiMap.php:293
TextPassDumper\readDump
readDump( $input)
Definition: TextPassDumper.php:443
BackupDumper\$startTime
int $startTime
Definition: BackupDumper.php:71
TextPassDumper\getTextSpawned
getTextSpawned( $address)
Definition: TextPassDumper.php:742
TextPassDumper\startElement
startElement( $parser, $name, $attribs)
Definition: TextPassDumper.php:895
TextPassDumper\__construct
__construct( $args=null)
Definition: TextPassDumper.php:114
MediaWiki\MediaWikiServices
MediaWikiServices is the service locator for the application scope of MediaWiki.
Definition: MediaWikiServices.php:154
TextPassDumper\$history
$history
Definition: TextPassDumper.php:58
TextPassDumper\getBlobStore
getBlobStore()
Definition: TextPassDumper.php:155
WikiExporter\CURRENT
const CURRENT
Definition: WikiExporter.php:54
TextPassDumper\processFileOpt
processFileOpt( $opt)
Definition: TextPassDumper.php:302
MediaWiki\Storage\SqlBlobStore
Service for storing and loading Content objects.
Definition: SqlBlobStore.php:51
Maintenance\addDescription
addDescription( $text)
Set the description text.
Definition: Maintenance.php:327
TextPassDumper\$spawnWrite
resource $spawnWrite
Definition: TextPassDumper.php:82
TextPassDumper\showReport
showReport()
Overridden to include prefetch ratio if enabled.
Definition: TextPassDumper.php:338
TextPassDumper\$db
IMaintainableDatabase $db
Definition: TextPassDumper.php:109
wfTimestamp
wfTimestamp( $outputtype=TS_UNIX, $ts=0)
Get a timestamp string in one of various formats.
Definition: GlobalFunctions.php:1808
TextPassDumper\$spawnProc
bool resource $spawnProc
Definition: TextPassDumper.php:77
BackupDumper\$revCountLast
$revCountLast
Definition: BackupDumper.php:63
TextPassDumper\$maxTimeAllowed
$maxTimeAllowed
Definition: TextPassDumper.php:55
TextPassDumper\$spawnErr
bool resource $spawnErr
Definition: TextPassDumper.php:92
wfMessage
wfMessage( $key,... $params)
This is the function for getting translated interface messages.
Definition: GlobalFunctions.php:1219
BackupDumper\$pageCountLast
$pageCountLast
Definition: BackupDumper.php:62
TextPassDumper\setTimeExceeded
setTimeExceeded()
Definition: TextPassDumper.php:400
TextPassDumper\rotateDb
rotateDb()
Drop the database connection $this->db and try to get a new one.
Definition: TextPassDumper.php:217
BackupDumper\$revCount
$revCount
Definition: BackupDumper.php:57
TextPassDumper\$xmlwriterobj
bool XmlDumpWriter $xmlwriterobj
Definition: TextPassDumper.php:97
CONTENT_MODEL_WIKITEXT
const CONTENT_MODEL_WIKITEXT
Definition: Defines.php:224
WikiMap\getCurrentWikiId
static getCurrentWikiId()
Definition: WikiMap.php:303
TextPassDumper\$thisRev
string bool $thisRev
Definition: TextPassDumper.php:48
TextPassDumper\exportTransform
exportTransform( $text, $model, $format=null)
Applies applicable export transformations to $text.
Definition: TextPassDumper.php:529
TextPassDumper\finalOptionCheck
finalOptionCheck()
Definition: TextPassDumper.php:414
TextPassDumper\$spawnRead
resource $spawnRead
Definition: TextPassDumper.php:87
BackupDumper\$buffer
string $buffer
Definition: BackupDumper.php:83
Maintenance\loadWithArgv
loadWithArgv( $argv)
Load params and arguments from a given array of command-line arguments.
Definition: Maintenance.php:854
TextPassDumper\isValidTextId
isValidTextId( $id)
Definition: TextPassDumper.php:1049
TextPassDumper\processOptions
processOptions()
Processes arguments and sets $this->$sink accordingly.
Definition: TextPassDumper.php:164
TextPassDumper\$checkpointFiles
string[] $checkpointFiles
Definition: TextPassDumper.php:104
TextPassDumper\$maxConsecutiveFailedTextRetrievals
$maxConsecutiveFailedTextRetrievals
Definition: TextPassDumper.php:65
MWException
MediaWiki exception.
Definition: MWException.php:29
TextPassDumper\$php
array $php
Definition: TextPassDumper.php:71
XmlDumpWriter\WRITE_CONTENT
const WRITE_CONTENT
Output serialized revision content.
Definition: XmlDumpWriter.php:42
TextPassDumper\$firstPageWritten
$firstPageWritten
Definition: TextPassDumper.php:100
WikiExporter\TEXT
const TEXT
Definition: WikiExporter.php:59
TextPassDumper\dump
dump( $history, $text=WikiExporter::TEXT)
Definition: TextPassDumper.php:261
BackupDumper\$maxCount
int $maxCount
Definition: BackupDumper.php:77
TextPassDumper\$prefetchCount
$prefetchCount
Definition: TextPassDumper.php:60
TextPassDumper\$timeExceeded
$timeExceeded
Definition: TextPassDumper.php:99
TextPassDumper\$maxFailures
$maxFailures
Definition: TextPassDumper.php:64
TextPassDumper\$fetchCountLast
$fetchCountLast
Definition: TextPassDumper.php:62
TextPassDumper\$fetchCount
$fetchCount
Definition: TextPassDumper.php:59
Maintenance\addOption
addOption( $name, $description, $required=false, $withArg=false, $shortName=false, $multiOccurrence=false)
Add a parameter to the script.
Definition: Maintenance.php:245
TextPassDumper\$failureTimeout
$failureTimeout
Definition: TextPassDumper.php:66
Xml\element
static element( $element, $attribs=null, $contents='', $allowShortTag=true)
Format an XML element with given attributes and, optionally, text content.
Definition: Xml.php:41
$args
if( $line===false) $args
Definition: mcc.php:124
DB_REPLICA
const DB_REPLICA
Definition: defines.php:25
TextPassDumper\checkIfTimeExceeded
checkIfTimeExceeded()
Definition: TextPassDumper.php:404
wfDebug
wfDebug( $text, $dest='all', array $context=[])
Sends a line to the debug log if enabled or, optionally, to a comment in output.
Definition: GlobalFunctions.php:909
TextPassDumper\getTextSpawnedOnce
getTextSpawnedOnce( $address)
Definition: TextPassDumper.php:825
TextPassDumper\$thisRole
string bool $thisRole
Definition: TextPassDumper.php:50
TextPassDumper\getText
getText( $id, $model=null, $format=null, $expSize=null)
Tries to load revision text.
Definition: TextPassDumper.php:569
TextPassDumper\initProgress
initProgress( $history=WikiExporter::FULL)
Initialise starting time and maximum revision count.
Definition: TextPassDumper.php:256
BackupDumper\report
report( $final=false)
Definition: BackupDumper.php:410
TextPassDumper\$prefetch
BaseDump $prefetch
Definition: TextPassDumper.php:44
WikiExporter\FULL
const FULL
Definition: WikiExporter.php:53
TextPassDumper\$input
$input
Definition: TextPassDumper.php:57
BackupDumper\progress
progress( $string)
Definition: BackupDumper.php:457
TextPassDumper\$prefetchCountLast
$prefetchCountLast
Definition: TextPassDumper.php:61
TextPassDumper\characterData
characterData( $parser, $data)
Definition: TextPassDumper.php:1010
TextPassDumper\getTextDb
getTextDb( $id)
Loads the serialized content from storage.
Definition: TextPassDumper.php:718
BackupDumper
Definition: BackupDumper.php:39
Maintenance\getOption
getOption( $name, $default=null)
Get an option, or return the default.
Definition: Maintenance.php:281
TextPassDumper\$lastPageWritten
$lastPageWritten
Definition: TextPassDumper.php:101
TextPassDumper\openSpawn
openSpawn()
Definition: TextPassDumper.php:754
XmlDumpWriter
Definition: XmlDumpWriter.php:39
BackupDumper\$lastTime
$lastTime
Definition: BackupDumper.php:61
TextPassDumper\$spawn
$spawn
Definition: TextPassDumper.php:72
TextPassDumper\$thisPage
string bool $thisPage
Definition: TextPassDumper.php:46
Maintenance\output
output( $out, $channel=null)
Throw some output to the user.
Definition: Maintenance.php:434
TextPassDumper\endElement
endElement( $parser, $name)
Definition: TextPassDumper.php:949
wfWarn
wfWarn( $msg, $callerOffset=1, $level=E_USER_NOTICE)
Send a warning either to the debug log or in a PHP error depending on $wgDevelopmentWarnings.
Definition: GlobalFunctions.php:1073
ExportProgressFilter
Definition: ExportProgressFilter.php:27
TextPassDumper
Definition: TextPassDumper.php:42
Maintenance\hasOption
hasOption( $name)
Checks to see if a particular option exists.
Definition: Maintenance.php:266
TextPassDumper\$checkpointJustWritten
$checkpointJustWritten
Definition: TextPassDumper.php:102
$fileinfo
$fileinfo
Definition: generateLocalAutoload.php:18
$IP
$IP
Definition: WebStart.php:49
Wikimedia\Rdbms\IMaintainableDatabase
Advanced database interface for IDatabase handles that include maintenance methods.
Definition: IMaintainableDatabase.php:38
BackupDumper\$forcedDb
IMaintainableDatabase null $forcedDb
The dependency-injected database to use.
Definition: BackupDumper.php:104
TextPassDumper\closeSpawn
closeSpawn()
Definition: TextPassDumper.php:800
Revision\SlotRecord
Value object representing a content slot associated with a page revision.
Definition: SlotRecord.php:39