| blob | 85ebd51bfecf75fb3070eb3a9c9207ecd81b2f48 |
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 */
27 require_once __DIR__ . '/backup.inc';
29 /**
30 * @ingroup Maintenance
31 */
32 class TextPassDumper extends BackupDumper {
33 public $prefetch = null;
35 // when we spend more than maxTimeAllowed seconds on this run, we continue
36 // processing until we write out the next complete page, then save output file(s),
37 // rename it/them and open new one(s)
38 public $maxTimeAllowed = 0; // 0 = no limit
40 protected $input = "php://stdin";
41 protected $history = WikiExporter::FULL;
42 protected $fetchCount = 0;
43 protected $prefetchCount = 0;
44 protected $prefetchCountLast = 0;
45 protected $fetchCountLast = 0;
47 protected $maxFailures = 5;
48 protected $maxConsecutiveFailedTextRetrievals = 200;
49 protected $failureTimeout = 5; // Seconds to sleep after db failure
51 protected $php = "php";
52 protected $spawn = false;
54 /**
55 * @var bool|resource
56 */
57 protected $spawnProc = false;
59 /**
60 * @var bool|resource
61 */
62 protected $spawnWrite = false;
64 /**
65 * @var bool|resource
66 */
67 protected $spawnRead = false;
69 /**
70 * @var bool|resource
71 */
72 protected $spawnErr = false;
74 protected $xmlwriterobj = false;
76 protected $timeExceeded = false;
77 protected $firstPageWritten = false;
78 protected $lastPageWritten = false;
79 protected $checkpointJustWritten = false;
80 protected $checkpointFiles = array();
82 /**
83 * @var DatabaseBase
84 */
85 protected $db;
87 /**
88 * Drop the database connection $this->db and try to get a new one.
89 *
90 * This function tries to get a /different/ connection if this is
91 * possible. Hence, (if this is possible) it switches to a different
92 * failover upon each call.
93 *
94 * This function resets $this->lb and closes all connections on it.
95 *
96 * @throws MWException
97 */
98 function rotateDb() {
99 // Cleaning up old connections
100 if ( isset( $this->lb ) ) {
101 $this->lb->closeAll();
102 unset( $this->lb );
103 }
105 if ( $this->forcedDb !== null ) {
106 $this->db = $this->forcedDb;
108 return;
109 }
111 if ( isset( $this->db ) && $this->db->isOpen() ) {
112 throw new MWException( 'DB is set and has not been closed by the Load Balancer' );
113 }
115 unset( $this->db );
117 // Trying to set up new connection.
118 // We do /not/ retry upon failure, but delegate to encapsulating logic, to avoid
119 // individually retrying at different layers of code.
121 // 1. The LoadBalancer.
122 try {
123 $this->lb = wfGetLBFactory()->newMainLB();
124 } catch ( Exception $e ) {
125 throw new MWException( __METHOD__
126 . " rotating DB failed to obtain new load balancer (" . $e->getMessage() . ")" );
127 }
129 // 2. The Connection, through the load balancer.
130 try {
131 $this->db = $this->lb->getConnection( DB_SLAVE, 'dump' );
132 } catch ( Exception $e ) {
133 throw new MWException( __METHOD__
134 . " rotating DB failed to obtain new database (" . $e->getMessage() . ")" );
135 }
136 }
138 function initProgress( $history = WikiExporter::FULL ) {
139 parent::initProgress();
140 $this->timeOfCheckpoint = $this->startTime;
141 }
143 function dump( $history, $text = WikiExporter::TEXT ) {
144 // Notice messages will foul up your XML output even if they're
145 // relatively harmless.
146 if ( ini_get( 'display_errors' ) ) {
147 ini_set( 'display_errors', 'stderr' );
148 }
150 $this->initProgress( $this->history );
152 // We are trying to get an initial database connection to avoid that the
153 // first try of this request's first call to getText fails. However, if
154 // obtaining a good DB connection fails it's not a serious issue, as
155 // getText does retry upon failure and can start without having a working
156 // DB connection.
157 try {
158 $this->rotateDb();
159 } catch ( Exception $e ) {
160 // We do not even count this as failure. Just let eventual
161 // watchdogs know.
162 $this->progress( "Getting initial DB connection failed (" .
163 $e->getMessage() . ")" );
164 }
166 $this->egress = new ExportProgressFilter( $this->sink, $this );
168 // it would be nice to do it in the constructor, oh well. need egress set
169 $this->finalOptionCheck();
171 // we only want this so we know how to close a stream :-P
172 $this->xmlwriterobj = new XmlDumpWriter();
174 $input = fopen( $this->input, "rt" );
175 $this->readDump( $input );
177 if ( $this->spawnProc ) {
178 $this->closeSpawn();
179 }
181 $this->report( true );
182 }
184 function processOption( $opt, $val, $param ) {
185 global $IP;
186 $url = $this->processFileOpt( $val, $param );
188 switch ( $opt ) {
189 case 'prefetch':
190 require_once "$IP/maintenance/backupPrefetch.inc";
191 $this->prefetch = new BaseDump( $url );
192 break;
193 case 'stub':
194 $this->input = $url;
195 break;
196 case 'maxtime':
197 $this->maxTimeAllowed = intval( $val ) * 60;
198 break;
199 case 'checkpointfile':
200 $this->checkpointFiles[] = $val;
201 break;
202 case 'current':
203 $this->history = WikiExporter::CURRENT;
204 break;
205 case 'full':
206 $this->history = WikiExporter::FULL;
207 break;
208 case 'spawn':
209 $this->spawn = true;
210 if ( $val ) {
211 $this->php = $val;
212 }
213 break;
214 }
215 }
217 function processFileOpt( $val, $param ) {
218 $fileURIs = explode( ';', $param );
219 foreach ( $fileURIs as $URI ) {
220 switch ( $val ) {
221 case "file":
222 $newURI = $URI;
223 break;
224 case "gzip":
225 $newURI = "compress.zlib://$URI";
226 break;
227 case "bzip2":
228 $newURI = "compress.bzip2://$URI";
229 break;
230 case "7zip":
231 $newURI = "mediawiki.compress.7z://$URI";
232 break;
233 default:
234 $newURI = $URI;
235 }
236 $newFileURIs[] = $newURI;
237 }
238 $val = implode( ';', $newFileURIs );
240 return $val;
241 }
243 /**
244 * Overridden to include prefetch ratio if enabled.
245 */
246 function showReport() {
247 if ( !$this->prefetch ) {
248 parent::showReport();
250 return;
251 }
253 if ( $this->reporting ) {
254 $now = wfTimestamp( TS_DB );
255 $nowts = microtime( true );
256 $deltaAll = $nowts - $this->startTime;
257 $deltaPart = $nowts - $this->lastTime;
258 $this->pageCountPart = $this->pageCount - $this->pageCountLast;
259 $this->revCountPart = $this->revCount - $this->revCountLast;
261 if ( $deltaAll ) {
262 $portion = $this->revCount / $this->maxCount;
263 $eta = $this->startTime + $deltaAll / $portion;
264 $etats = wfTimestamp( TS_DB, intval( $eta ) );
265 if ( $this->fetchCount ) {
266 $fetchRate = 100.0 * $this->prefetchCount / $this->fetchCount;
267 } else {
268 $fetchRate = '-';
269 }
270 $pageRate = $this->pageCount / $deltaAll;
271 $revRate = $this->revCount / $deltaAll;
272 } else {
273 $pageRate = '-';
274 $revRate = '-';
275 $etats = '-';
276 $fetchRate = '-';
277 }
278 if ( $deltaPart ) {
279 if ( $this->fetchCountLast ) {
280 $fetchRatePart = 100.0 * $this->prefetchCountLast / $this->fetchCountLast;
281 } else {
282 $fetchRatePart = '-';
283 }
284 $pageRatePart = $this->pageCountPart / $deltaPart;
285 $revRatePart = $this->revCountPart / $deltaPart;
286 } else {
287 $fetchRatePart = '-';
288 $pageRatePart = '-';
289 $revRatePart = '-';
290 }
291 $this->progress( sprintf(
292 "%s: %s (ID %d) %d pages (%0.1f|%0.1f/sec all|curr), "
293 . "%d revs (%0.1f|%0.1f/sec all|curr), %0.1f%%|%0.1f%% "
294 . "prefetched (all|curr), ETA %s [max %d]",
295 $now, wfWikiID(), $this->ID, $this->pageCount, $pageRate,
296 $pageRatePart, $this->revCount, $revRate, $revRatePart,
297 $fetchRate, $fetchRatePart, $etats, $this->maxCount
298 ) );
299 $this->lastTime = $nowts;
300 $this->revCountLast = $this->revCount;
301 $this->prefetchCountLast = $this->prefetchCount;
302 $this->fetchCountLast = $this->fetchCount;
303 }
304 }
306 function setTimeExceeded() {
307 $this->timeExceeded = true;
308 }
310 function checkIfTimeExceeded() {
311 if ( $this->maxTimeAllowed
312 && ( $this->lastTime - $this->timeOfCheckpoint > $this->maxTimeAllowed )
313 ) {
314 return true;
315 }
317 return false;
318 }
320 function finalOptionCheck() {
321 if ( ( $this->checkpointFiles && !$this->maxTimeAllowed )
322 || ( $this->maxTimeAllowed && !$this->checkpointFiles )
323 ) {
324 throw new MWException( "Options checkpointfile and maxtime must be specified together.\n" );
325 }
326 foreach ( $this->checkpointFiles as $checkpointFile ) {
327 $count = substr_count( $checkpointFile, "%s" );
328 if ( $count != 2 ) {
329 throw new MWException( "Option checkpointfile must contain two '%s' "
330 . "for substitution of first and last pageids, count is $count instead, "
331 . "file is $checkpointFile.\n" );
332 }
333 }
335 if ( $this->checkpointFiles ) {
336 $filenameList = (array)$this->egress->getFilenames();
337 if ( count( $filenameList ) != count( $this->checkpointFiles ) ) {
338 throw new MWException( "One checkpointfile must be specified "
339 . "for each output option, if maxtime is used.\n" );
340 }
341 }
342 }
344 /**
345 * @throws MWException Failure to parse XML input
346 * @param string $input
347 * @return bool
348 */
349 function readDump( $input ) {
350 $this->buffer = "";
351 $this->openElement = false;
352 $this->atStart = true;
353 $this->state = "";
354 $this->lastName = "";
355 $this->thisPage = 0;
356 $this->thisRev = 0;
357 $this->thisRevModel = null;
358 $this->thisRevFormat = null;
360 $parser = xml_parser_create( "UTF-8" );
361 xml_parser_set_option( $parser, XML_OPTION_CASE_FOLDING, false );
363 xml_set_element_handler(
364 $parser,
365 array( &$this, 'startElement' ),
366 array( &$this, 'endElement' )
367 );
368 xml_set_character_data_handler( $parser, array( &$this, 'characterData' ) );
370 $offset = 0; // for context extraction on error reporting
371 $bufferSize = 512 * 1024;
372 do {
373 if ( $this->checkIfTimeExceeded() ) {
374 $this->setTimeExceeded();
375 }
376 $chunk = fread( $input, $bufferSize );
377 if ( !xml_parse( $parser, $chunk, feof( $input ) ) ) {
378 wfDebug( "TextDumpPass::readDump encountered XML parsing error\n" );
380 $byte = xml_get_current_byte_index( $parser );
381 $msg = wfMessage( 'xml-error-string',
382 'XML import parse failure',
383 xml_get_current_line_number( $parser ),
384 xml_get_current_column_number( $parser ),
385 $byte . ( is_null( $chunk ) ? null : ( '; "' . substr( $chunk, $byte - $offset, 16 ) . '"' ) ),
386 xml_error_string( xml_get_error_code( $parser ) ) )->escaped();
388 xml_parser_free( $parser );
390 throw new MWException( $msg );
391 }
392 $offset += strlen( $chunk );
393 } while ( $chunk !== false && !feof( $input ) );
394 if ( $this->maxTimeAllowed ) {
395 $filenameList = (array)$this->egress->getFilenames();
396 // we wrote some stuff after last checkpoint that needs renamed
397 if ( file_exists( $filenameList[0] ) ) {
398 $newFilenames = array();
399 # we might have just written the header and footer and had no
400 # pages or revisions written... perhaps they were all deleted
401 # there's no pageID 0 so we use that. the caller is responsible
402 # for deciding what to do with a file containing only the
403 # siteinfo information and the mw tags.
404 if ( !$this->firstPageWritten ) {
405 $firstPageID = str_pad( 0, 9, "0", STR_PAD_LEFT );
406 $lastPageID = str_pad( 0, 9, "0", STR_PAD_LEFT );
407 } else {
408 $firstPageID = str_pad( $this->firstPageWritten, 9, "0", STR_PAD_LEFT );
409 $lastPageID = str_pad( $this->lastPageWritten, 9, "0", STR_PAD_LEFT );
410 }
412 $filenameCount = count( $filenameList );
413 for ( $i = 0; $i < $filenameCount; $i++ ) {
414 $checkpointNameFilledIn = sprintf( $this->checkpointFiles[$i], $firstPageID, $lastPageID );
415 $fileinfo = pathinfo( $filenameList[$i] );
416 $newFilenames[] = $fileinfo['dirname'] . '/' . $checkpointNameFilledIn;
417 }
418 $this->egress->closeAndRename( $newFilenames );
419 }
420 }
421 xml_parser_free( $parser );
423 return true;
424 }
426 /**
427 * Applies applicable export transformations to $text.
428 *
429 * @param string $text
430 * @param string $model
431 * @param string|null $format
432 *
433 * @return string
434 */
435 private function exportTransform( $text, $model, $format = null ) {
436 try {
437 $handler = ContentHandler::getForModelID( $model );
438 $text = $handler->exportTransform( $text, $format );
439 }
440 catch ( MWException $ex ) {
441 $this->progress(
442 "Unable to apply export transformation for content model '$model': " .
443 $ex->getMessage()
444 );
445 }
447 return $text;
448 }
450 /**
451 * Tries to get the revision text for a revision id.
452 * Export transformations are applied if the content model can is given or can be
453 * determined from the database.
454 *
455 * Upon errors, retries (Up to $this->maxFailures tries each call).
456 * If still no good revision get could be found even after this retrying, "" is returned.
457 * If no good revision text could be returned for
458 * $this->maxConsecutiveFailedTextRetrievals consecutive calls to getText, MWException
459 * is thrown.
460 *
461 * @param string $id The revision id to get the text for
462 * @param string|bool|null $model The content model used to determine applicable export transformations.
463 * If $model is null, it will be determined from the database.
464 * @param string|null $format The content format used when applying export transformations.
465 *
466 * @throws MWException
467 * @return string The revision text for $id, or ""
468 */
469 function getText( $id, $model = null, $format = null ) {
470 global $wgContentHandlerUseDB;
472 $prefetchNotTried = true; // Whether or not we already tried to get the text via prefetch.
473 $text = false; // The candidate for a good text. false if no proper value.
474 $failures = 0; // The number of times, this invocation of getText already failed.
476 // The number of times getText failed without yielding a good text in between.
477 static $consecutiveFailedTextRetrievals = 0;
479 $this->fetchCount++;
481 // To allow to simply return on success and do not have to worry about book keeping,
482 // we assume, this fetch works (possible after some retries). Nevertheless, we koop
483 // the old value, so we can restore it, if problems occur (See after the while loop).
484 $oldConsecutiveFailedTextRetrievals = $consecutiveFailedTextRetrievals;
485 $consecutiveFailedTextRetrievals = 0;
487 if ( $model === null && $wgContentHandlerUseDB ) {
488 $row = $this->db->selectRow(
489 'revision',
490 array( 'rev_content_model', 'rev_content_format' ),
491 array( 'rev_id' => $this->thisRev ),
492 __METHOD__
493 );
495 if ( $row ) {
496 $model = $row->rev_content_model;
497 $format = $row->rev_content_format;
498 }
499 }
501 if ( $model === null || $model === '' ) {
502 $model = false;
503 }
505 while ( $failures < $this->maxFailures ) {
507 // As soon as we found a good text for the $id, we will return immediately.
508 // Hence, if we make it past the try catch block, we know that we did not
509 // find a good text.
511 try {
512 // Step 1: Get some text (or reuse from previous iteratuon if checking
513 // for plausibility failed)
515 // Trying to get prefetch, if it has not been tried before
516 if ( $text === false && isset( $this->prefetch ) && $prefetchNotTried ) {
517 $prefetchNotTried = false;
518 $tryIsPrefetch = true;
519 $text = $this->prefetch->prefetch( intval( $this->thisPage ),
520 intval( $this->thisRev ) );
522 if ( $text === null ) {
523 $text = false;
524 }
526 if ( is_string( $text ) && $model !== false ) {
527 // Apply export transformation to text coming from an old dump.
528 // The purpose of this transformation is to convert up from legacy
529 // formats, which may still be used in the older dump that is used
530 // for pre-fetching. Applying the transformation again should not
531 // interfere with content that is already in the correct form.
532 $text = $this->exportTransform( $text, $model, $format );
533 }
534 }
536 if ( $text === false ) {
537 // Fallback to asking the database
538 $tryIsPrefetch = false;
539 if ( $this->spawn ) {
540 $text = $this->getTextSpawned( $id );
541 } else {
542 $text = $this->getTextDb( $id );
543 }
545 if ( $text !== false && $model !== false ) {
546 // Apply export transformation to text coming from the database.
547 // Prefetched text should already have transformations applied.
548 $text = $this->exportTransform( $text, $model, $format );
549 }
551 // No more checks for texts from DB for now.
552 // If we received something that is not false,
553 // We treat it as good text, regardless of whether it actually is or is not
554 if ( $text !== false ) {
555 return $text;
556 }
557 }
559 if ( $text === false ) {
560 throw new MWException( "Generic error while obtaining text for id " . $id );
561 }
563 // We received a good candidate for the text of $id via some method
565 // Step 2: Checking for plausibility and return the text if it is
566 // plausible
567 $revID = intval( $this->thisRev );
568 if ( !isset( $this->db ) ) {
569 throw new MWException( "No database available" );
570 }
572 if ( $model !== CONTENT_MODEL_WIKITEXT ) {
573 $revLength = strlen( $text );
574 } else {
575 $revLength = $this->db->selectField( 'revision', 'rev_len', array( 'rev_id' => $revID ) );
576 }
578 if ( strlen( $text ) == $revLength ) {
579 if ( $tryIsPrefetch ) {
580 $this->prefetchCount++;
581 }
583 return $text;
584 }
586 $text = false;
587 throw new MWException( "Received text is unplausible for id " . $id );
588 } catch ( Exception $e ) {
589 $msg = "getting/checking text " . $id . " failed (" . $e->getMessage() . ")";
590 if ( $failures + 1 < $this->maxFailures ) {
591 $msg .= " (Will retry " . ( $this->maxFailures - $failures - 1 ) . " more times)";
592 }
593 $this->progress( $msg );
594 }
596 // Something went wrong; we did not a text that was plausible :(
597 $failures++;
599 // A failure in a prefetch hit does not warrant resetting db connection etc.
600 if ( !$tryIsPrefetch ) {
601 // After backing off for some time, we try to reboot the whole process as
602 // much as possible to not carry over failures from one part to the other
603 // parts
604 sleep( $this->failureTimeout );
605 try {
606 $this->rotateDb();
607 if ( $this->spawn ) {
608 $this->closeSpawn();
609 $this->openSpawn();
610 }
611 } catch ( Exception $e ) {
612 $this->progress( "Rebooting getText infrastructure failed (" . $e->getMessage() . ")" .
613 " Trying to continue anyways" );
614 }
615 }
616 }
618 // Retirieving a good text for $id failed (at least) maxFailures times.
619 // We abort for this $id.
621 // Restoring the consecutive failures, and maybe aborting, if the dump
622 // is too broken.
623 $consecutiveFailedTextRetrievals = $oldConsecutiveFailedTextRetrievals + 1;
624 if ( $consecutiveFailedTextRetrievals > $this->maxConsecutiveFailedTextRetrievals ) {
625 throw new MWException( "Graceful storage failure" );
626 }
628 return "";
629 }
631 /**
632 * May throw a database error if, say, the server dies during query.
633 * @param int $id
634 * @return bool|string
635 * @throws MWException
636 */
637 private function getTextDb( $id ) {
638 global $wgContLang;
639 if ( !isset( $this->db ) ) {
640 throw new MWException( __METHOD__ . "No database available" );
641 }
642 $row = $this->db->selectRow( 'text',
643 array( 'old_text', 'old_flags' ),
644 array( 'old_id' => $id ),
645 __METHOD__ );
646 $text = Revision::getRevisionText( $row );
647 if ( $text === false ) {
648 return false;
649 }
650 $stripped = str_replace( "\r", "", $text );
651 $normalized = $wgContLang->normalize( $stripped );
653 return $normalized;
654 }
656 private function getTextSpawned( $id ) {
657 wfSuppressWarnings();
658 if ( !$this->spawnProc ) {
659 // First time?
660 $this->openSpawn();
661 }
662 $text = $this->getTextSpawnedOnce( $id );
663 wfRestoreWarnings();
665 return $text;
666 }
668 function openSpawn() {
669 global $IP;
671 if ( file_exists( "$IP/../multiversion/MWScript.php" ) ) {
672 $cmd = implode( " ",
673 array_map( 'wfEscapeShellArg',
674 array(
675 $this->php,
676 "$IP/../multiversion/MWScript.php",
677 "fetchText.php",
678 '--wiki', wfWikiID() ) ) );
679 } else {
680 $cmd = implode( " ",
681 array_map( 'wfEscapeShellArg',
682 array(
683 $this->php,
684 "$IP/maintenance/fetchText.php",
685 '--wiki', wfWikiID() ) ) );
686 }
687 $spec = array(
688 0 => array( "pipe", "r" ),
689 1 => array( "pipe", "w" ),
690 2 => array( "file", "/dev/null", "a" ) );
691 $pipes = array();
693 $this->progress( "Spawning database subprocess: $cmd" );
694 $this->spawnProc = proc_open( $cmd, $spec, $pipes );
695 if ( !$this->spawnProc ) {
696 // shit
697 $this->progress( "Subprocess spawn failed." );
699 return false;
700 }
701 list(
702 $this->spawnWrite, // -> stdin
703 $this->spawnRead, // <- stdout
704 ) = $pipes;
706 return true;
707 }
709 private function closeSpawn() {
710 wfSuppressWarnings();
711 if ( $this->spawnRead ) {
712 fclose( $this->spawnRead );
713 }
714 $this->spawnRead = false;
715 if ( $this->spawnWrite ) {
716 fclose( $this->spawnWrite );
717 }
718 $this->spawnWrite = false;
719 if ( $this->spawnErr ) {
720 fclose( $this->spawnErr );
721 }
722 $this->spawnErr = false;
723 if ( $this->spawnProc ) {
724 pclose( $this->spawnProc );
725 }
726 $this->spawnProc = false;
727 wfRestoreWarnings();
728 }
730 private function getTextSpawnedOnce( $id ) {
731 global $wgContLang;
733 $ok = fwrite( $this->spawnWrite, "$id\n" );
734 // $this->progress( ">> $id" );
735 if ( !$ok ) {
736 return false;
737 }
739 $ok = fflush( $this->spawnWrite );
740 // $this->progress( ">> [flush]" );
741 if ( !$ok ) {
742 return false;
743 }
745 // check that the text id they are sending is the one we asked for
746 // this avoids out of sync revision text errors we have encountered in the past
747 $newId = fgets( $this->spawnRead );
748 if ( $newId === false ) {
749 return false;
750 }
751 if ( $id != intval( $newId ) ) {
752 return false;
753 }
755 $len = fgets( $this->spawnRead );
756 // $this->progress( "<< " . trim( $len ) );
757 if ( $len === false ) {
758 return false;
759 }
761 $nbytes = intval( $len );
762 // actual error, not zero-length text
763 if ( $nbytes < 0 ) {
764 return false;
765 }
767 $text = "";
769 // Subprocess may not send everything at once, we have to loop.
770 while ( $nbytes > strlen( $text ) ) {
771 $buffer = fread( $this->spawnRead, $nbytes - strlen( $text ) );
772 if ( $buffer === false ) {
773 break;
774 }
775 $text .= $buffer;
776 }
778 $gotbytes = strlen( $text );
779 if ( $gotbytes != $nbytes ) {
780 $this->progress( "Expected $nbytes bytes from database subprocess, got $gotbytes " );
782 return false;
783 }
785 // Do normalization in the dump thread...
786 $stripped = str_replace( "\r", "", $text );
787 $normalized = $wgContLang->normalize( $stripped );
789 return $normalized;
790 }
792 function startElement( $parser, $name, $attribs ) {
793 $this->checkpointJustWritten = false;
795 $this->clearOpenElement( null );
796 $this->lastName = $name;
798 if ( $name == 'revision' ) {
799 $this->state = $name;
800 $this->egress->writeOpenPage( null, $this->buffer );
801 $this->buffer = "";
802 } elseif ( $name == 'page' ) {
803 $this->state = $name;
804 if ( $this->atStart ) {
805 $this->egress->writeOpenStream( $this->buffer );
806 $this->buffer = "";
807 $this->atStart = false;
808 }
809 }
811 if ( $name == "text" && isset( $attribs['id'] ) ) {
812 $id = $attribs['id'];
813 $model = trim( $this->thisRevModel );
814 $format = trim( $this->thisRevFormat );
816 $model = $model === '' ? null : $model;
817 $format = $format === '' ? null : $format;
819 $text = $this->getText( $id, $model, $format );
820 $this->openElement = array( $name, array( 'xml:space' => 'preserve' ) );
821 if ( strlen( $text ) > 0 ) {
822 $this->characterData( $parser, $text );
823 }
824 } else {
825 $this->openElement = array( $name, $attribs );
826 }
827 }
829 function endElement( $parser, $name ) {
830 $this->checkpointJustWritten = false;
832 if ( $this->openElement ) {
833 $this->clearOpenElement( "" );
834 } else {
835 $this->buffer .= "</$name>";
836 }
838 if ( $name == 'revision' ) {
839 $this->egress->writeRevision( null, $this->buffer );
840 $this->buffer = "";
841 $this->thisRev = "";
842 $this->thisRevModel = null;
843 $this->thisRevFormat = null;
844 } elseif ( $name == 'page' ) {
845 if ( !$this->firstPageWritten ) {
846 $this->firstPageWritten = trim( $this->thisPage );
847 }
848 $this->lastPageWritten = trim( $this->thisPage );
849 if ( $this->timeExceeded ) {
850 $this->egress->writeClosePage( $this->buffer );
851 // nasty hack, we can't just write the chardata after the
852 // page tag, it will include leading blanks from the next line
853 $this->egress->sink->write( "\n" );
855 $this->buffer = $this->xmlwriterobj->closeStream();
856 $this->egress->writeCloseStream( $this->buffer );
858 $this->buffer = "";
859 $this->thisPage = "";
860 // this could be more than one file if we had more than one output arg
862 $filenameList = (array)$this->egress->getFilenames();
863 $newFilenames = array();
864 $firstPageID = str_pad( $this->firstPageWritten, 9, "0", STR_PAD_LEFT );
865 $lastPageID = str_pad( $this->lastPageWritten, 9, "0", STR_PAD_LEFT );
866 $filenamesCount = count( $filenameList );
867 for ( $i = 0; $i < $filenamesCount; $i++ ) {
868 $checkpointNameFilledIn = sprintf( $this->checkpointFiles[$i], $firstPageID, $lastPageID );
869 $fileinfo = pathinfo( $filenameList[$i] );
870 $newFilenames[] = $fileinfo['dirname'] . '/' . $checkpointNameFilledIn;
871 }
872 $this->egress->closeRenameAndReopen( $newFilenames );
873 $this->buffer = $this->xmlwriterobj->openStream();
874 $this->timeExceeded = false;
875 $this->timeOfCheckpoint = $this->lastTime;
876 $this->firstPageWritten = false;
877 $this->checkpointJustWritten = true;
878 } else {
879 $this->egress->writeClosePage( $this->buffer );
880 $this->buffer = "";
881 $this->thisPage = "";
882 }
883 } elseif ( $name == 'mediawiki' ) {
884 $this->egress->writeCloseStream( $this->buffer );
885 $this->buffer = "";
886 }
887 }
889 function characterData( $parser, $data ) {
890 $this->clearOpenElement( null );
891 if ( $this->lastName == "id" ) {
892 if ( $this->state == "revision" ) {
893 $this->thisRev .= $data;
894 } elseif ( $this->state == "page" ) {
895 $this->thisPage .= $data;
896 }
897 }
898 elseif ( $this->lastName == "model" ) {
899 $this->thisRevModel .= $data;
900 }
901 elseif ( $this->lastName == "format" ) {
902 $this->thisRevFormat .= $data;
903 }
905 // have to skip the newline left over from closepagetag line of
906 // end of checkpoint files. nasty hack!!
907 if ( $this->checkpointJustWritten ) {
908 if ( $data[0] == "\n" ) {
909 $data = substr( $data, 1 );
910 }
911 $this->checkpointJustWritten = false;
912 }
913 $this->buffer .= htmlspecialchars( $data );
914 }
916 function clearOpenElement( $style ) {
917 if ( $this->openElement ) {
918 $this->buffer .= Xml::element( $this->openElement[0], $this->openElement[1], $style );
919 $this->openElement = false;
920 }
921 }
922 }