| blob | 5f7763730e63b99b5b0372e2af1b5b99213c3798 |
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;
358 $parser = xml_parser_create( "UTF-8" );
359 xml_parser_set_option( $parser, XML_OPTION_CASE_FOLDING, false );
361 xml_set_element_handler(
362 $parser,
363 array( &$this, 'startElement' ),
364 array( &$this, 'endElement' )
365 );
366 xml_set_character_data_handler( $parser, array( &$this, 'characterData' ) );
368 $offset = 0; // for context extraction on error reporting
369 $bufferSize = 512 * 1024;
370 do {
371 if ( $this->checkIfTimeExceeded() ) {
372 $this->setTimeExceeded();
373 }
374 $chunk = fread( $input, $bufferSize );
375 if ( !xml_parse( $parser, $chunk, feof( $input ) ) ) {
376 wfDebug( "TextDumpPass::readDump encountered XML parsing error\n" );
378 $byte = xml_get_current_byte_index( $parser );
379 $msg = wfMessage( 'xml-error-string',
380 'XML import parse failure',
381 xml_get_current_line_number( $parser ),
382 xml_get_current_column_number( $parser ),
383 $byte . ( is_null( $chunk ) ? null : ( '; "' . substr( $chunk, $byte - $offset, 16 ) . '"' ) ),
384 xml_error_string( xml_get_error_code( $parser ) ) )->escaped();
386 xml_parser_free( $parser );
388 throw new MWException( $msg );
389 }
390 $offset += strlen( $chunk );
391 } while ( $chunk !== false && !feof( $input ) );
392 if ( $this->maxTimeAllowed ) {
393 $filenameList = (array)$this->egress->getFilenames();
394 // we wrote some stuff after last checkpoint that needs renamed
395 if ( file_exists( $filenameList[0] ) ) {
396 $newFilenames = array();
397 # we might have just written the header and footer and had no
398 # pages or revisions written... perhaps they were all deleted
399 # there's no pageID 0 so we use that. the caller is responsible
400 # for deciding what to do with a file containing only the
401 # siteinfo information and the mw tags.
402 if ( !$this->firstPageWritten ) {
403 $firstPageID = str_pad( 0, 9, "0", STR_PAD_LEFT );
404 $lastPageID = str_pad( 0, 9, "0", STR_PAD_LEFT );
405 } else {
406 $firstPageID = str_pad( $this->firstPageWritten, 9, "0", STR_PAD_LEFT );
407 $lastPageID = str_pad( $this->lastPageWritten, 9, "0", STR_PAD_LEFT );
408 }
410 $filenameCount = count( $filenameList );
411 for ( $i = 0; $i < $filenameCount; $i++ ) {
412 $checkpointNameFilledIn = sprintf( $this->checkpointFiles[$i], $firstPageID, $lastPageID );
413 $fileinfo = pathinfo( $filenameList[$i] );
414 $newFilenames[] = $fileinfo['dirname'] . '/' . $checkpointNameFilledIn;
415 }
416 $this->egress->closeAndRename( $newFilenames );
417 }
418 }
419 xml_parser_free( $parser );
421 return true;
422 }
424 /**
425 * Tries to get the revision text for a revision id.
426 *
427 * Upon errors, retries (Up to $this->maxFailures tries each call).
428 * If still no good revision get could be found even after this retrying, "" is returned.
429 * If no good revision text could be returned for
430 * $this->maxConsecutiveFailedTextRetrievals consecutive calls to getText, MWException
431 * is thrown.
432 *
433 * @param string $id The revision id to get the text for
434 *
435 * @return string The revision text for $id, or ""
436 * @throws MWException
437 */
438 function getText( $id ) {
439 global $wgContentHandlerUseDB;
441 $prefetchNotTried = true; // Whether or not we already tried to get the text via prefetch.
442 $text = false; // The candidate for a good text. false if no proper value.
443 $failures = 0; // The number of times, this invocation of getText already failed.
445 // The number of times getText failed without yielding a good text in between.
446 static $consecutiveFailedTextRetrievals = 0;
448 $this->fetchCount++;
450 // To allow to simply return on success and do not have to worry about book keeping,
451 // we assume, this fetch works (possible after some retries). Nevertheless, we koop
452 // the old value, so we can restore it, if problems occur (See after the while loop).
453 $oldConsecutiveFailedTextRetrievals = $consecutiveFailedTextRetrievals;
454 $consecutiveFailedTextRetrievals = 0;
456 while ( $failures < $this->maxFailures ) {
458 // As soon as we found a good text for the $id, we will return immediately.
459 // Hence, if we make it past the try catch block, we know that we did not
460 // find a good text.
462 try {
463 // Step 1: Get some text (or reuse from previous iteratuon if checking
464 // for plausibility failed)
466 // Trying to get prefetch, if it has not been tried before
467 if ( $text === false && isset( $this->prefetch ) && $prefetchNotTried ) {
468 $prefetchNotTried = false;
469 $tryIsPrefetch = true;
470 $text = $this->prefetch->prefetch( intval( $this->thisPage ),
471 intval( $this->thisRev ) );
472 if ( $text === null ) {
473 $text = false;
474 }
475 }
477 if ( $text === false ) {
478 // Fallback to asking the database
479 $tryIsPrefetch = false;
480 if ( $this->spawn ) {
481 $text = $this->getTextSpawned( $id );
482 } else {
483 $text = $this->getTextDb( $id );
484 }
486 // No more checks for texts from DB for now.
487 // If we received something that is not false,
488 // We treat it as good text, regardless of whether it actually is or is not
489 if ( $text !== false ) {
490 return $text;
491 }
492 }
494 if ( $text === false ) {
495 throw new MWException( "Generic error while obtaining text for id " . $id );
496 }
498 // We received a good candidate for the text of $id via some method
500 // Step 2: Checking for plausibility and return the text if it is
501 // plausible
502 $revID = intval( $this->thisRev );
503 if ( !isset( $this->db ) ) {
504 throw new MWException( "No database available" );
505 }
507 $revLength = strlen( $text );
508 if ( $wgContentHandlerUseDB ) {
509 $row = $this->db->selectRow(
510 'revision',
511 array( 'rev_len', 'rev_content_model' ),
512 array( 'rev_id' => $revID ),
513 __METHOD__
514 );
515 if ( $row ) {
516 // only check the length for the wikitext content handler,
517 // it's a wasted (and failed) check otherwise
518 if ( $row->rev_content_model == CONTENT_MODEL_WIKITEXT ) {
519 $revLength = $row->rev_len;
520 }
521 }
522 } else {
523 $revLength = $this->db->selectField( 'revision', 'rev_len', array( 'rev_id' => $revID ) );
524 }
526 if ( strlen( $text ) == $revLength ) {
527 if ( $tryIsPrefetch ) {
528 $this->prefetchCount++;
529 }
531 return $text;
532 }
534 $text = false;
535 throw new MWException( "Received text is unplausible for id " . $id );
536 } catch ( Exception $e ) {
537 $msg = "getting/checking text " . $id . " failed (" . $e->getMessage() . ")";
538 if ( $failures + 1 < $this->maxFailures ) {
539 $msg .= " (Will retry " . ( $this->maxFailures - $failures - 1 ) . " more times)";
540 }
541 $this->progress( $msg );
542 }
544 // Something went wrong; we did not a text that was plausible :(
545 $failures++;
547 // A failure in a prefetch hit does not warrant resetting db connection etc.
548 if ( !$tryIsPrefetch ) {
549 // After backing off for some time, we try to reboot the whole process as
550 // much as possible to not carry over failures from one part to the other
551 // parts
552 sleep( $this->failureTimeout );
553 try {
554 $this->rotateDb();
555 if ( $this->spawn ) {
556 $this->closeSpawn();
557 $this->openSpawn();
558 }
559 } catch ( Exception $e ) {
560 $this->progress( "Rebooting getText infrastructure failed (" . $e->getMessage() . ")" .
561 " Trying to continue anyways" );
562 }
563 }
564 }
566 // Retirieving a good text for $id failed (at least) maxFailures times.
567 // We abort for this $id.
569 // Restoring the consecutive failures, and maybe aborting, if the dump
570 // is too broken.
571 $consecutiveFailedTextRetrievals = $oldConsecutiveFailedTextRetrievals + 1;
572 if ( $consecutiveFailedTextRetrievals > $this->maxConsecutiveFailedTextRetrievals ) {
573 throw new MWException( "Graceful storage failure" );
574 }
576 return "";
577 }
579 /**
580 * May throw a database error if, say, the server dies during query.
581 * @param int $id
582 * @return bool|string
583 * @throws MWException
584 */
585 private function getTextDb( $id ) {
586 global $wgContLang;
587 if ( !isset( $this->db ) ) {
588 throw new MWException( __METHOD__ . "No database available" );
589 }
590 $row = $this->db->selectRow( 'text',
591 array( 'old_text', 'old_flags' ),
592 array( 'old_id' => $id ),
593 __METHOD__ );
594 $text = Revision::getRevisionText( $row );
595 if ( $text === false ) {
596 return false;
597 }
598 $stripped = str_replace( "\r", "", $text );
599 $normalized = $wgContLang->normalize( $stripped );
601 return $normalized;
602 }
604 private function getTextSpawned( $id ) {
605 wfSuppressWarnings();
606 if ( !$this->spawnProc ) {
607 // First time?
608 $this->openSpawn();
609 }
610 $text = $this->getTextSpawnedOnce( $id );
611 wfRestoreWarnings();
613 return $text;
614 }
616 function openSpawn() {
617 global $IP;
619 if ( file_exists( "$IP/../multiversion/MWScript.php" ) ) {
620 $cmd = implode( " ",
621 array_map( 'wfEscapeShellArg',
622 array(
623 $this->php,
624 "$IP/../multiversion/MWScript.php",
625 "fetchText.php",
626 '--wiki', wfWikiID() ) ) );
627 } else {
628 $cmd = implode( " ",
629 array_map( 'wfEscapeShellArg',
630 array(
631 $this->php,
632 "$IP/maintenance/fetchText.php",
633 '--wiki', wfWikiID() ) ) );
634 }
635 $spec = array(
636 0 => array( "pipe", "r" ),
637 1 => array( "pipe", "w" ),
638 2 => array( "file", "/dev/null", "a" ) );
639 $pipes = array();
641 $this->progress( "Spawning database subprocess: $cmd" );
642 $this->spawnProc = proc_open( $cmd, $spec, $pipes );
643 if ( !$this->spawnProc ) {
644 // shit
645 $this->progress( "Subprocess spawn failed." );
647 return false;
648 }
649 list(
650 $this->spawnWrite, // -> stdin
651 $this->spawnRead, // <- stdout
652 ) = $pipes;
654 return true;
655 }
657 private function closeSpawn() {
658 wfSuppressWarnings();
659 if ( $this->spawnRead ) {
660 fclose( $this->spawnRead );
661 }
662 $this->spawnRead = false;
663 if ( $this->spawnWrite ) {
664 fclose( $this->spawnWrite );
665 }
666 $this->spawnWrite = false;
667 if ( $this->spawnErr ) {
668 fclose( $this->spawnErr );
669 }
670 $this->spawnErr = false;
671 if ( $this->spawnProc ) {
672 pclose( $this->spawnProc );
673 }
674 $this->spawnProc = false;
675 wfRestoreWarnings();
676 }
678 private function getTextSpawnedOnce( $id ) {
679 global $wgContLang;
681 $ok = fwrite( $this->spawnWrite, "$id\n" );
682 // $this->progress( ">> $id" );
683 if ( !$ok ) {
684 return false;
685 }
687 $ok = fflush( $this->spawnWrite );
688 // $this->progress( ">> [flush]" );
689 if ( !$ok ) {
690 return false;
691 }
693 // check that the text id they are sending is the one we asked for
694 // this avoids out of sync revision text errors we have encountered in the past
695 $newId = fgets( $this->spawnRead );
696 if ( $newId === false ) {
697 return false;
698 }
699 if ( $id != intval( $newId ) ) {
700 return false;
701 }
703 $len = fgets( $this->spawnRead );
704 // $this->progress( "<< " . trim( $len ) );
705 if ( $len === false ) {
706 return false;
707 }
709 $nbytes = intval( $len );
710 // actual error, not zero-length text
711 if ( $nbytes < 0 ) {
712 return false;
713 }
715 $text = "";
717 // Subprocess may not send everything at once, we have to loop.
718 while ( $nbytes > strlen( $text ) ) {
719 $buffer = fread( $this->spawnRead, $nbytes - strlen( $text ) );
720 if ( $buffer === false ) {
721 break;
722 }
723 $text .= $buffer;
724 }
726 $gotbytes = strlen( $text );
727 if ( $gotbytes != $nbytes ) {
728 $this->progress( "Expected $nbytes bytes from database subprocess, got $gotbytes " );
730 return false;
731 }
733 // Do normalization in the dump thread...
734 $stripped = str_replace( "\r", "", $text );
735 $normalized = $wgContLang->normalize( $stripped );
737 return $normalized;
738 }
740 function startElement( $parser, $name, $attribs ) {
741 $this->checkpointJustWritten = false;
743 $this->clearOpenElement( null );
744 $this->lastName = $name;
746 if ( $name == 'revision' ) {
747 $this->state = $name;
748 $this->egress->writeOpenPage( null, $this->buffer );
749 $this->buffer = "";
750 } elseif ( $name == 'page' ) {
751 $this->state = $name;
752 if ( $this->atStart ) {
753 $this->egress->writeOpenStream( $this->buffer );
754 $this->buffer = "";
755 $this->atStart = false;
756 }
757 }
759 if ( $name == "text" && isset( $attribs['id'] ) ) {
760 $text = $this->getText( $attribs['id'] );
761 $this->openElement = array( $name, array( 'xml:space' => 'preserve' ) );
762 if ( strlen( $text ) > 0 ) {
763 $this->characterData( $parser, $text );
764 }
765 } else {
766 $this->openElement = array( $name, $attribs );
767 }
768 }
770 function endElement( $parser, $name ) {
771 $this->checkpointJustWritten = false;
773 if ( $this->openElement ) {
774 $this->clearOpenElement( "" );
775 } else {
776 $this->buffer .= "</$name>";
777 }
779 if ( $name == 'revision' ) {
780 $this->egress->writeRevision( null, $this->buffer );
781 $this->buffer = "";
782 $this->thisRev = "";
783 } elseif ( $name == 'page' ) {
784 if ( !$this->firstPageWritten ) {
785 $this->firstPageWritten = trim( $this->thisPage );
786 }
787 $this->lastPageWritten = trim( $this->thisPage );
788 if ( $this->timeExceeded ) {
789 $this->egress->writeClosePage( $this->buffer );
790 // nasty hack, we can't just write the chardata after the
791 // page tag, it will include leading blanks from the next line
792 $this->egress->sink->write( "\n" );
794 $this->buffer = $this->xmlwriterobj->closeStream();
795 $this->egress->writeCloseStream( $this->buffer );
797 $this->buffer = "";
798 $this->thisPage = "";
799 // this could be more than one file if we had more than one output arg
801 $filenameList = (array)$this->egress->getFilenames();
802 $newFilenames = array();
803 $firstPageID = str_pad( $this->firstPageWritten, 9, "0", STR_PAD_LEFT );
804 $lastPageID = str_pad( $this->lastPageWritten, 9, "0", STR_PAD_LEFT );
805 $filenamesCount = count( $filenameList );
806 for ( $i = 0; $i < $filenamesCount; $i++ ) {
807 $checkpointNameFilledIn = sprintf( $this->checkpointFiles[$i], $firstPageID, $lastPageID );
808 $fileinfo = pathinfo( $filenameList[$i] );
809 $newFilenames[] = $fileinfo['dirname'] . '/' . $checkpointNameFilledIn;
810 }
811 $this->egress->closeRenameAndReopen( $newFilenames );
812 $this->buffer = $this->xmlwriterobj->openStream();
813 $this->timeExceeded = false;
814 $this->timeOfCheckpoint = $this->lastTime;
815 $this->firstPageWritten = false;
816 $this->checkpointJustWritten = true;
817 } else {
818 $this->egress->writeClosePage( $this->buffer );
819 $this->buffer = "";
820 $this->thisPage = "";
821 }
822 } elseif ( $name == 'mediawiki' ) {
823 $this->egress->writeCloseStream( $this->buffer );
824 $this->buffer = "";
825 }
826 }
828 function characterData( $parser, $data ) {
829 $this->clearOpenElement( null );
830 if ( $this->lastName == "id" ) {
831 if ( $this->state == "revision" ) {
832 $this->thisRev .= $data;
833 } elseif ( $this->state == "page" ) {
834 $this->thisPage .= $data;
835 }
836 }
837 // have to skip the newline left over from closepagetag line of
838 // end of checkpoint files. nasty hack!!
839 if ( $this->checkpointJustWritten ) {
840 if ( $data[0] == "\n" ) {
841 $data = substr( $data, 1 );
842 }
843 $this->checkpointJustWritten = false;
844 }
845 $this->buffer .= htmlspecialchars( $data );
846 }
848 function clearOpenElement( $style ) {
849 if ( $this->openElement ) {
850 $this->buffer .= Xml::element( $this->openElement[0], $this->openElement[1], $style );
851 $this->openElement = false;
852 }
853 }
854 }