| blob | 7fca377335a8e9771bd9e6228aa4591145fd2231 |
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 * @return bool
347 */
348 function readDump( $input ) {
349 $this->buffer = "";
350 $this->openElement = false;
351 $this->atStart = true;
352 $this->state = "";
353 $this->lastName = "";
354 $this->thisPage = 0;
355 $this->thisRev = 0;
357 $parser = xml_parser_create( "UTF-8" );
358 xml_parser_set_option( $parser, XML_OPTION_CASE_FOLDING, false );
360 xml_set_element_handler(
361 $parser,
362 array( &$this, 'startElement' ),
363 array( &$this, 'endElement' )
364 );
365 xml_set_character_data_handler( $parser, array( &$this, 'characterData' ) );
367 $offset = 0; // for context extraction on error reporting
368 $bufferSize = 512 * 1024;
369 do {
370 if ( $this->checkIfTimeExceeded() ) {
371 $this->setTimeExceeded();
372 }
373 $chunk = fread( $input, $bufferSize );
374 if ( !xml_parse( $parser, $chunk, feof( $input ) ) ) {
375 wfDebug( "TextDumpPass::readDump encountered XML parsing error\n" );
377 $byte = xml_get_current_byte_index( $parser );
378 $msg = wfMessage( 'xml-error-string',
379 'XML import parse failure',
380 xml_get_current_line_number( $parser ),
381 xml_get_current_column_number( $parser ),
382 $byte . ( is_null( $chunk ) ? null : ( '; "' . substr( $chunk, $byte - $offset, 16 ) . '"' ) ),
383 xml_error_string( xml_get_error_code( $parser ) ) )->escaped();
385 xml_parser_free( $parser );
387 throw new MWException( $msg );
388 }
389 $offset += strlen( $chunk );
390 } while ( $chunk !== false && !feof( $input ) );
391 if ( $this->maxTimeAllowed ) {
392 $filenameList = (array)$this->egress->getFilenames();
393 // we wrote some stuff after last checkpoint that needs renamed
394 if ( file_exists( $filenameList[0] ) ) {
395 $newFilenames = array();
396 # we might have just written the header and footer and had no
397 # pages or revisions written... perhaps they were all deleted
398 # there's no pageID 0 so we use that. the caller is responsible
399 # for deciding what to do with a file containing only the
400 # siteinfo information and the mw tags.
401 if ( !$this->firstPageWritten ) {
402 $firstPageID = str_pad( 0, 9, "0", STR_PAD_LEFT );
403 $lastPageID = str_pad( 0, 9, "0", STR_PAD_LEFT );
404 } else {
405 $firstPageID = str_pad( $this->firstPageWritten, 9, "0", STR_PAD_LEFT );
406 $lastPageID = str_pad( $this->lastPageWritten, 9, "0", STR_PAD_LEFT );
407 }
409 $filenameCount = count( $filenameList );
410 for ( $i = 0; $i < $filenameCount; $i++ ) {
411 $checkpointNameFilledIn = sprintf( $this->checkpointFiles[$i], $firstPageID, $lastPageID );
412 $fileinfo = pathinfo( $filenameList[$i] );
413 $newFilenames[] = $fileinfo['dirname'] . '/' . $checkpointNameFilledIn;
414 }
415 $this->egress->closeAndRename( $newFilenames );
416 }
417 }
418 xml_parser_free( $parser );
420 return true;
421 }
423 /**
424 * Tries to get the revision text for a revision id.
425 *
426 * Upon errors, retries (Up to $this->maxFailures tries each call).
427 * If still no good revision get could be found even after this retrying, "" is returned.
428 * If no good revision text could be returned for
429 * $this->maxConsecutiveFailedTextRetrievals consecutive calls to getText, MWException
430 * is thrown.
431 *
432 * @param string $id The revision id to get the text for
433 *
434 * @return string The revision text for $id, or ""
435 * @throws MWException
436 */
437 function getText( $id ) {
438 global $wgContentHandlerUseDB;
440 $prefetchNotTried = true; // Whether or not we already tried to get the text via prefetch.
441 $text = false; // The candidate for a good text. false if no proper value.
442 $failures = 0; // The number of times, this invocation of getText already failed.
444 // The number of times getText failed without yielding a good text in between.
445 static $consecutiveFailedTextRetrievals = 0;
447 $this->fetchCount++;
449 // To allow to simply return on success and do not have to worry about book keeping,
450 // we assume, this fetch works (possible after some retries). Nevertheless, we koop
451 // the old value, so we can restore it, if problems occur (See after the while loop).
452 $oldConsecutiveFailedTextRetrievals = $consecutiveFailedTextRetrievals;
453 $consecutiveFailedTextRetrievals = 0;
455 while ( $failures < $this->maxFailures ) {
457 // As soon as we found a good text for the $id, we will return immediately.
458 // Hence, if we make it past the try catch block, we know that we did not
459 // find a good text.
461 try {
462 // Step 1: Get some text (or reuse from previous iteratuon if checking
463 // for plausibility failed)
465 // Trying to get prefetch, if it has not been tried before
466 if ( $text === false && isset( $this->prefetch ) && $prefetchNotTried ) {
467 $prefetchNotTried = false;
468 $tryIsPrefetch = true;
469 $text = $this->prefetch->prefetch( intval( $this->thisPage ),
470 intval( $this->thisRev ) );
471 if ( $text === null ) {
472 $text = false;
473 }
474 }
476 if ( $text === false ) {
477 // Fallback to asking the database
478 $tryIsPrefetch = false;
479 if ( $this->spawn ) {
480 $text = $this->getTextSpawned( $id );
481 } else {
482 $text = $this->getTextDb( $id );
483 }
485 // No more checks for texts from DB for now.
486 // If we received something that is not false,
487 // We treat it as good text, regardless of whether it actually is or is not
488 if ( $text !== false ) {
489 return $text;
490 }
491 }
493 if ( $text === false ) {
494 throw new MWException( "Generic error while obtaining text for id " . $id );
495 }
497 // We received a good candidate for the text of $id via some method
499 // Step 2: Checking for plausibility and return the text if it is
500 // plausible
501 $revID = intval( $this->thisRev );
502 if ( !isset( $this->db ) ) {
503 throw new MWException( "No database available" );
504 }
506 $revLength = strlen( $text );
507 if ( $wgContentHandlerUseDB ) {
508 $row = $this->db->selectRow(
509 'revision',
510 array( 'rev_len', 'rev_content_model' ),
511 array( 'rev_id' => $revID ),
512 __METHOD__
513 );
514 if ( $row ) {
515 // only check the length for the wikitext content handler,
516 // it's a wasted (and failed) check otherwise
517 if ( $row->rev_content_model == CONTENT_MODEL_WIKITEXT ) {
518 $revLength = $row->rev_len;
519 }
520 }
521 } else {
522 $revLength = $this->db->selectField( 'revision', 'rev_len', array( 'rev_id' => $revID ) );
523 }
525 if ( strlen( $text ) == $revLength ) {
526 if ( $tryIsPrefetch ) {
527 $this->prefetchCount++;
528 }
530 return $text;
531 }
533 $text = false;
534 throw new MWException( "Received text is unplausible for id " . $id );
535 } catch ( Exception $e ) {
536 $msg = "getting/checking text " . $id . " failed (" . $e->getMessage() . ")";
537 if ( $failures + 1 < $this->maxFailures ) {
538 $msg .= " (Will retry " . ( $this->maxFailures - $failures - 1 ) . " more times)";
539 }
540 $this->progress( $msg );
541 }
543 // Something went wrong; we did not a text that was plausible :(
544 $failures++;
546 // A failure in a prefetch hit does not warrant resetting db connection etc.
547 if ( !$tryIsPrefetch ) {
548 // After backing off for some time, we try to reboot the whole process as
549 // much as possible to not carry over failures from one part to the other
550 // parts
551 sleep( $this->failureTimeout );
552 try {
553 $this->rotateDb();
554 if ( $this->spawn ) {
555 $this->closeSpawn();
556 $this->openSpawn();
557 }
558 } catch ( Exception $e ) {
559 $this->progress( "Rebooting getText infrastructure failed (" . $e->getMessage() . ")" .
560 " Trying to continue anyways" );
561 }
562 }
563 }
565 // Retirieving a good text for $id failed (at least) maxFailures times.
566 // We abort for this $id.
568 // Restoring the consecutive failures, and maybe aborting, if the dump
569 // is too broken.
570 $consecutiveFailedTextRetrievals = $oldConsecutiveFailedTextRetrievals + 1;
571 if ( $consecutiveFailedTextRetrievals > $this->maxConsecutiveFailedTextRetrievals ) {
572 throw new MWException( "Graceful storage failure" );
573 }
575 return "";
576 }
578 /**
579 * May throw a database error if, say, the server dies during query.
580 * @param int $id
581 * @return bool|string
582 * @throws MWException
583 */
584 private function getTextDb( $id ) {
585 global $wgContLang;
586 if ( !isset( $this->db ) ) {
587 throw new MWException( __METHOD__ . "No database available" );
588 }
589 $row = $this->db->selectRow( 'text',
590 array( 'old_text', 'old_flags' ),
591 array( 'old_id' => $id ),
592 __METHOD__ );
593 $text = Revision::getRevisionText( $row );
594 if ( $text === false ) {
595 return false;
596 }
597 $stripped = str_replace( "\r", "", $text );
598 $normalized = $wgContLang->normalize( $stripped );
600 return $normalized;
601 }
603 private function getTextSpawned( $id ) {
604 wfSuppressWarnings();
605 if ( !$this->spawnProc ) {
606 // First time?
607 $this->openSpawn();
608 }
609 $text = $this->getTextSpawnedOnce( $id );
610 wfRestoreWarnings();
612 return $text;
613 }
615 function openSpawn() {
616 global $IP;
618 if ( file_exists( "$IP/../multiversion/MWScript.php" ) ) {
619 $cmd = implode( " ",
620 array_map( 'wfEscapeShellArg',
621 array(
622 $this->php,
623 "$IP/../multiversion/MWScript.php",
624 "fetchText.php",
625 '--wiki', wfWikiID() ) ) );
626 } else {
627 $cmd = implode( " ",
628 array_map( 'wfEscapeShellArg',
629 array(
630 $this->php,
631 "$IP/maintenance/fetchText.php",
632 '--wiki', wfWikiID() ) ) );
633 }
634 $spec = array(
635 0 => array( "pipe", "r" ),
636 1 => array( "pipe", "w" ),
637 2 => array( "file", "/dev/null", "a" ) );
638 $pipes = array();
640 $this->progress( "Spawning database subprocess: $cmd" );
641 $this->spawnProc = proc_open( $cmd, $spec, $pipes );
642 if ( !$this->spawnProc ) {
643 // shit
644 $this->progress( "Subprocess spawn failed." );
646 return false;
647 }
648 list(
649 $this->spawnWrite, // -> stdin
650 $this->spawnRead, // <- stdout
651 ) = $pipes;
653 return true;
654 }
656 private function closeSpawn() {
657 wfSuppressWarnings();
658 if ( $this->spawnRead ) {
659 fclose( $this->spawnRead );
660 }
661 $this->spawnRead = false;
662 if ( $this->spawnWrite ) {
663 fclose( $this->spawnWrite );
664 }
665 $this->spawnWrite = false;
666 if ( $this->spawnErr ) {
667 fclose( $this->spawnErr );
668 }
669 $this->spawnErr = false;
670 if ( $this->spawnProc ) {
671 pclose( $this->spawnProc );
672 }
673 $this->spawnProc = false;
674 wfRestoreWarnings();
675 }
677 private function getTextSpawnedOnce( $id ) {
678 global $wgContLang;
680 $ok = fwrite( $this->spawnWrite, "$id\n" );
681 // $this->progress( ">> $id" );
682 if ( !$ok ) {
683 return false;
684 }
686 $ok = fflush( $this->spawnWrite );
687 // $this->progress( ">> [flush]" );
688 if ( !$ok ) {
689 return false;
690 }
692 // check that the text id they are sending is the one we asked for
693 // this avoids out of sync revision text errors we have encountered in the past
694 $newId = fgets( $this->spawnRead );
695 if ( $newId === false ) {
696 return false;
697 }
698 if ( $id != intval( $newId ) ) {
699 return false;
700 }
702 $len = fgets( $this->spawnRead );
703 // $this->progress( "<< " . trim( $len ) );
704 if ( $len === false ) {
705 return false;
706 }
708 $nbytes = intval( $len );
709 // actual error, not zero-length text
710 if ( $nbytes < 0 ) {
711 return false;
712 }
714 $text = "";
716 // Subprocess may not send everything at once, we have to loop.
717 while ( $nbytes > strlen( $text ) ) {
718 $buffer = fread( $this->spawnRead, $nbytes - strlen( $text ) );
719 if ( $buffer === false ) {
720 break;
721 }
722 $text .= $buffer;
723 }
725 $gotbytes = strlen( $text );
726 if ( $gotbytes != $nbytes ) {
727 $this->progress( "Expected $nbytes bytes from database subprocess, got $gotbytes " );
729 return false;
730 }
732 // Do normalization in the dump thread...
733 $stripped = str_replace( "\r", "", $text );
734 $normalized = $wgContLang->normalize( $stripped );
736 return $normalized;
737 }
739 function startElement( $parser, $name, $attribs ) {
740 $this->checkpointJustWritten = false;
742 $this->clearOpenElement( null );
743 $this->lastName = $name;
745 if ( $name == 'revision' ) {
746 $this->state = $name;
747 $this->egress->writeOpenPage( null, $this->buffer );
748 $this->buffer = "";
749 } elseif ( $name == 'page' ) {
750 $this->state = $name;
751 if ( $this->atStart ) {
752 $this->egress->writeOpenStream( $this->buffer );
753 $this->buffer = "";
754 $this->atStart = false;
755 }
756 }
758 if ( $name == "text" && isset( $attribs['id'] ) ) {
759 $text = $this->getText( $attribs['id'] );
760 $this->openElement = array( $name, array( 'xml:space' => 'preserve' ) );
761 if ( strlen( $text ) > 0 ) {
762 $this->characterData( $parser, $text );
763 }
764 } else {
765 $this->openElement = array( $name, $attribs );
766 }
767 }
769 function endElement( $parser, $name ) {
770 $this->checkpointJustWritten = false;
772 if ( $this->openElement ) {
773 $this->clearOpenElement( "" );
774 } else {
775 $this->buffer .= "</$name>";
776 }
778 if ( $name == 'revision' ) {
779 $this->egress->writeRevision( null, $this->buffer );
780 $this->buffer = "";
781 $this->thisRev = "";
782 } elseif ( $name == 'page' ) {
783 if ( !$this->firstPageWritten ) {
784 $this->firstPageWritten = trim( $this->thisPage );
785 }
786 $this->lastPageWritten = trim( $this->thisPage );
787 if ( $this->timeExceeded ) {
788 $this->egress->writeClosePage( $this->buffer );
789 // nasty hack, we can't just write the chardata after the
790 // page tag, it will include leading blanks from the next line
791 $this->egress->sink->write( "\n" );
793 $this->buffer = $this->xmlwriterobj->closeStream();
794 $this->egress->writeCloseStream( $this->buffer );
796 $this->buffer = "";
797 $this->thisPage = "";
798 // this could be more than one file if we had more than one output arg
800 $filenameList = (array)$this->egress->getFilenames();
801 $newFilenames = array();
802 $firstPageID = str_pad( $this->firstPageWritten, 9, "0", STR_PAD_LEFT );
803 $lastPageID = str_pad( $this->lastPageWritten, 9, "0", STR_PAD_LEFT );
804 $filenamesCount = count( $filenameList );
805 for ( $i = 0; $i < $filenamesCount; $i++ ) {
806 $checkpointNameFilledIn = sprintf( $this->checkpointFiles[$i], $firstPageID, $lastPageID );
807 $fileinfo = pathinfo( $filenameList[$i] );
808 $newFilenames[] = $fileinfo['dirname'] . '/' . $checkpointNameFilledIn;
809 }
810 $this->egress->closeRenameAndReopen( $newFilenames );
811 $this->buffer = $this->xmlwriterobj->openStream();
812 $this->timeExceeded = false;
813 $this->timeOfCheckpoint = $this->lastTime;
814 $this->firstPageWritten = false;
815 $this->checkpointJustWritten = true;
816 } else {
817 $this->egress->writeClosePage( $this->buffer );
818 $this->buffer = "";
819 $this->thisPage = "";
820 }
821 } elseif ( $name == 'mediawiki' ) {
822 $this->egress->writeCloseStream( $this->buffer );
823 $this->buffer = "";
824 }
825 }
827 function characterData( $parser, $data ) {
828 $this->clearOpenElement( null );
829 if ( $this->lastName == "id" ) {
830 if ( $this->state == "revision" ) {
831 $this->thisRev .= $data;
832 } elseif ( $this->state == "page" ) {
833 $this->thisPage .= $data;
834 }
835 }
836 // have to skip the newline left over from closepagetag line of
837 // end of checkpoint files. nasty hack!!
838 if ( $this->checkpointJustWritten ) {
839 if ( $data[0] == "\n" ) {
840 $data = substr( $data, 1 );
841 }
842 $this->checkpointJustWritten = false;
843 }
844 $this->buffer .= htmlspecialchars( $data );
845 }
847 function clearOpenElement( $style ) {
848 if ( $this->openElement ) {
849 $this->buffer .= Xml::element( $this->openElement[0], $this->openElement[1], $style );
850 $this->openElement = false;
851 }
852 }
853 }