maintenance/dumpTextPass.php

   1 <?php
   2 /**
   3  * Script that postprocesses XML dumps from dumpBackup.php to add page text
   4  *
   5  * Copyright (C) 2005 Brion Vibber <brion@pobox.com>
   6  * http://www.mediawiki.org/
   7  *
   8  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 2 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with this program; if not, write to the Free Software Foundation, Inc.,
  20  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  21  * http://www.gnu.org/copyleft/gpl.html
  22  *
  23  * @file
  24  * @ingroup Maintenance
  25  */
  26
  27 $originalDir = getcwd();
  28
  29 require_once( dirname( __FILE__ ) . '/commandLine.inc' );
  30 require_once( 'backup.inc' );
  31
  32 /**
  33  * @ingroup Maintenance
  34  */
  35 class TextPassDumper extends BackupDumper {
  36         var $prefetch = null;
  37         var $input = "php://stdin";
  38         var $history = WikiExporter::FULL;
  39         var $fetchCount = 0;
  40         var $prefetchCount = 0;
  41         var $prefetchCountLast = 0;
  42         var $fetchCountLast = 0;
  43
  44         var $failures = 0;
  45         var $maxFailures = 5;
  46         var $failedTextRetrievals = 0;
  47         var $maxConsecutiveFailedTextRetrievals = 200;
  48         var $failureTimeout = 5; // Seconds to sleep after db failure
  49
  50         var $php = "php";
  51         var $spawn = false;
  52         var $spawnProc = false;
  53         var $spawnWrite = false;
  54         var $spawnRead = false;
  55         var $spawnErr = false;
  56
  57         var $xmlwriterobj = false;
  58
  59         // when we spend more than maxTimeAllowed seconds on this run, we continue
  60         // processing until we write out the next complete page, then save output file(s),
  61         // rename it/them and open new one(s)
  62         var $maxTimeAllowed = 0;  // 0 = no limit
  63         var $timeExceeded = false;
  64         var $firstPageWritten = false;
  65         var $lastPageWritten = false;
  66         var $checkpointJustWritten = false;
  67         var $checkpointFiles = array();
  68
  69         /**
  70          * @var DatabaseBase
  71          */
  72         protected $db;
  73
  74         function initProgress( $history ) {
  75                 parent::initProgress();
  76                 $this->timeOfCheckpoint = $this->startTime;
  77         }
  78
  79         function dump( $history, $text = WikiExporter::TEXT ) {
  80                 // This shouldn't happen if on console... ;)
  81                 header( 'Content-type: text/html; charset=UTF-8' );
  82
  83                 // Notice messages will foul up your XML output even if they're
  84                 // relatively harmless.
  85                 if ( ini_get( 'display_errors' ) )
  86                         ini_set( 'display_errors', 'stderr' );
  87
  88                 $this->initProgress( $this->history );
  89
  90                 $this->db = $this->backupDb();
  91
  92                 $this->egress = new ExportProgressFilter( $this->sink, $this );
  93
  94                 // it would be nice to do it in the constructor, oh well. need egress set
  95                 $this->finalOptionCheck();
  96
  97                 // we only want this so we know how to close a stream :-P
  98                 $this->xmlwriterobj = new XmlDumpWriter();
  99
 100                 $input = fopen( $this->input, "rt" );
 101                 $result = $this->readDump( $input );
 102
 103                 if ( WikiError::isError( $result ) ) {
 104                         throw new MWException( $result->getMessage() );
 105                 }
 106
 107                 if ( $this->spawnProc ) {
 108                         $this->closeSpawn();
 109                 }
 110
 111                 $this->report( true );
 112         }
 113
 114         function processOption( $opt, $val, $param ) {
 115                 global $IP;
 116                 $url = $this->processFileOpt( $val, $param );
 117
 118                 switch( $opt ) {
 119                 case 'prefetch':
 120                         require_once "$IP/maintenance/backupPrefetch.inc";
 121                         $this->prefetch = new BaseDump( $url );
 122                         break;
 123                 case 'stub':
 124                         $this->input = $url;
 125                         break;
 126                 case 'maxtime':
 127                         $this->maxTimeAllowed = intval($val)*60;
 128                         break;
 129                 case 'checkpointfile':
 130                         $this->checkpointFiles[] = $val;
 131                         break;
 132                 case 'current':
 133                         $this->history = WikiExporter::CURRENT;
 134                         break;
 135                 case 'full':
 136                         $this->history = WikiExporter::FULL;
 137                         break;
 138                 case 'spawn':
 139                         $this->spawn = true;
 140                         if ( $val ) {
 141                                 $this->php = $val;
 142                         }
 143                         break;
 144                 }
 145         }
 146
 147         function processFileOpt( $val, $param ) {
 148                 $fileURIs = explode(';',$param);
 149                 foreach ( $fileURIs as $URI ) {
 150                         switch( $val ) {
 151                                 case "file":
 152                                         $newURI = $URI;
 153                                         break;
 154                                 case "gzip":
 155                                         $newURI = "compress.zlib://$URI";
 156                                         break;
 157                                 case "bzip2":
 158                                         $newURI = "compress.bzip2://$URI";
 159                                         break;
 160                                 case "7zip":
 161                                         $newURI = "mediawiki.compress.7z://$URI";
 162                                         break;
 163                                 default:
 164                                         $newURI = $URI;
 165                         }
 166                         $newFileURIs[] = $newURI;
 167                 }
 168                 $val = implode( ';', $newFileURIs );
 169                 return $val;
 170         }
 171
 172         /**
 173          * Overridden to include prefetch ratio if enabled.
 174          */
 175         function showReport() {
 176                 if ( !$this->prefetch ) {
 177                         parent::showReport();
 178                         return;
 179                 }
 180
 181                 if ( $this->reporting ) {
 182                         $now = wfTimestamp( TS_DB );
 183                         $nowts = wfTime();
 184                         $deltaAll = wfTime() - $this->startTime;
 185                         $deltaPart = wfTime() - $this->lastTime;
 186                         $this->pageCountPart = $this->pageCount - $this->pageCountLast;
 187                         $this->revCountPart = $this->revCount - $this->revCountLast;
 188
 189                         if ( $deltaAll ) {
 190                                 $portion = $this->revCount / $this->maxCount;
 191                                 $eta = $this->startTime + $deltaAll / $portion;
 192                                 $etats = wfTimestamp( TS_DB, intval( $eta ) );
 193                                 if ( $this->fetchCount ) {
 194                                         $fetchRate = 100.0 * $this->prefetchCount / $this->fetchCount;
 195                                 } else {
 196                                         $fetchRate = '-';
 197                                 }
 198                                 $pageRate = $this->pageCount / $deltaAll;
 199                                 $revRate = $this->revCount / $deltaAll;
 200                         } else {
 201                                 $pageRate = '-';
 202                                 $revRate = '-';
 203                                 $etats = '-';
 204                                 $fetchRate = '-';
 205                         }
 206                         if ( $deltaPart ) {
 207                                 if ( $this->fetchCountLast ) {
 208                                         $fetchRatePart = 100.0 * $this->prefetchCountLast / $this->fetchCountLast;
 209                                 } else {
 210                                         $fetchRatePart = '-';
 211                                 }
 212                                 $pageRatePart = $this->pageCountPart / $deltaPart;
 213                                 $revRatePart = $this->revCountPart / $deltaPart;
 214
 215                         } else {
 216                                 $fetchRatePart = '-';
 217                                 $pageRatePart = '-';
 218                                 $revRatePart = '-';
 219                         }
 220                         $this->progress( sprintf( "%s: %s (ID %d) %d pages (%0.1f|%0.1f/sec all|curr), %d revs (%0.1f|%0.1f/sec all|curr), %0.1f%%|%0.1f%% prefetched (all|curr), ETA %s [max %d]",
 221                                         $now, wfWikiID(), $this->ID, $this->pageCount, $pageRate, $pageRatePart, $this->revCount, $revRate, $revRatePart, $fetchRate, $fetchRatePart, $etats, $this->maxCount ) );
 222                         $this->lastTime = $nowts;
 223                         $this->revCountLast = $this->revCount;
 224                         $this->prefetchCountLast = $this->prefetchCount;
 225                         $this->fetchCountLast = $this->fetchCount;
 226                 }
 227         }
 228
 229         function setTimeExceeded() {
 230                 $this->timeExceeded = True;
 231         }
 232
 233         function checkIfTimeExceeded() {
 234                 if ( $this->maxTimeAllowed &&  ( $this->lastTime - $this->timeOfCheckpoint  > $this->maxTimeAllowed ) ) {
 235                         return true;
 236                 }
 237                 return false;
 238         }
 239
 240         function finalOptionCheck() {
 241                 if ( ( $this->checkpointFiles && ! $this->maxTimeAllowed ) ||
 242                         ( $this->maxTimeAllowed && !$this->checkpointFiles ) ) {
 243                         throw new MWException("Options checkpointfile and maxtime must be specified together.\n");
 244                 }
 245                 foreach ($this->checkpointFiles as $checkpointFile) {
 246                         $count = substr_count ( $checkpointFile,"%s" );
 247                         if ( $count != 2 ) {
 248                                 throw new MWException("Option checkpointfile must contain two '%s' for substitution of first and last pageids, count is $count instead, file is $checkpointFile.\n");
 249                         }
 250                 }
 251
 252                 if ( $this->checkpointFiles ) {
 253                         $filenameList = (array)$this->egress->getFilenames();
 254                         if ( count( $filenameList ) != count( $this->checkpointFiles ) ) {
 255                                 throw new MWException("One checkpointfile must be specified for each output option, if maxtime is used.\n");
 256                         }
 257                 }
 258         }
 259
 260         function readDump( $input ) {
 261                 $this->buffer = "";
 262                 $this->openElement = false;
 263                 $this->atStart = true;
 264                 $this->state = "";
 265                 $this->lastName = "";
 266                 $this->thisPage = 0;
 267                 $this->thisRev = 0;
 268
 269                 $parser = xml_parser_create( "UTF-8" );
 270                 xml_parser_set_option( $parser, XML_OPTION_CASE_FOLDING, false );
 271
 272                 xml_set_element_handler( $parser, array( &$this, 'startElement' ), array( &$this, 'endElement' ) );
 273                 xml_set_character_data_handler( $parser, array( &$this, 'characterData' ) );
 274
 275                 $offset = 0; // for context extraction on error reporting
 276                 $bufferSize = 512 * 1024;
 277                 do {
 278                         if ($this->checkIfTimeExceeded()) {
 279                                 $this->setTimeExceeded();
 280                         }
 281                         $chunk = fread( $input, $bufferSize );
 282                         if ( !xml_parse( $parser, $chunk, feof( $input ) ) ) {
 283                                 wfDebug( "TextDumpPass::readDump encountered XML parsing error\n" );
 284                                 return new WikiXmlError( $parser, 'XML import parse failure', $chunk, $offset );
 285                         }
 286                         $offset += strlen( $chunk );
 287                 } while ( $chunk !== false && !feof( $input ) );
 288                 if ($this->maxTimeAllowed) {
 289                         $filenameList = (array)$this->egress->getFilenames();
 290                         // we wrote some stuff after last checkpoint that needs renamed
 291                         if (file_exists($filenameList[0])) {
 292                                 $newFilenames = array();
 293                                 # we might have just written the header and footer and had no
 294                                 # pages or revisions written... perhaps they were all deleted
 295                                 # there's no pageID 0 so we use that. the caller is responsible
 296                                 # for deciding what to do with a file containing only the
 297                                 # siteinfo information and the mw tags.
 298                                 if (! $this->firstPageWritten) {
 299                                         $firstPageID = str_pad(0,9,"0",STR_PAD_LEFT);
 300                                         $lastPageID = str_pad(0,9,"0",STR_PAD_LEFT);
 301                                 }
 302                                 else {
 303                                         $firstPageID = str_pad($this->firstPageWritten,9,"0",STR_PAD_LEFT);
 304                                         $lastPageID = str_pad($this->lastPageWritten,9,"0",STR_PAD_LEFT);
 305                                 }
 306                                 for ( $i = 0; $i < count( $filenameList ); $i++ ) {
 307                                         $checkpointNameFilledIn = sprintf( $this->checkpointFiles[$i], $firstPageID, $lastPageID );
 308                                         $fileinfo = pathinfo($filenameList[$i]);
 309                                         $newFilenames[] = $fileinfo['dirname'] . '/' . $checkpointNameFilledIn;
 310                                 }
 311                                 $this->egress->closeAndRename( $newFilenames );
 312                         }
 313                 }
 314                 xml_parser_free( $parser );
 315
 316                 return true;
 317         }
 318
 319         function getText( $id ) {
 320                 $this->fetchCount++;
 321                 if ( isset( $this->prefetch ) ) {
 322                         $text = $this->prefetch->prefetch( $this->thisPage, $this->thisRev );
 323                         if ( $text !== null ) { // Entry missing from prefetch dump
 324                                 $dbr = wfGetDB( DB_SLAVE );
 325                                 $revID = intval( $this->thisRev );
 326                                 $revLength = $dbr->selectField( 'revision', 'rev_len', array( 'rev_id' => $revID ) );
 327                                 // if length of rev text in file doesn't match length in db, we reload
 328                                 // this avoids carrying forward broken data from previous xml dumps
 329                                 if( strlen( $text ) == $revLength ) {
 330                                         $this->prefetchCount++;
 331                                         return $text;
 332                                 }
 333                         }
 334                 }
 335                 return $this->doGetText( $id );
 336         }
 337
 338         private function doGetText( $id ) {
 339                 $id = intval( $id );
 340                 $this->failures = 0;
 341                 $ex = new MWException( "Graceful storage failure" );
 342                 while (true) {
 343                         if ( $this->spawn ) {
 344                                 if ($this->failures) {
 345                                         // we don't know why it failed, could be the child process
 346                                         // borked, could be db entry busted, could be db server out to lunch,
 347                                         // so cover all bases
 348                                         $this->closeSpawn();
 349                                         $this->openSpawn();
 350                                 }
 351                                 $text = $this->getTextSpawned( $id );
 352                         } else {
 353                                 $text = $this->getTextDbSafe( $id );
 354                         }
 355                         if ( $text === false ) {
 356                                 $this->failures++;
 357                                 if ( $this->failures > $this->maxFailures) {
 358                                         $this->progress( "Failed to retrieve revision text for text id ".
 359                                                                          "$id after $this->maxFailures tries, giving up" );
 360                                         // were there so many bad retrievals in a row we want to bail?
 361                                         // at some point we have to declare the dump irretrievably broken
 362                                         $this->failedTextRetrievals++;
 363                                         if ($this->failedTextRetrievals > $this->maxConsecutiveFailedTextRetrievals) {
 364                                                 throw $ex;
 365                                         } else {
 366                                                 // would be nice to return something better to the caller someday,
 367                                                 // log what we know about the failure and about the revision
 368                                                 return "";
 369                                         }
 370                                 } else {
 371                                         $this->progress( "Error $this->failures " .
 372                                                                  "of allowed $this->maxFailures retrieving revision text for text id $id! " .
 373                                                                  "Pausing $this->failureTimeout seconds before retry..." );
 374                                         sleep( $this->failureTimeout );
 375                                 }
 376                         } else {
 377                                 $this->failedTextRetrievals= 0;
 378                                 return $text;
 379                         }
 380                 }
 381                 return '';
 382         }
 383
 384         /**
 385          * Fetch a text revision from the database, retrying in case of failure.
 386          * This may survive some transitory errors by reconnecting, but
 387          * may not survive a long-term server outage.
 388          *
 389          * FIXME: WTF? Why is it using a loop and then returning unconditionally?
 390          */
 391         private function getTextDbSafe( $id ) {
 392                 while ( true ) {
 393                         try {
 394                                 $text = $this->getTextDb( $id );
 395                         } catch ( DBQueryError $ex ) {
 396                                 $text = false;
 397                         }
 398                         return $text;
 399                 }
 400         }
 401
 402         /**
 403          * May throw a database error if, say, the server dies during query.
 404          * @param $id
 405          * @return bool|string
 406          */
 407         private function getTextDb( $id ) {
 408                 global $wgContLang;
 409                 $row = $this->db->selectRow( 'text',
 410                         array( 'old_text', 'old_flags' ),
 411                         array( 'old_id' => $id ),
 412                         __METHOD__ );
 413                 $text = Revision::getRevisionText( $row );
 414                 if ( $text === false ) {
 415                         return false;
 416                 }
 417                 $stripped = str_replace( "\r", "", $text );
 418                 $normalized = $wgContLang->normalize( $stripped );
 419                 return $normalized;
 420         }
 421
 422         private function getTextSpawned( $id ) {
 423                 wfSuppressWarnings();
 424                 if ( !$this->spawnProc ) {
 425                         // First time?
 426                         $this->openSpawn();
 427                 }
 428                 $text = $this->getTextSpawnedOnce( $id );
 429                 wfRestoreWarnings();
 430                 return $text;
 431         }
 432
 433         function openSpawn() {
 434                 global $IP;
 435
 436                 if ( file_exists( "$IP/../multiversion/MWScript.php" ) ) {
 437                         $cmd = implode( " ",
 438                                 array_map( 'wfEscapeShellArg',
 439                                         array(
 440                                                 $this->php,
 441                                                 "$IP/../multiversion/MWScript.php",
 442                                                 "fetchText.php",
 443                                                 '--wiki', wfWikiID() ) ) );
 444                 }
 445                 else {
 446                         $cmd = implode( " ",
 447                                 array_map( 'wfEscapeShellArg',
 448                                         array(
 449                                                 $this->php,
 450                                                 "$IP/maintenance/fetchText.php",
 451                                                 '--wiki', wfWikiID() ) ) );
 452                 }
 453                 $spec = array(
 454                         0 => array( "pipe", "r" ),
 455                         1 => array( "pipe", "w" ),
 456                         2 => array( "file", "/dev/null", "a" ) );
 457                 $pipes = array();
 458
 459                 $this->progress( "Spawning database subprocess: $cmd" );
 460                 $this->spawnProc = proc_open( $cmd, $spec, $pipes );
 461                 if ( !$this->spawnProc ) {
 462                         // shit
 463                         $this->progress( "Subprocess spawn failed." );
 464                         return false;
 465                 }
 466                 list(
 467                         $this->spawnWrite, // -> stdin
 468                         $this->spawnRead,  // <- stdout
 469                 ) = $pipes;
 470
 471                 return true;
 472         }
 473
 474         private function closeSpawn() {
 475                 wfSuppressWarnings();
 476                 if ( $this->spawnRead )
 477                         fclose( $this->spawnRead );
 478                 $this->spawnRead = false;
 479                 if ( $this->spawnWrite )
 480                         fclose( $this->spawnWrite );
 481                 $this->spawnWrite = false;
 482                 if ( $this->spawnErr )
 483                         fclose( $this->spawnErr );
 484                 $this->spawnErr = false;
 485                 if ( $this->spawnProc )
 486                         pclose( $this->spawnProc );
 487                 $this->spawnProc = false;
 488                 wfRestoreWarnings();
 489         }
 490
 491         private function getTextSpawnedOnce( $id ) {
 492                 global $wgContLang;
 493
 494                 $ok = fwrite( $this->spawnWrite, "$id\n" );
 495                 // $this->progress( ">> $id" );
 496                 if ( !$ok ) return false;
 497
 498                 $ok = fflush( $this->spawnWrite );
 499                 // $this->progress( ">> [flush]" );
 500                 if ( !$ok ) return false;
 501
 502                 // check that the text id they are sending is the one we asked for
 503                 // this avoids out of sync revision text errors we have encountered in the past
 504                 $newId = fgets( $this->spawnRead );
 505                 if ( $newId === false ) {
 506                         return false;
 507                 }
 508                 if ( $id != intval( $newId ) ) {
 509                         return false;
 510                 }
 511
 512                 $len = fgets( $this->spawnRead );
 513                 // $this->progress( "<< " . trim( $len ) );
 514                 if ( $len === false ) return false;
 515
 516                 $nbytes = intval( $len );
 517                 // actual error, not zero-length text
 518                 if ($nbytes < 0 ) return false;
 519
 520                 $text = "";
 521
 522                 // Subprocess may not send everything at once, we have to loop.
 523                 while ( $nbytes > strlen( $text ) ) {
 524                         $buffer = fread( $this->spawnRead, $nbytes - strlen( $text ) );
 525                         if ( $buffer === false ) break;
 526                         $text .= $buffer;
 527                 }
 528
 529                 $gotbytes = strlen( $text );
 530                 if ( $gotbytes != $nbytes ) {
 531                         $this->progress( "Expected $nbytes bytes from database subprocess, got $gotbytes " );
 532                         return false;
 533                 }
 534
 535                 // Do normalization in the dump thread...
 536                 $stripped = str_replace( "\r", "", $text );
 537                 $normalized = $wgContLang->normalize( $stripped );
 538                 return $normalized;
 539         }
 540
 541         function startElement( $parser, $name, $attribs ) {
 542                 $this->checkpointJustWritten = false;
 543
 544                 $this->clearOpenElement( null );
 545                 $this->lastName = $name;
 546
 547                 if ( $name == 'revision' ) {
 548                         $this->state = $name;
 549                         $this->egress->writeOpenPage( null, $this->buffer );
 550                         $this->buffer = "";
 551                 } elseif ( $name == 'page' ) {
 552                         $this->state = $name;
 553                         if ( $this->atStart ) {
 554                                 $this->egress->writeOpenStream( $this->buffer );
 555                                 $this->buffer = "";
 556                                 $this->atStart = false;
 557                         }
 558                 }
 559
 560                 if ( $name == "text" && isset( $attribs['id'] ) ) {
 561                         $text = $this->getText( $attribs['id'] );
 562                         $this->openElement = array( $name, array( 'xml:space' => 'preserve' ) );
 563                         if ( strlen( $text ) > 0 ) {
 564                                 $this->characterData( $parser, $text );
 565                         }
 566                 } else {
 567                         $this->openElement = array( $name, $attribs );
 568                 }
 569         }
 570
 571         function endElement( $parser, $name ) {
 572                 $this->checkpointJustWritten = false;
 573
 574                 if ( $this->openElement ) {
 575                         $this->clearOpenElement( "" );
 576                 } else {
 577                         $this->buffer .= "</$name>";
 578                 }
 579
 580                 if ( $name == 'revision' ) {
 581                         $this->egress->writeRevision( null, $this->buffer );
 582                         $this->buffer = "";
 583                         $this->thisRev = "";
 584                 } elseif ( $name == 'page' ) {
 585                         if (! $this->firstPageWritten) {
 586                                 $this->firstPageWritten = trim($this->thisPage);
 587                         }
 588                         $this->lastPageWritten = trim($this->thisPage);
 589                         if ($this->timeExceeded) {
 590                                 $this->egress->writeClosePage( $this->buffer );
 591                                 // nasty hack, we can't just write the chardata after the
 592                                 // page tag, it will include leading blanks from the next line
 593                                 $this->egress->sink->write("\n");
 594
 595                                 $this->buffer = $this->xmlwriterobj->closeStream();
 596                                 $this->egress->writeCloseStream( $this->buffer );
 597
 598                                 $this->buffer = "";
 599                                 $this->thisPage = "";
 600                                 // this could be more than one file if we had more than one output arg
 601
 602                                 $filenameList = (array)$this->egress->getFilenames();
 603                                 $newFilenames = array();
 604                                 $firstPageID = str_pad($this->firstPageWritten,9,"0",STR_PAD_LEFT);
 605                                 $lastPageID = str_pad($this->lastPageWritten,9,"0",STR_PAD_LEFT);
 606                                 for ( $i = 0; $i < count( $filenameList ); $i++ ) {
 607                                         $checkpointNameFilledIn = sprintf( $this->checkpointFiles[$i], $firstPageID, $lastPageID );
 608                                         $fileinfo = pathinfo($filenameList[$i]);
 609                                         $newFilenames[] = $fileinfo['dirname'] . '/' . $checkpointNameFilledIn;
 610                                 }
 611                                 $this->egress->closeRenameAndReopen( $newFilenames );
 612                                 $this->buffer = $this->xmlwriterobj->openStream();
 613                                 $this->timeExceeded = false;
 614                                 $this->timeOfCheckpoint = $this->lastTime;
 615                                 $this->firstPageWritten = false;
 616                                 $this->checkpointJustWritten = true;
 617                         }
 618                         else {
 619                                 $this->egress->writeClosePage( $this->buffer );
 620                                 $this->buffer = "";
 621                                 $this->thisPage = "";
 622                         }
 623
 624                 } elseif ( $name == 'mediawiki' ) {
 625                         $this->egress->writeCloseStream( $this->buffer );
 626                         $this->buffer = "";
 627                 }
 628         }
 629
 630         function characterData( $parser, $data ) {
 631                 $this->clearOpenElement( null );
 632                 if ( $this->lastName == "id" ) {
 633                         if ( $this->state == "revision" ) {
 634                                 $this->thisRev .= $data;
 635                         } elseif ( $this->state == "page" ) {
 636                                 $this->thisPage .= $data;
 637                         }
 638                 }
 639                 // have to skip the newline left over from closepagetag line of
 640                 // end of checkpoint files. nasty hack!!
 641                 if ($this->checkpointJustWritten) {
 642                         if ($data[0] == "\n") {
 643                                 $data = substr($data,1);
 644                         }
 645                         $this->checkpointJustWritten = false;
 646                 }
 647                 $this->buffer .= htmlspecialchars( $data );
 648         }
 649
 650         function clearOpenElement( $style ) {
 651                 if ( $this->openElement ) {
 652                         $this->buffer .= Xml::element( $this->openElement[0], $this->openElement[1], $style );
 653                         $this->openElement = false;
 654                 }
 655         }
 656 }
 657
 658
 659 $dumper = new TextPassDumper( $argv );
 660
 661 if ( !isset( $options['help'] ) ) {
 662         $dumper->dump( true );
 663 } else {
 664         $dumper->progress( <<<ENDS
 665 This script postprocesses XML dumps from dumpBackup.php to add
 666 page text which was stubbed out (using --stub).
 667
 668 XML input is accepted on stdin.
 669 XML output is sent to stdout; progress reports are sent to stderr.
 670
 671 Usage: php dumpTextPass.php [<options>]
 672 Options:
 673   --stub=<type>:<file> To load a compressed stub dump instead of stdin
 674   --prefetch=<type>:<file> Use a prior dump file as a text source, to save
 675                           pressure on the database.
 676                           (Requires the XMLReader extension)
 677   --maxtime=<minutes> Write out checkpoint file after this many minutes (writing
 678                   out complete page, closing xml file properly, and opening new one
 679                   with header).  This option requires the checkpointfile option.
 680   --checkpointfile=<filenamepattern> Use this string for checkpoint filenames,
 681                       substituting first pageid written for the first %s (required) and the
 682               last pageid written for the second %s if it exists.
 683   --quiet         Don't dump status reports to stderr.
 684   --report=n  Report position and speed after every n pages processed.
 685                           (Default: 100)
 686   --server=h  Force reading from MySQL server h
 687   --current       Base ETA on number of pages in database instead of all revisions
 688   --spawn         Spawn a subprocess for loading text records
 689   --help      Display this help message
 690 ENDS
 691 );
 692 }
 693
 694