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
  42         var $failures = 0;
  43         var $maxFailures = 5;
  44         var $failedTextRetrievals = 0;
  45         var $maxConsecutiveFailedTextRetrievals = 200;
  46         var $failureTimeout = 5; // Seconds to sleep after db failure
  47
  48         var $php = "php";
  49         var $spawn = false;
  50         var $spawnProc = false;
  51         var $spawnWrite = false;
  52         var $spawnRead = false;
  53         var $spawnErr = false;
  54
  55         function dump( $history, $text = WikiExporter::TEXT ) {
  56                 # This shouldn't happen if on console... ;)
  57                 header( 'Content-type: text/html; charset=UTF-8' );
  58
  59                 # Notice messages will foul up your XML output even if they're
  60                 # relatively harmless.
  61                 if ( ini_get( 'display_errors' ) )
  62                         ini_set( 'display_errors', 'stderr' );
  63
  64                 $this->initProgress( $this->history );
  65
  66                 $this->db = $this->backupDb();
  67
  68                 $this->egress = new ExportProgressFilter( $this->sink, $this );
  69
  70                 $input = fopen( $this->input, "rt" );
  71                 $result = $this->readDump( $input );
  72
  73                 if ( WikiError::isError( $result ) ) {
  74                         wfDie( $result->getMessage() );
  75                 }
  76
  77                 if ( $this->spawnProc ) {
  78                         $this->closeSpawn();
  79                 }
  80
  81                 $this->report( true );
  82         }
  83
  84         function processOption( $opt, $val, $param ) {
  85                 global $IP;
  86                 $url = $this->processFileOpt( $val, $param );
  87
  88                 switch( $opt ) {
  89                 case 'prefetch':
  90                         require_once "$IP/maintenance/backupPrefetch.inc";
  91                         $this->prefetch = new BaseDump( $url );
  92                         break;
  93                 case 'stub':
  94                         $this->input = $url;
  95                         break;
  96                 case 'current':
  97                         $this->history = WikiExporter::CURRENT;
  98                         break;
  99                 case 'full':
 100                         $this->history = WikiExporter::FULL;
 101                         break;
 102                 case 'spawn':
 103                         $this->spawn = true;
 104                         if ( $val ) {
 105                                 $this->php = $val;
 106                         }
 107                         break;
 108                 }
 109         }
 110
 111         function processFileOpt( $val, $param ) {
 112                 switch( $val ) {
 113                 case "file":
 114                         return $param;
 115                 case "gzip":
 116                         return "compress.zlib://$param";
 117                 case "bzip2":
 118                         return "compress.bzip2://$param";
 119                 case "7zip":
 120                         return "mediawiki.compress.7z://$param";
 121                 default:
 122                         return $val;
 123                 }
 124         }
 125
 126         /**
 127          * Overridden to include prefetch ratio if enabled.
 128          */
 129         function showReport() {
 130                 if ( !$this->prefetch ) {
 131                         return parent::showReport();
 132                 }
 133
 134                 if ( $this->reporting ) {
 135                         $delta = wfTime() - $this->startTime;
 136                         $now = wfTimestamp( TS_DB );
 137                         if ( $delta ) {
 138                                 $rate = $this->pageCount / $delta;
 139                                 $revrate = $this->revCount / $delta;
 140                                 $portion = $this->revCount / $this->maxCount;
 141                                 $eta = $this->startTime + $delta / $portion;
 142                                 $etats = wfTimestamp( TS_DB, intval( $eta ) );
 143                                 $fetchrate = 100.0 * $this->prefetchCount / $this->fetchCount;
 144                         } else {
 145                                 $rate = '-';
 146                                 $revrate = '-';
 147                                 $etats = '-';
 148                                 $fetchrate = '-';
 149                         }
 150                         $this->progress( sprintf( "%s: %s %d pages (%0.3f/sec), %d revs (%0.3f/sec), %0.1f%% prefetched, ETA %s [max %d]",
 151                                 $now, wfWikiID(), $this->pageCount, $rate, $this->revCount, $revrate, $fetchrate, $etats, $this->maxCount ) );
 152                 }
 153         }
 154
 155         function readDump( $input ) {
 156                 $this->buffer = "";
 157                 $this->openElement = false;
 158                 $this->atStart = true;
 159                 $this->state = "";
 160                 $this->lastName = "";
 161                 $this->thisPage = 0;
 162                 $this->thisRev = 0;
 163
 164                 $parser = xml_parser_create( "UTF-8" );
 165                 xml_parser_set_option( $parser, XML_OPTION_CASE_FOLDING, false );
 166
 167                 xml_set_element_handler( $parser, array( &$this, 'startElement' ), array( &$this, 'endElement' ) );
 168                 xml_set_character_data_handler( $parser, array( &$this, 'characterData' ) );
 169
 170                 $offset = 0; // for context extraction on error reporting
 171                 $bufferSize = 512 * 1024;
 172                 do {
 173                         $chunk = fread( $input, $bufferSize );
 174                         if ( !xml_parse( $parser, $chunk, feof( $input ) ) ) {
 175                                 wfDebug( "TextDumpPass::readDump encountered XML parsing error\n" );
 176                                 return new WikiXmlError( $parser, 'XML import parse failure', $chunk, $offset );
 177                         }
 178                         $offset += strlen( $chunk );
 179                 } while ( $chunk !== false && !feof( $input ) );
 180                 xml_parser_free( $parser );
 181
 182                 return true;
 183         }
 184
 185         function getText( $id ) {
 186                 $this->fetchCount++;
 187                 if ( isset( $this->prefetch ) ) {
 188                         $text = $this->prefetch->prefetch( $this->thisPage, $this->thisRev );
 189                         if ( $text === null ) {
 190                                 // Entry missing from prefetch dump
 191                         } elseif ( $text === "" ) {
 192                                 // Blank entries may indicate that the prior dump was broken.
 193                                 // To be safe, reload it.
 194                         } else {
 195                                 $dbr = wfGetDB( DB_SLAVE );
 196                                 $revID = intval($this->thisRev);
 197                                 $revLength = $dbr->selectField( 'revision', 'rev_len', array('rev_id' => $revID ) );
 198                                 // if length of rev text in file doesn't match length in db, we reload
 199                                 // this avoids carrying forward broken data from previous xml dumps
 200                                 if( strlen($text) == $revLength ) {
 201                                         $this->prefetchCount++;
 202                                         return $text;
 203                                 }
 204                         }
 205                 }
 206                 return $this->doGetText( $id );
 207         }
 208
 209         private function doGetText( $id ) {
 210
 211                 $id = intval( $id );
 212                 $this->failures = 0;
 213                 $ex = new MWException( "Graceful storage failure" );
 214                 while (true) {
 215                         if ( $this->spawn ) {
 216                                 if ($this->failures) {
 217                                         // we don't know why it failed, could be the child process
 218                                         // borked, could be db entry busted, could be db server out to lunch,
 219                                         // so cover all bases
 220                                         $this->closeSpawn();
 221                                         $this->openSpawn();
 222                                 }
 223                                 $text =  $this->getTextSpawned( $id );
 224                         } else {
 225                                 $text =  $this->getTextDbSafe( $id );
 226                         }
 227                         if ( $text === false ) {
 228                                 $this->failures++;
 229                                 if ( $this->failures > $this->maxFailures) {
 230                                         $this->progress( "Failed to retrieve revision text for text id ".
 231                                                                          "$id after $this->maxFailures tries, giving up" );
 232                                         // were there so many bad retrievals in a row we want to bail?
 233                                         // at some point we have to declare the dump irretrievably broken
 234                                         $this->failedTextRetrievals++;
 235                                         if ($this->failedTextRetrievals > $this->maxConsecutiveFailedTextRetrievals) {
 236                                                 throw $ex;
 237                                         }
 238                                         else {
 239                                                 // would be nice to return something better to the caller someday,
 240                                                 // log what we know about the failure and about the revision
 241                                                 return("");
 242                                         }
 243                                 } else {
 244                                         $this->progress( "Error $this->failures " .
 245                                                                  "of allowed $this->maxFailures retrieving revision text for text id $id! " .
 246                                                                  "Pausing $this->failureTimeout seconds before retry..." );
 247                                         sleep( $this->failureTimeout );
 248                                 }
 249                         } else {
 250                                 $this->failedTextRetrievals= 0;
 251                                 return( $text );
 252                         }
 253                 }
 254
 255         }
 256
 257         /**
 258          * Fetch a text revision from the database, retrying in case of failure.
 259          * This may survive some transitory errors by reconnecting, but
 260          * may not survive a long-term server outage.
 261          */
 262         private function getTextDbSafe( $id ) {
 263                 while ( true ) {
 264                         try {
 265                                 $text = $this->getTextDb( $id );
 266                         } catch ( DBQueryError $ex ) {
 267                                 $text = false;
 268                         }
 269                         return $text;
 270                 }
 271         }
 272
 273         /**
 274          * May throw a database error if, say, the server dies during query.
 275          */
 276         private function getTextDb( $id ) {
 277                 global $wgContLang;
 278                 $row = $this->db->selectRow( 'text',
 279                         array( 'old_text', 'old_flags' ),
 280                         array( 'old_id' => $id ),
 281                         __METHOD__ );
 282                 $text = Revision::getRevisionText( $row );
 283                 if ( $text === false ) {
 284                         return false;
 285                 }
 286                 $stripped = str_replace( "\r", "", $text );
 287                 $normalized = $wgContLang->normalize( $stripped );
 288                 return $normalized;
 289         }
 290
 291         private function getTextSpawned( $id ) {
 292                 wfSuppressWarnings();
 293                 if ( !$this->spawnProc ) {
 294                         // First time?
 295                         $this->openSpawn();
 296                 }
 297                 $text = $this->getTextSpawnedOnce( $id );
 298                 wfRestoreWarnings();
 299                 return $text;
 300         }
 301
 302         function openSpawn() {
 303                 global $IP, $wgDBname;
 304
 305                 $cmd = implode( " ",
 306                         array_map( 'wfEscapeShellArg',
 307                                 array(
 308                                         $this->php,
 309                                         "$IP/maintenance/fetchText.php",
 310                                         $wgDBname ) ) );
 311                 $spec = array(
 312                         0 => array( "pipe", "r" ),
 313                         1 => array( "pipe", "w" ),
 314                         2 => array( "file", "/dev/null", "a" ) );
 315                 $pipes = array();
 316
 317                 $this->progress( "Spawning database subprocess: $cmd" );
 318                 $this->spawnProc = proc_open( $cmd, $spec, $pipes );
 319                 if ( !$this->spawnProc ) {
 320                         // shit
 321                         $this->progress( "Subprocess spawn failed." );
 322                         return false;
 323                 }
 324                 list(
 325                         $this->spawnWrite, // -> stdin
 326                         $this->spawnRead,  // <- stdout
 327                 ) = $pipes;
 328
 329                 return true;
 330         }
 331
 332         private function closeSpawn() {
 333                 wfSuppressWarnings();
 334                 if ( $this->spawnRead )
 335                         fclose( $this->spawnRead );
 336                 $this->spawnRead = false;
 337                 if ( $this->spawnWrite )
 338                         fclose( $this->spawnWrite );
 339                 $this->spawnWrite = false;
 340                 if ( $this->spawnErr )
 341                         fclose( $this->spawnErr );
 342                 $this->spawnErr = false;
 343                 if ( $this->spawnProc )
 344                         pclose( $this->spawnProc );
 345                 $this->spawnProc = false;
 346                 wfRestoreWarnings();
 347         }
 348
 349         private function getTextSpawnedOnce( $id ) {
 350                 global $wgContLang;
 351
 352                 $ok = fwrite( $this->spawnWrite, "$id\n" );
 353                 // $this->progress( ">> $id" );
 354                 if ( !$ok ) return false;
 355
 356                 $ok = fflush( $this->spawnWrite );
 357                 // $this->progress( ">> [flush]" );
 358                 if ( !$ok ) return false;
 359
 360                 // check that the text id they are sending is the one we asked for
 361                 // this avoids out of sync revision text errors we have encountered in the past
 362                 $newId = fgets( $this->spawnRead );
 363                 if ( $newId === false ) {
 364                         return false;
 365                 }
 366                 if ( $id != intval( $newId ) ) {
 367                         return false;
 368                 }
 369
 370                 $len = fgets( $this->spawnRead );
 371                 // $this->progress( "<< " . trim( $len ) );
 372                 if ( $len === false ) return false;
 373
 374                 $nbytes = intval( $len );
 375                 // actual error, not zero-length text
 376                 if ($nbytes < 0 ) return false;
 377
 378                 $text = "";
 379
 380                 // Subprocess may not send everything at once, we have to loop.
 381                 while ( $nbytes > strlen( $text ) ) {
 382                         $buffer = fread( $this->spawnRead, $nbytes - strlen( $text ) );
 383                         if ( $buffer === false ) break;
 384                         $text .= $buffer;
 385                 }
 386
 387                 $gotbytes = strlen( $text );
 388                 if ( $gotbytes != $nbytes ) {
 389                         $this->progress( "Expected $nbytes bytes from database subprocess, got $gotbytes " );
 390                         return false;
 391                 }
 392
 393                 // Do normalization in the dump thread...
 394                 $stripped = str_replace( "\r", "", $text );
 395                 $normalized = $wgContLang->normalize( $stripped );
 396                 return $normalized;
 397         }
 398
 399         function startElement( $parser, $name, $attribs ) {
 400                 $this->clearOpenElement( null );
 401                 $this->lastName = $name;
 402
 403                 if ( $name == 'revision' ) {
 404                         $this->state = $name;
 405                         $this->egress->writeOpenPage( null, $this->buffer );
 406                         $this->buffer = "";
 407                 } elseif ( $name == 'page' ) {
 408                         $this->state = $name;
 409                         if ( $this->atStart ) {
 410                                 $this->egress->writeOpenStream( $this->buffer );
 411                                 $this->buffer = "";
 412                                 $this->atStart = false;
 413                         }
 414                 }
 415
 416                 if ( $name == "text" && isset( $attribs['id'] ) ) {
 417                         $text = $this->getText( $attribs['id'] );
 418                         $this->openElement = array( $name, array( 'xml:space' => 'preserve' ) );
 419                         if ( strlen( $text ) > 0 ) {
 420                                 $this->characterData( $parser, $text );
 421                         }
 422                 } else {
 423                         $this->openElement = array( $name, $attribs );
 424                 }
 425         }
 426
 427         function endElement( $parser, $name ) {
 428                 if ( $this->openElement ) {
 429                         $this->clearOpenElement( "" );
 430                 } else {
 431                         $this->buffer .= "</$name>";
 432                 }
 433
 434                 if ( $name == 'revision' ) {
 435                         $this->egress->writeRevision( null, $this->buffer );
 436                         $this->buffer = "";
 437                         $this->thisRev = "";
 438                 } elseif ( $name == 'page' ) {
 439                         $this->egress->writeClosePage( $this->buffer );
 440                         $this->buffer = "";
 441                         $this->thisPage = "";
 442                 } elseif ( $name == 'mediawiki' ) {
 443                         $this->egress->writeCloseStream( $this->buffer );
 444                         $this->buffer = "";
 445                 }
 446         }
 447
 448         function characterData( $parser, $data ) {
 449                 $this->clearOpenElement( null );
 450                 if ( $this->lastName == "id" ) {
 451                         if ( $this->state == "revision" ) {
 452                                 $this->thisRev .= $data;
 453                         } elseif ( $this->state == "page" ) {
 454                                 $this->thisPage .= $data;
 455                         }
 456                 }
 457                 $this->buffer .= htmlspecialchars( $data );
 458         }
 459
 460         function clearOpenElement( $style ) {
 461                 if ( $this->openElement ) {
 462                         $this->buffer .= Xml::element( $this->openElement[0], $this->openElement[1], $style );
 463                         $this->openElement = false;
 464                 }
 465         }
 466 }
 467
 468
 469 $dumper = new TextPassDumper( $argv );
 470
 471 if ( !isset( $options['help'] ) ) {
 472         $dumper->dump( true );
 473 } else {
 474         $dumper->progress( <<<ENDS
 475 This script postprocesses XML dumps from dumpBackup.php to add
 476 page text which was stubbed out (using --stub).
 477
 478 XML input is accepted on stdin.
 479 XML output is sent to stdout; progress reports are sent to stderr.
 480
 481 Usage: php dumpTextPass.php [<options>]
 482 Options:
 483   --stub=<type>:<file> To load a compressed stub dump instead of stdin
 484   --prefetch=<type>:<file> Use a prior dump file as a text source, to save
 485                           pressure on the database.
 486                           (Requires the XMLReader extension)
 487   --quiet         Don't dump status reports to stderr.
 488   --report=n  Report position and speed after every n pages processed.
 489                           (Default: 100)
 490   --server=h  Force reading from MySQL server h
 491   --current       Base ETA on number of pages in database instead of all revisions
 492   --spawn         Spawn a subprocess for loading text records
 493   --help      Display this help message
 494 ENDS
 495 );
 496 }
 497
 498