includes/Export.php

   1 <?php
   2 /**
   3  * Base classes for dumps and export
   4  *
   5  * Copyright © 2003, 2005, 2006 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  */
  25
  26 /**
  27  * @defgroup Dump Dump
  28  */
  29
  30 /**
  31  * @ingroup SpecialPage Dump
  32  */
  33 class WikiExporter {
  34         var $list_authors = false ; # Return distinct author list (when not returning full history)
  35         var $author_list = "" ;
  36
  37         var $dumpUploads = false;
  38         var $dumpUploadFileContents = false;
  39
  40         const FULL = 1;
  41         const CURRENT = 2;
  42         const STABLE = 4; // extension defined
  43         const LOGS = 8;
  44
  45         const BUFFER = 0;
  46         const STREAM = 1;
  47
  48         const TEXT = 0;
  49         const STUB = 1;
  50
  51         /**
  52          * If using WikiExporter::STREAM to stream a large amount of data,
  53          * provide a database connection which is not managed by
  54          * LoadBalancer to read from: some history blob types will
  55          * make additional queries to pull source data while the
  56          * main query is still running.
  57          *
  58          * @param $db Database
  59          * @param $history Mixed: one of WikiExporter::FULL or WikiExporter::CURRENT,
  60          *                 or an associative array:
  61          *                   offset: non-inclusive offset at which to start the query
  62          *                   limit: maximum number of rows to return
  63          *                   dir: "asc" or "desc" timestamp order
  64          * @param $buffer Int: one of WikiExporter::BUFFER or WikiExporter::STREAM
  65          * @param $text Int: one of WikiExporter::TEXT or WikiExporter::STUB
  66          */
  67         function __construct( &$db, $history = WikiExporter::CURRENT,
  68                         $buffer = WikiExporter::BUFFER, $text = WikiExporter::TEXT ) {
  69                 $this->db =& $db;
  70                 $this->history = $history;
  71                 $this->buffer  = $buffer;
  72                 $this->writer  = new XmlDumpWriter();
  73                 $this->sink    = new DumpOutput();
  74                 $this->text    = $text;
  75         }
  76
  77         /**
  78          * Set the DumpOutput or DumpFilter object which will receive
  79          * various row objects and XML output for filtering. Filters
  80          * can be chained or used as callbacks.
  81          *
  82          * @param $sink mixed
  83          */
  84         public function setOutputSink( &$sink ) {
  85                 $this->sink =& $sink;
  86         }
  87
  88         public function openStream() {
  89                 $output = $this->writer->openStream();
  90                 $this->sink->writeOpenStream( $output );
  91         }
  92
  93         public function closeStream() {
  94                 $output = $this->writer->closeStream();
  95                 $this->sink->writeCloseStream( $output );
  96         }
  97
  98         /**
  99          * Dumps a series of page and revision records for all pages
 100          * in the database, either including complete history or only
 101          * the most recent version.
 102          */
 103         public function allPages() {
 104                 return $this->dumpFrom( '' );
 105         }
 106
 107         /**
 108          * Dumps a series of page and revision records for those pages
 109          * in the database falling within the page_id range given.
 110          * @param $start Int: inclusive lower limit (this id is included)
 111          * @param $end   Int: Exclusive upper limit (this id is not included)
 112          *                   If 0, no upper limit.
 113          */
 114         public function pagesByRange( $start, $end ) {
 115                 $condition = 'page_id >= ' . intval( $start );
 116                 if ( $end ) {
 117                         $condition .= ' AND page_id < ' . intval( $end );
 118                 }
 119                 return $this->dumpFrom( $condition );
 120         }
 121
 122         /**
 123          * @param $title Title
 124          */
 125         public function pageByTitle( $title ) {
 126                 return $this->dumpFrom(
 127                         'page_namespace=' . $title->getNamespace() .
 128                         ' AND page_title=' . $this->db->addQuotes( $title->getDBkey() ) );
 129         }
 130
 131         public function pageByName( $name ) {
 132                 $title = Title::newFromText( $name );
 133                 if ( is_null( $title ) ) {
 134                         throw new MWException( "Can't export invalid title" );
 135                 } else {
 136                         return $this->pageByTitle( $title );
 137                 }
 138         }
 139
 140         public function pagesByName( $names ) {
 141                 foreach ( $names as $name ) {
 142                         $this->pageByName( $name );
 143                 }
 144         }
 145
 146         public function allLogs() {
 147                 return $this->dumpFrom( '' );
 148         }
 149
 150         public function logsByRange( $start, $end ) {
 151                 $condition = 'log_id >= ' . intval( $start );
 152                 if ( $end ) {
 153                         $condition .= ' AND log_id < ' . intval( $end );
 154                 }
 155                 return $this->dumpFrom( $condition );
 156         }
 157
 158         # Generates the distinct list of authors of an article
 159         # Not called by default (depends on $this->list_authors)
 160         # Can be set by Special:Export when not exporting whole history
 161         protected function do_list_authors( $cond ) {
 162                 wfProfileIn( __METHOD__ );
 163                 $this->author_list = "<contributors>";
 164                 // rev_deleted
 165
 166                 $res = $this->db->select(
 167                         array( 'page', 'revision' ),
 168                         array( 'DISTINCT rev_user_text', 'rev_user' ),
 169                         array(
 170                                 $this->db->bitAnd( 'rev_deleted', Revision::DELETED_USER ) . ' = 0',
 171                                 $cond,
 172                                 'page_id = rev_id',
 173                         ),
 174                         __METHOD__
 175                 );
 176
 177                 foreach ( $res as $row ) {
 178                         $this->author_list .= "<contributor>" .
 179                                 "<username>" .
 180                                 htmlentities( $row->rev_user_text )  .
 181                                 "</username>" .
 182                                 "<id>" .
 183                                 $row->rev_user .
 184                                 "</id>" .
 185                                 "</contributor>";
 186                 }
 187                 $this->author_list .= "</contributors>";
 188                 wfProfileOut( __METHOD__ );
 189         }
 190
 191         protected function dumpFrom( $cond = '' ) {
 192                 wfProfileIn( __METHOD__ );
 193                 # For logging dumps...
 194                 if ( $this->history & self::LOGS ) {
 195                         if ( $this->buffer == WikiExporter::STREAM ) {
 196                                 $prev = $this->db->bufferResults( false );
 197                         }
 198                         $where = array( 'user_id = log_user' );
 199                         # Hide private logs
 200                         $hideLogs = LogEventsList::getExcludeClause( $this->db );
 201                         if ( $hideLogs ) $where[] = $hideLogs;
 202                         # Add on any caller specified conditions
 203                         if ( $cond ) $where[] = $cond;
 204                         # Get logging table name for logging.* clause
 205                         $logging = $this->db->tableName( 'logging' );
 206                         $result = $this->db->select( array( 'logging', 'user' ),
 207                                 array( "{$logging}.*", 'user_name' ), // grab the user name
 208                                 $where,
 209                                 __METHOD__,
 210                                 array( 'ORDER BY' => 'log_id', 'USE INDEX' => array( 'logging' => 'PRIMARY' ) )
 211                         );
 212                         $wrapper = $this->db->resultObject( $result );
 213                         $this->outputLogStream( $wrapper );
 214                         if ( $this->buffer == WikiExporter::STREAM ) {
 215                                 $this->db->bufferResults( $prev );
 216                         }
 217                 # For page dumps...
 218                 } else {
 219                         $tables = array( 'page', 'revision' );
 220                         $opts = array( 'ORDER BY' => 'page_id ASC' );
 221                         $opts['USE INDEX'] = array();
 222                         $join = array();
 223                         if ( is_array( $this->history ) ) {
 224                                 # Time offset/limit for all pages/history...
 225                                 $revJoin = 'page_id=rev_page';
 226                                 # Set time order
 227                                 if ( $this->history['dir'] == 'asc' ) {
 228                                         $op = '>';
 229                                         $opts['ORDER BY'] = 'rev_timestamp ASC';
 230                                 } else {
 231                                         $op = '<';
 232                                         $opts['ORDER BY'] = 'rev_timestamp DESC';
 233                                 }
 234                                 # Set offset
 235                                 if ( !empty( $this->history['offset'] ) ) {
 236                                         $revJoin .= " AND rev_timestamp $op " .
 237                                                 $this->db->addQuotes( $this->db->timestamp( $this->history['offset'] ) );
 238                                 }
 239                                 $join['revision'] = array( 'INNER JOIN', $revJoin );
 240                                 # Set query limit
 241                                 if ( !empty( $this->history['limit'] ) ) {
 242                                         $opts['LIMIT'] = intval( $this->history['limit'] );
 243                                 }
 244                         } elseif ( $this->history & WikiExporter::FULL ) {
 245                                 # Full history dumps...
 246                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page' );
 247                         } elseif ( $this->history & WikiExporter::CURRENT ) {
 248                                 # Latest revision dumps...
 249                                 if ( $this->list_authors && $cond != '' )  { // List authors, if so desired
 250                                         $this->do_list_authors( $cond );
 251                                 }
 252                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' );
 253                         } elseif ( $this->history & WikiExporter::STABLE ) {
 254                                 # "Stable" revision dumps...
 255                                 # Default JOIN, to be overridden...
 256                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' );
 257                                 # One, and only one hook should set this, and return false
 258                                 if ( wfRunHooks( 'WikiExporter::dumpStableQuery', array( &$tables, &$opts, &$join ) ) ) {
 259                                         wfProfileOut( __METHOD__ );
 260                                         throw new MWException( __METHOD__ . " given invalid history dump type." );
 261                                 }
 262                         } else {
 263                                 # Uknown history specification parameter?
 264                                 wfProfileOut( __METHOD__ );
 265                                 throw new MWException( __METHOD__ . " given invalid history dump type." );
 266                         }
 267                         # Query optimization hacks
 268                         if ( $cond == '' ) {
 269                                 $opts[] = 'STRAIGHT_JOIN';
 270                                 $opts['USE INDEX']['page'] = 'PRIMARY';
 271                         }
 272                         # Build text join options
 273                         if ( $this->text != WikiExporter::STUB ) { // 1-pass
 274                                 $tables[] = 'text';
 275                                 $join['text'] = array( 'INNER JOIN', 'rev_text_id=old_id' );
 276                         }
 277
 278                         if ( $this->buffer == WikiExporter::STREAM ) {
 279                                 $prev = $this->db->bufferResults( false );
 280                         }
 281
 282                         wfRunHooks( 'ModifyExportQuery',
 283                                                 array( $this->db, &$tables, &$cond, &$opts, &$join ) );
 284
 285                         # Do the query!
 286                         $result = $this->db->select( $tables, '*', $cond, __METHOD__, $opts, $join );
 287                         $wrapper = $this->db->resultObject( $result );
 288                         # Output dump results
 289                         $this->outputPageStream( $wrapper );
 290                         if ( $this->list_authors ) {
 291                                 $this->outputPageStream( $wrapper );
 292                         }
 293
 294                         if ( $this->buffer == WikiExporter::STREAM ) {
 295                                 $this->db->bufferResults( $prev );
 296                         }
 297                 }
 298                 wfProfileOut( __METHOD__ );
 299         }
 300
 301         /**
 302          * Runs through a query result set dumping page and revision records.
 303          * The result set should be sorted/grouped by page to avoid duplicate
 304          * page records in the output.
 305          *
 306          * The result set will be freed once complete. Should be safe for
 307          * streaming (non-buffered) queries, as long as it was made on a
 308          * separate database connection not managed by LoadBalancer; some
 309          * blob storage types will make queries to pull source data.
 310          *
 311          * @param $resultset ResultWrapper
 312          */
 313         protected function outputPageStream( $resultset ) {
 314                 $last = null;
 315                 foreach ( $resultset as $row ) {
 316                         if ( is_null( $last ) ||
 317                                 $last->page_namespace != $row->page_namespace ||
 318                                 $last->page_title     != $row->page_title ) {
 319                                 if ( isset( $last ) ) {
 320                                         $output = '';
 321                                         if ( $this->dumpUploads ) {
 322                                                 $output .= $this->writer->writeUploads( $last, $this->dumpUploadFileContents );
 323                                         }
 324                                         $output .= $this->writer->closePage();
 325                                         $this->sink->writeClosePage( $output );
 326                                 }
 327                                 $output = $this->writer->openPage( $row );
 328                                 $this->sink->writeOpenPage( $row, $output );
 329                                 $last = $row;
 330                         }
 331                         $output = $this->writer->writeRevision( $row );
 332                         $this->sink->writeRevision( $row, $output );
 333                 }
 334                 if ( isset( $last ) ) {
 335                         $output = '';
 336                         if ( $this->dumpUploads ) {
 337                                 $output .= $this->writer->writeUploads( $last, $this->dumpUploadFileContents );
 338                         }
 339                         $output .= $this->author_list;
 340                         $output .= $this->writer->closePage();
 341                         $this->sink->writeClosePage( $output );
 342                 }
 343         }
 344
 345         protected function outputLogStream( $resultset ) {
 346                 foreach ( $resultset as $row ) {
 347                         $output = $this->writer->writeLogItem( $row );
 348                         $this->sink->writeLogItem( $row, $output );
 349                 }
 350         }
 351 }
 352
 353 /**
 354  * @ingroup Dump
 355  */
 356 class XmlDumpWriter {
 357         protected $firstPageWritten = 0, $lastPageWritten = 0, $pageInProgress = 0;
 358
 359         /**
 360          * Returns the export schema version.
 361          * @return string
 362          */
 363         function schemaVersion() {
 364                 return "0.5";
 365         }
 366
 367         /**
 368          * Opens the XML output stream's root <mediawiki> element.
 369          * This does not include an xml directive, so is safe to include
 370          * as a subelement in a larger XML stream. Namespace and XML Schema
 371          * references are included.
 372          *
 373          * Output will be encoded in UTF-8.
 374          *
 375          * @return string
 376          */
 377         function openStream() {
 378                 global $wgLanguageCode;
 379                 $ver = $this->schemaVersion();
 380                 return Xml::element( 'mediawiki', array(
 381                         'xmlns'              => "http://www.mediawiki.org/xml/export-$ver/",
 382                         'xmlns:xsi'          => "http://www.w3.org/2001/XMLSchema-instance",
 383                         'xsi:schemaLocation' => "http://www.mediawiki.org/xml/export-$ver/ " .
 384                                                 "http://www.mediawiki.org/xml/export-$ver.xsd",
 385                         'version'            => $ver,
 386                         'xml:lang'           => $wgLanguageCode ),
 387                         null ) .
 388                         "\n" .
 389                         $this->siteInfo();
 390         }
 391
 392         function siteInfo() {
 393                 $info = array(
 394                         $this->sitename(),
 395                         $this->homelink(),
 396                         $this->generator(),
 397                         $this->caseSetting(),
 398                         $this->namespaces() );
 399                 return "  <siteinfo>\n    " .
 400                         implode( "\n    ", $info ) .
 401                         "\n  </siteinfo>\n";
 402         }
 403
 404         function sitename() {
 405                 global $wgSitename;
 406                 return Xml::element( 'sitename', array(), $wgSitename );
 407         }
 408
 409         function generator() {
 410                 global $wgVersion;
 411                 return Xml::element( 'generator', array(), "MediaWiki $wgVersion" );
 412         }
 413
 414         function homelink() {
 415                 return Xml::element( 'base', array(), Title::newMainPage()->getCanonicalUrl() );
 416         }
 417
 418         function caseSetting() {
 419                 global $wgCapitalLinks;
 420                 // "case-insensitive" option is reserved for future
 421                 $sensitivity = $wgCapitalLinks ? 'first-letter' : 'case-sensitive';
 422                 return Xml::element( 'case', array(), $sensitivity );
 423         }
 424
 425         function namespaces() {
 426                 global $wgContLang;
 427                 $spaces = "<namespaces>\n";
 428                 foreach ( $wgContLang->getFormattedNamespaces() as $ns => $title ) {
 429                         $spaces .= '      ' .
 430                                 Xml::element( 'namespace',
 431                                         array(  'key' => $ns,
 432                                                         'case' => MWNamespace::isCapitalized( $ns ) ? 'first-letter' : 'case-sensitive',
 433                                         ), $title ) . "\n";
 434                 }
 435                 $spaces .= "    </namespaces>";
 436                 return $spaces;
 437         }
 438
 439         /**
 440          * Closes the output stream with the closing root element.
 441          * Call when finished dumping things.
 442          *
 443          * @return string
 444          */
 445         function closeStream() {
 446                 return "</mediawiki>\n";
 447         }
 448
 449         /**
 450          * Opens a <page> section on the output stream, with data
 451          * from the given database row.
 452          *
 453          * @param $row object
 454          * @return string
 455          * @access private
 456          */
 457         function openPage( $row ) {
 458                 $out = "  <page>\n";
 459                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 460                 $out .= '    ' . Xml::elementClean( 'title', array(), $title->getPrefixedText() ) . "\n";
 461                 $out .= '    ' . Xml::element( 'id', array(), strval( $row->page_id ) ) . "\n";
 462                 $this->pageInProgress = $row->page_id;
 463                 if ( $row->page_is_redirect ) {
 464                         $out .= '    ' . Xml::element( 'redirect', array() ) . "\n";
 465                 }
 466                 if ( $row->page_restrictions != '' ) {
 467                         $out .= '    ' . Xml::element( 'restrictions', array(),
 468                                 strval( $row->page_restrictions ) ) . "\n";
 469                 }
 470
 471                 wfRunHooks( 'XmlDumpWriterOpenPage', array( $this, &$out, $row, $title ) );
 472
 473                 return $out;
 474         }
 475
 476         /**
 477          * Closes a <page> section on the output stream.
 478          *
 479          * @access private
 480          */
 481         function closePage() {
 482                 return "  </page>\n";
 483                 if ( !$this->firstPageWritten ) {
 484                         $this->firstPageWritten = $this->pageInProgress;
 485                 }
 486                 $this->lastPageWritten = $this->pageInProgress;
 487         }
 488
 489         /**
 490          * Dumps a <revision> section on the output stream, with
 491          * data filled in from the given database row.
 492          *
 493          * @param $row object
 494          * @return string
 495          * @access private
 496          */
 497         function writeRevision( $row ) {
 498                 wfProfileIn( __METHOD__ );
 499
 500                 $out  = "    <revision>\n";
 501                 $out .= "      " . Xml::element( 'id', null, strval( $row->rev_id ) ) . "\n";
 502
 503                 $out .= $this->writeTimestamp( $row->rev_timestamp );
 504
 505                 if ( $row->rev_deleted & Revision::DELETED_USER ) {
 506                         $out .= "      " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 507                 } else {
 508                         $out .= $this->writeContributor( $row->rev_user, $row->rev_user_text );
 509                 }
 510
 511                 if ( $row->rev_minor_edit ) {
 512                         $out .=  "      <minor/>\n";
 513                 }
 514                 if ( $row->rev_deleted & Revision::DELETED_COMMENT ) {
 515                         $out .= "      " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 516                 } elseif ( $row->rev_comment != '' ) {
 517                         $out .= "      " . Xml::elementClean( 'comment', null, strval( $row->rev_comment ) ) . "\n";
 518                 }
 519
 520                 $text = '';
 521                 if ( $row->rev_deleted & Revision::DELETED_TEXT ) {
 522                         $out .= "      " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 523                 } elseif ( isset( $row->old_text ) ) {
 524                         // Raw text from the database may have invalid chars
 525                         $text = strval( Revision::getRevisionText( $row ) );
 526                         $out .= "      " . Xml::elementClean( 'text',
 527                                 array( 'xml:space' => 'preserve', 'bytes' => $row->rev_len ),
 528                                 strval( $text ) ) . "\n";
 529                 } else {
 530                         // Stub output
 531                         $out .= "      " . Xml::element( 'text',
 532                                 array( 'id' => $row->rev_text_id, 'bytes' => $row->rev_len ),
 533                                 "" ) . "\n";
 534                 }
 535
 536                 wfRunHooks( 'XmlDumpWriterWriteRevision', array( &$this, &$out, $row, $text ) );
 537
 538                 $out .= "    </revision>\n";
 539
 540                 wfProfileOut( __METHOD__ );
 541                 return $out;
 542         }
 543
 544         /**
 545          * Dumps a <logitem> section on the output stream, with
 546          * data filled in from the given database row.
 547          *
 548          * @param $row object
 549          * @return string
 550          * @access private
 551          */
 552         function writeLogItem( $row ) {
 553                 wfProfileIn( __METHOD__ );
 554
 555                 $out  = "    <logitem>\n";
 556                 $out .= "      " . Xml::element( 'id', null, strval( $row->log_id ) ) . "\n";
 557
 558                 $out .= $this->writeTimestamp( $row->log_timestamp );
 559
 560                 if ( $row->log_deleted & LogPage::DELETED_USER ) {
 561                         $out .= "      " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 562                 } else {
 563                         $out .= $this->writeContributor( $row->log_user, $row->user_name );
 564                 }
 565
 566                 if ( $row->log_deleted & LogPage::DELETED_COMMENT ) {
 567                         $out .= "      " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 568                 } elseif ( $row->log_comment != '' ) {
 569                         $out .= "      " . Xml::elementClean( 'comment', null, strval( $row->log_comment ) ) . "\n";
 570                 }
 571
 572                 $out .= "      " . Xml::element( 'type', null, strval( $row->log_type ) ) . "\n";
 573                 $out .= "      " . Xml::element( 'action', null, strval( $row->log_action ) ) . "\n";
 574
 575                 if ( $row->log_deleted & LogPage::DELETED_ACTION ) {
 576                         $out .= "      " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 577                 } else {
 578                         $title = Title::makeTitle( $row->log_namespace, $row->log_title );
 579                         $out .= "      " . Xml::elementClean( 'logtitle', null, $title->getPrefixedText() ) . "\n";
 580                         $out .= "      " . Xml::elementClean( 'params',
 581                                 array( 'xml:space' => 'preserve' ),
 582                                 strval( $row->log_params ) ) . "\n";
 583                 }
 584
 585                 $out .= "    </logitem>\n";
 586
 587                 wfProfileOut( __METHOD__ );
 588                 return $out;
 589         }
 590
 591         function writeTimestamp( $timestamp ) {
 592                 $ts = wfTimestamp( TS_ISO_8601, $timestamp );
 593                 return "      " . Xml::element( 'timestamp', null, $ts ) . "\n";
 594         }
 595
 596         function writeContributor( $id, $text ) {
 597                 $out = "      <contributor>\n";
 598                 if ( $id ) {
 599                         $out .= "        " . Xml::elementClean( 'username', null, strval( $text ) ) . "\n";
 600                         $out .= "        " . Xml::element( 'id', null, strval( $id ) ) . "\n";
 601                 } else {
 602                         $out .= "        " . Xml::elementClean( 'ip', null, strval( $text ) ) . "\n";
 603                 }
 604                 $out .= "      </contributor>\n";
 605                 return $out;
 606         }
 607
 608         /**
 609          * Warning! This data is potentially inconsistent. :(
 610          */
 611         function writeUploads( $row, $dumpContents = false ) {
 612                 if ( $row->page_namespace == NS_IMAGE ) {
 613                         $img = wfLocalFile( $row->page_title );
 614                         if ( $img && $img->exists() ) {
 615                                 $out = '';
 616                                 foreach ( array_reverse( $img->getHistory() ) as $ver ) {
 617                                         $out .= $this->writeUpload( $ver, $dumpContents );
 618                                 }
 619                                 $out .= $this->writeUpload( $img, $dumpContents );
 620                                 return $out;
 621                         }
 622                 }
 623                 return '';
 624         }
 625
 626         /**
 627          * @param $file File
 628          * @param $dumpContents bool
 629          * @return string
 630          */
 631         function writeUpload( $file, $dumpContents = false ) {
 632                 if ( $file->isOld() ) {
 633                         $archiveName = "      " .
 634                                 Xml::element( 'archivename', null, $file->getArchiveName() ) . "\n";
 635                 } else {
 636                         $archiveName = '';
 637                 }
 638                 if ( $dumpContents ) {
 639                         # Dump file as base64
 640                         # Uses only XML-safe characters, so does not need escaping
 641                         $contents = '      <contents encoding="base64">' .
 642                                 chunk_split( base64_encode( file_get_contents( $file->getPath() ) ) ) .
 643                                 "      </contents>\n";
 644                 } else {
 645                         $contents = '';
 646                 }
 647                 return "    <upload>\n" .
 648                         $this->writeTimestamp( $file->getTimestamp() ) .
 649                         $this->writeContributor( $file->getUser( 'id' ), $file->getUser( 'text' ) ) .
 650                         "      " . Xml::elementClean( 'comment', null, $file->getDescription() ) . "\n" .
 651                         "      " . Xml::element( 'filename', null, $file->getName() ) . "\n" .
 652                         $archiveName .
 653                         "      " . Xml::element( 'src', null, $file->getCanonicalUrl() ) . "\n" .
 654                         "      " . Xml::element( 'size', null, $file->getSize() ) . "\n" .
 655                         "      " . Xml::element( 'sha1base36', null, $file->getSha1() ) . "\n" .
 656                         "      " . Xml::element( 'rel', null, $file->getRel() ) . "\n" .
 657                         $contents .
 658                         "    </upload>\n";
 659         }
 660
 661 }
 662
 663
 664 /**
 665  * Base class for output stream; prints to stdout or buffer or whereever.
 666  * @ingroup Dump
 667  */
 668 class DumpOutput {
 669         function writeOpenStream( $string ) {
 670                 $this->write( $string );
 671         }
 672
 673         function writeCloseStream( $string ) {
 674                 $this->write( $string );
 675         }
 676
 677         function writeOpenPage( $page, $string ) {
 678                 $this->write( $string );
 679         }
 680
 681         function writeClosePage( $string ) {
 682                 $this->write( $string );
 683         }
 684
 685         function writeRevision( $rev, $string ) {
 686                 $this->write( $string );
 687         }
 688
 689         function writeLogItem( $rev, $string ) {
 690                 $this->write( $string );
 691         }
 692
 693         /**
 694          * Override to write to a different stream type.
 695          * @return bool
 696          */
 697         function write( $string ) {
 698                 print $string;
 699         }
 700
 701         /**
 702          * Close the old file, move it to a specified name,
 703          * and reopen new file with the old name. Use this
 704          * for writing out a file in multiple pieces
 705          * at specified checkpoints (e.g. every n hours).
 706          * @param $newname mixed File name. May be a string or an array with one element
 707          */
 708         function closeRenameAndReopen( $newname ) {
 709                 return;
 710         }
 711
 712         // TODO: document
 713         function closeAndRename( $newname ) {
 714                 return;
 715         }
 716
 717         // TODO: document
 718         function rename( $newname ) {
 719                 return;
 720         }
 721
 722         // TODO: document
 723         function getFilename() {
 724                 return NULL;
 725         }
 726 }
 727
 728 /**
 729  * Stream outputter to send data to a file.
 730  * @ingroup Dump
 731  */
 732 class DumpFileOutput extends DumpOutput {
 733         var $handle;
 734         var $filename;
 735
 736         function __construct( $file ) {
 737                 $this->handle = fopen( $file, "wt" );
 738                 $this->filename = $file;
 739         }
 740
 741         function write( $string ) {
 742                 fputs( $this->handle, $string );
 743         }
 744
 745         function closeRenameAndReopen( $newname ) {
 746                 if ( is_array( $newname ) ) {
 747                         if ( count( $newname ) > 1 ) {
 748                                 throw new MWException( __METHOD__ . ": passed multiple arguments for rename of single file\n" );
 749                         } else {
 750                                 $newname = $newname[0];
 751                         }
 752                 }
 753                 if ( $newname ) {
 754                         fclose( $this->handle );
 755                         rename( $this->filename, $newname );
 756                         $this->handle = fopen( $this->filename, "wt" );
 757                 }
 758         }
 759
 760         function closeAndRename( $newname ) {
 761                 if ( is_array( $newname ) ) {
 762                         if ( count( $newname ) > 1 ) {
 763                                 throw new MWException( __METHOD__ . ": passed multiple arguments for rename of single file\n" );
 764                         } else {
 765                                 $newname = $newname[0];
 766                         }
 767                 }
 768                 if ( $newname ) {
 769                         fclose( $this->handle );
 770                         rename( $this->filename, $newname );
 771                 }
 772         }
 773
 774         function rename( $newname ) {
 775                 if ( is_array( $newname ) ) {
 776                         if ( count( $newname ) > 1 ) {
 777                                 throw new MWException( __METHOD__ . ": passed multiple arguments for rename of single file\n" );
 778                         } else {
 779                                 $newname = $newname[0];
 780                         }
 781                 }
 782                 if ( $newname ) {
 783                         rename( $this->filename, $newname );
 784                 }
 785         }
 786
 787         function getFilename() {
 788                 return $this->filename;
 789         }
 790 }
 791
 792 /**
 793  * Stream outputter to send data to a file via some filter program.
 794  * Even if compression is available in a library, using a separate
 795  * program can allow us to make use of a multi-processor system.
 796  * @ingroup Dump
 797  */
 798 class DumpPipeOutput extends DumpFileOutput {
 799         protected $command, $filename;
 800
 801         function __construct( $command, $file = null ) {
 802                 if ( !is_null( $file ) ) {
 803                         $command .=  " > " . wfEscapeShellArg( $file );
 804                 }
 805
 806                 $this->startCommand( $command );
 807                 $this->command = $command;
 808                 $this->filename = $file;
 809         }
 810
 811         function startCommand( $command ) {
 812                 $spec = array(
 813                         0 => array( "pipe", "r" ),
 814                 );
 815                 $pipes = array();
 816                 $this->procOpenResource = proc_open( $command, $spec, $pipes );
 817                 $this->handle = $pipes[0];
 818         }
 819
 820         /**
 821          * Close the old file, move it to a specified name,
 822          * and reopen new file with the old name.
 823          */
 824         function closeRenameAndReopen( $newname ) {
 825                 if ( is_array( $newname ) ) {
 826                         if ( count( $newname ) > 1 ) {
 827                                 throw new MWException( __METHOD__ . ": passed multiple arguments for rename of single file\n" );
 828                         } else {
 829                                 $newname = $newname[0];
 830                         }
 831                 }
 832                 if ( $newname ) {
 833                         fclose( $this->handle );
 834                         proc_close( $this->procOpenResource );
 835                         rename( $this->filename, $newname );
 836                         $command = $this->command;
 837                         $command .=  " > " . wfEscapeShellArg( $this->filename );
 838                         $this->startCommand( $command );
 839                 }
 840         }
 841
 842         function closeAndRename( $newname ) {
 843                 if ( is_array( $newname ) ) {
 844                         if ( count( $newname ) > 1 ) {
 845                                 throw new MWException( __METHOD__ . ": passed multiple arguments for rename of single file\n" );
 846                         } else {
 847                                 $newname = $newname[0];
 848                         }
 849                 }
 850                 if ( $newname ) {
 851 #                       pclose( $this->handle );
 852                         fclose( $this->handle );
 853                         proc_close( $this->procOpenResource );
 854                         rename( $this->filename, $newname );
 855                 }
 856         }
 857
 858         function rename( $newname ) {
 859                 if ( is_array( $newname ) ) {
 860                         if ( count( $newname ) > 1 ) {
 861                                 throw new MWException( __METHOD__ . ": passed multiple arguments for rename of single file\n" );
 862                         } else {
 863                                 $newname = $newname[0];
 864                         }
 865                 }
 866                 if ( $newname ) {
 867                         rename( $this->filename, $newname );
 868                 }
 869         }
 870 }
 871
 872 /**
 873  * Sends dump output via the gzip compressor.
 874  * @ingroup Dump
 875  */
 876 class DumpGZipOutput extends DumpPipeOutput {
 877         function __construct( $file ) {
 878                 parent::__construct( "gzip", $file );
 879         }
 880 }
 881
 882 /**
 883  * Sends dump output via the bgzip2 compressor.
 884  * @ingroup Dump
 885  */
 886 class DumpBZip2Output extends DumpPipeOutput {
 887         function __construct( $file ) {
 888                 parent::__construct( "bzip2", $file );
 889         }
 890 }
 891
 892 /**
 893  * Sends dump output via the p7zip compressor.
 894  * @ingroup Dump
 895  */
 896 class Dump7ZipOutput extends DumpPipeOutput {
 897         protected $filename;
 898
 899         function __construct( $file ) {
 900                 $command = "7za a -bd -si " . wfEscapeShellArg( $file );
 901                 // Suppress annoying useless crap from p7zip
 902                 // Unfortunately this could suppress real error messages too
 903                 $command .= ' >' . wfGetNull() . ' 2>&1';
 904                 parent::__construct( $command );
 905                 $this->filename = $file;
 906         }
 907
 908         function closeRenameAndReopen( $newname ) {
 909                 if ( is_array( $newname ) ) {
 910                         if ( count( $newname ) > 1 ) {
 911                                 throw new MWException( __METHOD__ . ": passed multiple arguments for rename of single file\n" );
 912                         }
 913                         else {
 914                                 $newname = $newname[0];
 915                         }
 916                 }
 917                 if ( $newname ) {
 918                         fclose( $this->handle );
 919                         proc_close( $this->procOpenResource );
 920                         rename( $this->filename, $newname );
 921                         $command = "7za a -bd -si " . wfEscapeShellArg( $file );
 922                         $command .= ' >' . wfGetNull() . ' 2>&1';
 923                         $this->startCommand( $command );
 924                 }
 925         }
 926
 927         function closeAndRename( $newname ) {
 928                 if ( is_array( $newname ) ) {
 929                         if ( count( $newname ) > 1 ) {
 930                                 throw new MWException( __METHOD__ . ": passed multiple arguments for rename of single file\n" );
 931                         } else {
 932                                 $newname = $newname[0];
 933                         }
 934                 }
 935                 if ( $newname ) {
 936                         fclose( $this->handle );
 937                         proc_close( $this->procOpenResource );
 938                         rename( $this->filename, $newname );
 939                 }
 940         }
 941
 942         function rename( $newname ) {
 943                 if ( is_array( $newname ) ) {
 944                         if ( count( $newname ) > 1 ) {
 945                                 throw new MWException( __METHOD__ . ": passed multiple arguments for rename of single file\n" );
 946                         }
 947                         else {
 948                                 $newname = $newname[0];
 949                         }
 950                 }
 951                 if ( $newname ) {
 952                         rename( $this->filename, $newname );
 953                 }
 954         }
 955 }
 956
 957
 958
 959 /**
 960  * Dump output filter class.
 961  * This just does output filtering and streaming; XML formatting is done
 962  * higher up, so be careful in what you do.
 963  * @ingroup Dump
 964  */
 965 class DumpFilter {
 966         function __construct( &$sink ) {
 967                 $this->sink =& $sink;
 968         }
 969
 970         function writeOpenStream( $string ) {
 971                 $this->sink->writeOpenStream( $string );
 972         }
 973
 974         function writeCloseStream( $string ) {
 975                 $this->sink->writeCloseStream( $string );
 976         }
 977
 978         function writeOpenPage( $page, $string ) {
 979                 $this->sendingThisPage = $this->pass( $page, $string );
 980                 if ( $this->sendingThisPage ) {
 981                         $this->sink->writeOpenPage( $page, $string );
 982                 }
 983         }
 984
 985         function writeClosePage( $string ) {
 986                 if ( $this->sendingThisPage ) {
 987                         $this->sink->writeClosePage( $string );
 988                         $this->sendingThisPage = false;
 989                 }
 990         }
 991
 992         function writeRevision( $rev, $string ) {
 993                 if ( $this->sendingThisPage ) {
 994                         $this->sink->writeRevision( $rev, $string );
 995                 }
 996         }
 997
 998         function writeLogItem( $rev, $string ) {
 999                 $this->sink->writeRevision( $rev, $string );
1000         }
1001
1002         function closeRenameAndReopen( $newname ) {
1003                 $this->sink->closeRenameAndReopen( $newname );
1004         }
1005
1006         function closeAndRename( $newname ) {
1007                 $this->sink->closeAndRename( $newname );
1008         }
1009
1010         function rename( $newname ) {
1011                 $this->sink->rename( $newname );
1012         }
1013
1014         function getFilename() {
1015                 return $this->sink->getFilename();
1016         }
1017
1018         /**
1019          * Override for page-based filter types.
1020          * @return bool
1021          */
1022         function pass( $page ) {
1023                 return true;
1024         }
1025 }
1026
1027 /**
1028  * Simple dump output filter to exclude all talk pages.
1029  * @ingroup Dump
1030  */
1031 class DumpNotalkFilter extends DumpFilter {
1032         function pass( $page ) {
1033                 return !MWNamespace::isTalk( $page->page_namespace );
1034         }
1035 }
1036
1037 /**
1038  * Dump output filter to include or exclude pages in a given set of namespaces.
1039  * @ingroup Dump
1040  */
1041 class DumpNamespaceFilter extends DumpFilter {
1042         var $invert = false;
1043         var $namespaces = array();
1044
1045         function __construct( &$sink, $param ) {
1046                 parent::__construct( $sink );
1047
1048                 $constants = array(
1049                         "NS_MAIN"           => NS_MAIN,
1050                         "NS_TALK"           => NS_TALK,
1051                         "NS_USER"           => NS_USER,
1052                         "NS_USER_TALK"      => NS_USER_TALK,
1053                         "NS_PROJECT"        => NS_PROJECT,
1054                         "NS_PROJECT_TALK"   => NS_PROJECT_TALK,
1055                         "NS_FILE"           => NS_FILE,
1056                         "NS_FILE_TALK"      => NS_FILE_TALK,
1057                         "NS_IMAGE"          => NS_IMAGE,  // NS_IMAGE is an alias for NS_FILE
1058                         "NS_IMAGE_TALK"     => NS_IMAGE_TALK,
1059                         "NS_MEDIAWIKI"      => NS_MEDIAWIKI,
1060                         "NS_MEDIAWIKI_TALK" => NS_MEDIAWIKI_TALK,
1061                         "NS_TEMPLATE"       => NS_TEMPLATE,
1062                         "NS_TEMPLATE_TALK"  => NS_TEMPLATE_TALK,
1063                         "NS_HELP"           => NS_HELP,
1064                         "NS_HELP_TALK"      => NS_HELP_TALK,
1065                         "NS_CATEGORY"       => NS_CATEGORY,
1066                         "NS_CATEGORY_TALK"  => NS_CATEGORY_TALK );
1067
1068                 if ( $param { 0 } == '!' ) {
1069                         $this->invert = true;
1070                         $param = substr( $param, 1 );
1071                 }
1072
1073                 foreach ( explode( ',', $param ) as $key ) {
1074                         $key = trim( $key );
1075                         if ( isset( $constants[$key] ) ) {
1076                                 $ns = $constants[$key];
1077                                 $this->namespaces[$ns] = true;
1078                         } elseif ( is_numeric( $key ) ) {
1079                                 $ns = intval( $key );
1080                                 $this->namespaces[$ns] = true;
1081                         } else {
1082                                 throw new MWException( "Unrecognized namespace key '$key'\n" );
1083                         }
1084                 }
1085         }
1086
1087         function pass( $page ) {
1088                 $match = isset( $this->namespaces[$page->page_namespace] );
1089                 return $this->invert xor $match;
1090         }
1091 }
1092
1093
1094 /**
1095  * Dump output filter to include only the last revision in each page sequence.
1096  * @ingroup Dump
1097  */
1098 class DumpLatestFilter extends DumpFilter {
1099         var $page, $pageString, $rev, $revString;
1100
1101         function writeOpenPage( $page, $string ) {
1102                 $this->page = $page;
1103                 $this->pageString = $string;
1104         }
1105
1106         function writeClosePage( $string ) {
1107                 if ( $this->rev ) {
1108                         $this->sink->writeOpenPage( $this->page, $this->pageString );
1109                         $this->sink->writeRevision( $this->rev, $this->revString );
1110                         $this->sink->writeClosePage( $string );
1111                 }
1112                 $this->rev = null;
1113                 $this->revString = null;
1114                 $this->page = null;
1115                 $this->pageString = null;
1116         }
1117
1118         function writeRevision( $rev, $string ) {
1119                 if ( $rev->rev_id == $this->page->page_latest ) {
1120                         $this->rev = $rev;
1121                         $this->revString = $string;
1122                 }
1123         }
1124 }
1125
1126 /**
1127  * Base class for output stream; prints to stdout or buffer or whereever.
1128  * @ingroup Dump
1129  */
1130 class DumpMultiWriter {
1131         function __construct( $sinks ) {
1132                 $this->sinks = $sinks;
1133                 $this->count = count( $sinks );
1134         }
1135
1136         function writeOpenStream( $string ) {
1137                 for ( $i = 0; $i < $this->count; $i++ ) {
1138                         $this->sinks[$i]->writeOpenStream( $string );
1139                 }
1140         }
1141
1142         function writeCloseStream( $string ) {
1143                 for ( $i = 0; $i < $this->count; $i++ ) {
1144                         $this->sinks[$i]->writeCloseStream( $string );
1145                 }
1146         }
1147
1148         function writeOpenPage( $page, $string ) {
1149                 for ( $i = 0; $i < $this->count; $i++ ) {
1150                         $this->sinks[$i]->writeOpenPage( $page, $string );
1151                 }
1152         }
1153
1154         function writeClosePage( $string ) {
1155                 for ( $i = 0; $i < $this->count; $i++ ) {
1156                         $this->sinks[$i]->writeClosePage( $string );
1157                 }
1158         }
1159
1160         function writeRevision( $rev, $string ) {
1161                 for ( $i = 0; $i < $this->count; $i++ ) {
1162                         $this->sinks[$i]->writeRevision( $rev, $string );
1163                 }
1164         }
1165
1166         function closeRenameAndReopen( $newnames ) {
1167                 for ( $i = 0; $i < $this->count; $i++ ) {
1168                         $this->sinks[$i]->closeRenameAndReopen( $newnames[$i] );
1169                 }
1170         }
1171
1172         function closeAndRename( $newname ) {
1173                 for ( $i = 0; $i < $this->count; $i++ ) {
1174                         $this->sinks[$i]->closeAndRename( $newnames[$i] );
1175                 }
1176         }
1177         function rename( $newnames ) {
1178                 for ( $i = 0; $i < $this->count; $i++ ) {
1179                         $this->sinks[$i]->rename( $newnames[$i] );
1180                 }
1181         }
1182
1183         function getFilename() {
1184                 $filenames = array();
1185                 for ( $i = 0; $i < $this->count; $i++ ) {
1186                         $filenames[] =  $this->sinks[$i]->getFilename();
1187                 }
1188                 return $filenames;
1189         }
1190
1191 }
1192
1193 function xmlsafe( $string ) {
1194         wfProfileIn( __FUNCTION__ );
1195
1196         /**
1197          * The page may contain old data which has not been properly normalized.
1198          * Invalid UTF-8 sequences or forbidden control characters will make our
1199          * XML output invalid, so be sure to strip them out.
1200          */
1201         $string = UtfNormal::cleanUp( $string );
1202
1203         $string = htmlspecialchars( $string );
1204         wfProfileOut( __FUNCTION__ );
1205         return $string;
1206 }