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