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( $cond ) {
 161                 wfProfileIn( __METHOD__ );
 162                 $this->author_list = "<contributors>";
 163                 // rev_deleted
 164
 165                 $res = $this->db->select(
 166                         array( 'page', 'revision' ),
 167                         array( 'DISTINCT rev_user_text', 'rev_user' ),
 168                         array(
 169                                 $this->db->bitAnd( 'rev_deleted', Revision::DELETED_USER ) . ' = 0',
 170                                 $cond,
 171                                 'page_id = rev_id',
 172                         ),
 173                         __METHOD__
 174                 );
 175
 176                 foreach ( $res as $row ) {
 177                         $this->author_list .= "<contributor>" .
 178                                 "<username>" .
 179                                 htmlentities( $row->rev_user_text )  .
 180                                 "</username>" .
 181                                 "<id>" .
 182                                 $row->rev_user .
 183                                 "</id>" .
 184                                 "</contributor>";
 185                 }
 186                 $this->author_list .= "</contributors>";
 187                 wfProfileOut( __METHOD__ );
 188         }
 189
 190         protected function dumpFrom( $cond = '' ) {
 191                 wfProfileIn( __METHOD__ );
 192                 # For logging dumps...
 193                 if ( $this->history & self::LOGS ) {
 194                         if ( $this->buffer == WikiExporter::STREAM ) {
 195                                 $prev = $this->db->bufferResults( false );
 196                         }
 197                         $where = array( 'user_id = log_user' );
 198                         # Hide private logs
 199                         $hideLogs = LogEventsList::getExcludeClause( $this->db );
 200                         if ( $hideLogs ) $where[] = $hideLogs;
 201                         # Add on any caller specified conditions
 202                         if ( $cond ) $where[] = $cond;
 203                         # Get logging table name for logging.* clause
 204                         $logging = $this->db->tableName( 'logging' );
 205                         $result = $this->db->select( array( 'logging', 'user' ),
 206                                 array( "{$logging}.*", 'user_name' ), // grab the user name
 207                                 $where,
 208                                 __METHOD__,
 209                                 array( 'ORDER BY' => 'log_id', 'USE INDEX' => array( 'logging' => 'PRIMARY' ) )
 210                         );
 211                         $wrapper = $this->db->resultObject( $result );
 212                         $this->outputLogStream( $wrapper );
 213                         if ( $this->buffer == WikiExporter::STREAM ) {
 214                                 $this->db->bufferResults( $prev );
 215                         }
 216                 # For page dumps...
 217                 } else {
 218                         $tables = array( 'page', 'revision' );
 219                         $opts = array( 'ORDER BY' => 'page_id ASC' );
 220                         $opts['USE INDEX'] = array();
 221                         $join = array();
 222                         if ( is_array( $this->history ) ) {
 223                                 # Time offset/limit for all pages/history...
 224                                 $revJoin = 'page_id=rev_page';
 225                                 # Set time order
 226                                 if ( $this->history['dir'] == 'asc' ) {
 227                                         $op = '>';
 228                                         $opts['ORDER BY'] = 'rev_timestamp ASC';
 229                                 } else {
 230                                         $op = '<';
 231                                         $opts['ORDER BY'] = 'rev_timestamp DESC';
 232                                 }
 233                                 # Set offset
 234                                 if ( !empty( $this->history['offset'] ) ) {
 235                                         $revJoin .= " AND rev_timestamp $op " .
 236                                                 $this->db->addQuotes( $this->db->timestamp( $this->history['offset'] ) );
 237                                 }
 238                                 $join['revision'] = array( 'INNER JOIN', $revJoin );
 239                                 # Set query limit
 240                                 if ( !empty( $this->history['limit'] ) ) {
 241                                         $opts['LIMIT'] = intval( $this->history['limit'] );
 242                                 }
 243                         } elseif ( $this->history & WikiExporter::FULL ) {
 244                                 # Full history dumps...
 245                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page' );
 246                         } elseif ( $this->history & WikiExporter::CURRENT ) {
 247                                 # Latest revision dumps...
 248                                 if ( $this->list_authors && $cond != '' )  { // List authors, if so desired
 249                                         $this->do_list_authors( $cond );
 250                                 }
 251                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' );
 252                         } elseif ( $this->history & WikiExporter::STABLE ) {
 253                                 # "Stable" revision dumps...
 254                                 # Default JOIN, to be overridden...
 255                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' );
 256                                 # One, and only one hook should set this, and return false
 257                                 if ( wfRunHooks( 'WikiExporter::dumpStableQuery', array( &$tables, &$opts, &$join ) ) ) {
 258                                         wfProfileOut( __METHOD__ );
 259                                         throw new MWException( __METHOD__ . " given invalid history dump type." );
 260                                 }
 261                         } else {
 262                                 # Uknown history specification parameter?
 263                                 wfProfileOut( __METHOD__ );
 264                                 throw new MWException( __METHOD__ . " given invalid history dump type." );
 265                         }
 266                         # Query optimization hacks
 267                         if ( $cond == '' ) {
 268                                 $opts[] = 'STRAIGHT_JOIN';
 269                                 $opts['USE INDEX']['page'] = 'PRIMARY';
 270                         }
 271                         # Build text join options
 272                         if ( $this->text != WikiExporter::STUB ) { // 1-pass
 273                                 $tables[] = 'text';
 274                                 $join['text'] = array( 'INNER JOIN', 'rev_text_id=old_id' );
 275                         }
 276
 277                         if ( $this->buffer == WikiExporter::STREAM ) {
 278                                 $prev = $this->db->bufferResults( false );
 279                         }
 280
 281                         wfRunHooks( 'ModifyExportQuery',
 282                                                 array( $this->db, &$tables, &$cond, &$opts, &$join ) );
 283
 284                         # Do the query!
 285                         $result = $this->db->select( $tables, '*', $cond, __METHOD__, $opts, $join );
 286                         $wrapper = $this->db->resultObject( $result );
 287                         # Output dump results
 288                         $this->outputPageStream( $wrapper );
 289                         if ( $this->list_authors ) {
 290                                 $this->outputPageStream( $wrapper );
 291                         }
 292
 293                         if ( $this->buffer == WikiExporter::STREAM ) {
 294                                 $this->db->bufferResults( $prev );
 295                         }
 296                 }
 297                 wfProfileOut( __METHOD__ );
 298         }
 299
 300         /**
 301          * Runs through a query result set dumping page and revision records.
 302          * The result set should be sorted/grouped by page to avoid duplicate
 303          * page records in the output.
 304          *
 305          * The result set will be freed once complete. Should be safe for
 306          * streaming (non-buffered) queries, as long as it was made on a
 307          * separate database connection not managed by LoadBalancer; some
 308          * blob storage types will make queries to pull source data.
 309          *
 310          * @param $resultset ResultWrapper
 311          */
 312         protected function outputPageStream( $resultset ) {
 313                 $last = null;
 314                 foreach ( $resultset as $row ) {
 315                         if ( is_null( $last ) ||
 316                                 $last->page_namespace != $row->page_namespace ||
 317                                 $last->page_title     != $row->page_title ) {
 318                                 if ( isset( $last ) ) {
 319                                         $output = '';
 320                                         if ( $this->dumpUploads ) {
 321                                                 $output .= $this->writer->writeUploads( $last );
 322                                         }
 323                                         $output .= $this->writer->closePage();
 324                                         $this->sink->writeClosePage( $output );
 325                                 }
 326                                 $output = $this->writer->openPage( $row );
 327                                 $this->sink->writeOpenPage( $row, $output );
 328                                 $last = $row;
 329                         }
 330                         $output = $this->writer->writeRevision( $row );
 331                         $this->sink->writeRevision( $row, $output );
 332                 }
 333                 if ( isset( $last ) ) {
 334                         $output = '';
 335                         if ( $this->dumpUploads ) {
 336                                 $output .= $this->writer->writeUploads( $last );
 337                         }
 338                         $output .= $this->author_list;
 339                         $output .= $this->writer->closePage();
 340                         $this->sink->writeClosePage( $output );
 341                 }
 342         }
 343
 344         protected function outputLogStream( $resultset ) {
 345                 foreach ( $resultset as $row ) {
 346                         $output = $this->writer->writeLogItem( $row );
 347                         $this->sink->writeLogItem( $row, $output );
 348                 }
 349         }
 350 }
 351
 352 /**
 353  * @ingroup Dump
 354  */
 355 class XmlDumpWriter {
 356
 357         /**
 358          * Returns the export schema version.
 359          * @return string
 360          */
 361         function schemaVersion() {
 362                 return "0.5";
 363         }
 364
 365         /**
 366          * Opens the XML output stream's root <mediawiki> element.
 367          * This does not include an xml directive, so is safe to include
 368          * as a subelement in a larger XML stream. Namespace and XML Schema
 369          * references are included.
 370          *
 371          * Output will be encoded in UTF-8.
 372          *
 373          * @return string
 374          */
 375         function openStream() {
 376                 global $wgLanguageCode;
 377                 $ver = $this->schemaVersion();
 378                 return Xml::element( 'mediawiki', array(
 379                         'xmlns'              => "http://www.mediawiki.org/xml/export-$ver/",
 380                         'xmlns:xsi'          => "http://www.w3.org/2001/XMLSchema-instance",
 381                         'xsi:schemaLocation' => "http://www.mediawiki.org/xml/export-$ver/ " .
 382                                                 "http://www.mediawiki.org/xml/export-$ver.xsd",
 383                         'version'            => $ver,
 384                         'xml:lang'           => $wgLanguageCode ),
 385                         null ) .
 386                         "\n" .
 387                         $this->siteInfo();
 388         }
 389
 390         function siteInfo() {
 391                 $info = array(
 392                         $this->sitename(),
 393                         $this->homelink(),
 394                         $this->generator(),
 395                         $this->caseSetting(),
 396                         $this->namespaces() );
 397                 return "  <siteinfo>\n    " .
 398                         implode( "\n    ", $info ) .
 399                         "\n  </siteinfo>\n";
 400         }
 401
 402         function sitename() {
 403                 global $wgSitename;
 404                 return Xml::element( 'sitename', array(), $wgSitename );
 405         }
 406
 407         function generator() {
 408                 global $wgVersion;
 409                 return Xml::element( 'generator', array(), "MediaWiki $wgVersion" );
 410         }
 411
 412         function homelink() {
 413                 return Xml::element( 'base', array(), Title::newMainPage()->getFullUrl() );
 414         }
 415
 416         function caseSetting() {
 417                 global $wgCapitalLinks;
 418                 // "case-insensitive" option is reserved for future
 419                 $sensitivity = $wgCapitalLinks ? 'first-letter' : 'case-sensitive';
 420                 return Xml::element( 'case', array(), $sensitivity );
 421         }
 422
 423         function namespaces() {
 424                 global $wgContLang;
 425                 $spaces = "<namespaces>\n";
 426                 foreach ( $wgContLang->getFormattedNamespaces() as $ns => $title ) {
 427                         $spaces .= '      ' .
 428                                 Xml::element( 'namespace',
 429                                         array(  'key' => $ns,
 430                                                         'case' => MWNamespace::isCapitalized( $ns ) ? 'first-letter' : 'case-sensitive',
 431                                         ), $title ) . "\n";
 432                 }
 433                 $spaces .= "    </namespaces>";
 434                 return $spaces;
 435         }
 436
 437         /**
 438          * Closes the output stream with the closing root element.
 439          * Call when finished dumping things.
 440          */
 441         function closeStream() {
 442                 return "</mediawiki>\n";
 443         }
 444
 445
 446         /**
 447          * Opens a <page> section on the output stream, with data
 448          * from the given database row.
 449          *
 450          * @param $row object
 451          * @return string
 452          * @access private
 453          */
 454         function openPage( $row ) {
 455                 $out = "  <page>\n";
 456                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 457                 $out .= '    ' . Xml::elementClean( 'title', array(), $title->getPrefixedText() ) . "\n";
 458                 $out .= '    ' . Xml::element( 'id', array(), strval( $row->page_id ) ) . "\n";
 459                 if ( $row->page_is_redirect ) {
 460                         $out .= '    ' . Xml::element( 'redirect', array() ) . "\n";
 461                 }
 462                 if ( $row->page_restrictions != '' ) {
 463                         $out .= '    ' . Xml::element( 'restrictions', array(),
 464                                 strval( $row->page_restrictions ) ) . "\n";
 465                 }
 466
 467                 wfRunHooks( 'XmlDumpWriterOpenPage', array( $this, &$out, $row, $title ) );
 468
 469                 return $out;
 470         }
 471
 472         /**
 473          * Closes a <page> section on the output stream.
 474          *
 475          * @access private
 476          */
 477         function closePage() {
 478                 return "  </page>\n";
 479         }
 480
 481         /**
 482          * Dumps a <revision> section on the output stream, with
 483          * data filled in from the given database row.
 484          *
 485          * @param $row object
 486          * @return string
 487          * @access private
 488          */
 489         function writeRevision( $row ) {
 490                 wfProfileIn( __METHOD__ );
 491
 492                 $out  = "    <revision>\n";
 493                 $out .= "      " . Xml::element( 'id', null, strval( $row->rev_id ) ) . "\n";
 494
 495                 $out .= $this->writeTimestamp( $row->rev_timestamp );
 496
 497                 if ( $row->rev_deleted & Revision::DELETED_USER ) {
 498                         $out .= "      " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 499                 } else {
 500                         $out .= $this->writeContributor( $row->rev_user, $row->rev_user_text );
 501                 }
 502
 503                 if ( $row->rev_minor_edit ) {
 504                         $out .=  "      <minor/>\n";
 505                 }
 506                 if ( $row->rev_deleted & Revision::DELETED_COMMENT ) {
 507                         $out .= "      " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 508                 } elseif ( $row->rev_comment != '' ) {
 509                         $out .= "      " . Xml::elementClean( 'comment', null, strval( $row->rev_comment ) ) . "\n";
 510                 }
 511
 512                 $text = '';
 513                 if ( $row->rev_deleted & Revision::DELETED_TEXT ) {
 514                         $out .= "      " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 515                 } elseif ( isset( $row->old_text ) ) {
 516                         // Raw text from the database may have invalid chars
 517                         $text = strval( Revision::getRevisionText( $row ) );
 518                         $out .= "      " . Xml::elementClean( 'text',
 519                                 array( 'xml:space' => 'preserve', 'bytes' => $row->rev_len ),
 520                                 strval( $text ) ) . "\n";
 521                 } else {
 522                         // Stub output
 523                         $out .= "      " . Xml::element( 'text',
 524                                 array( 'id' => $row->rev_text_id, 'bytes' => $row->rev_len ),
 525                                 "" ) . "\n";
 526                 }
 527
 528                 wfRunHooks( 'XmlDumpWriterWriteRevision', array( &$this, &$out, $row, $text ) );
 529
 530                 $out .= "    </revision>\n";
 531
 532                 wfProfileOut( __METHOD__ );
 533                 return $out;
 534         }
 535
 536         /**
 537          * Dumps a <logitem> section on the output stream, with
 538          * data filled in from the given database row.
 539          *
 540          * @param $row object
 541          * @return string
 542          * @access private
 543          */
 544         function writeLogItem( $row ) {
 545                 wfProfileIn( __METHOD__ );
 546
 547                 $out  = "    <logitem>\n";
 548                 $out .= "      " . Xml::element( 'id', null, strval( $row->log_id ) ) . "\n";
 549
 550                 $out .= $this->writeTimestamp( $row->log_timestamp );
 551
 552                 if ( $row->log_deleted & LogPage::DELETED_USER ) {
 553                         $out .= "      " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 554                 } else {
 555                         $out .= $this->writeContributor( $row->log_user, $row->user_name );
 556                 }
 557
 558                 if ( $row->log_deleted & LogPage::DELETED_COMMENT ) {
 559                         $out .= "      " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 560                 } elseif ( $row->log_comment != '' ) {
 561                         $out .= "      " . Xml::elementClean( 'comment', null, strval( $row->log_comment ) ) . "\n";
 562                 }
 563
 564                 $out .= "      " . Xml::element( 'type', null, strval( $row->log_type ) ) . "\n";
 565                 $out .= "      " . Xml::element( 'action', null, strval( $row->log_action ) ) . "\n";
 566
 567                 if ( $row->log_deleted & LogPage::DELETED_ACTION ) {
 568                         $out .= "      " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 569                 } else {
 570                         $title = Title::makeTitle( $row->log_namespace, $row->log_title );
 571                         $out .= "      " . Xml::elementClean( 'logtitle', null, $title->getPrefixedText() ) . "\n";
 572                         $out .= "      " . Xml::elementClean( 'params',
 573                                 array( 'xml:space' => 'preserve' ),
 574                                 strval( $row->log_params ) ) . "\n";
 575                 }
 576
 577                 $out .= "    </logitem>\n";
 578
 579                 wfProfileOut( __METHOD__ );
 580                 return $out;
 581         }
 582
 583         function writeTimestamp( $timestamp ) {
 584                 $ts = wfTimestamp( TS_ISO_8601, $timestamp );
 585                 return "      " . Xml::element( 'timestamp', null, $ts ) . "\n";
 586         }
 587
 588         function writeContributor( $id, $text ) {
 589                 $out = "      <contributor>\n";
 590                 if ( $id ) {
 591                         $out .= "        " . Xml::elementClean( 'username', null, strval( $text ) ) . "\n";
 592                         $out .= "        " . Xml::element( 'id', null, strval( $id ) ) . "\n";
 593                 } else {
 594                         $out .= "        " . Xml::elementClean( 'ip', null, strval( $text ) ) . "\n";
 595                 }
 596                 $out .= "      </contributor>\n";
 597                 return $out;
 598         }
 599
 600         /**
 601          * Warning! This data is potentially inconsistent. :(
 602          */
 603         function writeUploads( $row ) {
 604                 if ( $row->page_namespace == NS_IMAGE ) {
 605                         $img = wfFindFile( $row->page_title );
 606                         if ( $img ) {
 607                                 $out = '';
 608                                 foreach ( array_reverse( $img->getHistory() ) as $ver ) {
 609                                         $out .= $this->writeUpload( $ver );
 610                                 }
 611                                 $out .= $this->writeUpload( $img );
 612                                 return $out;
 613                         }
 614                 }
 615                 return '';
 616         }
 617
 618         function writeUpload( $file ) {
 619                 return "    <upload>\n" .
 620                         $this->writeTimestamp( $file->getTimestamp() ) .
 621                         $this->writeContributor( $file->getUser( 'id' ), $file->getUser( 'text' ) ) .
 622                         "      " . Xml::elementClean( 'comment', null, $file->getDescription() ) . "\n" .
 623                         "      " . Xml::element( 'filename', null, $file->getName() ) . "\n" .
 624                         "      " . Xml::element( 'src', null, $file->getFullUrl() ) . "\n" .
 625                         "      " . Xml::element( 'size', null, $file->getSize() ) . "\n" .
 626                         "    </upload>\n";
 627         }
 628
 629 }
 630
 631
 632 /**
 633  * Base class for output stream; prints to stdout or buffer or whereever.
 634  * @ingroup Dump
 635  */
 636 class DumpOutput {
 637         function writeOpenStream( $string ) {
 638                 $this->write( $string );
 639         }
 640
 641         function writeCloseStream( $string ) {
 642                 $this->write( $string );
 643         }
 644
 645         function writeOpenPage( $page, $string ) {
 646                 $this->write( $string );
 647         }
 648
 649         function writeClosePage( $string ) {
 650                 $this->write( $string );
 651         }
 652
 653         function writeRevision( $rev, $string ) {
 654                 $this->write( $string );
 655         }
 656
 657         function writeLogItem( $rev, $string ) {
 658                 $this->write( $string );
 659         }
 660
 661         /**
 662          * Override to write to a different stream type.
 663          * @return bool
 664          */
 665         function write( $string ) {
 666                 print $string;
 667         }
 668 }
 669
 670 /**
 671  * Stream outputter to send data to a file.
 672  * @ingroup Dump
 673  */
 674 class DumpFileOutput extends DumpOutput {
 675         var $handle;
 676
 677         function __construct( $file ) {
 678                 $this->handle = fopen( $file, "wt" );
 679         }
 680
 681         function write( $string ) {
 682                 fputs( $this->handle, $string );
 683         }
 684 }
 685
 686 /**
 687  * Stream outputter to send data to a file via some filter program.
 688  * Even if compression is available in a library, using a separate
 689  * program can allow us to make use of a multi-processor system.
 690  * @ingroup Dump
 691  */
 692 class DumpPipeOutput extends DumpFileOutput {
 693         function __construct( $command, $file = null ) {
 694                 if ( !is_null( $file ) ) {
 695                         $command .=  " > " . wfEscapeShellArg( $file );
 696                 }
 697                 $this->handle = popen( $command, "w" );
 698         }
 699 }
 700
 701 /**
 702  * Sends dump output via the gzip compressor.
 703  * @ingroup Dump
 704  */
 705 class DumpGZipOutput extends DumpPipeOutput {
 706         function __construct( $file ) {
 707                 parent::__construct( "gzip", $file );
 708         }
 709 }
 710
 711 /**
 712  * Sends dump output via the bgzip2 compressor.
 713  * @ingroup Dump
 714  */
 715 class DumpBZip2Output extends DumpPipeOutput {
 716         function __construct( $file ) {
 717                 parent::__construct( "bzip2", $file );
 718         }
 719 }
 720
 721 /**
 722  * Sends dump output via the p7zip compressor.
 723  * @ingroup Dump
 724  */
 725 class Dump7ZipOutput extends DumpPipeOutput {
 726         function __construct( $file ) {
 727                 $command = "7za a -bd -si " . wfEscapeShellArg( $file );
 728                 // Suppress annoying useless crap from p7zip
 729                 // Unfortunately this could suppress real error messages too
 730                 $command .= ' >' . wfGetNull() . ' 2>&1';
 731                 parent::__construct( $command );
 732         }
 733 }
 734
 735
 736
 737 /**
 738  * Dump output filter class.
 739  * This just does output filtering and streaming; XML formatting is done
 740  * higher up, so be careful in what you do.
 741  * @ingroup Dump
 742  */
 743 class DumpFilter {
 744         function __construct( &$sink ) {
 745                 $this->sink =& $sink;
 746         }
 747
 748         function writeOpenStream( $string ) {
 749                 $this->sink->writeOpenStream( $string );
 750         }
 751
 752         function writeCloseStream( $string ) {
 753                 $this->sink->writeCloseStream( $string );
 754         }
 755
 756         function writeOpenPage( $page, $string ) {
 757                 $this->sendingThisPage = $this->pass( $page, $string );
 758                 if ( $this->sendingThisPage ) {
 759                         $this->sink->writeOpenPage( $page, $string );
 760                 }
 761         }
 762
 763         function writeClosePage( $string ) {
 764                 if ( $this->sendingThisPage ) {
 765                         $this->sink->writeClosePage( $string );
 766                         $this->sendingThisPage = false;
 767                 }
 768         }
 769
 770         function writeRevision( $rev, $string ) {
 771                 if ( $this->sendingThisPage ) {
 772                         $this->sink->writeRevision( $rev, $string );
 773                 }
 774         }
 775
 776         function writeLogItem( $rev, $string ) {
 777                 $this->sink->writeRevision( $rev, $string );
 778         }
 779
 780         /**
 781          * Override for page-based filter types.
 782          * @return bool
 783          */
 784         function pass( $page ) {
 785                 return true;
 786         }
 787 }
 788
 789 /**
 790  * Simple dump output filter to exclude all talk pages.
 791  * @ingroup Dump
 792  */
 793 class DumpNotalkFilter extends DumpFilter {
 794         function pass( $page ) {
 795                 return !MWNamespace::isTalk( $page->page_namespace );
 796         }
 797 }
 798
 799 /**
 800  * Dump output filter to include or exclude pages in a given set of namespaces.
 801  * @ingroup Dump
 802  */
 803 class DumpNamespaceFilter extends DumpFilter {
 804         var $invert = false;
 805         var $namespaces = array();
 806
 807         function __construct( &$sink, $param ) {
 808                 parent::__construct( $sink );
 809
 810                 $constants = array(
 811                         "NS_MAIN"           => NS_MAIN,
 812                         "NS_TALK"           => NS_TALK,
 813                         "NS_USER"           => NS_USER,
 814                         "NS_USER_TALK"      => NS_USER_TALK,
 815                         "NS_PROJECT"        => NS_PROJECT,
 816                         "NS_PROJECT_TALK"   => NS_PROJECT_TALK,
 817                         "NS_FILE"           => NS_FILE,
 818                         "NS_FILE_TALK"      => NS_FILE_TALK,
 819                         "NS_IMAGE"          => NS_IMAGE,  // NS_IMAGE is an alias for NS_FILE
 820                         "NS_IMAGE_TALK"     => NS_IMAGE_TALK,
 821                         "NS_MEDIAWIKI"      => NS_MEDIAWIKI,
 822                         "NS_MEDIAWIKI_TALK" => NS_MEDIAWIKI_TALK,
 823                         "NS_TEMPLATE"       => NS_TEMPLATE,
 824                         "NS_TEMPLATE_TALK"  => NS_TEMPLATE_TALK,
 825                         "NS_HELP"           => NS_HELP,
 826                         "NS_HELP_TALK"      => NS_HELP_TALK,
 827                         "NS_CATEGORY"       => NS_CATEGORY,
 828                         "NS_CATEGORY_TALK"  => NS_CATEGORY_TALK );
 829
 830                 if ( $param { 0 } == '!' ) {
 831                         $this->invert = true;
 832                         $param = substr( $param, 1 );
 833                 }
 834
 835                 foreach ( explode( ',', $param ) as $key ) {
 836                         $key = trim( $key );
 837                         if ( isset( $constants[$key] ) ) {
 838                                 $ns = $constants[$key];
 839                                 $this->namespaces[$ns] = true;
 840                         } elseif ( is_numeric( $key ) ) {
 841                                 $ns = intval( $key );
 842                                 $this->namespaces[$ns] = true;
 843                         } else {
 844                                 throw new MWException( "Unrecognized namespace key '$key'\n" );
 845                         }
 846                 }
 847         }
 848
 849         function pass( $page ) {
 850                 $match = isset( $this->namespaces[$page->page_namespace] );
 851                 return $this->invert xor $match;
 852         }
 853 }
 854
 855
 856 /**
 857  * Dump output filter to include only the last revision in each page sequence.
 858  * @ingroup Dump
 859  */
 860 class DumpLatestFilter extends DumpFilter {
 861         var $page, $pageString, $rev, $revString;
 862
 863         function writeOpenPage( $page, $string ) {
 864                 $this->page = $page;
 865                 $this->pageString = $string;
 866         }
 867
 868         function writeClosePage( $string ) {
 869                 if ( $this->rev ) {
 870                         $this->sink->writeOpenPage( $this->page, $this->pageString );
 871                         $this->sink->writeRevision( $this->rev, $this->revString );
 872                         $this->sink->writeClosePage( $string );
 873                 }
 874                 $this->rev = null;
 875                 $this->revString = null;
 876                 $this->page = null;
 877                 $this->pageString = null;
 878         }
 879
 880         function writeRevision( $rev, $string ) {
 881                 if ( $rev->rev_id == $this->page->page_latest ) {
 882                         $this->rev = $rev;
 883                         $this->revString = $string;
 884                 }
 885         }
 886 }
 887
 888 /**
 889  * Base class for output stream; prints to stdout or buffer or whereever.
 890  * @ingroup Dump
 891  */
 892 class DumpMultiWriter {
 893         function __construct( $sinks ) {
 894                 $this->sinks = $sinks;
 895                 $this->count = count( $sinks );
 896         }
 897
 898         function writeOpenStream( $string ) {
 899                 for ( $i = 0; $i < $this->count; $i++ ) {
 900                         $this->sinks[$i]->writeOpenStream( $string );
 901                 }
 902         }
 903
 904         function writeCloseStream( $string ) {
 905                 for ( $i = 0; $i < $this->count; $i++ ) {
 906                         $this->sinks[$i]->writeCloseStream( $string );
 907                 }
 908         }
 909
 910         function writeOpenPage( $page, $string ) {
 911                 for ( $i = 0; $i < $this->count; $i++ ) {
 912                         $this->sinks[$i]->writeOpenPage( $page, $string );
 913                 }
 914         }
 915
 916         function writeClosePage( $string ) {
 917                 for ( $i = 0; $i < $this->count; $i++ ) {
 918                         $this->sinks[$i]->writeClosePage( $string );
 919                 }
 920         }
 921
 922         function writeRevision( $rev, $string ) {
 923                 for ( $i = 0; $i < $this->count; $i++ ) {
 924                         $this->sinks[$i]->writeRevision( $rev, $string );
 925                 }
 926         }
 927 }
 928
 929 function xmlsafe( $string ) {
 930         wfProfileIn( __FUNCTION__ );
 931
 932         /**
 933          * The page may contain old data which has not been properly normalized.
 934          * Invalid UTF-8 sequences or forbidden control characters will make our
 935          * XML output invalid, so be sure to strip them out.
 936          */
 937         $string = UtfNormal::cleanUp( $string );
 938
 939         $string = htmlspecialchars( $string );
 940         wfProfileOut( __FUNCTION__ );
 941         return $string;
 942 }