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
 358         /**
 359          * Returns the export schema version.
 360          * @return string
 361          */
 362         function schemaVersion() {
 363                 return "0.5";
 364         }
 365
 366         /**
 367          * Opens the XML output stream's root <mediawiki> element.
 368          * This does not include an xml directive, so is safe to include
 369          * as a subelement in a larger XML stream. Namespace and XML Schema
 370          * references are included.
 371          *
 372          * Output will be encoded in UTF-8.
 373          *
 374          * @return string
 375          */
 376         function openStream() {
 377                 global $wgLanguageCode;
 378                 $ver = $this->schemaVersion();
 379                 return Xml::element( 'mediawiki', array(
 380                         'xmlns'              => "http://www.mediawiki.org/xml/export-$ver/",
 381                         'xmlns:xsi'          => "http://www.w3.org/2001/XMLSchema-instance",
 382                         'xsi:schemaLocation' => "http://www.mediawiki.org/xml/export-$ver/ " .
 383                                                 "http://www.mediawiki.org/xml/export-$ver.xsd",
 384                         'version'            => $ver,
 385                         'xml:lang'           => $wgLanguageCode ),
 386                         null ) .
 387                         "\n" .
 388                         $this->siteInfo();
 389         }
 390
 391         function siteInfo() {
 392                 $info = array(
 393                         $this->sitename(),
 394                         $this->homelink(),
 395                         $this->generator(),
 396                         $this->caseSetting(),
 397                         $this->namespaces() );
 398                 return "  <siteinfo>\n    " .
 399                         implode( "\n    ", $info ) .
 400                         "\n  </siteinfo>\n";
 401         }
 402
 403         function sitename() {
 404                 global $wgSitename;
 405                 return Xml::element( 'sitename', array(), $wgSitename );
 406         }
 407
 408         function generator() {
 409                 return Xml::element( 'generator', array(), "MediaWiki " . MW_VERSION );
 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, $dumpContents = false ) {
 604                 if ( $row->page_namespace == NS_IMAGE ) {
 605                         $img = wfLocalFile( $row->page_title );
 606                         if ( $img && $img->exists() ) {
 607                                 $out = '';
 608                                 foreach ( array_reverse( $img->getHistory() ) as $ver ) {
 609                                         $out .= $this->writeUpload( $ver, $dumpContents );
 610                                 }
 611                                 $out .= $this->writeUpload( $img, $dumpContents );
 612                                 return $out;
 613                         }
 614                 }
 615                 return '';
 616         }
 617
 618         function writeUpload( $file, $dumpContents = false ) {
 619                 if ( $file->isOld() ) {
 620                         $archiveName = "      " .
 621                                 Xml::element( 'archivename', null, $file->getArchiveName() ) . "\n";
 622                 } else {
 623                         $archiveName = '';
 624                 }
 625                 if ( $dumpContents ) {
 626                         # Dump file as base64
 627                         # Uses only XML-safe characters, so does not need escaping
 628                         $contents = '      <contents encoding="base64">' .
 629                                 chunk_split( base64_encode( file_get_contents( $file->getPath() ) ) ) .
 630                                 "      </contents>\n";
 631                 } else {
 632                         $contents = '';
 633                 }
 634                 return "    <upload>\n" .
 635                         $this->writeTimestamp( $file->getTimestamp() ) .
 636                         $this->writeContributor( $file->getUser( 'id' ), $file->getUser( 'text' ) ) .
 637                         "      " . Xml::elementClean( 'comment', null, $file->getDescription() ) . "\n" .
 638                         "      " . Xml::element( 'filename', null, $file->getName() ) . "\n" .
 639                         $archiveName .
 640                         "      " . Xml::element( 'src', null, $file->getFullUrl() ) . "\n" .
 641                         "      " . Xml::element( 'size', null, $file->getSize() ) . "\n" .
 642                         "      " . Xml::element( 'sha1base36', null, $file->getSha1() ) . "\n" .
 643                         "      " . Xml::element( 'rel', null, $file->getRel() ) . "\n" .
 644                         $contents .
 645                         "    </upload>\n";
 646         }
 647
 648 }
 649
 650
 651 /**
 652  * Base class for output stream; prints to stdout or buffer or whereever.
 653  * @ingroup Dump
 654  */
 655 class DumpOutput {
 656         function writeOpenStream( $string ) {
 657                 $this->write( $string );
 658         }
 659
 660         function writeCloseStream( $string ) {
 661                 $this->write( $string );
 662         }
 663
 664         function writeOpenPage( $page, $string ) {
 665                 $this->write( $string );
 666         }
 667
 668         function writeClosePage( $string ) {
 669                 $this->write( $string );
 670         }
 671
 672         function writeRevision( $rev, $string ) {
 673                 $this->write( $string );
 674         }
 675
 676         function writeLogItem( $rev, $string ) {
 677                 $this->write( $string );
 678         }
 679
 680         /**
 681          * Override to write to a different stream type.
 682          * @return bool
 683          */
 684         function write( $string ) {
 685                 print $string;
 686         }
 687 }
 688
 689 /**
 690  * Stream outputter to send data to a file.
 691  * @ingroup Dump
 692  */
 693 class DumpFileOutput extends DumpOutput {
 694         var $handle;
 695
 696         function __construct( $file ) {
 697                 $this->handle = fopen( $file, "wt" );
 698         }
 699
 700         function write( $string ) {
 701                 fputs( $this->handle, $string );
 702         }
 703 }
 704
 705 /**
 706  * Stream outputter to send data to a file via some filter program.
 707  * Even if compression is available in a library, using a separate
 708  * program can allow us to make use of a multi-processor system.
 709  * @ingroup Dump
 710  */
 711 class DumpPipeOutput extends DumpFileOutput {
 712         function __construct( $command, $file = null ) {
 713                 if ( !is_null( $file ) ) {
 714                         $command .=  " > " . wfEscapeShellArg( $file );
 715                 }
 716                 $this->handle = popen( $command, "w" );
 717         }
 718 }
 719
 720 /**
 721  * Sends dump output via the gzip compressor.
 722  * @ingroup Dump
 723  */
 724 class DumpGZipOutput extends DumpPipeOutput {
 725         function __construct( $file ) {
 726                 parent::__construct( "gzip", $file );
 727         }
 728 }
 729
 730 /**
 731  * Sends dump output via the bgzip2 compressor.
 732  * @ingroup Dump
 733  */
 734 class DumpBZip2Output extends DumpPipeOutput {
 735         function __construct( $file ) {
 736                 parent::__construct( "bzip2", $file );
 737         }
 738 }
 739
 740 /**
 741  * Sends dump output via the p7zip compressor.
 742  * @ingroup Dump
 743  */
 744 class Dump7ZipOutput extends DumpPipeOutput {
 745         function __construct( $file ) {
 746                 $command = "7za a -bd -si " . wfEscapeShellArg( $file );
 747                 // Suppress annoying useless crap from p7zip
 748                 // Unfortunately this could suppress real error messages too
 749                 $command .= ' >' . wfGetNull() . ' 2>&1';
 750                 parent::__construct( $command );
 751         }
 752 }
 753
 754
 755
 756 /**
 757  * Dump output filter class.
 758  * This just does output filtering and streaming; XML formatting is done
 759  * higher up, so be careful in what you do.
 760  * @ingroup Dump
 761  */
 762 class DumpFilter {
 763         function __construct( &$sink ) {
 764                 $this->sink =& $sink;
 765         }
 766
 767         function writeOpenStream( $string ) {
 768                 $this->sink->writeOpenStream( $string );
 769         }
 770
 771         function writeCloseStream( $string ) {
 772                 $this->sink->writeCloseStream( $string );
 773         }
 774
 775         function writeOpenPage( $page, $string ) {
 776                 $this->sendingThisPage = $this->pass( $page, $string );
 777                 if ( $this->sendingThisPage ) {
 778                         $this->sink->writeOpenPage( $page, $string );
 779                 }
 780         }
 781
 782         function writeClosePage( $string ) {
 783                 if ( $this->sendingThisPage ) {
 784                         $this->sink->writeClosePage( $string );
 785                         $this->sendingThisPage = false;
 786                 }
 787         }
 788
 789         function writeRevision( $rev, $string ) {
 790                 if ( $this->sendingThisPage ) {
 791                         $this->sink->writeRevision( $rev, $string );
 792                 }
 793         }
 794
 795         function writeLogItem( $rev, $string ) {
 796                 $this->sink->writeRevision( $rev, $string );
 797         }
 798
 799         /**
 800          * Override for page-based filter types.
 801          * @return bool
 802          */
 803         function pass( $page ) {
 804                 return true;
 805         }
 806 }
 807
 808 /**
 809  * Simple dump output filter to exclude all talk pages.
 810  * @ingroup Dump
 811  */
 812 class DumpNotalkFilter extends DumpFilter {
 813         function pass( $page ) {
 814                 return !MWNamespace::isTalk( $page->page_namespace );
 815         }
 816 }
 817
 818 /**
 819  * Dump output filter to include or exclude pages in a given set of namespaces.
 820  * @ingroup Dump
 821  */
 822 class DumpNamespaceFilter extends DumpFilter {
 823         var $invert = false;
 824         var $namespaces = array();
 825
 826         function __construct( &$sink, $param ) {
 827                 parent::__construct( $sink );
 828
 829                 $constants = array(
 830                         "NS_MAIN"           => NS_MAIN,
 831                         "NS_TALK"           => NS_TALK,
 832                         "NS_USER"           => NS_USER,
 833                         "NS_USER_TALK"      => NS_USER_TALK,
 834                         "NS_PROJECT"        => NS_PROJECT,
 835                         "NS_PROJECT_TALK"   => NS_PROJECT_TALK,
 836                         "NS_FILE"           => NS_FILE,
 837                         "NS_FILE_TALK"      => NS_FILE_TALK,
 838                         "NS_IMAGE"          => NS_IMAGE,  // NS_IMAGE is an alias for NS_FILE
 839                         "NS_IMAGE_TALK"     => NS_IMAGE_TALK,
 840                         "NS_MEDIAWIKI"      => NS_MEDIAWIKI,
 841                         "NS_MEDIAWIKI_TALK" => NS_MEDIAWIKI_TALK,
 842                         "NS_TEMPLATE"       => NS_TEMPLATE,
 843                         "NS_TEMPLATE_TALK"  => NS_TEMPLATE_TALK,
 844                         "NS_HELP"           => NS_HELP,
 845                         "NS_HELP_TALK"      => NS_HELP_TALK,
 846                         "NS_CATEGORY"       => NS_CATEGORY,
 847                         "NS_CATEGORY_TALK"  => NS_CATEGORY_TALK );
 848
 849                 if ( $param { 0 } == '!' ) {
 850                         $this->invert = true;
 851                         $param = substr( $param, 1 );
 852                 }
 853
 854                 foreach ( explode( ',', $param ) as $key ) {
 855                         $key = trim( $key );
 856                         if ( isset( $constants[$key] ) ) {
 857                                 $ns = $constants[$key];
 858                                 $this->namespaces[$ns] = true;
 859                         } elseif ( is_numeric( $key ) ) {
 860                                 $ns = intval( $key );
 861                                 $this->namespaces[$ns] = true;
 862                         } else {
 863                                 throw new MWException( "Unrecognized namespace key '$key'\n" );
 864                         }
 865                 }
 866         }
 867
 868         function pass( $page ) {
 869                 $match = isset( $this->namespaces[$page->page_namespace] );
 870                 return $this->invert xor $match;
 871         }
 872 }
 873
 874
 875 /**
 876  * Dump output filter to include only the last revision in each page sequence.
 877  * @ingroup Dump
 878  */
 879 class DumpLatestFilter extends DumpFilter {
 880         var $page, $pageString, $rev, $revString;
 881
 882         function writeOpenPage( $page, $string ) {
 883                 $this->page = $page;
 884                 $this->pageString = $string;
 885         }
 886
 887         function writeClosePage( $string ) {
 888                 if ( $this->rev ) {
 889                         $this->sink->writeOpenPage( $this->page, $this->pageString );
 890                         $this->sink->writeRevision( $this->rev, $this->revString );
 891                         $this->sink->writeClosePage( $string );
 892                 }
 893                 $this->rev = null;
 894                 $this->revString = null;
 895                 $this->page = null;
 896                 $this->pageString = null;
 897         }
 898
 899         function writeRevision( $rev, $string ) {
 900                 if ( $rev->rev_id == $this->page->page_latest ) {
 901                         $this->rev = $rev;
 902                         $this->revString = $string;
 903                 }
 904         }
 905 }
 906
 907 /**
 908  * Base class for output stream; prints to stdout or buffer or whereever.
 909  * @ingroup Dump
 910  */
 911 class DumpMultiWriter {
 912         function __construct( $sinks ) {
 913                 $this->sinks = $sinks;
 914                 $this->count = count( $sinks );
 915         }
 916
 917         function writeOpenStream( $string ) {
 918                 for ( $i = 0; $i < $this->count; $i++ ) {
 919                         $this->sinks[$i]->writeOpenStream( $string );
 920                 }
 921         }
 922
 923         function writeCloseStream( $string ) {
 924                 for ( $i = 0; $i < $this->count; $i++ ) {
 925                         $this->sinks[$i]->writeCloseStream( $string );
 926                 }
 927         }
 928
 929         function writeOpenPage( $page, $string ) {
 930                 for ( $i = 0; $i < $this->count; $i++ ) {
 931                         $this->sinks[$i]->writeOpenPage( $page, $string );
 932                 }
 933         }
 934
 935         function writeClosePage( $string ) {
 936                 for ( $i = 0; $i < $this->count; $i++ ) {
 937                         $this->sinks[$i]->writeClosePage( $string );
 938                 }
 939         }
 940
 941         function writeRevision( $rev, $string ) {
 942                 for ( $i = 0; $i < $this->count; $i++ ) {
 943                         $this->sinks[$i]->writeRevision( $rev, $string );
 944                 }
 945         }
 946 }
 947
 948 function xmlsafe( $string ) {
 949         wfProfileIn( __FUNCTION__ );
 950
 951         /**
 952          * The page may contain old data which has not been properly normalized.
 953          * Invalid UTF-8 sequences or forbidden control characters will make our
 954          * XML output invalid, so be sure to strip them out.
 955          */
 956         $string = UtfNormal::cleanUp( $string );
 957
 958         $string = htmlspecialchars( $string );
 959         wfProfileOut( __FUNCTION__ );
 960         return $string;
 961 }