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         var $firstPageWritten = 0;
 358         var $lastPageWritten = 0;
 359         var $pageInProgress = 0;
 360
 361         /**
 362          * Returns the export schema version.
 363          * @return string
 364          */
 365         function schemaVersion() {
 366                 return "0.5";
 367         }
 368
 369         /**
 370          * Opens the XML output stream's root <mediawiki> element.
 371          * This does not include an xml directive, so is safe to include
 372          * as a subelement in a larger XML stream. Namespace and XML Schema
 373          * references are included.
 374          *
 375          * Output will be encoded in UTF-8.
 376          *
 377          * @return string
 378          */
 379         function openStream() {
 380                 global $wgLanguageCode;
 381                 $ver = $this->schemaVersion();
 382                 return Xml::element( 'mediawiki', array(
 383                         'xmlns'              => "http://www.mediawiki.org/xml/export-$ver/",
 384                         'xmlns:xsi'          => "http://www.w3.org/2001/XMLSchema-instance",
 385                         'xsi:schemaLocation' => "http://www.mediawiki.org/xml/export-$ver/ " .
 386                                                 "http://www.mediawiki.org/xml/export-$ver.xsd",
 387                         'version'            => $ver,
 388                         'xml:lang'           => $wgLanguageCode ),
 389                         null ) .
 390                         "\n" .
 391                         $this->siteInfo();
 392         }
 393
 394         function siteInfo() {
 395                 $info = array(
 396                         $this->sitename(),
 397                         $this->homelink(),
 398                         $this->generator(),
 399                         $this->caseSetting(),
 400                         $this->namespaces() );
 401                 return "  <siteinfo>\n    " .
 402                         implode( "\n    ", $info ) .
 403                         "\n  </siteinfo>\n";
 404         }
 405
 406         function sitename() {
 407                 global $wgSitename;
 408                 return Xml::element( 'sitename', array(), $wgSitename );
 409         }
 410
 411         function generator() {
 412                 global $wgVersion;
 413                 return Xml::element( 'generator', array(), "MediaWiki $wgVersion" );
 414         }
 415
 416         function homelink() {
 417                 return Xml::element( 'base', array(), Title::newMainPage()->getCanonicalUrl() );
 418         }
 419
 420         function caseSetting() {
 421                 global $wgCapitalLinks;
 422                 // "case-insensitive" option is reserved for future
 423                 $sensitivity = $wgCapitalLinks ? 'first-letter' : 'case-sensitive';
 424                 return Xml::element( 'case', array(), $sensitivity );
 425         }
 426
 427         function namespaces() {
 428                 global $wgContLang;
 429                 $spaces = "<namespaces>\n";
 430                 foreach ( $wgContLang->getFormattedNamespaces() as $ns => $title ) {
 431                         $spaces .= '      ' .
 432                                 Xml::element( 'namespace',
 433                                         array(  'key' => $ns,
 434                                                         'case' => MWNamespace::isCapitalized( $ns ) ? 'first-letter' : 'case-sensitive',
 435                                         ), $title ) . "\n";
 436                 }
 437                 $spaces .= "    </namespaces>";
 438                 return $spaces;
 439         }
 440
 441         /**
 442          * Closes the output stream with the closing root element.
 443          * Call when finished dumping things.
 444          *
 445          * @return string
 446          */
 447         function closeStream() {
 448                 return "</mediawiki>\n";
 449         }
 450
 451         /**
 452          * Opens a <page> section on the output stream, with data
 453          * from the given database row.
 454          *
 455          * @param $row object
 456          * @return string
 457          * @access private
 458          */
 459         function openPage( $row ) {
 460                 $out = "  <page>\n";
 461                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 462                 $out .= '    ' . Xml::elementClean( 'title', array(), $title->getPrefixedText() ) . "\n";
 463                 $out .= '    ' . Xml::element( 'id', array(), strval( $row->page_id ) ) . "\n";
 464                 $this->pageInProgress = $row->page_id;
 465                 if ( $row->page_is_redirect ) {
 466                         $out .= '    ' . Xml::element( 'redirect', array() ) . "\n";
 467                 }
 468                 if ( $row->page_restrictions != '' ) {
 469                         $out .= '    ' . Xml::element( 'restrictions', array(),
 470                                 strval( $row->page_restrictions ) ) . "\n";
 471                 }
 472
 473                 wfRunHooks( 'XmlDumpWriterOpenPage', array( $this, &$out, $row, $title ) );
 474
 475                 return $out;
 476         }
 477
 478         /**
 479          * Closes a <page> section on the output stream.
 480          *
 481          * @access private
 482          */
 483         function closePage() {
 484                 return "  </page>\n";
 485                 if (! $this->firstPageWritten) {
 486                         $this->firstPageWritten = $this->pageInProgress;
 487                 }
 488                 $this->lastPageWritten = $this->pageInProgress;
 489         }
 490
 491         /**
 492          * Dumps a <revision> section on the output stream, with
 493          * data filled in from the given database row.
 494          *
 495          * @param $row object
 496          * @return string
 497          * @access private
 498          */
 499         function writeRevision( $row ) {
 500                 wfProfileIn( __METHOD__ );
 501
 502                 $out  = "    <revision>\n";
 503                 $out .= "      " . Xml::element( 'id', null, strval( $row->rev_id ) ) . "\n";
 504
 505                 $out .= $this->writeTimestamp( $row->rev_timestamp );
 506
 507                 if ( $row->rev_deleted & Revision::DELETED_USER ) {
 508                         $out .= "      " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 509                 } else {
 510                         $out .= $this->writeContributor( $row->rev_user, $row->rev_user_text );
 511                 }
 512
 513                 if ( $row->rev_minor_edit ) {
 514                         $out .=  "      <minor/>\n";
 515                 }
 516                 if ( $row->rev_deleted & Revision::DELETED_COMMENT ) {
 517                         $out .= "      " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 518                 } elseif ( $row->rev_comment != '' ) {
 519                         $out .= "      " . Xml::elementClean( 'comment', null, strval( $row->rev_comment ) ) . "\n";
 520                 }
 521
 522                 $text = '';
 523                 if ( $row->rev_deleted & Revision::DELETED_TEXT ) {
 524                         $out .= "      " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 525                 } elseif ( isset( $row->old_text ) ) {
 526                         // Raw text from the database may have invalid chars
 527                         $text = strval( Revision::getRevisionText( $row ) );
 528                         $out .= "      " . Xml::elementClean( 'text',
 529                                 array( 'xml:space' => 'preserve', 'bytes' => $row->rev_len ),
 530                                 strval( $text ) ) . "\n";
 531                 } else {
 532                         // Stub output
 533                         $out .= "      " . Xml::element( 'text',
 534                                 array( 'id' => $row->rev_text_id, 'bytes' => $row->rev_len ),
 535                                 "" ) . "\n";
 536                 }
 537
 538                 wfRunHooks( 'XmlDumpWriterWriteRevision', array( &$this, &$out, $row, $text ) );
 539
 540                 $out .= "    </revision>\n";
 541
 542                 wfProfileOut( __METHOD__ );
 543                 return $out;
 544         }
 545
 546         /**
 547          * Dumps a <logitem> section on the output stream, with
 548          * data filled in from the given database row.
 549          *
 550          * @param $row object
 551          * @return string
 552          * @access private
 553          */
 554         function writeLogItem( $row ) {
 555                 wfProfileIn( __METHOD__ );
 556
 557                 $out  = "    <logitem>\n";
 558                 $out .= "      " . Xml::element( 'id', null, strval( $row->log_id ) ) . "\n";
 559
 560                 $out .= $this->writeTimestamp( $row->log_timestamp );
 561
 562                 if ( $row->log_deleted & LogPage::DELETED_USER ) {
 563                         $out .= "      " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 564                 } else {
 565                         $out .= $this->writeContributor( $row->log_user, $row->user_name );
 566                 }
 567
 568                 if ( $row->log_deleted & LogPage::DELETED_COMMENT ) {
 569                         $out .= "      " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 570                 } elseif ( $row->log_comment != '' ) {
 571                         $out .= "      " . Xml::elementClean( 'comment', null, strval( $row->log_comment ) ) . "\n";
 572                 }
 573
 574                 $out .= "      " . Xml::element( 'type', null, strval( $row->log_type ) ) . "\n";
 575                 $out .= "      " . Xml::element( 'action', null, strval( $row->log_action ) ) . "\n";
 576
 577                 if ( $row->log_deleted & LogPage::DELETED_ACTION ) {
 578                         $out .= "      " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 579                 } else {
 580                         $title = Title::makeTitle( $row->log_namespace, $row->log_title );
 581                         $out .= "      " . Xml::elementClean( 'logtitle', null, $title->getPrefixedText() ) . "\n";
 582                         $out .= "      " . Xml::elementClean( 'params',
 583                                 array( 'xml:space' => 'preserve' ),
 584                                 strval( $row->log_params ) ) . "\n";
 585                 }
 586
 587                 $out .= "    </logitem>\n";
 588
 589                 wfProfileOut( __METHOD__ );
 590                 return $out;
 591         }
 592
 593         function writeTimestamp( $timestamp ) {
 594                 $ts = wfTimestamp( TS_ISO_8601, $timestamp );
 595                 return "      " . Xml::element( 'timestamp', null, $ts ) . "\n";
 596         }
 597
 598         function writeContributor( $id, $text ) {
 599                 $out = "      <contributor>\n";
 600                 if ( $id ) {
 601                         $out .= "        " . Xml::elementClean( 'username', null, strval( $text ) ) . "\n";
 602                         $out .= "        " . Xml::element( 'id', null, strval( $id ) ) . "\n";
 603                 } else {
 604                         $out .= "        " . Xml::elementClean( 'ip', null, strval( $text ) ) . "\n";
 605                 }
 606                 $out .= "      </contributor>\n";
 607                 return $out;
 608         }
 609
 610         /**
 611          * Warning! This data is potentially inconsistent. :(
 612          */
 613         function writeUploads( $row, $dumpContents = false ) {
 614                 if ( $row->page_namespace == NS_IMAGE ) {
 615                         $img = wfLocalFile( $row->page_title );
 616                         if ( $img && $img->exists() ) {
 617                                 $out = '';
 618                                 foreach ( array_reverse( $img->getHistory() ) as $ver ) {
 619                                         $out .= $this->writeUpload( $ver, $dumpContents );
 620                                 }
 621                                 $out .= $this->writeUpload( $img, $dumpContents );
 622                                 return $out;
 623                         }
 624                 }
 625                 return '';
 626         }
 627
 628         /**
 629          * @param $file File
 630          * @param $dumpContents bool
 631          * @return string
 632          */
 633         function writeUpload( $file, $dumpContents = false ) {
 634                 if ( $file->isOld() ) {
 635                         $archiveName = "      " .
 636                                 Xml::element( 'archivename', null, $file->getArchiveName() ) . "\n";
 637                 } else {
 638                         $archiveName = '';
 639                 }
 640                 if ( $dumpContents ) {
 641                         # Dump file as base64
 642                         # Uses only XML-safe characters, so does not need escaping
 643                         $contents = '      <contents encoding="base64">' .
 644                                 chunk_split( base64_encode( file_get_contents( $file->getPath() ) ) ) .
 645                                 "      </contents>\n";
 646                 } else {
 647                         $contents = '';
 648                 }
 649                 return "    <upload>\n" .
 650                         $this->writeTimestamp( $file->getTimestamp() ) .
 651                         $this->writeContributor( $file->getUser( 'id' ), $file->getUser( 'text' ) ) .
 652                         "      " . Xml::elementClean( 'comment', null, $file->getDescription() ) . "\n" .
 653                         "      " . Xml::element( 'filename', null, $file->getName() ) . "\n" .
 654                         $archiveName .
 655                         "      " . Xml::element( 'src', null, $file->getCanonicalUrl() ) . "\n" .
 656                         "      " . Xml::element( 'size', null, $file->getSize() ) . "\n" .
 657                         "      " . Xml::element( 'sha1base36', null, $file->getSha1() ) . "\n" .
 658                         "      " . Xml::element( 'rel', null, $file->getRel() ) . "\n" .
 659                         $contents .
 660                         "    </upload>\n";
 661         }
 662
 663 }
 664
 665
 666 /**
 667  * Base class for output stream; prints to stdout or buffer or whereever.
 668  * @ingroup Dump
 669  */
 670 class DumpOutput {
 671         function writeOpenStream( $string ) {
 672                 $this->write( $string );
 673         }
 674
 675         function writeCloseStream( $string ) {
 676                 $this->write( $string );
 677         }
 678
 679         function writeOpenPage( $page, $string ) {
 680                 $this->write( $string );
 681         }
 682
 683         function writeClosePage( $string ) {
 684                 $this->write( $string );
 685         }
 686
 687         function writeRevision( $rev, $string ) {
 688                 $this->write( $string );
 689         }
 690
 691         function writeLogItem( $rev, $string ) {
 692                 $this->write( $string );
 693         }
 694
 695         /**
 696          * Override to write to a different stream type.
 697          * @return bool
 698          */
 699         function write( $string ) {
 700                 print $string;
 701         }
 702
 703         function closeRenameAndReopen( $newname ) {
 704                 return;
 705         }
 706
 707         function closeAndRename( $newname ) {
 708                 return;
 709         }
 710
 711         function rename( $newname ) {
 712                 return;
 713         }
 714
 715         function getFilename() {
 716                 return NULL;
 717         }
 718 }
 719
 720 /**
 721  * Stream outputter to send data to a file.
 722  * @ingroup Dump
 723  */
 724 class DumpFileOutput extends DumpOutput {
 725         var $handle;
 726         var $filename;
 727
 728         function __construct( $file ) {
 729                 $this->handle = fopen( $file, "wt" );
 730                 $this->filename = $file;
 731         }
 732
 733         function write( $string ) {
 734                 fputs( $this->handle, $string );
 735         }
 736
 737         /**
 738          * Close the old file, move it to a specified name,
 739          * and reopen new file with the old name. Use this
 740          * for writing out a file in multiple pieces
 741          * at specified checkpoints (e.g. every n hours).
 742          */
 743         function closeRenameAndReopen( $newname ) {
 744                 if ( is_array($newname) ) {
 745                         if (count($newname) > 1) {
 746                                 throw new MWException("Export closeRenameAndReopen: passed multiple argumnts for rename of single file\n");
 747                         }
 748                         else {
 749                                 $newname = $newname[0];
 750                         }
 751                 }
 752                 if ( $newname ) {
 753                         fclose( $this->handle );
 754                         rename( $this->filename, $newname );
 755                         $this->handle = fopen( $this->filename, "wt" );
 756                 }
 757         }
 758
 759         function closeAndRename( $newname ) {
 760                 if ( is_array($newname) ) {
 761                         if (count($newname) > 1) {
 762                                 throw new MWException("Export closeRenameAndReopen: passed multiple argumnts for rename of single file\n");
 763                         }
 764                         else {
 765                                 $newname = $newname[0];
 766                         }
 767                 }
 768                 if ( $newname ) {
 769                         fclose( $this->handle );
 770                         rename( $this->filename, $newname );
 771                 }
 772         }
 773
 774         function rename( $newname ) {
 775                 if ( is_array($newname) ) {
 776                         if (count($newname) > 1) {
 777                                 throw new MWException("Export closeRenameAndReopen: passed multiple argumnts for rename of single file\n");
 778                         }
 779                         else {
 780                                 $newname = $newname[0];
 781                         }
 782                 }
 783                 if ( $newname ) {
 784                         rename( $this->filename, $newname );
 785                 }
 786         }
 787
 788         function getFilename() {
 789                 return $this->filename;
 790         }
 791 }
 792
 793 /**
 794  * Stream outputter to send data to a file via some filter program.
 795  * Even if compression is available in a library, using a separate
 796  * program can allow us to make use of a multi-processor system.
 797  * @ingroup Dump
 798  */
 799 class DumpPipeOutput extends DumpFileOutput {
 800         var $command;
 801
 802         function __construct( $command, $file = null ) {
 803                 if ( !is_null( $file ) ) {
 804                         $command .=  " > " . wfEscapeShellArg( $file );
 805                 }
 806
 807                 $this->startCommand($command);
 808                 $this->command = $command;
 809                 $this->filename = $file;
 810         }
 811
 812         function startCommand($command) {
 813                 $spec = array(
 814                         0 => array( "pipe", "r" ),
 815                 );
 816                 $pipes = array();
 817                 $this->procOpenResource = proc_open( $command, $spec, $pipes );
 818                 $this->handle = $pipes[0];
 819         }
 820
 821         /**
 822          * Close the old file, move it to a specified name,
 823          * and reopen new file with the old name.
 824          */
 825         function closeRenameAndReopen( $newname ) {
 826                 if ( is_array($newname) ) {
 827                         if (count($newname) > 1) {
 828                                 throw new MWException("Export closeRenameAndReopen: passed multiple argumnts for rename of single file\n");
 829                         }
 830                         else {
 831                                 $newname = $newname[0];
 832                         }
 833                 }
 834                 if ( $newname ) {
 835                         fclose( $this->handle );
 836                         proc_close($this->procOpenResource);
 837                         rename( $this->filename, $newname );
 838                         $command = $this->command;
 839                         $command .=  " > " . wfEscapeShellArg( $this->filename );
 840                         $this->startCommand($command);
 841                 }
 842         }
 843
 844         function closeAndRename( $newname ) {
 845                 if ( is_array($newname) ) {
 846                         if (count($newname) > 1) {
 847                                 throw new MWException("Export closeRenameAndReopen: passed multiple argumnts for rename of single file\n");
 848                         }
 849                         else {
 850                                 $newname = $newname[0];
 851                         }
 852                 }
 853                 if ( $newname ) {
 854 #                       pclose( $this->handle );
 855                         fclose( $this->handle );
 856                         proc_close($this->procOpenResource);
 857                         rename( $this->filename, $newname );
 858                 }
 859         }
 860
 861         function rename( $newname ) {
 862                 if ( is_array($newname) ) {
 863                         if (count($newname) > 1) {
 864                                 throw new MWException("Export closeRenameAndReopen: passed multiple argumnts for rename of single file\n");
 865                         }
 866                         else {
 867                                 $newname = $newname[0];
 868                         }
 869                 }
 870                 if ( $newname ) {
 871                         rename( $this->filename, $newname );
 872                 }
 873         }
 874 }
 875
 876 /**
 877  * Sends dump output via the gzip compressor.
 878  * @ingroup Dump
 879  */
 880 class DumpGZipOutput extends DumpPipeOutput {
 881         function __construct( $file ) {
 882                 parent::__construct( "gzip", $file );
 883         }
 884 }
 885
 886 /**
 887  * Sends dump output via the bgzip2 compressor.
 888  * @ingroup Dump
 889  */
 890 class DumpBZip2Output extends DumpPipeOutput {
 891         function __construct( $file ) {
 892                 parent::__construct( "bzip2", $file );
 893         }
 894 }
 895
 896 /**
 897  * Sends dump output via the p7zip compressor.
 898  * @ingroup Dump
 899  */
 900 class Dump7ZipOutput extends DumpPipeOutput {
 901         var $filename;
 902
 903         function __construct( $file ) {
 904                 $command = "7za a -bd -si " . wfEscapeShellArg( $file );
 905                 // Suppress annoying useless crap from p7zip
 906                 // Unfortunately this could suppress real error messages too
 907                 $command .= ' >' . wfGetNull() . ' 2>&1';
 908                 parent::__construct( $command );
 909                 $this->filename = $file;
 910         }
 911
 912         function closeRenameAndReopen( $newname ) {
 913                 if ( is_array($newname) ) {
 914                         if (count($newname) > 1) {
 915                                 throw new MWException("Export closeRenameAndReopen: passed multiple argumnts for rename of single file\n");
 916                         }
 917                         else {
 918                                 $newname = $newname[0];
 919                         }
 920                 }
 921                 if ( $newname ) {
 922                         fclose( $this->handle );
 923                         proc_close($this->procOpenResource);
 924                         rename( $this->filename, $newname );
 925                         $command = "7za a -bd -si " . wfEscapeShellArg( $file );
 926                         $command .= ' >' . wfGetNull() . ' 2>&1';
 927                         $this->startCommand($command);
 928                 }
 929         }
 930
 931         function closeAndRename( $newname ) {
 932                 if ( is_array($newname) ) {
 933                         if (count($newname) > 1) {
 934                                 throw new MWException("Export closeRenameAndReopen: passed multiple argumnts for rename of single file\n");
 935                         }
 936                         else {
 937                                 $newname = $newname[0];
 938                         }
 939                 }
 940                 if ( $newname ) {
 941                         fclose( $this->handle );
 942                         proc_close($this->procOpenResource);
 943                         rename( $this->filename, $newname );
 944                 }
 945         }
 946
 947         function rename( $newname ) {
 948                 if ( is_array($newname) ) {
 949                         if (count($newname) > 1) {
 950                                 throw new MWException("Export closeRenameAndReopen: passed multiple argumnts for rename of single file\n");
 951                         }
 952                         else {
 953                                 $newname = $newname[0];
 954                         }
 955                 }
 956                 if ( $newname ) {
 957                         rename( $this->filename, $newname );
 958                 }
 959         }
 960 }
 961
 962
 963
 964 /**
 965  * Dump output filter class.
 966  * This just does output filtering and streaming; XML formatting is done
 967  * higher up, so be careful in what you do.
 968  * @ingroup Dump
 969  */
 970 class DumpFilter {
 971         function __construct( &$sink ) {
 972                 $this->sink =& $sink;
 973         }
 974
 975         function writeOpenStream( $string ) {
 976                 $this->sink->writeOpenStream( $string );
 977         }
 978
 979         function writeCloseStream( $string ) {
 980                 $this->sink->writeCloseStream( $string );
 981         }
 982
 983         function writeOpenPage( $page, $string ) {
 984                 $this->sendingThisPage = $this->pass( $page, $string );
 985                 if ( $this->sendingThisPage ) {
 986                         $this->sink->writeOpenPage( $page, $string );
 987                 }
 988         }
 989
 990         function writeClosePage( $string ) {
 991                 if ( $this->sendingThisPage ) {
 992                         $this->sink->writeClosePage( $string );
 993                         $this->sendingThisPage = false;
 994                 }
 995         }
 996
 997         function writeRevision( $rev, $string ) {
 998                 if ( $this->sendingThisPage ) {
 999                         $this->sink->writeRevision( $rev, $string );
1000                 }
1001         }
1002
1003         function writeLogItem( $rev, $string ) {
1004                 $this->sink->writeRevision( $rev, $string );
1005         }
1006
1007         function closeRenameAndReopen( $newname ) {
1008                 $this->sink->closeRenameAndReopen( $newname );
1009         }
1010
1011         function closeAndRename( $newname ) {
1012                 $this->sink->closeAndRename( $newname );
1013         }
1014
1015         function rename( $newname ) {
1016                 $this->sink->rename( $newname );
1017         }
1018
1019         function getFilename() {
1020                 return $this->sink->getFilename();
1021         }
1022
1023         /**
1024          * Override for page-based filter types.
1025          * @return bool
1026          */
1027         function pass( $page ) {
1028                 return true;
1029         }
1030 }
1031
1032 /**
1033  * Simple dump output filter to exclude all talk pages.
1034  * @ingroup Dump
1035  */
1036 class DumpNotalkFilter extends DumpFilter {
1037         function pass( $page ) {
1038                 return !MWNamespace::isTalk( $page->page_namespace );
1039         }
1040 }
1041
1042 /**
1043  * Dump output filter to include or exclude pages in a given set of namespaces.
1044  * @ingroup Dump
1045  */
1046 class DumpNamespaceFilter extends DumpFilter {
1047         var $invert = false;
1048         var $namespaces = array();
1049
1050         function __construct( &$sink, $param ) {
1051                 parent::__construct( $sink );
1052
1053                 $constants = array(
1054                         "NS_MAIN"           => NS_MAIN,
1055                         "NS_TALK"           => NS_TALK,
1056                         "NS_USER"           => NS_USER,
1057                         "NS_USER_TALK"      => NS_USER_TALK,
1058                         "NS_PROJECT"        => NS_PROJECT,
1059                         "NS_PROJECT_TALK"   => NS_PROJECT_TALK,
1060                         "NS_FILE"           => NS_FILE,
1061                         "NS_FILE_TALK"      => NS_FILE_TALK,
1062                         "NS_IMAGE"          => NS_IMAGE,  // NS_IMAGE is an alias for NS_FILE
1063                         "NS_IMAGE_TALK"     => NS_IMAGE_TALK,
1064                         "NS_MEDIAWIKI"      => NS_MEDIAWIKI,
1065                         "NS_MEDIAWIKI_TALK" => NS_MEDIAWIKI_TALK,
1066                         "NS_TEMPLATE"       => NS_TEMPLATE,
1067                         "NS_TEMPLATE_TALK"  => NS_TEMPLATE_TALK,
1068                         "NS_HELP"           => NS_HELP,
1069                         "NS_HELP_TALK"      => NS_HELP_TALK,
1070                         "NS_CATEGORY"       => NS_CATEGORY,
1071                         "NS_CATEGORY_TALK"  => NS_CATEGORY_TALK );
1072
1073                 if ( $param { 0 } == '!' ) {
1074                         $this->invert = true;
1075                         $param = substr( $param, 1 );
1076                 }
1077
1078                 foreach ( explode( ',', $param ) as $key ) {
1079                         $key = trim( $key );
1080                         if ( isset( $constants[$key] ) ) {
1081                                 $ns = $constants[$key];
1082                                 $this->namespaces[$ns] = true;
1083                         } elseif ( is_numeric( $key ) ) {
1084                                 $ns = intval( $key );
1085                                 $this->namespaces[$ns] = true;
1086                         } else {
1087                                 throw new MWException( "Unrecognized namespace key '$key'\n" );
1088                         }
1089                 }
1090         }
1091
1092         function pass( $page ) {
1093                 $match = isset( $this->namespaces[$page->page_namespace] );
1094                 return $this->invert xor $match;
1095         }
1096 }
1097
1098
1099 /**
1100  * Dump output filter to include only the last revision in each page sequence.
1101  * @ingroup Dump
1102  */
1103 class DumpLatestFilter extends DumpFilter {
1104         var $page, $pageString, $rev, $revString;
1105
1106         function writeOpenPage( $page, $string ) {
1107                 $this->page = $page;
1108                 $this->pageString = $string;
1109         }
1110
1111         function writeClosePage( $string ) {
1112                 if ( $this->rev ) {
1113                         $this->sink->writeOpenPage( $this->page, $this->pageString );
1114                         $this->sink->writeRevision( $this->rev, $this->revString );
1115                         $this->sink->writeClosePage( $string );
1116                 }
1117                 $this->rev = null;
1118                 $this->revString = null;
1119                 $this->page = null;
1120                 $this->pageString = null;
1121         }
1122
1123         function writeRevision( $rev, $string ) {
1124                 if ( $rev->rev_id == $this->page->page_latest ) {
1125                         $this->rev = $rev;
1126                         $this->revString = $string;
1127                 }
1128         }
1129 }
1130
1131 /**
1132  * Base class for output stream; prints to stdout or buffer or whereever.
1133  * @ingroup Dump
1134  */
1135 class DumpMultiWriter {
1136         function __construct( $sinks ) {
1137                 $this->sinks = $sinks;
1138                 $this->count = count( $sinks );
1139         }
1140
1141         function writeOpenStream( $string ) {
1142                 for ( $i = 0; $i < $this->count; $i++ ) {
1143                         $this->sinks[$i]->writeOpenStream( $string );
1144                 }
1145         }
1146
1147         function writeCloseStream( $string ) {
1148                 for ( $i = 0; $i < $this->count; $i++ ) {
1149                         $this->sinks[$i]->writeCloseStream( $string );
1150                 }
1151         }
1152
1153         function writeOpenPage( $page, $string ) {
1154                 for ( $i = 0; $i < $this->count; $i++ ) {
1155                         $this->sinks[$i]->writeOpenPage( $page, $string );
1156                 }
1157         }
1158
1159         function writeClosePage( $string ) {
1160                 for ( $i = 0; $i < $this->count; $i++ ) {
1161                         $this->sinks[$i]->writeClosePage( $string );
1162                 }
1163         }
1164
1165         function writeRevision( $rev, $string ) {
1166                 for ( $i = 0; $i < $this->count; $i++ ) {
1167                         $this->sinks[$i]->writeRevision( $rev, $string );
1168                 }
1169         }
1170
1171         function closeRenameAndReopen( $newnames ) {
1172                 for( $i = 0; $i < $this->count; $i++ ) {
1173                         $this->sinks[$i]->closeRenameAndReopen( $newnames[$i] );
1174                 }
1175         }
1176
1177         function closeAndRename( $newname ) {
1178                 for( $i = 0; $i < $this->count; $i++ ) {
1179                         $this->sinks[$i]->closeAndRename( $newnames[$i] );
1180                 }
1181         }
1182         function rename( $newnames ) {
1183                 for( $i = 0; $i < $this->count; $i++ ) {
1184                         $this->sinks[$i]->rename( $newnames[$i] );
1185                 }
1186         }
1187
1188         function getFilename() {
1189                 $filenames = array();
1190                 for( $i = 0; $i < $this->count; $i++ ) {
1191                         $filenames[] =  $this->sinks[$i]->getFilename();
1192                 }
1193                 return $filenames;
1194         }
1195
1196 }
1197
1198 function xmlsafe( $string ) {
1199         wfProfileIn( __FUNCTION__ );
1200
1201         /**
1202          * The page may contain old data which has not been properly normalized.
1203          * Invalid UTF-8 sequences or forbidden control characters will make our
1204          * XML output invalid, so be sure to strip them out.
1205          */
1206         $string = UtfNormal::cleanUp( $string );
1207
1208         $string = htmlspecialchars( $string );
1209         wfProfileOut( __FUNCTION__ );
1210         return $string;
1211 }