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         const RANGE = 16;
  45
  46         const BUFFER = 0;
  47         const STREAM = 1;
  48
  49         const TEXT = 0;
  50         const STUB = 1;
  51
  52         /**
  53          * If using WikiExporter::STREAM to stream a large amount of data,
  54          * provide a database connection which is not managed by
  55          * LoadBalancer to read from: some history blob types will
  56          * make additional queries to pull source data while the
  57          * main query is still running.
  58          *
  59          * @param $db DatabaseBase
  60          * @param $history Mixed: one of WikiExporter::FULL, WikiExporter::CURRENT,
  61          *                 WikiExporter::RANGE or WikiExporter::STABLE,
  62          *                 or an associative array:
  63          *                   offset: non-inclusive offset at which to start the query
  64          *                   limit: maximum number of rows to return
  65          *                   dir: "asc" or "desc" timestamp order
  66          * @param $buffer Int: one of WikiExporter::BUFFER or WikiExporter::STREAM
  67          * @param $text Int: one of WikiExporter::TEXT or WikiExporter::STUB
  68          */
  69         function __construct( &$db, $history = WikiExporter::CURRENT,
  70                         $buffer = WikiExporter::BUFFER, $text = WikiExporter::TEXT ) {
  71                 $this->db =& $db;
  72                 $this->history = $history;
  73                 $this->buffer  = $buffer;
  74                 $this->writer  = new XmlDumpWriter();
  75                 $this->sink    = new DumpOutput();
  76                 $this->text    = $text;
  77         }
  78
  79         /**
  80          * Set the DumpOutput or DumpFilter object which will receive
  81          * various row objects and XML output for filtering. Filters
  82          * can be chained or used as callbacks.
  83          *
  84          * @param $sink mixed
  85          */
  86         public function setOutputSink( &$sink ) {
  87                 $this->sink =& $sink;
  88         }
  89
  90         public function openStream() {
  91                 $output = $this->writer->openStream();
  92                 $this->sink->writeOpenStream( $output );
  93         }
  94
  95         public function closeStream() {
  96                 $output = $this->writer->closeStream();
  97                 $this->sink->writeCloseStream( $output );
  98         }
  99
 100         /**
 101          * Dumps a series of page and revision records for all pages
 102          * in the database, either including complete history or only
 103          * the most recent version.
 104          */
 105         public function allPages() {
 106                 return $this->dumpFrom( '' );
 107         }
 108
 109         /**
 110          * Dumps a series of page and revision records for those pages
 111          * in the database falling within the page_id range given.
 112          * @param $start Int: inclusive lower limit (this id is included)
 113          * @param $end   Int: Exclusive upper limit (this id is not included)
 114          *                   If 0, no upper limit.
 115          */
 116         public function pagesByRange( $start, $end ) {
 117                 $condition = 'page_id >= ' . intval( $start );
 118                 if ( $end ) {
 119                         $condition .= ' AND page_id < ' . intval( $end );
 120                 }
 121                 return $this->dumpFrom( $condition );
 122         }
 123
 124         /**
 125          * Dumps a series of page and revision records for those pages
 126          * in the database with revisions falling within the rev_id range given.
 127          * @param $start Int: inclusive lower limit (this id is included)
 128          * @param $end   Int: Exclusive upper limit (this id is not included)
 129          *                   If 0, no upper limit.
 130          */
 131         public function revsByRange( $start, $end ) {
 132                 $condition = 'rev_id >= ' . intval( $start );
 133                 if ( $end ) {
 134                         $condition .= ' AND rev_id < ' . intval( $end );
 135                 }
 136                 return $this->dumpFrom( $condition );
 137         }
 138
 139         /**
 140          * @param $title Title
 141          */
 142         public function pageByTitle( $title ) {
 143                 return $this->dumpFrom(
 144                         'page_namespace=' . $title->getNamespace() .
 145                         ' AND page_title=' . $this->db->addQuotes( $title->getDBkey() ) );
 146         }
 147
 148         public function pageByName( $name ) {
 149                 $title = Title::newFromText( $name );
 150                 if ( is_null( $title ) ) {
 151                         throw new MWException( "Can't export invalid title" );
 152                 } else {
 153                         return $this->pageByTitle( $title );
 154                 }
 155         }
 156
 157         public function pagesByName( $names ) {
 158                 foreach ( $names as $name ) {
 159                         $this->pageByName( $name );
 160                 }
 161         }
 162
 163         public function allLogs() {
 164                 return $this->dumpFrom( '' );
 165         }
 166
 167         public function logsByRange( $start, $end ) {
 168                 $condition = 'log_id >= ' . intval( $start );
 169                 if ( $end ) {
 170                         $condition .= ' AND log_id < ' . intval( $end );
 171                 }
 172                 return $this->dumpFrom( $condition );
 173         }
 174
 175         # Generates the distinct list of authors of an article
 176         # Not called by default (depends on $this->list_authors)
 177         # Can be set by Special:Export when not exporting whole history
 178         protected function do_list_authors( $cond ) {
 179                 wfProfileIn( __METHOD__ );
 180                 $this->author_list = "<contributors>";
 181                 // rev_deleted
 182
 183                 $res = $this->db->select(
 184                         array( 'page', 'revision' ),
 185                         array( 'DISTINCT rev_user_text', 'rev_user' ),
 186                         array(
 187                                 $this->db->bitAnd( 'rev_deleted', Revision::DELETED_USER ) . ' = 0',
 188                                 $cond,
 189                                 'page_id = rev_id',
 190                         ),
 191                         __METHOD__
 192                 );
 193
 194                 foreach ( $res as $row ) {
 195                         $this->author_list .= "<contributor>" .
 196                                 "<username>" .
 197                                 htmlentities( $row->rev_user_text )  .
 198                                 "</username>" .
 199                                 "<id>" .
 200                                 $row->rev_user .
 201                                 "</id>" .
 202                                 "</contributor>";
 203                 }
 204                 $this->author_list .= "</contributors>";
 205                 wfProfileOut( __METHOD__ );
 206         }
 207
 208         protected function dumpFrom( $cond = '' ) {
 209                 wfProfileIn( __METHOD__ );
 210                 # For logging dumps...
 211                 if ( $this->history & self::LOGS ) {
 212                         if ( $this->buffer == WikiExporter::STREAM ) {
 213                                 $prev = $this->db->bufferResults( false );
 214                         }
 215                         $where = array( 'user_id = log_user' );
 216                         # Hide private logs
 217                         $hideLogs = LogEventsList::getExcludeClause( $this->db );
 218                         if ( $hideLogs ) $where[] = $hideLogs;
 219                         # Add on any caller specified conditions
 220                         if ( $cond ) $where[] = $cond;
 221                         # Get logging table name for logging.* clause
 222                         $logging = $this->db->tableName( 'logging' );
 223                         $result = $this->db->select( array( 'logging', 'user' ),
 224                                 array( "{$logging}.*", 'user_name' ), // grab the user name
 225                                 $where,
 226                                 __METHOD__,
 227                                 array( 'ORDER BY' => 'log_id', 'USE INDEX' => array( 'logging' => 'PRIMARY' ) )
 228                         );
 229                         $wrapper = $this->db->resultObject( $result );
 230                         $this->outputLogStream( $wrapper );
 231                         if ( $this->buffer == WikiExporter::STREAM ) {
 232                                 $this->db->bufferResults( $prev );
 233                         }
 234                 # For page dumps...
 235                 } else {
 236                         $tables = array( 'page', 'revision' );
 237                         $opts = array( 'ORDER BY' => 'page_id ASC' );
 238                         $opts['USE INDEX'] = array();
 239                         $join = array();
 240                         if ( is_array( $this->history ) ) {
 241                                 # Time offset/limit for all pages/history...
 242                                 $revJoin = 'page_id=rev_page';
 243                                 # Set time order
 244                                 if ( $this->history['dir'] == 'asc' ) {
 245                                         $op = '>';
 246                                         $opts['ORDER BY'] = 'rev_timestamp ASC';
 247                                 } else {
 248                                         $op = '<';
 249                                         $opts['ORDER BY'] = 'rev_timestamp DESC';
 250                                 }
 251                                 # Set offset
 252                                 if ( !empty( $this->history['offset'] ) ) {
 253                                         $revJoin .= " AND rev_timestamp $op " .
 254                                                 $this->db->addQuotes( $this->db->timestamp( $this->history['offset'] ) );
 255                                 }
 256                                 $join['revision'] = array( 'INNER JOIN', $revJoin );
 257                                 # Set query limit
 258                                 if ( !empty( $this->history['limit'] ) ) {
 259                                         $opts['LIMIT'] = intval( $this->history['limit'] );
 260                                 }
 261                         } elseif ( $this->history & WikiExporter::FULL ) {
 262                                 # Full history dumps...
 263                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page' );
 264                         } elseif ( $this->history & WikiExporter::CURRENT ) {
 265                                 # Latest revision dumps...
 266                                 if ( $this->list_authors && $cond != '' )  { // List authors, if so desired
 267                                         $this->do_list_authors( $cond );
 268                                 }
 269                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' );
 270                         } elseif ( $this->history & WikiExporter::STABLE ) {
 271                                 # "Stable" revision dumps...
 272                                 # Default JOIN, to be overridden...
 273                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' );
 274                                 # One, and only one hook should set this, and return false
 275                                 if ( wfRunHooks( 'WikiExporter::dumpStableQuery', array( &$tables, &$opts, &$join ) ) ) {
 276                                         wfProfileOut( __METHOD__ );
 277                                         throw new MWException( __METHOD__ . " given invalid history dump type." );
 278                                 }
 279                         } elseif ( $this->history & WikiExporter::RANGE ) {
 280                                 # Dump of revisions within a specified range
 281                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page' );
 282                                 $opts['ORDER BY'] = 'rev_page ASC, rev_id ASC';
 283                         } else {
 284                                 # Uknown history specification parameter?
 285                                 wfProfileOut( __METHOD__ );
 286                                 throw new MWException( __METHOD__ . " given invalid history dump type." );
 287                         }
 288                         # Query optimization hacks
 289                         if ( $cond == '' ) {
 290                                 $opts[] = 'STRAIGHT_JOIN';
 291                                 $opts['USE INDEX']['page'] = 'PRIMARY';
 292                         }
 293                         # Build text join options
 294                         if ( $this->text != WikiExporter::STUB ) { // 1-pass
 295                                 $tables[] = 'text';
 296                                 $join['text'] = array( 'INNER JOIN', 'rev_text_id=old_id' );
 297                         }
 298
 299                         if ( $this->buffer == WikiExporter::STREAM ) {
 300                                 $prev = $this->db->bufferResults( false );
 301                         }
 302
 303                         wfRunHooks( 'ModifyExportQuery',
 304                                                 array( $this->db, &$tables, &$cond, &$opts, &$join ) );
 305
 306                         # Do the query!
 307                         $result = $this->db->select( $tables, '*', $cond, __METHOD__, $opts, $join );
 308                         $wrapper = $this->db->resultObject( $result );
 309                         # Output dump results
 310                         $this->outputPageStream( $wrapper );
 311                         if ( $this->list_authors ) {
 312                                 $this->outputPageStream( $wrapper );
 313                         }
 314
 315                         if ( $this->buffer == WikiExporter::STREAM ) {
 316                                 $this->db->bufferResults( $prev );
 317                         }
 318                 }
 319                 wfProfileOut( __METHOD__ );
 320         }
 321
 322         /**
 323          * Runs through a query result set dumping page and revision records.
 324          * The result set should be sorted/grouped by page to avoid duplicate
 325          * page records in the output.
 326          *
 327          * The result set will be freed once complete. Should be safe for
 328          * streaming (non-buffered) queries, as long as it was made on a
 329          * separate database connection not managed by LoadBalancer; some
 330          * blob storage types will make queries to pull source data.
 331          *
 332          * @param $resultset ResultWrapper
 333          */
 334         protected function outputPageStream( $resultset ) {
 335                 $last = null;
 336                 foreach ( $resultset as $row ) {
 337                         if ( is_null( $last ) ||
 338                                 $last->page_namespace != $row->page_namespace ||
 339                                 $last->page_title     != $row->page_title ) {
 340                                 if ( isset( $last ) ) {
 341                                         $output = '';
 342                                         if ( $this->dumpUploads ) {
 343                                                 $output .= $this->writer->writeUploads( $last, $this->dumpUploadFileContents );
 344                                         }
 345                                         $output .= $this->writer->closePage();
 346                                         $this->sink->writeClosePage( $output );
 347                                 }
 348                                 $output = $this->writer->openPage( $row );
 349                                 $this->sink->writeOpenPage( $row, $output );
 350                                 $last = $row;
 351                         }
 352                         $output = $this->writer->writeRevision( $row );
 353                         $this->sink->writeRevision( $row, $output );
 354                 }
 355                 if ( isset( $last ) ) {
 356                         $output = '';
 357                         if ( $this->dumpUploads ) {
 358                                 $output .= $this->writer->writeUploads( $last, $this->dumpUploadFileContents );
 359                         }
 360                         $output .= $this->author_list;
 361                         $output .= $this->writer->closePage();
 362                         $this->sink->writeClosePage( $output );
 363                 }
 364         }
 365
 366         protected function outputLogStream( $resultset ) {
 367                 foreach ( $resultset as $row ) {
 368                         $output = $this->writer->writeLogItem( $row );
 369                         $this->sink->writeLogItem( $row, $output );
 370                 }
 371         }
 372 }
 373
 374 /**
 375  * @ingroup Dump
 376  */
 377 class XmlDumpWriter {
 378         /**
 379          * Returns the export schema version.
 380          * @return string
 381          */
 382         function schemaVersion() {
 383                 return "0.6";
 384         }
 385
 386         /**
 387          * Opens the XML output stream's root <mediawiki> element.
 388          * This does not include an xml directive, so is safe to include
 389          * as a subelement in a larger XML stream. Namespace and XML Schema
 390          * references are included.
 391          *
 392          * Output will be encoded in UTF-8.
 393          *
 394          * @return string
 395          */
 396         function openStream() {
 397                 global $wgLanguageCode;
 398                 $ver = $this->schemaVersion();
 399                 return Xml::element( 'mediawiki', array(
 400                         'xmlns'              => "http://www.mediawiki.org/xml/export-$ver/",
 401                         'xmlns:xsi'          => "http://www.w3.org/2001/XMLSchema-instance",
 402                         'xsi:schemaLocation' => "http://www.mediawiki.org/xml/export-$ver/ " .
 403                                                 "http://www.mediawiki.org/xml/export-$ver.xsd",
 404                         'version'            => $ver,
 405                         'xml:lang'           => $wgLanguageCode ),
 406                         null ) .
 407                         "\n" .
 408                         $this->siteInfo();
 409         }
 410
 411         function siteInfo() {
 412                 $info = array(
 413                         $this->sitename(),
 414                         $this->homelink(),
 415                         $this->generator(),
 416                         $this->caseSetting(),
 417                         $this->namespaces() );
 418                 return "  <siteinfo>\n    " .
 419                         implode( "\n    ", $info ) .
 420                         "\n  </siteinfo>\n";
 421         }
 422
 423         function sitename() {
 424                 global $wgSitename;
 425                 return Xml::element( 'sitename', array(), $wgSitename );
 426         }
 427
 428         function generator() {
 429                 global $wgVersion;
 430                 return Xml::element( 'generator', array(), "MediaWiki $wgVersion" );
 431         }
 432
 433         function homelink() {
 434                 return Xml::element( 'base', array(), Title::newMainPage()->getCanonicalUrl() );
 435         }
 436
 437         function caseSetting() {
 438                 global $wgCapitalLinks;
 439                 // "case-insensitive" option is reserved for future
 440                 $sensitivity = $wgCapitalLinks ? 'first-letter' : 'case-sensitive';
 441                 return Xml::element( 'case', array(), $sensitivity );
 442         }
 443
 444         function namespaces() {
 445                 global $wgContLang;
 446                 $spaces = "<namespaces>\n";
 447                 foreach ( $wgContLang->getFormattedNamespaces() as $ns => $title ) {
 448                         $spaces .= '      ' .
 449                                 Xml::element( 'namespace',
 450                                         array(  'key' => $ns,
 451                                                         'case' => MWNamespace::isCapitalized( $ns ) ? 'first-letter' : 'case-sensitive',
 452                                         ), $title ) . "\n";
 453                 }
 454                 $spaces .= "    </namespaces>";
 455                 return $spaces;
 456         }
 457
 458         /**
 459          * Closes the output stream with the closing root element.
 460          * Call when finished dumping things.
 461          *
 462          * @return string
 463          */
 464         function closeStream() {
 465                 return "</mediawiki>\n";
 466         }
 467
 468         /**
 469          * Opens a <page> section on the output stream, with data
 470          * from the given database row.
 471          *
 472          * @param $row object
 473          * @return string
 474          * @access private
 475          */
 476         function openPage( $row ) {
 477                 $out = "  <page>\n";
 478                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 479                 $out .= '    ' . Xml::elementClean( 'title', array(), self::canonicalTitle( $title ) ) . "\n";
 480                 $out .= '    ' . Xml::element( 'ns', array(), strval( $row->page_namespace) ) . "\n";
 481                 $out .= '    ' . Xml::element( 'id', array(), strval( $row->page_id ) ) . "\n";
 482                 if ( $row->page_is_redirect ) {
 483                         $page = WikiPage::factory( $title );
 484                         $redirect = $page->getRedirectTarget();
 485                         if ( $redirect instanceOf Title && $redirect->isValidRedirectTarget() ) {
 486                                 $out .= '    ' . Xml::element( 'redirect', array( 'title' => self::canonicalTitle( $redirect ) ) ) . "\n";
 487                         }
 488                 }
 489                 if ( $row->page_restrictions != '' ) {
 490                         $out .= '    ' . Xml::element( 'restrictions', array(),
 491                                 strval( $row->page_restrictions ) ) . "\n";
 492                 }
 493
 494                 wfRunHooks( 'XmlDumpWriterOpenPage', array( $this, &$out, $row, $title ) );
 495
 496                 return $out;
 497         }
 498
 499         /**
 500          * Closes a <page> section on the output stream.
 501          *
 502          * @access private
 503          */
 504         function closePage() {
 505                 return "  </page>\n";
 506         }
 507
 508         /**
 509          * Dumps a <revision> section on the output stream, with
 510          * data filled in from the given database row.
 511          *
 512          * @param $row object
 513          * @return string
 514          * @access private
 515          */
 516         function writeRevision( $row ) {
 517                 wfProfileIn( __METHOD__ );
 518
 519                 $out  = "    <revision>\n";
 520                 $out .= "      " . Xml::element( 'id', null, strval( $row->rev_id ) ) . "\n";
 521
 522                 $out .= $this->writeTimestamp( $row->rev_timestamp );
 523
 524                 if ( $row->rev_deleted & Revision::DELETED_USER ) {
 525                         $out .= "      " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 526                 } else {
 527                         $out .= $this->writeContributor( $row->rev_user, $row->rev_user_text );
 528                 }
 529
 530                 if ( $row->rev_minor_edit ) {
 531                         $out .=  "      <minor/>\n";
 532                 }
 533                 if ( $row->rev_deleted & Revision::DELETED_COMMENT ) {
 534                         $out .= "      " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 535                 } elseif ( $row->rev_comment != '' ) {
 536                         $out .= "      " . Xml::elementClean( 'comment', null, strval( $row->rev_comment ) ) . "\n";
 537                 }
 538
 539                 $text = '';
 540                 if ( $row->rev_deleted & Revision::DELETED_TEXT ) {
 541                         $out .= "      " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 542                 } elseif ( isset( $row->old_text ) ) {
 543                         // Raw text from the database may have invalid chars
 544                         $text = strval( Revision::getRevisionText( $row ) );
 545                         $out .= "      " . Xml::elementClean( 'text',
 546                                 array( 'xml:space' => 'preserve', 'bytes' => intval( $row->rev_len ) ),
 547                                 strval( $text ) ) . "\n";
 548                 } else {
 549                         // Stub output
 550                         $out .= "      " . Xml::element( 'text',
 551                                 array( 'id' => $row->rev_text_id, 'bytes' => intval( $row->rev_len ) ),
 552                                 "" ) . "\n";
 553                 }
 554
 555                 wfRunHooks( 'XmlDumpWriterWriteRevision', array( &$this, &$out, $row, $text ) );
 556
 557                 $out .= "    </revision>\n";
 558
 559                 wfProfileOut( __METHOD__ );
 560                 return $out;
 561         }
 562
 563         /**
 564          * Dumps a <logitem> section on the output stream, with
 565          * data filled in from the given database row.
 566          *
 567          * @param $row object
 568          * @return string
 569          * @access private
 570          */
 571         function writeLogItem( $row ) {
 572                 wfProfileIn( __METHOD__ );
 573
 574                 $out  = "    <logitem>\n";
 575                 $out .= "      " . Xml::element( 'id', null, strval( $row->log_id ) ) . "\n";
 576
 577                 $out .= $this->writeTimestamp( $row->log_timestamp );
 578
 579                 if ( $row->log_deleted & LogPage::DELETED_USER ) {
 580                         $out .= "      " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 581                 } else {
 582                         $out .= $this->writeContributor( $row->log_user, $row->user_name );
 583                 }
 584
 585                 if ( $row->log_deleted & LogPage::DELETED_COMMENT ) {
 586                         $out .= "      " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 587                 } elseif ( $row->log_comment != '' ) {
 588                         $out .= "      " . Xml::elementClean( 'comment', null, strval( $row->log_comment ) ) . "\n";
 589                 }
 590
 591                 $out .= "      " . Xml::element( 'type', null, strval( $row->log_type ) ) . "\n";
 592                 $out .= "      " . Xml::element( 'action', null, strval( $row->log_action ) ) . "\n";
 593
 594                 if ( $row->log_deleted & LogPage::DELETED_ACTION ) {
 595                         $out .= "      " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 596                 } else {
 597                         $title = Title::makeTitle( $row->log_namespace, $row->log_title );
 598                         $out .= "      " . Xml::elementClean( 'logtitle', null, self::canonicalTitle( $title ) ) . "\n";
 599                         $out .= "      " . Xml::elementClean( 'params',
 600                                 array( 'xml:space' => 'preserve' ),
 601                                 strval( $row->log_params ) ) . "\n";
 602                 }
 603
 604                 $out .= "    </logitem>\n";
 605
 606                 wfProfileOut( __METHOD__ );
 607                 return $out;
 608         }
 609
 610         function writeTimestamp( $timestamp ) {
 611                 $ts = wfTimestamp( TS_ISO_8601, $timestamp );
 612                 return "      " . Xml::element( 'timestamp', null, $ts ) . "\n";
 613         }
 614
 615         function writeContributor( $id, $text ) {
 616                 $out = "      <contributor>\n";
 617                 if ( $id || !IP::isValid( $text ) ) {
 618                         $out .= "        " . Xml::elementClean( 'username', null, strval( $text ) ) . "\n";
 619                         $out .= "        " . Xml::element( 'id', null, strval( $id ) ) . "\n";
 620                 } else {
 621                         $out .= "        " . Xml::elementClean( 'ip', null, strval( $text ) ) . "\n";
 622                 }
 623                 $out .= "      </contributor>\n";
 624                 return $out;
 625         }
 626
 627         /**
 628          * Warning! This data is potentially inconsistent. :(
 629          */
 630         function writeUploads( $row, $dumpContents = false ) {
 631                 if ( $row->page_namespace == NS_IMAGE ) {
 632                         $img = wfLocalFile( $row->page_title );
 633                         if ( $img && $img->exists() ) {
 634                                 $out = '';
 635                                 foreach ( array_reverse( $img->getHistory() ) as $ver ) {
 636                                         $out .= $this->writeUpload( $ver, $dumpContents );
 637                                 }
 638                                 $out .= $this->writeUpload( $img, $dumpContents );
 639                                 return $out;
 640                         }
 641                 }
 642                 return '';
 643         }
 644
 645         /**
 646          * @param $file File
 647          * @param $dumpContents bool
 648          * @return string
 649          */
 650         function writeUpload( $file, $dumpContents = false ) {
 651                 if ( $file->isOld() ) {
 652                         $archiveName = "      " .
 653                                 Xml::element( 'archivename', null, $file->getArchiveName() ) . "\n";
 654                 } else {
 655                         $archiveName = '';
 656                 }
 657                 if ( $dumpContents ) {
 658                         # Dump file as base64
 659                         # Uses only XML-safe characters, so does not need escaping
 660                         $contents = '      <contents encoding="base64">' .
 661                                 chunk_split( base64_encode( file_get_contents( $file->getPath() ) ) ) .
 662                                 "      </contents>\n";
 663                 } else {
 664                         $contents = '';
 665                 }
 666                 return "    <upload>\n" .
 667                         $this->writeTimestamp( $file->getTimestamp() ) .
 668                         $this->writeContributor( $file->getUser( 'id' ), $file->getUser( 'text' ) ) .
 669                         "      " . Xml::elementClean( 'comment', null, $file->getDescription() ) . "\n" .
 670                         "      " . Xml::element( 'filename', null, $file->getName() ) . "\n" .
 671                         $archiveName .
 672                         "      " . Xml::element( 'src', null, $file->getCanonicalUrl() ) . "\n" .
 673                         "      " . Xml::element( 'size', null, $file->getSize() ) . "\n" .
 674                         "      " . Xml::element( 'sha1base36', null, $file->getSha1() ) . "\n" .
 675                         "      " . Xml::element( 'rel', null, $file->getRel() ) . "\n" .
 676                         $contents .
 677                         "    </upload>\n";
 678         }
 679
 680         /**
 681          * Return prefixed text form of title, but using the content language's
 682          * canonical namespace. This skips any special-casing such as gendered
 683          * user namespaces -- which while useful, are not yet listed in the
 684          * XML <siteinfo> data so are unsafe in export.
 685          *
 686          * @param Title $title
 687          * @return string
 688          */
 689         public static function canonicalTitle( Title $title ) {
 690                 if ( $title->getInterwiki() ) {
 691                         return $title->getPrefixedText();
 692                 }
 693
 694                 global $wgContLang;
 695                 $prefix = str_replace( '_', ' ', $wgContLang->getNsText( $title->getNamespace() ) );
 696
 697                 if ($prefix !== '') {
 698                         $prefix .= ':';
 699                 }
 700
 701                 return $prefix . $title->getText();
 702         }
 703 }
 704
 705
 706 /**
 707  * Base class for output stream; prints to stdout or buffer or whereever.
 708  * @ingroup Dump
 709  */
 710 class DumpOutput {
 711         function writeOpenStream( $string ) {
 712                 $this->write( $string );
 713         }
 714
 715         function writeCloseStream( $string ) {
 716                 $this->write( $string );
 717         }
 718
 719         function writeOpenPage( $page, $string ) {
 720                 $this->write( $string );
 721         }
 722
 723         function writeClosePage( $string ) {
 724                 $this->write( $string );
 725         }
 726
 727         function writeRevision( $rev, $string ) {
 728                 $this->write( $string );
 729         }
 730
 731         function writeLogItem( $rev, $string ) {
 732                 $this->write( $string );
 733         }
 734
 735         /**
 736          * Override to write to a different stream type.
 737          * @return bool
 738          */
 739         function write( $string ) {
 740                 print $string;
 741         }
 742
 743         /**
 744          * Close the old file, move it to a specified name,
 745          * and reopen new file with the old name. Use this
 746          * for writing out a file in multiple pieces
 747          * at specified checkpoints (e.g. every n hours).
 748          * @param $newname mixed File name. May be a string or an array with one element
 749          */
 750         function closeRenameAndReopen( $newname ) {
 751                 return;
 752         }
 753
 754         /**
 755          * Close the old file, and move it to a specified name.
 756          * Use this for the last piece of a file written out
 757          * at specified checkpoints (e.g. every n hours).
 758          * @param $newname mixed File name. May be a string or an array with one element
 759          * @param $open bool If true, a new file with the old filename will be opened again for writing (default: false)
 760          */
 761         function closeAndRename( $newname, $open = false ) {
 762                 return;
 763         }
 764
 765         /**
 766          * Returns the name of the file or files which are
 767          * being written to, if there are any.
 768          */
 769         function getFilenames() {
 770                 return NULL;
 771         }
 772 }
 773
 774 /**
 775  * Stream outputter to send data to a file.
 776  * @ingroup Dump
 777  */
 778 class DumpFileOutput extends DumpOutput {
 779         protected $handle, $filename;
 780
 781         function __construct( $file ) {
 782                 $this->handle = fopen( $file, "wt" );
 783                 $this->filename = $file;
 784         }
 785
 786         function write( $string ) {
 787                 fputs( $this->handle, $string );
 788         }
 789
 790         function closeRenameAndReopen( $newname ) {
 791                 $this->closeAndRename( $newname, true );
 792         }
 793
 794         function renameOrException( $newname ) {
 795                         if (! rename( $this->filename, $newname ) ) {
 796                                 throw new MWException( __METHOD__ . ": rename of file {$this->filename} to $newname failed\n" );
 797                         }
 798         }
 799
 800         function checkRenameArgCount( $newname ) {
 801                 if ( is_array( $newname ) ) {
 802                         if ( count( $newname ) > 1 ) {
 803                                 throw new MWException( __METHOD__ . ": passed multiple arguments for rename of single file\n" );
 804                         } else {
 805                                 $newname = $newname[0];
 806                         }
 807                 }
 808                 return $newname;
 809         }
 810
 811         function closeAndRename( $newname, $open = false ) {
 812                 $newname = $this->checkRenameArgCount( $newname );
 813                 if ( $newname ) {
 814                         fclose( $this->handle );
 815                         $this->renameOrException( $newname );
 816                         if ( $open ) {
 817                                 $this->handle = fopen( $this->filename, "wt" );
 818                         }
 819                 }
 820         }
 821
 822         function getFilenames() {
 823                 return $this->filename;
 824         }
 825 }
 826
 827 /**
 828  * Stream outputter to send data to a file via some filter program.
 829  * Even if compression is available in a library, using a separate
 830  * program can allow us to make use of a multi-processor system.
 831  * @ingroup Dump
 832  */
 833 class DumpPipeOutput extends DumpFileOutput {
 834         protected $command, $filename;
 835
 836         function __construct( $command, $file = null ) {
 837                 if ( !is_null( $file ) ) {
 838                         $command .=  " > " . wfEscapeShellArg( $file );
 839                 }
 840
 841                 $this->startCommand( $command );
 842                 $this->command = $command;
 843                 $this->filename = $file;
 844         }
 845
 846         function startCommand( $command ) {
 847                 $spec = array(
 848                         0 => array( "pipe", "r" ),
 849                 );
 850                 $pipes = array();
 851                 $this->procOpenResource = proc_open( $command, $spec, $pipes );
 852                 $this->handle = $pipes[0];
 853         }
 854
 855         function closeRenameAndReopen( $newname ) {
 856                 $this->closeAndRename( $newname, true );
 857         }
 858
 859         function closeAndRename( $newname, $open = false ) {
 860                 $newname = $this->checkRenameArgCount( $newname );
 861                 if ( $newname ) {
 862                         fclose( $this->handle );
 863                         proc_close( $this->procOpenResource );
 864                         $this->renameOrException( $newname );
 865                         if ( $open ) {
 866                                 $command = $this->command;
 867                                 $command .=  " > " . wfEscapeShellArg( $this->filename );
 868                                 $this->startCommand( $command );
 869                         }
 870                 }
 871         }
 872
 873 }
 874
 875 /**
 876  * Sends dump output via the gzip compressor.
 877  * @ingroup Dump
 878  */
 879 class DumpGZipOutput extends DumpPipeOutput {
 880         function __construct( $file ) {
 881                 parent::__construct( "gzip", $file );
 882         }
 883 }
 884
 885 /**
 886  * Sends dump output via the bgzip2 compressor.
 887  * @ingroup Dump
 888  */
 889 class DumpBZip2Output extends DumpPipeOutput {
 890         function __construct( $file ) {
 891                 parent::__construct( "bzip2", $file );
 892         }
 893 }
 894
 895 /**
 896  * Sends dump output via the p7zip compressor.
 897  * @ingroup Dump
 898  */
 899 class Dump7ZipOutput extends DumpPipeOutput {
 900         function __construct( $file ) {
 901                 $command = $this->setup7zCommand( $file );
 902                 parent::__construct( $command );
 903                 $this->filename = $file;
 904         }
 905
 906         function setup7zCommand( $file ) {
 907                 $command = "7za a -bd -si " . wfEscapeShellArg( $file );
 908                 // Suppress annoying useless crap from p7zip
 909                 // Unfortunately this could suppress real error messages too
 910                 $command .= ' >' . wfGetNull() . ' 2>&1';
 911                 return( $command );
 912         }
 913
 914         function closeAndRename( $newname, $open = false ) {
 915                 $newname = $this->checkRenameArgCount( $newname );
 916                 if ( $newname ) {
 917                         fclose( $this->handle );
 918                         proc_close( $this->procOpenResource );
 919                         $this->renameOrException( $newname );
 920                         if ( $open ) {
 921                                 $command = $this->setup7zCommand( $this->filename );
 922                                 $this->startCommand( $command );
 923                         }
 924                 }
 925         }
 926 }
 927
 928
 929
 930 /**
 931  * Dump output filter class.
 932  * This just does output filtering and streaming; XML formatting is done
 933  * higher up, so be careful in what you do.
 934  * @ingroup Dump
 935  */
 936 class DumpFilter {
 937         function __construct( &$sink ) {
 938                 $this->sink =& $sink;
 939         }
 940
 941         function writeOpenStream( $string ) {
 942                 $this->sink->writeOpenStream( $string );
 943         }
 944
 945         function writeCloseStream( $string ) {
 946                 $this->sink->writeCloseStream( $string );
 947         }
 948
 949         function writeOpenPage( $page, $string ) {
 950                 $this->sendingThisPage = $this->pass( $page, $string );
 951                 if ( $this->sendingThisPage ) {
 952                         $this->sink->writeOpenPage( $page, $string );
 953                 }
 954         }
 955
 956         function writeClosePage( $string ) {
 957                 if ( $this->sendingThisPage ) {
 958                         $this->sink->writeClosePage( $string );
 959                         $this->sendingThisPage = false;
 960                 }
 961         }
 962
 963         function writeRevision( $rev, $string ) {
 964                 if ( $this->sendingThisPage ) {
 965                         $this->sink->writeRevision( $rev, $string );
 966                 }
 967         }
 968
 969         function writeLogItem( $rev, $string ) {
 970                 $this->sink->writeRevision( $rev, $string );
 971         }
 972
 973         function closeRenameAndReopen( $newname ) {
 974                 $this->sink->closeRenameAndReopen( $newname );
 975         }
 976
 977         function closeAndRename( $newname, $open = false ) {
 978                 $this->sink->closeAndRename( $newname, $open );
 979         }
 980
 981         function getFilenames() {
 982                 return $this->sink->getFilenames();
 983         }
 984
 985         /**
 986          * Override for page-based filter types.
 987          * @return bool
 988          */
 989         function pass( $page ) {
 990                 return true;
 991         }
 992 }
 993
 994 /**
 995  * Simple dump output filter to exclude all talk pages.
 996  * @ingroup Dump
 997  */
 998 class DumpNotalkFilter extends DumpFilter {
 999         function pass( $page ) {
1000                 return !MWNamespace::isTalk( $page->page_namespace );
1001         }
1002 }
1003
1004 /**
1005  * Dump output filter to include or exclude pages in a given set of namespaces.
1006  * @ingroup Dump
1007  */
1008 class DumpNamespaceFilter extends DumpFilter {
1009         var $invert = false;
1010         var $namespaces = array();
1011
1012         function __construct( &$sink, $param ) {
1013                 parent::__construct( $sink );
1014
1015                 $constants = array(
1016                         "NS_MAIN"           => NS_MAIN,
1017                         "NS_TALK"           => NS_TALK,
1018                         "NS_USER"           => NS_USER,
1019                         "NS_USER_TALK"      => NS_USER_TALK,
1020                         "NS_PROJECT"        => NS_PROJECT,
1021                         "NS_PROJECT_TALK"   => NS_PROJECT_TALK,
1022                         "NS_FILE"           => NS_FILE,
1023                         "NS_FILE_TALK"      => NS_FILE_TALK,
1024                         "NS_IMAGE"          => NS_IMAGE,  // NS_IMAGE is an alias for NS_FILE
1025                         "NS_IMAGE_TALK"     => NS_IMAGE_TALK,
1026                         "NS_MEDIAWIKI"      => NS_MEDIAWIKI,
1027                         "NS_MEDIAWIKI_TALK" => NS_MEDIAWIKI_TALK,
1028                         "NS_TEMPLATE"       => NS_TEMPLATE,
1029                         "NS_TEMPLATE_TALK"  => NS_TEMPLATE_TALK,
1030                         "NS_HELP"           => NS_HELP,
1031                         "NS_HELP_TALK"      => NS_HELP_TALK,
1032                         "NS_CATEGORY"       => NS_CATEGORY,
1033                         "NS_CATEGORY_TALK"  => NS_CATEGORY_TALK );
1034
1035                 if ( $param { 0 } == '!' ) {
1036                         $this->invert = true;
1037                         $param = substr( $param, 1 );
1038                 }
1039
1040                 foreach ( explode( ',', $param ) as $key ) {
1041                         $key = trim( $key );
1042                         if ( isset( $constants[$key] ) ) {
1043                                 $ns = $constants[$key];
1044                                 $this->namespaces[$ns] = true;
1045                         } elseif ( is_numeric( $key ) ) {
1046                                 $ns = intval( $key );
1047                                 $this->namespaces[$ns] = true;
1048                         } else {
1049                                 throw new MWException( "Unrecognized namespace key '$key'\n" );
1050                         }
1051                 }
1052         }
1053
1054         function pass( $page ) {
1055                 $match = isset( $this->namespaces[$page->page_namespace] );
1056                 return $this->invert xor $match;
1057         }
1058 }
1059
1060
1061 /**
1062  * Dump output filter to include only the last revision in each page sequence.
1063  * @ingroup Dump
1064  */
1065 class DumpLatestFilter extends DumpFilter {
1066         var $page, $pageString, $rev, $revString;
1067
1068         function writeOpenPage( $page, $string ) {
1069                 $this->page = $page;
1070                 $this->pageString = $string;
1071         }
1072
1073         function writeClosePage( $string ) {
1074                 if ( $this->rev ) {
1075                         $this->sink->writeOpenPage( $this->page, $this->pageString );
1076                         $this->sink->writeRevision( $this->rev, $this->revString );
1077                         $this->sink->writeClosePage( $string );
1078                 }
1079                 $this->rev = null;
1080                 $this->revString = null;
1081                 $this->page = null;
1082                 $this->pageString = null;
1083         }
1084
1085         function writeRevision( $rev, $string ) {
1086                 if ( $rev->rev_id == $this->page->page_latest ) {
1087                         $this->rev = $rev;
1088                         $this->revString = $string;
1089                 }
1090         }
1091 }
1092
1093 /**
1094  * Base class for output stream; prints to stdout or buffer or whereever.
1095  * @ingroup Dump
1096  */
1097 class DumpMultiWriter {
1098         function __construct( $sinks ) {
1099                 $this->sinks = $sinks;
1100                 $this->count = count( $sinks );
1101         }
1102
1103         function writeOpenStream( $string ) {
1104                 for ( $i = 0; $i < $this->count; $i++ ) {
1105                         $this->sinks[$i]->writeOpenStream( $string );
1106                 }
1107         }
1108
1109         function writeCloseStream( $string ) {
1110                 for ( $i = 0; $i < $this->count; $i++ ) {
1111                         $this->sinks[$i]->writeCloseStream( $string );
1112                 }
1113         }
1114
1115         function writeOpenPage( $page, $string ) {
1116                 for ( $i = 0; $i < $this->count; $i++ ) {
1117                         $this->sinks[$i]->writeOpenPage( $page, $string );
1118                 }
1119         }
1120
1121         function writeClosePage( $string ) {
1122                 for ( $i = 0; $i < $this->count; $i++ ) {
1123                         $this->sinks[$i]->writeClosePage( $string );
1124                 }
1125         }
1126
1127         function writeRevision( $rev, $string ) {
1128                 for ( $i = 0; $i < $this->count; $i++ ) {
1129                         $this->sinks[$i]->writeRevision( $rev, $string );
1130                 }
1131         }
1132
1133         function closeRenameAndReopen( $newnames ) {
1134                 $this->closeAndRename( $newnames, true );
1135         }
1136
1137         function closeAndRename( $newnames, $open = false ) {
1138                 for ( $i = 0; $i < $this->count; $i++ ) {
1139                         $this->sinks[$i]->closeAndRename( $newnames[$i], $open );
1140                 }
1141         }
1142
1143         function getFilenames() {
1144                 $filenames = array();
1145                 for ( $i = 0; $i < $this->count; $i++ ) {
1146                         $filenames[] =  $this->sinks[$i]->getFilenames();
1147                 }
1148                 return $filenames;
1149         }
1150
1151 }
1152
1153 function xmlsafe( $string ) {
1154         wfProfileIn( __FUNCTION__ );
1155
1156         /**
1157          * The page may contain old data which has not been properly normalized.
1158          * Invalid UTF-8 sequences or forbidden control characters will make our
1159          * XML output invalid, so be sure to strip them out.
1160          */
1161         $string = UtfNormal::cleanUp( $string );
1162
1163         $string = htmlspecialchars( $string );
1164         wfProfileOut( __FUNCTION__ );
1165         return $string;
1166 }