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
 312                         if ( $this->buffer == WikiExporter::STREAM ) {
 313                                 $this->db->bufferResults( $prev );
 314                         }
 315                 }
 316                 wfProfileOut( __METHOD__ );
 317         }
 318
 319         /**
 320          * Runs through a query result set dumping page and revision records.
 321          * The result set should be sorted/grouped by page to avoid duplicate
 322          * page records in the output.
 323          *
 324          * The result set will be freed once complete. Should be safe for
 325          * streaming (non-buffered) queries, as long as it was made on a
 326          * separate database connection not managed by LoadBalancer; some
 327          * blob storage types will make queries to pull source data.
 328          *
 329          * @param $resultset ResultWrapper
 330          */
 331         protected function outputPageStream( $resultset ) {
 332                 $last = null;
 333                 foreach ( $resultset as $row ) {
 334                         if ( is_null( $last ) ||
 335                                 $last->page_namespace != $row->page_namespace ||
 336                                 $last->page_title     != $row->page_title ) {
 337                                 if ( isset( $last ) ) {
 338                                         $output = '';
 339                                         if ( $this->dumpUploads ) {
 340                                                 $output .= $this->writer->writeUploads( $last, $this->dumpUploadFileContents );
 341                                         }
 342                                         $output .= $this->writer->closePage();
 343                                         $this->sink->writeClosePage( $output );
 344                                 }
 345                                 $output = $this->writer->openPage( $row );
 346                                 $this->sink->writeOpenPage( $row, $output );
 347                                 $last = $row;
 348                         }
 349                         $output = $this->writer->writeRevision( $row );
 350                         $this->sink->writeRevision( $row, $output );
 351                 }
 352                 if ( isset( $last ) ) {
 353                         $output = '';
 354                         if ( $this->dumpUploads ) {
 355                                 $output .= $this->writer->writeUploads( $last, $this->dumpUploadFileContents );
 356                         }
 357                         $output .= $this->author_list;
 358                         $output .= $this->writer->closePage();
 359                         $this->sink->writeClosePage( $output );
 360                 }
 361         }
 362
 363         protected function outputLogStream( $resultset ) {
 364                 foreach ( $resultset as $row ) {
 365                         $output = $this->writer->writeLogItem( $row );
 366                         $this->sink->writeLogItem( $row, $output );
 367                 }
 368         }
 369 }
 370
 371 /**
 372  * @ingroup Dump
 373  */
 374 class XmlDumpWriter {
 375         /**
 376          * Returns the export schema version.
 377          * @return string
 378          */
 379         function schemaVersion() {
 380                 return "0.6";
 381         }
 382
 383         /**
 384          * Opens the XML output stream's root <mediawiki> element.
 385          * This does not include an xml directive, so is safe to include
 386          * as a subelement in a larger XML stream. Namespace and XML Schema
 387          * references are included.
 388          *
 389          * Output will be encoded in UTF-8.
 390          *
 391          * @return string
 392          */
 393         function openStream() {
 394                 global $wgLanguageCode;
 395                 $ver = $this->schemaVersion();
 396                 return Xml::element( 'mediawiki', array(
 397                         'xmlns'              => "http://www.mediawiki.org/xml/export-$ver/",
 398                         'xmlns:xsi'          => "http://www.w3.org/2001/XMLSchema-instance",
 399                         'xsi:schemaLocation' => "http://www.mediawiki.org/xml/export-$ver/ " .
 400                                                 "http://www.mediawiki.org/xml/export-$ver.xsd",
 401                         'version'            => $ver,
 402                         'xml:lang'           => $wgLanguageCode ),
 403                         null ) .
 404                         "\n" .
 405                         $this->siteInfo();
 406         }
 407
 408         function siteInfo() {
 409                 $info = array(
 410                         $this->sitename(),
 411                         $this->homelink(),
 412                         $this->generator(),
 413                         $this->caseSetting(),
 414                         $this->namespaces() );
 415                 return "  <siteinfo>\n    " .
 416                         implode( "\n    ", $info ) .
 417                         "\n  </siteinfo>\n";
 418         }
 419
 420         function sitename() {
 421                 global $wgSitename;
 422                 return Xml::element( 'sitename', array(), $wgSitename );
 423         }
 424
 425         function generator() {
 426                 global $wgVersion;
 427                 return Xml::element( 'generator', array(), "MediaWiki $wgVersion" );
 428         }
 429
 430         function homelink() {
 431                 return Xml::element( 'base', array(), Title::newMainPage()->getCanonicalUrl() );
 432         }
 433
 434         function caseSetting() {
 435                 global $wgCapitalLinks;
 436                 // "case-insensitive" option is reserved for future
 437                 $sensitivity = $wgCapitalLinks ? 'first-letter' : 'case-sensitive';
 438                 return Xml::element( 'case', array(), $sensitivity );
 439         }
 440
 441         function namespaces() {
 442                 global $wgContLang;
 443                 $spaces = "<namespaces>\n";
 444                 foreach ( $wgContLang->getFormattedNamespaces() as $ns => $title ) {
 445                         $spaces .= '      ' .
 446                                 Xml::element( 'namespace',
 447                                         array(  'key' => $ns,
 448                                                         'case' => MWNamespace::isCapitalized( $ns ) ? 'first-letter' : 'case-sensitive',
 449                                         ), $title ) . "\n";
 450                 }
 451                 $spaces .= "    </namespaces>";
 452                 return $spaces;
 453         }
 454
 455         /**
 456          * Closes the output stream with the closing root element.
 457          * Call when finished dumping things.
 458          *
 459          * @return string
 460          */
 461         function closeStream() {
 462                 return "</mediawiki>\n";
 463         }
 464
 465         /**
 466          * Opens a <page> section on the output stream, with data
 467          * from the given database row.
 468          *
 469          * @param $row object
 470          * @return string
 471          * @access private
 472          */
 473         function openPage( $row ) {
 474                 $out = "  <page>\n";
 475                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 476                 $out .= '    ' . Xml::elementClean( 'title', array(), self::canonicalTitle( $title ) ) . "\n";
 477                 $out .= '    ' . Xml::element( 'ns', array(), strval( $row->page_namespace) ) . "\n";
 478                 $out .= '    ' . Xml::element( 'id', array(), strval( $row->page_id ) ) . "\n";
 479                 if ( $row->page_is_redirect ) {
 480                         $page = WikiPage::factory( $title );
 481                         $redirect = $page->getRedirectTarget();
 482                         if ( $redirect instanceOf Title && $redirect->isValidRedirectTarget() ) {
 483                                 $out .= '    ' . Xml::element( 'redirect', array( 'title' => self::canonicalTitle( $redirect ) ) ) . "\n";
 484                         }
 485                 }
 486
 487                 if ( $row->page_restrictions != '' ) {
 488                         $out .= '    ' . Xml::element( 'restrictions', array(),
 489                                 strval( $row->page_restrictions ) ) . "\n";
 490                 }
 491
 492                 wfRunHooks( 'XmlDumpWriterOpenPage', array( $this, &$out, $row, $title ) );
 493
 494                 return $out;
 495         }
 496
 497         /**
 498          * Closes a <page> section on the output stream.
 499          *
 500          * @access private
 501          * @return string
 502          */
 503         function closePage() {
 504                 return "  </page>\n";
 505         }
 506
 507         /**
 508          * Dumps a <revision> section on the output stream, with
 509          * data filled in from the given database row.
 510          *
 511          * @param $row object
 512          * @return string
 513          * @access private
 514          */
 515         function writeRevision( $row ) {
 516                 wfProfileIn( __METHOD__ );
 517
 518                 $out  = "    <revision>\n";
 519                 $out .= "      " . Xml::element( 'id', null, strval( $row->rev_id ) ) . "\n";
 520
 521                 $out .= $this->writeTimestamp( $row->rev_timestamp );
 522
 523                 if ( $row->rev_deleted & Revision::DELETED_USER ) {
 524                         $out .= "      " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 525                 } else {
 526                         $out .= $this->writeContributor( $row->rev_user, $row->rev_user_text );
 527                 }
 528
 529                 if ( $row->rev_minor_edit ) {
 530                         $out .=  "      <minor/>\n";
 531                 }
 532                 if ( $row->rev_deleted & Revision::DELETED_COMMENT ) {
 533                         $out .= "      " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 534                 } elseif ( $row->rev_comment != '' ) {
 535                         $out .= "      " . Xml::elementClean( 'comment', null, strval( $row->rev_comment ) ) . "\n";
 536                 }
 537
 538                 $text = '';
 539                 if ( $row->rev_deleted & Revision::DELETED_TEXT ) {
 540                         $out .= "      " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 541                 } elseif ( isset( $row->old_text ) ) {
 542                         // Raw text from the database may have invalid chars
 543                         $text = strval( Revision::getRevisionText( $row ) );
 544                         $out .= "      " . Xml::elementClean( 'text',
 545                                 array( 'xml:space' => 'preserve', 'bytes' => intval( $row->rev_len ) ),
 546                                 strval( $text ) ) . "\n";
 547                 } else {
 548                         // Stub output
 549                         $out .= "      " . Xml::element( 'text',
 550                                 array( 'id' => $row->rev_text_id, 'bytes' => intval( $row->rev_len ) ),
 551                                 "" ) . "\n";
 552                 }
 553
 554                 if ( $row->rev_sha1 && !( $row->rev_deleted & Revision::DELETED_TEXT ) ) {
 555                         $out .= "      " . Xml::element('sha1', null, strval( $row->rev_sha1 ) ) . "\n";
 556                 } else {
 557                         $out .= "      <sha1/>\n";
 558                 }
 559
 560                 wfRunHooks( 'XmlDumpWriterWriteRevision', array( &$this, &$out, $row, $text ) );
 561
 562                 $out .= "    </revision>\n";
 563
 564                 wfProfileOut( __METHOD__ );
 565                 return $out;
 566         }
 567
 568         /**
 569          * Dumps a <logitem> section on the output stream, with
 570          * data filled in from the given database row.
 571          *
 572          * @param $row object
 573          * @return string
 574          * @access private
 575          */
 576         function writeLogItem( $row ) {
 577                 wfProfileIn( __METHOD__ );
 578
 579                 $out  = "    <logitem>\n";
 580                 $out .= "      " . Xml::element( 'id', null, strval( $row->log_id ) ) . "\n";
 581
 582                 $out .= $this->writeTimestamp( $row->log_timestamp );
 583
 584                 if ( $row->log_deleted & LogPage::DELETED_USER ) {
 585                         $out .= "      " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 586                 } else {
 587                         $out .= $this->writeContributor( $row->log_user, $row->user_name );
 588                 }
 589
 590                 if ( $row->log_deleted & LogPage::DELETED_COMMENT ) {
 591                         $out .= "      " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 592                 } elseif ( $row->log_comment != '' ) {
 593                         $out .= "      " . Xml::elementClean( 'comment', null, strval( $row->log_comment ) ) . "\n";
 594                 }
 595
 596                 $out .= "      " . Xml::element( 'type', null, strval( $row->log_type ) ) . "\n";
 597                 $out .= "      " . Xml::element( 'action', null, strval( $row->log_action ) ) . "\n";
 598
 599                 if ( $row->log_deleted & LogPage::DELETED_ACTION ) {
 600                         $out .= "      " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 601                 } else {
 602                         $title = Title::makeTitle( $row->log_namespace, $row->log_title );
 603                         $out .= "      " . Xml::elementClean( 'logtitle', null, self::canonicalTitle( $title ) ) . "\n";
 604                         $out .= "      " . Xml::elementClean( 'params',
 605                                 array( 'xml:space' => 'preserve' ),
 606                                 strval( $row->log_params ) ) . "\n";
 607                 }
 608
 609                 $out .= "    </logitem>\n";
 610
 611                 wfProfileOut( __METHOD__ );
 612                 return $out;
 613         }
 614
 615         function writeTimestamp( $timestamp ) {
 616                 $ts = wfTimestamp( TS_ISO_8601, $timestamp );
 617                 return "      " . Xml::element( 'timestamp', null, $ts ) . "\n";
 618         }
 619
 620         function writeContributor( $id, $text ) {
 621                 $out = "      <contributor>\n";
 622                 if ( $id || !IP::isValid( $text ) ) {
 623                         $out .= "        " . Xml::elementClean( 'username', null, strval( $text ) ) . "\n";
 624                         $out .= "        " . Xml::element( 'id', null, strval( $id ) ) . "\n";
 625                 } else {
 626                         $out .= "        " . Xml::elementClean( 'ip', null, strval( $text ) ) . "\n";
 627                 }
 628                 $out .= "      </contributor>\n";
 629                 return $out;
 630         }
 631
 632         /**
 633          * Warning! This data is potentially inconsistent. :(
 634          * @return string
 635          */
 636         function writeUploads( $row, $dumpContents = false ) {
 637                 if ( $row->page_namespace == NS_IMAGE ) {
 638                         $img = wfLocalFile( $row->page_title );
 639                         if ( $img && $img->exists() ) {
 640                                 $out = '';
 641                                 foreach ( array_reverse( $img->getHistory() ) as $ver ) {
 642                                         $out .= $this->writeUpload( $ver, $dumpContents );
 643                                 }
 644                                 $out .= $this->writeUpload( $img, $dumpContents );
 645                                 return $out;
 646                         }
 647                 }
 648                 return '';
 649         }
 650
 651         /**
 652          * @param $file File
 653          * @param $dumpContents bool
 654          * @return string
 655          */
 656         function writeUpload( $file, $dumpContents = false ) {
 657                 if ( $file->isOld() ) {
 658                         $archiveName = "      " .
 659                                 Xml::element( 'archivename', null, $file->getArchiveName() ) . "\n";
 660                 } else {
 661                         $archiveName = '';
 662                 }
 663                 if ( $dumpContents ) {
 664                         # Dump file as base64
 665                         # Uses only XML-safe characters, so does not need escaping
 666                         $contents = '      <contents encoding="base64">' .
 667                                 chunk_split( base64_encode( file_get_contents( $file->getPath() ) ) ) .
 668                                 "      </contents>\n";
 669                 } else {
 670                         $contents = '';
 671                 }
 672                 return "    <upload>\n" .
 673                         $this->writeTimestamp( $file->getTimestamp() ) .
 674                         $this->writeContributor( $file->getUser( 'id' ), $file->getUser( 'text' ) ) .
 675                         "      " . Xml::elementClean( 'comment', null, $file->getDescription() ) . "\n" .
 676                         "      " . Xml::element( 'filename', null, $file->getName() ) . "\n" .
 677                         $archiveName .
 678                         "      " . Xml::element( 'src', null, $file->getCanonicalUrl() ) . "\n" .
 679                         "      " . Xml::element( 'size', null, $file->getSize() ) . "\n" .
 680                         "      " . Xml::element( 'sha1base36', null, $file->getSha1() ) . "\n" .
 681                         "      " . Xml::element( 'rel', null, $file->getRel() ) . "\n" .
 682                         $contents .
 683                         "    </upload>\n";
 684         }
 685
 686         /**
 687          * Return prefixed text form of title, but using the content language's
 688          * canonical namespace. This skips any special-casing such as gendered
 689          * user namespaces -- which while useful, are not yet listed in the
 690          * XML <siteinfo> data so are unsafe in export.
 691          *
 692          * @param Title $title
 693          * @return string
 694          * @since 1.18
 695          */
 696         public static function canonicalTitle( Title $title ) {
 697                 if ( $title->getInterwiki() ) {
 698                         return $title->getPrefixedText();
 699                 }
 700
 701                 global $wgContLang;
 702                 $prefix = str_replace( '_', ' ', $wgContLang->getNsText( $title->getNamespace() ) );
 703
 704                 if ( $prefix !== '' ) {
 705                         $prefix .= ':';
 706                 }
 707
 708                 return $prefix . $title->getText();
 709         }
 710 }
 711
 712
 713 /**
 714  * Base class for output stream; prints to stdout or buffer or whereever.
 715  * @ingroup Dump
 716  */
 717 class DumpOutput {
 718         function writeOpenStream( $string ) {
 719                 $this->write( $string );
 720         }
 721
 722         function writeCloseStream( $string ) {
 723                 $this->write( $string );
 724         }
 725
 726         function writeOpenPage( $page, $string ) {
 727                 $this->write( $string );
 728         }
 729
 730         function writeClosePage( $string ) {
 731                 $this->write( $string );
 732         }
 733
 734         function writeRevision( $rev, $string ) {
 735                 $this->write( $string );
 736         }
 737
 738         function writeLogItem( $rev, $string ) {
 739                 $this->write( $string );
 740         }
 741
 742         /**
 743          * Override to write to a different stream type.
 744          * @return bool
 745          */
 746         function write( $string ) {
 747                 print $string;
 748         }
 749
 750         /**
 751          * Close the old file, move it to a specified name,
 752          * and reopen new file with the old name. Use this
 753          * for writing out a file in multiple pieces
 754          * at specified checkpoints (e.g. every n hours).
 755          * @param $newname mixed File name. May be a string or an array with one element
 756          */
 757         function closeRenameAndReopen( $newname ) {
 758                 return;
 759         }
 760
 761         /**
 762          * Close the old file, and move it to a specified name.
 763          * Use this for the last piece of a file written out
 764          * at specified checkpoints (e.g. every n hours).
 765          * @param $newname mixed File name. May be a string or an array with one element
 766          * @param $open bool If true, a new file with the old filename will be opened again for writing (default: false)
 767          */
 768         function closeAndRename( $newname, $open = false ) {
 769                 return;
 770         }
 771
 772         /**
 773          * Returns the name of the file or files which are
 774          * being written to, if there are any.
 775          * @return null
 776          */
 777         function getFilenames() {
 778                 return NULL;
 779         }
 780 }
 781
 782 /**
 783  * Stream outputter to send data to a file.
 784  * @ingroup Dump
 785  */
 786 class DumpFileOutput extends DumpOutput {
 787         protected $handle = false, $filename;
 788
 789         function __construct( $file ) {
 790                 $this->handle = fopen( $file, "wt" );
 791                 $this->filename = $file;
 792         }
 793
 794         function writeCloseStream( $string ) {
 795                 parent::writeCloseStream( $string );
 796                 if ( $this->handle ) {
 797                         fclose( $this->handle );
 798                         $this->handle = false;
 799                 }
 800         }
 801
 802         function write( $string ) {
 803                 fputs( $this->handle, $string );
 804         }
 805
 806         function closeRenameAndReopen( $newname ) {
 807                 $this->closeAndRename( $newname, true );
 808         }
 809
 810         function renameOrException( $newname ) {
 811                         if (! rename( $this->filename, $newname ) ) {
 812                                 throw new MWException( __METHOD__ . ": rename of file {$this->filename} to $newname failed\n" );
 813                         }
 814         }
 815
 816         function checkRenameArgCount( $newname ) {
 817                 if ( is_array( $newname ) ) {
 818                         if ( count( $newname ) > 1 ) {
 819                                 throw new MWException( __METHOD__ . ": passed multiple arguments for rename of single file\n" );
 820                         } else {
 821                                 $newname = $newname[0];
 822                         }
 823                 }
 824                 return $newname;
 825         }
 826
 827         function closeAndRename( $newname, $open = false ) {
 828                 $newname = $this->checkRenameArgCount( $newname );
 829                 if ( $newname ) {
 830                         if ( $this->handle ) {
 831                                 fclose( $this->handle );
 832                                 $this->handle = false;
 833                         }
 834                         $this->renameOrException( $newname );
 835                         if ( $open ) {
 836                                 $this->handle = fopen( $this->filename, "wt" );
 837                         }
 838                 }
 839         }
 840
 841         function getFilenames() {
 842                 return $this->filename;
 843         }
 844 }
 845
 846 /**
 847  * Stream outputter to send data to a file via some filter program.
 848  * Even if compression is available in a library, using a separate
 849  * program can allow us to make use of a multi-processor system.
 850  * @ingroup Dump
 851  */
 852 class DumpPipeOutput extends DumpFileOutput {
 853         protected $command, $filename;
 854         private $procOpenResource = false;
 855
 856         function __construct( $command, $file = null ) {
 857                 if ( !is_null( $file ) ) {
 858                         $command .=  " > " . wfEscapeShellArg( $file );
 859                 }
 860
 861                 $this->startCommand( $command );
 862                 $this->command = $command;
 863                 $this->filename = $file;
 864         }
 865
 866         function writeCloseStream( $string ) {
 867                 parent::writeCloseStream( $string );
 868                 if ( $this->procOpenResource ) {
 869                         proc_close( $this->procOpenResource );
 870                         $this->procOpenResource = false;
 871                 }
 872         }
 873
 874         function startCommand( $command ) {
 875                 $spec = array(
 876                         0 => array( "pipe", "r" ),
 877                 );
 878                 $pipes = array();
 879                 $this->procOpenResource = proc_open( $command, $spec, $pipes );
 880                 $this->handle = $pipes[0];
 881         }
 882
 883         function closeRenameAndReopen( $newname ) {
 884                 $this->closeAndRename( $newname, true );
 885         }
 886
 887         function closeAndRename( $newname, $open = false ) {
 888                 $newname = $this->checkRenameArgCount( $newname );
 889                 if ( $newname ) {
 890                         if ( $this->handle ) {
 891                                 fclose( $this->handle );
 892                                 $this->handle = false;
 893                         }
 894                         if ( $this->procOpenResource ) {
 895                                 proc_close( $this->procOpenResource );
 896                                 $this->procOpenResource = false;
 897                         }
 898                         $this->renameOrException( $newname );
 899                         if ( $open ) {
 900                                 $command = $this->command;
 901                                 $command .=  " > " . wfEscapeShellArg( $this->filename );
 902                                 $this->startCommand( $command );
 903                         }
 904                 }
 905         }
 906
 907 }
 908
 909 /**
 910  * Sends dump output via the gzip compressor.
 911  * @ingroup Dump
 912  */
 913 class DumpGZipOutput extends DumpPipeOutput {
 914         function __construct( $file ) {
 915                 parent::__construct( "gzip", $file );
 916         }
 917 }
 918
 919 /**
 920  * Sends dump output via the bgzip2 compressor.
 921  * @ingroup Dump
 922  */
 923 class DumpBZip2Output extends DumpPipeOutput {
 924         function __construct( $file ) {
 925                 parent::__construct( "bzip2", $file );
 926         }
 927 }
 928
 929 /**
 930  * Sends dump output via the p7zip compressor.
 931  * @ingroup Dump
 932  */
 933 class Dump7ZipOutput extends DumpPipeOutput {
 934         function __construct( $file ) {
 935                 $command = $this->setup7zCommand( $file );
 936                 parent::__construct( $command );
 937                 $this->filename = $file;
 938         }
 939
 940         function setup7zCommand( $file ) {
 941                 $command = "7za a -bd -si " . wfEscapeShellArg( $file );
 942                 // Suppress annoying useless crap from p7zip
 943                 // Unfortunately this could suppress real error messages too
 944                 $command .= ' >' . wfGetNull() . ' 2>&1';
 945                 return( $command );
 946         }
 947
 948         function closeAndRename( $newname, $open = false ) {
 949                 $newname = $this->checkRenameArgCount( $newname );
 950                 if ( $newname ) {
 951                         fclose( $this->handle );
 952                         proc_close( $this->procOpenResource );
 953                         $this->renameOrException( $newname );
 954                         if ( $open ) {
 955                                 $command = $this->setup7zCommand( $this->filename );
 956                                 $this->startCommand( $command );
 957                         }
 958                 }
 959         }
 960 }
 961
 962
 963
 964 /**
 965  * Dump output filter class.
 966  * This just does output filtering and streaming; XML formatting is done
 967  * higher up, so be careful in what you do.
 968  * @ingroup Dump
 969  */
 970 class DumpFilter {
 971         function __construct( &$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, $open = false ) {
1012                 $this->sink->closeAndRename( $newname, $open );
1013         }
1014
1015         function getFilenames() {
1016                 return $this->sink->getFilenames();
1017         }
1018
1019         /**
1020          * Override for page-based filter types.
1021          * @return bool
1022          */
1023         function pass( $page ) {
1024                 return true;
1025         }
1026 }
1027
1028 /**
1029  * Simple dump output filter to exclude all talk pages.
1030  * @ingroup Dump
1031  */
1032 class DumpNotalkFilter extends DumpFilter {
1033         function pass( $page ) {
1034                 return !MWNamespace::isTalk( $page->page_namespace );
1035         }
1036 }
1037
1038 /**
1039  * Dump output filter to include or exclude pages in a given set of namespaces.
1040  * @ingroup Dump
1041  */
1042 class DumpNamespaceFilter extends DumpFilter {
1043         var $invert = false;
1044         var $namespaces = array();
1045
1046         function __construct( &$sink, $param ) {
1047                 parent::__construct( $sink );
1048
1049                 $constants = array(
1050                         "NS_MAIN"           => NS_MAIN,
1051                         "NS_TALK"           => NS_TALK,
1052                         "NS_USER"           => NS_USER,
1053                         "NS_USER_TALK"      => NS_USER_TALK,
1054                         "NS_PROJECT"        => NS_PROJECT,
1055                         "NS_PROJECT_TALK"   => NS_PROJECT_TALK,
1056                         "NS_FILE"           => NS_FILE,
1057                         "NS_FILE_TALK"      => NS_FILE_TALK,
1058                         "NS_IMAGE"          => NS_IMAGE,  // NS_IMAGE is an alias for NS_FILE
1059                         "NS_IMAGE_TALK"     => NS_IMAGE_TALK,
1060                         "NS_MEDIAWIKI"      => NS_MEDIAWIKI,
1061                         "NS_MEDIAWIKI_TALK" => NS_MEDIAWIKI_TALK,
1062                         "NS_TEMPLATE"       => NS_TEMPLATE,
1063                         "NS_TEMPLATE_TALK"  => NS_TEMPLATE_TALK,
1064                         "NS_HELP"           => NS_HELP,
1065                         "NS_HELP_TALK"      => NS_HELP_TALK,
1066                         "NS_CATEGORY"       => NS_CATEGORY,
1067                         "NS_CATEGORY_TALK"  => NS_CATEGORY_TALK );
1068
1069                 if ( $param { 0 } == '!' ) {
1070                         $this->invert = true;
1071                         $param = substr( $param, 1 );
1072                 }
1073
1074                 foreach ( explode( ',', $param ) as $key ) {
1075                         $key = trim( $key );
1076                         if ( isset( $constants[$key] ) ) {
1077                                 $ns = $constants[$key];
1078                                 $this->namespaces[$ns] = true;
1079                         } elseif ( is_numeric( $key ) ) {
1080                                 $ns = intval( $key );
1081                                 $this->namespaces[$ns] = true;
1082                         } else {
1083                                 throw new MWException( "Unrecognized namespace key '$key'\n" );
1084                         }
1085                 }
1086         }
1087
1088         function pass( $page ) {
1089                 $match = isset( $this->namespaces[$page->page_namespace] );
1090                 return $this->invert xor $match;
1091         }
1092 }
1093
1094
1095 /**
1096  * Dump output filter to include only the last revision in each page sequence.
1097  * @ingroup Dump
1098  */
1099 class DumpLatestFilter extends DumpFilter {
1100         var $page, $pageString, $rev, $revString;
1101
1102         function writeOpenPage( $page, $string ) {
1103                 $this->page = $page;
1104                 $this->pageString = $string;
1105         }
1106
1107         function writeClosePage( $string ) {
1108                 if ( $this->rev ) {
1109                         $this->sink->writeOpenPage( $this->page, $this->pageString );
1110                         $this->sink->writeRevision( $this->rev, $this->revString );
1111                         $this->sink->writeClosePage( $string );
1112                 }
1113                 $this->rev = null;
1114                 $this->revString = null;
1115                 $this->page = null;
1116                 $this->pageString = null;
1117         }
1118
1119         function writeRevision( $rev, $string ) {
1120                 if ( $rev->rev_id == $this->page->page_latest ) {
1121                         $this->rev = $rev;
1122                         $this->revString = $string;
1123                 }
1124         }
1125 }
1126
1127 /**
1128  * Base class for output stream; prints to stdout or buffer or whereever.
1129  * @ingroup Dump
1130  */
1131 class DumpMultiWriter {
1132         function __construct( $sinks ) {
1133                 $this->sinks = $sinks;
1134                 $this->count = count( $sinks );
1135         }
1136
1137         function writeOpenStream( $string ) {
1138                 for ( $i = 0; $i < $this->count; $i++ ) {
1139                         $this->sinks[$i]->writeOpenStream( $string );
1140                 }
1141         }
1142
1143         function writeCloseStream( $string ) {
1144                 for ( $i = 0; $i < $this->count; $i++ ) {
1145                         $this->sinks[$i]->writeCloseStream( $string );
1146                 }
1147         }
1148
1149         function writeOpenPage( $page, $string ) {
1150                 for ( $i = 0; $i < $this->count; $i++ ) {
1151                         $this->sinks[$i]->writeOpenPage( $page, $string );
1152                 }
1153         }
1154
1155         function writeClosePage( $string ) {
1156                 for ( $i = 0; $i < $this->count; $i++ ) {
1157                         $this->sinks[$i]->writeClosePage( $string );
1158                 }
1159         }
1160
1161         function writeRevision( $rev, $string ) {
1162                 for ( $i = 0; $i < $this->count; $i++ ) {
1163                         $this->sinks[$i]->writeRevision( $rev, $string );
1164                 }
1165         }
1166
1167         function closeRenameAndReopen( $newnames ) {
1168                 $this->closeAndRename( $newnames, true );
1169         }
1170
1171         function closeAndRename( $newnames, $open = false ) {
1172                 for ( $i = 0; $i < $this->count; $i++ ) {
1173                         $this->sinks[$i]->closeAndRename( $newnames[$i], $open );
1174                 }
1175         }
1176
1177         function getFilenames() {
1178                 $filenames = array();
1179                 for ( $i = 0; $i < $this->count; $i++ ) {
1180                         $filenames[] =  $this->sinks[$i]->getFilenames();
1181                 }
1182                 return $filenames;
1183         }
1184
1185 }
1186
1187 function xmlsafe( $string ) {
1188         wfProfileIn( __FUNCTION__ );
1189
1190         /**
1191          * The page may contain old data which has not been properly normalized.
1192          * Invalid UTF-8 sequences or forbidden control characters will make our
1193          * XML output invalid, so be sure to strip them out.
1194          */
1195         $string = UtfNormal::cleanUp( $string );
1196
1197         $string = htmlspecialchars( $string );
1198         wfProfileOut( __FUNCTION__ );
1199         return $string;
1200 }