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
 490                 if ( $row->rev_sha1 ) {
 491                         $out .= "      " . Xml::element('sha1', null, strval($row->rev_sha1) ) . "\n";
 492                 } else {
 493                         $out .= "      <sha1/>\n";
 494                 }
 495
 496                 if ( $row->page_restrictions != '' ) {
 497                         $out .= '    ' . Xml::element( 'restrictions', array(),
 498                                 strval( $row->page_restrictions ) ) . "\n";
 499                 }
 500
 501                 wfRunHooks( 'XmlDumpWriterOpenPage', array( $this, &$out, $row, $title ) );
 502
 503                 return $out;
 504         }
 505
 506         /**
 507          * Closes a <page> section on the output stream.
 508          *
 509          * @access private
 510          */
 511         function closePage() {
 512                 return "  </page>\n";
 513         }
 514
 515         /**
 516          * Dumps a <revision> section on the output stream, with
 517          * data filled in from the given database row.
 518          *
 519          * @param $row object
 520          * @return string
 521          * @access private
 522          */
 523         function writeRevision( $row ) {
 524                 wfProfileIn( __METHOD__ );
 525
 526                 $out  = "    <revision>\n";
 527                 $out .= "      " . Xml::element( 'id', null, strval( $row->rev_id ) ) . "\n";
 528
 529                 $out .= $this->writeTimestamp( $row->rev_timestamp );
 530
 531                 if ( $row->rev_deleted & Revision::DELETED_USER ) {
 532                         $out .= "      " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 533                 } else {
 534                         $out .= $this->writeContributor( $row->rev_user, $row->rev_user_text );
 535                 }
 536
 537                 if ( $row->rev_minor_edit ) {
 538                         $out .=  "      <minor/>\n";
 539                 }
 540                 if ( $row->rev_deleted & Revision::DELETED_COMMENT ) {
 541                         $out .= "      " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 542                 } elseif ( $row->rev_comment != '' ) {
 543                         $out .= "      " . Xml::elementClean( 'comment', null, strval( $row->rev_comment ) ) . "\n";
 544                 }
 545
 546                 $text = '';
 547                 if ( $row->rev_deleted & Revision::DELETED_TEXT ) {
 548                         $out .= "      " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 549                 } elseif ( isset( $row->old_text ) ) {
 550                         // Raw text from the database may have invalid chars
 551                         $text = strval( Revision::getRevisionText( $row ) );
 552                         $out .= "      " . Xml::elementClean( 'text',
 553                                 array( 'xml:space' => 'preserve', 'bytes' => intval( $row->rev_len ) ),
 554                                 strval( $text ) ) . "\n";
 555                 } else {
 556                         // Stub output
 557                         $out .= "      " . Xml::element( 'text',
 558                                 array( 'id' => $row->rev_text_id, 'bytes' => intval( $row->rev_len ) ),
 559                                 "" ) . "\n";
 560                 }
 561
 562                 wfRunHooks( 'XmlDumpWriterWriteRevision', array( &$this, &$out, $row, $text ) );
 563
 564                 $out .= "    </revision>\n";
 565
 566                 wfProfileOut( __METHOD__ );
 567                 return $out;
 568         }
 569
 570         /**
 571          * Dumps a <logitem> section on the output stream, with
 572          * data filled in from the given database row.
 573          *
 574          * @param $row object
 575          * @return string
 576          * @access private
 577          */
 578         function writeLogItem( $row ) {
 579                 wfProfileIn( __METHOD__ );
 580
 581                 $out  = "    <logitem>\n";
 582                 $out .= "      " . Xml::element( 'id', null, strval( $row->log_id ) ) . "\n";
 583
 584                 $out .= $this->writeTimestamp( $row->log_timestamp );
 585
 586                 if ( $row->log_deleted & LogPage::DELETED_USER ) {
 587                         $out .= "      " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 588                 } else {
 589                         $out .= $this->writeContributor( $row->log_user, $row->user_name );
 590                 }
 591
 592                 if ( $row->log_deleted & LogPage::DELETED_COMMENT ) {
 593                         $out .= "      " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 594                 } elseif ( $row->log_comment != '' ) {
 595                         $out .= "      " . Xml::elementClean( 'comment', null, strval( $row->log_comment ) ) . "\n";
 596                 }
 597
 598                 $out .= "      " . Xml::element( 'type', null, strval( $row->log_type ) ) . "\n";
 599                 $out .= "      " . Xml::element( 'action', null, strval( $row->log_action ) ) . "\n";
 600
 601                 if ( $row->log_deleted & LogPage::DELETED_ACTION ) {
 602                         $out .= "      " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 603                 } else {
 604                         $title = Title::makeTitle( $row->log_namespace, $row->log_title );
 605                         $out .= "      " . Xml::elementClean( 'logtitle', null, self::canonicalTitle( $title ) ) . "\n";
 606                         $out .= "      " . Xml::elementClean( 'params',
 607                                 array( 'xml:space' => 'preserve' ),
 608                                 strval( $row->log_params ) ) . "\n";
 609                 }
 610
 611                 $out .= "    </logitem>\n";
 612
 613                 wfProfileOut( __METHOD__ );
 614                 return $out;
 615         }
 616
 617         function writeTimestamp( $timestamp ) {
 618                 $ts = wfTimestamp( TS_ISO_8601, $timestamp );
 619                 return "      " . Xml::element( 'timestamp', null, $ts ) . "\n";
 620         }
 621
 622         function writeContributor( $id, $text ) {
 623                 $out = "      <contributor>\n";
 624                 if ( $id || !IP::isValid( $text ) ) {
 625                         $out .= "        " . Xml::elementClean( 'username', null, strval( $text ) ) . "\n";
 626                         $out .= "        " . Xml::element( 'id', null, strval( $id ) ) . "\n";
 627                 } else {
 628                         $out .= "        " . Xml::elementClean( 'ip', null, strval( $text ) ) . "\n";
 629                 }
 630                 $out .= "      </contributor>\n";
 631                 return $out;
 632         }
 633
 634         /**
 635          * Warning! This data is potentially inconsistent. :(
 636          */
 637         function writeUploads( $row, $dumpContents = false ) {
 638                 if ( $row->page_namespace == NS_IMAGE ) {
 639                         $img = wfLocalFile( $row->page_title );
 640                         if ( $img && $img->exists() ) {
 641                                 $out = '';
 642                                 foreach ( array_reverse( $img->getHistory() ) as $ver ) {
 643                                         $out .= $this->writeUpload( $ver, $dumpContents );
 644                                 }
 645                                 $out .= $this->writeUpload( $img, $dumpContents );
 646                                 return $out;
 647                         }
 648                 }
 649                 return '';
 650         }
 651
 652         /**
 653          * @param $file File
 654          * @param $dumpContents bool
 655          * @return string
 656          */
 657         function writeUpload( $file, $dumpContents = false ) {
 658                 if ( $file->isOld() ) {
 659                         $archiveName = "      " .
 660                                 Xml::element( 'archivename', null, $file->getArchiveName() ) . "\n";
 661                 } else {
 662                         $archiveName = '';
 663                 }
 664                 if ( $dumpContents ) {
 665                         # Dump file as base64
 666                         # Uses only XML-safe characters, so does not need escaping
 667                         $contents = '      <contents encoding="base64">' .
 668                                 chunk_split( base64_encode( file_get_contents( $file->getPath() ) ) ) .
 669                                 "      </contents>\n";
 670                 } else {
 671                         $contents = '';
 672                 }
 673                 return "    <upload>\n" .
 674                         $this->writeTimestamp( $file->getTimestamp() ) .
 675                         $this->writeContributor( $file->getUser( 'id' ), $file->getUser( 'text' ) ) .
 676                         "      " . Xml::elementClean( 'comment', null, $file->getDescription() ) . "\n" .
 677                         "      " . Xml::element( 'filename', null, $file->getName() ) . "\n" .
 678                         $archiveName .
 679                         "      " . Xml::element( 'src', null, $file->getCanonicalUrl() ) . "\n" .
 680                         "      " . Xml::element( 'size', null, $file->getSize() ) . "\n" .
 681                         "      " . Xml::element( 'sha1base36', null, $file->getSha1() ) . "\n" .
 682                         "      " . Xml::element( 'rel', null, $file->getRel() ) . "\n" .
 683                         $contents .
 684                         "    </upload>\n";
 685         }
 686
 687         /**
 688          * Return prefixed text form of title, but using the content language's
 689          * canonical namespace. This skips any special-casing such as gendered
 690          * user namespaces -- which while useful, are not yet listed in the
 691          * XML <siteinfo> data so are unsafe in export.
 692          *
 693          * @param Title $title
 694          * @return string
 695          * @since 1.18
 696          */
 697         public static function canonicalTitle( Title $title ) {
 698                 if ( $title->getInterwiki() ) {
 699                         return $title->getPrefixedText();
 700                 }
 701
 702                 global $wgContLang;
 703                 $prefix = str_replace( '_', ' ', $wgContLang->getNsText( $title->getNamespace() ) );
 704
 705                 if ( $prefix !== '' ) {
 706                         $prefix .= ':';
 707                 }
 708
 709                 return $prefix . $title->getText();
 710         }
 711 }
 712
 713
 714 /**
 715  * Base class for output stream; prints to stdout or buffer or whereever.
 716  * @ingroup Dump
 717  */
 718 class DumpOutput {
 719         function writeOpenStream( $string ) {
 720                 $this->write( $string );
 721         }
 722
 723         function writeCloseStream( $string ) {
 724                 $this->write( $string );
 725         }
 726
 727         function writeOpenPage( $page, $string ) {
 728                 $this->write( $string );
 729         }
 730
 731         function writeClosePage( $string ) {
 732                 $this->write( $string );
 733         }
 734
 735         function writeRevision( $rev, $string ) {
 736                 $this->write( $string );
 737         }
 738
 739         function writeLogItem( $rev, $string ) {
 740                 $this->write( $string );
 741         }
 742
 743         /**
 744          * Override to write to a different stream type.
 745          * @return bool
 746          */
 747         function write( $string ) {
 748                 print $string;
 749         }
 750
 751         /**
 752          * Close the old file, move it to a specified name,
 753          * and reopen new file with the old name. Use this
 754          * for writing out a file in multiple pieces
 755          * at specified checkpoints (e.g. every n hours).
 756          * @param $newname mixed File name. May be a string or an array with one element
 757          */
 758         function closeRenameAndReopen( $newname ) {
 759                 return;
 760         }
 761
 762         /**
 763          * Close the old file, and move it to a specified name.
 764          * Use this for the last piece of a file written out
 765          * at specified checkpoints (e.g. every n hours).
 766          * @param $newname mixed File name. May be a string or an array with one element
 767          * @param $open bool If true, a new file with the old filename will be opened again for writing (default: false)
 768          */
 769         function closeAndRename( $newname, $open = false ) {
 770                 return;
 771         }
 772
 773         /**
 774          * Returns the name of the file or files which are
 775          * being written to, if there are any.
 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, $filename;
 788
 789         function __construct( $file ) {
 790                 $this->handle = fopen( $file, "wt" );
 791                 $this->filename = $file;
 792         }
 793
 794         function write( $string ) {
 795                 fputs( $this->handle, $string );
 796         }
 797
 798         function closeRenameAndReopen( $newname ) {
 799                 $this->closeAndRename( $newname, true );
 800         }
 801
 802         function renameOrException( $newname ) {
 803                         if (! rename( $this->filename, $newname ) ) {
 804                                 throw new MWException( __METHOD__ . ": rename of file {$this->filename} to $newname failed\n" );
 805                         }
 806         }
 807
 808         function checkRenameArgCount( $newname ) {
 809                 if ( is_array( $newname ) ) {
 810                         if ( count( $newname ) > 1 ) {
 811                                 throw new MWException( __METHOD__ . ": passed multiple arguments for rename of single file\n" );
 812                         } else {
 813                                 $newname = $newname[0];
 814                         }
 815                 }
 816                 return $newname;
 817         }
 818
 819         function closeAndRename( $newname, $open = false ) {
 820                 $newname = $this->checkRenameArgCount( $newname );
 821                 if ( $newname ) {
 822                         fclose( $this->handle );
 823                         $this->renameOrException( $newname );
 824                         if ( $open ) {
 825                                 $this->handle = fopen( $this->filename, "wt" );
 826                         }
 827                 }
 828         }
 829
 830         function getFilenames() {
 831                 return $this->filename;
 832         }
 833 }
 834
 835 /**
 836  * Stream outputter to send data to a file via some filter program.
 837  * Even if compression is available in a library, using a separate
 838  * program can allow us to make use of a multi-processor system.
 839  * @ingroup Dump
 840  */
 841 class DumpPipeOutput extends DumpFileOutput {
 842         protected $command, $filename;
 843
 844         function __construct( $command, $file = null ) {
 845                 if ( !is_null( $file ) ) {
 846                         $command .=  " > " . wfEscapeShellArg( $file );
 847                 }
 848
 849                 $this->startCommand( $command );
 850                 $this->command = $command;
 851                 $this->filename = $file;
 852         }
 853
 854         function startCommand( $command ) {
 855                 $spec = array(
 856                         0 => array( "pipe", "r" ),
 857                 );
 858                 $pipes = array();
 859                 $this->procOpenResource = proc_open( $command, $spec, $pipes );
 860                 $this->handle = $pipes[0];
 861         }
 862
 863         function closeRenameAndReopen( $newname ) {
 864                 $this->closeAndRename( $newname, true );
 865         }
 866
 867         function closeAndRename( $newname, $open = false ) {
 868                 $newname = $this->checkRenameArgCount( $newname );
 869                 if ( $newname ) {
 870                         fclose( $this->handle );
 871                         proc_close( $this->procOpenResource );
 872                         $this->renameOrException( $newname );
 873                         if ( $open ) {
 874                                 $command = $this->command;
 875                                 $command .=  " > " . wfEscapeShellArg( $this->filename );
 876                                 $this->startCommand( $command );
 877                         }
 878                 }
 879         }
 880
 881 }
 882
 883 /**
 884  * Sends dump output via the gzip compressor.
 885  * @ingroup Dump
 886  */
 887 class DumpGZipOutput extends DumpPipeOutput {
 888         function __construct( $file ) {
 889                 parent::__construct( "gzip", $file );
 890         }
 891 }
 892
 893 /**
 894  * Sends dump output via the bgzip2 compressor.
 895  * @ingroup Dump
 896  */
 897 class DumpBZip2Output extends DumpPipeOutput {
 898         function __construct( $file ) {
 899                 parent::__construct( "bzip2", $file );
 900         }
 901 }
 902
 903 /**
 904  * Sends dump output via the p7zip compressor.
 905  * @ingroup Dump
 906  */
 907 class Dump7ZipOutput extends DumpPipeOutput {
 908         function __construct( $file ) {
 909                 $command = $this->setup7zCommand( $file );
 910                 parent::__construct( $command );
 911                 $this->filename = $file;
 912         }
 913
 914         function setup7zCommand( $file ) {
 915                 $command = "7za a -bd -si " . wfEscapeShellArg( $file );
 916                 // Suppress annoying useless crap from p7zip
 917                 // Unfortunately this could suppress real error messages too
 918                 $command .= ' >' . wfGetNull() . ' 2>&1';
 919                 return( $command );
 920         }
 921
 922         function closeAndRename( $newname, $open = false ) {
 923                 $newname = $this->checkRenameArgCount( $newname );
 924                 if ( $newname ) {
 925                         fclose( $this->handle );
 926                         proc_close( $this->procOpenResource );
 927                         $this->renameOrException( $newname );
 928                         if ( $open ) {
 929                                 $command = $this->setup7zCommand( $this->filename );
 930                                 $this->startCommand( $command );
 931                         }
 932                 }
 933         }
 934 }
 935
 936
 937
 938 /**
 939  * Dump output filter class.
 940  * This just does output filtering and streaming; XML formatting is done
 941  * higher up, so be careful in what you do.
 942  * @ingroup Dump
 943  */
 944 class DumpFilter {
 945         function __construct( &$sink ) {
 946                 $this->sink =& $sink;
 947         }
 948
 949         function writeOpenStream( $string ) {
 950                 $this->sink->writeOpenStream( $string );
 951         }
 952
 953         function writeCloseStream( $string ) {
 954                 $this->sink->writeCloseStream( $string );
 955         }
 956
 957         function writeOpenPage( $page, $string ) {
 958                 $this->sendingThisPage = $this->pass( $page, $string );
 959                 if ( $this->sendingThisPage ) {
 960                         $this->sink->writeOpenPage( $page, $string );
 961                 }
 962         }
 963
 964         function writeClosePage( $string ) {
 965                 if ( $this->sendingThisPage ) {
 966                         $this->sink->writeClosePage( $string );
 967                         $this->sendingThisPage = false;
 968                 }
 969         }
 970
 971         function writeRevision( $rev, $string ) {
 972                 if ( $this->sendingThisPage ) {
 973                         $this->sink->writeRevision( $rev, $string );
 974                 }
 975         }
 976
 977         function writeLogItem( $rev, $string ) {
 978                 $this->sink->writeRevision( $rev, $string );
 979         }
 980
 981         function closeRenameAndReopen( $newname ) {
 982                 $this->sink->closeRenameAndReopen( $newname );
 983         }
 984
 985         function closeAndRename( $newname, $open = false ) {
 986                 $this->sink->closeAndRename( $newname, $open );
 987         }
 988
 989         function getFilenames() {
 990                 return $this->sink->getFilenames();
 991         }
 992
 993         /**
 994          * Override for page-based filter types.
 995          * @return bool
 996          */
 997         function pass( $page ) {
 998                 return true;
 999         }
1000 }
1001
1002 /**
1003  * Simple dump output filter to exclude all talk pages.
1004  * @ingroup Dump
1005  */
1006 class DumpNotalkFilter extends DumpFilter {
1007         function pass( $page ) {
1008                 return !MWNamespace::isTalk( $page->page_namespace );
1009         }
1010 }
1011
1012 /**
1013  * Dump output filter to include or exclude pages in a given set of namespaces.
1014  * @ingroup Dump
1015  */
1016 class DumpNamespaceFilter extends DumpFilter {
1017         var $invert = false;
1018         var $namespaces = array();
1019
1020         function __construct( &$sink, $param ) {
1021                 parent::__construct( $sink );
1022
1023                 $constants = array(
1024                         "NS_MAIN"           => NS_MAIN,
1025                         "NS_TALK"           => NS_TALK,
1026                         "NS_USER"           => NS_USER,
1027                         "NS_USER_TALK"      => NS_USER_TALK,
1028                         "NS_PROJECT"        => NS_PROJECT,
1029                         "NS_PROJECT_TALK"   => NS_PROJECT_TALK,
1030                         "NS_FILE"           => NS_FILE,
1031                         "NS_FILE_TALK"      => NS_FILE_TALK,
1032                         "NS_IMAGE"          => NS_IMAGE,  // NS_IMAGE is an alias for NS_FILE
1033                         "NS_IMAGE_TALK"     => NS_IMAGE_TALK,
1034                         "NS_MEDIAWIKI"      => NS_MEDIAWIKI,
1035                         "NS_MEDIAWIKI_TALK" => NS_MEDIAWIKI_TALK,
1036                         "NS_TEMPLATE"       => NS_TEMPLATE,
1037                         "NS_TEMPLATE_TALK"  => NS_TEMPLATE_TALK,
1038                         "NS_HELP"           => NS_HELP,
1039                         "NS_HELP_TALK"      => NS_HELP_TALK,
1040                         "NS_CATEGORY"       => NS_CATEGORY,
1041                         "NS_CATEGORY_TALK"  => NS_CATEGORY_TALK );
1042
1043                 if ( $param { 0 } == '!' ) {
1044                         $this->invert = true;
1045                         $param = substr( $param, 1 );
1046                 }
1047
1048                 foreach ( explode( ',', $param ) as $key ) {
1049                         $key = trim( $key );
1050                         if ( isset( $constants[$key] ) ) {
1051                                 $ns = $constants[$key];
1052                                 $this->namespaces[$ns] = true;
1053                         } elseif ( is_numeric( $key ) ) {
1054                                 $ns = intval( $key );
1055                                 $this->namespaces[$ns] = true;
1056                         } else {
1057                                 throw new MWException( "Unrecognized namespace key '$key'\n" );
1058                         }
1059                 }
1060         }
1061
1062         function pass( $page ) {
1063                 $match = isset( $this->namespaces[$page->page_namespace] );
1064                 return $this->invert xor $match;
1065         }
1066 }
1067
1068
1069 /**
1070  * Dump output filter to include only the last revision in each page sequence.
1071  * @ingroup Dump
1072  */
1073 class DumpLatestFilter extends DumpFilter {
1074         var $page, $pageString, $rev, $revString;
1075
1076         function writeOpenPage( $page, $string ) {
1077                 $this->page = $page;
1078                 $this->pageString = $string;
1079         }
1080
1081         function writeClosePage( $string ) {
1082                 if ( $this->rev ) {
1083                         $this->sink->writeOpenPage( $this->page, $this->pageString );
1084                         $this->sink->writeRevision( $this->rev, $this->revString );
1085                         $this->sink->writeClosePage( $string );
1086                 }
1087                 $this->rev = null;
1088                 $this->revString = null;
1089                 $this->page = null;
1090                 $this->pageString = null;
1091         }
1092
1093         function writeRevision( $rev, $string ) {
1094                 if ( $rev->rev_id == $this->page->page_latest ) {
1095                         $this->rev = $rev;
1096                         $this->revString = $string;
1097                 }
1098         }
1099 }
1100
1101 /**
1102  * Base class for output stream; prints to stdout or buffer or whereever.
1103  * @ingroup Dump
1104  */
1105 class DumpMultiWriter {
1106         function __construct( $sinks ) {
1107                 $this->sinks = $sinks;
1108                 $this->count = count( $sinks );
1109         }
1110
1111         function writeOpenStream( $string ) {
1112                 for ( $i = 0; $i < $this->count; $i++ ) {
1113                         $this->sinks[$i]->writeOpenStream( $string );
1114                 }
1115         }
1116
1117         function writeCloseStream( $string ) {
1118                 for ( $i = 0; $i < $this->count; $i++ ) {
1119                         $this->sinks[$i]->writeCloseStream( $string );
1120                 }
1121         }
1122
1123         function writeOpenPage( $page, $string ) {
1124                 for ( $i = 0; $i < $this->count; $i++ ) {
1125                         $this->sinks[$i]->writeOpenPage( $page, $string );
1126                 }
1127         }
1128
1129         function writeClosePage( $string ) {
1130                 for ( $i = 0; $i < $this->count; $i++ ) {
1131                         $this->sinks[$i]->writeClosePage( $string );
1132                 }
1133         }
1134
1135         function writeRevision( $rev, $string ) {
1136                 for ( $i = 0; $i < $this->count; $i++ ) {
1137                         $this->sinks[$i]->writeRevision( $rev, $string );
1138                 }
1139         }
1140
1141         function closeRenameAndReopen( $newnames ) {
1142                 $this->closeAndRename( $newnames, true );
1143         }
1144
1145         function closeAndRename( $newnames, $open = false ) {
1146                 for ( $i = 0; $i < $this->count; $i++ ) {
1147                         $this->sinks[$i]->closeAndRename( $newnames[$i], $open );
1148                 }
1149         }
1150
1151         function getFilenames() {
1152                 $filenames = array();
1153                 for ( $i = 0; $i < $this->count; $i++ ) {
1154                         $filenames[] =  $this->sinks[$i]->getFilenames();
1155                 }
1156                 return $filenames;
1157         }
1158
1159 }
1160
1161 function xmlsafe( $string ) {
1162         wfProfileIn( __FUNCTION__ );
1163
1164         /**
1165          * The page may contain old data which has not been properly normalized.
1166          * Invalid UTF-8 sequences or forbidden control characters will make our
1167          * XML output invalid, so be sure to strip them out.
1168          */
1169         $string = UtfNormal::cleanUp( $string );
1170
1171         $string = htmlspecialchars( $string );
1172         wfProfileOut( __FUNCTION__ );
1173         return $string;
1174 }