includes/Export.php

   1 <?php
   2 # Copyright (C) 2003, 2005, 2006 Brion Vibber <brion@pobox.com>
   3 # http://www.mediawiki.org/
   4 #
   5 # This program is free software; you can redistribute it and/or modify
   6 # it under the terms of the GNU General Public License as published by
   7 # the Free Software Foundation; either version 2 of the License, or
   8 # (at your option) any later version.
   9 #
  10 # This program is distributed in the hope that it will be useful,
  11 # but WITHOUT ANY WARRANTY; without even the implied warranty of
  12 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13 # GNU General Public License for more details.
  14 #
  15 # You should have received a copy of the GNU General Public License along
  16 # with this program; if not, write to the Free Software Foundation, Inc.,
  17 # 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18 # http://www.gnu.org/copyleft/gpl.html
  19
  20
  21 /**
  22  *
  23  * @addtogroup SpecialPage
  24  */
  25 class WikiExporter {
  26         var $list_authors = false ; # Return distinct author list (when not returning full history)
  27         var $author_list = "" ;
  28
  29         const FULL = 0;
  30         const CURRENT = 1;
  31
  32         const BUFFER = 0;
  33         const STREAM = 1;
  34
  35         const TEXT = 0;
  36         const STUB = 1;
  37
  38         /**
  39          * If using WikiExporter::STREAM to stream a large amount of data,
  40          * provide a database connection which is not managed by
  41          * LoadBalancer to read from: some history blob types will
  42          * make additional queries to pull source data while the
  43          * main query is still running.
  44          *
  45          * @param Database $db
  46          * @param mixed $history one of WikiExporter::FULL or WikiExporter::CURRENT, or an
  47          *                       associative array:
  48          *                         offset: non-inclusive offset at which to start the query
  49          *                         limit: maximum number of rows to return
  50          *                         dir: "asc" or "desc" timestamp order
  51          * @param int $buffer one of WikiExporter::BUFFER or WikiExporter::STREAM
  52          */
  53         function __construct( &$db, $history = WikiExporter::CURRENT,
  54                         $buffer = WikiExporter::BUFFER, $text = WikiExporter::TEXT ) {
  55                 $this->db =& $db;
  56                 $this->history = $history;
  57                 $this->buffer  = $buffer;
  58                 $this->writer  = new XmlDumpWriter();
  59                 $this->sink    = new DumpOutput();
  60                 $this->text    = $text;
  61         }
  62
  63         /**
  64          * Set the DumpOutput or DumpFilter object which will receive
  65          * various row objects and XML output for filtering. Filters
  66          * can be chained or used as callbacks.
  67          *
  68          * @param mixed $callback
  69          */
  70         function setOutputSink( &$sink ) {
  71                 $this->sink =& $sink;
  72         }
  73
  74         function openStream() {
  75                 $output = $this->writer->openStream();
  76                 $this->sink->writeOpenStream( $output );
  77         }
  78
  79         function closeStream() {
  80                 $output = $this->writer->closeStream();
  81                 $this->sink->writeCloseStream( $output );
  82         }
  83
  84         /**
  85          * Dumps a series of page and revision records for all pages
  86          * in the database, either including complete history or only
  87          * the most recent version.
  88          */
  89         function allPages() {
  90                 return $this->dumpFrom( '' );
  91         }
  92
  93         /**
  94          * Dumps a series of page and revision records for those pages
  95          * in the database falling within the page_id range given.
  96          * @param int $start Inclusive lower limit (this id is included)
  97          * @param int $end   Exclusive upper limit (this id is not included)
  98          *                   If 0, no upper limit.
  99          */
 100         function pagesByRange( $start, $end ) {
 101                 $condition = 'page_id >= ' . intval( $start );
 102                 if( $end ) {
 103                         $condition .= ' AND page_id < ' . intval( $end );
 104                 }
 105                 return $this->dumpFrom( $condition );
 106         }
 107
 108         /**
 109          * @param Title $title
 110          */
 111         function pageByTitle( $title ) {
 112                 return $this->dumpFrom(
 113                         'page_namespace=' . $title->getNamespace() .
 114                         ' AND page_title=' . $this->db->addQuotes( $title->getDbKey() ) );
 115         }
 116
 117         function pageByName( $name ) {
 118                 $title = Title::newFromText( $name );
 119                 if( is_null( $title ) ) {
 120                         return new WikiError( "Can't export invalid title" );
 121                 } else {
 122                         return $this->pageByTitle( $title );
 123                 }
 124         }
 125
 126         function pagesByName( $names ) {
 127                 foreach( $names as $name ) {
 128                         $this->pageByName( $name );
 129                 }
 130         }
 131
 132
 133         // -------------------- private implementation below --------------------
 134
 135         # Generates the distinct list of authors of an article
 136         # Not called by default (depends on $this->list_authors)
 137         # Can be set by Special:Export when not exporting whole history
 138         function do_list_authors ( $page , $revision , $cond ) {
 139                 $fname = "do_list_authors" ;
 140                 wfProfileIn( $fname );
 141                 $this->author_list = "<contributors>";
 142                 //rev_deleted
 143                 $nothidden = '(rev_deleted & '.Revision::DELETED_USER.') = 0';
 144
 145                 $sql = "SELECT DISTINCT rev_user_text,rev_user FROM {$page},{$revision} WHERE page_id=rev_page AND $nothidden AND " . $cond ;
 146                 $result = $this->db->query( $sql, $fname );
 147                 $resultset = $this->db->resultObject( $result );
 148                 while( $row = $resultset->fetchObject() ) {
 149                         $this->author_list .= "<contributor>" .
 150                                 "<username>" .
 151                                 htmlentities( $row->rev_user_text )  .
 152                                 "</username>" .
 153                                 "<id>" .
 154                                 $row->rev_user .
 155                                 "</id>" .
 156                                 "</contributor>";
 157                 }
 158                 wfProfileOut( $fname );
 159                 $this->author_list .= "</contributors>";
 160         }
 161
 162         function dumpFrom( $cond = '' ) {
 163                 $fname = 'WikiExporter::dumpFrom';
 164                 wfProfileIn( $fname );
 165
 166                 $page     = $this->db->tableName( 'page' );
 167                 $revision = $this->db->tableName( 'revision' );
 168                 $text     = $this->db->tableName( 'text' );
 169
 170                 $order = 'ORDER BY page_id';
 171                 $limit = '';
 172
 173                 if( $this->history == WikiExporter::FULL ) {
 174                         $join = 'page_id=rev_page';
 175                 } elseif( $this->history == WikiExporter::CURRENT ) {
 176                         if ( $this->list_authors && $cond != '' )  { // List authors, if so desired
 177                                 $this->do_list_authors ( $page , $revision , $cond );
 178                         }
 179                         $join = 'page_id=rev_page AND page_latest=rev_id';
 180                 } elseif ( is_array( $this->history ) ) {
 181                         $join = 'page_id=rev_page';
 182                         if ( $this->history['dir'] == 'asc' ) {
 183                                 $op = '>';
 184                                 $order .= ', rev_timestamp';
 185                         } else {
 186                                 $op = '<';
 187                                 $order .= ', rev_timestamp DESC';
 188                         }
 189                         if ( !empty( $this->history['offset'] ) ) {
 190                                 $join .= " AND rev_timestamp $op " . $this->db->addQuotes(
 191                                         $this->db->timestamp( $this->history['offset'] ) );
 192                         }
 193                         if ( !empty( $this->history['limit'] ) ) {
 194                                 $limitNum = intval( $this->history['limit'] );
 195                                 if ( $limitNum > 0 ) {
 196                                         $limit = "LIMIT $limitNum";
 197                                 }
 198                         }
 199                 } else {
 200                         wfProfileOut( $fname );
 201                         return new WikiError( "$fname given invalid history dump type." );
 202                 }
 203                 $where = ( $cond == '' ) ? '' : "$cond AND";
 204
 205                 if( $this->buffer == WikiExporter::STREAM ) {
 206                         $prev = $this->db->bufferResults( false );
 207                 }
 208                 if( $cond == '' ) {
 209                         // Optimization hack for full-database dump
 210                         $revindex = $pageindex = $this->db->useIndexClause("PRIMARY");
 211                         $straight = ' /*! STRAIGHT_JOIN */ ';
 212                 } else {
 213                         $pageindex = '';
 214                         $revindex = '';
 215                         $straight = '';
 216                 }
 217                 if( $this->text == WikiExporter::STUB ) {
 218                         $sql = "SELECT $straight * FROM
 219                                         $page $pageindex,
 220                                         $revision $revindex
 221                                         WHERE $where $join
 222                                         $order $limit";
 223                 } else {
 224                         $sql = "SELECT $straight * FROM
 225                                         $page $pageindex,
 226                                         $revision $revindex,
 227                                         $text
 228                                         WHERE $where $join AND rev_text_id=old_id
 229                                         $order $limit";
 230                 }
 231                 $result = $this->db->query( $sql, $fname );
 232                 $wrapper = $this->db->resultObject( $result );
 233                 $this->outputStream( $wrapper );
 234
 235                 if ( $this->list_authors ) {
 236                         $this->outputStream( $wrapper );
 237                 }
 238
 239                 if( $this->buffer == WikiExporter::STREAM ) {
 240                         $this->db->bufferResults( $prev );
 241                 }
 242
 243                 wfProfileOut( $fname );
 244         }
 245
 246         /**
 247          * Runs through a query result set dumping page and revision records.
 248          * The result set should be sorted/grouped by page to avoid duplicate
 249          * page records in the output.
 250          *
 251          * The result set will be freed once complete. Should be safe for
 252          * streaming (non-buffered) queries, as long as it was made on a
 253          * separate database connection not managed by LoadBalancer; some
 254          * blob storage types will make queries to pull source data.
 255          *
 256          * @param ResultWrapper $resultset
 257          * @access private
 258          */
 259         function outputStream( $resultset ) {
 260                 $last = null;
 261                 while( $row = $resultset->fetchObject() ) {
 262                         if( is_null( $last ) ||
 263                                 $last->page_namespace != $row->page_namespace ||
 264                                 $last->page_title     != $row->page_title ) {
 265                                 if( isset( $last ) ) {
 266                                         $output = $this->writer->closePage();
 267                                         $this->sink->writeClosePage( $output );
 268                                 }
 269                                 $output = $this->writer->openPage( $row );
 270                                 $this->sink->writeOpenPage( $row, $output );
 271                                 $last = $row;
 272                         }
 273                         $output = $this->writer->writeRevision( $row );
 274                         $this->sink->writeRevision( $row, $output );
 275                 }
 276                 if( isset( $last ) ) {
 277                         $output = $this->author_list . $this->writer->closePage();
 278                         $this->sink->writeClosePage( $output );
 279                 }
 280                 $resultset->free();
 281         }
 282 }
 283
 284 /**
 285  * @addtogroup Dump
 286  */
 287 class XmlDumpWriter {
 288
 289         /**
 290          * Returns the export schema version.
 291          * @return string
 292          */
 293         function schemaVersion() {
 294                 return "0.3"; // FIXME: upgrade to 0.4 when updated XSD is ready, for the revision deletion bits
 295         }
 296
 297         /**
 298          * Opens the XML output stream's root <mediawiki> element.
 299          * This does not include an xml directive, so is safe to include
 300          * as a subelement in a larger XML stream. Namespace and XML Schema
 301          * references are included.
 302          *
 303          * Output will be encoded in UTF-8.
 304          *
 305          * @return string
 306          */
 307         function openStream() {
 308                 global $wgContLanguageCode;
 309                 $ver = $this->schemaVersion();
 310                 return wfElement( 'mediawiki', array(
 311                         'xmlns'              => "http://www.mediawiki.org/xml/export-$ver/",
 312                         'xmlns:xsi'          => "http://www.w3.org/2001/XMLSchema-instance",
 313                         'xsi:schemaLocation' => "http://www.mediawiki.org/xml/export-$ver/ " .
 314                                                 "http://www.mediawiki.org/xml/export-$ver.xsd",
 315                         'version'            => $ver,
 316                         'xml:lang'           => $wgContLanguageCode ),
 317                         null ) .
 318                         "\n" .
 319                         $this->siteInfo();
 320         }
 321
 322         function siteInfo() {
 323                 $info = array(
 324                         $this->sitename(),
 325                         $this->homelink(),
 326                         $this->generator(),
 327                         $this->caseSetting(),
 328                         $this->namespaces() );
 329                 return "  <siteinfo>\n    " .
 330                         implode( "\n    ", $info ) .
 331                         "\n  </siteinfo>\n";
 332         }
 333
 334         function sitename() {
 335                 global $wgSitename;
 336                 return wfElement( 'sitename', array(), $wgSitename );
 337         }
 338
 339         function generator() {
 340                 global $wgVersion;
 341                 return wfElement( 'generator', array(), "MediaWiki $wgVersion" );
 342         }
 343
 344         function homelink() {
 345                 return wfElement( 'base', array(), Title::newMainPage()->getFullUrl() );
 346         }
 347
 348         function caseSetting() {
 349                 global $wgCapitalLinks;
 350                 // "case-insensitive" option is reserved for future
 351                 $sensitivity = $wgCapitalLinks ? 'first-letter' : 'case-sensitive';
 352                 return wfElement( 'case', array(), $sensitivity );
 353         }
 354
 355         function namespaces() {
 356                 global $wgContLang;
 357                 $spaces = "  <namespaces>\n";
 358                 foreach( $wgContLang->getFormattedNamespaces() as $ns => $title ) {
 359                         $spaces .= '      ' . wfElement( 'namespace', array( 'key' => $ns ), $title ) . "\n";
 360                 }
 361                 $spaces .= "    </namespaces>";
 362                 return $spaces;
 363         }
 364
 365         /**
 366          * Closes the output stream with the closing root element.
 367          * Call when finished dumping things.
 368          */
 369         function closeStream() {
 370                 return "</mediawiki>\n";
 371         }
 372
 373
 374         /**
 375          * Opens a <page> section on the output stream, with data
 376          * from the given database row.
 377          *
 378          * @param object $row
 379          * @return string
 380          * @access private
 381          */
 382         function openPage( $row ) {
 383                 $out = "  <page>\n";
 384                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 385                 $out .= '    ' . wfElementClean( 'title', array(), $title->getPrefixedText() ) . "\n";
 386                 $out .= '    ' . wfElement( 'id', array(), strval( $row->page_id ) ) . "\n";
 387                 if( '' != $row->page_restrictions ) {
 388                         $out .= '    ' . wfElement( 'restrictions', array(),
 389                                 strval( $row->page_restrictions ) ) . "\n";
 390                 }
 391                 return $out;
 392         }
 393
 394         /**
 395          * Closes a <page> section on the output stream.
 396          *
 397          * @access private
 398          */
 399         function closePage() {
 400                 return "  </page>\n";
 401         }
 402
 403         /**
 404          * Dumps a <revision> section on the output stream, with
 405          * data filled in from the given database row.
 406          *
 407          * @param object $row
 408          * @return string
 409          * @access private
 410          */
 411         function writeRevision( $row ) {
 412                 $fname = 'WikiExporter::dumpRev';
 413                 wfProfileIn( $fname );
 414
 415                 $out  = "    <revision>\n";
 416                 $out .= "      " . wfElement( 'id', null, strval( $row->rev_id ) ) . "\n";
 417
 418                 $ts = wfTimestamp( TS_ISO_8601, $row->rev_timestamp );
 419                 $out .= "      " . wfElement( 'timestamp', null, $ts ) . "\n";
 420
 421                 if( $row->rev_deleted & Revision::DELETED_USER ) {
 422                         $out .= "      " . wfElement( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 423                 } else {
 424                         $out .= "      <contributor>\n";
 425                         if( $row->rev_user ) {
 426                                 $out .= "        " . wfElementClean( 'username', null, strval( $row->rev_user_text ) ) . "\n";
 427                                 $out .= "        " . wfElement( 'id', null, strval( $row->rev_user ) ) . "\n";
 428                         } else {
 429                                 $out .= "        " . wfElementClean( 'ip', null, strval( $row->rev_user_text ) ) . "\n";
 430                         }
 431                         $out .= "      </contributor>\n";
 432                 }
 433
 434                 if( $row->rev_minor_edit ) {
 435                         $out .=  "      <minor/>\n";
 436                 }
 437                 if( $row->rev_deleted & Revision::DELETED_COMMENT ) {
 438                         $out .= "      " . wfElement( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 439                 } elseif( $row->rev_comment != '' ) {
 440                         $out .= "      " . wfElementClean( 'comment', null, strval( $row->rev_comment ) ) . "\n";
 441                 }
 442
 443                 if( $row->rev_deleted & Revision::DELETED_TEXT ) {
 444                         $out .= "      " . wfElement( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 445                 } elseif( isset( $row->old_text ) ) {
 446                         // Raw text from the database may have invalid chars
 447                         $text = strval( Revision::getRevisionText( $row ) );
 448                         $out .= "      " . wfElementClean( 'text',
 449                                 array( 'xml:space' => 'preserve' ),
 450                                 strval( $text ) ) . "\n";
 451                 } else {
 452                         // Stub output
 453                         $out .= "      " . wfElement( 'text',
 454                                 array( 'id' => $row->rev_text_id ),
 455                                 "" ) . "\n";
 456                 }
 457
 458                 $out .= "    </revision>\n";
 459
 460                 wfProfileOut( $fname );
 461                 return $out;
 462         }
 463
 464 }
 465
 466
 467 /**
 468  * Base class for output stream; prints to stdout or buffer or whereever.
 469  * @addtogroup Dump
 470  */
 471 class DumpOutput {
 472         function writeOpenStream( $string ) {
 473                 $this->write( $string );
 474         }
 475
 476         function writeCloseStream( $string ) {
 477                 $this->write( $string );
 478         }
 479
 480         function writeOpenPage( $page, $string ) {
 481                 $this->write( $string );
 482         }
 483
 484         function writeClosePage( $string ) {
 485                 $this->write( $string );
 486         }
 487
 488         function writeRevision( $rev, $string ) {
 489                 $this->write( $string );
 490         }
 491
 492         /**
 493          * Override to write to a different stream type.
 494          * @return bool
 495          */
 496         function write( $string ) {
 497                 print $string;
 498         }
 499 }
 500
 501 /**
 502  * Stream outputter to send data to a file.
 503  * @addtogroup Dump
 504  */
 505 class DumpFileOutput extends DumpOutput {
 506         var $handle;
 507
 508         function DumpFileOutput( $file ) {
 509                 $this->handle = fopen( $file, "wt" );
 510         }
 511
 512         function write( $string ) {
 513                 fputs( $this->handle, $string );
 514         }
 515 }
 516
 517 /**
 518  * Stream outputter to send data to a file via some filter program.
 519  * Even if compression is available in a library, using a separate
 520  * program can allow us to make use of a multi-processor system.
 521  * @addtogroup Dump
 522  */
 523 class DumpPipeOutput extends DumpFileOutput {
 524         function DumpPipeOutput( $command, $file = null ) {
 525                 if( !is_null( $file ) ) {
 526                         $command .=  " > " . wfEscapeShellArg( $file );
 527                 }
 528                 $this->handle = popen( $command, "w" );
 529         }
 530 }
 531
 532 /**
 533  * Sends dump output via the gzip compressor.
 534  * @addtogroup Dump
 535  */
 536 class DumpGZipOutput extends DumpPipeOutput {
 537         function DumpGZipOutput( $file ) {
 538                 parent::DumpPipeOutput( "gzip", $file );
 539         }
 540 }
 541
 542 /**
 543  * Sends dump output via the bgzip2 compressor.
 544  * @addtogroup Dump
 545  */
 546 class DumpBZip2Output extends DumpPipeOutput {
 547         function DumpBZip2Output( $file ) {
 548                 parent::DumpPipeOutput( "bzip2", $file );
 549         }
 550 }
 551
 552 /**
 553  * Sends dump output via the p7zip compressor.
 554  * @addtogroup Dump
 555  */
 556 class Dump7ZipOutput extends DumpPipeOutput {
 557         function Dump7ZipOutput( $file ) {
 558                 $command = "7za a -bd -si " . wfEscapeShellArg( $file );
 559                 // Suppress annoying useless crap from p7zip
 560                 // Unfortunately this could suppress real error messages too
 561                 $command .= ' >' . wfGetNull() . ' 2>&1';
 562                 parent::DumpPipeOutput( $command );
 563         }
 564 }
 565
 566
 567
 568 /**
 569  * Dump output filter class.
 570  * This just does output filtering and streaming; XML formatting is done
 571  * higher up, so be careful in what you do.
 572  * @addtogroup Dump
 573  */
 574 class DumpFilter {
 575         function DumpFilter( &$sink ) {
 576                 $this->sink =& $sink;
 577         }
 578
 579         function writeOpenStream( $string ) {
 580                 $this->sink->writeOpenStream( $string );
 581         }
 582
 583         function writeCloseStream( $string ) {
 584                 $this->sink->writeCloseStream( $string );
 585         }
 586
 587         function writeOpenPage( $page, $string ) {
 588                 $this->sendingThisPage = $this->pass( $page, $string );
 589                 if( $this->sendingThisPage ) {
 590                         $this->sink->writeOpenPage( $page, $string );
 591                 }
 592         }
 593
 594         function writeClosePage( $string ) {
 595                 if( $this->sendingThisPage ) {
 596                         $this->sink->writeClosePage( $string );
 597                         $this->sendingThisPage = false;
 598                 }
 599         }
 600
 601         function writeRevision( $rev, $string ) {
 602                 if( $this->sendingThisPage ) {
 603                         $this->sink->writeRevision( $rev, $string );
 604                 }
 605         }
 606
 607         /**
 608          * Override for page-based filter types.
 609          * @return bool
 610          */
 611         function pass( $page ) {
 612                 return true;
 613         }
 614 }
 615
 616 /**
 617  * Simple dump output filter to exclude all talk pages.
 618  * @addtogroup Dump
 619  */
 620 class DumpNotalkFilter extends DumpFilter {
 621         function pass( $page ) {
 622                 return !Namespace::isTalk( $page->page_namespace );
 623         }
 624 }
 625
 626 /**
 627  * Dump output filter to include or exclude pages in a given set of namespaces.
 628  * @addtogroup Dump
 629  */
 630 class DumpNamespaceFilter extends DumpFilter {
 631         var $invert = false;
 632         var $namespaces = array();
 633
 634         function DumpNamespaceFilter( &$sink, $param ) {
 635                 parent::DumpFilter( $sink );
 636
 637                 $constants = array(
 638                         "NS_MAIN"           => NS_MAIN,
 639                         "NS_TALK"           => NS_TALK,
 640                         "NS_USER"           => NS_USER,
 641                         "NS_USER_TALK"      => NS_USER_TALK,
 642                         "NS_PROJECT"        => NS_PROJECT,
 643                         "NS_PROJECT_TALK"   => NS_PROJECT_TALK,
 644                         "NS_IMAGE"          => NS_IMAGE,
 645                         "NS_IMAGE_TALK"     => NS_IMAGE_TALK,
 646                         "NS_MEDIAWIKI"      => NS_MEDIAWIKI,
 647                         "NS_MEDIAWIKI_TALK" => NS_MEDIAWIKI_TALK,
 648                         "NS_TEMPLATE"       => NS_TEMPLATE,
 649                         "NS_TEMPLATE_TALK"  => NS_TEMPLATE_TALK,
 650                         "NS_HELP"           => NS_HELP,
 651                         "NS_HELP_TALK"      => NS_HELP_TALK,
 652                         "NS_CATEGORY"       => NS_CATEGORY,
 653                         "NS_CATEGORY_TALK"  => NS_CATEGORY_TALK );
 654
 655                 if( $param{0} == '!' ) {
 656                         $this->invert = true;
 657                         $param = substr( $param, 1 );
 658                 }
 659
 660                 foreach( explode( ',', $param ) as $key ) {
 661                         $key = trim( $key );
 662                         if( isset( $constants[$key] ) ) {
 663                                 $ns = $constants[$key];
 664                                 $this->namespaces[$ns] = true;
 665                         } elseif( is_numeric( $key ) ) {
 666                                 $ns = intval( $key );
 667                                 $this->namespaces[$ns] = true;
 668                         } else {
 669                                 throw new MWException( "Unrecognized namespace key '$key'\n" );
 670                         }
 671                 }
 672         }
 673
 674         function pass( $page ) {
 675                 $match = isset( $this->namespaces[$page->page_namespace] );
 676                 return $this->invert xor $match;
 677         }
 678 }
 679
 680
 681 /**
 682  * Dump output filter to include only the last revision in each page sequence.
 683  * @addtogroup Dump
 684  */
 685 class DumpLatestFilter extends DumpFilter {
 686         var $page, $pageString, $rev, $revString;
 687
 688         function writeOpenPage( $page, $string ) {
 689                 $this->page = $page;
 690                 $this->pageString = $string;
 691         }
 692
 693         function writeClosePage( $string ) {
 694                 if( $this->rev ) {
 695                         $this->sink->writeOpenPage( $this->page, $this->pageString );
 696                         $this->sink->writeRevision( $this->rev, $this->revString );
 697                         $this->sink->writeClosePage( $string );
 698                 }
 699                 $this->rev = null;
 700                 $this->revString = null;
 701                 $this->page = null;
 702                 $this->pageString = null;
 703         }
 704
 705         function writeRevision( $rev, $string ) {
 706                 if( $rev->rev_id == $this->page->page_latest ) {
 707                         $this->rev = $rev;
 708                         $this->revString = $string;
 709                 }
 710         }
 711 }
 712
 713 /**
 714  * Base class for output stream; prints to stdout or buffer or whereever.
 715  * @addtogroup Dump
 716  */
 717 class DumpMultiWriter {
 718         function DumpMultiWriter( $sinks ) {
 719                 $this->sinks = $sinks;
 720                 $this->count = count( $sinks );
 721         }
 722
 723         function writeOpenStream( $string ) {
 724                 for( $i = 0; $i < $this->count; $i++ ) {
 725                         $this->sinks[$i]->writeOpenStream( $string );
 726                 }
 727         }
 728
 729         function writeCloseStream( $string ) {
 730                 for( $i = 0; $i < $this->count; $i++ ) {
 731                         $this->sinks[$i]->writeCloseStream( $string );
 732                 }
 733         }
 734
 735         function writeOpenPage( $page, $string ) {
 736                 for( $i = 0; $i < $this->count; $i++ ) {
 737                         $this->sinks[$i]->writeOpenPage( $page, $string );
 738                 }
 739         }
 740
 741         function writeClosePage( $string ) {
 742                 for( $i = 0; $i < $this->count; $i++ ) {
 743                         $this->sinks[$i]->writeClosePage( $string );
 744                 }
 745         }
 746
 747         function writeRevision( $rev, $string ) {
 748                 for( $i = 0; $i < $this->count; $i++ ) {
 749                         $this->sinks[$i]->writeRevision( $rev, $string );
 750                 }
 751         }
 752 }
 753
 754 function xmlsafe( $string ) {
 755         $fname = 'xmlsafe';
 756         wfProfileIn( $fname );
 757
 758         /**
 759          * The page may contain old data which has not been properly normalized.
 760          * Invalid UTF-8 sequences or forbidden control characters will make our
 761          * XML output invalid, so be sure to strip them out.
 762          */
 763         $string = UtfNormal::cleanUp( $string );
 764
 765         $string = htmlspecialchars( $string );
 766         wfProfileOut( $fname );
 767         return $string;
 768 }
 769
 770