includes/Import.php

   1 <?php
   2 /**
   3  * MediaWiki page data importer
   4  * Copyright (C) 2003,2005 Brion Vibber <brion@pobox.com>
   5  * http://www.mediawiki.org/
   6  *
   7  * This program is free software; you can redistribute it and/or modify
   8  * it under the terms of the GNU General Public License as published by
   9  * the Free Software Foundation; either version 2 of the License, or
  10  * (at your option) any later version.
  11  *
  12  * This program is distributed in the hope that it will be useful,
  13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  15  * GNU General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU General Public License along
  18  * with this program; if not, write to the Free Software Foundation, Inc.,
  19  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  20  * http://www.gnu.org/copyleft/gpl.html
  21  *
  22  * @file
  23  * @ingroup SpecialPage
  24  */
  25
  26 /**
  27  *
  28  * @ingroup SpecialPage
  29  */
  30 class WikiRevision {
  31         var $title = null;
  32         var $id = 0;
  33         var $timestamp = "20010115000000";
  34         var $user = 0;
  35         var $user_text = "";
  36         var $text = "";
  37         var $comment = "";
  38         var $minor = false;
  39         var $type = "";
  40         var $action = "";
  41         var $params = "";
  42
  43         function setTitle( $title ) {
  44                 if( is_object( $title ) ) {
  45                         $this->title = $title;
  46                 } elseif( is_null( $title ) ) {
  47                         throw new MWException( "WikiRevision given a null title in import. You may need to adjust \$wgLegalTitleChars." );
  48                 } else {
  49                         throw new MWException( "WikiRevision given non-object title in import." );
  50                 }
  51         }
  52
  53         function setID( $id ) {
  54                 $this->id = $id;
  55         }
  56
  57         function setTimestamp( $ts ) {
  58                 # 2003-08-05T18:30:02Z
  59                 $this->timestamp = wfTimestamp( TS_MW, $ts );
  60         }
  61
  62         function setUsername( $user ) {
  63                 $this->user_text = $user;
  64         }
  65
  66         function setUserIP( $ip ) {
  67                 $this->user_text = $ip;
  68         }
  69
  70         function setText( $text ) {
  71                 $this->text = $text;
  72         }
  73
  74         function setComment( $text ) {
  75                 $this->comment = $text;
  76         }
  77
  78         function setMinor( $minor ) {
  79                 $this->minor = (bool)$minor;
  80         }
  81
  82         function setSrc( $src ) {
  83                 $this->src = $src;
  84         }
  85
  86         function setFilename( $filename ) {
  87                 $this->filename = $filename;
  88         }
  89
  90         function setSize( $size ) {
  91                 $this->size = intval( $size );
  92         }
  93
  94         function setType( $type ) {
  95                 $this->type = $type;
  96         }
  97
  98         function setAction( $action ) {
  99                 $this->action = $action;
 100         }
 101
 102         function setParams( $params ) {
 103                 $this->params = $params;
 104         }
 105
 106         function getTitle() {
 107                 return $this->title;
 108         }
 109
 110         function getID() {
 111                 return $this->id;
 112         }
 113
 114         function getTimestamp() {
 115                 return $this->timestamp;
 116         }
 117
 118         function getUser() {
 119                 return $this->user_text;
 120         }
 121
 122         function getText() {
 123                 return $this->text;
 124         }
 125
 126         function getComment() {
 127                 return $this->comment;
 128         }
 129
 130         function getMinor() {
 131                 return $this->minor;
 132         }
 133
 134         function getSrc() {
 135                 return $this->src;
 136         }
 137
 138         function getFilename() {
 139                 return $this->filename;
 140         }
 141
 142         function getSize() {
 143                 return $this->size;
 144         }
 145
 146         function getType() {
 147                 return $this->type;
 148         }
 149
 150         function getAction() {
 151                 return $this->action;
 152         }
 153
 154         function getParams() {
 155                 return $this->params;
 156         }
 157
 158         function importOldRevision() {
 159                 $dbw = wfGetDB( DB_MASTER );
 160
 161                 # Sneak a single revision into place
 162                 $user = User::newFromName( $this->getUser() );
 163                 if( $user ) {
 164                         $userId = intval( $user->getId() );
 165                         $userText = $user->getName();
 166                 } else {
 167                         $userId = 0;
 168                         $userText = $this->getUser();
 169                 }
 170
 171                 // avoid memory leak...?
 172                 $linkCache = LinkCache::singleton();
 173                 $linkCache->clear();
 174
 175                 $article = new Article( $this->title );
 176                 $pageId = $article->getId();
 177                 if( $pageId == 0 ) {
 178                         # must create the page...
 179                         $pageId = $article->insertOn( $dbw );
 180                         $created = true;
 181                 } else {
 182                         $created = false;
 183
 184                         $prior = $dbw->selectField( 'revision', '1',
 185                                 array( 'rev_page' => $pageId,
 186                                         'rev_timestamp' => $dbw->timestamp( $this->timestamp ),
 187                                         'rev_user_text' => $userText,
 188                                         'rev_comment'   => $this->getComment() ),
 189                                 __METHOD__
 190                         );
 191                         if( $prior ) {
 192                                 // FIXME: this could fail slightly for multiple matches :P
 193                                 wfDebug( __METHOD__ . ": skipping existing revision for [[" .
 194                                         $this->title->getPrefixedText() . "]], timestamp " . $this->timestamp . "\n" );
 195                                 return false;
 196                         }
 197                 }
 198
 199                 # FIXME: Use original rev_id optionally (better for backups)
 200                 # Insert the row
 201                 $revision = new Revision( array(
 202                         'page'       => $pageId,
 203                         'text'       => $this->getText(),
 204                         'comment'    => $this->getComment(),
 205                         'user'       => $userId,
 206                         'user_text'  => $userText,
 207                         'timestamp'  => $this->timestamp,
 208                         'minor_edit' => $this->minor,
 209                         ) );
 210                 $revId = $revision->insertOn( $dbw );
 211                 $changed = $article->updateIfNewerOn( $dbw, $revision );
 212
 213                 if( $created ) {
 214                         wfDebug( __METHOD__ . ": running onArticleCreate\n" );
 215                         Article::onArticleCreate( $this->title );
 216
 217                         wfDebug( __METHOD__ . ": running create updates\n" );
 218                         $article->createUpdates( $revision );
 219
 220                 } elseif( $changed ) {
 221                         wfDebug( __METHOD__ . ": running onArticleEdit\n" );
 222                         Article::onArticleEdit( $this->title );
 223
 224                         wfDebug( __METHOD__ . ": running edit updates\n" );
 225                         $article->editUpdates(
 226                                 $this->getText(),
 227                                 $this->getComment(),
 228                                 $this->minor,
 229                                 $this->timestamp,
 230                                 $revId );
 231                 }
 232
 233                 return true;
 234         }
 235
 236         function importLogItem() {
 237                 $dbw = wfGetDB( DB_MASTER );
 238                 # FIXME: this will not record autoblocks
 239                 if( !$this->getTitle() ) {
 240                         wfDebug( __METHOD__ . ": skipping invalid {$this->type}/{$this->action} log time, timestamp " .
 241                                 $this->timestamp . "\n" );
 242                         return;
 243                 }
 244                 # Check if it exists already
 245                 // FIXME: use original log ID (better for backups)
 246                 $prior = $dbw->selectField( 'logging', '1',
 247                         array( 'log_type' => $this->getType(),
 248                                 'log_action'    => $this->getAction(),
 249                                 'log_timestamp' => $dbw->timestamp( $this->timestamp ),
 250                                 #'log_user_text' => $this->user_text,
 251                                 'log_params'    => $this->params ),
 252                         __METHOD__
 253                 );
 254                 // FIXME: this could fail slightly for multiple matches :P
 255                 if( $prior ) {
 256                         wfDebug( __METHOD__ . ": skipping existing item for Log:{$this->type}/{$this->action}, timestamp " .
 257                                 $this->timestamp . "\n" );
 258                         return false;
 259                 }
 260                 $log_id = $dbw->nextSequenceValue( 'log_log_id_seq' );
 261                 $data = array(
 262                         'log_id' => $log_id,
 263                         'log_type' => $this->type,
 264                         'log_action' => $this->action,
 265                         'log_timestamp' => $dbw->timestamp( $this->timestamp ),
 266                         'log_user' => User::idFromName( $this->user_text ),
 267                         #'log_user_text' => $this->user_text,
 268                         'log_namespace' => $this->getTitle()->getNamespace(),
 269                         'log_title' => $this->getTitle()->getDBkey(),
 270                         'log_comment' => $this->getComment(),
 271                         'log_params' => $this->params
 272                 );
 273                 $dbw->insert( 'logging', $data, __METHOD__ );
 274         }
 275
 276         function importUpload() {
 277                 wfDebug( __METHOD__ . ": STUB\n" );
 278
 279                 /**
 280                         // from file revert...
 281                         $source = $this->file->getArchiveVirtualUrl( $this->oldimage );
 282                         $comment = $wgRequest->getText( 'wpComment' );
 283                         // TODO: Preserve file properties from database instead of reloading from file
 284                         $status = $this->file->upload( $source, $comment, $comment );
 285                         if( $status->isGood() ) {
 286                 */
 287
 288                 /**
 289                         // from file upload...
 290                 $this->mLocalFile = wfLocalFile( $nt );
 291                 $this->mDestName = $this->mLocalFile->getName();
 292                 //....
 293                         $status = $this->mLocalFile->upload( $this->mTempPath, $this->mComment, $pageText,
 294                         File::DELETE_SOURCE, $this->mFileProps );
 295                         if ( !$status->isGood() ) {
 296                                 $resultDetails = array( 'internal' => $status->getWikiText() );
 297                 */
 298
 299                 // @fixme upload() uses $wgUser, which is wrong here
 300                 // it may also create a page without our desire, also wrong potentially.
 301                 // and, it will record a *current* upload, but we might want an archive version here
 302
 303                 $file = wfLocalFile( $this->getTitle() );
 304                 if( !$file ) {
 305                         var_dump( $file );
 306                         wfDebug( "IMPORT: Bad file. :(\n" );
 307                         return false;
 308                 }
 309
 310                 $source = $this->downloadSource();
 311                 if( !$source ) {
 312                         wfDebug( "IMPORT: Could not fetch remote file. :(\n" );
 313                         return false;
 314                 }
 315
 316                 $status = $file->upload( $source,
 317                         $this->getComment(),
 318                         $this->getComment(), // Initial page, if none present...
 319                         File::DELETE_SOURCE,
 320                         false, // props...
 321                         $this->getTimestamp() );
 322
 323                 if( $status->isGood() ) {
 324                         // yay?
 325                         wfDebug( "IMPORT: is ok?\n" );
 326                         return true;
 327                 }
 328
 329                 wfDebug( "IMPORT: is bad? " . $status->getXml() . "\n" );
 330                 return false;
 331
 332         }
 333
 334         function downloadSource() {
 335                 global $wgEnableUploads;
 336                 if( !$wgEnableUploads ) {
 337                         return false;
 338                 }
 339
 340                 $tempo = tempnam( wfTempDir(), 'download' );
 341                 $f = fopen( $tempo, 'wb' );
 342                 if( !$f ) {
 343                         wfDebug( "IMPORT: couldn't write to temp file $tempo\n" );
 344                         return false;
 345                 }
 346
 347                 // @fixme!
 348                 $src = $this->getSrc();
 349                 $data = Http::get( $src );
 350                 if( !$data ) {
 351                         wfDebug( "IMPORT: couldn't fetch source $src\n" );
 352                         fclose( $f );
 353                         unlink( $tempo );
 354                         return false;
 355                 }
 356
 357                 fwrite( $f, $data );
 358                 fclose( $f );
 359
 360                 return $tempo;
 361         }
 362
 363 }
 364
 365 /**
 366  * implements Special:Import
 367  * @ingroup SpecialPage
 368  */
 369 class WikiImporter {
 370         var $mDebug = false;
 371         var $mSource = null;
 372         var $mPageCallback = null;
 373         var $mPageOutCallback = null;
 374         var $mRevisionCallback = null;
 375         var $mLogItemCallback = null;
 376         var $mUploadCallback = null;
 377         var $mTargetNamespace = null;
 378         var $lastfield;
 379         var $tagStack = array();
 380
 381         function __construct( $source ) {
 382                 $this->setRevisionCallback( array( $this, "importRevision" ) );
 383                 $this->setUploadCallback( array( $this, "importUpload" ) );
 384                 $this->setLogItemCallback( array( $this, "importLogItem" ) );
 385                 $this->mSource = $source;
 386         }
 387
 388         function throwXmlError( $err ) {
 389                 $this->debug( "FAILURE: $err" );
 390                 wfDebug( "WikiImporter XML error: $err\n" );
 391         }
 392
 393         # --------------
 394
 395         function doImport() {
 396                 if( empty( $this->mSource ) ) {
 397                         return new WikiErrorMsg( "importnotext" );
 398                 }
 399
 400                 $parser = xml_parser_create( "UTF-8" );
 401
 402                 # case folding violates XML standard, turn it off
 403                 xml_parser_set_option( $parser, XML_OPTION_CASE_FOLDING, false );
 404
 405                 xml_set_object( $parser, $this );
 406                 xml_set_element_handler( $parser, "in_start", "" );
 407
 408                 $offset = 0; // for context extraction on error reporting
 409                 do {
 410                         $chunk = $this->mSource->readChunk();
 411                         if( !xml_parse( $parser, $chunk, $this->mSource->atEnd() ) ) {
 412                                 wfDebug( "WikiImporter::doImport encountered XML parsing error\n" );
 413                                 return new WikiXmlError( $parser, wfMsgHtml( 'import-parse-failure' ), $chunk, $offset );
 414                         }
 415                         $offset += strlen( $chunk );
 416                 } while( $chunk !== false && !$this->mSource->atEnd() );
 417                 xml_parser_free( $parser );
 418
 419                 return true;
 420         }
 421
 422         function debug( $data ) {
 423                 if( $this->mDebug ) {
 424                         wfDebug( "IMPORT: $data\n" );
 425                 }
 426         }
 427
 428         function notice( $data ) {
 429                 global $wgCommandLineMode;
 430                 if( $wgCommandLineMode ) {
 431                         print "$data\n";
 432                 } else {
 433                         global $wgOut;
 434                         $wgOut->addHTML( "<li>" . htmlspecialchars( $data ) . "</li>\n" );
 435                 }
 436         }
 437
 438         /**
 439          * Set debug mode...
 440          */
 441         function setDebug( $debug ) {
 442                 $this->mDebug = $debug;
 443         }
 444
 445         /**
 446          * Sets the action to perform as each new page in the stream is reached.
 447          * @param $callback callback
 448          * @return callback
 449          */
 450         function setPageCallback( $callback ) {
 451                 $previous = $this->mPageCallback;
 452                 $this->mPageCallback = $callback;
 453                 return $previous;
 454         }
 455
 456         /**
 457          * Sets the action to perform as each page in the stream is completed.
 458          * Callback accepts the page title (as a Title object), a second object
 459          * with the original title form (in case it's been overridden into a
 460          * local namespace), and a count of revisions.
 461          *
 462          * @param $callback callback
 463          * @return callback
 464          */
 465         function setPageOutCallback( $callback ) {
 466                 $previous = $this->mPageOutCallback;
 467                 $this->mPageOutCallback = $callback;
 468                 return $previous;
 469         }
 470
 471         /**
 472          * Sets the action to perform as each page revision is reached.
 473          * @param $callback callback
 474          * @return callback
 475          */
 476         function setRevisionCallback( $callback ) {
 477                 $previous = $this->mRevisionCallback;
 478                 $this->mRevisionCallback = $callback;
 479                 return $previous;
 480         }
 481
 482         /**
 483          * Sets the action to perform as each file upload version is reached.
 484          * @param $callback callback
 485          * @return callback
 486          */
 487         function setUploadCallback( $callback ) {
 488                 $previous = $this->mUploadCallback;
 489                 $this->mUploadCallback = $callback;
 490                 return $previous;
 491         }
 492
 493         /**
 494          * Sets the action to perform as each log item reached.
 495          * @param $callback callback
 496          * @return callback
 497          */
 498         function setLogItemCallback( $callback ) {
 499                 $previous = $this->mLogItemCallback;
 500                 $this->mLogItemCallback = $callback;
 501                 return $previous;
 502         }
 503
 504         /**
 505          * Set a target namespace to override the defaults
 506          */
 507         function setTargetNamespace( $namespace ) {
 508                 if( is_null( $namespace ) ) {
 509                         // Don't override namespaces
 510                         $this->mTargetNamespace = null;
 511                 } elseif( $namespace >= 0 ) {
 512                         // FIXME: Check for validity
 513                         $this->mTargetNamespace = intval( $namespace );
 514                 } else {
 515                         return false;
 516                 }
 517         }
 518
 519         /**
 520          * Default per-revision callback, performs the import.
 521          * @param $revision WikiRevision
 522          * @private
 523          */
 524         function importRevision( $revision ) {
 525                 $dbw = wfGetDB( DB_MASTER );
 526                 return $dbw->deadlockLoop( array( $revision, 'importOldRevision' ) );
 527         }
 528
 529         /**
 530          * Default per-revision callback, performs the import.
 531          * @param $revision WikiRevision
 532          * @private
 533          */
 534         function importLogItem( $rev ) {
 535                 $dbw = wfGetDB( DB_MASTER );
 536                 return $dbw->deadlockLoop( array( $rev, 'importLogItem' ) );
 537         }
 538
 539         /**
 540          * Dummy for now...
 541          */
 542         function importUpload( $revision ) {
 543                 //$dbw = wfGetDB( DB_MASTER );
 544                 //return $dbw->deadlockLoop( array( $revision, 'importUpload' ) );
 545                 return false;
 546         }
 547
 548         /**
 549          * Alternate per-revision callback, for debugging.
 550          * @param $revision WikiRevision
 551          * @private
 552          */
 553         function debugRevisionHandler( &$revision ) {
 554                 $this->debug( "Got revision:" );
 555                 if( is_object( $revision->title ) ) {
 556                         $this->debug( "-- Title: " . $revision->title->getPrefixedText() );
 557                 } else {
 558                         $this->debug( "-- Title: <invalid>" );
 559                 }
 560                 $this->debug( "-- User: " . $revision->user_text );
 561                 $this->debug( "-- Timestamp: " . $revision->timestamp );
 562                 $this->debug( "-- Comment: " . $revision->comment );
 563                 $this->debug( "-- Text: " . $revision->text );
 564         }
 565
 566         /**
 567          * Notify the callback function when a new <page> is reached.
 568          * @param $title Title
 569          * @private
 570          */
 571         function pageCallback( $title ) {
 572                 if( is_callable( $this->mPageCallback ) ) {
 573                         call_user_func( $this->mPageCallback, $title );
 574                 }
 575         }
 576
 577         /**
 578          * Notify the callback function when a </page> is closed.
 579          * @param $title Title
 580          * @param $origTitle Title
 581          * @param $revisionCount int
 582          * @param $successCount Int: number of revisions for which callback returned true
 583          * @private
 584          */
 585         function pageOutCallback( $title, $origTitle, $revisionCount, $successCount ) {
 586                 if( is_callable( $this->mPageOutCallback ) ) {
 587                         call_user_func( $this->mPageOutCallback, $title, $origTitle,
 588                                 $revisionCount, $successCount );
 589                 }
 590         }
 591
 592         # XML parser callbacks from here out -- beware!
 593         function donothing( $parser, $x, $y="" ) {
 594                 #$this->debug( "donothing" );
 595         }
 596
 597         function in_start( $parser, $name, $attribs ) {
 598                 $this->debug( "in_start $name" );
 599                 if( $name != "mediawiki" ) {
 600                         return $this->throwXMLerror( "Expected <mediawiki>, got <$name>" );
 601                 }
 602                 xml_set_element_handler( $parser, "in_mediawiki", "out_mediawiki" );
 603         }
 604
 605         function in_mediawiki( $parser, $name, $attribs ) {
 606                 $this->debug( "in_mediawiki $name" );
 607                 if( $name == 'siteinfo' ) {
 608                         xml_set_element_handler( $parser, "in_siteinfo", "out_siteinfo" );
 609                 } elseif( $name == 'page' ) {
 610                         $this->push( $name );
 611                         $this->workRevisionCount = 0;
 612                         $this->workSuccessCount = 0;
 613                         $this->uploadCount = 0;
 614                         $this->uploadSuccessCount = 0;
 615                         xml_set_element_handler( $parser, "in_page", "out_page" );
 616                 } elseif( $name == 'logitem' ) {
 617                         $this->push( $name );
 618                         $this->workRevision = new WikiRevision;
 619                         xml_set_element_handler( $parser, "in_logitem", "out_logitem" );
 620                 } else {
 621                         return $this->throwXMLerror( "Expected <page>, got <$name>" );
 622                 }
 623         }
 624         function out_mediawiki( $parser, $name ) {
 625                 $this->debug( "out_mediawiki $name" );
 626                 if( $name != "mediawiki" ) {
 627                         return $this->throwXMLerror( "Expected </mediawiki>, got </$name>" );
 628                 }
 629                 xml_set_element_handler( $parser, "donothing", "donothing" );
 630         }
 631
 632
 633         function in_siteinfo( $parser, $name, $attribs ) {
 634                 // no-ops for now
 635                 $this->debug( "in_siteinfo $name" );
 636                 switch( $name ) {
 637                 case "sitename":
 638                 case "base":
 639                 case "generator":
 640                 case "case":
 641                 case "namespaces":
 642                 case "namespace":
 643                         break;
 644                 default:
 645                         return $this->throwXMLerror( "Element <$name> not allowed in <siteinfo>." );
 646                 }
 647         }
 648
 649         function out_siteinfo( $parser, $name ) {
 650                 if( $name == "siteinfo" ) {
 651                         xml_set_element_handler( $parser, "in_mediawiki", "out_mediawiki" );
 652                 }
 653         }
 654
 655
 656         function in_page( $parser, $name, $attribs ) {
 657                 $this->debug( "in_page $name" );
 658                 switch( $name ) {
 659                 case "id":
 660                 case "title":
 661                 case "restrictions":
 662                         $this->appendfield = $name;
 663                         $this->appenddata = "";
 664                         xml_set_element_handler( $parser, "in_nothing", "out_append" );
 665                         xml_set_character_data_handler( $parser, "char_append" );
 666                         break;
 667                 case "revision":
 668                         $this->push( "revision" );
 669                         if( is_object( $this->pageTitle ) ) {
 670                                 $this->workRevision = new WikiRevision;
 671                                 $this->workRevision->setTitle( $this->pageTitle );
 672                                 $this->workRevisionCount++;
 673                         } else {
 674                                 // Skipping items due to invalid page title
 675                                 $this->workRevision = null;
 676                         }
 677                         xml_set_element_handler( $parser, "in_revision", "out_revision" );
 678                         break;
 679                 case "upload":
 680                         $this->push( "upload" );
 681                         if( is_object( $this->pageTitle ) ) {
 682                                 $this->workRevision = new WikiRevision;
 683                                 $this->workRevision->setTitle( $this->pageTitle );
 684                                 $this->uploadCount++;
 685                         } else {
 686                                 // Skipping items due to invalid page title
 687                                 $this->workRevision = null;
 688                         }
 689                         xml_set_element_handler( $parser, "in_upload", "out_upload" );
 690                         break;
 691                 default:
 692                         return $this->throwXMLerror( "Element <$name> not allowed in a <page>." );
 693                 }
 694         }
 695
 696         function out_page( $parser, $name ) {
 697                 $this->debug( "out_page $name" );
 698                 $this->pop();
 699                 if( $name != "page" ) {
 700                         return $this->throwXMLerror( "Expected </page>, got </$name>" );
 701                 }
 702                 xml_set_element_handler( $parser, "in_mediawiki", "out_mediawiki" );
 703
 704                 $this->pageOutCallback( $this->pageTitle, $this->origTitle,
 705                         $this->workRevisionCount, $this->workSuccessCount );
 706
 707                 $this->workTitle = null;
 708                 $this->workRevision = null;
 709                 $this->workRevisionCount = 0;
 710                 $this->workSuccessCount = 0;
 711                 $this->pageTitle = null;
 712                 $this->origTitle = null;
 713         }
 714
 715         function in_nothing( $parser, $name, $attribs ) {
 716                 $this->debug( "in_nothing $name" );
 717                 return $this->throwXMLerror( "No child elements allowed here; got <$name>" );
 718         }
 719
 720         function char_append( $parser, $data ) {
 721                 $this->debug( "char_append '$data'" );
 722                 $this->appenddata .= $data;
 723         }
 724
 725         function out_append( $parser, $name ) {
 726                 $this->debug( "out_append $name" );
 727                 if( $name != $this->appendfield ) {
 728                         return $this->throwXMLerror( "Expected </{$this->appendfield}>, got </$name>" );
 729                 }
 730
 731                 switch( $this->appendfield ) {
 732                 case "title":
 733                         $this->workTitle = $this->appenddata;
 734                         $this->origTitle = Title::newFromText( $this->workTitle );
 735                         if( !is_null( $this->mTargetNamespace ) && !is_null( $this->origTitle ) ) {
 736                                 $this->pageTitle = Title::makeTitle( $this->mTargetNamespace,
 737                                         $this->origTitle->getDBkey() );
 738                         } else {
 739                                 $this->pageTitle = Title::newFromText( $this->workTitle );
 740                         }
 741                         if( is_null( $this->pageTitle ) ) {
 742                                 // Invalid page title? Ignore the page
 743                                 $this->notice( "Skipping invalid page title '$this->workTitle'" );
 744                         } else {
 745                                 $this->pageCallback( $this->workTitle );
 746                         }
 747                         break;
 748                 case "id":
 749                         if ( $this->parentTag() == 'revision' || $this->parentTag() == 'logitem' ) {
 750                                 if( $this->workRevision )
 751                                         $this->workRevision->setID( $this->appenddata );
 752                         }
 753                         break;
 754                 case "text":
 755                         if( $this->workRevision )
 756                                 $this->workRevision->setText( $this->appenddata );
 757                         break;
 758                 case "username":
 759                         if( $this->workRevision )
 760                                 $this->workRevision->setUsername( $this->appenddata );
 761                         break;
 762                 case "ip":
 763                         if( $this->workRevision )
 764                                 $this->workRevision->setUserIP( $this->appenddata );
 765                         break;
 766                 case "timestamp":
 767                         if( $this->workRevision )
 768                                 $this->workRevision->setTimestamp( $this->appenddata );
 769                         break;
 770                 case "comment":
 771                         if( $this->workRevision )
 772                                 $this->workRevision->setComment( $this->appenddata );
 773                         break;
 774                 case "type":
 775                         if( $this->workRevision )
 776                                 $this->workRevision->setType( $this->appenddata );
 777                         break;
 778                 case "action":
 779                         if( $this->workRevision )
 780                                 $this->workRevision->setAction( $this->appenddata );
 781                         break;
 782                 case "logtitle":
 783                         if( $this->workRevision )
 784                                 $this->workRevision->setTitle( Title::newFromText( $this->appenddata ) );
 785                         break;
 786                 case "params":
 787                         if( $this->workRevision )
 788                                 $this->workRevision->setParams( $this->appenddata );
 789                         break;
 790                 case "minor":
 791                         if( $this->workRevision )
 792                                 $this->workRevision->setMinor( true );
 793                         break;
 794                 case "filename":
 795                         if( $this->workRevision )
 796                                 $this->workRevision->setFilename( $this->appenddata );
 797                         break;
 798                 case "src":
 799                         if( $this->workRevision )
 800                                 $this->workRevision->setSrc( $this->appenddata );
 801                         break;
 802                 case "size":
 803                         if( $this->workRevision )
 804                                 $this->workRevision->setSize( intval( $this->appenddata ) );
 805                         break;
 806                 default:
 807                         $this->debug( "Bad append: {$this->appendfield}" );
 808                 }
 809                 $this->appendfield = "";
 810                 $this->appenddata = "";
 811
 812                 $parent = $this->parentTag();
 813                 xml_set_element_handler( $parser, "in_$parent", "out_$parent" );
 814                 xml_set_character_data_handler( $parser, "donothing" );
 815         }
 816
 817         function in_revision( $parser, $name, $attribs ) {
 818                 $this->debug( "in_revision $name" );
 819                 switch( $name ) {
 820                 case "id":
 821                 case "timestamp":
 822                 case "comment":
 823                 case "minor":
 824                 case "text":
 825                         $this->appendfield = $name;
 826                         xml_set_element_handler( $parser, "in_nothing", "out_append" );
 827                         xml_set_character_data_handler( $parser, "char_append" );
 828                         break;
 829                 case "contributor":
 830                         $this->push( "contributor" );
 831                         xml_set_element_handler( $parser, "in_contributor", "out_contributor" );
 832                         break;
 833                 default:
 834                         return $this->throwXMLerror( "Element <$name> not allowed in a <revision>." );
 835                 }
 836         }
 837
 838         function out_revision( $parser, $name ) {
 839                 $this->debug( "out_revision $name" );
 840                 $this->pop();
 841                 if( $name != "revision" ) {
 842                         return $this->throwXMLerror( "Expected </revision>, got </$name>" );
 843                 }
 844                 xml_set_element_handler( $parser, "in_page", "out_page" );
 845
 846                 if( $this->workRevision ) {
 847                         $ok = call_user_func_array( $this->mRevisionCallback,
 848                                 array( $this->workRevision, $this ) );
 849                         if( $ok ) {
 850                                 $this->workSuccessCount++;
 851                         }
 852                 }
 853         }
 854
 855         function in_logitem( $parser, $name, $attribs ) {
 856                 $this->debug( "in_logitem $name" );
 857                 switch( $name ) {
 858                 case "id":
 859                 case "timestamp":
 860                 case "comment":
 861                 case "type":
 862                 case "action":
 863                 case "logtitle":
 864                 case "params":
 865                         $this->appendfield = $name;
 866                         xml_set_element_handler( $parser, "in_nothing", "out_append" );
 867                         xml_set_character_data_handler( $parser, "char_append" );
 868                         break;
 869                 case "contributor":
 870                         $this->push( "contributor" );
 871                         xml_set_element_handler( $parser, "in_contributor", "out_contributor" );
 872                         break;
 873                 default:
 874                         return $this->throwXMLerror( "Element <$name> not allowed in a <revision>." );
 875                 }
 876         }
 877
 878         function out_logitem( $parser, $name ) {
 879                 $this->debug( "out_logitem $name" );
 880                 $this->pop();
 881                 if( $name != "logitem" ) {
 882                         return $this->throwXMLerror( "Expected </logitem>, got </$name>" );
 883                 }
 884                 xml_set_element_handler( $parser, "in_mediawiki", "out_mediawiki" );
 885
 886                 if( $this->workRevision ) {
 887                         $ok = call_user_func_array( $this->mLogItemCallback,
 888                                 array( $this->workRevision, $this ) );
 889                         if( $ok ) {
 890                                 $this->workSuccessCount++;
 891                         }
 892                 }
 893         }
 894
 895         function in_upload( $parser, $name, $attribs ) {
 896                 $this->debug( "in_upload $name" );
 897                 switch( $name ) {
 898                 case "timestamp":
 899                 case "comment":
 900                 case "text":
 901                 case "filename":
 902                 case "src":
 903                 case "size":
 904                         $this->appendfield = $name;
 905                         xml_set_element_handler( $parser, "in_nothing", "out_append" );
 906                         xml_set_character_data_handler( $parser, "char_append" );
 907                         break;
 908                 case "contributor":
 909                         $this->push( "contributor" );
 910                         xml_set_element_handler( $parser, "in_contributor", "out_contributor" );
 911                         break;
 912                 default:
 913                         return $this->throwXMLerror( "Element <$name> not allowed in an <upload>." );
 914                 }
 915         }
 916
 917         function out_upload( $parser, $name ) {
 918                 $this->debug( "out_revision $name" );
 919                 $this->pop();
 920                 if( $name != "upload" ) {
 921                         return $this->throwXMLerror( "Expected </upload>, got </$name>" );
 922                 }
 923                 xml_set_element_handler( $parser, "in_page", "out_page" );
 924
 925                 if( $this->workRevision ) {
 926                         $ok = call_user_func_array( $this->mUploadCallback,
 927                                 array( $this->workRevision, $this ) );
 928                         if( $ok ) {
 929                                 $this->workUploadSuccessCount++;
 930                         }
 931                 }
 932         }
 933
 934         function in_contributor( $parser, $name, $attribs ) {
 935                 $this->debug( "in_contributor $name" );
 936                 switch( $name ) {
 937                 case "username":
 938                 case "ip":
 939                 case "id":
 940                         $this->appendfield = $name;
 941                         xml_set_element_handler( $parser, "in_nothing", "out_append" );
 942                         xml_set_character_data_handler( $parser, "char_append" );
 943                         break;
 944                 default:
 945                         $this->throwXMLerror( "Invalid tag <$name> in <contributor>" );
 946                 }
 947         }
 948
 949         function out_contributor( $parser, $name ) {
 950                 $this->debug( "out_contributor $name" );
 951                 $this->pop();
 952                 if( $name != "contributor" ) {
 953                         return $this->throwXMLerror( "Expected </contributor>, got </$name>" );
 954                 }
 955                 $parent = $this->parentTag();
 956                 xml_set_element_handler( $parser, "in_$parent", "out_$parent" );
 957         }
 958
 959         private function push( $name ) {
 960                 array_push( $this->tagStack, $name );
 961                 $this->debug( "PUSH $name" );
 962         }
 963
 964         private function pop() {
 965                 $name = array_pop( $this->tagStack );
 966                 $this->debug( "POP $name" );
 967                 return $name;
 968         }
 969
 970         private function parentTag() {
 971                 $name = $this->tagStack[count( $this->tagStack ) - 1];
 972                 $this->debug( "PARENT $name" );
 973                 return $name;
 974         }
 975
 976 }
 977
 978 /**
 979  * @todo document (e.g. one-sentence class description).
 980  * @ingroup SpecialPage
 981  */
 982 class ImportStringSource {
 983         function __construct( $string ) {
 984                 $this->mString = $string;
 985                 $this->mRead = false;
 986         }
 987
 988         function atEnd() {
 989                 return $this->mRead;
 990         }
 991
 992         function readChunk() {
 993                 if( $this->atEnd() ) {
 994                         return false;
 995                 } else {
 996                         $this->mRead = true;
 997                         return $this->mString;
 998                 }
 999         }
1000 }
1001
1002 /**
1003  * @todo document (e.g. one-sentence class description).
1004  * @ingroup SpecialPage
1005  */
1006 class ImportStreamSource {
1007         function __construct( $handle ) {
1008                 $this->mHandle = $handle;
1009         }
1010
1011         function atEnd() {
1012                 return feof( $this->mHandle );
1013         }
1014
1015         function readChunk() {
1016                 return fread( $this->mHandle, 32768 );
1017         }
1018
1019         static function newFromFile( $filename ) {
1020                 $file = @fopen( $filename, 'rt' );
1021                 if( !$file ) {
1022                         return new WikiErrorMsg( "importcantopen" );
1023                 }
1024                 return new ImportStreamSource( $file );
1025         }
1026
1027         static function newFromUpload( $fieldname = "xmlimport" ) {
1028                 $upload =& $_FILES[$fieldname];
1029
1030                 if( !isset( $upload ) || !$upload['name'] ) {
1031                         return new WikiErrorMsg( 'importnofile' );
1032                 }
1033                 if( !empty( $upload['error'] ) ) {
1034                         switch($upload['error']){
1035                                 case 1: # The uploaded file exceeds the upload_max_filesize directive in php.ini.
1036                                         return new WikiErrorMsg( 'importuploaderrorsize' );
1037                                 case 2: # The uploaded file exceeds the MAX_FILE_SIZE directive that was specified in the HTML form.
1038                                         return new WikiErrorMsg( 'importuploaderrorsize' );
1039                                 case 3: # The uploaded file was only partially uploaded
1040                                         return new WikiErrorMsg( 'importuploaderrorpartial' );
1041                             case 6: #Missing a temporary folder. Introduced in PHP 4.3.10 and PHP 5.0.3.
1042                                 return new WikiErrorMsg( 'importuploaderrortemp' );
1043                             # case else: # Currently impossible
1044                         }
1045
1046                 }
1047                 $fname = $upload['tmp_name'];
1048                 if( is_uploaded_file( $fname ) ) {
1049                         return ImportStreamSource::newFromFile( $fname );
1050                 } else {
1051                         return new WikiErrorMsg( 'importnofile' );
1052                 }
1053         }
1054
1055         static function newFromURL( $url, $method = 'GET' ) {
1056                 wfDebug( __METHOD__ . ": opening $url\n" );
1057                 # Use the standard HTTP fetch function; it times out
1058                 # quicker and sorts out user-agent problems which might
1059                 # otherwise prevent importing from large sites, such
1060                 # as the Wikimedia cluster, etc.
1061                 $data = Http::request( $method, $url );
1062                 if( $data !== false ) {
1063                         $file = tmpfile();
1064                         fwrite( $file, $data );
1065                         fflush( $file );
1066                         fseek( $file, 0 );
1067                         return new ImportStreamSource( $file );
1068                 } else {
1069                         return new WikiErrorMsg( 'importcantopen' );
1070                 }
1071         }
1072
1073         public static function newFromInterwiki( $interwiki, $page, $history=false ) {
1074                 if( $page == '' ) {
1075                         return new WikiErrorMsg( 'import-noarticle' );
1076                 }
1077                 $link = Title::newFromText( "$interwiki:Special:Export/$page" );
1078                 if( is_null( $link ) || $link->getInterwiki() == '' ) {
1079                         return new WikiErrorMsg( 'importbadinterwiki' );
1080                 } else {
1081                         $params = $history ? 'history=1' : '';
1082                         $url = $link->getFullUrl( $params );
1083                         # For interwikis, use POST to avoid redirects.
1084                         return ImportStreamSource::newFromURL( $url, "POST" );
1085                 }
1086         }
1087 }