includes/Import.php

   1 <?php
   2 /**
   3  * MediaWiki page data importer.
   4  *
   5  * Copyright © 2003,2005 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  * @ingroup SpecialPage
  25  */
  26
  27 /**
  28  * XML file reader for the page data importer
  29  *
  30  * implements Special:Import
  31  * @ingroup SpecialPage
  32  */
  33 class WikiImporter {
  34         private $reader = null;
  35         private $mLogItemCallback, $mUploadCallback, $mRevisionCallback, $mPageCallback;
  36         private $mSiteInfoCallback, $mTargetNamespace, $mTargetRootPage, $mPageOutCallback;
  37         private $mNoticeCallback, $mDebug;
  38         private $mImportUploads, $mImageBasePath;
  39         private $mNoUpdates = false;
  40
  41         /**
  42          * Creates an ImportXMLReader drawing from the source provided
  43          * @param $source
  44          */
  45         function __construct( $source ) {
  46                 $this->reader = new XMLReader();
  47
  48                 stream_wrapper_register( 'uploadsource', 'UploadSourceAdapter' );
  49                 $id = UploadSourceAdapter::registerSource( $source );
  50                 if (defined( 'LIBXML_PARSEHUGE' ) ) {
  51                         $this->reader->open( "uploadsource://$id", null, LIBXML_PARSEHUGE );
  52                 } else {
  53                         $this->reader->open( "uploadsource://$id" );
  54                 }
  55
  56                 // Default callbacks
  57                 $this->setRevisionCallback( array( $this, "importRevision" ) );
  58                 $this->setUploadCallback( array( $this, 'importUpload' ) );
  59                 $this->setLogItemCallback( array( $this, 'importLogItem' ) );
  60                 $this->setPageOutCallback( array( $this, 'finishImportPage' ) );
  61         }
  62
  63         private function throwXmlError( $err ) {
  64                 $this->debug( "FAILURE: $err" );
  65                 wfDebug( "WikiImporter XML error: $err\n" );
  66         }
  67
  68         private function debug( $data ) {
  69                 if( $this->mDebug ) {
  70                         wfDebug( "IMPORT: $data\n" );
  71                 }
  72         }
  73
  74         private function warn( $data ) {
  75                 wfDebug( "IMPORT: $data\n" );
  76         }
  77
  78         private function notice( $msg /*, $param, ...*/ ) {
  79                 $params = func_get_args();
  80                 array_shift( $params );
  81
  82                 if ( is_callable( $this->mNoticeCallback ) ) {
  83                         call_user_func( $this->mNoticeCallback, $msg, $params );
  84                 } else { # No ImportReporter -> CLI
  85                         echo wfMessage( $msg, $params )->text() . "\n";
  86                 }
  87         }
  88
  89         /**
  90          * Set debug mode...
  91          * @param $debug bool
  92          */
  93         function setDebug( $debug ) {
  94                 $this->mDebug = $debug;
  95         }
  96
  97         /**
  98          * Set 'no updates' mode. In this mode, the link tables will not be updated by the importer
  99          * @param $noupdates bool
 100          */
 101         function setNoUpdates( $noupdates ) {
 102                 $this->mNoUpdates = $noupdates;
 103         }
 104
 105         /**
 106          * Set a callback that displays notice messages
 107          *
 108          * @param $callback callback
 109          * @return callback
 110          */
 111         public function setNoticeCallback( $callback ) {
 112                 return wfSetVar( $this->mNoticeCallback, $callback );
 113         }
 114
 115         /**
 116          * Sets the action to perform as each new page in the stream is reached.
 117          * @param $callback callback
 118          * @return callback
 119          */
 120         public function setPageCallback( $callback ) {
 121                 $previous = $this->mPageCallback;
 122                 $this->mPageCallback = $callback;
 123                 return $previous;
 124         }
 125
 126         /**
 127          * Sets the action to perform as each page in the stream is completed.
 128          * Callback accepts the page title (as a Title object), a second object
 129          * with the original title form (in case it's been overridden into a
 130          * local namespace), and a count of revisions.
 131          *
 132          * @param $callback callback
 133          * @return callback
 134          */
 135         public function setPageOutCallback( $callback ) {
 136                 $previous = $this->mPageOutCallback;
 137                 $this->mPageOutCallback = $callback;
 138                 return $previous;
 139         }
 140
 141         /**
 142          * Sets the action to perform as each page revision is reached.
 143          * @param $callback callback
 144          * @return callback
 145          */
 146         public function setRevisionCallback( $callback ) {
 147                 $previous = $this->mRevisionCallback;
 148                 $this->mRevisionCallback = $callback;
 149                 return $previous;
 150         }
 151
 152         /**
 153          * Sets the action to perform as each file upload version is reached.
 154          * @param $callback callback
 155          * @return callback
 156          */
 157         public function setUploadCallback( $callback ) {
 158                 $previous = $this->mUploadCallback;
 159                 $this->mUploadCallback = $callback;
 160                 return $previous;
 161         }
 162
 163         /**
 164          * Sets the action to perform as each log item reached.
 165          * @param $callback callback
 166          * @return callback
 167          */
 168         public function setLogItemCallback( $callback ) {
 169                 $previous = $this->mLogItemCallback;
 170                 $this->mLogItemCallback = $callback;
 171                 return $previous;
 172         }
 173
 174         /**
 175          * Sets the action to perform when site info is encountered
 176          * @param $callback callback
 177          * @return callback
 178          */
 179         public function setSiteInfoCallback( $callback ) {
 180                 $previous = $this->mSiteInfoCallback;
 181                 $this->mSiteInfoCallback = $callback;
 182                 return $previous;
 183         }
 184
 185         /**
 186          * Set a target namespace to override the defaults
 187          * @param $namespace
 188          * @return bool
 189          */
 190         public function setTargetNamespace( $namespace ) {
 191                 if( is_null( $namespace ) ) {
 192                         // Don't override namespaces
 193                         $this->mTargetNamespace = null;
 194                 } elseif( $namespace >= 0 ) {
 195                         // @todo FIXME: Check for validity
 196                         $this->mTargetNamespace = intval( $namespace );
 197                 } else {
 198                         return false;
 199                 }
 200         }
 201
 202         /**
 203          * Set a target root page under which all pages are imported
 204          * @param $rootpage
 205          * @return status object
 206          */
 207         public function setTargetRootPage( $rootpage ) {
 208                 $status = Status::newGood();
 209                 if( is_null( $rootpage ) ) {
 210                         // No rootpage
 211                         $this->mTargetRootPage = null;
 212                 } elseif( $rootpage !== '' ) {
 213                         $rootpage = rtrim( $rootpage, '/' ); //avoid double slashes
 214                         $title = Title::newFromText( $rootpage, !is_null( $this->mTargetNamespace ) ? $this->mTargetNamespace : NS_MAIN );
 215                         if( !$title || $title->isExternal() ) {
 216                                 $status->fatal( 'import-rootpage-invalid' );
 217                         } else {
 218                                 if( !MWNamespace::hasSubpages( $title->getNamespace() ) ) {
 219                                         global $wgContLang;
 220
 221                                         $displayNSText = $title->getNamespace() == NS_MAIN
 222                                                 ? wfMessage( 'blanknamespace' )->text()
 223                                                 : $wgContLang->getNsText( $title->getNamespace() );
 224                                         $status->fatal( 'import-rootpage-nosubpage', $displayNSText );
 225                                 } else {
 226                                         // set namespace to 'all', so the namespace check in processTitle() can passed
 227                                         $this->setTargetNamespace( null );
 228                                         $this->mTargetRootPage = $title->getPrefixedDBKey();
 229                                 }
 230                         }
 231                 }
 232                 return $status;
 233         }
 234
 235         /**
 236          * @param $dir
 237          */
 238         public function setImageBasePath( $dir ) {
 239                 $this->mImageBasePath = $dir;
 240         }
 241
 242         /**
 243          * @param $import
 244          */
 245         public function setImportUploads( $import ) {
 246                 $this->mImportUploads = $import;
 247         }
 248
 249         /**
 250          * Default per-revision callback, performs the import.
 251          * @param $revision WikiRevision
 252          * @return bool
 253          */
 254         public function importRevision( $revision ) {
 255                 try {
 256                         $dbw = wfGetDB( DB_MASTER );
 257                         return $dbw->deadlockLoop( array( $revision, 'importOldRevision' ) );
 258                 } catch ( MWContentSerializationException $ex ) {
 259                         $this->notice( 'import-error-unserialize',
 260                                 $revision->getTitle()->getPrefixedText(),
 261                                 $revision->getID(),
 262                                 $revision->getModel(),
 263                                 $revision->getFormat() );
 264                 }
 265         }
 266
 267         /**
 268          * Default per-revision callback, performs the import.
 269          * @param $rev WikiRevision
 270          * @return bool
 271          */
 272         public function importLogItem( $rev ) {
 273                 $dbw = wfGetDB( DB_MASTER );
 274                 return $dbw->deadlockLoop( array( $rev, 'importLogItem' ) );
 275         }
 276
 277         /**
 278          * Dummy for now...
 279          * @param $revision
 280          * @return bool
 281          */
 282         public function importUpload( $revision ) {
 283                 $dbw = wfGetDB( DB_MASTER );
 284                 return $dbw->deadlockLoop( array( $revision, 'importUpload' ) );
 285         }
 286
 287         /**
 288          * Mostly for hook use
 289          * @param $title
 290          * @param $origTitle
 291          * @param $revCount
 292          * @param $sRevCount
 293          * @param $pageInfo
 294          * @return
 295          */
 296         public function finishImportPage( $title, $origTitle, $revCount, $sRevCount, $pageInfo ) {
 297                 $args = func_get_args();
 298                 return wfRunHooks( 'AfterImportPage', $args );
 299         }
 300
 301         /**
 302          * Alternate per-revision callback, for debugging.
 303          * @param $revision WikiRevision
 304          */
 305         public function debugRevisionHandler( &$revision ) {
 306                 $this->debug( "Got revision:" );
 307                 if( is_object( $revision->title ) ) {
 308                         $this->debug( "-- Title: " . $revision->title->getPrefixedText() );
 309                 } else {
 310                         $this->debug( "-- Title: <invalid>" );
 311                 }
 312                 $this->debug( "-- User: " . $revision->user_text );
 313                 $this->debug( "-- Timestamp: " . $revision->timestamp );
 314                 $this->debug( "-- Comment: " . $revision->comment );
 315                 $this->debug( "-- Text: " . $revision->text );
 316         }
 317
 318         /**
 319          * Notify the callback function when a new "<page>" is reached.
 320          * @param $title Title
 321          */
 322         function pageCallback( $title ) {
 323                 if( isset( $this->mPageCallback ) ) {
 324                         call_user_func( $this->mPageCallback, $title );
 325                 }
 326         }
 327
 328         /**
 329          * Notify the callback function when a "</page>" is closed.
 330          * @param $title Title
 331          * @param $origTitle Title
 332          * @param $revCount Integer
 333          * @param $sucCount Int: number of revisions for which callback returned true
 334          * @param $pageInfo Array: associative array of page information
 335          */
 336         private function pageOutCallback( $title, $origTitle, $revCount, $sucCount, $pageInfo ) {
 337                 if( isset( $this->mPageOutCallback ) ) {
 338                         $args = func_get_args();
 339                         call_user_func_array( $this->mPageOutCallback, $args );
 340                 }
 341         }
 342
 343         /**
 344          * Notify the callback function of a revision
 345          * @param $revision WikiRevision object
 346          * @return bool|mixed
 347          */
 348         private function revisionCallback( $revision ) {
 349                 if ( isset( $this->mRevisionCallback ) ) {
 350                         return call_user_func_array( $this->mRevisionCallback,
 351                                         array( $revision, $this ) );
 352                 } else {
 353                         return false;
 354                 }
 355         }
 356
 357         /**
 358          * Notify the callback function of a new log item
 359          * @param $revision WikiRevision object
 360          * @return bool|mixed
 361          */
 362         private function logItemCallback( $revision ) {
 363                 if ( isset( $this->mLogItemCallback ) ) {
 364                         return call_user_func_array( $this->mLogItemCallback,
 365                                         array( $revision, $this ) );
 366                 } else {
 367                         return false;
 368                 }
 369         }
 370
 371         /**
 372          * Shouldn't something like this be built-in to XMLReader?
 373          * Fetches text contents of the current element, assuming
 374          * no sub-elements or such scary things.
 375          * @return string
 376          * @access private
 377          */
 378         private function nodeContents() {
 379                 if( $this->reader->isEmptyElement ) {
 380                         return "";
 381                 }
 382                 $buffer = "";
 383                 while( $this->reader->read() ) {
 384                         switch( $this->reader->nodeType ) {
 385                         case XmlReader::TEXT:
 386                         case XmlReader::SIGNIFICANT_WHITESPACE:
 387                                 $buffer .= $this->reader->value;
 388                                 break;
 389                         case XmlReader::END_ELEMENT:
 390                                 return $buffer;
 391                         }
 392                 }
 393
 394                 $this->reader->close();
 395                 return '';
 396         }
 397
 398         # --------------
 399
 400         /** Left in for debugging */
 401         private function dumpElement() {
 402                 static $lookup = null;
 403                 if (!$lookup) {
 404                         $xmlReaderConstants = array(
 405                                 "NONE",
 406                                 "ELEMENT",
 407                                 "ATTRIBUTE",
 408                                 "TEXT",
 409                                 "CDATA",
 410                                 "ENTITY_REF",
 411                                 "ENTITY",
 412                                 "PI",
 413                                 "COMMENT",
 414                                 "DOC",
 415                                 "DOC_TYPE",
 416                                 "DOC_FRAGMENT",
 417                                 "NOTATION",
 418                                 "WHITESPACE",
 419                                 "SIGNIFICANT_WHITESPACE",
 420                                 "END_ELEMENT",
 421                                 "END_ENTITY",
 422                                 "XML_DECLARATION",
 423                                 );
 424                         $lookup = array();
 425
 426                         foreach( $xmlReaderConstants as $name ) {
 427                                 $lookup[constant("XmlReader::$name")] = $name;
 428                         }
 429                 }
 430
 431                 print( var_dump(
 432                         $lookup[$this->reader->nodeType],
 433                         $this->reader->name,
 434                         $this->reader->value
 435                 )."\n\n" );
 436         }
 437
 438         /**
 439          * Primary entry point
 440          * @throws MWException
 441          * @return bool
 442          */
 443         public function doImport() {
 444                 $this->reader->read();
 445
 446                 if ( $this->reader->name != 'mediawiki' ) {
 447                         throw new MWException( "Expected <mediawiki> tag, got ".
 448                                 $this->reader->name );
 449                 }
 450                 $this->debug( "<mediawiki> tag is correct." );
 451
 452                 $this->debug( "Starting primary dump processing loop." );
 453
 454                 $keepReading = $this->reader->read();
 455                 $skip = false;
 456                 while ( $keepReading ) {
 457                         $tag = $this->reader->name;
 458                         $type = $this->reader->nodeType;
 459
 460                         if ( !wfRunHooks( 'ImportHandleToplevelXMLTag', $this ) ) {
 461                                 // Do nothing
 462                         } elseif ( $tag == 'mediawiki' && $type == XmlReader::END_ELEMENT ) {
 463                                 break;
 464                         } elseif ( $tag == 'siteinfo' ) {
 465                                 $this->handleSiteInfo();
 466                         } elseif ( $tag == 'page' ) {
 467                                 $this->handlePage();
 468                         } elseif ( $tag == 'logitem' ) {
 469                                 $this->handleLogItem();
 470                         } elseif ( $tag != '#text' ) {
 471                                 $this->warn( "Unhandled top-level XML tag $tag" );
 472
 473                                 $skip = true;
 474                         }
 475
 476                         if ($skip) {
 477                                 $keepReading = $this->reader->next();
 478                                 $skip = false;
 479                                 $this->debug( "Skip" );
 480                         } else {
 481                                 $keepReading = $this->reader->read();
 482                         }
 483                 }
 484
 485                 return true;
 486         }
 487
 488         /**
 489          * @return bool
 490          * @throws MWException
 491          */
 492         private function handleSiteInfo() {
 493                 // Site info is useful, but not actually used for dump imports.
 494                 // Includes a quick short-circuit to save performance.
 495                 if ( ! $this->mSiteInfoCallback ) {
 496                         $this->reader->next();
 497                         return true;
 498                 }
 499                 throw new MWException( "SiteInfo tag is not yet handled, do not set mSiteInfoCallback" );
 500         }
 501
 502         private function handleLogItem() {
 503                 $this->debug( "Enter log item handler." );
 504                 $logInfo = array();
 505
 506                 // Fields that can just be stuffed in the pageInfo object
 507                 $normalFields = array( 'id', 'comment', 'type', 'action', 'timestamp',
 508                                         'logtitle', 'params' );
 509
 510                 while ( $this->reader->read() ) {
 511                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 512                                         $this->reader->name == 'logitem') {
 513                                 break;
 514                         }
 515
 516                         $tag = $this->reader->name;
 517
 518                         if ( !wfRunHooks( 'ImportHandleLogItemXMLTag',
 519                                                 $this, $logInfo ) ) {
 520                                 // Do nothing
 521                         } elseif ( in_array( $tag, $normalFields ) ) {
 522                                 $logInfo[$tag] = $this->nodeContents();
 523                         } elseif ( $tag == 'contributor' ) {
 524                                 $logInfo['contributor'] = $this->handleContributor();
 525                         } elseif ( $tag != '#text' ) {
 526                                 $this->warn( "Unhandled log-item XML tag $tag" );
 527                         }
 528                 }
 529
 530                 $this->processLogItem( $logInfo );
 531         }
 532
 533         /**
 534          * @param $logInfo
 535          * @return bool|mixed
 536          */
 537         private function processLogItem( $logInfo ) {
 538                 $revision = new WikiRevision;
 539
 540                 $revision->setID( $logInfo['id'] );
 541                 $revision->setType( $logInfo['type'] );
 542                 $revision->setAction( $logInfo['action'] );
 543                 $revision->setTimestamp( $logInfo['timestamp'] );
 544                 $revision->setParams( $logInfo['params'] );
 545                 $revision->setTitle( Title::newFromText( $logInfo['logtitle'] ) );
 546                 $revision->setNoUpdates( $this->mNoUpdates );
 547
 548                 if ( isset( $logInfo['comment'] ) ) {
 549                         $revision->setComment( $logInfo['comment'] );
 550                 }
 551
 552                 if ( isset( $logInfo['contributor']['ip'] ) ) {
 553                         $revision->setUserIP( $logInfo['contributor']['ip'] );
 554                 }
 555                 if ( isset( $logInfo['contributor']['username'] ) ) {
 556                         $revision->setUserName( $logInfo['contributor']['username'] );
 557                 }
 558
 559                 return $this->logItemCallback( $revision );
 560         }
 561
 562         private function handlePage() {
 563                 // Handle page data.
 564                 $this->debug( "Enter page handler." );
 565                 $pageInfo = array( 'revisionCount' => 0, 'successfulRevisionCount' => 0 );
 566
 567                 // Fields that can just be stuffed in the pageInfo object
 568                 $normalFields = array( 'title', 'id', 'redirect', 'restrictions' );
 569
 570                 $skip = false;
 571                 $badTitle = false;
 572
 573                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 574                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 575                                         $this->reader->name == 'page') {
 576                                 break;
 577                         }
 578
 579                         $tag = $this->reader->name;
 580
 581                         if ( $badTitle ) {
 582                                 // The title is invalid, bail out of this page
 583                                 $skip = true;
 584                         } elseif ( !wfRunHooks( 'ImportHandlePageXMLTag', array( $this,
 585                                                 &$pageInfo ) ) ) {
 586                                 // Do nothing
 587                         } elseif ( in_array( $tag, $normalFields ) ) {
 588                                 $pageInfo[$tag] = $this->nodeContents();
 589                                 if ( $tag == 'title' ) {
 590                                         $title = $this->processTitle( $pageInfo['title'] );
 591
 592                                         if ( !$title ) {
 593                                                 $badTitle = true;
 594                                                 $skip = true;
 595                                         }
 596
 597                                         $this->pageCallback( $title );
 598                                         list( $pageInfo['_title'], $origTitle ) = $title;
 599                                 }
 600                         } elseif ( $tag == 'revision' ) {
 601                                 $this->handleRevision( $pageInfo );
 602                         } elseif ( $tag == 'upload' ) {
 603                                 $this->handleUpload( $pageInfo );
 604                         } elseif ( $tag != '#text' ) {
 605                                 $this->warn( "Unhandled page XML tag $tag" );
 606                                 $skip = true;
 607                         }
 608                 }
 609
 610                 $this->pageOutCallback( $pageInfo['_title'], $origTitle,
 611                                         $pageInfo['revisionCount'],
 612                                         $pageInfo['successfulRevisionCount'],
 613                                         $pageInfo );
 614         }
 615
 616         /**
 617          * @param $pageInfo array
 618          */
 619         private function handleRevision( &$pageInfo ) {
 620                 $this->debug( "Enter revision handler" );
 621                 $revisionInfo = array();
 622
 623                 $normalFields = array( 'id', 'timestamp', 'comment', 'minor', 'model', 'format', 'text' );
 624
 625                 $skip = false;
 626
 627                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 628                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 629                                         $this->reader->name == 'revision') {
 630                                 break;
 631                         }
 632
 633                         $tag = $this->reader->name;
 634
 635                         if ( !wfRunHooks( 'ImportHandleRevisionXMLTag', $this,
 636                                                 $pageInfo, $revisionInfo ) ) {
 637                                 // Do nothing
 638                         } elseif ( in_array( $tag, $normalFields ) ) {
 639                                 $revisionInfo[$tag] = $this->nodeContents();
 640                         } elseif ( $tag == 'contributor' ) {
 641                                 $revisionInfo['contributor'] = $this->handleContributor();
 642                         } elseif ( $tag != '#text' ) {
 643                                 $this->warn( "Unhandled revision XML tag $tag" );
 644                                 $skip = true;
 645                         }
 646                 }
 647
 648                 $pageInfo['revisionCount']++;
 649                 if ( $this->processRevision( $pageInfo, $revisionInfo ) ) {
 650                         $pageInfo['successfulRevisionCount']++;
 651                 }
 652         }
 653
 654         /**
 655          * @param $pageInfo
 656          * @param $revisionInfo
 657          * @return bool|mixed
 658          */
 659         private function processRevision( $pageInfo, $revisionInfo ) {
 660                 $revision = new WikiRevision;
 661
 662                 if( isset( $revisionInfo['id'] ) ) {
 663                         $revision->setID( $revisionInfo['id'] );
 664                 }
 665                 if ( isset( $revisionInfo['text'] ) ) {
 666                         $revision->setText( $revisionInfo['text'] );
 667                 }
 668                 if ( isset( $revisionInfo['model'] ) ) {
 669                         $revision->setModel( $revisionInfo['model'] );
 670                 }
 671                 if ( isset( $revisionInfo['format'] ) ) {
 672                         $revision->setFormat( $revisionInfo['format'] );
 673                 }
 674                 $revision->setTitle( $pageInfo['_title'] );
 675
 676                 if ( isset( $revisionInfo['timestamp'] ) ) {
 677                         $revision->setTimestamp( $revisionInfo['timestamp'] );
 678                 } else {
 679                         $revision->setTimestamp( wfTimestampNow() );
 680                 }
 681
 682                 if ( isset( $revisionInfo['comment'] ) ) {
 683                         $revision->setComment( $revisionInfo['comment'] );
 684                 }
 685
 686                 if ( isset( $revisionInfo['minor'] ) ) {
 687                         $revision->setMinor( true );
 688                 }
 689                 if ( isset( $revisionInfo['contributor']['ip'] ) ) {
 690                         $revision->setUserIP( $revisionInfo['contributor']['ip'] );
 691                 }
 692                 if ( isset( $revisionInfo['contributor']['username'] ) ) {
 693                         $revision->setUserName( $revisionInfo['contributor']['username'] );
 694                 }
 695                 $revision->setNoUpdates( $this->mNoUpdates );
 696
 697                 return $this->revisionCallback( $revision );
 698         }
 699
 700         /**
 701          * @param $pageInfo
 702          * @return mixed
 703          */
 704         private function handleUpload( &$pageInfo ) {
 705                 $this->debug( "Enter upload handler" );
 706                 $uploadInfo = array();
 707
 708                 $normalFields = array( 'timestamp', 'comment', 'filename', 'text',
 709                                         'src', 'size', 'sha1base36', 'archivename', 'rel' );
 710
 711                 $skip = false;
 712
 713                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 714                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 715                                         $this->reader->name == 'upload') {
 716                                 break;
 717                         }
 718
 719                         $tag = $this->reader->name;
 720
 721                         if ( !wfRunHooks( 'ImportHandleUploadXMLTag', $this,
 722                                                 $pageInfo ) ) {
 723                                 // Do nothing
 724                         } elseif ( in_array( $tag, $normalFields ) ) {
 725                                 $uploadInfo[$tag] = $this->nodeContents();
 726                         } elseif ( $tag == 'contributor' ) {
 727                                 $uploadInfo['contributor'] = $this->handleContributor();
 728                         } elseif ( $tag == 'contents' ) {
 729                                 $contents = $this->nodeContents();
 730                                 $encoding = $this->reader->getAttribute( 'encoding' );
 731                                 if ( $encoding === 'base64' ) {
 732                                         $uploadInfo['fileSrc'] = $this->dumpTemp( base64_decode( $contents ) );
 733                                         $uploadInfo['isTempSrc'] = true;
 734                                 }
 735                         } elseif ( $tag != '#text' ) {
 736                                 $this->warn( "Unhandled upload XML tag $tag" );
 737                                 $skip = true;
 738                         }
 739                 }
 740
 741                 if ( $this->mImageBasePath && isset( $uploadInfo['rel'] ) ) {
 742                         $path = "{$this->mImageBasePath}/{$uploadInfo['rel']}";
 743                         if ( file_exists( $path ) ) {
 744                                 $uploadInfo['fileSrc'] = $path;
 745                                 $uploadInfo['isTempSrc'] = false;
 746                         }
 747                 }
 748
 749                 if ( $this->mImportUploads ) {
 750                         return $this->processUpload( $pageInfo, $uploadInfo );
 751                 }
 752         }
 753
 754         /**
 755          * @param $contents
 756          * @return string
 757          */
 758         private function dumpTemp( $contents ) {
 759                 $filename = tempnam( wfTempDir(), 'importupload' );
 760                 file_put_contents( $filename, $contents );
 761                 return $filename;
 762         }
 763
 764         /**
 765          * @param $pageInfo
 766          * @param $uploadInfo
 767          * @return mixed
 768          */
 769         private function processUpload( $pageInfo, $uploadInfo ) {
 770                 $revision = new WikiRevision;
 771                 $text = isset( $uploadInfo['text'] ) ? $uploadInfo['text'] : '';
 772
 773                 $revision->setTitle( $pageInfo['_title'] );
 774                 $revision->setID( $pageInfo['id'] );
 775                 $revision->setTimestamp( $uploadInfo['timestamp'] );
 776                 $revision->setText( $text );
 777                 $revision->setFilename( $uploadInfo['filename'] );
 778                 if ( isset( $uploadInfo['archivename'] ) ) {
 779                         $revision->setArchiveName( $uploadInfo['archivename'] );
 780                 }
 781                 $revision->setSrc( $uploadInfo['src'] );
 782                 if ( isset( $uploadInfo['fileSrc'] ) ) {
 783                         $revision->setFileSrc( $uploadInfo['fileSrc'],
 784                                 !empty( $uploadInfo['isTempSrc'] ) );
 785                 }
 786                 if ( isset( $uploadInfo['sha1base36'] ) ) {
 787                         $revision->setSha1Base36( $uploadInfo['sha1base36'] );
 788                 }
 789                 $revision->setSize( intval( $uploadInfo['size'] ) );
 790                 $revision->setComment( $uploadInfo['comment'] );
 791
 792                 if ( isset( $uploadInfo['contributor']['ip'] ) ) {
 793                         $revision->setUserIP( $uploadInfo['contributor']['ip'] );
 794                 }
 795                 if ( isset( $uploadInfo['contributor']['username'] ) ) {
 796                         $revision->setUserName( $uploadInfo['contributor']['username'] );
 797                 }
 798                 $revision->setNoUpdates( $this->mNoUpdates );
 799
 800                 return call_user_func( $this->mUploadCallback, $revision );
 801         }
 802
 803         /**
 804          * @return array
 805          */
 806         private function handleContributor() {
 807                 $fields = array( 'id', 'ip', 'username' );
 808                 $info = array();
 809
 810                 while ( $this->reader->read() ) {
 811                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 812                                         $this->reader->name == 'contributor') {
 813                                 break;
 814                         }
 815
 816                         $tag = $this->reader->name;
 817
 818                         if ( in_array( $tag, $fields ) ) {
 819                                 $info[$tag] = $this->nodeContents();
 820                         }
 821                 }
 822
 823                 return $info;
 824         }
 825
 826         /**
 827          * @param $text string
 828          * @return Array or false
 829          */
 830         private function processTitle( $text ) {
 831                 global $wgCommandLineMode;
 832
 833                 $workTitle = $text;
 834                 $origTitle = Title::newFromText( $workTitle );
 835
 836                 if( !is_null( $this->mTargetNamespace ) && !is_null( $origTitle ) ) {
 837                         # makeTitleSafe, because $origTitle can have a interwiki (different setting of interwiki map)
 838                         # and than dbKey can begin with a lowercase char
 839                         $title = Title::makeTitleSafe( $this->mTargetNamespace,
 840                                 $origTitle->getDBkey() );
 841                 } else {
 842                         if( !is_null( $this->mTargetRootPage ) ) {
 843                                 $workTitle = $this->mTargetRootPage . '/' . $workTitle;
 844                         }
 845                         $title = Title::newFromText( $workTitle );
 846                 }
 847
 848                 if( is_null( $title ) ) {
 849                         # Invalid page title? Ignore the page
 850                         $this->notice( 'import-error-invalid', $workTitle );
 851                         return false;
 852                 } elseif( $title->isExternal() ) {
 853                         $this->notice( 'import-error-interwiki', $title->getPrefixedText() );
 854                         return false;
 855                 } elseif( !$title->canExist() ) {
 856                         $this->notice( 'import-error-special', $title->getPrefixedText() );
 857                         return false;
 858                 } elseif( !$title->userCan( 'edit' ) && !$wgCommandLineMode ) {
 859                         # Do not import if the importing wiki user cannot edit this page
 860                         $this->notice( 'import-error-edit', $title->getPrefixedText() );
 861                         return false;
 862                 } elseif( !$title->exists() && !$title->userCan( 'create' ) && !$wgCommandLineMode ) {
 863                         # Do not import if the importing wiki user cannot create this page
 864                         $this->notice( 'import-error-create', $title->getPrefixedText() );
 865                         return false;
 866                 }
 867
 868                 return array( $title, $origTitle );
 869         }
 870 }
 871
 872 /** This is a horrible hack used to keep source compatibility */
 873 class UploadSourceAdapter {
 874         static $sourceRegistrations = array();
 875
 876         private $mSource;
 877         private $mBuffer;
 878         private $mPosition;
 879
 880         /**
 881          * @param $source
 882          * @return string
 883          */
 884         static function registerSource( $source ) {
 885                 $id = wfRandomString();
 886
 887                 self::$sourceRegistrations[$id] = $source;
 888
 889                 return $id;
 890         }
 891
 892         /**
 893          * @param $path
 894          * @param $mode
 895          * @param $options
 896          * @param $opened_path
 897          * @return bool
 898          */
 899         function stream_open( $path, $mode, $options, &$opened_path ) {
 900                 $url = parse_url($path);
 901                 $id = $url['host'];
 902
 903                 if ( !isset( self::$sourceRegistrations[$id] ) ) {
 904                         return false;
 905                 }
 906
 907                 $this->mSource = self::$sourceRegistrations[$id];
 908
 909                 return true;
 910         }
 911
 912         /**
 913          * @param $count
 914          * @return string
 915          */
 916         function stream_read( $count ) {
 917                 $return = '';
 918                 $leave = false;
 919
 920                 while ( !$leave && !$this->mSource->atEnd() &&
 921                                 strlen($this->mBuffer) < $count ) {
 922                         $read = $this->mSource->readChunk();
 923
 924                         if ( !strlen($read) ) {
 925                                 $leave = true;
 926                         }
 927
 928                         $this->mBuffer .= $read;
 929                 }
 930
 931                 if ( strlen($this->mBuffer) ) {
 932                         $return = substr( $this->mBuffer, 0, $count );
 933                         $this->mBuffer = substr( $this->mBuffer, $count );
 934                 }
 935
 936                 $this->mPosition += strlen($return);
 937
 938                 return $return;
 939         }
 940
 941         /**
 942          * @param $data
 943          * @return bool
 944          */
 945         function stream_write( $data ) {
 946                 return false;
 947         }
 948
 949         /**
 950          * @return mixed
 951          */
 952         function stream_tell() {
 953                 return $this->mPosition;
 954         }
 955
 956         /**
 957          * @return bool
 958          */
 959         function stream_eof() {
 960                 return $this->mSource->atEnd();
 961         }
 962
 963         /**
 964          * @return array
 965          */
 966         function url_stat() {
 967                 $result = array();
 968
 969                 $result['dev'] = $result[0] = 0;
 970                 $result['ino'] = $result[1] = 0;
 971                 $result['mode'] = $result[2] = 0;
 972                 $result['nlink'] = $result[3] = 0;
 973                 $result['uid'] = $result[4] = 0;
 974                 $result['gid'] = $result[5] = 0;
 975                 $result['rdev'] = $result[6] = 0;
 976                 $result['size'] = $result[7] = 0;
 977                 $result['atime'] = $result[8] = 0;
 978                 $result['mtime'] = $result[9] = 0;
 979                 $result['ctime'] = $result[10] = 0;
 980                 $result['blksize'] = $result[11] = 0;
 981                 $result['blocks'] = $result[12] = 0;
 982
 983                 return $result;
 984         }
 985 }
 986
 987 class XMLReader2 extends XMLReader {
 988
 989         /**
 990          * @return bool|string
 991          */
 992         function nodeContents() {
 993                 if( $this->isEmptyElement ) {
 994                         return "";
 995                 }
 996                 $buffer = "";
 997                 while( $this->read() ) {
 998                         switch( $this->nodeType ) {
 999                         case XmlReader::TEXT:
1000                         case XmlReader::SIGNIFICANT_WHITESPACE:
1001                                 $buffer .= $this->value;
1002                                 break;
1003                         case XmlReader::END_ELEMENT:
1004                                 return $buffer;
1005                         }
1006                 }
1007                 return $this->close();
1008         }
1009 }
1010
1011 /**
1012  * @todo document (e.g. one-sentence class description).
1013  * @ingroup SpecialPage
1014  */
1015 class WikiRevision {
1016         var $importer = null;
1017
1018         /**
1019          * @var Title
1020          */
1021         var $title = null;
1022         var $id = 0;
1023         var $timestamp = "20010115000000";
1024         var $user = 0;
1025         var $user_text = "";
1026         var $model = null;
1027         var $format = null;
1028         var $text = "";
1029         var $content = null;
1030         var $comment = "";
1031         var $minor = false;
1032         var $type = "";
1033         var $action = "";
1034         var $params = "";
1035         var $fileSrc = '';
1036         var $sha1base36 = false;
1037         var $isTemp = false;
1038         var $archiveName = '';
1039         var $fileIsTemp;
1040         private $mNoUpdates = false;
1041
1042         /**
1043          * @param $title
1044          * @throws MWException
1045          */
1046         function setTitle( $title ) {
1047                 if( is_object( $title ) ) {
1048                         $this->title = $title;
1049                 } elseif( is_null( $title ) ) {
1050                         throw new MWException( "WikiRevision given a null title in import. You may need to adjust \$wgLegalTitleChars." );
1051                 } else {
1052                         throw new MWException( "WikiRevision given non-object title in import." );
1053                 }
1054         }
1055
1056         /**
1057          * @param $id
1058          */
1059         function setID( $id ) {
1060                 $this->id = $id;
1061         }
1062
1063         /**
1064          * @param $ts
1065          */
1066         function setTimestamp( $ts ) {
1067                 # 2003-08-05T18:30:02Z
1068                 $this->timestamp = wfTimestamp( TS_MW, $ts );
1069         }
1070
1071         /**
1072          * @param $user
1073          */
1074         function setUsername( $user ) {
1075                 $this->user_text = $user;
1076         }
1077
1078         /**
1079          * @param $ip
1080          */
1081         function setUserIP( $ip ) {
1082                 $this->user_text = $ip;
1083         }
1084
1085         /**
1086          * @param $model
1087          */
1088         function setModel( $model ) {
1089                 $this->model = $model;
1090         }
1091
1092         /**
1093          * @param $format
1094          */
1095         function setFormat( $format ) {
1096                 $this->format = $format;
1097         }
1098
1099         /**
1100          * @param $text
1101          */
1102         function setText( $text ) {
1103                 $this->text = $text;
1104         }
1105
1106         /**
1107          * @param $text
1108          */
1109         function setComment( $text ) {
1110                 $this->comment = $text;
1111         }
1112
1113         /**
1114          * @param $minor
1115          */
1116         function setMinor( $minor ) {
1117                 $this->minor = (bool)$minor;
1118         }
1119
1120         /**
1121          * @param $src
1122          */
1123         function setSrc( $src ) {
1124                 $this->src = $src;
1125         }
1126
1127         /**
1128          * @param $src
1129          * @param $isTemp
1130          */
1131         function setFileSrc( $src, $isTemp ) {
1132                 $this->fileSrc = $src;
1133                 $this->fileIsTemp = $isTemp;
1134         }
1135
1136         /**
1137          * @param $sha1base36
1138          */
1139         function setSha1Base36( $sha1base36 ) {
1140                 $this->sha1base36 = $sha1base36;
1141         }
1142
1143         /**
1144          * @param $filename
1145          */
1146         function setFilename( $filename ) {
1147                 $this->filename = $filename;
1148         }
1149
1150         /**
1151          * @param $archiveName
1152          */
1153         function setArchiveName( $archiveName ) {
1154                 $this->archiveName = $archiveName;
1155         }
1156
1157         /**
1158          * @param $size
1159          */
1160         function setSize( $size ) {
1161                 $this->size = intval( $size );
1162         }
1163
1164         /**
1165          * @param $type
1166          */
1167         function setType( $type ) {
1168                 $this->type = $type;
1169         }
1170
1171         /**
1172          * @param $action
1173          */
1174         function setAction( $action ) {
1175                 $this->action = $action;
1176         }
1177
1178         /**
1179          * @param $params
1180          */
1181         function setParams( $params ) {
1182                 $this->params = $params;
1183         }
1184
1185         /**
1186          * @param $noupdates
1187          */
1188         public function setNoUpdates( $noupdates ) {
1189                 $this->mNoUpdates = $noupdates;
1190         }
1191
1192         /**
1193          * @return Title
1194          */
1195         function getTitle() {
1196                 return $this->title;
1197         }
1198
1199         /**
1200          * @return int
1201          */
1202         function getID() {
1203                 return $this->id;
1204         }
1205
1206         /**
1207          * @return string
1208          */
1209         function getTimestamp() {
1210                 return $this->timestamp;
1211         }
1212
1213         /**
1214          * @return string
1215          */
1216         function getUser() {
1217                 return $this->user_text;
1218         }
1219
1220         /**
1221          * @return string
1222          *
1223          * @deprecated Since 1.21, use getContent() instead.
1224          */
1225         function getText() {
1226                 ContentHandler::deprecated( __METHOD__, '1.21' );
1227
1228                 return $this->text;
1229         }
1230
1231         /**
1232          * @return Content
1233          */
1234         function getContent() {
1235                 if ( is_null( $this->content ) ) {
1236                         $this->content =
1237                                 ContentHandler::makeContent(
1238                                         $this->text,
1239                                         $this->getTitle(),
1240                                         $this->getModel(),
1241                                         $this->getFormat()
1242                                 );
1243                 }
1244
1245                 return $this->content;
1246         }
1247
1248         /**
1249          * @return String
1250          */
1251         function getModel() {
1252                 if ( is_null( $this->model ) ) {
1253                         $this->model = $this->getTitle()->getContentModel();
1254                 }
1255
1256                 return $this->model;
1257         }
1258
1259         /**
1260          * @return String
1261          */
1262         function getFormat() {
1263                 if ( is_null( $this->model ) ) {
1264                         $this->format = ContentHandler::getForTitle( $this->getTitle() )->getDefaultFormat();
1265                 }
1266
1267                 return $this->format;
1268         }
1269
1270         /**
1271          * @return string
1272          */
1273         function getComment() {
1274                 return $this->comment;
1275         }
1276
1277         /**
1278          * @return bool
1279          */
1280         function getMinor() {
1281                 return $this->minor;
1282         }
1283
1284         /**
1285          * @return mixed
1286          */
1287         function getSrc() {
1288                 return $this->src;
1289         }
1290
1291         /**
1292          * @return bool|String
1293          */
1294         function getSha1() {
1295                 if ( $this->sha1base36 ) {
1296                         return wfBaseConvert( $this->sha1base36, 36, 16 );
1297                 }
1298                 return false;
1299         }
1300
1301         /**
1302          * @return string
1303          */
1304         function getFileSrc() {
1305                 return $this->fileSrc;
1306         }
1307
1308         /**
1309          * @return bool
1310          */
1311         function isTempSrc() {
1312                 return $this->isTemp;
1313         }
1314
1315         /**
1316          * @return mixed
1317          */
1318         function getFilename() {
1319                 return $this->filename;
1320         }
1321
1322         /**
1323          * @return string
1324          */
1325         function getArchiveName() {
1326                 return $this->archiveName;
1327         }
1328
1329         /**
1330          * @return mixed
1331          */
1332         function getSize() {
1333                 return $this->size;
1334         }
1335
1336         /**
1337          * @return string
1338          */
1339         function getType() {
1340                 return $this->type;
1341         }
1342
1343         /**
1344          * @return string
1345          */
1346         function getAction() {
1347                 return $this->action;
1348         }
1349
1350         /**
1351          * @return string
1352          */
1353         function getParams() {
1354                 return $this->params;
1355         }
1356
1357         /**
1358          * @return bool
1359          */
1360         function importOldRevision() {
1361                 $dbw = wfGetDB( DB_MASTER );
1362
1363                 # Sneak a single revision into place
1364                 $user = User::newFromName( $this->getUser() );
1365                 if( $user ) {
1366                         $userId = intval( $user->getId() );
1367                         $userText = $user->getName();
1368                         $userObj = $user;
1369                 } else {
1370                         $userId = 0;
1371                         $userText = $this->getUser();
1372                         $userObj = new User;
1373                 }
1374
1375                 // avoid memory leak...?
1376                 $linkCache = LinkCache::singleton();
1377                 $linkCache->clear();
1378
1379                 $page = WikiPage::factory( $this->title );
1380                 if( !$page->exists() ) {
1381                         # must create the page...
1382                         $pageId = $page->insertOn( $dbw );
1383                         $created = true;
1384                         $oldcountable = null;
1385                 } else {
1386                         $pageId = $page->getId();
1387                         $created = false;
1388
1389                         $prior = $dbw->selectField( 'revision', '1',
1390                                 array( 'rev_page' => $pageId,
1391                                         'rev_timestamp' => $dbw->timestamp( $this->timestamp ),
1392                                         'rev_user_text' => $userText,
1393                                         'rev_comment'   => $this->getComment() ),
1394                                 __METHOD__
1395                         );
1396                         if( $prior ) {
1397                                 // @todo FIXME: This could fail slightly for multiple matches :P
1398                                 wfDebug( __METHOD__ . ": skipping existing revision for [[" .
1399                                         $this->title->getPrefixedText() . "]], timestamp " . $this->timestamp . "\n" );
1400                                 return false;
1401                         }
1402                         $oldcountable = $page->isCountable();
1403                 }
1404
1405                 # @todo FIXME: Use original rev_id optionally (better for backups)
1406                 # Insert the row
1407                 $revision = new Revision( array(
1408                         'title'      => $this->title,
1409                         'page'       => $pageId,
1410                         'content_model'  => $this->getModel(),
1411                         'content_format' => $this->getFormat(),
1412                         'text'       => $this->getContent()->serialize( $this->getFormat() ), //XXX: just set 'content' => $this->getContent()?
1413                         'comment'    => $this->getComment(),
1414                         'user'       => $userId,
1415                         'user_text'  => $userText,
1416                         'timestamp'  => $this->timestamp,
1417                         'minor_edit' => $this->minor,
1418                         ) );
1419                 $revision->insertOn( $dbw );
1420                 $changed = $page->updateIfNewerOn( $dbw, $revision );
1421
1422                 if ( $changed !== false && !$this->mNoUpdates ) {
1423                         wfDebug( __METHOD__ . ": running updates\n" );
1424                         $page->doEditUpdates( $revision, $userObj, array( 'created' => $created, 'oldcountable' => $oldcountable ) );
1425                 }
1426
1427                 return true;
1428         }
1429
1430         /**
1431          * @return mixed
1432          */
1433         function importLogItem() {
1434                 $dbw = wfGetDB( DB_MASTER );
1435                 # @todo FIXME: This will not record autoblocks
1436                 if( !$this->getTitle() ) {
1437                         wfDebug( __METHOD__ . ": skipping invalid {$this->type}/{$this->action} log time, timestamp " .
1438                                 $this->timestamp . "\n" );
1439                         return;
1440                 }
1441                 # Check if it exists already
1442                 // @todo FIXME: Use original log ID (better for backups)
1443                 $prior = $dbw->selectField( 'logging', '1',
1444                         array( 'log_type' => $this->getType(),
1445                                 'log_action'    => $this->getAction(),
1446                                 'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1447                                 'log_namespace' => $this->getTitle()->getNamespace(),
1448                                 'log_title'     => $this->getTitle()->getDBkey(),
1449                                 'log_comment'   => $this->getComment(),
1450                                 #'log_user_text' => $this->user_text,
1451                                 'log_params'    => $this->params ),
1452                         __METHOD__
1453                 );
1454                 // @todo FIXME: This could fail slightly for multiple matches :P
1455                 if( $prior ) {
1456                         wfDebug( __METHOD__ . ": skipping existing item for Log:{$this->type}/{$this->action}, timestamp " .
1457                                 $this->timestamp . "\n" );
1458                         return;
1459                 }
1460                 $log_id = $dbw->nextSequenceValue( 'logging_log_id_seq' );
1461                 $data = array(
1462                         'log_id' => $log_id,
1463                         'log_type' => $this->type,
1464                         'log_action' => $this->action,
1465                         'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1466                         'log_user' => User::idFromName( $this->user_text ),
1467                         #'log_user_text' => $this->user_text,
1468                         'log_namespace' => $this->getTitle()->getNamespace(),
1469                         'log_title' => $this->getTitle()->getDBkey(),
1470                         'log_comment' => $this->getComment(),
1471                         'log_params' => $this->params
1472                 );
1473                 $dbw->insert( 'logging', $data, __METHOD__ );
1474         }
1475
1476         /**
1477          * @return bool
1478          */
1479         function importUpload() {
1480                 # Construct a file
1481                 $archiveName = $this->getArchiveName();
1482                 if ( $archiveName ) {
1483                         wfDebug( __METHOD__ . "Importing archived file as $archiveName\n" );
1484                         $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1485                                 RepoGroup::singleton()->getLocalRepo(), $archiveName );
1486                 } else {
1487                         $file = wfLocalFile( $this->getTitle() );
1488                         wfDebug( __METHOD__ . 'Importing new file as ' . $file->getName() . "\n" );
1489                         if ( $file->exists() && $file->getTimestamp() > $this->getTimestamp() ) {
1490                                 $archiveName = $file->getTimestamp() . '!' . $file->getName();
1491                                 $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1492                                         RepoGroup::singleton()->getLocalRepo(), $archiveName );
1493                                 wfDebug( __METHOD__ . "File already exists; importing as $archiveName\n" );
1494                         }
1495                 }
1496                 if( !$file ) {
1497                         wfDebug( __METHOD__ . ': Bad file for ' . $this->getTitle() . "\n" );
1498                         return false;
1499                 }
1500
1501                 # Get the file source or download if necessary
1502                 $source = $this->getFileSrc();
1503                 $flags = $this->isTempSrc() ? File::DELETE_SOURCE : 0;
1504                 if ( !$source ) {
1505                         $source = $this->downloadSource();
1506                         $flags |= File::DELETE_SOURCE;
1507                 }
1508                 if( !$source ) {
1509                         wfDebug( __METHOD__ . ": Could not fetch remote file.\n" );
1510                         return false;
1511                 }
1512                 $sha1 = $this->getSha1();
1513                 if ( $sha1 && ( $sha1 !== sha1_file( $source ) ) ) {
1514                         if ( $flags & File::DELETE_SOURCE ) {
1515                                 # Broken file; delete it if it is a temporary file
1516                                 unlink( $source );
1517                         }
1518                         wfDebug( __METHOD__ . ": Corrupt file $source.\n" );
1519                         return false;
1520                 }
1521
1522                 $user = User::newFromName( $this->user_text );
1523
1524                 # Do the actual upload
1525                 if ( $archiveName ) {
1526                         $status = $file->uploadOld( $source, $archiveName,
1527                                 $this->getTimestamp(), $this->getComment(), $user, $flags );
1528                 } else {
1529                         $status = $file->upload( $source, $this->getComment(), $this->getComment(),
1530                                 $flags, false, $this->getTimestamp(), $user );
1531                 }
1532
1533                 if ( $status->isGood() ) {
1534                         wfDebug( __METHOD__ . ": Succesful\n" );
1535                         return true;
1536                 } else {
1537                         wfDebug( __METHOD__ . ': failed: ' . $status->getXml() . "\n" );
1538                         return false;
1539                 }
1540         }
1541
1542         /**
1543          * @return bool|string
1544          */
1545         function downloadSource() {
1546                 global $wgEnableUploads;
1547                 if( !$wgEnableUploads ) {
1548                         return false;
1549                 }
1550
1551                 $tempo = tempnam( wfTempDir(), 'download' );
1552                 $f = fopen( $tempo, 'wb' );
1553                 if( !$f ) {
1554                         wfDebug( "IMPORT: couldn't write to temp file $tempo\n" );
1555                         return false;
1556                 }
1557
1558                 // @todo FIXME!
1559                 $src = $this->getSrc();
1560                 $data = Http::get( $src );
1561                 if( !$data ) {
1562                         wfDebug( "IMPORT: couldn't fetch source $src\n" );
1563                         fclose( $f );
1564                         unlink( $tempo );
1565                         return false;
1566                 }
1567
1568                 fwrite( $f, $data );
1569                 fclose( $f );
1570
1571                 return $tempo;
1572         }
1573
1574 }
1575
1576 /**
1577  * @todo document (e.g. one-sentence class description).
1578  * @ingroup SpecialPage
1579  */
1580 class ImportStringSource {
1581         function __construct( $string ) {
1582                 $this->mString = $string;
1583                 $this->mRead = false;
1584         }
1585
1586         /**
1587          * @return bool
1588          */
1589         function atEnd() {
1590                 return $this->mRead;
1591         }
1592
1593         /**
1594          * @return bool|string
1595          */
1596         function readChunk() {
1597                 if( $this->atEnd() ) {
1598                         return false;
1599                 }
1600                 $this->mRead = true;
1601                 return $this->mString;
1602         }
1603 }
1604
1605 /**
1606  * @todo document (e.g. one-sentence class description).
1607  * @ingroup SpecialPage
1608  */
1609 class ImportStreamSource {
1610         function __construct( $handle ) {
1611                 $this->mHandle = $handle;
1612         }
1613
1614         /**
1615          * @return bool
1616          */
1617         function atEnd() {
1618                 return feof( $this->mHandle );
1619         }
1620
1621         /**
1622          * @return string
1623          */
1624         function readChunk() {
1625                 return fread( $this->mHandle, 32768 );
1626         }
1627
1628         /**
1629          * @param $filename string
1630          * @return Status
1631          */
1632         static function newFromFile( $filename ) {
1633                 wfSuppressWarnings();
1634                 $file = fopen( $filename, 'rt' );
1635                 wfRestoreWarnings();
1636                 if( !$file ) {
1637                         return Status::newFatal( "importcantopen" );
1638                 }
1639                 return Status::newGood( new ImportStreamSource( $file ) );
1640         }
1641
1642         /**
1643          * @param $fieldname string
1644          * @return Status
1645          */
1646         static function newFromUpload( $fieldname = "xmlimport" ) {
1647                 $upload =& $_FILES[$fieldname];
1648
1649                 if( !isset( $upload ) || !$upload['name'] ) {
1650                         return Status::newFatal( 'importnofile' );
1651                 }
1652                 if( !empty( $upload['error'] ) ) {
1653                         switch( $upload['error'] ) {
1654                                 case 1: # The uploaded file exceeds the upload_max_filesize directive in php.ini.
1655                                         return Status::newFatal( 'importuploaderrorsize' );
1656                                 case 2: # The uploaded file exceeds the MAX_FILE_SIZE directive that was specified in the HTML form.
1657                                         return Status::newFatal( 'importuploaderrorsize' );
1658                                 case 3: # The uploaded file was only partially uploaded
1659                                         return Status::newFatal( 'importuploaderrorpartial' );
1660                                 case 6: #Missing a temporary folder.
1661                                         return Status::newFatal( 'importuploaderrortemp' );
1662                                 # case else: # Currently impossible
1663                         }
1664
1665                 }
1666                 $fname = $upload['tmp_name'];
1667                 if( is_uploaded_file( $fname ) ) {
1668                         return ImportStreamSource::newFromFile( $fname );
1669                 } else {
1670                         return Status::newFatal( 'importnofile' );
1671                 }
1672         }
1673
1674         /**
1675          * @param $url
1676          * @param $method string
1677          * @return Status
1678          */
1679         static function newFromURL( $url, $method = 'GET' ) {
1680                 wfDebug( __METHOD__ . ": opening $url\n" );
1681                 # Use the standard HTTP fetch function; it times out
1682                 # quicker and sorts out user-agent problems which might
1683                 # otherwise prevent importing from large sites, such
1684                 # as the Wikimedia cluster, etc.
1685                 $data = Http::request( $method, $url, array( 'followRedirects' => true ) );
1686                 if( $data !== false ) {
1687                         $file = tmpfile();
1688                         fwrite( $file, $data );
1689                         fflush( $file );
1690                         fseek( $file, 0 );
1691                         return Status::newGood( new ImportStreamSource( $file ) );
1692                 } else {
1693                         return Status::newFatal( 'importcantopen' );
1694                 }
1695         }
1696
1697         /**
1698          * @param $interwiki
1699          * @param $page
1700          * @param $history bool
1701          * @param $templates bool
1702          * @param $pageLinkDepth int
1703          * @return Status
1704          */
1705         public static function newFromInterwiki( $interwiki, $page, $history = false, $templates = false, $pageLinkDepth = 0 ) {
1706                 if( $page == '' ) {
1707                         return Status::newFatal( 'import-noarticle' );
1708                 }
1709                 $link = Title::newFromText( "$interwiki:Special:Export/$page" );
1710                 if( is_null( $link ) || $link->getInterwiki() == '' ) {
1711                         return Status::newFatal( 'importbadinterwiki' );
1712                 } else {
1713                         $params = array();
1714                         if ( $history ) $params['history'] = 1;
1715                         if ( $templates ) $params['templates'] = 1;
1716                         if ( $pageLinkDepth ) $params['pagelink-depth'] = $pageLinkDepth;
1717                         $url = $link->getFullUrl( $params );
1718                         # For interwikis, use POST to avoid redirects.
1719                         return ImportStreamSource::newFromURL( $url, "POST" );
1720                 }
1721         }
1722 }