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