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