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', '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                 $revision->setTitle( $pageInfo['_title'] );
 627
 628                 if ( isset( $revisionInfo['timestamp'] ) ) {
 629                         $revision->setTimestamp( $revisionInfo['timestamp'] );
 630                 } else {
 631                         $revision->setTimestamp( wfTimestampNow() );
 632                 }
 633
 634                 if ( isset( $revisionInfo['comment'] ) ) {
 635                         $revision->setComment( $revisionInfo['comment'] );
 636                 }
 637
 638                 if ( isset( $revisionInfo['minor'] ) ) {
 639                         $revision->setMinor( true );
 640                 }
 641                 if ( isset( $revisionInfo['contributor']['ip'] ) ) {
 642                         $revision->setUserIP( $revisionInfo['contributor']['ip'] );
 643                 }
 644                 if ( isset( $revisionInfo['contributor']['username'] ) ) {
 645                         $revision->setUserName( $revisionInfo['contributor']['username'] );
 646                 }
 647                 $revision->setNoUpdates( $this->mNoUpdates );
 648
 649                 return $this->revisionCallback( $revision );
 650         }
 651
 652         /**
 653          * @param $pageInfo
 654          * @return mixed
 655          */
 656         private function handleUpload( &$pageInfo ) {
 657                 $this->debug( "Enter upload handler" );
 658                 $uploadInfo = array();
 659
 660                 $normalFields = array( 'timestamp', 'comment', 'filename', 'text',
 661                                         'src', 'size', 'sha1base36', 'archivename', 'rel' );
 662
 663                 $skip = false;
 664
 665                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 666                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 667                                         $this->reader->name == 'upload') {
 668                                 break;
 669                         }
 670
 671                         $tag = $this->reader->name;
 672
 673                         if ( !wfRunHooks( 'ImportHandleUploadXMLTag', $this,
 674                                                 $pageInfo ) ) {
 675                                 // Do nothing
 676                         } elseif ( in_array( $tag, $normalFields ) ) {
 677                                 $uploadInfo[$tag] = $this->nodeContents();
 678                         } elseif ( $tag == 'contributor' ) {
 679                                 $uploadInfo['contributor'] = $this->handleContributor();
 680                         } elseif ( $tag == 'contents' ) {
 681                                 $contents = $this->nodeContents();
 682                                 $encoding = $this->reader->getAttribute( 'encoding' );
 683                                 if ( $encoding === 'base64' ) {
 684                                         $uploadInfo['fileSrc'] = $this->dumpTemp( base64_decode( $contents ) );
 685                                         $uploadInfo['isTempSrc'] = true;
 686                                 }
 687                         } elseif ( $tag != '#text' ) {
 688                                 $this->warn( "Unhandled upload XML tag $tag" );
 689                                 $skip = true;
 690                         }
 691                 }
 692
 693                 if ( $this->mImageBasePath && isset( $uploadInfo['rel'] ) ) {
 694                         $path = "{$this->mImageBasePath}/{$uploadInfo['rel']}";
 695                         if ( file_exists( $path ) ) {
 696                                 $uploadInfo['fileSrc'] = $path;
 697                                 $uploadInfo['isTempSrc'] = false;
 698                         }
 699                 }
 700
 701                 if ( $this->mImportUploads ) {
 702                         return $this->processUpload( $pageInfo, $uploadInfo );
 703                 }
 704         }
 705
 706         /**
 707          * @param $contents
 708          * @return string
 709          */
 710         private function dumpTemp( $contents ) {
 711                 $filename = tempnam( wfTempDir(), 'importupload' );
 712                 file_put_contents( $filename, $contents );
 713                 return $filename;
 714         }
 715
 716         /**
 717          * @param $pageInfo
 718          * @param $uploadInfo
 719          * @return mixed
 720          */
 721         private function processUpload( $pageInfo, $uploadInfo ) {
 722                 $revision = new WikiRevision;
 723                 $text = isset( $uploadInfo['text'] ) ? $uploadInfo['text'] : '';
 724
 725                 $revision->setTitle( $pageInfo['_title'] );
 726                 $revision->setID( $pageInfo['id'] );
 727                 $revision->setTimestamp( $uploadInfo['timestamp'] );
 728                 $revision->setText( $text );
 729                 $revision->setFilename( $uploadInfo['filename'] );
 730                 if ( isset( $uploadInfo['archivename'] ) ) {
 731                         $revision->setArchiveName( $uploadInfo['archivename'] );
 732                 }
 733                 $revision->setSrc( $uploadInfo['src'] );
 734                 if ( isset( $uploadInfo['fileSrc'] ) ) {
 735                         $revision->setFileSrc( $uploadInfo['fileSrc'],
 736                                 !empty( $uploadInfo['isTempSrc'] ) );
 737                 }
 738                 if ( isset( $uploadInfo['sha1base36'] ) ) {
 739                         $revision->setSha1Base36( $uploadInfo['sha1base36'] );
 740                 }
 741                 $revision->setSize( intval( $uploadInfo['size'] ) );
 742                 $revision->setComment( $uploadInfo['comment'] );
 743
 744                 if ( isset( $uploadInfo['contributor']['ip'] ) ) {
 745                         $revision->setUserIP( $uploadInfo['contributor']['ip'] );
 746                 }
 747                 if ( isset( $uploadInfo['contributor']['username'] ) ) {
 748                         $revision->setUserName( $uploadInfo['contributor']['username'] );
 749                 }
 750                 $revision->setNoUpdates( $this->mNoUpdates );
 751
 752                 return call_user_func( $this->mUploadCallback, $revision );
 753         }
 754
 755         /**
 756          * @return array
 757          */
 758         private function handleContributor() {
 759                 $fields = array( 'id', 'ip', 'username' );
 760                 $info = array();
 761
 762                 while ( $this->reader->read() ) {
 763                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 764                                         $this->reader->name == 'contributor') {
 765                                 break;
 766                         }
 767
 768                         $tag = $this->reader->name;
 769
 770                         if ( in_array( $tag, $fields ) ) {
 771                                 $info[$tag] = $this->nodeContents();
 772                         }
 773                 }
 774
 775                 return $info;
 776         }
 777
 778         /**
 779          * @param $text string
 780          * @return Array or false
 781          */
 782         private function processTitle( $text ) {
 783                 global $wgCommandLineMode;
 784
 785                 $workTitle = $text;
 786                 $origTitle = Title::newFromText( $workTitle );
 787
 788                 if( !is_null( $this->mTargetNamespace ) && !is_null( $origTitle ) ) {
 789                         $title = Title::makeTitle( $this->mTargetNamespace,
 790                                 $origTitle->getDBkey() );
 791                 } else {
 792                         $title = Title::newFromText( $workTitle );
 793                 }
 794
 795                 if( is_null( $title ) ) {
 796                         # Invalid page title? Ignore the page
 797                         $this->notice( 'import-error-invalid', $workTitle );
 798                         return false;
 799                 } elseif( $title->isExternal() ) {
 800                         $this->notice( 'import-error-interwiki', $title->getPrefixedText() );
 801                         return false;
 802                 } elseif( !$title->canExist() ) {
 803                         $this->notice( 'import-error-special', $title->getPrefixedText() );
 804                         return false;
 805                 } elseif( !$title->userCan( 'edit' ) && !$wgCommandLineMode ) {
 806                         # Do not import if the importing wiki user cannot edit this page
 807                         $this->notice( 'import-error-edit', $title->getPrefixedText() );
 808                         return false;
 809                 } elseif( !$title->exists() && !$title->userCan( 'create' ) && !$wgCommandLineMode ) {
 810                         # Do not import if the importing wiki user cannot create this page
 811                         $this->notice( 'import-error-create', $title->getPrefixedText() );
 812                         return false;
 813                 }
 814
 815                 return array( $title, $origTitle );
 816         }
 817 }
 818
 819 /** This is a horrible hack used to keep source compatibility */
 820 class UploadSourceAdapter {
 821         static $sourceRegistrations = array();
 822
 823         private $mSource;
 824         private $mBuffer;
 825         private $mPosition;
 826
 827         /**
 828          * @param $source
 829          * @return string
 830          */
 831         static function registerSource( $source ) {
 832                 $id = wfRandomString();
 833
 834                 self::$sourceRegistrations[$id] = $source;
 835
 836                 return $id;
 837         }
 838
 839         /**
 840          * @param $path
 841          * @param $mode
 842          * @param $options
 843          * @param $opened_path
 844          * @return bool
 845          */
 846         function stream_open( $path, $mode, $options, &$opened_path ) {
 847                 $url = parse_url($path);
 848                 $id = $url['host'];
 849
 850                 if ( !isset( self::$sourceRegistrations[$id] ) ) {
 851                         return false;
 852                 }
 853
 854                 $this->mSource = self::$sourceRegistrations[$id];
 855
 856                 return true;
 857         }
 858
 859         /**
 860          * @param $count
 861          * @return string
 862          */
 863         function stream_read( $count ) {
 864                 $return = '';
 865                 $leave = false;
 866
 867                 while ( !$leave && !$this->mSource->atEnd() &&
 868                                 strlen($this->mBuffer) < $count ) {
 869                         $read = $this->mSource->readChunk();
 870
 871                         if ( !strlen($read) ) {
 872                                 $leave = true;
 873                         }
 874
 875                         $this->mBuffer .= $read;
 876                 }
 877
 878                 if ( strlen($this->mBuffer) ) {
 879                         $return = substr( $this->mBuffer, 0, $count );
 880                         $this->mBuffer = substr( $this->mBuffer, $count );
 881                 }
 882
 883                 $this->mPosition += strlen($return);
 884
 885                 return $return;
 886         }
 887
 888         /**
 889          * @param $data
 890          * @return bool
 891          */
 892         function stream_write( $data ) {
 893                 return false;
 894         }
 895
 896         /**
 897          * @return mixed
 898          */
 899         function stream_tell() {
 900                 return $this->mPosition;
 901         }
 902
 903         /**
 904          * @return bool
 905          */
 906         function stream_eof() {
 907                 return $this->mSource->atEnd();
 908         }
 909
 910         /**
 911          * @return array
 912          */
 913         function url_stat() {
 914                 $result = array();
 915
 916                 $result['dev'] = $result[0] = 0;
 917                 $result['ino'] = $result[1] = 0;
 918                 $result['mode'] = $result[2] = 0;
 919                 $result['nlink'] = $result[3] = 0;
 920                 $result['uid'] = $result[4] = 0;
 921                 $result['gid'] = $result[5] = 0;
 922                 $result['rdev'] = $result[6] = 0;
 923                 $result['size'] = $result[7] = 0;
 924                 $result['atime'] = $result[8] = 0;
 925                 $result['mtime'] = $result[9] = 0;
 926                 $result['ctime'] = $result[10] = 0;
 927                 $result['blksize'] = $result[11] = 0;
 928                 $result['blocks'] = $result[12] = 0;
 929
 930                 return $result;
 931         }
 932 }
 933
 934 class XMLReader2 extends XMLReader {
 935
 936         /**
 937          * @return bool|string
 938          */
 939         function nodeContents() {
 940                 if( $this->isEmptyElement ) {
 941                         return "";
 942                 }
 943                 $buffer = "";
 944                 while( $this->read() ) {
 945                         switch( $this->nodeType ) {
 946                         case XmlReader::TEXT:
 947                         case XmlReader::SIGNIFICANT_WHITESPACE:
 948                                 $buffer .= $this->value;
 949                                 break;
 950                         case XmlReader::END_ELEMENT:
 951                                 return $buffer;
 952                         }
 953                 }
 954                 return $this->close();
 955         }
 956 }
 957
 958 /**
 959  * @todo document (e.g. one-sentence class description).
 960  * @ingroup SpecialPage
 961  */
 962 class WikiRevision {
 963         var $importer = null;
 964
 965         /**
 966          * @var Title
 967          */
 968         var $title = null;
 969         var $id = 0;
 970         var $timestamp = "20010115000000";
 971         var $user = 0;
 972         var $user_text = "";
 973         var $text = "";
 974         var $comment = "";
 975         var $minor = false;
 976         var $type = "";
 977         var $action = "";
 978         var $params = "";
 979         var $fileSrc = '';
 980         var $sha1base36 = false;
 981         var $isTemp = false;
 982         var $archiveName = '';
 983         var $fileIsTemp;
 984         private $mNoUpdates = false;
 985
 986         /**
 987          * @param $title
 988          * @throws MWException
 989          */
 990         function setTitle( $title ) {
 991                 if( is_object( $title ) ) {
 992                         $this->title = $title;
 993                 } elseif( is_null( $title ) ) {
 994                         throw new MWException( "WikiRevision given a null title in import. You may need to adjust \$wgLegalTitleChars." );
 995                 } else {
 996                         throw new MWException( "WikiRevision given non-object title in import." );
 997                 }
 998         }
 999
1000         /**
1001          * @param $id
1002          */
1003         function setID( $id ) {
1004                 $this->id = $id;
1005         }
1006
1007         /**
1008          * @param $ts
1009          */
1010         function setTimestamp( $ts ) {
1011                 # 2003-08-05T18:30:02Z
1012                 $this->timestamp = wfTimestamp( TS_MW, $ts );
1013         }
1014
1015         /**
1016          * @param $user
1017          */
1018         function setUsername( $user ) {
1019                 $this->user_text = $user;
1020         }
1021
1022         /**
1023          * @param $ip
1024          */
1025         function setUserIP( $ip ) {
1026                 $this->user_text = $ip;
1027         }
1028
1029         /**
1030          * @param $text
1031          */
1032         function setText( $text ) {
1033                 $this->text = $text;
1034         }
1035
1036         /**
1037          * @param $text
1038          */
1039         function setComment( $text ) {
1040                 $this->comment = $text;
1041         }
1042
1043         /**
1044          * @param $minor
1045          */
1046         function setMinor( $minor ) {
1047                 $this->minor = (bool)$minor;
1048         }
1049
1050         /**
1051          * @param $src
1052          */
1053         function setSrc( $src ) {
1054                 $this->src = $src;
1055         }
1056
1057         /**
1058          * @param $src
1059          * @param $isTemp
1060          */
1061         function setFileSrc( $src, $isTemp ) {
1062                 $this->fileSrc = $src;
1063                 $this->fileIsTemp = $isTemp;
1064         }
1065
1066         /**
1067          * @param $sha1base36
1068          */
1069         function setSha1Base36( $sha1base36 ) {
1070                 $this->sha1base36 = $sha1base36;
1071         }
1072
1073         /**
1074          * @param $filename
1075          */
1076         function setFilename( $filename ) {
1077                 $this->filename = $filename;
1078         }
1079
1080         /**
1081          * @param $archiveName
1082          */
1083         function setArchiveName( $archiveName ) {
1084                 $this->archiveName = $archiveName;
1085         }
1086
1087         /**
1088          * @param $size
1089          */
1090         function setSize( $size ) {
1091                 $this->size = intval( $size );
1092         }
1093
1094         /**
1095          * @param $type
1096          */
1097         function setType( $type ) {
1098                 $this->type = $type;
1099         }
1100
1101         /**
1102          * @param $action
1103          */
1104         function setAction( $action ) {
1105                 $this->action = $action;
1106         }
1107
1108         /**
1109          * @param $params
1110          */
1111         function setParams( $params ) {
1112                 $this->params = $params;
1113         }
1114
1115         /**
1116          * @param $noupdates
1117          */
1118         public function setNoUpdates( $noupdates ) {
1119                 $this->mNoUpdates = $noupdates;
1120         }
1121
1122         /**
1123          * @return Title
1124          */
1125         function getTitle() {
1126                 return $this->title;
1127         }
1128
1129         /**
1130          * @return int
1131          */
1132         function getID() {
1133                 return $this->id;
1134         }
1135
1136         /**
1137          * @return string
1138          */
1139         function getTimestamp() {
1140                 return $this->timestamp;
1141         }
1142
1143         /**
1144          * @return string
1145          */
1146         function getUser() {
1147                 return $this->user_text;
1148         }
1149
1150         /**
1151          * @return string
1152          */
1153         function getText() {
1154                 return $this->text;
1155         }
1156
1157         /**
1158          * @return string
1159          */
1160         function getComment() {
1161                 return $this->comment;
1162         }
1163
1164         /**
1165          * @return bool
1166          */
1167         function getMinor() {
1168                 return $this->minor;
1169         }
1170
1171         /**
1172          * @return mixed
1173          */
1174         function getSrc() {
1175                 return $this->src;
1176         }
1177
1178         /**
1179          * @return bool|String
1180          */
1181         function getSha1() {
1182                 if ( $this->sha1base36 ) {
1183                         return wfBaseConvert( $this->sha1base36, 36, 16 );
1184                 }
1185                 return false;
1186         }
1187
1188         /**
1189          * @return string
1190          */
1191         function getFileSrc() {
1192                 return $this->fileSrc;
1193         }
1194
1195         /**
1196          * @return bool
1197          */
1198         function isTempSrc() {
1199                 return $this->isTemp;
1200         }
1201
1202         /**
1203          * @return mixed
1204          */
1205         function getFilename() {
1206                 return $this->filename;
1207         }
1208
1209         /**
1210          * @return string
1211          */
1212         function getArchiveName() {
1213                 return $this->archiveName;
1214         }
1215
1216         /**
1217          * @return mixed
1218          */
1219         function getSize() {
1220                 return $this->size;
1221         }
1222
1223         /**
1224          * @return string
1225          */
1226         function getType() {
1227                 return $this->type;
1228         }
1229
1230         /**
1231          * @return string
1232          */
1233         function getAction() {
1234                 return $this->action;
1235         }
1236
1237         /**
1238          * @return string
1239          */
1240         function getParams() {
1241                 return $this->params;
1242         }
1243
1244         /**
1245          * @return bool
1246          */
1247         function importOldRevision() {
1248                 $dbw = wfGetDB( DB_MASTER );
1249
1250                 # Sneak a single revision into place
1251                 $user = User::newFromName( $this->getUser() );
1252                 if( $user ) {
1253                         $userId = intval( $user->getId() );
1254                         $userText = $user->getName();
1255                         $userObj = $user;
1256                 } else {
1257                         $userId = 0;
1258                         $userText = $this->getUser();
1259                         $userObj = new User;
1260                 }
1261
1262                 // avoid memory leak...?
1263                 $linkCache = LinkCache::singleton();
1264                 $linkCache->clear();
1265
1266                 $page = WikiPage::factory( $this->title );
1267                 if( !$page->exists() ) {
1268                         # must create the page...
1269                         $pageId = $page->insertOn( $dbw );
1270                         $created = true;
1271                         $oldcountable = null;
1272                 } else {
1273                         $pageId = $page->getId();
1274                         $created = false;
1275
1276                         $prior = $dbw->selectField( 'revision', '1',
1277                                 array( 'rev_page' => $pageId,
1278                                         'rev_timestamp' => $dbw->timestamp( $this->timestamp ),
1279                                         'rev_user_text' => $userText,
1280                                         'rev_comment'   => $this->getComment() ),
1281                                 __METHOD__
1282                         );
1283                         if( $prior ) {
1284                                 // @todo FIXME: This could fail slightly for multiple matches :P
1285                                 wfDebug( __METHOD__ . ": skipping existing revision for [[" .
1286                                         $this->title->getPrefixedText() . "]], timestamp " . $this->timestamp . "\n" );
1287                                 return false;
1288                         }
1289                         $oldcountable = $page->isCountable();
1290                 }
1291
1292                 # @todo FIXME: Use original rev_id optionally (better for backups)
1293                 # Insert the row
1294                 $revision = new Revision( array(
1295                         'page'       => $pageId,
1296                         'text'       => $this->getText(),
1297                         'comment'    => $this->getComment(),
1298                         'user'       => $userId,
1299                         'user_text'  => $userText,
1300                         'timestamp'  => $this->timestamp,
1301                         'minor_edit' => $this->minor,
1302                         ) );
1303                 $revision->insertOn( $dbw );
1304                 $changed = $page->updateIfNewerOn( $dbw, $revision );
1305
1306                 if ( $changed !== false && !$this->mNoUpdates ) {
1307                         wfDebug( __METHOD__ . ": running updates\n" );
1308                         $page->doEditUpdates( $revision, $userObj, array( 'created' => $created, 'oldcountable' => $oldcountable ) );
1309                 }
1310
1311                 return true;
1312         }
1313
1314         /**
1315          * @return mixed
1316          */
1317         function importLogItem() {
1318                 $dbw = wfGetDB( DB_MASTER );
1319                 # @todo FIXME: This will not record autoblocks
1320                 if( !$this->getTitle() ) {
1321                         wfDebug( __METHOD__ . ": skipping invalid {$this->type}/{$this->action} log time, timestamp " .
1322                                 $this->timestamp . "\n" );
1323                         return;
1324                 }
1325                 # Check if it exists already
1326                 // @todo FIXME: Use original log ID (better for backups)
1327                 $prior = $dbw->selectField( 'logging', '1',
1328                         array( 'log_type' => $this->getType(),
1329                                 'log_action'    => $this->getAction(),
1330                                 'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1331                                 'log_namespace' => $this->getTitle()->getNamespace(),
1332                                 'log_title'     => $this->getTitle()->getDBkey(),
1333                                 'log_comment'   => $this->getComment(),
1334                                 #'log_user_text' => $this->user_text,
1335                                 'log_params'    => $this->params ),
1336                         __METHOD__
1337                 );
1338                 // @todo FIXME: This could fail slightly for multiple matches :P
1339                 if( $prior ) {
1340                         wfDebug( __METHOD__ . ": skipping existing item for Log:{$this->type}/{$this->action}, timestamp " .
1341                                 $this->timestamp . "\n" );
1342                         return;
1343                 }
1344                 $log_id = $dbw->nextSequenceValue( 'logging_log_id_seq' );
1345                 $data = array(
1346                         'log_id' => $log_id,
1347                         'log_type' => $this->type,
1348                         'log_action' => $this->action,
1349                         'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1350                         'log_user' => User::idFromName( $this->user_text ),
1351                         #'log_user_text' => $this->user_text,
1352                         'log_namespace' => $this->getTitle()->getNamespace(),
1353                         'log_title' => $this->getTitle()->getDBkey(),
1354                         'log_comment' => $this->getComment(),
1355                         'log_params' => $this->params
1356                 );
1357                 $dbw->insert( 'logging', $data, __METHOD__ );
1358         }
1359
1360         /**
1361          * @return bool
1362          */
1363         function importUpload() {
1364                 # Construct a file
1365                 $archiveName = $this->getArchiveName();
1366                 if ( $archiveName ) {
1367                         wfDebug( __METHOD__ . "Importing archived file as $archiveName\n" );
1368                         $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1369                                 RepoGroup::singleton()->getLocalRepo(), $archiveName );
1370                 } else {
1371                         $file = wfLocalFile( $this->getTitle() );
1372                         wfDebug( __METHOD__ . 'Importing new file as ' . $file->getName() . "\n" );
1373                         if ( $file->exists() && $file->getTimestamp() > $this->getTimestamp() ) {
1374                                 $archiveName = $file->getTimestamp() . '!' . $file->getName();
1375                                 $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1376                                         RepoGroup::singleton()->getLocalRepo(), $archiveName );
1377                                 wfDebug( __METHOD__ . "File already exists; importing as $archiveName\n" );
1378                         }
1379                 }
1380                 if( !$file ) {
1381                         wfDebug( __METHOD__ . ': Bad file for ' . $this->getTitle() . "\n" );
1382                         return false;
1383                 }
1384
1385                 # Get the file source or download if necessary
1386                 $source = $this->getFileSrc();
1387                 $flags = $this->isTempSrc() ? File::DELETE_SOURCE : 0;
1388                 if ( !$source ) {
1389                         $source = $this->downloadSource();
1390                         $flags |= File::DELETE_SOURCE;
1391                 }
1392                 if( !$source ) {
1393                         wfDebug( __METHOD__ . ": Could not fetch remote file.\n" );
1394                         return false;
1395                 }
1396                 $sha1 = $this->getSha1();
1397                 if ( $sha1 && ( $sha1 !== sha1_file( $source ) ) ) {
1398                         if ( $flags & File::DELETE_SOURCE ) {
1399                                 # Broken file; delete it if it is a temporary file
1400                                 unlink( $source );
1401                         }
1402                         wfDebug( __METHOD__ . ": Corrupt file $source.\n" );
1403                         return false;
1404                 }
1405
1406                 $user = User::newFromName( $this->user_text );
1407
1408                 # Do the actual upload
1409                 if ( $archiveName ) {
1410                         $status = $file->uploadOld( $source, $archiveName,
1411                                 $this->getTimestamp(), $this->getComment(), $user, $flags );
1412                 } else {
1413                         $status = $file->upload( $source, $this->getComment(), $this->getComment(),
1414                                 $flags, false, $this->getTimestamp(), $user );
1415                 }
1416
1417                 if ( $status->isGood() ) {
1418                         wfDebug( __METHOD__ . ": Succesful\n" );
1419                         return true;
1420                 } else {
1421                         wfDebug( __METHOD__ . ': failed: ' . $status->getXml() . "\n" );
1422                         return false;
1423                 }
1424         }
1425
1426         /**
1427          * @return bool|string
1428          */
1429         function downloadSource() {
1430                 global $wgEnableUploads;
1431                 if( !$wgEnableUploads ) {
1432                         return false;
1433                 }
1434
1435                 $tempo = tempnam( wfTempDir(), 'download' );
1436                 $f = fopen( $tempo, 'wb' );
1437                 if( !$f ) {
1438                         wfDebug( "IMPORT: couldn't write to temp file $tempo\n" );
1439                         return false;
1440                 }
1441
1442                 // @todo FIXME!
1443                 $src = $this->getSrc();
1444                 $data = Http::get( $src );
1445                 if( !$data ) {
1446                         wfDebug( "IMPORT: couldn't fetch source $src\n" );
1447                         fclose( $f );
1448                         unlink( $tempo );
1449                         return false;
1450                 }
1451
1452                 fwrite( $f, $data );
1453                 fclose( $f );
1454
1455                 return $tempo;
1456         }
1457
1458 }
1459
1460 /**
1461  * @todo document (e.g. one-sentence class description).
1462  * @ingroup SpecialPage
1463  */
1464 class ImportStringSource {
1465         function __construct( $string ) {
1466                 $this->mString = $string;
1467                 $this->mRead = false;
1468         }
1469
1470         /**
1471          * @return bool
1472          */
1473         function atEnd() {
1474                 return $this->mRead;
1475         }
1476
1477         /**
1478          * @return bool|string
1479          */
1480         function readChunk() {
1481                 if( $this->atEnd() ) {
1482                         return false;
1483                 }
1484                 $this->mRead = true;
1485                 return $this->mString;
1486         }
1487 }
1488
1489 /**
1490  * @todo document (e.g. one-sentence class description).
1491  * @ingroup SpecialPage
1492  */
1493 class ImportStreamSource {
1494         function __construct( $handle ) {
1495                 $this->mHandle = $handle;
1496         }
1497
1498         /**
1499          * @return bool
1500          */
1501         function atEnd() {
1502                 return feof( $this->mHandle );
1503         }
1504
1505         /**
1506          * @return string
1507          */
1508         function readChunk() {
1509                 return fread( $this->mHandle, 32768 );
1510         }
1511
1512         /**
1513          * @param $filename string
1514          * @return Status
1515          */
1516         static function newFromFile( $filename ) {
1517                 wfSuppressWarnings();
1518                 $file = fopen( $filename, 'rt' );
1519                 wfRestoreWarnings();
1520                 if( !$file ) {
1521                         return Status::newFatal( "importcantopen" );
1522                 }
1523                 return Status::newGood( new ImportStreamSource( $file ) );
1524         }
1525
1526         /**
1527          * @param $fieldname string
1528          * @return Status
1529          */
1530         static function newFromUpload( $fieldname = "xmlimport" ) {
1531                 $upload =& $_FILES[$fieldname];
1532
1533                 if( !isset( $upload ) || !$upload['name'] ) {
1534                         return Status::newFatal( 'importnofile' );
1535                 }
1536                 if( !empty( $upload['error'] ) ) {
1537                         switch($upload['error']){
1538                                 case 1: # The uploaded file exceeds the upload_max_filesize directive in php.ini.
1539                                         return Status::newFatal( 'importuploaderrorsize' );
1540                                 case 2: # The uploaded file exceeds the MAX_FILE_SIZE directive that was specified in the HTML form.
1541                                         return Status::newFatal( 'importuploaderrorsize' );
1542                                 case 3: # The uploaded file was only partially uploaded
1543                                         return Status::newFatal( 'importuploaderrorpartial' );
1544                                 case 6: #Missing a temporary folder.
1545                                         return Status::newFatal( 'importuploaderrortemp' );
1546                                 # case else: # Currently impossible
1547                         }
1548
1549                 }
1550                 $fname = $upload['tmp_name'];
1551                 if( is_uploaded_file( $fname ) ) {
1552                         return ImportStreamSource::newFromFile( $fname );
1553                 } else {
1554                         return Status::newFatal( 'importnofile' );
1555                 }
1556         }
1557
1558         /**
1559          * @param $url
1560          * @param $method string
1561          * @return Status
1562          */
1563         static function newFromURL( $url, $method = 'GET' ) {
1564                 wfDebug( __METHOD__ . ": opening $url\n" );
1565                 # Use the standard HTTP fetch function; it times out
1566                 # quicker and sorts out user-agent problems which might
1567                 # otherwise prevent importing from large sites, such
1568                 # as the Wikimedia cluster, etc.
1569                 $data = Http::request( $method, $url, array( 'followRedirects' => true ) );
1570                 if( $data !== false ) {
1571                         $file = tmpfile();
1572                         fwrite( $file, $data );
1573                         fflush( $file );
1574                         fseek( $file, 0 );
1575                         return Status::newGood( new ImportStreamSource( $file ) );
1576                 } else {
1577                         return Status::newFatal( 'importcantopen' );
1578                 }
1579         }
1580
1581         /**
1582          * @param $interwiki
1583          * @param $page
1584          * @param $history bool
1585          * @param $templates bool
1586          * @param $pageLinkDepth int
1587          * @return Status
1588          */
1589         public static function newFromInterwiki( $interwiki, $page, $history = false, $templates = false, $pageLinkDepth = 0 ) {
1590                 if( $page == '' ) {
1591                         return Status::newFatal( 'import-noarticle' );
1592                 }
1593                 $link = Title::newFromText( "$interwiki:Special:Export/$page" );
1594                 if( is_null( $link ) || $link->getInterwiki() == '' ) {
1595                         return Status::newFatal( 'importbadinterwiki' );
1596                 } else {
1597                         $params = array();
1598                         if ( $history ) $params['history'] = 1;
1599                         if ( $templates ) $params['templates'] = 1;
1600                         if ( $pageLinkDepth ) $params['pagelink-depth'] = $pageLinkDepth;
1601                         $url = $link->getFullUrl( $params );
1602                         # For interwikis, use POST to avoid redirects.
1603                         return ImportStreamSource::newFromURL( $url, "POST" );
1604                 }
1605         }
1606 }