includes/Import.php

   1 <?php
   2 /**
   3  * MediaWiki page data importer.
   4  *
   5  * Copyright © 2003,2005 Brion Vibber <brion@pobox.com>
   6  * http://www.mediawiki.org/
   7  *
   8  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 2 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with this program; if not, write to the Free Software Foundation, Inc.,
  20  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  21  * http://www.gnu.org/copyleft/gpl.html
  22  *
  23  * @file
  24  * @ingroup SpecialPage
  25  */
  26
  27 /**
  28  * XML file reader for the page data importer
  29  *
  30  * implements Special:Import
  31  * @ingroup SpecialPage
  32  */
  33 class WikiImporter {
  34         private $reader = null;
  35         private $mLogItemCallback, $mUploadCallback, $mRevisionCallback, $mPageCallback;
  36         private $mSiteInfoCallback, $mTargetNamespace, $mTargetRootPage, $mPageOutCallback;
  37         private $mNoticeCallback, $mDebug;
  38         private $mImportUploads, $mImageBasePath;
  39         private $mNoUpdates = false;
  40
  41         /**
  42          * Creates an ImportXMLReader drawing from the source provided
  43          * @param $source
  44          */
  45         function __construct( $source ) {
  46                 $this->reader = new XMLReader();
  47
  48                 stream_wrapper_register( 'uploadsource', 'UploadSourceAdapter' );
  49                 $id = UploadSourceAdapter::registerSource( $source );
  50                 if ( defined( 'LIBXML_PARSEHUGE' ) ) {
  51                         $this->reader->open( "uploadsource://$id", null, LIBXML_PARSEHUGE );
  52                 } else {
  53                         $this->reader->open( "uploadsource://$id" );
  54                 }
  55
  56                 // Default callbacks
  57                 $this->setRevisionCallback( array( $this, "importRevision" ) );
  58                 $this->setUploadCallback( array( $this, 'importUpload' ) );
  59                 $this->setLogItemCallback( array( $this, 'importLogItem' ) );
  60                 $this->setPageOutCallback( array( $this, 'finishImportPage' ) );
  61         }
  62
  63         private function throwXmlError( $err ) {
  64                 $this->debug( "FAILURE: $err" );
  65                 wfDebug( "WikiImporter XML error: $err\n" );
  66         }
  67
  68         private function debug( $data ) {
  69                 if ( $this->mDebug ) {
  70                         wfDebug( "IMPORT: $data\n" );
  71                 }
  72         }
  73
  74         private function warn( $data ) {
  75                 wfDebug( "IMPORT: $data\n" );
  76         }
  77
  78         private function notice( $msg /*, $param, ...*/ ) {
  79                 $params = func_get_args();
  80                 array_shift( $params );
  81
  82                 if ( is_callable( $this->mNoticeCallback ) ) {
  83                         call_user_func( $this->mNoticeCallback, $msg, $params );
  84                 } else { # No ImportReporter -> CLI
  85                         echo wfMessage( $msg, $params )->text() . "\n";
  86                 }
  87         }
  88
  89         /**
  90          * Set debug mode...
  91          * @param $debug bool
  92          */
  93         function setDebug( $debug ) {
  94                 $this->mDebug = $debug;
  95         }
  96
  97         /**
  98          * Set 'no updates' mode. In this mode, the link tables will not be updated by the importer
  99          * @param $noupdates bool
 100          */
 101         function setNoUpdates( $noupdates ) {
 102                 $this->mNoUpdates = $noupdates;
 103         }
 104
 105         /**
 106          * Set a callback that displays notice messages
 107          *
 108          * @param $callback callback
 109          * @return callback
 110          */
 111         public function setNoticeCallback( $callback ) {
 112                 return wfSetVar( $this->mNoticeCallback, $callback );
 113         }
 114
 115         /**
 116          * Sets the action to perform as each new page in the stream is reached.
 117          * @param $callback callback
 118          * @return callback
 119          */
 120         public function setPageCallback( $callback ) {
 121                 $previous = $this->mPageCallback;
 122                 $this->mPageCallback = $callback;
 123                 return $previous;
 124         }
 125
 126         /**
 127          * Sets the action to perform as each page in the stream is completed.
 128          * Callback accepts the page title (as a Title object), a second object
 129          * with the original title form (in case it's been overridden into a
 130          * local namespace), and a count of revisions.
 131          *
 132          * @param $callback callback
 133          * @return callback
 134          */
 135         public function setPageOutCallback( $callback ) {
 136                 $previous = $this->mPageOutCallback;
 137                 $this->mPageOutCallback = $callback;
 138                 return $previous;
 139         }
 140
 141         /**
 142          * Sets the action to perform as each page revision is reached.
 143          * @param $callback callback
 144          * @return callback
 145          */
 146         public function setRevisionCallback( $callback ) {
 147                 $previous = $this->mRevisionCallback;
 148                 $this->mRevisionCallback = $callback;
 149                 return $previous;
 150         }
 151
 152         /**
 153          * Sets the action to perform as each file upload version is reached.
 154          * @param $callback callback
 155          * @return callback
 156          */
 157         public function setUploadCallback( $callback ) {
 158                 $previous = $this->mUploadCallback;
 159                 $this->mUploadCallback = $callback;
 160                 return $previous;
 161         }
 162
 163         /**
 164          * Sets the action to perform as each log item reached.
 165          * @param $callback callback
 166          * @return callback
 167          */
 168         public function setLogItemCallback( $callback ) {
 169                 $previous = $this->mLogItemCallback;
 170                 $this->mLogItemCallback = $callback;
 171                 return $previous;
 172         }
 173
 174         /**
 175          * Sets the action to perform when site info is encountered
 176          * @param $callback callback
 177          * @return callback
 178          */
 179         public function setSiteInfoCallback( $callback ) {
 180                 $previous = $this->mSiteInfoCallback;
 181                 $this->mSiteInfoCallback = $callback;
 182                 return $previous;
 183         }
 184
 185         /**
 186          * Set a target namespace to override the defaults
 187          * @param $namespace
 188          * @return bool
 189          */
 190         public function setTargetNamespace( $namespace ) {
 191                 if ( is_null( $namespace ) ) {
 192                         // Don't override namespaces
 193                         $this->mTargetNamespace = null;
 194                 } elseif ( $namespace >= 0 ) {
 195                         // @todo FIXME: Check for validity
 196                         $this->mTargetNamespace = intval( $namespace );
 197                 } else {
 198                         return false;
 199                 }
 200         }
 201
 202         /**
 203          * Set a target root page under which all pages are imported
 204          * @param $rootpage
 205          * @return status object
 206          */
 207         public function setTargetRootPage( $rootpage ) {
 208                 $status = Status::newGood();
 209                 if ( is_null( $rootpage ) ) {
 210                         // No rootpage
 211                         $this->mTargetRootPage = null;
 212                 } elseif ( $rootpage !== '' ) {
 213                         $rootpage = rtrim( $rootpage, '/' ); //avoid double slashes
 214                         $title = Title::newFromText( $rootpage, !is_null( $this->mTargetNamespace ) ? $this->mTargetNamespace : NS_MAIN );
 215                         if ( !$title || $title->isExternal() ) {
 216                                 $status->fatal( 'import-rootpage-invalid' );
 217                         } else {
 218                                 if ( !MWNamespace::hasSubpages( $title->getNamespace() ) ) {
 219                                         global $wgContLang;
 220
 221                                         $displayNSText = $title->getNamespace() == NS_MAIN
 222                                                 ? wfMessage( 'blanknamespace' )->text()
 223                                                 : $wgContLang->getNsText( $title->getNamespace() );
 224                                         $status->fatal( 'import-rootpage-nosubpage', $displayNSText );
 225                                 } else {
 226                                         // set namespace to 'all', so the namespace check in processTitle() can passed
 227                                         $this->setTargetNamespace( null );
 228                                         $this->mTargetRootPage = $title->getPrefixedDBkey();
 229                                 }
 230                         }
 231                 }
 232                 return $status;
 233         }
 234
 235         /**
 236          * @param $dir
 237          */
 238         public function setImageBasePath( $dir ) {
 239                 $this->mImageBasePath = $dir;
 240         }
 241
 242         /**
 243          * @param $import
 244          */
 245         public function setImportUploads( $import ) {
 246                 $this->mImportUploads = $import;
 247         }
 248
 249         /**
 250          * Default per-revision callback, performs the import.
 251          * @param $revision WikiRevision
 252          * @return bool
 253          */
 254         public function importRevision( $revision ) {
 255                 try {
 256                         $dbw = wfGetDB( DB_MASTER );
 257                         return $dbw->deadlockLoop( array( $revision, 'importOldRevision' ) );
 258                 } catch ( MWContentSerializationException $ex ) {
 259                         $this->notice( 'import-error-unserialize',
 260                                 $revision->getTitle()->getPrefixedText(),
 261                                 $revision->getID(),
 262                                 $revision->getModel(),
 263                                 $revision->getFormat() );
 264                 }
 265         }
 266
 267         /**
 268          * Default per-revision callback, performs the import.
 269          * @param $rev WikiRevision
 270          * @return bool
 271          */
 272         public function importLogItem( $rev ) {
 273                 $dbw = wfGetDB( DB_MASTER );
 274                 return $dbw->deadlockLoop( array( $rev, 'importLogItem' ) );
 275         }
 276
 277         /**
 278          * Dummy for now...
 279          * @param $revision
 280          * @return bool
 281          */
 282         public function importUpload( $revision ) {
 283                 $dbw = wfGetDB( DB_MASTER );
 284                 return $dbw->deadlockLoop( array( $revision, 'importUpload' ) );
 285         }
 286
 287         /**
 288          * Mostly for hook use
 289          * @param $title
 290          * @param $origTitle
 291          * @param $revCount
 292          * @param $sRevCount
 293          * @param $pageInfo
 294          * @return
 295          */
 296         public function finishImportPage( $title, $origTitle, $revCount, $sRevCount, $pageInfo ) {
 297                 $args = func_get_args();
 298                 return wfRunHooks( 'AfterImportPage', $args );
 299         }
 300
 301         /**
 302          * Alternate per-revision callback, for debugging.
 303          * @param $revision WikiRevision
 304          */
 305         public function debugRevisionHandler( &$revision ) {
 306                 $this->debug( "Got revision:" );
 307                 if ( is_object( $revision->title ) ) {
 308                         $this->debug( "-- Title: " . $revision->title->getPrefixedText() );
 309                 } else {
 310                         $this->debug( "-- Title: <invalid>" );
 311                 }
 312                 $this->debug( "-- User: " . $revision->user_text );
 313                 $this->debug( "-- Timestamp: " . $revision->timestamp );
 314                 $this->debug( "-- Comment: " . $revision->comment );
 315                 $this->debug( "-- Text: " . $revision->text );
 316         }
 317
 318         /**
 319          * Notify the callback function when a new "<page>" is reached.
 320          * @param $title Title
 321          */
 322         function pageCallback( $title ) {
 323                 if ( isset( $this->mPageCallback ) ) {
 324                         call_user_func( $this->mPageCallback, $title );
 325                 }
 326         }
 327
 328         /**
 329          * Notify the callback function when a "</page>" is closed.
 330          * @param $title Title
 331          * @param $origTitle Title
 332          * @param $revCount Integer
 333          * @param int $sucCount number of revisions for which callback returned true
 334          * @param array $pageInfo associative array of page information
 335          */
 336         private function pageOutCallback( $title, $origTitle, $revCount, $sucCount, $pageInfo ) {
 337                 if ( isset( $this->mPageOutCallback ) ) {
 338                         $args = func_get_args();
 339                         call_user_func_array( $this->mPageOutCallback, $args );
 340                 }
 341         }
 342
 343         /**
 344          * Notify the callback function of a revision
 345          * @param $revision WikiRevision object
 346          * @return bool|mixed
 347          */
 348         private function revisionCallback( $revision ) {
 349                 if ( isset( $this->mRevisionCallback ) ) {
 350                         return call_user_func_array( $this->mRevisionCallback,
 351                                         array( $revision, $this ) );
 352                 } else {
 353                         return false;
 354                 }
 355         }
 356
 357         /**
 358          * Notify the callback function of a new log item
 359          * @param $revision WikiRevision object
 360          * @return bool|mixed
 361          */
 362         private function logItemCallback( $revision ) {
 363                 if ( isset( $this->mLogItemCallback ) ) {
 364                         return call_user_func_array( $this->mLogItemCallback,
 365                                         array( $revision, $this ) );
 366                 } else {
 367                         return false;
 368                 }
 369         }
 370
 371         /**
 372          * Shouldn't something like this be built-in to XMLReader?
 373          * Fetches text contents of the current element, assuming
 374          * no sub-elements or such scary things.
 375          * @return string
 376          * @access private
 377          */
 378         private function nodeContents() {
 379                 if ( $this->reader->isEmptyElement ) {
 380                         return "";
 381                 }
 382                 $buffer = "";
 383                 while ( $this->reader->read() ) {
 384                         switch ( $this->reader->nodeType ) {
 385                         case XmlReader::TEXT:
 386                         case XmlReader::SIGNIFICANT_WHITESPACE:
 387                                 $buffer .= $this->reader->value;
 388                                 break;
 389                         case XmlReader::END_ELEMENT:
 390                                 return $buffer;
 391                         }
 392                 }
 393
 394                 $this->reader->close();
 395                 return '';
 396         }
 397
 398         # --------------
 399
 400         /** Left in for debugging */
 401         private function dumpElement() {
 402                 static $lookup = null;
 403                 if ( !$lookup ) {
 404                         $xmlReaderConstants = array(
 405                                 "NONE",
 406                                 "ELEMENT",
 407                                 "ATTRIBUTE",
 408                                 "TEXT",
 409                                 "CDATA",
 410                                 "ENTITY_REF",
 411                                 "ENTITY",
 412                                 "PI",
 413                                 "COMMENT",
 414                                 "DOC",
 415                                 "DOC_TYPE",
 416                                 "DOC_FRAGMENT",
 417                                 "NOTATION",
 418                                 "WHITESPACE",
 419                                 "SIGNIFICANT_WHITESPACE",
 420                                 "END_ELEMENT",
 421                                 "END_ENTITY",
 422                                 "XML_DECLARATION",
 423                         );
 424                         $lookup = array();
 425
 426                         foreach ( $xmlReaderConstants as $name ) {
 427                                 $lookup[constant( "XmlReader::$name" )] = $name;
 428                         }
 429                 }
 430
 431                 print var_dump(
 432                         $lookup[$this->reader->nodeType],
 433                         $this->reader->name,
 434                         $this->reader->value
 435                 ) . "\n\n";
 436         }
 437
 438         /**
 439          * Primary entry point
 440          * @throws MWException
 441          * @return bool
 442          */
 443         public function doImport() {
 444
 445                 // Calls to reader->read need to be wrapped in calls to
 446                 // libxml_disable_entity_loader() to avoid local file
 447                 // inclusion attacks (bug 46932).
 448                 $oldDisable = libxml_disable_entity_loader( true );
 449                 $this->reader->read();
 450
 451                 if ( $this->reader->name != 'mediawiki' ) {
 452                         libxml_disable_entity_loader( $oldDisable );
 453                         throw new MWException( "Expected <mediawiki> tag, got " .
 454                                 $this->reader->name );
 455                 }
 456                 $this->debug( "<mediawiki> tag is correct." );
 457
 458                 $this->debug( "Starting primary dump processing loop." );
 459
 460                 $keepReading = $this->reader->read();
 461                 $skip = false;
 462                 while ( $keepReading ) {
 463                         $tag = $this->reader->name;
 464                         $type = $this->reader->nodeType;
 465
 466                         if ( !wfRunHooks( 'ImportHandleToplevelXMLTag', array( $this ) ) ) {
 467                                 // Do nothing
 468                         } elseif ( $tag == 'mediawiki' && $type == XmlReader::END_ELEMENT ) {
 469                                 break;
 470                         } elseif ( $tag == 'siteinfo' ) {
 471                                 $this->handleSiteInfo();
 472                         } elseif ( $tag == 'page' ) {
 473                                 $this->handlePage();
 474                         } elseif ( $tag == 'logitem' ) {
 475                                 $this->handleLogItem();
 476                         } elseif ( $tag != '#text' ) {
 477                                 $this->warn( "Unhandled top-level XML tag $tag" );
 478
 479                                 $skip = true;
 480                         }
 481
 482                         if ( $skip ) {
 483                                 $keepReading = $this->reader->next();
 484                                 $skip = false;
 485                                 $this->debug( "Skip" );
 486                         } else {
 487                                 $keepReading = $this->reader->read();
 488                         }
 489                 }
 490
 491                 libxml_disable_entity_loader( $oldDisable );
 492                 return true;
 493         }
 494
 495         /**
 496          * @return bool
 497          * @throws MWException
 498          */
 499         private function handleSiteInfo() {
 500                 // Site info is useful, but not actually used for dump imports.
 501                 // Includes a quick short-circuit to save performance.
 502                 if ( ! $this->mSiteInfoCallback ) {
 503                         $this->reader->next();
 504                         return true;
 505                 }
 506                 throw new MWException( "SiteInfo tag is not yet handled, do not set mSiteInfoCallback" );
 507         }
 508
 509         private function handleLogItem() {
 510                 $this->debug( "Enter log item handler." );
 511                 $logInfo = array();
 512
 513                 // Fields that can just be stuffed in the pageInfo object
 514                 $normalFields = array( 'id', 'comment', 'type', 'action', 'timestamp',
 515                                         'logtitle', 'params' );
 516
 517                 while ( $this->reader->read() ) {
 518                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 519                                         $this->reader->name == 'logitem' ) {
 520                                 break;
 521                         }
 522
 523                         $tag = $this->reader->name;
 524
 525                         if ( !wfRunHooks( 'ImportHandleLogItemXMLTag', array(
 526                                 $this, $logInfo
 527                         ) ) ) {
 528                                 // Do nothing
 529                         } elseif ( in_array( $tag, $normalFields ) ) {
 530                                 $logInfo[$tag] = $this->nodeContents();
 531                         } elseif ( $tag == 'contributor' ) {
 532                                 $logInfo['contributor'] = $this->handleContributor();
 533                         } elseif ( $tag != '#text' ) {
 534                                 $this->warn( "Unhandled log-item XML tag $tag" );
 535                         }
 536                 }
 537
 538                 $this->processLogItem( $logInfo );
 539         }
 540
 541         /**
 542          * @param $logInfo
 543          * @return bool|mixed
 544          */
 545         private function processLogItem( $logInfo ) {
 546                 $revision = new WikiRevision;
 547
 548                 $revision->setID( $logInfo['id'] );
 549                 $revision->setType( $logInfo['type'] );
 550                 $revision->setAction( $logInfo['action'] );
 551                 $revision->setTimestamp( $logInfo['timestamp'] );
 552                 $revision->setParams( $logInfo['params'] );
 553                 $revision->setTitle( Title::newFromText( $logInfo['logtitle'] ) );
 554                 $revision->setNoUpdates( $this->mNoUpdates );
 555
 556                 if ( isset( $logInfo['comment'] ) ) {
 557                         $revision->setComment( $logInfo['comment'] );
 558                 }
 559
 560                 if ( isset( $logInfo['contributor']['ip'] ) ) {
 561                         $revision->setUserIP( $logInfo['contributor']['ip'] );
 562                 }
 563                 if ( isset( $logInfo['contributor']['username'] ) ) {
 564                         $revision->setUserName( $logInfo['contributor']['username'] );
 565                 }
 566
 567                 return $this->logItemCallback( $revision );
 568         }
 569
 570         private function handlePage() {
 571                 // Handle page data.
 572                 $this->debug( "Enter page handler." );
 573                 $pageInfo = array( 'revisionCount' => 0, 'successfulRevisionCount' => 0 );
 574
 575                 // Fields that can just be stuffed in the pageInfo object
 576                 $normalFields = array( 'title', 'id', 'redirect', 'restrictions' );
 577
 578                 $skip = false;
 579                 $badTitle = false;
 580
 581                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 582                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 583                                         $this->reader->name == 'page' ) {
 584                                 break;
 585                         }
 586
 587                         $tag = $this->reader->name;
 588
 589                         if ( $badTitle ) {
 590                                 // The title is invalid, bail out of this page
 591                                 $skip = true;
 592                         } elseif ( !wfRunHooks( 'ImportHandlePageXMLTag', array( $this,
 593                                                 &$pageInfo ) ) ) {
 594                                 // Do nothing
 595                         } elseif ( in_array( $tag, $normalFields ) ) {
 596                                 $pageInfo[$tag] = $this->nodeContents();
 597                                 if ( $tag == 'title' ) {
 598                                         $title = $this->processTitle( $pageInfo['title'] );
 599
 600                                         if ( !$title ) {
 601                                                 $badTitle = true;
 602                                                 $skip = true;
 603                                         }
 604
 605                                         $this->pageCallback( $title );
 606                                         list( $pageInfo['_title'], $origTitle ) = $title;
 607                                 }
 608                         } elseif ( $tag == 'revision' ) {
 609                                 $this->handleRevision( $pageInfo );
 610                         } elseif ( $tag == 'upload' ) {
 611                                 $this->handleUpload( $pageInfo );
 612                         } elseif ( $tag != '#text' ) {
 613                                 $this->warn( "Unhandled page XML tag $tag" );
 614                                 $skip = true;
 615                         }
 616                 }
 617
 618                 $this->pageOutCallback( $pageInfo['_title'], $origTitle,
 619                                         $pageInfo['revisionCount'],
 620                                         $pageInfo['successfulRevisionCount'],
 621                                         $pageInfo );
 622         }
 623
 624         /**
 625          * @param $pageInfo array
 626          */
 627         private function handleRevision( &$pageInfo ) {
 628                 $this->debug( "Enter revision handler" );
 629                 $revisionInfo = array();
 630
 631                 $normalFields = array( 'id', 'timestamp', 'comment', 'minor', 'model', 'format', 'text' );
 632
 633                 $skip = false;
 634
 635                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 636                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 637                                         $this->reader->name == 'revision' ) {
 638                                 break;
 639                         }
 640
 641                         $tag = $this->reader->name;
 642
 643                         if ( !wfRunHooks( 'ImportHandleRevisionXMLTag', array(
 644                                 $this, $pageInfo, $revisionInfo
 645                         ) ) ) {
 646                                 // Do nothing
 647                         } elseif ( in_array( $tag, $normalFields ) ) {
 648                                 $revisionInfo[$tag] = $this->nodeContents();
 649                         } elseif ( $tag == 'contributor' ) {
 650                                 $revisionInfo['contributor'] = $this->handleContributor();
 651                         } elseif ( $tag != '#text' ) {
 652                                 $this->warn( "Unhandled revision XML tag $tag" );
 653                                 $skip = true;
 654                         }
 655                 }
 656
 657                 $pageInfo['revisionCount']++;
 658                 if ( $this->processRevision( $pageInfo, $revisionInfo ) ) {
 659                         $pageInfo['successfulRevisionCount']++;
 660                 }
 661         }
 662
 663         /**
 664          * @param $pageInfo
 665          * @param $revisionInfo
 666          * @return bool|mixed
 667          */
 668         private function processRevision( $pageInfo, $revisionInfo ) {
 669                 $revision = new WikiRevision;
 670
 671                 if ( isset( $revisionInfo['id'] ) ) {
 672                         $revision->setID( $revisionInfo['id'] );
 673                 }
 674                 if ( isset( $revisionInfo['text'] ) ) {
 675                         $revision->setText( $revisionInfo['text'] );
 676                 }
 677                 if ( isset( $revisionInfo['model'] ) ) {
 678                         $revision->setModel( $revisionInfo['model'] );
 679                 }
 680                 if ( isset( $revisionInfo['format'] ) ) {
 681                         $revision->setFormat( $revisionInfo['format'] );
 682                 }
 683                 $revision->setTitle( $pageInfo['_title'] );
 684
 685                 if ( isset( $revisionInfo['timestamp'] ) ) {
 686                         $revision->setTimestamp( $revisionInfo['timestamp'] );
 687                 } else {
 688                         $revision->setTimestamp( wfTimestampNow() );
 689                 }
 690
 691                 if ( isset( $revisionInfo['comment'] ) ) {
 692                         $revision->setComment( $revisionInfo['comment'] );
 693                 }
 694
 695                 if ( isset( $revisionInfo['minor'] ) ) {
 696                         $revision->setMinor( true );
 697                 }
 698                 if ( isset( $revisionInfo['contributor']['ip'] ) ) {
 699                         $revision->setUserIP( $revisionInfo['contributor']['ip'] );
 700                 }
 701                 if ( isset( $revisionInfo['contributor']['username'] ) ) {
 702                         $revision->setUserName( $revisionInfo['contributor']['username'] );
 703                 }
 704                 $revision->setNoUpdates( $this->mNoUpdates );
 705
 706                 return $this->revisionCallback( $revision );
 707         }
 708
 709         /**
 710          * @param $pageInfo
 711          * @return mixed
 712          */
 713         private function handleUpload( &$pageInfo ) {
 714                 $this->debug( "Enter upload handler" );
 715                 $uploadInfo = array();
 716
 717                 $normalFields = array( 'timestamp', 'comment', 'filename', 'text',
 718                                         'src', 'size', 'sha1base36', 'archivename', 'rel' );
 719
 720                 $skip = false;
 721
 722                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 723                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 724                                         $this->reader->name == 'upload' ) {
 725                                 break;
 726                         }
 727
 728                         $tag = $this->reader->name;
 729
 730                         if ( !wfRunHooks( 'ImportHandleUploadXMLTag', array(
 731                                 $this, $pageInfo
 732                         ) ) ) {
 733                                 // Do nothing
 734                         } elseif ( in_array( $tag, $normalFields ) ) {
 735                                 $uploadInfo[$tag] = $this->nodeContents();
 736                         } elseif ( $tag == 'contributor' ) {
 737                                 $uploadInfo['contributor'] = $this->handleContributor();
 738                         } elseif ( $tag == 'contents' ) {
 739                                 $contents = $this->nodeContents();
 740                                 $encoding = $this->reader->getAttribute( 'encoding' );
 741                                 if ( $encoding === 'base64' ) {
 742                                         $uploadInfo['fileSrc'] = $this->dumpTemp( base64_decode( $contents ) );
 743                                         $uploadInfo['isTempSrc'] = true;
 744                                 }
 745                         } elseif ( $tag != '#text' ) {
 746                                 $this->warn( "Unhandled upload XML tag $tag" );
 747                                 $skip = true;
 748                         }
 749                 }
 750
 751                 if ( $this->mImageBasePath && isset( $uploadInfo['rel'] ) ) {
 752                         $path = "{$this->mImageBasePath}/{$uploadInfo['rel']}";
 753                         if ( file_exists( $path ) ) {
 754                                 $uploadInfo['fileSrc'] = $path;
 755                                 $uploadInfo['isTempSrc'] = false;
 756                         }
 757                 }
 758
 759                 if ( $this->mImportUploads ) {
 760                         return $this->processUpload( $pageInfo, $uploadInfo );
 761                 }
 762         }
 763
 764         /**
 765          * @param $contents
 766          * @return string
 767          */
 768         private function dumpTemp( $contents ) {
 769                 $filename = tempnam( wfTempDir(), 'importupload' );
 770                 file_put_contents( $filename, $contents );
 771                 return $filename;
 772         }
 773
 774         /**
 775          * @param $pageInfo
 776          * @param $uploadInfo
 777          * @return mixed
 778          */
 779         private function processUpload( $pageInfo, $uploadInfo ) {
 780                 $revision = new WikiRevision;
 781                 $text = isset( $uploadInfo['text'] ) ? $uploadInfo['text'] : '';
 782
 783                 $revision->setTitle( $pageInfo['_title'] );
 784                 $revision->setID( $pageInfo['id'] );
 785                 $revision->setTimestamp( $uploadInfo['timestamp'] );
 786                 $revision->setText( $text );
 787                 $revision->setFilename( $uploadInfo['filename'] );
 788                 if ( isset( $uploadInfo['archivename'] ) ) {
 789                         $revision->setArchiveName( $uploadInfo['archivename'] );
 790                 }
 791                 $revision->setSrc( $uploadInfo['src'] );
 792                 if ( isset( $uploadInfo['fileSrc'] ) ) {
 793                         $revision->setFileSrc( $uploadInfo['fileSrc'],
 794                                 !empty( $uploadInfo['isTempSrc'] ) );
 795                 }
 796                 if ( isset( $uploadInfo['sha1base36'] ) ) {
 797                         $revision->setSha1Base36( $uploadInfo['sha1base36'] );
 798                 }
 799                 $revision->setSize( intval( $uploadInfo['size'] ) );
 800                 $revision->setComment( $uploadInfo['comment'] );
 801
 802                 if ( isset( $uploadInfo['contributor']['ip'] ) ) {
 803                         $revision->setUserIP( $uploadInfo['contributor']['ip'] );
 804                 }
 805                 if ( isset( $uploadInfo['contributor']['username'] ) ) {
 806                         $revision->setUserName( $uploadInfo['contributor']['username'] );
 807                 }
 808                 $revision->setNoUpdates( $this->mNoUpdates );
 809
 810                 return call_user_func( $this->mUploadCallback, $revision );
 811         }
 812
 813         /**
 814          * @return array
 815          */
 816         private function handleContributor() {
 817                 $fields = array( 'id', 'ip', 'username' );
 818                 $info = array();
 819
 820                 while ( $this->reader->read() ) {
 821                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 822                                         $this->reader->name == 'contributor' ) {
 823                                 break;
 824                         }
 825
 826                         $tag = $this->reader->name;
 827
 828                         if ( in_array( $tag, $fields ) ) {
 829                                 $info[$tag] = $this->nodeContents();
 830                         }
 831                 }
 832
 833                 return $info;
 834         }
 835
 836         /**
 837          * @param $text string
 838          * @return Array or false
 839          */
 840         private function processTitle( $text ) {
 841                 global $wgCommandLineMode;
 842
 843                 $workTitle = $text;
 844                 $origTitle = Title::newFromText( $workTitle );
 845
 846                 if ( !is_null( $this->mTargetNamespace ) && !is_null( $origTitle ) ) {
 847                         # makeTitleSafe, because $origTitle can have a interwiki (different setting of interwiki map)
 848                         # and than dbKey can begin with a lowercase char
 849                         $title = Title::makeTitleSafe( $this->mTargetNamespace,
 850                                 $origTitle->getDBkey() );
 851                 } else {
 852                         if ( !is_null( $this->mTargetRootPage ) ) {
 853                                 $workTitle = $this->mTargetRootPage . '/' . $workTitle;
 854                         }
 855                         $title = Title::newFromText( $workTitle );
 856                 }
 857
 858                 if ( is_null( $title ) ) {
 859                         # Invalid page title? Ignore the page
 860                         $this->notice( 'import-error-invalid', $workTitle );
 861                         return false;
 862                 } elseif ( $title->isExternal() ) {
 863                         $this->notice( 'import-error-interwiki', $title->getPrefixedText() );
 864                         return false;
 865                 } elseif ( !$title->canExist() ) {
 866                         $this->notice( 'import-error-special', $title->getPrefixedText() );
 867                         return false;
 868                 } elseif ( !$title->userCan( 'edit' ) && !$wgCommandLineMode ) {
 869                         # Do not import if the importing wiki user cannot edit this page
 870                         $this->notice( 'import-error-edit', $title->getPrefixedText() );
 871                         return false;
 872                 } elseif ( !$title->exists() && !$title->userCan( 'create' ) && !$wgCommandLineMode ) {
 873                         # Do not import if the importing wiki user cannot create this page
 874                         $this->notice( 'import-error-create', $title->getPrefixedText() );
 875                         return false;
 876                 }
 877
 878                 return array( $title, $origTitle );
 879         }
 880 }
 881
 882 /** This is a horrible hack used to keep source compatibility */
 883 class UploadSourceAdapter {
 884         static $sourceRegistrations = array();
 885
 886         private $mSource;
 887         private $mBuffer;
 888         private $mPosition;
 889
 890         /**
 891          * @param $source
 892          * @return string
 893          */
 894         static function registerSource( $source ) {
 895                 $id = wfRandomString();
 896
 897                 self::$sourceRegistrations[$id] = $source;
 898
 899                 return $id;
 900         }
 901
 902         /**
 903          * @param $path
 904          * @param $mode
 905          * @param $options
 906          * @param $opened_path
 907          * @return bool
 908          */
 909         function stream_open( $path, $mode, $options, &$opened_path ) {
 910                 $url = parse_url( $path );
 911                 $id = $url['host'];
 912
 913                 if ( !isset( self::$sourceRegistrations[$id] ) ) {
 914                         return false;
 915                 }
 916
 917                 $this->mSource = self::$sourceRegistrations[$id];
 918
 919                 return true;
 920         }
 921
 922         /**
 923          * @param $count
 924          * @return string
 925          */
 926         function stream_read( $count ) {
 927                 $return = '';
 928                 $leave = false;
 929
 930                 while ( !$leave && !$this->mSource->atEnd() &&
 931                                 strlen( $this->mBuffer ) < $count ) {
 932                         $read = $this->mSource->readChunk();
 933
 934                         if ( !strlen( $read ) ) {
 935                                 $leave = true;
 936                         }
 937
 938                         $this->mBuffer .= $read;
 939                 }
 940
 941                 if ( strlen( $this->mBuffer ) ) {
 942                         $return = substr( $this->mBuffer, 0, $count );
 943                         $this->mBuffer = substr( $this->mBuffer, $count );
 944                 }
 945
 946                 $this->mPosition += strlen( $return );
 947
 948                 return $return;
 949         }
 950
 951         /**
 952          * @param $data
 953          * @return bool
 954          */
 955         function stream_write( $data ) {
 956                 return false;
 957         }
 958
 959         /**
 960          * @return mixed
 961          */
 962         function stream_tell() {
 963                 return $this->mPosition;
 964         }
 965
 966         /**
 967          * @return bool
 968          */
 969         function stream_eof() {
 970                 return $this->mSource->atEnd();
 971         }
 972
 973         /**
 974          * @return array
 975          */
 976         function url_stat() {
 977                 $result = array();
 978
 979                 $result['dev'] = $result[0] = 0;
 980                 $result['ino'] = $result[1] = 0;
 981                 $result['mode'] = $result[2] = 0;
 982                 $result['nlink'] = $result[3] = 0;
 983                 $result['uid'] = $result[4] = 0;
 984                 $result['gid'] = $result[5] = 0;
 985                 $result['rdev'] = $result[6] = 0;
 986                 $result['size'] = $result[7] = 0;
 987                 $result['atime'] = $result[8] = 0;
 988                 $result['mtime'] = $result[9] = 0;
 989                 $result['ctime'] = $result[10] = 0;
 990                 $result['blksize'] = $result[11] = 0;
 991                 $result['blocks'] = $result[12] = 0;
 992
 993                 return $result;
 994         }
 995 }
 996
 997 class XMLReader2 extends XMLReader {
 998
 999         /**
1000          * @return bool|string
1001          */
1002         function nodeContents() {
1003                 if ( $this->isEmptyElement ) {
1004                         return "";
1005                 }
1006                 $buffer = "";
1007                 while ( $this->read() ) {
1008                         switch ( $this->nodeType ) {
1009                         case XmlReader::TEXT:
1010                         case XmlReader::SIGNIFICANT_WHITESPACE:
1011                                 $buffer .= $this->value;
1012                                 break;
1013                         case XmlReader::END_ELEMENT:
1014                                 return $buffer;
1015                         }
1016                 }
1017                 return $this->close();
1018         }
1019 }
1020
1021 /**
1022  * @todo document (e.g. one-sentence class description).
1023  * @ingroup SpecialPage
1024  */
1025 class WikiRevision {
1026         var $importer = null;
1027
1028         /**
1029          * @var Title
1030          */
1031         var $title = null;
1032         var $id = 0;
1033         var $timestamp = "20010115000000";
1034         var $user = 0;
1035         var $user_text = "";
1036         var $model = null;
1037         var $format = null;
1038         var $text = "";
1039         var $content = null;
1040         var $comment = "";
1041         var $minor = false;
1042         var $type = "";
1043         var $action = "";
1044         var $params = "";
1045         var $fileSrc = '';
1046         var $sha1base36 = false;
1047         var $isTemp = false;
1048         var $archiveName = '';
1049         var $fileIsTemp;
1050         private $mNoUpdates = false;
1051
1052         /**
1053          * @param $title
1054          * @throws MWException
1055          */
1056         function setTitle( $title ) {
1057                 if ( is_object( $title ) ) {
1058                         $this->title = $title;
1059                 } elseif ( is_null( $title ) ) {
1060                         throw new MWException( "WikiRevision given a null title in import. You may need to adjust \$wgLegalTitleChars." );
1061                 } else {
1062                         throw new MWException( "WikiRevision given non-object title in import." );
1063                 }
1064         }
1065
1066         /**
1067          * @param $id
1068          */
1069         function setID( $id ) {
1070                 $this->id = $id;
1071         }
1072
1073         /**
1074          * @param $ts
1075          */
1076         function setTimestamp( $ts ) {
1077                 # 2003-08-05T18:30:02Z
1078                 $this->timestamp = wfTimestamp( TS_MW, $ts );
1079         }
1080
1081         /**
1082          * @param $user
1083          */
1084         function setUsername( $user ) {
1085                 $this->user_text = $user;
1086         }
1087
1088         /**
1089          * @param $ip
1090          */
1091         function setUserIP( $ip ) {
1092                 $this->user_text = $ip;
1093         }
1094
1095         /**
1096          * @param $model
1097          */
1098         function setModel( $model ) {
1099                 $this->model = $model;
1100         }
1101
1102         /**
1103          * @param $format
1104          */
1105         function setFormat( $format ) {
1106                 $this->format = $format;
1107         }
1108
1109         /**
1110          * @param $text
1111          */
1112         function setText( $text ) {
1113                 $this->text = $text;
1114         }
1115
1116         /**
1117          * @param $text
1118          */
1119         function setComment( $text ) {
1120                 $this->comment = $text;
1121         }
1122
1123         /**
1124          * @param $minor
1125          */
1126         function setMinor( $minor ) {
1127                 $this->minor = (bool)$minor;
1128         }
1129
1130         /**
1131          * @param $src
1132          */
1133         function setSrc( $src ) {
1134                 $this->src = $src;
1135         }
1136
1137         /**
1138          * @param $src
1139          * @param $isTemp
1140          */
1141         function setFileSrc( $src, $isTemp ) {
1142                 $this->fileSrc = $src;
1143                 $this->fileIsTemp = $isTemp;
1144         }
1145
1146         /**
1147          * @param $sha1base36
1148          */
1149         function setSha1Base36( $sha1base36 ) {
1150                 $this->sha1base36 = $sha1base36;
1151         }
1152
1153         /**
1154          * @param $filename
1155          */
1156         function setFilename( $filename ) {
1157                 $this->filename = $filename;
1158         }
1159
1160         /**
1161          * @param $archiveName
1162          */
1163         function setArchiveName( $archiveName ) {
1164                 $this->archiveName = $archiveName;
1165         }
1166
1167         /**
1168          * @param $size
1169          */
1170         function setSize( $size ) {
1171                 $this->size = intval( $size );
1172         }
1173
1174         /**
1175          * @param $type
1176          */
1177         function setType( $type ) {
1178                 $this->type = $type;
1179         }
1180
1181         /**
1182          * @param $action
1183          */
1184         function setAction( $action ) {
1185                 $this->action = $action;
1186         }
1187
1188         /**
1189          * @param $params
1190          */
1191         function setParams( $params ) {
1192                 $this->params = $params;
1193         }
1194
1195         /**
1196          * @param $noupdates
1197          */
1198         public function setNoUpdates( $noupdates ) {
1199                 $this->mNoUpdates = $noupdates;
1200         }
1201
1202         /**
1203          * @return Title
1204          */
1205         function getTitle() {
1206                 return $this->title;
1207         }
1208
1209         /**
1210          * @return int
1211          */
1212         function getID() {
1213                 return $this->id;
1214         }
1215
1216         /**
1217          * @return string
1218          */
1219         function getTimestamp() {
1220                 return $this->timestamp;
1221         }
1222
1223         /**
1224          * @return string
1225          */
1226         function getUser() {
1227                 return $this->user_text;
1228         }
1229
1230         /**
1231          * @return string
1232          *
1233          * @deprecated Since 1.21, use getContent() instead.
1234          */
1235         function getText() {
1236                 ContentHandler::deprecated( __METHOD__, '1.21' );
1237
1238                 return $this->text;
1239         }
1240
1241         /**
1242          * @return Content
1243          */
1244         function getContent() {
1245                 if ( is_null( $this->content ) ) {
1246                         $this->content =
1247                                 ContentHandler::makeContent(
1248                                         $this->text,
1249                                         $this->getTitle(),
1250                                         $this->getModel(),
1251                                         $this->getFormat()
1252                                 );
1253                 }
1254
1255                 return $this->content;
1256         }
1257
1258         /**
1259          * @return String
1260          */
1261         function getModel() {
1262                 if ( is_null( $this->model ) ) {
1263                         $this->model = $this->getTitle()->getContentModel();
1264                 }
1265
1266                 return $this->model;
1267         }
1268
1269         /**
1270          * @return String
1271          */
1272         function getFormat() {
1273                 if ( is_null( $this->model ) ) {
1274                         $this->format = ContentHandler::getForTitle( $this->getTitle() )->getDefaultFormat();
1275                 }
1276
1277                 return $this->format;
1278         }
1279
1280         /**
1281          * @return string
1282          */
1283         function getComment() {
1284                 return $this->comment;
1285         }
1286
1287         /**
1288          * @return bool
1289          */
1290         function getMinor() {
1291                 return $this->minor;
1292         }
1293
1294         /**
1295          * @return mixed
1296          */
1297         function getSrc() {
1298                 return $this->src;
1299         }
1300
1301         /**
1302          * @return bool|String
1303          */
1304         function getSha1() {
1305                 if ( $this->sha1base36 ) {
1306                         return wfBaseConvert( $this->sha1base36, 36, 16 );
1307                 }
1308                 return false;
1309         }
1310
1311         /**
1312          * @return string
1313          */
1314         function getFileSrc() {
1315                 return $this->fileSrc;
1316         }
1317
1318         /**
1319          * @return bool
1320          */
1321         function isTempSrc() {
1322                 return $this->isTemp;
1323         }
1324
1325         /**
1326          * @return mixed
1327          */
1328         function getFilename() {
1329                 return $this->filename;
1330         }
1331
1332         /**
1333          * @return string
1334          */
1335         function getArchiveName() {
1336                 return $this->archiveName;
1337         }
1338
1339         /**
1340          * @return mixed
1341          */
1342         function getSize() {
1343                 return $this->size;
1344         }
1345
1346         /**
1347          * @return string
1348          */
1349         function getType() {
1350                 return $this->type;
1351         }
1352
1353         /**
1354          * @return string
1355          */
1356         function getAction() {
1357                 return $this->action;
1358         }
1359
1360         /**
1361          * @return string
1362          */
1363         function getParams() {
1364                 return $this->params;
1365         }
1366
1367         /**
1368          * @return bool
1369          */
1370         function importOldRevision() {
1371                 $dbw = wfGetDB( DB_MASTER );
1372
1373                 # Sneak a single revision into place
1374                 $user = User::newFromName( $this->getUser() );
1375                 if ( $user ) {
1376                         $userId = intval( $user->getId() );
1377                         $userText = $user->getName();
1378                         $userObj = $user;
1379                 } else {
1380                         $userId = 0;
1381                         $userText = $this->getUser();
1382                         $userObj = new User;
1383                 }
1384
1385                 // avoid memory leak...?
1386                 $linkCache = LinkCache::singleton();
1387                 $linkCache->clear();
1388
1389                 $page = WikiPage::factory( $this->title );
1390                 if ( !$page->exists() ) {
1391                         # must create the page...
1392                         $pageId = $page->insertOn( $dbw );
1393                         $created = true;
1394                         $oldcountable = null;
1395                 } else {
1396                         $pageId = $page->getId();
1397                         $created = false;
1398
1399                         $prior = $dbw->selectField( 'revision', '1',
1400                                 array( 'rev_page' => $pageId,
1401                                         'rev_timestamp' => $dbw->timestamp( $this->timestamp ),
1402                                         'rev_user_text' => $userText,
1403                                         'rev_comment' => $this->getComment() ),
1404                                 __METHOD__
1405                         );
1406                         if ( $prior ) {
1407                                 // @todo FIXME: This could fail slightly for multiple matches :P
1408                                 wfDebug( __METHOD__ . ": skipping existing revision for [[" .
1409                                         $this->title->getPrefixedText() . "]], timestamp " . $this->timestamp . "\n" );
1410                                 return false;
1411                         }
1412                         $oldcountable = $page->isCountable();
1413                 }
1414
1415                 # @todo FIXME: Use original rev_id optionally (better for backups)
1416                 # Insert the row
1417                 $revision = new Revision( array(
1418                         'title' => $this->title,
1419                         'page' => $pageId,
1420                         'content_model' => $this->getModel(),
1421                         'content_format' => $this->getFormat(),
1422                         'text' => $this->getContent()->serialize( $this->getFormat() ), //XXX: just set 'content' => $this->getContent()?
1423                         'comment' => $this->getComment(),
1424                         'user' => $userId,
1425                         'user_text' => $userText,
1426                         'timestamp' => $this->timestamp,
1427                         'minor_edit' => $this->minor,
1428                         ) );
1429                 $revision->insertOn( $dbw );
1430                 $changed = $page->updateIfNewerOn( $dbw, $revision );
1431
1432                 if ( $changed !== false && !$this->mNoUpdates ) {
1433                         wfDebug( __METHOD__ . ": running updates\n" );
1434                         $page->doEditUpdates( $revision, $userObj, array( 'created' => $created, 'oldcountable' => $oldcountable ) );
1435                 }
1436
1437                 return true;
1438         }
1439
1440         /**
1441          * @return mixed
1442          */
1443         function importLogItem() {
1444                 $dbw = wfGetDB( DB_MASTER );
1445                 # @todo FIXME: This will not record autoblocks
1446                 if ( !$this->getTitle() ) {
1447                         wfDebug( __METHOD__ . ": skipping invalid {$this->type}/{$this->action} log time, timestamp " .
1448                                 $this->timestamp . "\n" );
1449                         return;
1450                 }
1451                 # Check if it exists already
1452                 // @todo FIXME: Use original log ID (better for backups)
1453                 $prior = $dbw->selectField( 'logging', '1',
1454                         array( 'log_type' => $this->getType(),
1455                                 'log_action' => $this->getAction(),
1456                                 'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1457                                 'log_namespace' => $this->getTitle()->getNamespace(),
1458                                 'log_title' => $this->getTitle()->getDBkey(),
1459                                 'log_comment' => $this->getComment(),
1460                                 #'log_user_text' => $this->user_text,
1461                                 'log_params' => $this->params ),
1462                         __METHOD__
1463                 );
1464                 // @todo FIXME: This could fail slightly for multiple matches :P
1465                 if ( $prior ) {
1466                         wfDebug( __METHOD__ . ": skipping existing item for Log:{$this->type}/{$this->action}, timestamp " .
1467                                 $this->timestamp . "\n" );
1468                         return;
1469                 }
1470                 $log_id = $dbw->nextSequenceValue( 'logging_log_id_seq' );
1471                 $data = array(
1472                         'log_id' => $log_id,
1473                         'log_type' => $this->type,
1474                         'log_action' => $this->action,
1475                         'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1476                         'log_user' => User::idFromName( $this->user_text ),
1477                         #'log_user_text' => $this->user_text,
1478                         'log_namespace' => $this->getTitle()->getNamespace(),
1479                         'log_title' => $this->getTitle()->getDBkey(),
1480                         'log_comment' => $this->getComment(),
1481                         'log_params' => $this->params
1482                 );
1483                 $dbw->insert( 'logging', $data, __METHOD__ );
1484         }
1485
1486         /**
1487          * @return bool
1488          */
1489         function importUpload() {
1490                 # Construct a file
1491                 $archiveName = $this->getArchiveName();
1492                 if ( $archiveName ) {
1493                         wfDebug( __METHOD__ . "Importing archived file as $archiveName\n" );
1494                         $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1495                                 RepoGroup::singleton()->getLocalRepo(), $archiveName );
1496                 } else {
1497                         $file = wfLocalFile( $this->getTitle() );
1498                         wfDebug( __METHOD__ . 'Importing new file as ' . $file->getName() . "\n" );
1499                         if ( $file->exists() && $file->getTimestamp() > $this->getTimestamp() ) {
1500                                 $archiveName = $file->getTimestamp() . '!' . $file->getName();
1501                                 $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1502                                         RepoGroup::singleton()->getLocalRepo(), $archiveName );
1503                                 wfDebug( __METHOD__ . "File already exists; importing as $archiveName\n" );
1504                         }
1505                 }
1506                 if ( !$file ) {
1507                         wfDebug( __METHOD__ . ': Bad file for ' . $this->getTitle() . "\n" );
1508                         return false;
1509                 }
1510
1511                 # Get the file source or download if necessary
1512                 $source = $this->getFileSrc();
1513                 $flags = $this->isTempSrc() ? File::DELETE_SOURCE : 0;
1514                 if ( !$source ) {
1515                         $source = $this->downloadSource();
1516                         $flags |= File::DELETE_SOURCE;
1517                 }
1518                 if ( !$source ) {
1519                         wfDebug( __METHOD__ . ": Could not fetch remote file.\n" );
1520                         return false;
1521                 }
1522                 $sha1 = $this->getSha1();
1523                 if ( $sha1 && ( $sha1 !== sha1_file( $source ) ) ) {
1524                         if ( $flags & File::DELETE_SOURCE ) {
1525                                 # Broken file; delete it if it is a temporary file
1526                                 unlink( $source );
1527                         }
1528                         wfDebug( __METHOD__ . ": Corrupt file $source.\n" );
1529                         return false;
1530                 }
1531
1532                 $user = User::newFromName( $this->user_text );
1533
1534                 # Do the actual upload
1535                 if ( $archiveName ) {
1536                         $status = $file->uploadOld( $source, $archiveName,
1537                                 $this->getTimestamp(), $this->getComment(), $user, $flags );
1538                 } else {
1539                         $status = $file->upload( $source, $this->getComment(), $this->getComment(),
1540                                 $flags, false, $this->getTimestamp(), $user );
1541                 }
1542
1543                 if ( $status->isGood() ) {
1544                         wfDebug( __METHOD__ . ": Successful\n" );
1545                         return true;
1546                 } else {
1547                         wfDebug( __METHOD__ . ': failed: ' . $status->getXml() . "\n" );
1548                         return false;
1549                 }
1550         }
1551
1552         /**
1553          * @return bool|string
1554          */
1555         function downloadSource() {
1556                 global $wgEnableUploads;
1557                 if ( !$wgEnableUploads ) {
1558                         return false;
1559                 }
1560
1561                 $tempo = tempnam( wfTempDir(), 'download' );
1562                 $f = fopen( $tempo, 'wb' );
1563                 if ( !$f ) {
1564                         wfDebug( "IMPORT: couldn't write to temp file $tempo\n" );
1565                         return false;
1566                 }
1567
1568                 // @todo FIXME!
1569                 $src = $this->getSrc();
1570                 $data = Http::get( $src );
1571                 if ( !$data ) {
1572                         wfDebug( "IMPORT: couldn't fetch source $src\n" );
1573                         fclose( $f );
1574                         unlink( $tempo );
1575                         return false;
1576                 }
1577
1578                 fwrite( $f, $data );
1579                 fclose( $f );
1580
1581                 return $tempo;
1582         }
1583
1584 }
1585
1586 /**
1587  * @todo document (e.g. one-sentence class description).
1588  * @ingroup SpecialPage
1589  */
1590 class ImportStringSource {
1591         function __construct( $string ) {
1592                 $this->mString = $string;
1593                 $this->mRead = false;
1594         }
1595
1596         /**
1597          * @return bool
1598          */
1599         function atEnd() {
1600                 return $this->mRead;
1601         }
1602
1603         /**
1604          * @return bool|string
1605          */
1606         function readChunk() {
1607                 if ( $this->atEnd() ) {
1608                         return false;
1609                 }
1610                 $this->mRead = true;
1611                 return $this->mString;
1612         }
1613 }
1614
1615 /**
1616  * @todo document (e.g. one-sentence class description).
1617  * @ingroup SpecialPage
1618  */
1619 class ImportStreamSource {
1620         function __construct( $handle ) {
1621                 $this->mHandle = $handle;
1622         }
1623
1624         /**
1625          * @return bool
1626          */
1627         function atEnd() {
1628                 return feof( $this->mHandle );
1629         }
1630
1631         /**
1632          * @return string
1633          */
1634         function readChunk() {
1635                 return fread( $this->mHandle, 32768 );
1636         }
1637
1638         /**
1639          * @param $filename string
1640          * @return Status
1641          */
1642         static function newFromFile( $filename ) {
1643                 wfSuppressWarnings();
1644                 $file = fopen( $filename, 'rt' );
1645                 wfRestoreWarnings();
1646                 if ( !$file ) {
1647                         return Status::newFatal( "importcantopen" );
1648                 }
1649                 return Status::newGood( new ImportStreamSource( $file ) );
1650         }
1651
1652         /**
1653          * @param $fieldname string
1654          * @return Status
1655          */
1656         static function newFromUpload( $fieldname = "xmlimport" ) {
1657                 $upload =& $_FILES[$fieldname];
1658
1659                 if ( $upload === null || !$upload['name'] ) {
1660                         return Status::newFatal( 'importnofile' );
1661                 }
1662                 if ( !empty( $upload['error'] ) ) {
1663                         switch ( $upload['error'] ) {
1664                                 case 1: # The uploaded file exceeds the upload_max_filesize directive in php.ini.
1665                                         return Status::newFatal( 'importuploaderrorsize' );
1666                                 case 2: # The uploaded file exceeds the MAX_FILE_SIZE directive that was specified in the HTML form.
1667                                         return Status::newFatal( 'importuploaderrorsize' );
1668                                 case 3: # The uploaded file was only partially uploaded
1669                                         return Status::newFatal( 'importuploaderrorpartial' );
1670                                 case 6: #Missing a temporary folder.
1671                                         return Status::newFatal( 'importuploaderrortemp' );
1672                                 # case else: # Currently impossible
1673                         }
1674
1675                 }
1676                 $fname = $upload['tmp_name'];
1677                 if ( is_uploaded_file( $fname ) ) {
1678                         return ImportStreamSource::newFromFile( $fname );
1679                 } else {
1680                         return Status::newFatal( 'importnofile' );
1681                 }
1682         }
1683
1684         /**
1685          * @param $url
1686          * @param $method string
1687          * @return Status
1688          */
1689         static function newFromURL( $url, $method = 'GET' ) {
1690                 wfDebug( __METHOD__ . ": opening $url\n" );
1691                 # Use the standard HTTP fetch function; it times out
1692                 # quicker and sorts out user-agent problems which might
1693                 # otherwise prevent importing from large sites, such
1694                 # as the Wikimedia cluster, etc.
1695                 $data = Http::request( $method, $url, array( 'followRedirects' => true ) );
1696                 if ( $data !== false ) {
1697                         $file = tmpfile();
1698                         fwrite( $file, $data );
1699                         fflush( $file );
1700                         fseek( $file, 0 );
1701                         return Status::newGood( new ImportStreamSource( $file ) );
1702                 } else {
1703                         return Status::newFatal( 'importcantopen' );
1704                 }
1705         }
1706
1707         /**
1708          * @param $interwiki
1709          * @param $page
1710          * @param $history bool
1711          * @param $templates bool
1712          * @param $pageLinkDepth int
1713          * @return Status
1714          */
1715         public static function newFromInterwiki( $interwiki, $page, $history = false, $templates = false, $pageLinkDepth = 0 ) {
1716                 if ( $page == '' ) {
1717                         return Status::newFatal( 'import-noarticle' );
1718                 }
1719                 $link = Title::newFromText( "$interwiki:Special:Export/$page" );
1720                 if ( is_null( $link ) || $link->getInterwiki() == '' ) {
1721                         return Status::newFatal( 'importbadinterwiki' );
1722                 } else {
1723                         $params = array();
1724                         if ( $history ) {
1725                                 $params['history'] = 1;
1726                         }
1727                         if ( $templates ) {
1728                                 $params['templates'] = 1;
1729                         }
1730                         if ( $pageLinkDepth ) {
1731                                 $params['pagelink-depth'] = $pageLinkDepth;
1732                         }
1733                         $url = $link->getFullURL( $params );
1734                         # For interwikis, use POST to avoid redirects.
1735                         return ImportStreamSource::newFromURL( $url, "POST" );
1736                 }
1737         }
1738 }