includes/Import.php

   1 <?php
   2 /**
   3  * MediaWiki page data importer.
   4  *
   5  * Copyright © 2003,2005 Brion Vibber <brion@pobox.com>
   6  * https://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 ImportStreamSource $source
  44          */
  45         function __construct( ImportStreamSource $source ) {
  46                 $this->reader = new XMLReader();
  47
  48                 if ( !in_array( 'uploadsource', stream_get_wrappers() ) ) {
  49                         stream_wrapper_register( 'uploadsource', 'UploadSourceAdapter' );
  50                 }
  51                 $id = UploadSourceAdapter::registerSource( $source );
  52                 if ( defined( 'LIBXML_PARSEHUGE' ) ) {
  53                         $this->reader->open( "uploadsource://$id", null, LIBXML_PARSEHUGE );
  54                 } else {
  55                         $this->reader->open( "uploadsource://$id" );
  56                 }
  57
  58                 // Default callbacks
  59                 $this->setRevisionCallback( array( $this, "importRevision" ) );
  60                 $this->setUploadCallback( array( $this, 'importUpload' ) );
  61                 $this->setLogItemCallback( array( $this, 'importLogItem' ) );
  62                 $this->setPageOutCallback( array( $this, 'finishImportPage' ) );
  63         }
  64
  65         /**
  66          * @return null|XMLReader
  67          */
  68         public function getReader() {
  69                 return $this->reader;
  70         }
  71
  72         public function throwXmlError( $err ) {
  73                 $this->debug( "FAILURE: $err" );
  74                 wfDebug( "WikiImporter XML error: $err\n" );
  75         }
  76
  77         public function debug( $data ) {
  78                 if ( $this->mDebug ) {
  79                         wfDebug( "IMPORT: $data\n" );
  80                 }
  81         }
  82
  83         public function warn( $data ) {
  84                 wfDebug( "IMPORT: $data\n" );
  85         }
  86
  87         public function notice( $msg /*, $param, ...*/ ) {
  88                 $params = func_get_args();
  89                 array_shift( $params );
  90
  91                 if ( is_callable( $this->mNoticeCallback ) ) {
  92                         call_user_func( $this->mNoticeCallback, $msg, $params );
  93                 } else { # No ImportReporter -> CLI
  94                         echo wfMessage( $msg, $params )->text() . "\n";
  95                 }
  96         }
  97
  98         /**
  99          * Set debug mode...
 100          * @param bool $debug
 101          */
 102         function setDebug( $debug ) {
 103                 $this->mDebug = $debug;
 104         }
 105
 106         /**
 107          * Set 'no updates' mode. In this mode, the link tables will not be updated by the importer
 108          * @param bool $noupdates
 109          */
 110         function setNoUpdates( $noupdates ) {
 111                 $this->mNoUpdates = $noupdates;
 112         }
 113
 114         /**
 115          * Set a callback that displays notice messages
 116          *
 117          * @param callable $callback
 118          * @return callable
 119          */
 120         public function setNoticeCallback( $callback ) {
 121                 return wfSetVar( $this->mNoticeCallback, $callback );
 122         }
 123
 124         /**
 125          * Sets the action to perform as each new page in the stream is reached.
 126          * @param callable $callback
 127          * @return callable
 128          */
 129         public function setPageCallback( $callback ) {
 130                 $previous = $this->mPageCallback;
 131                 $this->mPageCallback = $callback;
 132                 return $previous;
 133         }
 134
 135         /**
 136          * Sets the action to perform as each page in the stream is completed.
 137          * Callback accepts the page title (as a Title object), a second object
 138          * with the original title form (in case it's been overridden into a
 139          * local namespace), and a count of revisions.
 140          *
 141          * @param callable $callback
 142          * @return callable
 143          */
 144         public function setPageOutCallback( $callback ) {
 145                 $previous = $this->mPageOutCallback;
 146                 $this->mPageOutCallback = $callback;
 147                 return $previous;
 148         }
 149
 150         /**
 151          * Sets the action to perform as each page revision is reached.
 152          * @param callable $callback
 153          * @return callable
 154          */
 155         public function setRevisionCallback( $callback ) {
 156                 $previous = $this->mRevisionCallback;
 157                 $this->mRevisionCallback = $callback;
 158                 return $previous;
 159         }
 160
 161         /**
 162          * Sets the action to perform as each file upload version is reached.
 163          * @param callable $callback
 164          * @return callable
 165          */
 166         public function setUploadCallback( $callback ) {
 167                 $previous = $this->mUploadCallback;
 168                 $this->mUploadCallback = $callback;
 169                 return $previous;
 170         }
 171
 172         /**
 173          * Sets the action to perform as each log item reached.
 174          * @param callable $callback
 175          * @return callable
 176          */
 177         public function setLogItemCallback( $callback ) {
 178                 $previous = $this->mLogItemCallback;
 179                 $this->mLogItemCallback = $callback;
 180                 return $previous;
 181         }
 182
 183         /**
 184          * Sets the action to perform when site info is encountered
 185          * @param callable $callback
 186          * @return callable
 187          */
 188         public function setSiteInfoCallback( $callback ) {
 189                 $previous = $this->mSiteInfoCallback;
 190                 $this->mSiteInfoCallback = $callback;
 191                 return $previous;
 192         }
 193
 194         /**
 195          * Set a target namespace to override the defaults
 196          * @param null|int $namespace
 197          * @return bool
 198          */
 199         public function setTargetNamespace( $namespace ) {
 200                 if ( is_null( $namespace ) ) {
 201                         // Don't override namespaces
 202                         $this->mTargetNamespace = null;
 203                 } elseif ( $namespace >= 0 ) {
 204                         // @todo FIXME: Check for validity
 205                         $this->mTargetNamespace = intval( $namespace );
 206                 } else {
 207                         return false;
 208                 }
 209         }
 210
 211         /**
 212          * Set a target root page under which all pages are imported
 213          * @param null|string $rootpage
 214          * @return Status
 215          */
 216         public function setTargetRootPage( $rootpage ) {
 217                 $status = Status::newGood();
 218                 if ( is_null( $rootpage ) ) {
 219                         // No rootpage
 220                         $this->mTargetRootPage = null;
 221                 } elseif ( $rootpage !== '' ) {
 222                         $rootpage = rtrim( $rootpage, '/' ); //avoid double slashes
 223                         $title = Title::newFromText( $rootpage, !is_null( $this->mTargetNamespace )
 224                                 ? $this->mTargetNamespace
 225                                 : NS_MAIN
 226                         );
 227
 228                         if ( !$title || $title->isExternal() ) {
 229                                 $status->fatal( 'import-rootpage-invalid' );
 230                         } else {
 231                                 if ( !MWNamespace::hasSubpages( $title->getNamespace() ) ) {
 232                                         global $wgContLang;
 233
 234                                         $displayNSText = $title->getNamespace() == NS_MAIN
 235                                                 ? wfMessage( 'blanknamespace' )->text()
 236                                                 : $wgContLang->getNsText( $title->getNamespace() );
 237                                         $status->fatal( 'import-rootpage-nosubpage', $displayNSText );
 238                                 } else {
 239                                         // set namespace to 'all', so the namespace check in processTitle() can passed
 240                                         $this->setTargetNamespace( null );
 241                                         $this->mTargetRootPage = $title->getPrefixedDBkey();
 242                                 }
 243                         }
 244                 }
 245                 return $status;
 246         }
 247
 248         /**
 249          * @param string $dir
 250          */
 251         public function setImageBasePath( $dir ) {
 252                 $this->mImageBasePath = $dir;
 253         }
 254
 255         /**
 256          * @param bool $import
 257          */
 258         public function setImportUploads( $import ) {
 259                 $this->mImportUploads = $import;
 260         }
 261
 262         /**
 263          * Default per-revision callback, performs the import.
 264          * @param WikiRevision $revision
 265          * @return bool
 266          */
 267         public function importRevision( $revision ) {
 268                 if ( !$revision->getContentHandler()->canBeUsedOn( $revision->getTitle() ) ) {
 269                         $this->notice( 'import-error-bad-location',
 270                                 $revision->getTitle()->getPrefixedText(),
 271                                 $revision->getID(),
 272                                 $revision->getModel(),
 273                                 $revision->getFormat() );
 274
 275                         return false;
 276                 }
 277
 278                 try {
 279                         $dbw = wfGetDB( DB_MASTER );
 280                         return $dbw->deadlockLoop( array( $revision, 'importOldRevision' ) );
 281                 } catch ( MWContentSerializationException $ex ) {
 282                         $this->notice( 'import-error-unserialize',
 283                                 $revision->getTitle()->getPrefixedText(),
 284                                 $revision->getID(),
 285                                 $revision->getModel(),
 286                                 $revision->getFormat() );
 287                 }
 288
 289                 return false;
 290         }
 291
 292         /**
 293          * Default per-revision callback, performs the import.
 294          * @param WikiRevision $revision
 295          * @return bool
 296          */
 297         public function importLogItem( $revision ) {
 298                 $dbw = wfGetDB( DB_MASTER );
 299                 return $dbw->deadlockLoop( array( $revision, 'importLogItem' ) );
 300         }
 301
 302         /**
 303          * Dummy for now...
 304          * @param WikiRevision $revision
 305          * @return bool
 306          */
 307         public function importUpload( $revision ) {
 308                 $dbw = wfGetDB( DB_MASTER );
 309                 return $dbw->deadlockLoop( array( $revision, 'importUpload' ) );
 310         }
 311
 312         /**
 313          * Mostly for hook use
 314          * @param Title $title
 315          * @param string $origTitle
 316          * @param int $revCount
 317          * @param int $sRevCount
 318          * @param array $pageInfo
 319          * @return bool
 320          */
 321         public function finishImportPage( $title, $origTitle, $revCount, $sRevCount, $pageInfo ) {
 322                 $args = func_get_args();
 323                 return wfRunHooks( 'AfterImportPage', $args );
 324         }
 325
 326         /**
 327          * Alternate per-revision callback, for debugging.
 328          * @param WikiRevision $revision
 329          */
 330         public function debugRevisionHandler( &$revision ) {
 331                 $this->debug( "Got revision:" );
 332                 if ( is_object( $revision->title ) ) {
 333                         $this->debug( "-- Title: " . $revision->title->getPrefixedText() );
 334                 } else {
 335                         $this->debug( "-- Title: <invalid>" );
 336                 }
 337                 $this->debug( "-- User: " . $revision->user_text );
 338                 $this->debug( "-- Timestamp: " . $revision->timestamp );
 339                 $this->debug( "-- Comment: " . $revision->comment );
 340                 $this->debug( "-- Text: " . $revision->text );
 341         }
 342
 343         /**
 344          * Notify the callback function when a new "<page>" is reached.
 345          * @param Title $title
 346          */
 347         function pageCallback( $title ) {
 348                 if ( isset( $this->mPageCallback ) ) {
 349                         call_user_func( $this->mPageCallback, $title );
 350                 }
 351         }
 352
 353         /**
 354          * Notify the callback function when a "</page>" is closed.
 355          * @param Title $title
 356          * @param Title $origTitle
 357          * @param int $revCount
 358          * @param int $sucCount Number of revisions for which callback returned true
 359          * @param array $pageInfo Associative array of page information
 360          */
 361         private function pageOutCallback( $title, $origTitle, $revCount, $sucCount, $pageInfo ) {
 362                 if ( isset( $this->mPageOutCallback ) ) {
 363                         $args = func_get_args();
 364                         call_user_func_array( $this->mPageOutCallback, $args );
 365                 }
 366         }
 367
 368         /**
 369          * Notify the callback function of a revision
 370          * @param WikiRevision $revision
 371          * @return bool|mixed
 372          */
 373         private function revisionCallback( $revision ) {
 374                 if ( isset( $this->mRevisionCallback ) ) {
 375                         return call_user_func_array( $this->mRevisionCallback,
 376                                         array( $revision, $this ) );
 377                 } else {
 378                         return false;
 379                 }
 380         }
 381
 382         /**
 383          * Notify the callback function of a new log item
 384          * @param WikiRevision $revision
 385          * @return bool|mixed
 386          */
 387         private function logItemCallback( $revision ) {
 388                 if ( isset( $this->mLogItemCallback ) ) {
 389                         return call_user_func_array( $this->mLogItemCallback,
 390                                         array( $revision, $this ) );
 391                 } else {
 392                         return false;
 393                 }
 394         }
 395
 396         /**
 397          * Retrieves the contents of the named attribute of the current element.
 398          * @param string $attr The name of the attribute
 399          * @return string The value of the attribute or an empty string if it is not set in the current element.
 400          */
 401         public function nodeAttribute( $attr ) {
 402                 return $this->reader->getAttribute( $attr );
 403         }
 404
 405         /**
 406          * Shouldn't something like this be built-in to XMLReader?
 407          * Fetches text contents of the current element, assuming
 408          * no sub-elements or such scary things.
 409          * @return string
 410          * @access private
 411          */
 412         public function nodeContents() {
 413                 if ( $this->reader->isEmptyElement ) {
 414                         return "";
 415                 }
 416                 $buffer = "";
 417                 while ( $this->reader->read() ) {
 418                         switch ( $this->reader->nodeType ) {
 419                         case XmlReader::TEXT:
 420                         case XmlReader::SIGNIFICANT_WHITESPACE:
 421                                 $buffer .= $this->reader->value;
 422                                 break;
 423                         case XmlReader::END_ELEMENT:
 424                                 return $buffer;
 425                         }
 426                 }
 427
 428                 $this->reader->close();
 429                 return '';
 430         }
 431
 432         /**
 433          * Primary entry point
 434          * @throws MWException
 435          * @return bool
 436          */
 437         public function doImport() {
 438                 // Calls to reader->read need to be wrapped in calls to
 439                 // libxml_disable_entity_loader() to avoid local file
 440                 // inclusion attacks (bug 46932).
 441                 $oldDisable = libxml_disable_entity_loader( true );
 442                 $this->reader->read();
 443
 444                 if ( $this->reader->name != 'mediawiki' ) {
 445                         libxml_disable_entity_loader( $oldDisable );
 446                         throw new MWException( "Expected <mediawiki> tag, got " .
 447                                 $this->reader->name );
 448                 }
 449                 $this->debug( "<mediawiki> tag is correct." );
 450
 451                 $this->debug( "Starting primary dump processing loop." );
 452
 453                 $keepReading = $this->reader->read();
 454                 $skip = false;
 455                 while ( $keepReading ) {
 456                         $tag = $this->reader->name;
 457                         $type = $this->reader->nodeType;
 458
 459                         if ( !wfRunHooks( 'ImportHandleToplevelXMLTag', array( $this ) ) ) {
 460                                 // Do nothing
 461                         } elseif ( $tag == 'mediawiki' && $type == XmlReader::END_ELEMENT ) {
 462                                 break;
 463                         } elseif ( $tag == 'siteinfo' ) {
 464                                 $this->handleSiteInfo();
 465                         } elseif ( $tag == 'page' ) {
 466                                 $this->handlePage();
 467                         } elseif ( $tag == 'logitem' ) {
 468                                 $this->handleLogItem();
 469                         } elseif ( $tag != '#text' ) {
 470                                 $this->warn( "Unhandled top-level XML tag $tag" );
 471
 472                                 $skip = true;
 473                         }
 474
 475                         if ( $skip ) {
 476                                 $keepReading = $this->reader->next();
 477                                 $skip = false;
 478                                 $this->debug( "Skip" );
 479                         } else {
 480                                 $keepReading = $this->reader->read();
 481                         }
 482                 }
 483
 484                 libxml_disable_entity_loader( $oldDisable );
 485                 return true;
 486         }
 487
 488         /**
 489          * @return bool
 490          * @throws MWException
 491          */
 492         private function handleSiteInfo() {
 493                 // Site info is useful, but not actually used for dump imports.
 494                 // Includes a quick short-circuit to save performance.
 495                 if ( !$this->mSiteInfoCallback ) {
 496                         $this->reader->next();
 497                         return true;
 498                 }
 499                 throw new MWException( "SiteInfo tag is not yet handled, do not set mSiteInfoCallback" );
 500         }
 501
 502         private function handleLogItem() {
 503                 $this->debug( "Enter log item handler." );
 504                 $logInfo = array();
 505
 506                 // Fields that can just be stuffed in the pageInfo object
 507                 $normalFields = array( 'id', 'comment', 'type', 'action', 'timestamp',
 508                                         'logtitle', 'params' );
 509
 510                 while ( $this->reader->read() ) {
 511                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 512                                         $this->reader->name == 'logitem' ) {
 513                                 break;
 514                         }
 515
 516                         $tag = $this->reader->name;
 517
 518                         if ( !wfRunHooks( 'ImportHandleLogItemXMLTag', array(
 519                                 $this, $logInfo
 520                         ) ) ) {
 521                                 // Do nothing
 522                         } elseif ( in_array( $tag, $normalFields ) ) {
 523                                 $logInfo[$tag] = $this->nodeContents();
 524                         } elseif ( $tag == 'contributor' ) {
 525                                 $logInfo['contributor'] = $this->handleContributor();
 526                         } elseif ( $tag != '#text' ) {
 527                                 $this->warn( "Unhandled log-item XML tag $tag" );
 528                         }
 529                 }
 530
 531                 $this->processLogItem( $logInfo );
 532         }
 533
 534         /**
 535          * @param array $logInfo
 536          * @return bool|mixed
 537          */
 538         private function processLogItem( $logInfo ) {
 539                 $revision = new WikiRevision;
 540
 541                 $revision->setID( $logInfo['id'] );
 542                 $revision->setType( $logInfo['type'] );
 543                 $revision->setAction( $logInfo['action'] );
 544                 $revision->setTimestamp( $logInfo['timestamp'] );
 545                 $revision->setParams( $logInfo['params'] );
 546                 $revision->setTitle( Title::newFromText( $logInfo['logtitle'] ) );
 547                 $revision->setNoUpdates( $this->mNoUpdates );
 548
 549                 if ( isset( $logInfo['comment'] ) ) {
 550                         $revision->setComment( $logInfo['comment'] );
 551                 }
 552
 553                 if ( isset( $logInfo['contributor']['ip'] ) ) {
 554                         $revision->setUserIP( $logInfo['contributor']['ip'] );
 555                 }
 556                 if ( isset( $logInfo['contributor']['username'] ) ) {
 557                         $revision->setUserName( $logInfo['contributor']['username'] );
 558                 }
 559
 560                 return $this->logItemCallback( $revision );
 561         }
 562
 563         private function handlePage() {
 564                 // Handle page data.
 565                 $this->debug( "Enter page handler." );
 566                 $pageInfo = array( 'revisionCount' => 0, 'successfulRevisionCount' => 0 );
 567
 568                 // Fields that can just be stuffed in the pageInfo object
 569                 $normalFields = array( 'title', 'id', 'redirect', 'restrictions' );
 570
 571                 $skip = false;
 572                 $badTitle = false;
 573
 574                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 575                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 576                                         $this->reader->name == 'page' ) {
 577                                 break;
 578                         }
 579
 580                         $tag = $this->reader->name;
 581
 582                         if ( $badTitle ) {
 583                                 // The title is invalid, bail out of this page
 584                                 $skip = true;
 585                         } elseif ( !wfRunHooks( 'ImportHandlePageXMLTag', array( $this,
 586                                                 &$pageInfo ) ) ) {
 587                                 // Do nothing
 588                         } elseif ( in_array( $tag, $normalFields ) ) {
 589                                 // An XML snippet:
 590                                 // <page>
 591                                 //     <id>123</id>
 592                                 //     <title>Page</title>
 593                                 //     <redirect title="NewTitle"/>
 594                                 //     ...
 595                                 // Because the redirect tag is built differently, we need special handling for that case.
 596                                 if ( $tag == 'redirect' ) {
 597                                         $pageInfo[$tag] = $this->nodeAttribute( 'title' );
 598                                 } else {
 599                                         $pageInfo[$tag] = $this->nodeContents();
 600                                         if ( $tag == 'title' ) {
 601                                                 $title = $this->processTitle( $pageInfo['title'] );
 602
 603                                                 if ( !$title ) {
 604                                                         $badTitle = true;
 605                                                         $skip = true;
 606                                                 }
 607
 608                                                 $this->pageCallback( $title );
 609                                                 list( $pageInfo['_title'], $origTitle ) = $title;
 610                                         }
 611                                 }
 612                         } elseif ( $tag == 'revision' ) {
 613                                 $this->handleRevision( $pageInfo );
 614                         } elseif ( $tag == 'upload' ) {
 615                                 $this->handleUpload( $pageInfo );
 616                         } elseif ( $tag != '#text' ) {
 617                                 $this->warn( "Unhandled page XML tag $tag" );
 618                                 $skip = true;
 619                         }
 620                 }
 621
 622                 $this->pageOutCallback( $pageInfo['_title'], $origTitle,
 623                                         $pageInfo['revisionCount'],
 624                                         $pageInfo['successfulRevisionCount'],
 625                                         $pageInfo );
 626         }
 627
 628         /**
 629          * @param array $pageInfo
 630          */
 631         private function handleRevision( &$pageInfo ) {
 632                 $this->debug( "Enter revision handler" );
 633                 $revisionInfo = array();
 634
 635                 $normalFields = array( 'id', 'timestamp', 'comment', 'minor', 'model', 'format', 'text' );
 636
 637                 $skip = false;
 638
 639                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 640                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 641                                         $this->reader->name == 'revision' ) {
 642                                 break;
 643                         }
 644
 645                         $tag = $this->reader->name;
 646
 647                         if ( !wfRunHooks( 'ImportHandleRevisionXMLTag', array(
 648                                 $this, $pageInfo, $revisionInfo
 649                         ) ) ) {
 650                                 // Do nothing
 651                         } elseif ( in_array( $tag, $normalFields ) ) {
 652                                 $revisionInfo[$tag] = $this->nodeContents();
 653                         } elseif ( $tag == 'contributor' ) {
 654                                 $revisionInfo['contributor'] = $this->handleContributor();
 655                         } elseif ( $tag != '#text' ) {
 656                                 $this->warn( "Unhandled revision XML tag $tag" );
 657                                 $skip = true;
 658                         }
 659                 }
 660
 661                 $pageInfo['revisionCount']++;
 662                 if ( $this->processRevision( $pageInfo, $revisionInfo ) ) {
 663                         $pageInfo['successfulRevisionCount']++;
 664                 }
 665         }
 666
 667         /**
 668          * @param array $pageInfo
 669          * @param array $revisionInfo
 670          * @return bool|mixed
 671          */
 672         private function processRevision( $pageInfo, $revisionInfo ) {
 673                 $revision = new WikiRevision;
 674
 675                 if ( isset( $revisionInfo['id'] ) ) {
 676                         $revision->setID( $revisionInfo['id'] );
 677                 }
 678                 if ( isset( $revisionInfo['model'] ) ) {
 679                         $revision->setModel( $revisionInfo['model'] );
 680                 }
 681                 if ( isset( $revisionInfo['format'] ) ) {
 682                         $revision->setFormat( $revisionInfo['format'] );
 683                 }
 684                 $revision->setTitle( $pageInfo['_title'] );
 685
 686                 if ( isset( $revisionInfo['text'] ) ) {
 687                         $handler = $revision->getContentHandler();
 688                         $text = $handler->importTransform(
 689                                 $revisionInfo['text'],
 690                                 $revision->getFormat() );
 691
 692                         $revision->setText( $text );
 693                 }
 694                 if ( isset( $revisionInfo['timestamp'] ) ) {
 695                         $revision->setTimestamp( $revisionInfo['timestamp'] );
 696                 } else {
 697                         $revision->setTimestamp( wfTimestampNow() );
 698                 }
 699
 700                 if ( isset( $revisionInfo['comment'] ) ) {
 701                         $revision->setComment( $revisionInfo['comment'] );
 702                 }
 703
 704                 if ( isset( $revisionInfo['minor'] ) ) {
 705                         $revision->setMinor( true );
 706                 }
 707                 if ( isset( $revisionInfo['contributor']['ip'] ) ) {
 708                         $revision->setUserIP( $revisionInfo['contributor']['ip'] );
 709                 }
 710                 if ( isset( $revisionInfo['contributor']['username'] ) ) {
 711                         $revision->setUserName( $revisionInfo['contributor']['username'] );
 712                 }
 713                 $revision->setNoUpdates( $this->mNoUpdates );
 714
 715                 return $this->revisionCallback( $revision );
 716         }
 717
 718         /**
 719          * @param array $pageInfo
 720          * @return mixed
 721          */
 722         private function handleUpload( &$pageInfo ) {
 723                 $this->debug( "Enter upload handler" );
 724                 $uploadInfo = array();
 725
 726                 $normalFields = array( 'timestamp', 'comment', 'filename', 'text',
 727                                         'src', 'size', 'sha1base36', 'archivename', 'rel' );
 728
 729                 $skip = false;
 730
 731                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 732                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 733                                         $this->reader->name == 'upload' ) {
 734                                 break;
 735                         }
 736
 737                         $tag = $this->reader->name;
 738
 739                         if ( !wfRunHooks( 'ImportHandleUploadXMLTag', array(
 740                                 $this, $pageInfo
 741                         ) ) ) {
 742                                 // Do nothing
 743                         } elseif ( in_array( $tag, $normalFields ) ) {
 744                                 $uploadInfo[$tag] = $this->nodeContents();
 745                         } elseif ( $tag == 'contributor' ) {
 746                                 $uploadInfo['contributor'] = $this->handleContributor();
 747                         } elseif ( $tag == 'contents' ) {
 748                                 $contents = $this->nodeContents();
 749                                 $encoding = $this->reader->getAttribute( 'encoding' );
 750                                 if ( $encoding === 'base64' ) {
 751                                         $uploadInfo['fileSrc'] = $this->dumpTemp( base64_decode( $contents ) );
 752                                         $uploadInfo['isTempSrc'] = true;
 753                                 }
 754                         } elseif ( $tag != '#text' ) {
 755                                 $this->warn( "Unhandled upload XML tag $tag" );
 756                                 $skip = true;
 757                         }
 758                 }
 759
 760                 if ( $this->mImageBasePath && isset( $uploadInfo['rel'] ) ) {
 761                         $path = "{$this->mImageBasePath}/{$uploadInfo['rel']}";
 762                         if ( file_exists( $path ) ) {
 763                                 $uploadInfo['fileSrc'] = $path;
 764                                 $uploadInfo['isTempSrc'] = false;
 765                         }
 766                 }
 767
 768                 if ( $this->mImportUploads ) {
 769                         return $this->processUpload( $pageInfo, $uploadInfo );
 770                 }
 771         }
 772
 773         /**
 774          * @param string $contents
 775          * @return string
 776          */
 777         private function dumpTemp( $contents ) {
 778                 $filename = tempnam( wfTempDir(), 'importupload' );
 779                 file_put_contents( $filename, $contents );
 780                 return $filename;
 781         }
 782
 783         /**
 784          * @param array $pageInfo
 785          * @param array $uploadInfo
 786          * @return mixed
 787          */
 788         private function processUpload( $pageInfo, $uploadInfo ) {
 789                 $revision = new WikiRevision;
 790                 $text = isset( $uploadInfo['text'] ) ? $uploadInfo['text'] : '';
 791
 792                 $revision->setTitle( $pageInfo['_title'] );
 793                 $revision->setID( $pageInfo['id'] );
 794                 $revision->setTimestamp( $uploadInfo['timestamp'] );
 795                 $revision->setText( $text );
 796                 $revision->setFilename( $uploadInfo['filename'] );
 797                 if ( isset( $uploadInfo['archivename'] ) ) {
 798                         $revision->setArchiveName( $uploadInfo['archivename'] );
 799                 }
 800                 $revision->setSrc( $uploadInfo['src'] );
 801                 if ( isset( $uploadInfo['fileSrc'] ) ) {
 802                         $revision->setFileSrc( $uploadInfo['fileSrc'],
 803                                 !empty( $uploadInfo['isTempSrc'] ) );
 804                 }
 805                 if ( isset( $uploadInfo['sha1base36'] ) ) {
 806                         $revision->setSha1Base36( $uploadInfo['sha1base36'] );
 807                 }
 808                 $revision->setSize( intval( $uploadInfo['size'] ) );
 809                 $revision->setComment( $uploadInfo['comment'] );
 810
 811                 if ( isset( $uploadInfo['contributor']['ip'] ) ) {
 812                         $revision->setUserIP( $uploadInfo['contributor']['ip'] );
 813                 }
 814                 if ( isset( $uploadInfo['contributor']['username'] ) ) {
 815                         $revision->setUserName( $uploadInfo['contributor']['username'] );
 816                 }
 817                 $revision->setNoUpdates( $this->mNoUpdates );
 818
 819                 return call_user_func( $this->mUploadCallback, $revision );
 820         }
 821
 822         /**
 823          * @return array
 824          */
 825         private function handleContributor() {
 826                 $fields = array( 'id', 'ip', 'username' );
 827                 $info = array();
 828
 829                 while ( $this->reader->read() ) {
 830                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 831                                         $this->reader->name == 'contributor' ) {
 832                                 break;
 833                         }
 834
 835                         $tag = $this->reader->name;
 836
 837                         if ( in_array( $tag, $fields ) ) {
 838                                 $info[$tag] = $this->nodeContents();
 839                         }
 840                 }
 841
 842                 return $info;
 843         }
 844
 845         /**
 846          * @param string $text
 847          * @return array|bool
 848          */
 849         private function processTitle( $text ) {
 850                 global $wgCommandLineMode;
 851
 852                 $workTitle = $text;
 853                 $origTitle = Title::newFromText( $workTitle );
 854
 855                 if ( !is_null( $this->mTargetNamespace ) && !is_null( $origTitle ) ) {
 856                         # makeTitleSafe, because $origTitle can have a interwiki (different setting of interwiki map)
 857                         # and than dbKey can begin with a lowercase char
 858                         $title = Title::makeTitleSafe( $this->mTargetNamespace,
 859                                 $origTitle->getDBkey() );
 860                 } else {
 861                         if ( !is_null( $this->mTargetRootPage ) ) {
 862                                 $workTitle = $this->mTargetRootPage . '/' . $workTitle;
 863                         }
 864                         $title = Title::newFromText( $workTitle );
 865                 }
 866
 867                 if ( is_null( $title ) ) {
 868                         # Invalid page title? Ignore the page
 869                         $this->notice( 'import-error-invalid', $workTitle );
 870                         return false;
 871                 } elseif ( $title->isExternal() ) {
 872                         $this->notice( 'import-error-interwiki', $title->getPrefixedText() );
 873                         return false;
 874                 } elseif ( !$title->canExist() ) {
 875                         $this->notice( 'import-error-special', $title->getPrefixedText() );
 876                         return false;
 877                 } elseif ( !$title->userCan( 'edit' ) && !$wgCommandLineMode ) {
 878                         # Do not import if the importing wiki user cannot edit this page
 879                         $this->notice( 'import-error-edit', $title->getPrefixedText() );
 880                         return false;
 881                 } elseif ( !$title->exists() && !$title->userCan( 'create' ) && !$wgCommandLineMode ) {
 882                         # Do not import if the importing wiki user cannot create this page
 883                         $this->notice( 'import-error-create', $title->getPrefixedText() );
 884                         return false;
 885                 }
 886
 887                 return array( $title, $origTitle );
 888         }
 889 }
 890
 891 /** This is a horrible hack used to keep source compatibility */
 892 class UploadSourceAdapter {
 893         /** @var array */
 894         private static $sourceRegistrations = array();
 895
 896         /** @var string */
 897         private $mSource;
 898
 899         /** @var string */
 900         private $mBuffer;
 901
 902         /** @var int */
 903         private $mPosition;
 904
 905         /**
 906          * @param ImportStreamSource $source
 907          * @return string
 908          */
 909         static function registerSource( ImportStreamSource $source ) {
 910                 $id = wfRandomString();
 911
 912                 self::$sourceRegistrations[$id] = $source;
 913
 914                 return $id;
 915         }
 916
 917         /**
 918          * @param string $path
 919          * @param string $mode
 920          * @param array $options
 921          * @param string $opened_path
 922          * @return bool
 923          */
 924         function stream_open( $path, $mode, $options, &$opened_path ) {
 925                 $url = parse_url( $path );
 926                 $id = $url['host'];
 927
 928                 if ( !isset( self::$sourceRegistrations[$id] ) ) {
 929                         return false;
 930                 }
 931
 932                 $this->mSource = self::$sourceRegistrations[$id];
 933
 934                 return true;
 935         }
 936
 937         /**
 938          * @param int $count
 939          * @return string
 940          */
 941         function stream_read( $count ) {
 942                 $return = '';
 943                 $leave = false;
 944
 945                 while ( !$leave && !$this->mSource->atEnd() &&
 946                                 strlen( $this->mBuffer ) < $count ) {
 947                         $read = $this->mSource->readChunk();
 948
 949                         if ( !strlen( $read ) ) {
 950                                 $leave = true;
 951                         }
 952
 953                         $this->mBuffer .= $read;
 954                 }
 955
 956                 if ( strlen( $this->mBuffer ) ) {
 957                         $return = substr( $this->mBuffer, 0, $count );
 958                         $this->mBuffer = substr( $this->mBuffer, $count );
 959                 }
 960
 961                 $this->mPosition += strlen( $return );
 962
 963                 return $return;
 964         }
 965
 966         /**
 967          * @param string $data
 968          * @return bool
 969          */
 970         function stream_write( $data ) {
 971                 return false;
 972         }
 973
 974         /**
 975          * @return mixed
 976          */
 977         function stream_tell() {
 978                 return $this->mPosition;
 979         }
 980
 981         /**
 982          * @return bool
 983          */
 984         function stream_eof() {
 985                 return $this->mSource->atEnd();
 986         }
 987
 988         /**
 989          * @return array
 990          */
 991         function url_stat() {
 992                 $result = array();
 993
 994                 $result['dev'] = $result[0] = 0;
 995                 $result['ino'] = $result[1] = 0;
 996                 $result['mode'] = $result[2] = 0;
 997                 $result['nlink'] = $result[3] = 0;
 998                 $result['uid'] = $result[4] = 0;
 999                 $result['gid'] = $result[5] = 0;
1000                 $result['rdev'] = $result[6] = 0;
1001                 $result['size'] = $result[7] = 0;
1002                 $result['atime'] = $result[8] = 0;
1003                 $result['mtime'] = $result[9] = 0;
1004                 $result['ctime'] = $result[10] = 0;
1005                 $result['blksize'] = $result[11] = 0;
1006                 $result['blocks'] = $result[12] = 0;
1007
1008                 return $result;
1009         }
1010 }
1011
1012 /**
1013  * @todo document (e.g. one-sentence class description).
1014  * @ingroup SpecialPage
1015  */
1016 class WikiRevision {
1017         /** @todo Unused? */
1018         private $importer = null;
1019
1020         /** @var Title */
1021         public $title = null;
1022
1023         /** @var int */
1024         private $id = 0;
1025
1026         /** @var string */
1027         public $timestamp = "20010115000000";
1028
1029         /**
1030          * @var int
1031          * @todo Can't find any uses. Public, because that's suspicious. Get clarity. */
1032         public $user = 0;
1033
1034         /** @var string */
1035         public $user_text = "";
1036
1037         /** @var string */
1038         protected $model = null;
1039
1040         /** @var string */
1041         protected $format = null;
1042
1043         /** @var string */
1044         public $text = "";
1045
1046         /** @var int */
1047         protected $size;
1048
1049         /** @var Content */
1050         protected $content = null;
1051
1052         /** @var ContentHandler */
1053         protected $contentHandler = null;
1054
1055         /** @var string */
1056         public $comment = "";
1057
1058         /** @var bool */
1059         protected $minor = false;
1060
1061         /** @var string */
1062         protected $type = "";
1063
1064         /** @var string */
1065         protected $action = "";
1066
1067         /** @var string */
1068         protected $params = "";
1069
1070         /** @var string */
1071         protected $fileSrc = '';
1072
1073         /** @var bool|string */
1074         protected $sha1base36 = false;
1075
1076         /**
1077          * @var bool
1078          * @todo Unused?
1079          */
1080         private $isTemp = false;
1081
1082         /** @var string */
1083         protected $archiveName = '';
1084
1085         protected $filename;
1086
1087         /** @var mixed */
1088         protected $src;
1089
1090         /** @todo Unused? */
1091         private $fileIsTemp;
1092
1093         /** @var bool */
1094         private $mNoUpdates = false;
1095
1096         /**
1097          * @param Title $title
1098          * @throws MWException
1099          */
1100         function setTitle( $title ) {
1101                 if ( is_object( $title ) ) {
1102                         $this->title = $title;
1103                 } elseif ( is_null( $title ) ) {
1104                         throw new MWException( "WikiRevision given a null title in import. "
1105                                 . "You may need to adjust \$wgLegalTitleChars." );
1106                 } else {
1107                         throw new MWException( "WikiRevision given non-object title in import." );
1108                 }
1109         }
1110
1111         /**
1112          * @param int $id
1113          */
1114         function setID( $id ) {
1115                 $this->id = $id;
1116         }
1117
1118         /**
1119          * @param string $ts
1120          */
1121         function setTimestamp( $ts ) {
1122                 # 2003-08-05T18:30:02Z
1123                 $this->timestamp = wfTimestamp( TS_MW, $ts );
1124         }
1125
1126         /**
1127          * @param string $user
1128          */
1129         function setUsername( $user ) {
1130                 $this->user_text = $user;
1131         }
1132
1133         /**
1134          * @param string $ip
1135          */
1136         function setUserIP( $ip ) {
1137                 $this->user_text = $ip;
1138         }
1139
1140         /**
1141          * @param string $model
1142          */
1143         function setModel( $model ) {
1144                 $this->model = $model;
1145         }
1146
1147         /**
1148          * @param string $format
1149          */
1150         function setFormat( $format ) {
1151                 $this->format = $format;
1152         }
1153
1154         /**
1155          * @param string $text
1156          */
1157         function setText( $text ) {
1158                 $this->text = $text;
1159         }
1160
1161         /**
1162          * @param string $text
1163          */
1164         function setComment( $text ) {
1165                 $this->comment = $text;
1166         }
1167
1168         /**
1169          * @param bool $minor
1170          */
1171         function setMinor( $minor ) {
1172                 $this->minor = (bool)$minor;
1173         }
1174
1175         /**
1176          * @param mixed $src
1177          */
1178         function setSrc( $src ) {
1179                 $this->src = $src;
1180         }
1181
1182         /**
1183          * @param string $src
1184          * @param bool $isTemp
1185          */
1186         function setFileSrc( $src, $isTemp ) {
1187                 $this->fileSrc = $src;
1188                 $this->fileIsTemp = $isTemp;
1189         }
1190
1191         /**
1192          * @param string $sha1base36
1193          */
1194         function setSha1Base36( $sha1base36 ) {
1195                 $this->sha1base36 = $sha1base36;
1196         }
1197
1198         /**
1199          * @param string $filename
1200          */
1201         function setFilename( $filename ) {
1202                 $this->filename = $filename;
1203         }
1204
1205         /**
1206          * @param string $archiveName
1207          */
1208         function setArchiveName( $archiveName ) {
1209                 $this->archiveName = $archiveName;
1210         }
1211
1212         /**
1213          * @param int $size
1214          */
1215         function setSize( $size ) {
1216                 $this->size = intval( $size );
1217         }
1218
1219         /**
1220          * @param string $type
1221          */
1222         function setType( $type ) {
1223                 $this->type = $type;
1224         }
1225
1226         /**
1227          * @param string $action
1228          */
1229         function setAction( $action ) {
1230                 $this->action = $action;
1231         }
1232
1233         /**
1234          * @param array $params
1235          */
1236         function setParams( $params ) {
1237                 $this->params = $params;
1238         }
1239
1240         /**
1241          * @param bool $noupdates
1242          */
1243         public function setNoUpdates( $noupdates ) {
1244                 $this->mNoUpdates = $noupdates;
1245         }
1246
1247         /**
1248          * @return Title
1249          */
1250         function getTitle() {
1251                 return $this->title;
1252         }
1253
1254         /**
1255          * @return int
1256          */
1257         function getID() {
1258                 return $this->id;
1259         }
1260
1261         /**
1262          * @return string
1263          */
1264         function getTimestamp() {
1265                 return $this->timestamp;
1266         }
1267
1268         /**
1269          * @return string
1270          */
1271         function getUser() {
1272                 return $this->user_text;
1273         }
1274
1275         /**
1276          * @return string
1277          *
1278          * @deprecated Since 1.21, use getContent() instead.
1279          */
1280         function getText() {
1281                 ContentHandler::deprecated( __METHOD__, '1.21' );
1282
1283                 return $this->text;
1284         }
1285
1286         /**
1287          * @return ContentHandler
1288          */
1289         function getContentHandler() {
1290                 if ( is_null( $this->contentHandler ) ) {
1291                         $this->contentHandler = ContentHandler::getForModelID( $this->getModel() );
1292                 }
1293
1294                 return $this->contentHandler;
1295         }
1296
1297         /**
1298          * @return Content
1299          */
1300         function getContent() {
1301                 if ( is_null( $this->content ) ) {
1302                         $handler = $this->getContentHandler();
1303                         $this->content = $handler->unserializeContent( $this->text, $this->getFormat() );
1304                 }
1305
1306                 return $this->content;
1307         }
1308
1309         /**
1310          * @return string
1311          */
1312         function getModel() {
1313                 if ( is_null( $this->model ) ) {
1314                         $this->model = $this->getTitle()->getContentModel();
1315                 }
1316
1317                 return $this->model;
1318         }
1319
1320         /**
1321          * @return string
1322          */
1323         function getFormat() {
1324                 if ( is_null( $this->format ) ) {
1325                         $this->format = $this->getContentHandler()->getDefaultFormat();
1326                 }
1327
1328                 return $this->format;
1329         }
1330
1331         /**
1332          * @return string
1333          */
1334         function getComment() {
1335                 return $this->comment;
1336         }
1337
1338         /**
1339          * @return bool
1340          */
1341         function getMinor() {
1342                 return $this->minor;
1343         }
1344
1345         /**
1346          * @return mixed
1347          */
1348         function getSrc() {
1349                 return $this->src;
1350         }
1351
1352         /**
1353          * @return bool|string
1354          */
1355         function getSha1() {
1356                 if ( $this->sha1base36 ) {
1357                         return wfBaseConvert( $this->sha1base36, 36, 16 );
1358                 }
1359                 return false;
1360         }
1361
1362         /**
1363          * @return string
1364          */
1365         function getFileSrc() {
1366                 return $this->fileSrc;
1367         }
1368
1369         /**
1370          * @return bool
1371          */
1372         function isTempSrc() {
1373                 return $this->isTemp;
1374         }
1375
1376         /**
1377          * @return mixed
1378          */
1379         function getFilename() {
1380                 return $this->filename;
1381         }
1382
1383         /**
1384          * @return string
1385          */
1386         function getArchiveName() {
1387                 return $this->archiveName;
1388         }
1389
1390         /**
1391          * @return mixed
1392          */
1393         function getSize() {
1394                 return $this->size;
1395         }
1396
1397         /**
1398          * @return string
1399          */
1400         function getType() {
1401                 return $this->type;
1402         }
1403
1404         /**
1405          * @return string
1406          */
1407         function getAction() {
1408                 return $this->action;
1409         }
1410
1411         /**
1412          * @return string
1413          */
1414         function getParams() {
1415                 return $this->params;
1416         }
1417
1418         /**
1419          * @return bool
1420          */
1421         function importOldRevision() {
1422                 $dbw = wfGetDB( DB_MASTER );
1423
1424                 # Sneak a single revision into place
1425                 $user = User::newFromName( $this->getUser() );
1426                 if ( $user ) {
1427                         $userId = intval( $user->getId() );
1428                         $userText = $user->getName();
1429                         $userObj = $user;
1430                 } else {
1431                         $userId = 0;
1432                         $userText = $this->getUser();
1433                         $userObj = new User;
1434                 }
1435
1436                 // avoid memory leak...?
1437                 $linkCache = LinkCache::singleton();
1438                 $linkCache->clear();
1439
1440                 $page = WikiPage::factory( $this->title );
1441                 $page->loadPageData( 'fromdbmaster' );
1442                 if ( !$page->exists() ) {
1443                         # must create the page...
1444                         $pageId = $page->insertOn( $dbw );
1445                         $created = true;
1446                         $oldcountable = null;
1447                 } else {
1448                         $pageId = $page->getId();
1449                         $created = false;
1450
1451                         $prior = $dbw->selectField( 'revision', '1',
1452                                 array( 'rev_page' => $pageId,
1453                                         'rev_timestamp' => $dbw->timestamp( $this->timestamp ),
1454                                         'rev_user_text' => $userText,
1455                                         'rev_comment' => $this->getComment() ),
1456                                 __METHOD__
1457                         );
1458                         if ( $prior ) {
1459                                 // @todo FIXME: This could fail slightly for multiple matches :P
1460                                 wfDebug( __METHOD__ . ": skipping existing revision for [[" .
1461                                         $this->title->getPrefixedText() . "]], timestamp " . $this->timestamp . "\n" );
1462                                 return false;
1463                         }
1464                         $oldcountable = $page->isCountable();
1465                 }
1466
1467                 # @todo FIXME: Use original rev_id optionally (better for backups)
1468                 # Insert the row
1469                 $revision = new Revision( array(
1470                         'title' => $this->title,
1471                         'page' => $pageId,
1472                         'content_model' => $this->getModel(),
1473                         'content_format' => $this->getFormat(),
1474                         //XXX: just set 'content' => $this->getContent()?
1475                         'text' => $this->getContent()->serialize( $this->getFormat() ),
1476                         'comment' => $this->getComment(),
1477                         'user' => $userId,
1478                         'user_text' => $userText,
1479                         'timestamp' => $this->timestamp,
1480                         'minor_edit' => $this->minor,
1481                         ) );
1482                 $revision->insertOn( $dbw );
1483                 $changed = $page->updateIfNewerOn( $dbw, $revision );
1484
1485                 if ( $changed !== false && !$this->mNoUpdates ) {
1486                         wfDebug( __METHOD__ . ": running updates\n" );
1487                         $page->doEditUpdates(
1488                                 $revision,
1489                                 $userObj,
1490                                 array( 'created' => $created, 'oldcountable' => $oldcountable )
1491                         );
1492                 }
1493
1494                 return true;
1495         }
1496
1497         function importLogItem() {
1498                 $dbw = wfGetDB( DB_MASTER );
1499                 # @todo FIXME: This will not record autoblocks
1500                 if ( !$this->getTitle() ) {
1501                         wfDebug( __METHOD__ . ": skipping invalid {$this->type}/{$this->action} log time, timestamp " .
1502                                 $this->timestamp . "\n" );
1503                         return;
1504                 }
1505                 # Check if it exists already
1506                 // @todo FIXME: Use original log ID (better for backups)
1507                 $prior = $dbw->selectField( 'logging', '1',
1508                         array( 'log_type' => $this->getType(),
1509                                 'log_action' => $this->getAction(),
1510                                 'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1511                                 'log_namespace' => $this->getTitle()->getNamespace(),
1512                                 'log_title' => $this->getTitle()->getDBkey(),
1513                                 'log_comment' => $this->getComment(),
1514                                 #'log_user_text' => $this->user_text,
1515                                 'log_params' => $this->params ),
1516                         __METHOD__
1517                 );
1518                 // @todo FIXME: This could fail slightly for multiple matches :P
1519                 if ( $prior ) {
1520                         wfDebug( __METHOD__
1521                                 . ": skipping existing item for Log:{$this->type}/{$this->action}, timestamp "
1522                                 . $this->timestamp . "\n" );
1523                         return;
1524                 }
1525                 $log_id = $dbw->nextSequenceValue( 'logging_log_id_seq' );
1526                 $data = array(
1527                         'log_id' => $log_id,
1528                         'log_type' => $this->type,
1529                         'log_action' => $this->action,
1530                         'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1531                         'log_user' => User::idFromName( $this->user_text ),
1532                         #'log_user_text' => $this->user_text,
1533                         'log_namespace' => $this->getTitle()->getNamespace(),
1534                         'log_title' => $this->getTitle()->getDBkey(),
1535                         'log_comment' => $this->getComment(),
1536                         'log_params' => $this->params
1537                 );
1538                 $dbw->insert( 'logging', $data, __METHOD__ );
1539         }
1540
1541         /**
1542          * @return bool
1543          */
1544         function importUpload() {
1545                 # Construct a file
1546                 $archiveName = $this->getArchiveName();
1547                 if ( $archiveName ) {
1548                         wfDebug( __METHOD__ . "Importing archived file as $archiveName\n" );
1549                         $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1550                                 RepoGroup::singleton()->getLocalRepo(), $archiveName );
1551                 } else {
1552                         $file = wfLocalFile( $this->getTitle() );
1553                         wfDebug( __METHOD__ . 'Importing new file as ' . $file->getName() . "\n" );
1554                         if ( $file->exists() && $file->getTimestamp() > $this->getTimestamp() ) {
1555                                 $archiveName = $file->getTimestamp() . '!' . $file->getName();
1556                                 $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1557                                         RepoGroup::singleton()->getLocalRepo(), $archiveName );
1558                                 wfDebug( __METHOD__ . "File already exists; importing as $archiveName\n" );
1559                         }
1560                 }
1561                 if ( !$file ) {
1562                         wfDebug( __METHOD__ . ': Bad file for ' . $this->getTitle() . "\n" );
1563                         return false;
1564                 }
1565
1566                 # Get the file source or download if necessary
1567                 $source = $this->getFileSrc();
1568                 $flags = $this->isTempSrc() ? File::DELETE_SOURCE : 0;
1569                 if ( !$source ) {
1570                         $source = $this->downloadSource();
1571                         $flags |= File::DELETE_SOURCE;
1572                 }
1573                 if ( !$source ) {
1574                         wfDebug( __METHOD__ . ": Could not fetch remote file.\n" );
1575                         return false;
1576                 }
1577                 $sha1 = $this->getSha1();
1578                 if ( $sha1 && ( $sha1 !== sha1_file( $source ) ) ) {
1579                         if ( $flags & File::DELETE_SOURCE ) {
1580                                 # Broken file; delete it if it is a temporary file
1581                                 unlink( $source );
1582                         }
1583                         wfDebug( __METHOD__ . ": Corrupt file $source.\n" );
1584                         return false;
1585                 }
1586
1587                 $user = User::newFromName( $this->user_text );
1588
1589                 # Do the actual upload
1590                 if ( $archiveName ) {
1591                         $status = $file->uploadOld( $source, $archiveName,
1592                                 $this->getTimestamp(), $this->getComment(), $user, $flags );
1593                 } else {
1594                         $status = $file->upload( $source, $this->getComment(), $this->getComment(),
1595                                 $flags, false, $this->getTimestamp(), $user );
1596                 }
1597
1598                 if ( $status->isGood() ) {
1599                         wfDebug( __METHOD__ . ": Successful\n" );
1600                         return true;
1601                 } else {
1602                         wfDebug( __METHOD__ . ': failed: ' . $status->getXml() . "\n" );
1603                         return false;
1604                 }
1605         }
1606
1607         /**
1608          * @return bool|string
1609          */
1610         function downloadSource() {
1611                 global $wgEnableUploads;
1612                 if ( !$wgEnableUploads ) {
1613                         return false;
1614                 }
1615
1616                 $tempo = tempnam( wfTempDir(), 'download' );
1617                 $f = fopen( $tempo, 'wb' );
1618                 if ( !$f ) {
1619                         wfDebug( "IMPORT: couldn't write to temp file $tempo\n" );
1620                         return false;
1621                 }
1622
1623                 // @todo FIXME!
1624                 $src = $this->getSrc();
1625                 $data = Http::get( $src );
1626                 if ( !$data ) {
1627                         wfDebug( "IMPORT: couldn't fetch source $src\n" );
1628                         fclose( $f );
1629                         unlink( $tempo );
1630                         return false;
1631                 }
1632
1633                 fwrite( $f, $data );
1634                 fclose( $f );
1635
1636                 return $tempo;
1637         }
1638
1639 }
1640
1641 /**
1642  * @todo document (e.g. one-sentence class description).
1643  * @ingroup SpecialPage
1644  */
1645 class ImportStringSource {
1646         function __construct( $string ) {
1647                 $this->mString = $string;
1648                 $this->mRead = false;
1649         }
1650
1651         /**
1652          * @return bool
1653          */
1654         function atEnd() {
1655                 return $this->mRead;
1656         }
1657
1658         /**
1659          * @return bool|string
1660          */
1661         function readChunk() {
1662                 if ( $this->atEnd() ) {
1663                         return false;
1664                 }
1665                 $this->mRead = true;
1666                 return $this->mString;
1667         }
1668 }
1669
1670 /**
1671  * @todo document (e.g. one-sentence class description).
1672  * @ingroup SpecialPage
1673  */
1674 class ImportStreamSource {
1675         function __construct( $handle ) {
1676                 $this->mHandle = $handle;
1677         }
1678
1679         /**
1680          * @return bool
1681          */
1682         function atEnd() {
1683                 return feof( $this->mHandle );
1684         }
1685
1686         /**
1687          * @return string
1688          */
1689         function readChunk() {
1690                 return fread( $this->mHandle, 32768 );
1691         }
1692
1693         /**
1694          * @param string $filename
1695          * @return Status
1696          */
1697         static function newFromFile( $filename ) {
1698                 wfSuppressWarnings();
1699                 $file = fopen( $filename, 'rt' );
1700                 wfRestoreWarnings();
1701                 if ( !$file ) {
1702                         return Status::newFatal( "importcantopen" );
1703                 }
1704                 return Status::newGood( new ImportStreamSource( $file ) );
1705         }
1706
1707         /**
1708          * @param string $fieldname
1709          * @return Status
1710          */
1711         static function newFromUpload( $fieldname = "xmlimport" ) {
1712                 $upload =& $_FILES[$fieldname];
1713
1714                 if ( $upload === null || !$upload['name'] ) {
1715                         return Status::newFatal( 'importnofile' );
1716                 }
1717                 if ( !empty( $upload['error'] ) ) {
1718                         switch ( $upload['error'] ) {
1719                                 case 1:
1720                                         # The uploaded file exceeds the upload_max_filesize directive in php.ini.
1721                                         return Status::newFatal( 'importuploaderrorsize' );
1722                                 case 2:
1723                                         # The uploaded file exceeds the MAX_FILE_SIZE directive that
1724                                         # was specified in the HTML form.
1725                                         return Status::newFatal( 'importuploaderrorsize' );
1726                                 case 3:
1727                                         # The uploaded file was only partially uploaded
1728                                         return Status::newFatal( 'importuploaderrorpartial' );
1729                                 case 6:
1730                                         # Missing a temporary folder.
1731                                         return Status::newFatal( 'importuploaderrortemp' );
1732                                 # case else: # Currently impossible
1733                         }
1734
1735                 }
1736                 $fname = $upload['tmp_name'];
1737                 if ( is_uploaded_file( $fname ) ) {
1738                         return ImportStreamSource::newFromFile( $fname );
1739                 } else {
1740                         return Status::newFatal( 'importnofile' );
1741                 }
1742         }
1743
1744         /**
1745          * @param string $url
1746          * @param string $method
1747          * @return Status
1748          */
1749         static function newFromURL( $url, $method = 'GET' ) {
1750                 wfDebug( __METHOD__ . ": opening $url\n" );
1751                 # Use the standard HTTP fetch function; it times out
1752                 # quicker and sorts out user-agent problems which might
1753                 # otherwise prevent importing from large sites, such
1754                 # as the Wikimedia cluster, etc.
1755                 $data = Http::request( $method, $url, array( 'followRedirects' => true ) );
1756                 if ( $data !== false ) {
1757                         $file = tmpfile();
1758                         fwrite( $file, $data );
1759                         fflush( $file );
1760                         fseek( $file, 0 );
1761                         return Status::newGood( new ImportStreamSource( $file ) );
1762                 } else {
1763                         return Status::newFatal( 'importcantopen' );
1764                 }
1765         }
1766
1767         /**
1768          * @param string $interwiki
1769          * @param string $page
1770          * @param bool $history
1771          * @param bool $templates
1772          * @param int $pageLinkDepth
1773          * @return Status
1774          */
1775         public static function newFromInterwiki( $interwiki, $page, $history = false,
1776                 $templates = false, $pageLinkDepth = 0
1777         ) {
1778                 if ( $page == '' ) {
1779                         return Status::newFatal( 'import-noarticle' );
1780                 }
1781                 $link = Title::newFromText( "$interwiki:Special:Export/$page" );
1782                 if ( is_null( $link ) || !$link->isExternal() ) {
1783                         return Status::newFatal( 'importbadinterwiki' );
1784                 } else {
1785                         $params = array();
1786                         if ( $history ) {
1787                                 $params['history'] = 1;
1788                         }
1789                         if ( $templates ) {
1790                                 $params['templates'] = 1;
1791                         }
1792                         if ( $pageLinkDepth ) {
1793                                 $params['pagelink-depth'] = $pageLinkDepth;
1794                         }
1795                         $url = $link->getFullURL( $params );
1796                         # For interwikis, use POST to avoid redirects.
1797                         return ImportStreamSource::newFromURL( $url, "POST" );
1798                 }
1799         }
1800 }