includes/Import.php

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