includes/Import.php

   1 <?php
   2 /**
   3  * MediaWiki page data importer
   4  *
   5  * Copyright © 2003,2005 Brion Vibber <brion@pobox.com>
   6  * http://www.mediawiki.org/
   7  *
   8  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 2 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with this program; if not, write to the Free Software Foundation, Inc.,
  20  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  21  * http://www.gnu.org/copyleft/gpl.html
  22  *
  23  * @file
  24  * @ingroup SpecialPage
  25  */
  26
  27 /**
  28  * XML file reader for the page data importer
  29  *
  30  * implements Special:Import
  31  * @ingroup SpecialPage
  32  */
  33 class WikiImporter {
  34         private $reader = null;
  35         private $mLogItemCallback, $mUploadCallback, $mRevisionCallback, $mPageCallback;
  36         private $mSiteInfoCallback, $mTargetNamespace, $mPageOutCallback;
  37         private $mDebug;
  38
  39         /**
  40          * Creates an ImportXMLReader drawing from the source provided
  41         */
  42         function __construct( $source ) {
  43                 $this->reader = new XMLReader();
  44
  45                 stream_wrapper_register( 'uploadsource', 'UploadSourceAdapter' );
  46                 $id = UploadSourceAdapter::registerSource( $source );
  47                 $this->reader->open( "uploadsource://$id" );
  48
  49                 // Default callbacks
  50                 $this->setRevisionCallback( array( $this, "importRevision" ) );
  51                 $this->setUploadCallback( array( $this, 'importUpload' ) );
  52                 $this->setLogItemCallback( array( $this, 'importLogItem' ) );
  53                 $this->setPageOutCallback( array( $this, 'finishImportPage' ) );
  54         }
  55
  56         private function throwXmlError( $err ) {
  57                 $this->debug( "FAILURE: $err" );
  58                 wfDebug( "WikiImporter XML error: $err\n" );
  59         }
  60
  61         private function debug( $data ) {
  62                 if( $this->mDebug ) {
  63                         wfDebug( "IMPORT: $data\n" );
  64                 }
  65         }
  66
  67         private function warn( $data ) {
  68                 wfDebug( "IMPORT: $data\n" );
  69         }
  70
  71         private function notice( $data ) {
  72                 global $wgCommandLineMode;
  73                 if( $wgCommandLineMode ) {
  74                         print "$data\n";
  75                 } else {
  76                         global $wgOut;
  77                         $wgOut->addHTML( "<li>" . htmlspecialchars( $data ) . "</li>\n" );
  78                 }
  79         }
  80
  81         /**
  82          * Set debug mode...
  83          */
  84         function setDebug( $debug ) {
  85                 $this->mDebug = $debug;
  86         }
  87
  88         /**
  89          * Sets the action to perform as each new page in the stream is reached.
  90          * @param $callback callback
  91          * @return callback
  92          */
  93         public function setPageCallback( $callback ) {
  94                 $previous = $this->mPageCallback;
  95                 $this->mPageCallback = $callback;
  96                 return $previous;
  97         }
  98
  99         /**
 100          * Sets the action to perform as each page in the stream is completed.
 101          * Callback accepts the page title (as a Title object), a second object
 102          * with the original title form (in case it's been overridden into a
 103          * local namespace), and a count of revisions.
 104          *
 105          * @param $callback callback
 106          * @return callback
 107          */
 108         public function setPageOutCallback( $callback ) {
 109                 $previous = $this->mPageOutCallback;
 110                 $this->mPageOutCallback = $callback;
 111                 return $previous;
 112         }
 113
 114         /**
 115          * Sets the action to perform as each page revision is reached.
 116          * @param $callback callback
 117          * @return callback
 118          */
 119         public function setRevisionCallback( $callback ) {
 120                 $previous = $this->mRevisionCallback;
 121                 $this->mRevisionCallback = $callback;
 122                 return $previous;
 123         }
 124
 125         /**
 126          * Sets the action to perform as each file upload version is reached.
 127          * @param $callback callback
 128          * @return callback
 129          */
 130         public function setUploadCallback( $callback ) {
 131                 $previous = $this->mUploadCallback;
 132                 $this->mUploadCallback = $callback;
 133                 return $previous;
 134         }
 135
 136         /**
 137          * Sets the action to perform as each log item reached.
 138          * @param $callback callback
 139          * @return callback
 140          */
 141         public function setLogItemCallback( $callback ) {
 142                 $previous = $this->mLogItemCallback;
 143                 $this->mLogItemCallback = $callback;
 144                 return $previous;
 145         }
 146
 147         /**
 148          * Sets the action to perform when site info is encountered
 149          * @param $callback callback
 150          * @return callback
 151          */
 152         public function setSiteInfoCallback( $callback ) {
 153                 $previous = $this->mSiteInfoCallback;
 154                 $this->mSiteInfoCallback = $callback;
 155                 return $previous;
 156         }
 157
 158         /**
 159          * Set a target namespace to override the defaults
 160          */
 161         public function setTargetNamespace( $namespace ) {
 162                 if( is_null( $namespace ) ) {
 163                         // Don't override namespaces
 164                         $this->mTargetNamespace = null;
 165                 } elseif( $namespace >= 0 ) {
 166                         // FIXME: Check for validity
 167                         $this->mTargetNamespace = intval( $namespace );
 168                 } else {
 169                         return false;
 170                 }
 171         }
 172
 173         /**
 174          * Default per-revision callback, performs the import.
 175          * @param $revision WikiRevision
 176          */
 177         public function importRevision( $revision ) {
 178                 $dbw = wfGetDB( DB_MASTER );
 179                 return $dbw->deadlockLoop( array( $revision, 'importOldRevision' ) );
 180         }
 181
 182         /**
 183          * Default per-revision callback, performs the import.
 184          * @param $rev WikiRevision
 185          */
 186         public function importLogItem( $rev ) {
 187                 $dbw = wfGetDB( DB_MASTER );
 188                 return $dbw->deadlockLoop( array( $rev, 'importLogItem' ) );
 189         }
 190
 191         /**
 192          * Dummy for now...
 193          */
 194         public function importUpload( $revision ) {
 195                            $revision->importUpload();
 196                 //$dbw = wfGetDB( DB_MASTER );
 197                 //return $dbw->deadlockLoop( array( $revision, 'importUpload' ) );
 198                 return false;
 199         }
 200
 201         /**
 202          * Mostly for hook use
 203          */
 204         public function finishImportPage( $title, $origTitle, $revCount, $sRevCount, $pageInfo ) {
 205                 $args = func_get_args();
 206                 return wfRunHooks( 'AfterImportPage', $args );
 207         }
 208
 209         /**
 210          * Alternate per-revision callback, for debugging.
 211          * @param $revision WikiRevision
 212          */
 213         public function debugRevisionHandler( &$revision ) {
 214                 $this->debug( "Got revision:" );
 215                 if( is_object( $revision->title ) ) {
 216                         $this->debug( "-- Title: " . $revision->title->getPrefixedText() );
 217                 } else {
 218                         $this->debug( "-- Title: <invalid>" );
 219                 }
 220                 $this->debug( "-- User: " . $revision->user_text );
 221                 $this->debug( "-- Timestamp: " . $revision->timestamp );
 222                 $this->debug( "-- Comment: " . $revision->comment );
 223                 $this->debug( "-- Text: " . $revision->text );
 224         }
 225
 226         /**
 227          * Notify the callback function when a new <page> is reached.
 228          * @param $title Title
 229          */
 230         function pageCallback( $title ) {
 231                 if( isset( $this->mPageCallback ) ) {
 232                         call_user_func( $this->mPageCallback, $title );
 233                 }
 234         }
 235
 236         /**
 237          * Notify the callback function when a </page> is closed.
 238          * @param $title Title
 239          * @param $origTitle Title
 240          * @param $revCount Integer
 241          * @param $sucCount Int: number of revisions for which callback returned true
 242          * @param $pageInfo Array: associative array of page information
 243          */
 244         private function pageOutCallback( $title, $origTitle, $revCount, $sucCount, $pageInfo ) {
 245                 if( isset( $this->mPageOutCallback ) ) {
 246                         $args = func_get_args();
 247                         call_user_func_array( $this->mPageOutCallback, $args );
 248                 }
 249         }
 250
 251         /**
 252          * Notify the callback function of a revision
 253          * @param $revision A WikiRevision object
 254          */
 255         private function revisionCallback( $revision ) {
 256                 if ( isset( $this->mRevisionCallback ) ) {
 257                         return call_user_func_array( $this->mRevisionCallback,
 258                                         array( $revision, $this ) );
 259                 } else {
 260                         return false;
 261                 }
 262         }
 263
 264         /**
 265          * Notify the callback function of a new log item
 266          * @param $revision A WikiRevision object
 267          */
 268         private function logItemCallback( $revision ) {
 269                 if ( isset( $this->mLogItemCallback ) ) {
 270                         return call_user_func_array( $this->mLogItemCallback,
 271                                         array( $revision, $this ) );
 272                 } else {
 273                         return false;
 274                 }
 275         }
 276
 277         /**
 278          * Shouldn't something like this be built-in to XMLReader?
 279          * Fetches text contents of the current element, assuming
 280          * no sub-elements or such scary things.
 281          * @return string
 282          * @access private
 283          */
 284         private function nodeContents() {
 285                 if( $this->reader->isEmptyElement ) {
 286                         return "";
 287                 }
 288                 $buffer = "";
 289                 while( $this->reader->read() ) {
 290                         switch( $this->reader->nodeType ) {
 291                         case XmlReader::TEXT:
 292                         case XmlReader::SIGNIFICANT_WHITESPACE:
 293                                 $buffer .= $this->reader->value;
 294                                 break;
 295                         case XmlReader::END_ELEMENT:
 296                                 return $buffer;
 297                         }
 298                 }
 299
 300                 $this->reader->close();
 301                 return '';
 302         }
 303
 304         # --------------
 305
 306         /** Left in for debugging */
 307         private function dumpElement() {
 308                 static $lookup = null;
 309                 if (!$lookup) {
 310                         $xmlReaderConstants = array(
 311                                 "NONE",
 312                                 "ELEMENT",
 313                                 "ATTRIBUTE",
 314                                 "TEXT",
 315                                 "CDATA",
 316                                 "ENTITY_REF",
 317                                 "ENTITY",
 318                                 "PI",
 319                                 "COMMENT",
 320                                 "DOC",
 321                                 "DOC_TYPE",
 322                                 "DOC_FRAGMENT",
 323                                 "NOTATION",
 324                                 "WHITESPACE",
 325                                 "SIGNIFICANT_WHITESPACE",
 326                                 "END_ELEMENT",
 327                                 "END_ENTITY",
 328                                 "XML_DECLARATION",
 329                                 );
 330                         $lookup = array();
 331
 332                         foreach( $xmlReaderConstants as $name ) {
 333                                 $lookup[constant("XmlReader::$name")] = $name;
 334                         }
 335                 }
 336
 337                 print( var_dump(
 338                         $lookup[$this->reader->nodeType],
 339                         $this->reader->name,
 340                         $this->reader->value
 341                 )."\n\n" );
 342         }
 343
 344         /**
 345          * Primary entry point
 346          */
 347         public function doImport() {
 348                 $this->reader->read();
 349
 350                 if ( $this->reader->name != 'mediawiki' ) {
 351                         throw new MWException( "Expected <mediawiki> tag, got ".
 352                                 $this->reader->name );
 353                 }
 354                 $this->debug( "<mediawiki> tag is correct." );
 355
 356                 $this->debug( "Starting primary dump processing loop." );
 357
 358                 $keepReading = $this->reader->read();
 359                 $skip = false;
 360                 while ( $keepReading ) {
 361                         $tag = $this->reader->name;
 362                         $type = $this->reader->nodeType;
 363
 364                         if ( !wfRunHooks( 'ImportHandleToplevelXMLTag', $this ) ) {
 365                                 // Do nothing
 366                         } elseif ( $tag == 'mediawiki' && $type == XmlReader::END_ELEMENT ) {
 367                                 break;
 368                         } elseif ( $tag == 'siteinfo' ) {
 369                                 $this->handleSiteInfo();
 370                         } elseif ( $tag == 'page' ) {
 371                                 $this->handlePage();
 372                         } elseif ( $tag == 'logitem' ) {
 373                                 $this->handleLogItem();
 374                         } elseif ( $tag != '#text' ) {
 375                                 $this->warn( "Unhandled top-level XML tag $tag" );
 376
 377                                 $skip = true;
 378                         }
 379
 380                         if ($skip) {
 381                                 $keepReading = $this->reader->next();
 382                                 $skip = false;
 383                                 $this->debug( "Skip" );
 384                         } else {
 385                                 $keepReading = $this->reader->read();
 386                         }
 387                 }
 388
 389                 return true;
 390         }
 391
 392         private function handleSiteInfo() {
 393                 // Site info is useful, but not actually used for dump imports.
 394                 // Includes a quick short-circuit to save performance.
 395                 if ( ! $this->mSiteInfoCallback ) {
 396                         $this->reader->next();
 397                         return true;
 398                 }
 399                 throw new MWException( "SiteInfo tag is not yet handled, do not set mSiteInfoCallback" );
 400         }
 401
 402         private function handleLogItem() {
 403                 $this->debug( "Enter log item handler." );
 404                 $logInfo = array();
 405
 406                 // Fields that can just be stuffed in the pageInfo object
 407                 $normalFields = array( 'id', 'comment', 'type', 'action', 'timestamp',
 408                                         'logtitle', 'params' );
 409
 410                 while ( $this->reader->read() ) {
 411                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 412                                         $this->reader->name == 'logitem') {
 413                                 break;
 414                         }
 415
 416                         $tag = $this->reader->name;
 417
 418                         if ( !wfRunHooks( 'ImportHandleLogItemXMLTag',
 419                                                 $this, $logInfo ) ) {
 420                                 // Do nothing
 421                         } elseif ( in_array( $tag, $normalFields ) ) {
 422                                 $logInfo[$tag] = $this->nodeContents();
 423                         } elseif ( $tag == 'contributor' ) {
 424                                 $logInfo['contributor'] = $this->handleContributor();
 425                         } elseif ( $tag != '#text' ) {
 426                                 $this->warn( "Unhandled log-item XML tag $tag" );
 427                         }
 428                 }
 429
 430                 $this->processLogItem( $logInfo );
 431         }
 432
 433         private function processLogItem( $logInfo ) {
 434                 $revision = new WikiRevision;
 435
 436                 $revision->setID( $logInfo['id'] );
 437                 $revision->setType( $logInfo['type'] );
 438                 $revision->setAction( $logInfo['action'] );
 439                 $revision->setTimestamp( $logInfo['timestamp'] );
 440                 $revision->setParams( $logInfo['params'] );
 441                 $revision->setTitle( Title::newFromText( $logInfo['logtitle'] ) );
 442
 443                 if ( isset( $logInfo['comment'] ) ) {
 444                         $revision->setComment( $logInfo['comment'] );
 445                 }
 446
 447                 if ( isset( $logInfo['contributor']['ip'] ) ) {
 448                         $revision->setUserIP( $logInfo['contributor']['ip'] );
 449                 }
 450                 if ( isset( $logInfo['contributor']['username'] ) ) {
 451                         $revision->setUserName( $logInfo['contributor']['username'] );
 452                 }
 453
 454                 return $this->logItemCallback( $revision );
 455         }
 456
 457         private function handlePage() {
 458                 // Handle page data.
 459                 $this->debug( "Enter page handler." );
 460                 $pageInfo = array( 'revisionCount' => 0, 'successfulRevisionCount' => 0 );
 461
 462                 // Fields that can just be stuffed in the pageInfo object
 463                 $normalFields = array( 'title', 'id', 'redirect', 'restrictions' );
 464
 465                 $skip = false;
 466                 $badTitle = false;
 467
 468                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 469                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 470                                         $this->reader->name == 'page') {
 471                                 break;
 472                         }
 473
 474                         $tag = $this->reader->name;
 475
 476                         if ( $badTitle ) {
 477                                 // The title is invalid, bail out of this page
 478                                 $skip = true;
 479                         } elseif ( !wfRunHooks( 'ImportHandlePageXMLTag', array( $this,
 480                                                 &$pageInfo ) ) ) {
 481                                 // Do nothing
 482                         } elseif ( in_array( $tag, $normalFields ) ) {
 483                                 $pageInfo[$tag] = $this->nodeContents();
 484                                 if ( $tag == 'title' ) {
 485                                         $title = $this->processTitle( $pageInfo['title'] );
 486
 487                                         if ( !$title ) {
 488                                                 $badTitle = true;
 489                                                 $skip = true;
 490                                         }
 491
 492                                         $this->pageCallback( $title );
 493                                         list( $pageInfo['_title'], $origTitle ) = $title;
 494                                 }
 495                         } elseif ( $tag == 'revision' ) {
 496                                 $this->handleRevision( $pageInfo );
 497                         } elseif ( $tag == 'upload' ) {
 498                                 $this->handleUpload( $pageInfo );
 499                         } elseif ( $tag != '#text' ) {
 500                                 $this->warn( "Unhandled page XML tag $tag" );
 501                                 $skip = true;
 502                         }
 503                 }
 504
 505                 $this->pageOutCallback( $pageInfo['_title'], $origTitle,
 506                                         $pageInfo['revisionCount'],
 507                                         $pageInfo['successfulRevisionCount'],
 508                                         $pageInfo );
 509         }
 510
 511         private function handleRevision( &$pageInfo ) {
 512                 $this->debug( "Enter revision handler" );
 513                 $revisionInfo = array();
 514
 515                 $normalFields = array( 'id', 'timestamp', 'comment', 'minor', 'text' );
 516
 517                 $skip = false;
 518
 519                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 520                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 521                                         $this->reader->name == 'revision') {
 522                                 break;
 523                         }
 524
 525                         $tag = $this->reader->name;
 526
 527                         if ( !wfRunHooks( 'ImportHandleRevisionXMLTag', $this,
 528                                                 $pageInfo, $revisionInfo ) ) {
 529                                 // Do nothing
 530                         } elseif ( in_array( $tag, $normalFields ) ) {
 531                                 $revisionInfo[$tag] = $this->nodeContents();
 532                         } elseif ( $tag == 'contributor' ) {
 533                                 $revisionInfo['contributor'] = $this->handleContributor();
 534                         } elseif ( $tag != '#text' ) {
 535                                 $this->warn( "Unhandled revision XML tag $tag" );
 536                                 $skip = true;
 537                         }
 538                 }
 539
 540                 $pageInfo['revisionCount']++;
 541                 if ( $this->processRevision( $pageInfo, $revisionInfo ) ) {
 542                         $pageInfo['successfulRevisionCount']++;
 543                 }
 544         }
 545
 546         private function processRevision( $pageInfo, $revisionInfo ) {
 547                 $revision = new WikiRevision;
 548
 549                 if( isset( $revisionInfo['id'] ) ) {
 550                         $revision->setID( $revisionInfo['id'] );
 551                 }
 552                 if ( isset( $revisionInfo['text'] ) ) {
 553                         $revision->setText( $revisionInfo['text'] );
 554                 }
 555                 $revision->setTitle( $pageInfo['_title'] );
 556
 557                 if ( isset( $revisionInfo['timestamp'] ) ) {
 558                         $revision->setTimestamp( $revisionInfo['timestamp'] );
 559                 } else {
 560                         $revision->setTimestamp( wfTimestampNow() );
 561                 }
 562
 563                 if ( isset( $revisionInfo['comment'] ) ) {
 564                         $revision->setComment( $revisionInfo['comment'] );
 565                 }
 566
 567                 if ( isset( $revisionInfo['minor'] ) ) {
 568                         $revision->setMinor( true );
 569                 }
 570                 if ( isset( $revisionInfo['contributor']['ip'] ) ) {
 571                         $revision->setUserIP( $revisionInfo['contributor']['ip'] );
 572                 }
 573                 if ( isset( $revisionInfo['contributor']['username'] ) ) {
 574                         $revision->setUserName( $revisionInfo['contributor']['username'] );
 575                 }
 576
 577                 return $this->revisionCallback( $revision );
 578         }
 579
 580         private function handleUpload( &$pageInfo ) {
 581                 $this->debug( "Enter upload handler" );
 582                 $uploadInfo = array();
 583
 584                 $normalFields = array( 'timestamp', 'comment', 'filename', 'text',
 585                                         'src', 'size' );
 586
 587                 $skip = false;
 588
 589                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 590                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 591                                         $this->reader->name == 'upload') {
 592                                 break;
 593                         }
 594
 595                         $tag = $this->reader->name;
 596
 597                         if ( !wfRunHooks( 'ImportHandleUploadXMLTag', $this,
 598                                                 $pageInfo ) ) {
 599                                 // Do nothing
 600                         } elseif ( in_array( $tag, $normalFields ) ) {
 601                                 $uploadInfo[$tag] = $this->nodeContents();
 602                         } elseif ( $tag == 'contributor' ) {
 603                                 $uploadInfo['contributor'] = $this->handleContributor();
 604                         } elseif ( $tag != '#text' ) {
 605                                 $this->warn( "Unhandled upload XML tag $tag" );
 606                                 $skip = true;
 607                         }
 608                 }
 609
 610                 return $this->processUpload( $pageInfo, $uploadInfo );
 611         }
 612
 613
 614         private function processUpload( $pageInfo, $uploadInfo ) {
 615                 $revision = new WikiRevision;
 616                            $text = isset( $uploadInfo['text'] ) ? $uploadInfo['text'] : '';
 617
 618                 $revision->setTitle( $pageInfo['_title'] );
 619                            $revision->setID( $pageInfo['id'] );
 620                 $revision->setTimestamp( $uploadInfo['timestamp'] );
 621                            $revision->setText( $text );
 622                 $revision->setFilename( $uploadInfo['filename'] );
 623                 $revision->setSrc( $uploadInfo['src'] );
 624                 $revision->setSize( intval( $uploadInfo['size'] ) );
 625                 $revision->setComment( $uploadInfo['comment'] );
 626
 627                 if ( isset( $uploadInfo['contributor']['ip'] ) ) {
 628                         $revision->setUserIP( $uploadInfo['contributor']['ip'] );
 629                 }
 630                 if ( isset( $uploadInfo['contributor']['username'] ) ) {
 631                         $revision->setUserName( $uploadInfo['contributor']['username'] );
 632                 }
 633
 634                            return call_user_func( $this->mUploadCallback, $revision );
 635         }
 636
 637         private function handleContributor() {
 638                 $fields = array( 'id', 'ip', 'username' );
 639                 $info = array();
 640
 641                 while ( $this->reader->read() ) {
 642                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 643                                         $this->reader->name == 'contributor') {
 644                                 break;
 645                         }
 646
 647                         $tag = $this->reader->name;
 648
 649                         if ( in_array( $tag, $fields ) ) {
 650                                 $info[$tag] = $this->nodeContents();
 651                         }
 652                 }
 653
 654                 return $info;
 655         }
 656
 657         private function processTitle( $text ) {
 658                 $workTitle = $text;
 659                 $origTitle = Title::newFromText( $workTitle );
 660
 661                 if( !is_null( $this->mTargetNamespace ) && !is_null( $origTitle ) ) {
 662                         $title = Title::makeTitle( $this->mTargetNamespace,
 663                                 $origTitle->getDBkey() );
 664                 } else {
 665                         $title = Title::newFromText( $workTitle );
 666                 }
 667
 668                 if( is_null( $title ) ) {
 669                         // Invalid page title? Ignore the page
 670                         $this->notice( "Skipping invalid page title '$workTitle'" );
 671                         return false;
 672                 } elseif( $title->getInterwiki() != '' ) {
 673                         $this->notice( "Skipping interwiki page title '$workTitle'" );
 674                         return false;
 675                 }
 676
 677                 return array( $title, $origTitle );
 678         }
 679 }
 680
 681 /** This is a horrible hack used to keep source compatibility */
 682 class UploadSourceAdapter {
 683         static $sourceRegistrations = array();
 684
 685         private $mSource;
 686         private $mBuffer;
 687         private $mPosition;
 688
 689         static function registerSource( $source ) {
 690                 $id = wfGenerateToken();
 691
 692                 self::$sourceRegistrations[$id] = $source;
 693
 694                 return $id;
 695         }
 696
 697         function stream_open( $path, $mode, $options, &$opened_path ) {
 698                 $url = parse_url($path);
 699                 $id = $url['host'];
 700
 701                 if ( !isset( self::$sourceRegistrations[$id] ) ) {
 702                         return false;
 703                 }
 704
 705                 $this->mSource = self::$sourceRegistrations[$id];
 706
 707                 return true;
 708         }
 709
 710         function stream_read( $count ) {
 711                 $return = '';
 712                 $leave = false;
 713
 714                 while ( !$leave && !$this->mSource->atEnd() &&
 715                                 strlen($this->mBuffer) < $count ) {
 716                         $read = $this->mSource->readChunk();
 717
 718                         if ( !strlen($read) ) {
 719                                 $leave = true;
 720                         }
 721
 722                         $this->mBuffer .= $read;
 723                 }
 724
 725                 if ( strlen($this->mBuffer) ) {
 726                         $return = substr( $this->mBuffer, 0, $count );
 727                         $this->mBuffer = substr( $this->mBuffer, $count );
 728                 }
 729
 730                 $this->mPosition += strlen($return);
 731
 732                 return $return;
 733         }
 734
 735         function stream_write( $data ) {
 736                 return false;
 737         }
 738
 739         function stream_tell() {
 740                 return $this->mPosition;
 741         }
 742
 743         function stream_eof() {
 744                 return $this->mSource->atEnd();
 745         }
 746
 747         function url_stat() {
 748                 $result = array();
 749
 750                 $result['dev'] = $result[0] = 0;
 751                 $result['ino'] = $result[1] = 0;
 752                 $result['mode'] = $result[2] = 0;
 753                 $result['nlink'] = $result[3] = 0;
 754                 $result['uid'] = $result[4] = 0;
 755                 $result['gid'] = $result[5] = 0;
 756                 $result['rdev'] = $result[6] = 0;
 757                 $result['size'] = $result[7] = 0;
 758                 $result['atime'] = $result[8] = 0;
 759                 $result['mtime'] = $result[9] = 0;
 760                 $result['ctime'] = $result[10] = 0;
 761                 $result['blksize'] = $result[11] = 0;
 762                 $result['blocks'] = $result[12] = 0;
 763
 764                 return $result;
 765         }
 766 }
 767
 768 class XMLReader2 extends XMLReader {
 769         function nodeContents() {
 770                 if( $this->isEmptyElement ) {
 771                         return "";
 772                 }
 773                 $buffer = "";
 774                 while( $this->read() ) {
 775                         switch( $this->nodeType ) {
 776                         case XmlReader::TEXT:
 777                         case XmlReader::SIGNIFICANT_WHITESPACE:
 778                                 $buffer .= $this->value;
 779                                 break;
 780                         case XmlReader::END_ELEMENT:
 781                                 return $buffer;
 782                         }
 783                 }
 784                 return $this->close();
 785         }
 786 }
 787
 788 /**
 789  * @todo document (e.g. one-sentence class description).
 790  * @ingroup SpecialPage
 791  */
 792 class WikiRevision {
 793         var $title = null;
 794         var $id = 0;
 795         var $timestamp = "20010115000000";
 796         var $user = 0;
 797         var $user_text = "";
 798         var $text = "";
 799         var $comment = "";
 800         var $minor = false;
 801         var $type = "";
 802         var $action = "";
 803         var $params = "";
 804
 805         function setTitle( $title ) {
 806                 if( is_object( $title ) ) {
 807                         $this->title = $title;
 808                 } elseif( is_null( $title ) ) {
 809                         throw new MWException( "WikiRevision given a null title in import. You may need to adjust \$wgLegalTitleChars." );
 810                 } else {
 811                         throw new MWException( "WikiRevision given non-object title in import." );
 812                 }
 813         }
 814
 815         function setID( $id ) {
 816                 $this->id = $id;
 817         }
 818
 819         function setTimestamp( $ts ) {
 820                 # 2003-08-05T18:30:02Z
 821                 $this->timestamp = wfTimestamp( TS_MW, $ts );
 822         }
 823
 824         function setUsername( $user ) {
 825                 $this->user_text = $user;
 826         }
 827
 828         function setUserIP( $ip ) {
 829                 $this->user_text = $ip;
 830         }
 831
 832         function setText( $text ) {
 833                 $this->text = $text;
 834         }
 835
 836         function setComment( $text ) {
 837                 $this->comment = $text;
 838         }
 839
 840         function setMinor( $minor ) {
 841                 $this->minor = (bool)$minor;
 842         }
 843
 844         function setSrc( $src ) {
 845                 $this->src = $src;
 846         }
 847
 848         function setFilename( $filename ) {
 849                 $this->filename = $filename;
 850         }
 851
 852         function setSize( $size ) {
 853                 $this->size = intval( $size );
 854         }
 855
 856         function setType( $type ) {
 857                 $this->type = $type;
 858         }
 859
 860         function setAction( $action ) {
 861                 $this->action = $action;
 862         }
 863
 864         function setParams( $params ) {
 865                 $this->params = $params;
 866         }
 867
 868         function getTitle() {
 869                 return $this->title;
 870         }
 871
 872         function getID() {
 873                 return $this->id;
 874         }
 875
 876         function getTimestamp() {
 877                 return $this->timestamp;
 878         }
 879
 880         function getUser() {
 881                 return $this->user_text;
 882         }
 883
 884         function getText() {
 885                 return $this->text;
 886         }
 887
 888         function getComment() {
 889                 return $this->comment;
 890         }
 891
 892         function getMinor() {
 893                 return $this->minor;
 894         }
 895
 896         function getSrc() {
 897                 return $this->src;
 898         }
 899
 900         function getFilename() {
 901                 return $this->filename;
 902         }
 903
 904         function getSize() {
 905                 return $this->size;
 906         }
 907
 908         function getType() {
 909                 return $this->type;
 910         }
 911
 912         function getAction() {
 913                 return $this->action;
 914         }
 915
 916         function getParams() {
 917                 return $this->params;
 918         }
 919
 920         function importOldRevision() {
 921                 $dbw = wfGetDB( DB_MASTER );
 922
 923                 # Sneak a single revision into place
 924                 $user = User::newFromName( $this->getUser() );
 925                 if( $user ) {
 926                         $userId = intval( $user->getId() );
 927                         $userText = $user->getName();
 928                 } else {
 929                         $userId = 0;
 930                         $userText = $this->getUser();
 931                 }
 932
 933                 // avoid memory leak...?
 934                 $linkCache = LinkCache::singleton();
 935                 $linkCache->clear();
 936
 937                 $article = new Article( $this->title );
 938                 $pageId = $article->getId();
 939                 if( $pageId == 0 ) {
 940                         # must create the page...
 941                         $pageId = $article->insertOn( $dbw );
 942                         $created = true;
 943                 } else {
 944                         $created = false;
 945
 946                         $prior = $dbw->selectField( 'revision', '1',
 947                                 array( 'rev_page' => $pageId,
 948                                         'rev_timestamp' => $dbw->timestamp( $this->timestamp ),
 949                                         'rev_user_text' => $userText,
 950                                         'rev_comment'   => $this->getComment() ),
 951                                 __METHOD__
 952                         );
 953                         if( $prior ) {
 954                                 // FIXME: this could fail slightly for multiple matches :P
 955                                 wfDebug( __METHOD__ . ": skipping existing revision for [[" .
 956                                         $this->title->getPrefixedText() . "]], timestamp " . $this->timestamp . "\n" );
 957                                 return false;
 958                         }
 959                 }
 960
 961                 # FIXME: Use original rev_id optionally (better for backups)
 962                 # Insert the row
 963                 $revision = new Revision( array(
 964                         'page'       => $pageId,
 965                         'text'       => $this->getText(),
 966                         'comment'    => $this->getComment(),
 967                         'user'       => $userId,
 968                         'user_text'  => $userText,
 969                         'timestamp'  => $this->timestamp,
 970                         'minor_edit' => $this->minor,
 971                         ) );
 972                 $revId = $revision->insertOn( $dbw );
 973                 $changed = $article->updateIfNewerOn( $dbw, $revision );
 974
 975                 # To be on the safe side...
 976                 $tempTitle = $GLOBALS['wgTitle'];
 977                 $GLOBALS['wgTitle'] = $this->title;
 978
 979                 if( $created ) {
 980                         wfDebug( __METHOD__ . ": running onArticleCreate\n" );
 981                         Article::onArticleCreate( $this->title );
 982
 983                         wfDebug( __METHOD__ . ": running create updates\n" );
 984                         $article->createUpdates( $revision );
 985
 986                 } elseif( $changed ) {
 987                         wfDebug( __METHOD__ . ": running onArticleEdit\n" );
 988                         Article::onArticleEdit( $this->title );
 989
 990                         wfDebug( __METHOD__ . ": running edit updates\n" );
 991                         $article->editUpdates(
 992                                 $this->getText(),
 993                                 $this->getComment(),
 994                                 $this->minor,
 995                                 $this->timestamp,
 996                                 $revId );
 997                 }
 998                 $GLOBALS['wgTitle'] = $tempTitle;
 999
1000                 return true;
1001         }
1002
1003         function importLogItem() {
1004                 $dbw = wfGetDB( DB_MASTER );
1005                 # FIXME: this will not record autoblocks
1006                 if( !$this->getTitle() ) {
1007                         wfDebug( __METHOD__ . ": skipping invalid {$this->type}/{$this->action} log time, timestamp " .
1008                                 $this->timestamp . "\n" );
1009                         return;
1010                 }
1011                 # Check if it exists already
1012                 // FIXME: use original log ID (better for backups)
1013                 $prior = $dbw->selectField( 'logging', '1',
1014                         array( 'log_type' => $this->getType(),
1015                                 'log_action'    => $this->getAction(),
1016                                 'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1017                                 'log_namespace' => $this->getTitle()->getNamespace(),
1018                                 'log_title'     => $this->getTitle()->getDBkey(),
1019                                 'log_comment'   => $this->getComment(),
1020                                 #'log_user_text' => $this->user_text,
1021                                 'log_params'    => $this->params ),
1022                         __METHOD__
1023                 );
1024                 // FIXME: this could fail slightly for multiple matches :P
1025                 if( $prior ) {
1026                         wfDebug( __METHOD__ . ": skipping existing item for Log:{$this->type}/{$this->action}, timestamp " .
1027                                 $this->timestamp . "\n" );
1028                         return false;
1029                 }
1030                 $log_id = $dbw->nextSequenceValue( 'logging_log_id_seq' );
1031                 $data = array(
1032                         'log_id' => $log_id,
1033                         'log_type' => $this->type,
1034                         'log_action' => $this->action,
1035                         'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1036                         'log_user' => User::idFromName( $this->user_text ),
1037                         #'log_user_text' => $this->user_text,
1038                         'log_namespace' => $this->getTitle()->getNamespace(),
1039                         'log_title' => $this->getTitle()->getDBkey(),
1040                         'log_comment' => $this->getComment(),
1041                         'log_params' => $this->params
1042                 );
1043                 $dbw->insert( 'logging', $data, __METHOD__ );
1044         }
1045
1046         function importUpload() {
1047                 wfDebug( __METHOD__ . ": STUB\n" );
1048
1049                 /**
1050                         // from file revert...
1051                         $source = $this->file->getArchiveVirtualUrl( $this->oldimage );
1052                         $comment = $wgRequest->getText( 'wpComment' );
1053                         // TODO: Preserve file properties from database instead of reloading from file
1054                         $status = $this->file->upload( $source, $comment, $comment );
1055                         if( $status->isGood() ) {
1056                 */
1057
1058                 /**
1059                         // from file upload...
1060                 $this->mLocalFile = wfLocalFile( $nt );
1061                 $this->mDestName = $this->mLocalFile->getName();
1062                 //....
1063                         $status = $this->mLocalFile->upload( $this->mTempPath, $this->mComment, $pageText,
1064                         File::DELETE_SOURCE, $this->mFileProps );
1065                         if ( !$status->isGood() ) {
1066                                 $resultDetails = array( 'internal' => $status->getWikiText() );
1067                 */
1068
1069                            // @todo Fixme: it may create a page without our desire, also wrong potentially.
1070                 // and, it will record a *current* upload, but we might want an archive version here
1071
1072                 $file = wfLocalFile( $this->getTitle() );
1073                 if( !$file ) {
1074                         wfDebug( "IMPORT: Bad file. :(\n" );
1075                         return false;
1076                 }
1077
1078                 $source = $this->downloadSource();
1079                 if( !$source ) {
1080                         wfDebug( "IMPORT: Could not fetch remote file. :(\n" );
1081                         return false;
1082                 }
1083
1084                 $user = User::newFromName( $this->user_text );
1085
1086                 $status = $file->upload( $source,
1087                         $this->getComment(),
1088                         $this->getComment(), // Initial page, if none present...
1089                         File::DELETE_SOURCE,
1090                         false, // props...
1091                                            $this->getTimestamp(),
1092                                            is_object( $user ) ? ( $user->isLoggedIn() ? $user : null ) : null );
1093
1094                 if( $status->isGood() ) {
1095                         // yay?
1096                         wfDebug( "IMPORT: is ok?\n" );
1097                         return true;
1098                 }
1099
1100                 wfDebug( "IMPORT: is bad? " . $status->getXml() . "\n" );
1101                 return false;
1102
1103         }
1104
1105         function downloadSource() {
1106                 global $wgEnableUploads;
1107                 if( !$wgEnableUploads ) {
1108                         return false;
1109                 }
1110
1111                 $tempo = tempnam( wfTempDir(), 'download' );
1112                 $f = fopen( $tempo, 'wb' );
1113                 if( !$f ) {
1114                         wfDebug( "IMPORT: couldn't write to temp file $tempo\n" );
1115                         return false;
1116                 }
1117
1118                 // @todo Fixme!
1119                 $src = $this->getSrc();
1120                 $data = Http::get( $src );
1121                 if( !$data ) {
1122                         wfDebug( "IMPORT: couldn't fetch source $src\n" );
1123                         fclose( $f );
1124                         unlink( $tempo );
1125                         return false;
1126                 }
1127
1128                 fwrite( $f, $data );
1129                 fclose( $f );
1130
1131                 return $tempo;
1132         }
1133
1134 }
1135
1136 /**
1137  * @todo document (e.g. one-sentence class description).
1138  * @ingroup SpecialPage
1139  */
1140 class ImportStringSource {
1141         function __construct( $string ) {
1142                 $this->mString = $string;
1143                 $this->mRead = false;
1144         }
1145
1146         function atEnd() {
1147                 return $this->mRead;
1148         }
1149
1150         function readChunk() {
1151                 if( $this->atEnd() ) {
1152                         return false;
1153                 } else {
1154                         $this->mRead = true;
1155                         return $this->mString;
1156                 }
1157         }
1158 }
1159
1160 /**
1161  * @todo document (e.g. one-sentence class description).
1162  * @ingroup SpecialPage
1163  */
1164 class ImportStreamSource {
1165         function __construct( $handle ) {
1166                 $this->mHandle = $handle;
1167         }
1168
1169         function atEnd() {
1170                 return feof( $this->mHandle );
1171         }
1172
1173         function readChunk() {
1174                 return fread( $this->mHandle, 32768 );
1175         }
1176
1177         static function newFromFile( $filename ) {
1178                 $file = @fopen( $filename, 'rt' );
1179                 if( !$file ) {
1180                         return Status::newFatal( "importcantopen" );
1181                 }
1182                 return Status::newGood( new ImportStreamSource( $file ) );
1183         }
1184
1185         static function newFromUpload( $fieldname = "xmlimport" ) {
1186                 $upload =& $_FILES[$fieldname];
1187
1188                 if( !isset( $upload ) || !$upload['name'] ) {
1189                         return Status::newFatal( 'importnofile' );
1190                 }
1191                 if( !empty( $upload['error'] ) ) {
1192                         switch($upload['error']){
1193                                 case 1: # The uploaded file exceeds the upload_max_filesize directive in php.ini.
1194                                         return Status::newFatal( 'importuploaderrorsize' );
1195                                 case 2: # The uploaded file exceeds the MAX_FILE_SIZE directive that was specified in the HTML form.
1196                                         return Status::newFatal( 'importuploaderrorsize' );
1197                                 case 3: # The uploaded file was only partially uploaded
1198                                         return Status::newFatal( 'importuploaderrorpartial' );
1199                                 case 6: #Missing a temporary folder.
1200                                         return Status::newFatal( 'importuploaderrortemp' );
1201                                 # case else: # Currently impossible
1202                         }
1203
1204                 }
1205                 $fname = $upload['tmp_name'];
1206                 if( is_uploaded_file( $fname ) ) {
1207                         return ImportStreamSource::newFromFile( $fname );
1208                 } else {
1209                         return Status::newFatal( 'importnofile' );
1210                 }
1211         }
1212
1213         static function newFromURL( $url, $method = 'GET' ) {
1214                 wfDebug( __METHOD__ . ": opening $url\n" );
1215                 # Use the standard HTTP fetch function; it times out
1216                 # quicker and sorts out user-agent problems which might
1217                 # otherwise prevent importing from large sites, such
1218                 # as the Wikimedia cluster, etc.
1219                 $data = Http::request( $method, $url );
1220                 if( $data !== false ) {
1221                         $file = tmpfile();
1222                         fwrite( $file, $data );
1223                         fflush( $file );
1224                         fseek( $file, 0 );
1225                         return Status::newGood( new ImportStreamSource( $file ) );
1226                 } else {
1227                         return Status::newFatal( 'importcantopen' );
1228                 }
1229         }
1230
1231         public static function newFromInterwiki( $interwiki, $page, $history = false, $templates = false, $pageLinkDepth = 0 ) {
1232                 if( $page == '' ) {
1233                         return Status::newFatal( 'import-noarticle' );
1234                 }
1235                 $link = Title::newFromText( "$interwiki:Special:Export/$page" );
1236                 if( is_null( $link ) || $link->getInterwiki() == '' ) {
1237                         return Status::newFatal( 'importbadinterwiki' );
1238                 } else {
1239                         $params = array();
1240                         if ( $history ) $params['history'] = 1;
1241                         if ( $templates ) $params['templates'] = 1;
1242                         if ( $pageLinkDepth ) $params['pagelink-depth'] = $pageLinkDepth;
1243                         $url = $link->getFullUrl( $params );
1244                         # For interwikis, use POST to avoid redirects.
1245                         return ImportStreamSource::newFromURL( $url, "POST" );
1246                 }
1247         }
1248 }