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