includes/import/WikiRevision.php

   1 <?php
   2 /**
   3  * MediaWiki page data importer.
   4  *
   5  * Copyright © 2003,2005 Brion Vibber <brion@pobox.com>
   6  * https://www.mediawiki.org/
   7  *
   8  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 2 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with this program; if not, write to the Free Software Foundation, Inc.,
  20  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  21  * http://www.gnu.org/copyleft/gpl.html
  22  *
  23  * @file
  24  * @ingroup SpecialPage
  25  */
  26
  27 /**
  28  * Represents a revision, log entry or upload during the import process.
  29  * This class sticks closely to the structure of the XML dump.
  30  *
  31  * @ingroup SpecialPage
  32  */
  33 class WikiRevision {
  34         /** @todo Unused? */
  35         public $importer = null;
  36
  37         /** @var Title */
  38         public $title = null;
  39
  40         /** @var int */
  41         public $id = 0;
  42
  43         /** @var string */
  44         public $timestamp = "20010115000000";
  45
  46         /**
  47          * @var int
  48          * @todo Can't find any uses. Public, because that's suspicious. Get clarity. */
  49         public $user = 0;
  50
  51         /** @var string */
  52         public $user_text = "";
  53
  54         /** @var User */
  55         public $userObj = null;
  56
  57         /** @var string */
  58         public $model = null;
  59
  60         /** @var string */
  61         public $format = null;
  62
  63         /** @var string */
  64         public $text = "";
  65
  66         /** @var int */
  67         protected $size;
  68
  69         /** @var Content */
  70         public $content = null;
  71
  72         /** @var ContentHandler */
  73         protected $contentHandler = null;
  74
  75         /** @var string */
  76         public $comment = "";
  77
  78         /** @var bool */
  79         public $minor = false;
  80
  81         /** @var string */
  82         public $type = "";
  83
  84         /** @var string */
  85         public $action = "";
  86
  87         /** @var string */
  88         public $params = "";
  89
  90         /** @var string */
  91         public $fileSrc = '';
  92
  93         /** @var bool|string */
  94         public $sha1base36 = false;
  95
  96         /**
  97          * @var bool
  98          * @todo Unused?
  99          */
 100         public $isTemp = false;
 101
 102         /** @var string */
 103         public $archiveName = '';
 104
 105         protected $filename;
 106
 107         /** @var mixed */
 108         protected $src;
 109
 110         /** @todo Unused? */
 111         public $fileIsTemp;
 112
 113         /** @var bool */
 114         private $mNoUpdates = false;
 115
 116         /** @var Config $config */
 117         private $config;
 118
 119         public function __construct( Config $config ) {
 120                 $this->config = $config;
 121         }
 122
 123         /**
 124          * @param Title $title
 125          * @throws MWException
 126          */
 127         public function setTitle( $title ) {
 128                 if ( is_object( $title ) ) {
 129                         $this->title = $title;
 130                 } elseif ( is_null( $title ) ) {
 131                         throw new MWException( "WikiRevision given a null title in import. "
 132                                 . "You may need to adjust \$wgLegalTitleChars." );
 133                 } else {
 134                         throw new MWException( "WikiRevision given non-object title in import." );
 135                 }
 136         }
 137
 138         /**
 139          * @param int $id
 140          */
 141         public function setID( $id ) {
 142                 $this->id = $id;
 143         }
 144
 145         /**
 146          * @param string $ts
 147          */
 148         public function setTimestamp( $ts ) {
 149                 # 2003-08-05T18:30:02Z
 150                 $this->timestamp = wfTimestamp( TS_MW, $ts );
 151         }
 152
 153         /**
 154          * @param string $user
 155          */
 156         public function setUsername( $user ) {
 157                 $this->user_text = $user;
 158         }
 159
 160         /**
 161          * @param User $user
 162          */
 163         public function setUserObj( $user ) {
 164                 $this->userObj = $user;
 165         }
 166
 167         /**
 168          * @param string $ip
 169          */
 170         public function setUserIP( $ip ) {
 171                 $this->user_text = $ip;
 172         }
 173
 174         /**
 175          * @param string $model
 176          */
 177         public function setModel( $model ) {
 178                 $this->model = $model;
 179         }
 180
 181         /**
 182          * @param string $format
 183          */
 184         public function setFormat( $format ) {
 185                 $this->format = $format;
 186         }
 187
 188         /**
 189          * @param string $text
 190          */
 191         public function setText( $text ) {
 192                 $this->text = $text;
 193         }
 194
 195         /**
 196          * @param string $text
 197          */
 198         public function setComment( $text ) {
 199                 $this->comment = $text;
 200         }
 201
 202         /**
 203          * @param bool $minor
 204          */
 205         public function setMinor( $minor ) {
 206                 $this->minor = (bool)$minor;
 207         }
 208
 209         /**
 210          * @param mixed $src
 211          */
 212         public function setSrc( $src ) {
 213                 $this->src = $src;
 214         }
 215
 216         /**
 217          * @param string $src
 218          * @param bool $isTemp
 219          */
 220         public function setFileSrc( $src, $isTemp ) {
 221                 $this->fileSrc = $src;
 222                 $this->fileIsTemp = $isTemp;
 223         }
 224
 225         /**
 226          * @param string $sha1base36
 227          */
 228         public function setSha1Base36( $sha1base36 ) {
 229                 $this->sha1base36 = $sha1base36;
 230         }
 231
 232         /**
 233          * @param string $filename
 234          */
 235         public function setFilename( $filename ) {
 236                 $this->filename = $filename;
 237         }
 238
 239         /**
 240          * @param string $archiveName
 241          */
 242         public function setArchiveName( $archiveName ) {
 243                 $this->archiveName = $archiveName;
 244         }
 245
 246         /**
 247          * @param int $size
 248          */
 249         public function setSize( $size ) {
 250                 $this->size = intval( $size );
 251         }
 252
 253         /**
 254          * @param string $type
 255          */
 256         public function setType( $type ) {
 257                 $this->type = $type;
 258         }
 259
 260         /**
 261          * @param string $action
 262          */
 263         public function setAction( $action ) {
 264                 $this->action = $action;
 265         }
 266
 267         /**
 268          * @param array $params
 269          */
 270         public function setParams( $params ) {
 271                 $this->params = $params;
 272         }
 273
 274         /**
 275          * @param bool $noupdates
 276          */
 277         public function setNoUpdates( $noupdates ) {
 278                 $this->mNoUpdates = $noupdates;
 279         }
 280
 281         /**
 282          * @return Title
 283          */
 284         public function getTitle() {
 285                 return $this->title;
 286         }
 287
 288         /**
 289          * @return int
 290          */
 291         public function getID() {
 292                 return $this->id;
 293         }
 294
 295         /**
 296          * @return string
 297          */
 298         public function getTimestamp() {
 299                 return $this->timestamp;
 300         }
 301
 302         /**
 303          * @return string
 304          */
 305         public function getUser() {
 306                 return $this->user_text;
 307         }
 308
 309         /**
 310          * @return User
 311          */
 312         public function getUserObj() {
 313                 return $this->userObj;
 314         }
 315
 316         /**
 317          * @return string
 318          */
 319         public function getText() {
 320                 return $this->text;
 321         }
 322
 323         /**
 324          * @return ContentHandler
 325          */
 326         public function getContentHandler() {
 327                 if ( is_null( $this->contentHandler ) ) {
 328                         $this->contentHandler = ContentHandler::getForModelID( $this->getModel() );
 329                 }
 330
 331                 return $this->contentHandler;
 332         }
 333
 334         /**
 335          * @return Content
 336          */
 337         public function getContent() {
 338                 if ( is_null( $this->content ) ) {
 339                         $handler = $this->getContentHandler();
 340                         $this->content = $handler->unserializeContent( $this->text, $this->getFormat() );
 341                 }
 342
 343                 return $this->content;
 344         }
 345
 346         /**
 347          * @return string
 348          */
 349         public function getModel() {
 350                 if ( is_null( $this->model ) ) {
 351                         $this->model = $this->getTitle()->getContentModel();
 352                 }
 353
 354                 return $this->model;
 355         }
 356
 357         /**
 358          * @return string
 359          */
 360         public function getFormat() {
 361                 if ( is_null( $this->format ) ) {
 362                         $this->format = $this->getContentHandler()->getDefaultFormat();
 363                 }
 364
 365                 return $this->format;
 366         }
 367
 368         /**
 369          * @return string
 370          */
 371         public function getComment() {
 372                 return $this->comment;
 373         }
 374
 375         /**
 376          * @return bool
 377          */
 378         public function getMinor() {
 379                 return $this->minor;
 380         }
 381
 382         /**
 383          * @return mixed
 384          */
 385         public function getSrc() {
 386                 return $this->src;
 387         }
 388
 389         /**
 390          * @return bool|string
 391          */
 392         public function getSha1() {
 393                 if ( $this->sha1base36 ) {
 394                         return Wikimedia\base_convert( $this->sha1base36, 36, 16 );
 395                 }
 396                 return false;
 397         }
 398
 399         /**
 400          * @return string
 401          */
 402         public function getFileSrc() {
 403                 return $this->fileSrc;
 404         }
 405
 406         /**
 407          * @return bool
 408          */
 409         public function isTempSrc() {
 410                 return $this->isTemp;
 411         }
 412
 413         /**
 414          * @return mixed
 415          */
 416         public function getFilename() {
 417                 return $this->filename;
 418         }
 419
 420         /**
 421          * @return string
 422          */
 423         public function getArchiveName() {
 424                 return $this->archiveName;
 425         }
 426
 427         /**
 428          * @return mixed
 429          */
 430         public function getSize() {
 431                 return $this->size;
 432         }
 433
 434         /**
 435          * @return string
 436          */
 437         public function getType() {
 438                 return $this->type;
 439         }
 440
 441         /**
 442          * @return string
 443          */
 444         public function getAction() {
 445                 return $this->action;
 446         }
 447
 448         /**
 449          * @return string
 450          */
 451         public function getParams() {
 452                 return $this->params;
 453         }
 454
 455         /**
 456          * @return bool
 457          */
 458         public function importOldRevision() {
 459                 $dbw = wfGetDB( DB_MASTER );
 460
 461                 # Sneak a single revision into place
 462                 $user = $this->getUserObj() ?: User::newFromName( $this->getUser() );
 463                 if ( $user ) {
 464                         $userId = intval( $user->getId() );
 465                         $userText = $user->getName();
 466                 } else {
 467                         $userId = 0;
 468                         $userText = $this->getUser();
 469                         $user = new User;
 470                 }
 471
 472                 // avoid memory leak...?
 473                 Title::clearCaches();
 474
 475                 $page = WikiPage::factory( $this->title );
 476                 $page->loadPageData( 'fromdbmaster' );
 477                 if ( !$page->exists() ) {
 478                         // must create the page...
 479                         $pageId = $page->insertOn( $dbw );
 480                         $created = true;
 481                         $oldcountable = null;
 482                 } else {
 483                         $pageId = $page->getId();
 484                         $created = false;
 485
 486                         $prior = $dbw->selectField( 'revision', '1',
 487                                 [ 'rev_page' => $pageId,
 488                                         'rev_timestamp' => $dbw->timestamp( $this->timestamp ),
 489                                         'rev_user_text' => $userText,
 490                                         'rev_comment' => $this->getComment() ],
 491                                 __METHOD__
 492                         );
 493                         if ( $prior ) {
 494                                 // @todo FIXME: This could fail slightly for multiple matches :P
 495                                 wfDebug( __METHOD__ . ": skipping existing revision for [[" .
 496                                         $this->title->getPrefixedText() . "]], timestamp " . $this->timestamp . "\n" );
 497                                 return false;
 498                         }
 499                 }
 500
 501                 if ( !$pageId ) {
 502                         // This seems to happen if two clients simultaneously try to import the
 503                         // same page
 504                         wfDebug( __METHOD__ . ': got invalid $pageId when importing revision of [[' .
 505                                 $this->title->getPrefixedText() . ']], timestamp ' . $this->timestamp . "\n" );
 506                         return false;
 507                 }
 508
 509                 // Select previous version to make size diffs correct
 510                 // @todo This assumes that multiple revisions of the same page are imported
 511                 // in order from oldest to newest.
 512                 $prevId = $dbw->selectField( 'revision', 'rev_id',
 513                         [
 514                                 'rev_page' => $pageId,
 515                                 'rev_timestamp <= ' . $dbw->addQuotes( $dbw->timestamp( $this->timestamp ) ),
 516                         ],
 517                         __METHOD__,
 518                         [ 'ORDER BY' => [
 519                                         'rev_timestamp DESC',
 520                                         'rev_id DESC', // timestamp is not unique per page
 521                                 ]
 522                         ]
 523                 );
 524
 525                 # @todo FIXME: Use original rev_id optionally (better for backups)
 526                 # Insert the row
 527                 $revision = new Revision( [
 528                         'title' => $this->title,
 529                         'page' => $pageId,
 530                         'content_model' => $this->getModel(),
 531                         'content_format' => $this->getFormat(),
 532                         // XXX: just set 'content' => $this->getContent()?
 533                         'text' => $this->getContent()->serialize( $this->getFormat() ),
 534                         'comment' => $this->getComment(),
 535                         'user' => $userId,
 536                         'user_text' => $userText,
 537                         'timestamp' => $this->timestamp,
 538                         'minor_edit' => $this->minor,
 539                         'parent_id' => $prevId,
 540                         ] );
 541                 $revision->insertOn( $dbw );
 542                 $changed = $page->updateIfNewerOn( $dbw, $revision );
 543
 544                 if ( $changed !== false && !$this->mNoUpdates ) {
 545                         wfDebug( __METHOD__ . ": running updates\n" );
 546                         // countable/oldcountable stuff is handled in WikiImporter::finishImportPage
 547                         $page->doEditUpdates(
 548                                 $revision,
 549                                 $user,
 550                                 [ 'created' => $created, 'oldcountable' => 'no-change' ]
 551                         );
 552                 }
 553
 554                 return true;
 555         }
 556
 557         public function importLogItem() {
 558                 $dbw = wfGetDB( DB_MASTER );
 559
 560                 $user = $this->getUserObj() ?: User::newFromName( $this->getUser() );
 561                 if ( $user ) {
 562                         $userId = intval( $user->getId() );
 563                         $userText = $user->getName();
 564                 } else {
 565                         $userId = 0;
 566                         $userText = $this->getUser();
 567                 }
 568
 569                 # @todo FIXME: This will not record autoblocks
 570                 if ( !$this->getTitle() ) {
 571                         wfDebug( __METHOD__ . ": skipping invalid {$this->type}/{$this->action} log time, timestamp " .
 572                                 $this->timestamp . "\n" );
 573                         return false;
 574                 }
 575                 # Check if it exists already
 576                 // @todo FIXME: Use original log ID (better for backups)
 577                 $prior = $dbw->selectField( 'logging', '1',
 578                         [ 'log_type' => $this->getType(),
 579                                 'log_action' => $this->getAction(),
 580                                 'log_timestamp' => $dbw->timestamp( $this->timestamp ),
 581                                 'log_namespace' => $this->getTitle()->getNamespace(),
 582                                 'log_title' => $this->getTitle()->getDBkey(),
 583                                 'log_comment' => $this->getComment(),
 584                                 # 'log_user_text' => $this->user_text,
 585                                 'log_params' => $this->params ],
 586                         __METHOD__
 587                 );
 588                 // @todo FIXME: This could fail slightly for multiple matches :P
 589                 if ( $prior ) {
 590                         wfDebug( __METHOD__
 591                                 . ": skipping existing item for Log:{$this->type}/{$this->action}, timestamp "
 592                                 . $this->timestamp . "\n" );
 593                         return false;
 594                 }
 595                 $log_id = $dbw->nextSequenceValue( 'logging_log_id_seq' );
 596                 $data = [
 597                         'log_id' => $log_id,
 598                         'log_type' => $this->type,
 599                         'log_action' => $this->action,
 600                         'log_timestamp' => $dbw->timestamp( $this->timestamp ),
 601                         'log_user' => $userId,
 602                         'log_user_text' => $userText,
 603                         'log_namespace' => $this->getTitle()->getNamespace(),
 604                         'log_title' => $this->getTitle()->getDBkey(),
 605                         'log_comment' => $this->getComment(),
 606                         'log_params' => $this->params
 607                 ];
 608                 $dbw->insert( 'logging', $data, __METHOD__ );
 609
 610                 return true;
 611         }
 612
 613         /**
 614          * @return bool
 615          */
 616         public function importUpload() {
 617                 # Construct a file
 618                 $archiveName = $this->getArchiveName();
 619                 if ( $archiveName ) {
 620                         wfDebug( __METHOD__ . "Importing archived file as $archiveName\n" );
 621                         $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
 622                                 RepoGroup::singleton()->getLocalRepo(), $archiveName );
 623                 } else {
 624                         $file = wfLocalFile( $this->getTitle() );
 625                         $file->load( File::READ_LATEST );
 626                         wfDebug( __METHOD__ . 'Importing new file as ' . $file->getName() . "\n" );
 627                         if ( $file->exists() && $file->getTimestamp() > $this->getTimestamp() ) {
 628                                 $archiveName = $file->getTimestamp() . '!' . $file->getName();
 629                                 $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
 630                                         RepoGroup::singleton()->getLocalRepo(), $archiveName );
 631                                 wfDebug( __METHOD__ . "File already exists; importing as $archiveName\n" );
 632                         }
 633                 }
 634                 if ( !$file ) {
 635                         wfDebug( __METHOD__ . ': Bad file for ' . $this->getTitle() . "\n" );
 636                         return false;
 637                 }
 638
 639                 # Get the file source or download if necessary
 640                 $source = $this->getFileSrc();
 641                 $autoDeleteSource = $this->isTempSrc();
 642                 if ( !strlen( $source ) ) {
 643                         $source = $this->downloadSource();
 644                         $autoDeleteSource = true;
 645                 }
 646                 if ( !strlen( $source ) ) {
 647                         wfDebug( __METHOD__ . ": Could not fetch remote file.\n" );
 648                         return false;
 649                 }
 650
 651                 $tmpFile = new TempFSFile( $source );
 652                 if ( $autoDeleteSource ) {
 653                         $tmpFile->autocollect();
 654                 }
 655
 656                 $sha1File = ltrim( sha1_file( $source ), '0' );
 657                 $sha1 = $this->getSha1();
 658                 if ( $sha1 && ( $sha1 !== $sha1File ) ) {
 659                         wfDebug( __METHOD__ . ": Corrupt file $source.\n" );
 660                         return false;
 661                 }
 662
 663                 $user = $this->getUserObj() ?: User::newFromName( $this->getUser() );
 664
 665                 # Do the actual upload
 666                 if ( $archiveName ) {
 667                         $status = $file->uploadOld( $source, $archiveName,
 668                                 $this->getTimestamp(), $this->getComment(), $user );
 669                 } else {
 670                         $flags = 0;
 671                         $status = $file->upload( $source, $this->getComment(), $this->getComment(),
 672                                 $flags, false, $this->getTimestamp(), $user );
 673                 }
 674
 675                 if ( $status->isGood() ) {
 676                         wfDebug( __METHOD__ . ": Successful\n" );
 677                         return true;
 678                 } else {
 679                         wfDebug( __METHOD__ . ': failed: ' . $status->getHTML() . "\n" );
 680                         return false;
 681                 }
 682         }
 683
 684         /**
 685          * @return bool|string
 686          */
 687         public function downloadSource() {
 688                 if ( !$this->config->get( 'EnableUploads' ) ) {
 689                         return false;
 690                 }
 691
 692                 $tempo = tempnam( wfTempDir(), 'download' );
 693                 $f = fopen( $tempo, 'wb' );
 694                 if ( !$f ) {
 695                         wfDebug( "IMPORT: couldn't write to temp file $tempo\n" );
 696                         return false;
 697                 }
 698
 699                 // @todo FIXME!
 700                 $src = $this->getSrc();
 701                 $data = Http::get( $src, [], __METHOD__ );
 702                 if ( !$data ) {
 703                         wfDebug( "IMPORT: couldn't fetch source $src\n" );
 704                         fclose( $f );
 705                         unlink( $tempo );
 706                         return false;
 707                 }
 708
 709                 fwrite( $f, $data );
 710                 fclose( $f );
 711
 712                 return $tempo;
 713         }
 714
 715 }