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 string */
  98         public $archiveName = '';
  99
 100         protected $filename;
 101
 102         /** @var mixed */
 103         protected $src;
 104
 105         /**
 106          * @since 1.18
 107          * @var bool
 108          */
 109         public $isTemp = false;
 110
 111         /**
 112          * @since 1.18
 113          * @deprecated 1.29 use Wikirevision::isTempSrc()
 114          * First written to in 43d5d3b682cc1733ad01a837d11af4a402d57e6a
 115          * Actually introduced in 52cd34acf590e5be946b7885ffdc13a157c1c6cf
 116          */
 117         public $fileIsTemp;
 118
 119         /** @var bool */
 120         private $mNoUpdates = false;
 121
 122         /** @var Config $config */
 123         private $config;
 124
 125         public function __construct( Config $config ) {
 126                 $this->config = $config;
 127         }
 128
 129         /**
 130          * @param Title $title
 131          * @throws MWException
 132          */
 133         public function setTitle( $title ) {
 134                 if ( is_object( $title ) ) {
 135                         $this->title = $title;
 136                 } elseif ( is_null( $title ) ) {
 137                         throw new MWException( "WikiRevision given a null title in import. "
 138                                 . "You may need to adjust \$wgLegalTitleChars." );
 139                 } else {
 140                         throw new MWException( "WikiRevision given non-object title in import." );
 141                 }
 142         }
 143
 144         /**
 145          * @param int $id
 146          */
 147         public function setID( $id ) {
 148                 $this->id = $id;
 149         }
 150
 151         /**
 152          * @param string $ts
 153          */
 154         public function setTimestamp( $ts ) {
 155                 # 2003-08-05T18:30:02Z
 156                 $this->timestamp = wfTimestamp( TS_MW, $ts );
 157         }
 158
 159         /**
 160          * @param string $user
 161          */
 162         public function setUsername( $user ) {
 163                 $this->user_text = $user;
 164         }
 165
 166         /**
 167          * @param User $user
 168          */
 169         public function setUserObj( $user ) {
 170                 $this->userObj = $user;
 171         }
 172
 173         /**
 174          * @param string $ip
 175          */
 176         public function setUserIP( $ip ) {
 177                 $this->user_text = $ip;
 178         }
 179
 180         /**
 181          * @param string $model
 182          */
 183         public function setModel( $model ) {
 184                 $this->model = $model;
 185         }
 186
 187         /**
 188          * @param string $format
 189          */
 190         public function setFormat( $format ) {
 191                 $this->format = $format;
 192         }
 193
 194         /**
 195          * @param string $text
 196          */
 197         public function setText( $text ) {
 198                 $this->text = $text;
 199         }
 200
 201         /**
 202          * @param string $text
 203          */
 204         public function setComment( $text ) {
 205                 $this->comment = $text;
 206         }
 207
 208         /**
 209          * @param bool $minor
 210          */
 211         public function setMinor( $minor ) {
 212                 $this->minor = (bool)$minor;
 213         }
 214
 215         /**
 216          * @param mixed $src
 217          */
 218         public function setSrc( $src ) {
 219                 $this->src = $src;
 220         }
 221
 222         /**
 223          * @param string $src
 224          * @param bool $isTemp
 225          */
 226         public function setFileSrc( $src, $isTemp ) {
 227                 $this->fileSrc = $src;
 228                 $this->fileIsTemp = $isTemp;
 229                 $this->isTemp = $isTemp;
 230         }
 231
 232         /**
 233          * @param string $sha1base36
 234          */
 235         public function setSha1Base36( $sha1base36 ) {
 236                 $this->sha1base36 = $sha1base36;
 237         }
 238
 239         /**
 240          * @param string $filename
 241          */
 242         public function setFilename( $filename ) {
 243                 $this->filename = $filename;
 244         }
 245
 246         /**
 247          * @param string $archiveName
 248          */
 249         public function setArchiveName( $archiveName ) {
 250                 $this->archiveName = $archiveName;
 251         }
 252
 253         /**
 254          * @param int $size
 255          */
 256         public function setSize( $size ) {
 257                 $this->size = intval( $size );
 258         }
 259
 260         /**
 261          * @param string $type
 262          */
 263         public function setType( $type ) {
 264                 $this->type = $type;
 265         }
 266
 267         /**
 268          * @param string $action
 269          */
 270         public function setAction( $action ) {
 271                 $this->action = $action;
 272         }
 273
 274         /**
 275          * @param array $params
 276          */
 277         public function setParams( $params ) {
 278                 $this->params = $params;
 279         }
 280
 281         /**
 282          * @param bool $noupdates
 283          */
 284         public function setNoUpdates( $noupdates ) {
 285                 $this->mNoUpdates = $noupdates;
 286         }
 287
 288         /**
 289          * @return Title
 290          */
 291         public function getTitle() {
 292                 return $this->title;
 293         }
 294
 295         /**
 296          * @return int
 297          */
 298         public function getID() {
 299                 return $this->id;
 300         }
 301
 302         /**
 303          * @return string
 304          */
 305         public function getTimestamp() {
 306                 return $this->timestamp;
 307         }
 308
 309         /**
 310          * @return string
 311          */
 312         public function getUser() {
 313                 return $this->user_text;
 314         }
 315
 316         /**
 317          * @return User
 318          */
 319         public function getUserObj() {
 320                 return $this->userObj;
 321         }
 322
 323         /**
 324          * @return string
 325          */
 326         public function getText() {
 327                 return $this->text;
 328         }
 329
 330         /**
 331          * @return ContentHandler
 332          */
 333         public function getContentHandler() {
 334                 if ( is_null( $this->contentHandler ) ) {
 335                         $this->contentHandler = ContentHandler::getForModelID( $this->getModel() );
 336                 }
 337
 338                 return $this->contentHandler;
 339         }
 340
 341         /**
 342          * @return Content
 343          */
 344         public function getContent() {
 345                 if ( is_null( $this->content ) ) {
 346                         $handler = $this->getContentHandler();
 347                         $this->content = $handler->unserializeContent( $this->text, $this->getFormat() );
 348                 }
 349
 350                 return $this->content;
 351         }
 352
 353         /**
 354          * @return string
 355          */
 356         public function getModel() {
 357                 if ( is_null( $this->model ) ) {
 358                         $this->model = $this->getTitle()->getContentModel();
 359                 }
 360
 361                 return $this->model;
 362         }
 363
 364         /**
 365          * @return string
 366          */
 367         public function getFormat() {
 368                 if ( is_null( $this->format ) ) {
 369                         $this->format = $this->getContentHandler()->getDefaultFormat();
 370                 }
 371
 372                 return $this->format;
 373         }
 374
 375         /**
 376          * @return string
 377          */
 378         public function getComment() {
 379                 return $this->comment;
 380         }
 381
 382         /**
 383          * @return bool
 384          */
 385         public function getMinor() {
 386                 return $this->minor;
 387         }
 388
 389         /**
 390          * @return mixed
 391          */
 392         public function getSrc() {
 393                 return $this->src;
 394         }
 395
 396         /**
 397          * @return bool|string
 398          */
 399         public function getSha1() {
 400                 if ( $this->sha1base36 ) {
 401                         return Wikimedia\base_convert( $this->sha1base36, 36, 16 );
 402                 }
 403                 return false;
 404         }
 405
 406         /**
 407          * @return string
 408          */
 409         public function getFileSrc() {
 410                 return $this->fileSrc;
 411         }
 412
 413         /**
 414          * @return bool
 415          */
 416         public function isTempSrc() {
 417                 return $this->isTemp;
 418         }
 419
 420         /**
 421          * @return mixed
 422          */
 423         public function getFilename() {
 424                 return $this->filename;
 425         }
 426
 427         /**
 428          * @return string
 429          */
 430         public function getArchiveName() {
 431                 return $this->archiveName;
 432         }
 433
 434         /**
 435          * @return mixed
 436          */
 437         public function getSize() {
 438                 return $this->size;
 439         }
 440
 441         /**
 442          * @return string
 443          */
 444         public function getType() {
 445                 return $this->type;
 446         }
 447
 448         /**
 449          * @return string
 450          */
 451         public function getAction() {
 452                 return $this->action;
 453         }
 454
 455         /**
 456          * @return string
 457          */
 458         public function getParams() {
 459                 return $this->params;
 460         }
 461
 462         /**
 463          * @return bool
 464          */
 465         public function importOldRevision() {
 466                 $dbw = wfGetDB( DB_MASTER );
 467
 468                 # Sneak a single revision into place
 469                 $user = $this->getUserObj() ?: User::newFromName( $this->getUser() );
 470                 if ( $user ) {
 471                         $userId = intval( $user->getId() );
 472                         $userText = $user->getName();
 473                 } else {
 474                         $userId = 0;
 475                         $userText = $this->getUser();
 476                         $user = new User;
 477                 }
 478
 479                 // avoid memory leak...?
 480                 Title::clearCaches();
 481
 482                 $page = WikiPage::factory( $this->title );
 483                 $page->loadPageData( 'fromdbmaster' );
 484                 if ( !$page->exists() ) {
 485                         // must create the page...
 486                         $pageId = $page->insertOn( $dbw );
 487                         $created = true;
 488                         $oldcountable = null;
 489                 } else {
 490                         $pageId = $page->getId();
 491                         $created = false;
 492
 493                         $prior = $dbw->selectField( 'revision', '1',
 494                                 [ 'rev_page' => $pageId,
 495                                         'rev_timestamp' => $dbw->timestamp( $this->timestamp ),
 496                                         'rev_user_text' => $userText,
 497                                         'rev_comment' => $this->getComment() ],
 498                                 __METHOD__
 499                         );
 500                         if ( $prior ) {
 501                                 // @todo FIXME: This could fail slightly for multiple matches :P
 502                                 wfDebug( __METHOD__ . ": skipping existing revision for [[" .
 503                                         $this->title->getPrefixedText() . "]], timestamp " . $this->timestamp . "\n" );
 504                                 return false;
 505                         }
 506                 }
 507
 508                 if ( !$pageId ) {
 509                         // This seems to happen if two clients simultaneously try to import the
 510                         // same page
 511                         wfDebug( __METHOD__ . ': got invalid $pageId when importing revision of [[' .
 512                                 $this->title->getPrefixedText() . ']], timestamp ' . $this->timestamp . "\n" );
 513                         return false;
 514                 }
 515
 516                 // Select previous version to make size diffs correct
 517                 // @todo This assumes that multiple revisions of the same page are imported
 518                 // in order from oldest to newest.
 519                 $prevId = $dbw->selectField( 'revision', 'rev_id',
 520                         [
 521                                 'rev_page' => $pageId,
 522                                 'rev_timestamp <= ' . $dbw->addQuotes( $dbw->timestamp( $this->timestamp ) ),
 523                         ],
 524                         __METHOD__,
 525                         [ 'ORDER BY' => [
 526                                         'rev_timestamp DESC',
 527                                         'rev_id DESC', // timestamp is not unique per page
 528                                 ]
 529                         ]
 530                 );
 531
 532                 # @todo FIXME: Use original rev_id optionally (better for backups)
 533                 # Insert the row
 534                 $revision = new Revision( [
 535                         'title' => $this->title,
 536                         'page' => $pageId,
 537                         'content_model' => $this->getModel(),
 538                         'content_format' => $this->getFormat(),
 539                         // XXX: just set 'content' => $this->getContent()?
 540                         'text' => $this->getContent()->serialize( $this->getFormat() ),
 541                         'comment' => $this->getComment(),
 542                         'user' => $userId,
 543                         'user_text' => $userText,
 544                         'timestamp' => $this->timestamp,
 545                         'minor_edit' => $this->minor,
 546                         'parent_id' => $prevId,
 547                         ] );
 548                 $revision->insertOn( $dbw );
 549                 $changed = $page->updateIfNewerOn( $dbw, $revision );
 550
 551                 if ( $changed !== false && !$this->mNoUpdates ) {
 552                         wfDebug( __METHOD__ . ": running updates\n" );
 553                         // countable/oldcountable stuff is handled in WikiImporter::finishImportPage
 554                         $page->doEditUpdates(
 555                                 $revision,
 556                                 $user,
 557                                 [ 'created' => $created, 'oldcountable' => 'no-change' ]
 558                         );
 559                 }
 560
 561                 return true;
 562         }
 563
 564         public function importLogItem() {
 565                 $dbw = wfGetDB( DB_MASTER );
 566
 567                 $user = $this->getUserObj() ?: User::newFromName( $this->getUser() );
 568                 if ( $user ) {
 569                         $userId = intval( $user->getId() );
 570                         $userText = $user->getName();
 571                 } else {
 572                         $userId = 0;
 573                         $userText = $this->getUser();
 574                 }
 575
 576                 # @todo FIXME: This will not record autoblocks
 577                 if ( !$this->getTitle() ) {
 578                         wfDebug( __METHOD__ . ": skipping invalid {$this->type}/{$this->action} log time, timestamp " .
 579                                 $this->timestamp . "\n" );
 580                         return false;
 581                 }
 582                 # Check if it exists already
 583                 // @todo FIXME: Use original log ID (better for backups)
 584                 $prior = $dbw->selectField( 'logging', '1',
 585                         [ 'log_type' => $this->getType(),
 586                                 'log_action' => $this->getAction(),
 587                                 'log_timestamp' => $dbw->timestamp( $this->timestamp ),
 588                                 'log_namespace' => $this->getTitle()->getNamespace(),
 589                                 'log_title' => $this->getTitle()->getDBkey(),
 590                                 'log_comment' => $this->getComment(),
 591                                 # 'log_user_text' => $this->user_text,
 592                                 'log_params' => $this->params ],
 593                         __METHOD__
 594                 );
 595                 // @todo FIXME: This could fail slightly for multiple matches :P
 596                 if ( $prior ) {
 597                         wfDebug( __METHOD__
 598                                 . ": skipping existing item for Log:{$this->type}/{$this->action}, timestamp "
 599                                 . $this->timestamp . "\n" );
 600                         return false;
 601                 }
 602                 $log_id = $dbw->nextSequenceValue( 'logging_log_id_seq' );
 603                 $data = [
 604                         'log_id' => $log_id,
 605                         'log_type' => $this->type,
 606                         'log_action' => $this->action,
 607                         'log_timestamp' => $dbw->timestamp( $this->timestamp ),
 608                         'log_user' => $userId,
 609                         'log_user_text' => $userText,
 610                         'log_namespace' => $this->getTitle()->getNamespace(),
 611                         'log_title' => $this->getTitle()->getDBkey(),
 612                         'log_comment' => $this->getComment(),
 613                         'log_params' => $this->params
 614                 ];
 615                 $dbw->insert( 'logging', $data, __METHOD__ );
 616
 617                 return true;
 618         }
 619
 620         /**
 621          * @return bool
 622          */
 623         public function importUpload() {
 624                 # Construct a file
 625                 $archiveName = $this->getArchiveName();
 626                 if ( $archiveName ) {
 627                         wfDebug( __METHOD__ . "Importing archived file as $archiveName\n" );
 628                         $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
 629                                 RepoGroup::singleton()->getLocalRepo(), $archiveName );
 630                 } else {
 631                         $file = wfLocalFile( $this->getTitle() );
 632                         $file->load( File::READ_LATEST );
 633                         wfDebug( __METHOD__ . 'Importing new file as ' . $file->getName() . "\n" );
 634                         if ( $file->exists() && $file->getTimestamp() > $this->getTimestamp() ) {
 635                                 $archiveName = $file->getTimestamp() . '!' . $file->getName();
 636                                 $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
 637                                         RepoGroup::singleton()->getLocalRepo(), $archiveName );
 638                                 wfDebug( __METHOD__ . "File already exists; importing as $archiveName\n" );
 639                         }
 640                 }
 641                 if ( !$file ) {
 642                         wfDebug( __METHOD__ . ': Bad file for ' . $this->getTitle() . "\n" );
 643                         return false;
 644                 }
 645
 646                 # Get the file source or download if necessary
 647                 $source = $this->getFileSrc();
 648                 $autoDeleteSource = $this->isTempSrc();
 649                 if ( !strlen( $source ) ) {
 650                         $source = $this->downloadSource();
 651                         $autoDeleteSource = true;
 652                 }
 653                 if ( !strlen( $source ) ) {
 654                         wfDebug( __METHOD__ . ": Could not fetch remote file.\n" );
 655                         return false;
 656                 }
 657
 658                 $tmpFile = new TempFSFile( $source );
 659                 if ( $autoDeleteSource ) {
 660                         $tmpFile->autocollect();
 661                 }
 662
 663                 $sha1File = ltrim( sha1_file( $source ), '0' );
 664                 $sha1 = $this->getSha1();
 665                 if ( $sha1 && ( $sha1 !== $sha1File ) ) {
 666                         wfDebug( __METHOD__ . ": Corrupt file $source.\n" );
 667                         return false;
 668                 }
 669
 670                 $user = $this->getUserObj() ?: User::newFromName( $this->getUser() );
 671
 672                 # Do the actual upload
 673                 if ( $archiveName ) {
 674                         $status = $file->uploadOld( $source, $archiveName,
 675                                 $this->getTimestamp(), $this->getComment(), $user );
 676                 } else {
 677                         $flags = 0;
 678                         $status = $file->upload( $source, $this->getComment(), $this->getComment(),
 679                                 $flags, false, $this->getTimestamp(), $user );
 680                 }
 681
 682                 if ( $status->isGood() ) {
 683                         wfDebug( __METHOD__ . ": Successful\n" );
 684                         return true;
 685                 } else {
 686                         wfDebug( __METHOD__ . ': failed: ' . $status->getHTML() . "\n" );
 687                         return false;
 688                 }
 689         }
 690
 691         /**
 692          * @return bool|string
 693          */
 694         public function downloadSource() {
 695                 if ( !$this->config->get( 'EnableUploads' ) ) {
 696                         return false;
 697                 }
 698
 699                 $tempo = tempnam( wfTempDir(), 'download' );
 700                 $f = fopen( $tempo, 'wb' );
 701                 if ( !$f ) {
 702                         wfDebug( "IMPORT: couldn't write to temp file $tempo\n" );
 703                         return false;
 704                 }
 705
 706                 // @todo FIXME!
 707                 $src = $this->getSrc();
 708                 $data = Http::get( $src, [], __METHOD__ );
 709                 if ( !$data ) {
 710                         wfDebug( "IMPORT: couldn't fetch source $src\n" );
 711                         fclose( $f );
 712                         unlink( $tempo );
 713                         return false;
 714                 }
 715
 716                 fwrite( $f, $data );
 717                 fclose( $f );
 718
 719                 return $tempo;
 720         }
 721
 722 }