includes/Revision.php

   1 <?php
   2 /**
   3  * Representation of a page version.
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  */
  22
  23 use Wikimedia\Rdbms\Database;
  24 use Wikimedia\Rdbms\IDatabase;
  25 use MediaWiki\Linker\LinkTarget;
  26 use MediaWiki\MediaWikiServices;
  27 use Wikimedia\Rdbms\ResultWrapper;
  28 use Wikimedia\Rdbms\FakeResultWrapper;
  29
  30 /**
  31  * @todo document
  32  */
  33 class Revision implements IDBAccessObject {
  34         /** @var int|null */
  35         protected $mId;
  36         /** @var int|null */
  37         protected $mPage;
  38         /** @var string */
  39         protected $mUserText;
  40         /** @var string */
  41         protected $mOrigUserText;
  42         /** @var int */
  43         protected $mUser;
  44         /** @var bool */
  45         protected $mMinorEdit;
  46         /** @var string */
  47         protected $mTimestamp;
  48         /** @var int */
  49         protected $mDeleted;
  50         /** @var int */
  51         protected $mSize;
  52         /** @var string */
  53         protected $mSha1;
  54         /** @var int */
  55         protected $mParentId;
  56         /** @var string */
  57         protected $mComment;
  58         /** @var string */
  59         protected $mText;
  60         /** @var int */
  61         protected $mTextId;
  62         /** @var int */
  63         protected $mUnpatrolled;
  64
  65         /** @var stdClass|null */
  66         protected $mTextRow;
  67
  68         /**  @var null|Title */
  69         protected $mTitle;
  70         /** @var bool */
  71         protected $mCurrent;
  72         /** @var string */
  73         protected $mContentModel;
  74         /** @var string */
  75         protected $mContentFormat;
  76
  77         /** @var Content|null|bool */
  78         protected $mContent;
  79         /** @var null|ContentHandler */
  80         protected $mContentHandler;
  81
  82         /** @var int */
  83         protected $mQueryFlags = 0;
  84         /** @var bool Used for cached values to reload user text and rev_deleted */
  85         protected $mRefreshMutableFields = false;
  86         /** @var string Wiki ID; false means the current wiki */
  87         protected $mWiki = false;
  88
  89         // Revision deletion constants
  90         const DELETED_TEXT = 1;
  91         const DELETED_COMMENT = 2;
  92         const DELETED_USER = 4;
  93         const DELETED_RESTRICTED = 8;
  94         const SUPPRESSED_USER = 12; // convenience
  95         const SUPPRESSED_ALL = 15; // convenience
  96
  97         // Audience options for accessors
  98         const FOR_PUBLIC = 1;
  99         const FOR_THIS_USER = 2;
 100         const RAW = 3;
 101
 102         const TEXT_CACHE_GROUP = 'revisiontext:10'; // process cache name and max key count
 103
 104         /**
 105          * Load a page revision from a given revision ID number.
 106          * Returns null if no such revision can be found.
 107          *
 108          * $flags include:
 109          *      Revision::READ_LATEST  : Select the data from the master
 110          *      Revision::READ_LOCKING : Select & lock the data from the master
 111          *
 112          * @param int $id
 113          * @param int $flags (optional)
 114          * @return Revision|null
 115          */
 116         public static function newFromId( $id, $flags = 0 ) {
 117                 return self::newFromConds( [ 'rev_id' => intval( $id ) ], $flags );
 118         }
 119
 120         /**
 121          * Load either the current, or a specified, revision
 122          * that's attached to a given link target. If not attached
 123          * to that link target, will return null.
 124          *
 125          * $flags include:
 126          *      Revision::READ_LATEST  : Select the data from the master
 127          *      Revision::READ_LOCKING : Select & lock the data from the master
 128          *
 129          * @param LinkTarget $linkTarget
 130          * @param int $id (optional)
 131          * @param int $flags Bitfield (optional)
 132          * @return Revision|null
 133          */
 134         public static function newFromTitle( LinkTarget $linkTarget, $id = 0, $flags = 0 ) {
 135                 $conds = [
 136                         'page_namespace' => $linkTarget->getNamespace(),
 137                         'page_title' => $linkTarget->getDBkey()
 138                 ];
 139                 if ( $id ) {
 140                         // Use the specified ID
 141                         $conds['rev_id'] = $id;
 142                         return self::newFromConds( $conds, $flags );
 143                 } else {
 144                         // Use a join to get the latest revision
 145                         $conds[] = 'rev_id=page_latest';
 146                         $db = wfGetDB( ( $flags & self::READ_LATEST ) ? DB_MASTER : DB_REPLICA );
 147                         return self::loadFromConds( $db, $conds, $flags );
 148                 }
 149         }
 150
 151         /**
 152          * Load either the current, or a specified, revision
 153          * that's attached to a given page ID.
 154          * Returns null if no such revision can be found.
 155          *
 156          * $flags include:
 157          *      Revision::READ_LATEST  : Select the data from the master (since 1.20)
 158          *      Revision::READ_LOCKING : Select & lock the data from the master
 159          *
 160          * @param int $pageId
 161          * @param int $revId (optional)
 162          * @param int $flags Bitfield (optional)
 163          * @return Revision|null
 164          */
 165         public static function newFromPageId( $pageId, $revId = 0, $flags = 0 ) {
 166                 $conds = [ 'page_id' => $pageId ];
 167                 if ( $revId ) {
 168                         $conds['rev_id'] = $revId;
 169                         return self::newFromConds( $conds, $flags );
 170                 } else {
 171                         // Use a join to get the latest revision
 172                         $conds[] = 'rev_id = page_latest';
 173                         $db = wfGetDB( ( $flags & self::READ_LATEST ) ? DB_MASTER : DB_REPLICA );
 174                         return self::loadFromConds( $db, $conds, $flags );
 175                 }
 176         }
 177
 178         /**
 179          * Make a fake revision object from an archive table row. This is queried
 180          * for permissions or even inserted (as in Special:Undelete)
 181          * @todo FIXME: Should be a subclass for RevisionDelete. [TS]
 182          *
 183          * @param object $row
 184          * @param array $overrides
 185          *
 186          * @throws MWException
 187          * @return Revision
 188          */
 189         public static function newFromArchiveRow( $row, $overrides = [] ) {
 190                 global $wgContentHandlerUseDB;
 191
 192                 $attribs = $overrides + [
 193                         'page'       => isset( $row->ar_page_id ) ? $row->ar_page_id : null,
 194                         'id'         => isset( $row->ar_rev_id ) ? $row->ar_rev_id : null,
 195                         'comment'    => CommentStore::newKey( 'ar_comment' )
 196                                 // Legacy because $row probably came from self::selectArchiveFields()
 197                                 ->getCommentLegacy( wfGetDB( DB_REPLICA ), $row, true )->text,
 198                         'user'       => $row->ar_user,
 199                         'user_text'  => $row->ar_user_text,
 200                         'timestamp'  => $row->ar_timestamp,
 201                         'minor_edit' => $row->ar_minor_edit,
 202                         'text_id'    => isset( $row->ar_text_id ) ? $row->ar_text_id : null,
 203                         'deleted'    => $row->ar_deleted,
 204                         'len'        => $row->ar_len,
 205                         'sha1'       => isset( $row->ar_sha1 ) ? $row->ar_sha1 : null,
 206                         'content_model'   => isset( $row->ar_content_model ) ? $row->ar_content_model : null,
 207                         'content_format'  => isset( $row->ar_content_format ) ? $row->ar_content_format : null,
 208                 ];
 209
 210                 if ( !$wgContentHandlerUseDB ) {
 211                         unset( $attribs['content_model'] );
 212                         unset( $attribs['content_format'] );
 213                 }
 214
 215                 if ( !isset( $attribs['title'] )
 216                         && isset( $row->ar_namespace )
 217                         && isset( $row->ar_title )
 218                 ) {
 219                         $attribs['title'] = Title::makeTitle( $row->ar_namespace, $row->ar_title );
 220                 }
 221
 222                 if ( isset( $row->ar_text ) && !$row->ar_text_id ) {
 223                         // Pre-1.5 ar_text row
 224                         $attribs['text'] = self::getRevisionText( $row, 'ar_' );
 225                         if ( $attribs['text'] === false ) {
 226                                 throw new MWException( 'Unable to load text from archive row (possibly T24624)' );
 227                         }
 228                 }
 229                 return new self( $attribs );
 230         }
 231
 232         /**
 233          * @since 1.19
 234          *
 235          * @param object $row
 236          * @return Revision
 237          */
 238         public static function newFromRow( $row ) {
 239                 return new self( $row );
 240         }
 241
 242         /**
 243          * Load a page revision from a given revision ID number.
 244          * Returns null if no such revision can be found.
 245          *
 246          * @param IDatabase $db
 247          * @param int $id
 248          * @return Revision|null
 249          */
 250         public static function loadFromId( $db, $id ) {
 251                 return self::loadFromConds( $db, [ 'rev_id' => intval( $id ) ] );
 252         }
 253
 254         /**
 255          * Load either the current, or a specified, revision
 256          * that's attached to a given page. If not attached
 257          * to that page, will return null.
 258          *
 259          * @param IDatabase $db
 260          * @param int $pageid
 261          * @param int $id
 262          * @return Revision|null
 263          */
 264         public static function loadFromPageId( $db, $pageid, $id = 0 ) {
 265                 $conds = [ 'rev_page' => intval( $pageid ), 'page_id' => intval( $pageid ) ];
 266                 if ( $id ) {
 267                         $conds['rev_id'] = intval( $id );
 268                 } else {
 269                         $conds[] = 'rev_id=page_latest';
 270                 }
 271                 return self::loadFromConds( $db, $conds );
 272         }
 273
 274         /**
 275          * Load either the current, or a specified, revision
 276          * that's attached to a given page. If not attached
 277          * to that page, will return null.
 278          *
 279          * @param IDatabase $db
 280          * @param Title $title
 281          * @param int $id
 282          * @return Revision|null
 283          */
 284         public static function loadFromTitle( $db, $title, $id = 0 ) {
 285                 if ( $id ) {
 286                         $matchId = intval( $id );
 287                 } else {
 288                         $matchId = 'page_latest';
 289                 }
 290                 return self::loadFromConds( $db,
 291                         [
 292                                 "rev_id=$matchId",
 293                                 'page_namespace' => $title->getNamespace(),
 294                                 'page_title' => $title->getDBkey()
 295                         ]
 296                 );
 297         }
 298
 299         /**
 300          * Load the revision for the given title with the given timestamp.
 301          * WARNING: Timestamps may in some circumstances not be unique,
 302          * so this isn't the best key to use.
 303          *
 304          * @param IDatabase $db
 305          * @param Title $title
 306          * @param string $timestamp
 307          * @return Revision|null
 308          */
 309         public static function loadFromTimestamp( $db, $title, $timestamp ) {
 310                 return self::loadFromConds( $db,
 311                         [
 312                                 'rev_timestamp' => $db->timestamp( $timestamp ),
 313                                 'page_namespace' => $title->getNamespace(),
 314                                 'page_title' => $title->getDBkey()
 315                         ]
 316                 );
 317         }
 318
 319         /**
 320          * Given a set of conditions, fetch a revision
 321          *
 322          * This method is used then a revision ID is qualified and
 323          * will incorporate some basic replica DB/master fallback logic
 324          *
 325          * @param array $conditions
 326          * @param int $flags (optional)
 327          * @return Revision|null
 328          */
 329         private static function newFromConds( $conditions, $flags = 0 ) {
 330                 $db = wfGetDB( ( $flags & self::READ_LATEST ) ? DB_MASTER : DB_REPLICA );
 331
 332                 $rev = self::loadFromConds( $db, $conditions, $flags );
 333                 // Make sure new pending/committed revision are visibile later on
 334                 // within web requests to certain avoid bugs like T93866 and T94407.
 335                 if ( !$rev
 336                         && !( $flags & self::READ_LATEST )
 337                         && wfGetLB()->getServerCount() > 1
 338                         && wfGetLB()->hasOrMadeRecentMasterChanges()
 339                 ) {
 340                         $flags = self::READ_LATEST;
 341                         $db = wfGetDB( DB_MASTER );
 342                         $rev = self::loadFromConds( $db, $conditions, $flags );
 343                 }
 344
 345                 if ( $rev ) {
 346                         $rev->mQueryFlags = $flags;
 347                 }
 348
 349                 return $rev;
 350         }
 351
 352         /**
 353          * Given a set of conditions, fetch a revision from
 354          * the given database connection.
 355          *
 356          * @param IDatabase $db
 357          * @param array $conditions
 358          * @param int $flags (optional)
 359          * @return Revision|null
 360          */
 361         private static function loadFromConds( $db, $conditions, $flags = 0 ) {
 362                 $row = self::fetchFromConds( $db, $conditions, $flags );
 363                 if ( $row ) {
 364                         $rev = new Revision( $row );
 365                         $rev->mWiki = $db->getDomainID();
 366
 367                         return $rev;
 368                 }
 369
 370                 return null;
 371         }
 372
 373         /**
 374          * Return a wrapper for a series of database rows to
 375          * fetch all of a given page's revisions in turn.
 376          * Each row can be fed to the constructor to get objects.
 377          *
 378          * @param LinkTarget $title
 379          * @return ResultWrapper
 380          * @deprecated Since 1.28
 381          */
 382         public static function fetchRevision( LinkTarget $title ) {
 383                 $row = self::fetchFromConds(
 384                         wfGetDB( DB_REPLICA ),
 385                         [
 386                                 'rev_id=page_latest',
 387                                 'page_namespace' => $title->getNamespace(),
 388                                 'page_title' => $title->getDBkey()
 389                         ]
 390                 );
 391
 392                 return new FakeResultWrapper( $row ? [ $row ] : [] );
 393         }
 394
 395         /**
 396          * Given a set of conditions, return a ResultWrapper
 397          * which will return matching database rows with the
 398          * fields necessary to build Revision objects.
 399          *
 400          * @param IDatabase $db
 401          * @param array $conditions
 402          * @param int $flags (optional)
 403          * @return stdClass
 404          */
 405         private static function fetchFromConds( $db, $conditions, $flags = 0 ) {
 406                 $fields = array_merge(
 407                         self::selectFields(),
 408                         self::selectPageFields(),
 409                         self::selectUserFields()
 410                 );
 411                 $options = [];
 412                 if ( ( $flags & self::READ_LOCKING ) == self::READ_LOCKING ) {
 413                         $options[] = 'FOR UPDATE';
 414                 }
 415                 return $db->selectRow(
 416                         [ 'revision', 'page', 'user' ],
 417                         $fields,
 418                         $conditions,
 419                         __METHOD__,
 420                         $options,
 421                         [ 'page' => self::pageJoinCond(), 'user' => self::userJoinCond() ]
 422                 );
 423         }
 424
 425         /**
 426          * Return the value of a select() JOIN conds array for the user table.
 427          * This will get user table rows for logged-in users.
 428          * @since 1.19
 429          * @return array
 430          */
 431         public static function userJoinCond() {
 432                 return [ 'LEFT JOIN', [ 'rev_user != 0', 'user_id = rev_user' ] ];
 433         }
 434
 435         /**
 436          * Return the value of a select() page conds array for the page table.
 437          * This will assure that the revision(s) are not orphaned from live pages.
 438          * @since 1.19
 439          * @return array
 440          */
 441         public static function pageJoinCond() {
 442                 return [ 'INNER JOIN', [ 'page_id = rev_page' ] ];
 443         }
 444
 445         /**
 446          * Return the list of revision fields that should be selected to create
 447          * a new revision.
 448          * @todo Deprecate this in favor of a method that returns tables and joins
 449          *  as well, and use CommentStore::getJoin().
 450          * @return array
 451          */
 452         public static function selectFields() {
 453                 global $wgContentHandlerUseDB;
 454
 455                 $fields = [
 456                         'rev_id',
 457                         'rev_page',
 458                         'rev_text_id',
 459                         'rev_timestamp',
 460                         'rev_user_text',
 461                         'rev_user',
 462                         'rev_minor_edit',
 463                         'rev_deleted',
 464                         'rev_len',
 465                         'rev_parent_id',
 466                         'rev_sha1',
 467                 ];
 468
 469                 $fields += CommentStore::newKey( 'rev_comment' )->getFields();
 470
 471                 if ( $wgContentHandlerUseDB ) {
 472                         $fields[] = 'rev_content_format';
 473                         $fields[] = 'rev_content_model';
 474                 }
 475
 476                 return $fields;
 477         }
 478
 479         /**
 480          * Return the list of revision fields that should be selected to create
 481          * a new revision from an archive row.
 482          * @todo Deprecate this in favor of a method that returns tables and joins
 483          *  as well, and use CommentStore::getJoin().
 484          * @return array
 485          */
 486         public static function selectArchiveFields() {
 487                 global $wgContentHandlerUseDB;
 488                 $fields = [
 489                         'ar_id',
 490                         'ar_page_id',
 491                         'ar_rev_id',
 492                         'ar_text',
 493                         'ar_text_id',
 494                         'ar_timestamp',
 495                         'ar_user_text',
 496                         'ar_user',
 497                         'ar_minor_edit',
 498                         'ar_deleted',
 499                         'ar_len',
 500                         'ar_parent_id',
 501                         'ar_sha1',
 502                 ];
 503
 504                 $fields += CommentStore::newKey( 'ar_comment' )->getFields();
 505
 506                 if ( $wgContentHandlerUseDB ) {
 507                         $fields[] = 'ar_content_format';
 508                         $fields[] = 'ar_content_model';
 509                 }
 510                 return $fields;
 511         }
 512
 513         /**
 514          * Return the list of text fields that should be selected to read the
 515          * revision text
 516          * @return array
 517          */
 518         public static function selectTextFields() {
 519                 return [
 520                         'old_text',
 521                         'old_flags'
 522                 ];
 523         }
 524
 525         /**
 526          * Return the list of page fields that should be selected from page table
 527          * @return array
 528          */
 529         public static function selectPageFields() {
 530                 return [
 531                         'page_namespace',
 532                         'page_title',
 533                         'page_id',
 534                         'page_latest',
 535                         'page_is_redirect',
 536                         'page_len',
 537                 ];
 538         }
 539
 540         /**
 541          * Return the list of user fields that should be selected from user table
 542          * @return array
 543          */
 544         public static function selectUserFields() {
 545                 return [ 'user_name' ];
 546         }
 547
 548         /**
 549          * Do a batched query to get the parent revision lengths
 550          * @param IDatabase $db
 551          * @param array $revIds
 552          * @return array
 553          */
 554         public static function getParentLengths( $db, array $revIds ) {
 555                 $revLens = [];
 556                 if ( !$revIds ) {
 557                         return $revLens; // empty
 558                 }
 559                 $res = $db->select( 'revision',
 560                         [ 'rev_id', 'rev_len' ],
 561                         [ 'rev_id' => $revIds ],
 562                         __METHOD__ );
 563                 foreach ( $res as $row ) {
 564                         $revLens[$row->rev_id] = $row->rev_len;
 565                 }
 566                 return $revLens;
 567         }
 568
 569         /**
 570          * @param object|array $row Either a database row or an array
 571          * @throws MWException
 572          * @access private
 573          */
 574         public function __construct( $row ) {
 575                 if ( is_object( $row ) ) {
 576                         $this->constructFromDbRowObject( $row );
 577                 } elseif ( is_array( $row ) ) {
 578                         $this->constructFromRowArray( $row );
 579                 } else {
 580                         throw new MWException( 'Revision constructor passed invalid row format.' );
 581                 }
 582                 $this->mUnpatrolled = null;
 583         }
 584
 585         /**
 586          * @param object $row
 587          */
 588         private function constructFromDbRowObject( $row ) {
 589                 $this->mId = intval( $row->rev_id );
 590                 $this->mPage = intval( $row->rev_page );
 591                 $this->mTextId = intval( $row->rev_text_id );
 592                 $this->mComment = CommentStore::newKey( 'rev_comment' )
 593                         // Legacy because $row probably came from self::selectFields()
 594                         ->getCommentLegacy( wfGetDB( DB_REPLICA ), $row, true )->text;
 595                 $this->mUser = intval( $row->rev_user );
 596                 $this->mMinorEdit = intval( $row->rev_minor_edit );
 597                 $this->mTimestamp = $row->rev_timestamp;
 598                 $this->mDeleted = intval( $row->rev_deleted );
 599
 600                 if ( !isset( $row->rev_parent_id ) ) {
 601                         $this->mParentId = null;
 602                 } else {
 603                         $this->mParentId = intval( $row->rev_parent_id );
 604                 }
 605
 606                 if ( !isset( $row->rev_len ) ) {
 607                         $this->mSize = null;
 608                 } else {
 609                         $this->mSize = intval( $row->rev_len );
 610                 }
 611
 612                 if ( !isset( $row->rev_sha1 ) ) {
 613                         $this->mSha1 = null;
 614                 } else {
 615                         $this->mSha1 = $row->rev_sha1;
 616                 }
 617
 618                 if ( isset( $row->page_latest ) ) {
 619                         $this->mCurrent = ( $row->rev_id == $row->page_latest );
 620                         $this->mTitle = Title::newFromRow( $row );
 621                 } else {
 622                         $this->mCurrent = false;
 623                         $this->mTitle = null;
 624                 }
 625
 626                 if ( !isset( $row->rev_content_model ) ) {
 627                         $this->mContentModel = null; # determine on demand if needed
 628                 } else {
 629                         $this->mContentModel = strval( $row->rev_content_model );
 630                 }
 631
 632                 if ( !isset( $row->rev_content_format ) ) {
 633                         $this->mContentFormat = null; # determine on demand if needed
 634                 } else {
 635                         $this->mContentFormat = strval( $row->rev_content_format );
 636                 }
 637
 638                 // Lazy extraction...
 639                 $this->mText = null;
 640                 if ( isset( $row->old_text ) ) {
 641                         $this->mTextRow = $row;
 642                 } else {
 643                         // 'text' table row entry will be lazy-loaded
 644                         $this->mTextRow = null;
 645                 }
 646
 647                 // Use user_name for users and rev_user_text for IPs...
 648                 $this->mUserText = null; // lazy load if left null
 649                 if ( $this->mUser == 0 ) {
 650                         $this->mUserText = $row->rev_user_text; // IP user
 651                 } elseif ( isset( $row->user_name ) ) {
 652                         $this->mUserText = $row->user_name; // logged-in user
 653                 }
 654                 $this->mOrigUserText = $row->rev_user_text;
 655         }
 656
 657         /**
 658          * @param array $row
 659          *
 660          * @throws MWException
 661          */
 662         private function constructFromRowArray( array $row ) {
 663                 // Build a new revision to be saved...
 664                 global $wgUser; // ugh
 665
 666                 # if we have a content object, use it to set the model and type
 667                 if ( !empty( $row['content'] ) ) {
 668                         if ( !( $row['content'] instanceof Content ) ) {
 669                                 throw new MWException( '`content` field must contain a Content object.' );
 670                         }
 671
 672                         // @todo when is that set? test with external store setup! check out insertOn() [dk]
 673                         if ( !empty( $row['text_id'] ) ) {
 674                                 throw new MWException( "Text already stored in external store (id {$row['text_id']}), " .
 675                                         "can't serialize content object" );
 676                         }
 677
 678                         $row['content_model'] = $row['content']->getModel();
 679                         # note: mContentFormat is initializes later accordingly
 680                         # note: content is serialized later in this method!
 681                         # also set text to null?
 682                 }
 683
 684                 $this->mId = isset( $row['id'] ) ? intval( $row['id'] ) : null;
 685                 $this->mPage = isset( $row['page'] ) ? intval( $row['page'] ) : null;
 686                 $this->mTextId = isset( $row['text_id'] ) ? intval( $row['text_id'] ) : null;
 687                 $this->mUserText = isset( $row['user_text'] )
 688                         ? strval( $row['user_text'] ) : $wgUser->getName();
 689                 $this->mUser = isset( $row['user'] ) ? intval( $row['user'] ) : $wgUser->getId();
 690                 $this->mMinorEdit = isset( $row['minor_edit'] ) ? intval( $row['minor_edit'] ) : 0;
 691                 $this->mTimestamp = isset( $row['timestamp'] )
 692                         ? strval( $row['timestamp'] ) : wfTimestampNow();
 693                 $this->mDeleted = isset( $row['deleted'] ) ? intval( $row['deleted'] ) : 0;
 694                 $this->mSize = isset( $row['len'] ) ? intval( $row['len'] ) : null;
 695                 $this->mParentId = isset( $row['parent_id'] ) ? intval( $row['parent_id'] ) : null;
 696                 $this->mSha1 = isset( $row['sha1'] ) ? strval( $row['sha1'] ) : null;
 697
 698                 $this->mContentModel = isset( $row['content_model'] )
 699                         ? strval( $row['content_model'] ) : null;
 700                 $this->mContentFormat = isset( $row['content_format'] )
 701                         ? strval( $row['content_format'] ) : null;
 702
 703                 // Enforce spacing trimming on supplied text
 704                 $this->mComment = isset( $row['comment'] ) ? trim( strval( $row['comment'] ) ) : null;
 705                 $this->mText = isset( $row['text'] ) ? rtrim( strval( $row['text'] ) ) : null;
 706                 $this->mTextRow = null;
 707
 708                 $this->mTitle = isset( $row['title'] ) ? $row['title'] : null;
 709
 710                 // if we have a Content object, override mText and mContentModel
 711                 if ( !empty( $row['content'] ) ) {
 712                         $handler = $this->getContentHandler();
 713                         $this->mContent = $row['content'];
 714
 715                         $this->mContentModel = $this->mContent->getModel();
 716                         $this->mContentHandler = null;
 717
 718                         $this->mText = $handler->serializeContent( $row['content'], $this->getContentFormat() );
 719                 } elseif ( $this->mText !== null ) {
 720                         $handler = $this->getContentHandler();
 721                         $this->mContent = $handler->unserializeContent( $this->mText );
 722                 }
 723
 724                 // If we have a Title object, make sure it is consistent with mPage.
 725                 if ( $this->mTitle && $this->mTitle->exists() ) {
 726                         if ( $this->mPage === null ) {
 727                                 // if the page ID wasn't known, set it now
 728                                 $this->mPage = $this->mTitle->getArticleID();
 729                         } elseif ( $this->mTitle->getArticleID() !== $this->mPage ) {
 730                                 // Got different page IDs. This may be legit (e.g. during undeletion),
 731                                 // but it seems worth mentioning it in the log.
 732                                 wfDebug( "Page ID " . $this->mPage . " mismatches the ID " .
 733                                         $this->mTitle->getArticleID() . " provided by the Title object." );
 734                         }
 735                 }
 736
 737                 $this->mCurrent = false;
 738
 739                 // If we still have no length, see it we have the text to figure it out
 740                 if ( !$this->mSize && $this->mContent !== null ) {
 741                         $this->mSize = $this->mContent->getSize();
 742                 }
 743
 744                 // Same for sha1
 745                 if ( $this->mSha1 === null ) {
 746                         $this->mSha1 = $this->mText === null ? null : self::base36Sha1( $this->mText );
 747                 }
 748
 749                 // force lazy init
 750                 $this->getContentModel();
 751                 $this->getContentFormat();
 752         }
 753
 754         /**
 755          * Get revision ID
 756          *
 757          * @return int|null
 758          */
 759         public function getId() {
 760                 return $this->mId;
 761         }
 762
 763         /**
 764          * Set the revision ID
 765          *
 766          * This should only be used for proposed revisions that turn out to be null edits
 767          *
 768          * @since 1.19
 769          * @param int $id
 770          */
 771         public function setId( $id ) {
 772                 $this->mId = (int)$id;
 773         }
 774
 775         /**
 776          * Set the user ID/name
 777          *
 778          * This should only be used for proposed revisions that turn out to be null edits
 779          *
 780          * @since 1.28
 781          * @param int $id User ID
 782          * @param string $name User name
 783          */
 784         public function setUserIdAndName( $id, $name ) {
 785                 $this->mUser = (int)$id;
 786                 $this->mUserText = $name;
 787                 $this->mOrigUserText = $name;
 788         }
 789
 790         /**
 791          * Get text row ID
 792          *
 793          * @return int|null
 794          */
 795         public function getTextId() {
 796                 return $this->mTextId;
 797         }
 798
 799         /**
 800          * Get parent revision ID (the original previous page revision)
 801          *
 802          * @return int|null
 803          */
 804         public function getParentId() {
 805                 return $this->mParentId;
 806         }
 807
 808         /**
 809          * Returns the length of the text in this revision, or null if unknown.
 810          *
 811          * @return int|null
 812          */
 813         public function getSize() {
 814                 return $this->mSize;
 815         }
 816
 817         /**
 818          * Returns the base36 sha1 of the text in this revision, or null if unknown.
 819          *
 820          * @return string|null
 821          */
 822         public function getSha1() {
 823                 return $this->mSha1;
 824         }
 825
 826         /**
 827          * Returns the title of the page associated with this entry or null.
 828          *
 829          * Will do a query, when title is not set and id is given.
 830          *
 831          * @return Title|null
 832          */
 833         public function getTitle() {
 834                 if ( $this->mTitle !== null ) {
 835                         return $this->mTitle;
 836                 }
 837                 // rev_id is defined as NOT NULL, but this revision may not yet have been inserted.
 838                 if ( $this->mId !== null ) {
 839                         $dbr = wfGetLB( $this->mWiki )->getConnectionRef( DB_REPLICA, [], $this->mWiki );
 840                         $row = $dbr->selectRow(
 841                                 [ 'page', 'revision' ],
 842                                 self::selectPageFields(),
 843                                 [ 'page_id=rev_page', 'rev_id' => $this->mId ],
 844                                 __METHOD__
 845                         );
 846                         if ( $row ) {
 847                                 // @TODO: better foreign title handling
 848                                 $this->mTitle = Title::newFromRow( $row );
 849                         }
 850                 }
 851
 852                 if ( $this->mWiki === false || $this->mWiki === wfWikiID() ) {
 853                         // Loading by ID is best, though not possible for foreign titles
 854                         if ( !$this->mTitle && $this->mPage !== null && $this->mPage > 0 ) {
 855                                 $this->mTitle = Title::newFromID( $this->mPage );
 856                         }
 857                 }
 858
 859                 return $this->mTitle;
 860         }
 861
 862         /**
 863          * Set the title of the revision
 864          *
 865          * @param Title $title
 866          */
 867         public function setTitle( $title ) {
 868                 $this->mTitle = $title;
 869         }
 870
 871         /**
 872          * Get the page ID
 873          *
 874          * @return int|null
 875          */
 876         public function getPage() {
 877                 return $this->mPage;
 878         }
 879
 880         /**
 881          * Fetch revision's user id if it's available to the specified audience.
 882          * If the specified audience does not have access to it, zero will be
 883          * returned.
 884          *
 885          * @param int $audience One of:
 886          *   Revision::FOR_PUBLIC       to be displayed to all users
 887          *   Revision::FOR_THIS_USER    to be displayed to the given user
 888          *   Revision::RAW              get the ID regardless of permissions
 889          * @param User|null $user User object to check for, only if FOR_THIS_USER is passed
 890          *   to the $audience parameter
 891          * @return int
 892          */
 893         public function getUser( $audience = self::FOR_PUBLIC, User $user = null ) {
 894                 if ( $audience == self::FOR_PUBLIC && $this->isDeleted( self::DELETED_USER ) ) {
 895                         return 0;
 896                 } elseif ( $audience == self::FOR_THIS_USER && !$this->userCan( self::DELETED_USER, $user ) ) {
 897                         return 0;
 898                 } else {
 899                         return $this->mUser;
 900                 }
 901         }
 902
 903         /**
 904          * Fetch revision's user id without regard for the current user's permissions
 905          *
 906          * @return int
 907          * @deprecated since 1.25, use getUser( Revision::RAW )
 908          */
 909         public function getRawUser() {
 910                 wfDeprecated( __METHOD__, '1.25' );
 911                 return $this->getUser( self::RAW );
 912         }
 913
 914         /**
 915          * Fetch revision's username if it's available to the specified audience.
 916          * If the specified audience does not have access to the username, an
 917          * empty string will be returned.
 918          *
 919          * @param int $audience One of:
 920          *   Revision::FOR_PUBLIC       to be displayed to all users
 921          *   Revision::FOR_THIS_USER    to be displayed to the given user
 922          *   Revision::RAW              get the text regardless of permissions
 923          * @param User|null $user User object to check for, only if FOR_THIS_USER is passed
 924          *   to the $audience parameter
 925          * @return string
 926          */
 927         public function getUserText( $audience = self::FOR_PUBLIC, User $user = null ) {
 928                 $this->loadMutableFields();
 929
 930                 if ( $audience == self::FOR_PUBLIC && $this->isDeleted( self::DELETED_USER ) ) {
 931                         return '';
 932                 } elseif ( $audience == self::FOR_THIS_USER && !$this->userCan( self::DELETED_USER, $user ) ) {
 933                         return '';
 934                 } else {
 935                         if ( $this->mUserText === null ) {
 936                                 $this->mUserText = User::whoIs( $this->mUser ); // load on demand
 937                                 if ( $this->mUserText === false ) {
 938                                         # This shouldn't happen, but it can if the wiki was recovered
 939                                         # via importing revs and there is no user table entry yet.
 940                                         $this->mUserText = $this->mOrigUserText;
 941                                 }
 942                         }
 943                         return $this->mUserText;
 944                 }
 945         }
 946
 947         /**
 948          * Fetch revision's username without regard for view restrictions
 949          *
 950          * @return string
 951          * @deprecated since 1.25, use getUserText( Revision::RAW )
 952          */
 953         public function getRawUserText() {
 954                 wfDeprecated( __METHOD__, '1.25' );
 955                 return $this->getUserText( self::RAW );
 956         }
 957
 958         /**
 959          * Fetch revision comment if it's available to the specified audience.
 960          * If the specified audience does not have access to the comment, an
 961          * empty string will be returned.
 962          *
 963          * @param int $audience One of:
 964          *   Revision::FOR_PUBLIC       to be displayed to all users
 965          *   Revision::FOR_THIS_USER    to be displayed to the given user
 966          *   Revision::RAW              get the text regardless of permissions
 967          * @param User|null $user User object to check for, only if FOR_THIS_USER is passed
 968          *   to the $audience parameter
 969          * @return string
 970          */
 971         function getComment( $audience = self::FOR_PUBLIC, User $user = null ) {
 972                 if ( $audience == self::FOR_PUBLIC && $this->isDeleted( self::DELETED_COMMENT ) ) {
 973                         return '';
 974                 } elseif ( $audience == self::FOR_THIS_USER && !$this->userCan( self::DELETED_COMMENT, $user ) ) {
 975                         return '';
 976                 } else {
 977                         return $this->mComment;
 978                 }
 979         }
 980
 981         /**
 982          * Fetch revision comment without regard for the current user's permissions
 983          *
 984          * @return string
 985          * @deprecated since 1.25, use getComment( Revision::RAW )
 986          */
 987         public function getRawComment() {
 988                 wfDeprecated( __METHOD__, '1.25' );
 989                 return $this->getComment( self::RAW );
 990         }
 991
 992         /**
 993          * @return bool
 994          */
 995         public function isMinor() {
 996                 return (bool)$this->mMinorEdit;
 997         }
 998
 999         /**
1000          * @return int Rcid of the unpatrolled row, zero if there isn't one
1001          */
1002         public function isUnpatrolled() {
1003                 if ( $this->mUnpatrolled !== null ) {
1004                         return $this->mUnpatrolled;
1005                 }
1006                 $rc = $this->getRecentChange();
1007                 if ( $rc && $rc->getAttribute( 'rc_patrolled' ) == 0 ) {
1008                         $this->mUnpatrolled = $rc->getAttribute( 'rc_id' );
1009                 } else {
1010                         $this->mUnpatrolled = 0;
1011                 }
1012                 return $this->mUnpatrolled;
1013         }
1014
1015         /**
1016          * Get the RC object belonging to the current revision, if there's one
1017          *
1018          * @param int $flags (optional) $flags include:
1019          *      Revision::READ_LATEST  : Select the data from the master
1020          *
1021          * @since 1.22
1022          * @return RecentChange|null
1023          */
1024         public function getRecentChange( $flags = 0 ) {
1025                 $dbr = wfGetDB( DB_REPLICA );
1026
1027                 list( $dbType, ) = DBAccessObjectUtils::getDBOptions( $flags );
1028
1029                 return RecentChange::newFromConds(
1030                         [
1031                                 'rc_user_text' => $this->getUserText( self::RAW ),
1032                                 'rc_timestamp' => $dbr->timestamp( $this->getTimestamp() ),
1033                                 'rc_this_oldid' => $this->getId()
1034                         ],
1035                         __METHOD__,
1036                         $dbType
1037                 );
1038         }
1039
1040         /**
1041          * @param int $field One of DELETED_* bitfield constants
1042          *
1043          * @return bool
1044          */
1045         public function isDeleted( $field ) {
1046                 if ( $this->isCurrent() && $field === self::DELETED_TEXT ) {
1047                         // Current revisions of pages cannot have the content hidden. Skipping this
1048                         // check is very useful for Parser as it fetches templates using newKnownCurrent().
1049                         // Calling getVisibility() in that case triggers a verification database query.
1050                         return false; // no need to check
1051                 }
1052
1053                 return ( $this->getVisibility() & $field ) == $field;
1054         }
1055
1056         /**
1057          * Get the deletion bitfield of the revision
1058          *
1059          * @return int
1060          */
1061         public function getVisibility() {
1062                 $this->loadMutableFields();
1063
1064                 return (int)$this->mDeleted;
1065         }
1066
1067         /**
1068          * Fetch revision content if it's available to the specified audience.
1069          * If the specified audience does not have the ability to view this
1070          * revision, null will be returned.
1071          *
1072          * @param int $audience One of:
1073          *   Revision::FOR_PUBLIC       to be displayed to all users
1074          *   Revision::FOR_THIS_USER    to be displayed to $wgUser
1075          *   Revision::RAW              get the text regardless of permissions
1076          * @param User $user User object to check for, only if FOR_THIS_USER is passed
1077          *   to the $audience parameter
1078          * @since 1.21
1079          * @return Content|null
1080          */
1081         public function getContent( $audience = self::FOR_PUBLIC, User $user = null ) {
1082                 if ( $audience == self::FOR_PUBLIC && $this->isDeleted( self::DELETED_TEXT ) ) {
1083                         return null;
1084                 } elseif ( $audience == self::FOR_THIS_USER && !$this->userCan( self::DELETED_TEXT, $user ) ) {
1085                         return null;
1086                 } else {
1087                         return $this->getContentInternal();
1088                 }
1089         }
1090
1091         /**
1092          * Get original serialized data (without checking view restrictions)
1093          *
1094          * @since 1.21
1095          * @return string
1096          */
1097         public function getSerializedData() {
1098                 if ( $this->mText === null ) {
1099                         // Revision is immutable. Load on demand.
1100                         $this->mText = $this->loadText();
1101                 }
1102
1103                 return $this->mText;
1104         }
1105
1106         /**
1107          * Gets the content object for the revision (or null on failure).
1108          *
1109          * Note that for mutable Content objects, each call to this method will return a
1110          * fresh clone.
1111          *
1112          * @since 1.21
1113          * @return Content|null The Revision's content, or null on failure.
1114          */
1115         protected function getContentInternal() {
1116                 if ( $this->mContent === null ) {
1117                         $text = $this->getSerializedData();
1118
1119                         if ( $text !== null && $text !== false ) {
1120                                 // Unserialize content
1121                                 $handler = $this->getContentHandler();
1122                                 $format = $this->getContentFormat();
1123
1124                                 $this->mContent = $handler->unserializeContent( $text, $format );
1125                         }
1126                 }
1127
1128                 // NOTE: copy() will return $this for immutable content objects
1129                 return $this->mContent ? $this->mContent->copy() : null;
1130         }
1131
1132         /**
1133          * Returns the content model for this revision.
1134          *
1135          * If no content model was stored in the database, the default content model for the title is
1136          * used to determine the content model to use. If no title is know, CONTENT_MODEL_WIKITEXT
1137          * is used as a last resort.
1138          *
1139          * @return string The content model id associated with this revision,
1140          *     see the CONTENT_MODEL_XXX constants.
1141          */
1142         public function getContentModel() {
1143                 if ( !$this->mContentModel ) {
1144                         $title = $this->getTitle();
1145                         if ( $title ) {
1146                                 $this->mContentModel = ContentHandler::getDefaultModelFor( $title );
1147                         } else {
1148                                 $this->mContentModel = CONTENT_MODEL_WIKITEXT;
1149                         }
1150
1151                         assert( !empty( $this->mContentModel ) );
1152                 }
1153
1154                 return $this->mContentModel;
1155         }
1156
1157         /**
1158          * Returns the content format for this revision.
1159          *
1160          * If no content format was stored in the database, the default format for this
1161          * revision's content model is returned.
1162          *
1163          * @return string The content format id associated with this revision,
1164          *     see the CONTENT_FORMAT_XXX constants.
1165          */
1166         public function getContentFormat() {
1167                 if ( !$this->mContentFormat ) {
1168                         $handler = $this->getContentHandler();
1169                         $this->mContentFormat = $handler->getDefaultFormat();
1170
1171                         assert( !empty( $this->mContentFormat ) );
1172                 }
1173
1174                 return $this->mContentFormat;
1175         }
1176
1177         /**
1178          * Returns the content handler appropriate for this revision's content model.
1179          *
1180          * @throws MWException
1181          * @return ContentHandler
1182          */
1183         public function getContentHandler() {
1184                 if ( !$this->mContentHandler ) {
1185                         $model = $this->getContentModel();
1186                         $this->mContentHandler = ContentHandler::getForModelID( $model );
1187
1188                         $format = $this->getContentFormat();
1189
1190                         if ( !$this->mContentHandler->isSupportedFormat( $format ) ) {
1191                                 throw new MWException( "Oops, the content format $format is not supported for "
1192                                         . "this content model, $model" );
1193                         }
1194                 }
1195
1196                 return $this->mContentHandler;
1197         }
1198
1199         /**
1200          * @return string
1201          */
1202         public function getTimestamp() {
1203                 return wfTimestamp( TS_MW, $this->mTimestamp );
1204         }
1205
1206         /**
1207          * @return bool
1208          */
1209         public function isCurrent() {
1210                 return $this->mCurrent;
1211         }
1212
1213         /**
1214          * Get previous revision for this title
1215          *
1216          * @return Revision|null
1217          */
1218         public function getPrevious() {
1219                 if ( $this->getTitle() ) {
1220                         $prev = $this->getTitle()->getPreviousRevisionID( $this->getId() );
1221                         if ( $prev ) {
1222                                 return self::newFromTitle( $this->getTitle(), $prev );
1223                         }
1224                 }
1225                 return null;
1226         }
1227
1228         /**
1229          * Get next revision for this title
1230          *
1231          * @return Revision|null
1232          */
1233         public function getNext() {
1234                 if ( $this->getTitle() ) {
1235                         $next = $this->getTitle()->getNextRevisionID( $this->getId() );
1236                         if ( $next ) {
1237                                 return self::newFromTitle( $this->getTitle(), $next );
1238                         }
1239                 }
1240                 return null;
1241         }
1242
1243         /**
1244          * Get previous revision Id for this page_id
1245          * This is used to populate rev_parent_id on save
1246          *
1247          * @param IDatabase $db
1248          * @return int
1249          */
1250         private function getPreviousRevisionId( $db ) {
1251                 if ( $this->mPage === null ) {
1252                         return 0;
1253                 }
1254                 # Use page_latest if ID is not given
1255                 if ( !$this->mId ) {
1256                         $prevId = $db->selectField( 'page', 'page_latest',
1257                                 [ 'page_id' => $this->mPage ],
1258                                 __METHOD__ );
1259                 } else {
1260                         $prevId = $db->selectField( 'revision', 'rev_id',
1261                                 [ 'rev_page' => $this->mPage, 'rev_id < ' . $this->mId ],
1262                                 __METHOD__,
1263                                 [ 'ORDER BY' => 'rev_id DESC' ] );
1264                 }
1265                 return intval( $prevId );
1266         }
1267
1268         /**
1269          * Get revision text associated with an old or archive row
1270          *
1271          * Both the flags and the text field must be included. Including the old_id
1272          * field will activate cache usage as long as the $wiki parameter is not set.
1273          *
1274          * @param stdClass $row The text data
1275          * @param string $prefix Table prefix (default 'old_')
1276          * @param string|bool $wiki The name of the wiki to load the revision text from
1277          *   (same as the the wiki $row was loaded from) or false to indicate the local
1278          *   wiki (this is the default). Otherwise, it must be a symbolic wiki database
1279          *   identifier as understood by the LoadBalancer class.
1280          * @return string|false Text the text requested or false on failure
1281          */
1282         public static function getRevisionText( $row, $prefix = 'old_', $wiki = false ) {
1283                 $textField = $prefix . 'text';
1284                 $flagsField = $prefix . 'flags';
1285
1286                 if ( isset( $row->$flagsField ) ) {
1287                         $flags = explode( ',', $row->$flagsField );
1288                 } else {
1289                         $flags = [];
1290                 }
1291
1292                 if ( isset( $row->$textField ) ) {
1293                         $text = $row->$textField;
1294                 } else {
1295                         return false;
1296                 }
1297
1298                 // Use external methods for external objects, text in table is URL-only then
1299                 if ( in_array( 'external', $flags ) ) {
1300                         $url = $text;
1301                         $parts = explode( '://', $url, 2 );
1302                         if ( count( $parts ) == 1 || $parts[1] == '' ) {
1303                                 return false;
1304                         }
1305
1306                         if ( isset( $row->old_id ) && $wiki === false ) {
1307                                 // Make use of the wiki-local revision text cache
1308                                 $cache = MediaWikiServices::getInstance()->getMainWANObjectCache();
1309                                 // The cached value should be decompressed, so handle that and return here
1310                                 return $cache->getWithSetCallback(
1311                                         $cache->makeKey( 'revisiontext', 'textid', $row->old_id ),
1312                                         self::getCacheTTL( $cache ),
1313                                         function () use ( $url, $wiki, $flags ) {
1314                                                 // No negative caching per Revision::loadText()
1315                                                 $text = ExternalStore::fetchFromURL( $url, [ 'wiki' => $wiki ] );
1316
1317                                                 return self::decompressRevisionText( $text, $flags );
1318                                         },
1319                                         [ 'pcGroup' => self::TEXT_CACHE_GROUP, 'pcTTL' => $cache::TTL_PROC_LONG ]
1320                                 );
1321                         } else {
1322                                 $text = ExternalStore::fetchFromURL( $url, [ 'wiki' => $wiki ] );
1323                         }
1324                 }
1325
1326                 return self::decompressRevisionText( $text, $flags );
1327         }
1328
1329         /**
1330          * If $wgCompressRevisions is enabled, we will compress data.
1331          * The input string is modified in place.
1332          * Return value is the flags field: contains 'gzip' if the
1333          * data is compressed, and 'utf-8' if we're saving in UTF-8
1334          * mode.
1335          *
1336          * @param mixed &$text Reference to a text
1337          * @return string
1338          */
1339         public static function compressRevisionText( &$text ) {
1340                 global $wgCompressRevisions;
1341                 $flags = [];
1342
1343                 # Revisions not marked this way will be converted
1344                 # on load if $wgLegacyCharset is set in the future.
1345                 $flags[] = 'utf-8';
1346
1347                 if ( $wgCompressRevisions ) {
1348                         if ( function_exists( 'gzdeflate' ) ) {
1349                                 $deflated = gzdeflate( $text );
1350
1351                                 if ( $deflated === false ) {
1352                                         wfLogWarning( __METHOD__ . ': gzdeflate() failed' );
1353                                 } else {
1354                                         $text = $deflated;
1355                                         $flags[] = 'gzip';
1356                                 }
1357                         } else {
1358                                 wfDebug( __METHOD__ . " -- no zlib support, not compressing\n" );
1359                         }
1360                 }
1361                 return implode( ',', $flags );
1362         }
1363
1364         /**
1365          * Re-converts revision text according to it's flags.
1366          *
1367          * @param mixed $text Reference to a text
1368          * @param array $flags Compression flags
1369          * @return string|bool Decompressed text, or false on failure
1370          */
1371         public static function decompressRevisionText( $text, $flags ) {
1372                 global $wgLegacyEncoding, $wgContLang;
1373
1374                 if ( $text === false ) {
1375                         // Text failed to be fetched; nothing to do
1376                         return false;
1377                 }
1378
1379                 if ( in_array( 'gzip', $flags ) ) {
1380                         # Deal with optional compression of archived pages.
1381                         # This can be done periodically via maintenance/compressOld.php, and
1382                         # as pages are saved if $wgCompressRevisions is set.
1383                         $text = gzinflate( $text );
1384
1385                         if ( $text === false ) {
1386                                 wfLogWarning( __METHOD__ . ': gzinflate() failed' );
1387                                 return false;
1388                         }
1389                 }
1390
1391                 if ( in_array( 'object', $flags ) ) {
1392                         # Generic compressed storage
1393                         $obj = unserialize( $text );
1394                         if ( !is_object( $obj ) ) {
1395                                 // Invalid object
1396                                 return false;
1397                         }
1398                         $text = $obj->getText();
1399                 }
1400
1401                 if ( $text !== false && $wgLegacyEncoding
1402                         && !in_array( 'utf-8', $flags ) && !in_array( 'utf8', $flags )
1403                 ) {
1404                         # Old revisions kept around in a legacy encoding?
1405                         # Upconvert on demand.
1406                         # ("utf8" checked for compatibility with some broken
1407                         #  conversion scripts 2008-12-30)
1408                         $text = $wgContLang->iconv( $wgLegacyEncoding, 'UTF-8', $text );
1409                 }
1410
1411                 return $text;
1412         }
1413
1414         /**
1415          * Insert a new revision into the database, returning the new revision ID
1416          * number on success and dies horribly on failure.
1417          *
1418          * @param IDatabase $dbw (master connection)
1419          * @throws MWException
1420          * @return int The revision ID
1421          */
1422         public function insertOn( $dbw ) {
1423                 global $wgDefaultExternalStore, $wgContentHandlerUseDB;
1424
1425                 // We're inserting a new revision, so we have to use master anyway.
1426                 // If it's a null revision, it may have references to rows that
1427                 // are not in the replica yet (the text row).
1428                 $this->mQueryFlags |= self::READ_LATEST;
1429
1430                 // Not allowed to have rev_page equal to 0, false, etc.
1431                 if ( !$this->mPage ) {
1432                         $title = $this->getTitle();
1433                         if ( $title instanceof Title ) {
1434                                 $titleText = ' for page ' . $title->getPrefixedText();
1435                         } else {
1436                                 $titleText = '';
1437                         }
1438                         throw new MWException( "Cannot insert revision$titleText: page ID must be nonzero" );
1439                 }
1440
1441                 $this->checkContentModel();
1442
1443                 $data = $this->mText;
1444                 $flags = self::compressRevisionText( $data );
1445
1446                 # Write to external storage if required
1447                 if ( $wgDefaultExternalStore ) {
1448                         // Store and get the URL
1449                         $data = ExternalStore::insertToDefault( $data );
1450                         if ( !$data ) {
1451                                 throw new MWException( "Unable to store text to external storage" );
1452                         }
1453                         if ( $flags ) {
1454                                 $flags .= ',';
1455                         }
1456                         $flags .= 'external';
1457                 }
1458
1459                 # Record the text (or external storage URL) to the text table
1460                 if ( $this->mTextId === null ) {
1461                         $dbw->insert( 'text',
1462                                 [
1463                                         'old_text' => $data,
1464                                         'old_flags' => $flags,
1465                                 ], __METHOD__
1466                         );
1467                         $this->mTextId = $dbw->insertId();
1468                 }
1469
1470                 if ( $this->mComment === null ) {
1471                         $this->mComment = "";
1472                 }
1473
1474                 # Record the edit in revisions
1475                 $row = [
1476                         'rev_page'       => $this->mPage,
1477                         'rev_text_id'    => $this->mTextId,
1478                         'rev_minor_edit' => $this->mMinorEdit ? 1 : 0,
1479                         'rev_user'       => $this->mUser,
1480                         'rev_user_text'  => $this->mUserText,
1481                         'rev_timestamp'  => $dbw->timestamp( $this->mTimestamp ),
1482                         'rev_deleted'    => $this->mDeleted,
1483                         'rev_len'        => $this->mSize,
1484                         'rev_parent_id'  => $this->mParentId === null
1485                                 ? $this->getPreviousRevisionId( $dbw )
1486                                 : $this->mParentId,
1487                         'rev_sha1'       => $this->mSha1 === null
1488                                 ? self::base36Sha1( $this->mText )
1489                                 : $this->mSha1,
1490                 ];
1491                 if ( $this->mId !== null ) {
1492                         $row['rev_id'] = $this->mId;
1493                 }
1494
1495                 list( $commentFields, $commentCallback ) =
1496                         CommentStore::newKey( 'rev_comment' )->insertWithTempTable( $dbw, $this->mComment );
1497                 $row += $commentFields;
1498
1499                 if ( $wgContentHandlerUseDB ) {
1500                         // NOTE: Store null for the default model and format, to save space.
1501                         // XXX: Makes the DB sensitive to changed defaults.
1502                         // Make this behavior optional? Only in miser mode?
1503
1504                         $model = $this->getContentModel();
1505                         $format = $this->getContentFormat();
1506
1507                         $title = $this->getTitle();
1508
1509                         if ( $title === null ) {
1510                                 throw new MWException( "Insufficient information to determine the title of the "
1511                                         . "revision's page!" );
1512                         }
1513
1514                         $defaultModel = ContentHandler::getDefaultModelFor( $title );
1515                         $defaultFormat = ContentHandler::getForModelID( $defaultModel )->getDefaultFormat();
1516
1517                         $row['rev_content_model'] = ( $model === $defaultModel ) ? null : $model;
1518                         $row['rev_content_format'] = ( $format === $defaultFormat ) ? null : $format;
1519                 }
1520
1521                 $dbw->insert( 'revision', $row, __METHOD__ );
1522
1523                 if ( $this->mId === null ) {
1524                         // Only if auto-increment was used
1525                         $this->mId = $dbw->insertId();
1526                 }
1527                 $commentCallback( $this->mId );
1528
1529                 // Assertion to try to catch T92046
1530                 if ( (int)$this->mId === 0 ) {
1531                         throw new UnexpectedValueException(
1532                                 'After insert, Revision mId is ' . var_export( $this->mId, 1 ) . ': ' .
1533                                         var_export( $row, 1 )
1534                         );
1535                 }
1536
1537                 // Insert IP revision into ip_changes for use when querying for a range.
1538                 if ( $this->mUser === 0 && IP::isValid( $this->mUserText ) ) {
1539                         $ipcRow = [
1540                                 'ipc_rev_id'        => $this->mId,
1541                                 'ipc_rev_timestamp' => $row['rev_timestamp'],
1542                                 'ipc_hex'           => IP::toHex( $row['rev_user_text'] ),
1543                         ];
1544                         $dbw->insert( 'ip_changes', $ipcRow, __METHOD__ );
1545                 }
1546
1547                 // Avoid PHP 7.1 warning of passing $this by reference
1548                 $revision = $this;
1549                 Hooks::run( 'RevisionInsertComplete', [ &$revision, $data, $flags ] );
1550
1551                 return $this->mId;
1552         }
1553
1554         protected function checkContentModel() {
1555                 global $wgContentHandlerUseDB;
1556
1557                 // Note: may return null for revisions that have not yet been inserted
1558                 $title = $this->getTitle();
1559
1560                 $model = $this->getContentModel();
1561                 $format = $this->getContentFormat();
1562                 $handler = $this->getContentHandler();
1563
1564                 if ( !$handler->isSupportedFormat( $format ) ) {
1565                         $t = $title->getPrefixedDBkey();
1566
1567                         throw new MWException( "Can't use format $format with content model $model on $t" );
1568                 }
1569
1570                 if ( !$wgContentHandlerUseDB && $title ) {
1571                         // if $wgContentHandlerUseDB is not set,
1572                         // all revisions must use the default content model and format.
1573
1574                         $defaultModel = ContentHandler::getDefaultModelFor( $title );
1575                         $defaultHandler = ContentHandler::getForModelID( $defaultModel );
1576                         $defaultFormat = $defaultHandler->getDefaultFormat();
1577
1578                         if ( $this->getContentModel() != $defaultModel ) {
1579                                 $t = $title->getPrefixedDBkey();
1580
1581                                 throw new MWException( "Can't save non-default content model with "
1582                                         . "\$wgContentHandlerUseDB disabled: model is $model, "
1583                                         . "default for $t is $defaultModel" );
1584                         }
1585
1586                         if ( $this->getContentFormat() != $defaultFormat ) {
1587                                 $t = $title->getPrefixedDBkey();
1588
1589                                 throw new MWException( "Can't use non-default content format with "
1590                                         . "\$wgContentHandlerUseDB disabled: format is $format, "
1591                                         . "default for $t is $defaultFormat" );
1592                         }
1593                 }
1594
1595                 $content = $this->getContent( self::RAW );
1596                 $prefixedDBkey = $title->getPrefixedDBkey();
1597                 $revId = $this->mId;
1598
1599                 if ( !$content ) {
1600                         throw new MWException(
1601                                 "Content of revision $revId ($prefixedDBkey) could not be loaded for validation!"
1602                         );
1603                 }
1604                 if ( !$content->isValid() ) {
1605                         throw new MWException(
1606                                 "Content of revision $revId ($prefixedDBkey) is not valid! Content model is $model"
1607                         );
1608                 }
1609         }
1610
1611         /**
1612          * Get the base 36 SHA-1 value for a string of text
1613          * @param string $text
1614          * @return string
1615          */
1616         public static function base36Sha1( $text ) {
1617                 return Wikimedia\base_convert( sha1( $text ), 16, 36, 31 );
1618         }
1619
1620         /**
1621          * Get the text cache TTL
1622          *
1623          * @param WANObjectCache $cache
1624          * @return int
1625          */
1626         private static function getCacheTTL( WANObjectCache $cache ) {
1627                 global $wgRevisionCacheExpiry;
1628
1629                 if ( $cache->getQoS( $cache::ATTR_EMULATION ) <= $cache::QOS_EMULATION_SQL ) {
1630                         // Do not cache RDBMs blobs in...the RDBMs store
1631                         $ttl = $cache::TTL_UNCACHEABLE;
1632                 } else {
1633                         $ttl = $wgRevisionCacheExpiry ?: $cache::TTL_UNCACHEABLE;
1634                 }
1635
1636                 return $ttl;
1637         }
1638
1639         /**
1640          * Lazy-load the revision's text.
1641          * Currently hardcoded to the 'text' table storage engine.
1642          *
1643          * @return string|bool The revision's text, or false on failure
1644          */
1645         private function loadText() {
1646                 $cache = ObjectCache::getMainWANInstance();
1647
1648                 // No negative caching; negative hits on text rows may be due to corrupted replica DBs
1649                 return $cache->getWithSetCallback(
1650                         $cache->makeKey( 'revisiontext', 'textid', $this->getTextId() ),
1651                         self::getCacheTTL( $cache ),
1652                         function () {
1653                                 return $this->fetchText();
1654                         },
1655                         [ 'pcGroup' => self::TEXT_CACHE_GROUP, 'pcTTL' => $cache::TTL_PROC_LONG ]
1656                 );
1657         }
1658
1659         private function fetchText() {
1660                 $textId = $this->getTextId();
1661
1662                 // If we kept data for lazy extraction, use it now...
1663                 if ( $this->mTextRow !== null ) {
1664                         $row = $this->mTextRow;
1665                         $this->mTextRow = null;
1666                 } else {
1667                         $row = null;
1668                 }
1669
1670                 // Callers doing updates will pass in READ_LATEST as usual. Since the text/blob tables
1671                 // do not normally get rows changed around, set READ_LATEST_IMMUTABLE in those cases.
1672                 $flags = $this->mQueryFlags;
1673                 $flags |= DBAccessObjectUtils::hasFlags( $flags, self::READ_LATEST )
1674                         ? self::READ_LATEST_IMMUTABLE
1675                         : 0;
1676
1677                 list( $index, $options, $fallbackIndex, $fallbackOptions ) =
1678                         DBAccessObjectUtils::getDBOptions( $flags );
1679
1680                 if ( !$row ) {
1681                         // Text data is immutable; check replica DBs first.
1682                         $row = wfGetDB( $index )->selectRow(
1683                                 'text',
1684                                 [ 'old_text', 'old_flags' ],
1685                                 [ 'old_id' => $textId ],
1686                                 __METHOD__,
1687                                 $options
1688                         );
1689                 }
1690
1691                 // Fallback to DB_MASTER in some cases if the row was not found
1692                 if ( !$row && $fallbackIndex !== null ) {
1693                         // Use FOR UPDATE if it was used to fetch this revision. This avoids missing the row
1694                         // due to REPEATABLE-READ. Also fallback to the master if READ_LATEST is provided.
1695                         $row = wfGetDB( $fallbackIndex )->selectRow(
1696                                 'text',
1697                                 [ 'old_text', 'old_flags' ],
1698                                 [ 'old_id' => $textId ],
1699                                 __METHOD__,
1700                                 $fallbackOptions
1701                         );
1702                 }
1703
1704                 if ( !$row ) {
1705                         wfDebugLog( 'Revision', "No text row with ID '$textId' (revision {$this->getId()})." );
1706                 }
1707
1708                 $text = self::getRevisionText( $row );
1709                 if ( $row && $text === false ) {
1710                         wfDebugLog( 'Revision', "No blob for text row '$textId' (revision {$this->getId()})." );
1711                 }
1712
1713                 return is_string( $text ) ? $text : false;
1714         }
1715
1716         /**
1717          * Create a new null-revision for insertion into a page's
1718          * history. This will not re-save the text, but simply refer
1719          * to the text from the previous version.
1720          *
1721          * Such revisions can for instance identify page rename
1722          * operations and other such meta-modifications.
1723          *
1724          * @param IDatabase $dbw
1725          * @param int $pageId ID number of the page to read from
1726          * @param string $summary Revision's summary
1727          * @param bool $minor Whether the revision should be considered as minor
1728          * @param User|null $user User object to use or null for $wgUser
1729          * @return Revision|null Revision or null on error
1730          */
1731         public static function newNullRevision( $dbw, $pageId, $summary, $minor, $user = null ) {
1732                 global $wgContentHandlerUseDB;
1733
1734                 $fields = [ 'page_latest', 'page_namespace', 'page_title',
1735                                                 'rev_text_id', 'rev_len', 'rev_sha1' ];
1736
1737                 if ( $wgContentHandlerUseDB ) {
1738                         $fields[] = 'rev_content_model';
1739                         $fields[] = 'rev_content_format';
1740                 }
1741
1742                 $current = $dbw->selectRow(
1743                         [ 'page', 'revision' ],
1744                         $fields,
1745                         [
1746                                 'page_id' => $pageId,
1747                                 'page_latest=rev_id',
1748                         ],
1749                         __METHOD__,
1750                         [ 'FOR UPDATE' ] // T51581
1751                 );
1752
1753                 if ( $current ) {
1754                         if ( !$user ) {
1755                                 global $wgUser;
1756                                 $user = $wgUser;
1757                         }
1758
1759                         $row = [
1760                                 'page'       => $pageId,
1761                                 'user_text'  => $user->getName(),
1762                                 'user'       => $user->getId(),
1763                                 'comment'    => $summary,
1764                                 'minor_edit' => $minor,
1765                                 'text_id'    => $current->rev_text_id,
1766                                 'parent_id'  => $current->page_latest,
1767                                 'len'        => $current->rev_len,
1768                                 'sha1'       => $current->rev_sha1
1769                         ];
1770
1771                         if ( $wgContentHandlerUseDB ) {
1772                                 $row['content_model'] = $current->rev_content_model;
1773                                 $row['content_format'] = $current->rev_content_format;
1774                         }
1775
1776                         $row['title'] = Title::makeTitle( $current->page_namespace, $current->page_title );
1777
1778                         $revision = new Revision( $row );
1779                 } else {
1780                         $revision = null;
1781                 }
1782
1783                 return $revision;
1784         }
1785
1786         /**
1787          * Determine if the current user is allowed to view a particular
1788          * field of this revision, if it's marked as deleted.
1789          *
1790          * @param int $field One of self::DELETED_TEXT,
1791          *                              self::DELETED_COMMENT,
1792          *                              self::DELETED_USER
1793          * @param User|null $user User object to check, or null to use $wgUser
1794          * @return bool
1795          */
1796         public function userCan( $field, User $user = null ) {
1797                 return self::userCanBitfield( $this->getVisibility(), $field, $user );
1798         }
1799
1800         /**
1801          * Determine if the current user is allowed to view a particular
1802          * field of this revision, if it's marked as deleted. This is used
1803          * by various classes to avoid duplication.
1804          *
1805          * @param int $bitfield Current field
1806          * @param int $field One of self::DELETED_TEXT = File::DELETED_FILE,
1807          *                               self::DELETED_COMMENT = File::DELETED_COMMENT,
1808          *                               self::DELETED_USER = File::DELETED_USER
1809          * @param User|null $user User object to check, or null to use $wgUser
1810          * @param Title|null $title A Title object to check for per-page restrictions on,
1811          *                          instead of just plain userrights
1812          * @return bool
1813          */
1814         public static function userCanBitfield( $bitfield, $field, User $user = null,
1815                 Title $title = null
1816         ) {
1817                 if ( $bitfield & $field ) { // aspect is deleted
1818                         if ( $user === null ) {
1819                                 global $wgUser;
1820                                 $user = $wgUser;
1821                         }
1822                         if ( $bitfield & self::DELETED_RESTRICTED ) {
1823                                 $permissions = [ 'suppressrevision', 'viewsuppressed' ];
1824                         } elseif ( $field & self::DELETED_TEXT ) {
1825                                 $permissions = [ 'deletedtext' ];
1826                         } else {
1827                                 $permissions = [ 'deletedhistory' ];
1828                         }
1829                         $permissionlist = implode( ', ', $permissions );
1830                         if ( $title === null ) {
1831                                 wfDebug( "Checking for $permissionlist due to $field match on $bitfield\n" );
1832                                 return call_user_func_array( [ $user, 'isAllowedAny' ], $permissions );
1833                         } else {
1834                                 $text = $title->getPrefixedText();
1835                                 wfDebug( "Checking for $permissionlist on $text due to $field match on $bitfield\n" );
1836                                 foreach ( $permissions as $perm ) {
1837                                         if ( $title->userCan( $perm, $user ) ) {
1838                                                 return true;
1839                                         }
1840                                 }
1841                                 return false;
1842                         }
1843                 } else {
1844                         return true;
1845                 }
1846         }
1847
1848         /**
1849          * Get rev_timestamp from rev_id, without loading the rest of the row
1850          *
1851          * @param Title $title
1852          * @param int $id
1853          * @param int $flags
1854          * @return string|bool False if not found
1855          */
1856         static function getTimestampFromId( $title, $id, $flags = 0 ) {
1857                 $db = ( $flags & self::READ_LATEST )
1858                         ? wfGetDB( DB_MASTER )
1859                         : wfGetDB( DB_REPLICA );
1860                 // Casting fix for databases that can't take '' for rev_id
1861                 if ( $id == '' ) {
1862                         $id = 0;
1863                 }
1864                 $conds = [ 'rev_id' => $id ];
1865                 $conds['rev_page'] = $title->getArticleID();
1866                 $timestamp = $db->selectField( 'revision', 'rev_timestamp', $conds, __METHOD__ );
1867
1868                 return ( $timestamp !== false ) ? wfTimestamp( TS_MW, $timestamp ) : false;
1869         }
1870
1871         /**
1872          * Get count of revisions per page...not very efficient
1873          *
1874          * @param IDatabase $db
1875          * @param int $id Page id
1876          * @return int
1877          */
1878         static function countByPageId( $db, $id ) {
1879                 $row = $db->selectRow( 'revision', [ 'revCount' => 'COUNT(*)' ],
1880                         [ 'rev_page' => $id ], __METHOD__ );
1881                 if ( $row ) {
1882                         return $row->revCount;
1883                 }
1884                 return 0;
1885         }
1886
1887         /**
1888          * Get count of revisions per page...not very efficient
1889          *
1890          * @param IDatabase $db
1891          * @param Title $title
1892          * @return int
1893          */
1894         static function countByTitle( $db, $title ) {
1895                 $id = $title->getArticleID();
1896                 if ( $id ) {
1897                         return self::countByPageId( $db, $id );
1898                 }
1899                 return 0;
1900         }
1901
1902         /**
1903          * Check if no edits were made by other users since
1904          * the time a user started editing the page. Limit to
1905          * 50 revisions for the sake of performance.
1906          *
1907          * @since 1.20
1908          * @deprecated since 1.24
1909          *
1910          * @param IDatabase|int $db The Database to perform the check on. May be given as a
1911          *        Database object or a database identifier usable with wfGetDB.
1912          * @param int $pageId The ID of the page in question
1913          * @param int $userId The ID of the user in question
1914          * @param string $since Look at edits since this time
1915          *
1916          * @return bool True if the given user was the only one to edit since the given timestamp
1917          */
1918         public static function userWasLastToEdit( $db, $pageId, $userId, $since ) {
1919                 if ( !$userId ) {
1920                         return false;
1921                 }
1922
1923                 if ( is_int( $db ) ) {
1924                         $db = wfGetDB( $db );
1925                 }
1926
1927                 $res = $db->select( 'revision',
1928                         'rev_user',
1929                         [
1930                                 'rev_page' => $pageId,
1931                                 'rev_timestamp > ' . $db->addQuotes( $db->timestamp( $since ) )
1932                         ],
1933                         __METHOD__,
1934                         [ 'ORDER BY' => 'rev_timestamp ASC', 'LIMIT' => 50 ] );
1935                 foreach ( $res as $row ) {
1936                         if ( $row->rev_user != $userId ) {
1937                                 return false;
1938                         }
1939                 }
1940                 return true;
1941         }
1942
1943         /**
1944          * Load a revision based on a known page ID and current revision ID from the DB
1945          *
1946          * This method allows for the use of caching, though accessing anything that normally
1947          * requires permission checks (aside from the text) will trigger a small DB lookup.
1948          * The title will also be lazy loaded, though setTitle() can be used to preload it.
1949          *
1950          * @param IDatabase $db
1951          * @param int $pageId Page ID
1952          * @param int $revId Known current revision of this page
1953          * @return Revision|bool Returns false if missing
1954          * @since 1.28
1955          */
1956         public static function newKnownCurrent( IDatabase $db, $pageId, $revId ) {
1957                 $cache = MediaWikiServices::getInstance()->getMainWANObjectCache();
1958                 return $cache->getWithSetCallback(
1959                         // Page/rev IDs passed in from DB to reflect history merges
1960                         $cache->makeGlobalKey( 'revision', $db->getDomainID(), $pageId, $revId ),
1961                         $cache::TTL_WEEK,
1962                         function ( $curValue, &$ttl, array &$setOpts ) use ( $db, $pageId, $revId ) {
1963                                 $setOpts += Database::getCacheSetOptions( $db );
1964
1965                                 $rev = Revision::loadFromPageId( $db, $pageId, $revId );
1966                                 // Reflect revision deletion and user renames
1967                                 if ( $rev ) {
1968                                         $rev->mTitle = null; // mutable; lazy-load
1969                                         $rev->mRefreshMutableFields = true;
1970                                 }
1971
1972                                 return $rev ?: false; // don't cache negatives
1973                         }
1974                 );
1975         }
1976
1977         /**
1978          * For cached revisions, make sure the user name and rev_deleted is up-to-date
1979          */
1980         private function loadMutableFields() {
1981                 if ( !$this->mRefreshMutableFields ) {
1982                         return; // not needed
1983                 }
1984
1985                 $this->mRefreshMutableFields = false;
1986                 $dbr = wfGetLB( $this->mWiki )->getConnectionRef( DB_REPLICA, [], $this->mWiki );
1987                 $row = $dbr->selectRow(
1988                         [ 'revision', 'user' ],
1989                         [ 'rev_deleted', 'user_name' ],
1990                         [ 'rev_id' => $this->mId, 'user_id = rev_user' ],
1991                         __METHOD__
1992                 );
1993                 if ( $row ) { // update values
1994                         $this->mDeleted = (int)$row->rev_deleted;
1995                         $this->mUserText = $row->user_name;
1996                 }
1997         }
1998 }