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