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