includes/Storage/RevisionStore.php

   1 <?php
   2 /**
   3  * Service for looking up page revisions.
   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  * Attribution notice: when this file was created, much of its content was taken
  21  * from the Revision.php file as present in release 1.30. Refer to the history
  22  * of that file for original authorship.
  23  *
  24  * @file
  25  */
  26
  27 namespace MediaWiki\Storage;
  28
  29 use CommentStore;
  30 use CommentStoreComment;
  31 use Content;
  32 use ContentHandler;
  33 use DBAccessObjectUtils;
  34 use Hooks;
  35 use IDBAccessObject;
  36 use InvalidArgumentException;
  37 use IP;
  38 use LogicException;
  39 use MediaWiki\Linker\LinkTarget;
  40 use MediaWiki\User\UserIdentity;
  41 use MediaWiki\User\UserIdentityValue;
  42 use Message;
  43 use MWException;
  44 use MWUnknownContentModelException;
  45 use RecentChange;
  46 use stdClass;
  47 use Title;
  48 use User;
  49 use WANObjectCache;
  50 use Wikimedia\Assert\Assert;
  51 use Wikimedia\Rdbms\Database;
  52 use Wikimedia\Rdbms\DBConnRef;
  53 use Wikimedia\Rdbms\IDatabase;
  54 use Wikimedia\Rdbms\LoadBalancer;
  55
  56 /**
  57  * Service for looking up page revisions.
  58  *
  59  * @since 1.31
  60  *
  61  * @note This was written to act as a drop-in replacement for the corresponding
  62  *       static methods in Revision.
  63  */
  64 class RevisionStore implements IDBAccessObject, RevisionFactory, RevisionLookup {
  65
  66         /**
  67          * @var SqlBlobStore
  68          */
  69         private $blobStore;
  70
  71         /**
  72          * @var bool|string
  73          */
  74         private $wikiId;
  75
  76         /**
  77          * @var boolean
  78          */
  79         private $contentHandlerUseDB = true;
  80
  81         /**
  82          * @var LoadBalancer
  83          */
  84         private $loadBalancer;
  85
  86         /**
  87          * @var WANObjectCache
  88          */
  89         private $cache;
  90
  91         /**
  92          * @todo $blobStore should be allowed to be any BlobStore!
  93          *
  94          * @param LoadBalancer $loadBalancer
  95          * @param SqlBlobStore $blobStore
  96          * @param WANObjectCache $cache
  97          * @param bool|string $wikiId
  98          */
  99         public function __construct(
 100                 LoadBalancer $loadBalancer,
 101                 SqlBlobStore $blobStore,
 102                 WANObjectCache $cache,
 103                 $wikiId = false
 104         ) {
 105                 Assert::parameterType( 'string|boolean', $wikiId, '$wikiId' );
 106
 107                 $this->loadBalancer = $loadBalancer;
 108                 $this->blobStore = $blobStore;
 109                 $this->cache = $cache;
 110                 $this->wikiId = $wikiId;
 111         }
 112
 113         /**
 114          * @return bool
 115          */
 116         public function getContentHandlerUseDB() {
 117                 return $this->contentHandlerUseDB;
 118         }
 119
 120         /**
 121          * @param bool $contentHandlerUseDB
 122          */
 123         public function setContentHandlerUseDB( $contentHandlerUseDB ) {
 124                 $this->contentHandlerUseDB = $contentHandlerUseDB;
 125         }
 126
 127         /**
 128          * @return LoadBalancer
 129          */
 130         private function getDBLoadBalancer() {
 131                 return $this->loadBalancer;
 132         }
 133
 134         /**
 135          * @param int $mode DB_MASTER or DB_REPLICA
 136          *
 137          * @return IDatabase
 138          */
 139         private function getDBConnection( $mode ) {
 140                 $lb = $this->getDBLoadBalancer();
 141                 return $lb->getConnection( $mode, [], $this->wikiId );
 142         }
 143
 144         /**
 145          * @param IDatabase $connection
 146          */
 147         private function releaseDBConnection( IDatabase $connection ) {
 148                 $lb = $this->getDBLoadBalancer();
 149                 $lb->reuseConnection( $connection );
 150         }
 151
 152         /**
 153          * @param int $mode DB_MASTER or DB_REPLICA
 154          *
 155          * @return DBConnRef
 156          */
 157         private function getDBConnectionRef( $mode ) {
 158                 $lb = $this->getDBLoadBalancer();
 159                 return $lb->getConnectionRef( $mode, [], $this->wikiId );
 160         }
 161
 162         /**
 163          * Determines the page Title based on the available information.
 164          *
 165          * MCR migration note: this corresponds to Revision::getTitle
 166          *
 167          * @param int|null $pageId
 168          * @param int|null $revId
 169          * @param int $queryFlags
 170          *
 171          * @return Title
 172          * @throws RevisionAccessException
 173          */
 174         private function getTitle( $pageId, $revId, $queryFlags = 0 ) {
 175                 if ( !$pageId && !$revId ) {
 176                         throw new InvalidArgumentException( '$pageId and $revId cannot both be 0 or null' );
 177                 }
 178
 179                 $title = null;
 180
 181                 // Loading by ID is best, but Title::newFromID does not support that for foreign IDs.
 182                 if ( $pageId !== null && $pageId > 0 && $this->wikiId === false ) {
 183                         // TODO: better foreign title handling (introduce TitleFactory)
 184                         $title = Title::newFromID( $pageId, $queryFlags );
 185                 }
 186
 187                 // rev_id is defined as NOT NULL, but this revision may not yet have been inserted.
 188                 if ( !$title && $revId !== null && $revId > 0 ) {
 189                         list( $dbMode, $dbOptions, , ) = DBAccessObjectUtils::getDBOptions( $queryFlags );
 190
 191                         $dbr = $this->getDbConnectionRef( $dbMode );
 192                         // @todo: Title::getSelectFields(), or Title::getQueryInfo(), or something like that
 193                         $row = $dbr->selectRow(
 194                                 [ 'revision', 'page' ],
 195                                 [
 196                                         'page_namespace',
 197                                         'page_title',
 198                                         'page_id',
 199                                         'page_latest',
 200                                         'page_is_redirect',
 201                                         'page_len',
 202                                 ],
 203                                 [ 'rev_id' => $revId ],
 204                                 __METHOD__,
 205                                 $dbOptions,
 206                                 [ 'page' => [ 'JOIN', 'page_id=rev_page' ] ]
 207                         );
 208                         if ( $row ) {
 209                                 // TODO: better foreign title handling (introduce TitleFactory)
 210                                 $title = Title::newFromRow( $row );
 211                         }
 212                 }
 213
 214                 if ( !$title ) {
 215                         throw new RevisionAccessException(
 216                                 "Could not determine title for page ID $pageId and revision ID $revId"
 217                         );
 218                 }
 219
 220                 return $title;
 221         }
 222
 223         /**
 224          * @param mixed $value
 225          * @param string $name
 226          *
 227          * @throw IncompleteRevisionException if $value is null
 228          * @return mixed $value, if $value is not null
 229          */
 230         private function failOnNull( $value, $name ) {
 231                 if ( $value === null ) {
 232                         throw new IncompleteRevisionException(
 233                                 "$name must not be " . var_export( $value, true ) . "!"
 234                         );
 235                 }
 236
 237                 return $value;
 238         }
 239
 240         /**
 241          * @param mixed $value
 242          * @param string $name
 243          *
 244          * @throw IncompleteRevisionException if $value is empty
 245          * @return mixed $value, if $value is not null
 246          */
 247         private function failOnEmpty( $value, $name ) {
 248                 if ( $value === null || $value === 0 || $value === '' ) {
 249                         throw new IncompleteRevisionException(
 250                                 "$name must not be " . var_export( $value, true ) . "!"
 251                         );
 252                 }
 253
 254                 return $value;
 255         }
 256
 257         /**
 258          * Insert a new revision into the database, returning the new revision ID
 259          * number on success and dies horribly on failure.
 260          *
 261          * MCR migration note: this replaces Revision::insertOn
 262          *
 263          * @param RevisionRecord $rev
 264          * @param IDatabase $dbw (master connection)
 265          *
 266          * @throws InvalidArgumentException
 267          * @return RevisionRecord the new revision record.
 268          */
 269         public function insertRevisionOn( RevisionRecord $rev, IDatabase $dbw ) {
 270                 // TODO: pass in a DBTransactionContext instead of a database connection.
 271                 $this->checkDatabaseWikiId( $dbw );
 272
 273                 if ( !$rev->getSlotRoles() ) {
 274                         throw new InvalidArgumentException( 'At least one slot needs to be defined!' );
 275                 }
 276
 277                 if ( $rev->getSlotRoles() !== [ 'main' ] ) {
 278                         throw new InvalidArgumentException( 'Only the main slot is supported for now!' );
 279                 }
 280
 281                 // TODO: we shouldn't need an actual Title here.
 282                 $title = Title::newFromLinkTarget( $rev->getPageAsLinkTarget() );
 283                 $pageId = $this->failOnEmpty( $rev->getPageId(), 'rev_page field' ); // check this early
 284
 285                 $parentId = $rev->getParentId() === null
 286                         ? $this->getPreviousRevisionId( $dbw, $rev )
 287                         : $rev->getParentId();
 288
 289                 // Record the text (or external storage URL) to the blob store
 290                 $slot = $rev->getSlot( 'main', RevisionRecord::RAW );
 291
 292                 $size = $this->failOnNull( $rev->getSize(), 'size field' );
 293                 $sha1 = $this->failOnEmpty( $rev->getSha1(), 'sha1 field' );
 294
 295                 if ( !$slot->hasAddress() ) {
 296                         $content = $slot->getContent();
 297                         $format = $content->getDefaultFormat();
 298                         $model = $content->getModel();
 299
 300                         $this->checkContentModel( $content, $title );
 301
 302                         $data = $content->serialize( $format );
 303
 304                         // Hints allow the blob store to optimize by "leaking" application level information to it.
 305                         // TODO: with the new MCR storage schema, we rev_id have this before storing the blobs.
 306                         // When we have it, add rev_id as a hint. Can be used with rev_parent_id for
 307                         // differential storage or compression of subsequent revisions.
 308                         $blobHints = [
 309                                 BlobStore::DESIGNATION_HINT => 'page-content', // BlobStore may be used for other things too.
 310                                 BlobStore::PAGE_HINT => $pageId,
 311                                 BlobStore::ROLE_HINT => $slot->getRole(),
 312                                 BlobStore::PARENT_HINT => $parentId,
 313                                 BlobStore::SHA1_HINT => $slot->getSha1(),
 314                                 BlobStore::MODEL_HINT => $model,
 315                                 BlobStore::FORMAT_HINT => $format,
 316                         ];
 317
 318                         $blobAddress = $this->blobStore->storeBlob( $data, $blobHints );
 319                 } else {
 320                         $blobAddress = $slot->getAddress();
 321                         $model = $slot->getModel();
 322                         $format = $slot->getFormat();
 323                 }
 324
 325                 $textId = $this->blobStore->getTextIdFromAddress( $blobAddress );
 326
 327                 if ( !$textId ) {
 328                         throw new LogicException(
 329                                 'Blob address not supported in 1.29 database schema: ' . $blobAddress
 330                         );
 331                 }
 332
 333                 // getTextIdFromAddress() is free to insert something into the text table, so $textId
 334                 // may be a new value, not anything already contained in $blobAddress.
 335                 $blobAddress = 'tt:' . $textId;
 336
 337                 $comment = $this->failOnNull( $rev->getComment( RevisionRecord::RAW ), 'comment' );
 338                 $user = $this->failOnNull( $rev->getUser( RevisionRecord::RAW ), 'user' );
 339                 $timestamp = $this->failOnEmpty( $rev->getTimestamp(), 'timestamp field' );
 340
 341                 # Record the edit in revisions
 342                 $row = [
 343                         'rev_page'       => $pageId,
 344                         'rev_parent_id'  => $parentId,
 345                         'rev_text_id'    => $textId,
 346                         'rev_minor_edit' => $rev->isMinor() ? 1 : 0,
 347                         'rev_user'       => $this->failOnNull( $user->getId(), 'user field' ),
 348                         'rev_user_text'  => $this->failOnEmpty( $user->getName(), 'user_text field' ),
 349                         'rev_timestamp'  => $dbw->timestamp( $timestamp ),
 350                         'rev_deleted'    => $rev->getVisibility(),
 351                         'rev_len'        => $size,
 352                         'rev_sha1'       => $sha1,
 353                 ];
 354
 355                 if ( $rev->getId() !== null ) {
 356                         // Needed to restore revisions with their original ID
 357                         $row['rev_id'] = $rev->getId();
 358                 }
 359
 360                 list( $commentFields, $commentCallback ) =
 361                         CommentStore::newKey( 'rev_comment' )->insertWithTempTable( $dbw, $comment );
 362                 $row += $commentFields;
 363
 364                 if ( $this->contentHandlerUseDB ) {
 365                         // MCR migration note: rev_content_model and rev_content_format will go away
 366
 367                         $defaultModel = ContentHandler::getDefaultModelFor( $title );
 368                         $defaultFormat = ContentHandler::getForModelID( $defaultModel )->getDefaultFormat();
 369
 370                         $row['rev_content_model'] = ( $model === $defaultModel ) ? null : $model;
 371                         $row['rev_content_format'] = ( $format === $defaultFormat ) ? null : $format;
 372                 }
 373
 374                 $dbw->insert( 'revision', $row, __METHOD__ );
 375
 376                 if ( !isset( $row['rev_id'] ) ) {
 377                         // only if auto-increment was used
 378                         $row['rev_id'] = intval( $dbw->insertId() );
 379                 }
 380                 $commentCallback( $row['rev_id'] );
 381
 382                 // Insert IP revision into ip_changes for use when querying for a range.
 383                 if ( $row['rev_user'] === 0 && IP::isValid( $row['rev_user_text'] ) ) {
 384                         $ipcRow = [
 385                                 'ipc_rev_id'        => $row['rev_id'],
 386                                 'ipc_rev_timestamp' => $row['rev_timestamp'],
 387                                 'ipc_hex'           => IP::toHex( $row['rev_user_text'] ),
 388                         ];
 389                         $dbw->insert( 'ip_changes', $ipcRow, __METHOD__ );
 390                 }
 391
 392                 $newSlot = SlotRecord::newSaved( $row['rev_id'], $blobAddress, $slot );
 393                 $slots = new RevisionSlots( [ 'main' => $newSlot ] );
 394
 395                 $user = new UserIdentityValue( intval( $row['rev_user'] ), $row['rev_user_text'] );
 396
 397                 $rev = new RevisionStoreRecord(
 398                         $title,
 399                         $user,
 400                         $comment,
 401                         (object)$row,
 402                         $slots,
 403                         $this->wikiId
 404                 );
 405
 406                 $newSlot = $rev->getSlot( 'main', RevisionRecord::RAW );
 407
 408                 // sanity checks
 409                 Assert::postcondition( $rev->getId() > 0, 'revision must have an ID' );
 410                 Assert::postcondition( $rev->getPageId() > 0, 'revision must have a page ID' );
 411                 Assert::postcondition(
 412                         $rev->getComment( RevisionRecord::RAW ) !== null,
 413                         'revision must have a comment'
 414                 );
 415                 Assert::postcondition(
 416                         $rev->getUser( RevisionRecord::RAW ) !== null,
 417                         'revision must have a user'
 418                 );
 419
 420                 Assert::postcondition( $newSlot !== null, 'revision must have a main slot' );
 421                 Assert::postcondition(
 422                         $newSlot->getAddress() !== null,
 423                         'main slot must have an addess'
 424                 );
 425
 426                 Hooks::run( 'RevisionRecordInserted', [ $rev ] );
 427
 428                 return $rev;
 429         }
 430
 431         /**
 432          * MCR migration note: this corresponds to Revision::checkContentModel
 433          *
 434          * @param Content $content
 435          * @param Title $title
 436          *
 437          * @throws MWException
 438          * @throws MWUnknownContentModelException
 439          */
 440         private function checkContentModel( Content $content, Title $title ) {
 441                 // Note: may return null for revisions that have not yet been inserted
 442
 443                 $model = $content->getModel();
 444                 $format = $content->getDefaultFormat();
 445                 $handler = $content->getContentHandler();
 446
 447                 $name = "$title";
 448
 449                 if ( !$handler->isSupportedFormat( $format ) ) {
 450                         throw new MWException( "Can't use format $format with content model $model on $name" );
 451                 }
 452
 453                 if ( !$this->contentHandlerUseDB ) {
 454                         // if $wgContentHandlerUseDB is not set,
 455                         // all revisions must use the default content model and format.
 456
 457                         $defaultModel = ContentHandler::getDefaultModelFor( $title );
 458                         $defaultHandler = ContentHandler::getForModelID( $defaultModel );
 459                         $defaultFormat = $defaultHandler->getDefaultFormat();
 460
 461                         if ( $model != $defaultModel ) {
 462                                 throw new MWException( "Can't save non-default content model with "
 463                                         . "\$wgContentHandlerUseDB disabled: model is $model, "
 464                                         . "default for $name is $defaultModel"
 465                                 );
 466                         }
 467
 468                         if ( $format != $defaultFormat ) {
 469                                 throw new MWException( "Can't use non-default content format with "
 470                                         . "\$wgContentHandlerUseDB disabled: format is $format, "
 471                                         . "default for $name is $defaultFormat"
 472                                 );
 473                         }
 474                 }
 475
 476                 if ( !$content->isValid() ) {
 477                         throw new MWException(
 478                                 "New content for $name is not valid! Content model is $model"
 479                         );
 480                 }
 481         }
 482
 483         /**
 484          * Create a new null-revision for insertion into a page's
 485          * history. This will not re-save the text, but simply refer
 486          * to the text from the previous version.
 487          *
 488          * Such revisions can for instance identify page rename
 489          * operations and other such meta-modifications.
 490          *
 491          * MCR migration note: this replaces Revision::newNullRevision
 492          *
 493          * @todo Introduce newFromParentRevision(). newNullRevision can then be based on that
 494          * (or go away).
 495          *
 496          * @param IDatabase $dbw
 497          * @param Title $title Title of the page to read from
 498          * @param CommentStoreComment $comment RevisionRecord's summary
 499          * @param bool $minor Whether the revision should be considered as minor
 500          * @param User $user The user to attribute the revision to
 501          * @return RevisionRecord|null RevisionRecord or null on error
 502          */
 503         public function newNullRevision(
 504                 IDatabase $dbw,
 505                 Title $title,
 506                 CommentStoreComment $comment,
 507                 $minor,
 508                 User $user
 509         ) {
 510                 $this->checkDatabaseWikiId( $dbw );
 511
 512                 $fields = [ 'page_latest', 'page_namespace', 'page_title',
 513                         'rev_id', 'rev_text_id', 'rev_len', 'rev_sha1' ];
 514
 515                 if ( $this->contentHandlerUseDB ) {
 516                         $fields[] = 'rev_content_model';
 517                         $fields[] = 'rev_content_format';
 518                 }
 519
 520                 $current = $dbw->selectRow(
 521                         [ 'page', 'revision' ],
 522                         $fields,
 523                         [
 524                                 'page_id' => $title->getArticleID(),
 525                                 'page_latest=rev_id',
 526                         ],
 527                         __METHOD__,
 528                         [ 'FOR UPDATE' ] // T51581
 529                 );
 530
 531                 if ( $current ) {
 532                         $fields = [
 533                                 'page'       => $title->getArticleID(),
 534                                 'user_text'  => $user->getName(),
 535                                 'user'       => $user->getId(),
 536                                 'comment'    => $comment,
 537                                 'minor_edit' => $minor,
 538                                 'text_id'    => $current->rev_text_id,
 539                                 'parent_id'  => $current->page_latest,
 540                                 'len'        => $current->rev_len,
 541                                 'sha1'       => $current->rev_sha1
 542                         ];
 543
 544                         if ( $this->contentHandlerUseDB ) {
 545                                 $fields['content_model'] = $current->rev_content_model;
 546                                 $fields['content_format'] = $current->rev_content_format;
 547                         }
 548
 549                         $fields['title'] = Title::makeTitle( $current->page_namespace, $current->page_title );
 550
 551                         $mainSlot = $this->emulateMainSlot_1_29( $fields, 0, $title );
 552                         $revision = new MutableRevisionRecord( $title, $this->wikiId );
 553                         $this->initializeMutableRevisionFromArray( $revision, $fields );
 554                         $revision->setSlot( $mainSlot );
 555                 } else {
 556                         $revision = null;
 557                 }
 558
 559                 return $revision;
 560         }
 561
 562         /**
 563          * MCR migration note: this replaces Revision::isUnpatrolled
 564          *
 565          * @return int Rcid of the unpatrolled row, zero if there isn't one
 566          */
 567         public function isUnpatrolled( RevisionRecord $rev ) {
 568                 $rc = $this->getRecentChange( $rev );
 569                 if ( $rc && $rc->getAttribute( 'rc_patrolled' ) == 0 ) {
 570                         return $rc->getAttribute( 'rc_id' );
 571                 } else {
 572                         return 0;
 573                 }
 574         }
 575
 576         /**
 577          * Get the RC object belonging to the current revision, if there's one
 578          *
 579          * MCR migration note: this replaces Revision::getRecentChange
 580          *
 581          * @todo move this somewhere else?
 582          *
 583          * @param RevisionRecord $rev
 584          * @param int $flags (optional) $flags include:
 585          *      IDBAccessObject::READ_LATEST: Select the data from the master
 586          *
 587          * @return null|RecentChange
 588          */
 589         public function getRecentChange( RevisionRecord $rev, $flags = 0 ) {
 590                 $dbr = $this->getDBConnection( DB_REPLICA );
 591
 592                 list( $dbType, ) = DBAccessObjectUtils::getDBOptions( $flags );
 593
 594                 $userIdentity = $rev->getUser( RevisionRecord::RAW );
 595
 596                 if ( !$userIdentity ) {
 597                         // If the revision has no user identity, chances are it never went
 598                         // into the database, and doesn't have an RC entry.
 599                         return null;
 600                 }
 601
 602                 // TODO: Select by rc_this_oldid alone - but as of Nov 2017, there is no index on that!
 603                 $rc = RecentChange::newFromConds(
 604                         [
 605                                 'rc_user_text' => $userIdentity->getName(),
 606                                 'rc_timestamp' => $dbr->timestamp( $rev->getTimestamp() ),
 607                                 'rc_this_oldid' => $rev->getId()
 608                         ],
 609                         __METHOD__,
 610                         $dbType
 611                 );
 612
 613                 $this->releaseDBConnection( $dbr );
 614
 615                 // XXX: cache this locally? Glue it to the RevisionRecord?
 616                 return $rc;
 617         }
 618
 619         /**
 620          * Maps fields of the archive row to corresponding revision rows.
 621          *
 622          * @param object $archiveRow
 623          *
 624          * @return object a revision row object, corresponding to $archiveRow.
 625          */
 626         private static function mapArchiveFields( $archiveRow ) {
 627                 $fieldMap = [
 628                         // keep with ar prefix:
 629                         'ar_id'        => 'ar_id',
 630
 631                         // not the same suffix:
 632                         'ar_page_id'        => 'rev_page',
 633                         'ar_rev_id'         => 'rev_id',
 634
 635                         // same suffix:
 636                         'ar_text_id'        => 'rev_text_id',
 637                         'ar_timestamp'      => 'rev_timestamp',
 638                         'ar_user_text'      => 'rev_user_text',
 639                         'ar_user'           => 'rev_user',
 640                         'ar_minor_edit'     => 'rev_minor_edit',
 641                         'ar_deleted'        => 'rev_deleted',
 642                         'ar_len'            => 'rev_len',
 643                         'ar_parent_id'      => 'rev_parent_id',
 644                         'ar_sha1'           => 'rev_sha1',
 645                         'ar_comment'        => 'rev_comment',
 646                         'ar_comment_cid'    => 'rev_comment_cid',
 647                         'ar_comment_id'     => 'rev_comment_id',
 648                         'ar_comment_text'   => 'rev_comment_text',
 649                         'ar_comment_data'   => 'rev_comment_data',
 650                         'ar_comment_old'    => 'rev_comment_old',
 651                         'ar_content_format' => 'rev_content_format',
 652                         'ar_content_model'  => 'rev_content_model',
 653                 ];
 654
 655                 if ( empty( $archiveRow->ar_text_id ) ) {
 656                         $fieldMap['ar_text'] = 'old_text';
 657                         $fieldMap['ar_flags'] = 'old_flags';
 658                 }
 659
 660                 $revRow = new stdClass();
 661                 foreach ( $fieldMap as $arKey => $revKey ) {
 662                         if ( property_exists( $archiveRow, $arKey ) ) {
 663                                 $revRow->$revKey = $archiveRow->$arKey;
 664                         }
 665                 }
 666
 667                 return $revRow;
 668         }
 669
 670         /**
 671          * Constructs a RevisionRecord for the revisions main slot, based on the MW1.29 schema.
 672          *
 673          * @param object|array $row Either a database row or an array
 674          * @param int $queryFlags for callbacks
 675          * @param Title $title
 676          *
 677          * @return SlotRecord The main slot, extracted from the MW 1.29 style row.
 678          * @throws MWException
 679          */
 680         private function emulateMainSlot_1_29( $row, $queryFlags, Title $title ) {
 681                 $mainSlotRow = new stdClass();
 682                 $mainSlotRow->role_name = 'main';
 683
 684                 $content = null;
 685                 $blobData = null;
 686                 $blobFlags = '';
 687
 688                 if ( is_object( $row ) ) {
 689                         // archive row
 690                         if ( !isset( $row->rev_id ) && isset( $row->ar_user ) ) {
 691                                 $row = $this->mapArchiveFields( $row );
 692                         }
 693
 694                         if ( isset( $row->rev_text_id ) && $row->rev_text_id > 0 ) {
 695                                 $mainSlotRow->cont_address = 'tt:' . $row->rev_text_id;
 696                         } elseif ( isset( $row->ar_id ) ) {
 697                                 $mainSlotRow->cont_address = 'ar:' . $row->ar_id;
 698                         }
 699
 700                         if ( isset( $row->old_text ) ) {
 701                                 // this happens when the text-table gets joined directly, in the pre-1.30 schema
 702                                 $blobData = isset( $row->old_text ) ? strval( $row->old_text ) : null;
 703                                 $blobFlags = isset( $row->old_flags ) ? strval( $row->old_flags ) : '';
 704                         }
 705
 706                         $mainSlotRow->slot_revision = intval( $row->rev_id );
 707
 708                         $mainSlotRow->cont_size = isset( $row->rev_len ) ? intval( $row->rev_len ) : null;
 709                         $mainSlotRow->cont_sha1 = isset( $row->rev_sha1 ) ? strval( $row->rev_sha1 ) : null;
 710                         $mainSlotRow->model_name = isset( $row->rev_content_model )
 711                                 ? strval( $row->rev_content_model )
 712                                 : null;
 713                         // XXX: in the future, we'll probably always use the default format, and drop content_format
 714                         $mainSlotRow->format_name = isset( $row->rev_content_format )
 715                                 ? strval( $row->rev_content_format )
 716                                 : null;
 717                 } elseif ( is_array( $row ) ) {
 718                         $mainSlotRow->slot_revision = isset( $row['id'] ) ? intval( $row['id'] ) : null;
 719
 720                         $mainSlotRow->cont_address = isset( $row['text_id'] )
 721                                 ? 'tt:' . intval( $row['text_id'] )
 722                                 : null;
 723                         $mainSlotRow->cont_size = isset( $row['len'] ) ? intval( $row['len'] ) : null;
 724                         $mainSlotRow->cont_sha1 = isset( $row['sha1'] ) ? strval( $row['sha1'] ) : null;
 725
 726                         $mainSlotRow->model_name = isset( $row['content_model'] )
 727                                 ? strval( $row['content_model'] ) : null;  // XXX: must be a string!
 728                         // XXX: in the future, we'll probably always use the default format, and drop content_format
 729                         $mainSlotRow->format_name = isset( $row['content_format'] )
 730                                 ? strval( $row['content_format'] ) : null;
 731                         $blobData = isset( $row['text'] ) ? rtrim( strval( $row['text'] ) ) : null;
 732                         $blobFlags = isset( $row['flags'] ) ? trim( strval( $row['flags'] ) ) : '';
 733
 734                         // if we have a Content object, override mText and mContentModel
 735                         if ( !empty( $row['content'] ) ) {
 736                                 if ( !( $row['content'] instanceof Content ) ) {
 737                                         throw new MWException( 'content field must contain a Content object.' );
 738                                 }
 739
 740                                 /** @var Content $content */
 741                                 $content = $row['content'];
 742                                 $handler = $content->getContentHandler();
 743
 744                                 $mainSlotRow->model_name = $content->getModel();
 745
 746                                 // XXX: in the future, we'll probably always use the default format.
 747                                 if ( $mainSlotRow->format_name === null ) {
 748                                         $mainSlotRow->format_name = $handler->getDefaultFormat();
 749                                 }
 750                         }
 751                 } else {
 752                         throw new MWException( 'Revision constructor passed invalid row format.' );
 753                 }
 754
 755                 // With the old schema, the content changes with every revision.
 756                 // ...except for null-revisions. Would be nice if we could detect them.
 757                 $mainSlotRow->slot_inherited = 0;
 758
 759                 if ( $mainSlotRow->model_name === null ) {
 760                         $mainSlotRow->model_name = function ( SlotRecord $slot ) use ( $title ) {
 761                                 // TODO: MCR: consider slot role in getDefaultModelFor()! Use LinkTarget!
 762                                 // TODO: MCR: deprecate $title->getModel().
 763                                 return ContentHandler::getDefaultModelFor( $title );
 764                         };
 765                 }
 766
 767                 if ( !$content ) {
 768                         $content = function ( SlotRecord $slot )
 769                                 use ( $blobData, $blobFlags, $queryFlags, $mainSlotRow )
 770                         {
 771                                 return $this->loadSlotContent(
 772                                         $slot,
 773                                         $blobData,
 774                                         $blobFlags,
 775                                         $mainSlotRow->format_name,
 776                                         $queryFlags
 777                                 );
 778                         };
 779                 }
 780
 781                 return new SlotRecord( $mainSlotRow, $content );
 782         }
 783
 784         /**
 785          * Loads a Content object based on a slot row.
 786          *
 787          * This method does not call $slot->getContent(), and may be used as a callback
 788          * called by $slot->getContent().
 789          *
 790          * MCR migration note: this roughly corresponds to Revision::getContentInternal
 791          *
 792          * @param SlotRecord $slot The SlotRecord to load content for
 793          * @param string|null $blobData The content blob, in the form indicated by $blobFlags
 794          * @param string $blobFlags Flags indicating how $blobData needs to be processed
 795          * @param string|null $blobFormat MIME type indicating how $dataBlob is encoded
 796          * @param int $queryFlags
 797          *
 798          * @throw RevisionAccessException
 799          * @return Content
 800          */
 801         private function loadSlotContent(
 802                 SlotRecord $slot,
 803                 $blobData = null,
 804                 $blobFlags = '',
 805                 $blobFormat = null,
 806                 $queryFlags = 0
 807         ) {
 808                 if ( $blobData !== null ) {
 809                         Assert::parameterType( 'string', $blobData, '$blobData' );
 810                         Assert::parameterType( 'string', $blobFlags, '$blobFlags' );
 811
 812                         $cacheKey = $slot->hasAddress() ? $slot->getAddress() : null;
 813
 814                         $data = $this->blobStore->expandBlob( $blobData, $blobFlags, $cacheKey );
 815
 816                         if ( $data === false ) {
 817                                 throw new RevisionAccessException(
 818                                         "Failed to expand blob data using flags $blobFlags (key: $cacheKey)"
 819                                 );
 820                         }
 821                 } else {
 822                         $address = $slot->getAddress();
 823                         try {
 824                                 $data = $this->blobStore->getBlob( $address, $queryFlags );
 825                         } catch ( BlobAccessException $e ) {
 826                                 throw new RevisionAccessException(
 827                                         "Failed to load data blob from $address: " . $e->getMessage(), 0, $e
 828                                 );
 829                         }
 830                 }
 831
 832                 // Unserialize content
 833                 $handler = ContentHandler::getForModelID( $slot->getModel() );
 834
 835                 $content = $handler->unserializeContent( $data, $blobFormat );
 836                 return $content;
 837         }
 838
 839         /**
 840          * Load a page revision from a given revision ID number.
 841          * Returns null if no such revision can be found.
 842          *
 843          * MCR migration note: this replaces Revision::newFromId
 844          *
 845          * $flags include:
 846          *      IDBAccessObject::READ_LATEST: Select the data from the master
 847          *      IDBAccessObject::READ_LOCKING : Select & lock the data from the master
 848          *
 849          * @param int $id
 850          * @param int $flags (optional)
 851          * @return RevisionRecord|null
 852          */
 853         public function getRevisionById( $id, $flags = 0 ) {
 854                 return $this->newRevisionFromConds( [ 'rev_id' => intval( $id ) ], $flags );
 855         }
 856
 857         /**
 858          * Load either the current, or a specified, revision
 859          * that's attached to a given link target. If not attached
 860          * to that link target, will return null.
 861          *
 862          * MCR migration note: this replaces Revision::newFromTitle
 863          *
 864          * $flags include:
 865          *      IDBAccessObject::READ_LATEST: Select the data from the master
 866          *      IDBAccessObject::READ_LOCKING : Select & lock the data from the master
 867          *
 868          * @param LinkTarget $linkTarget
 869          * @param int $revId (optional)
 870          * @param int $flags Bitfield (optional)
 871          * @return RevisionRecord|null
 872          */
 873         public function getRevisionByTitle( LinkTarget $linkTarget, $revId = 0, $flags = 0 ) {
 874                 $conds = [
 875                         'page_namespace' => $linkTarget->getNamespace(),
 876                         'page_title' => $linkTarget->getDBkey()
 877                 ];
 878                 if ( $revId ) {
 879                         // Use the specified revision ID.
 880                         // Note that we use newRevisionFromConds here because we want to retry
 881                         // and fall back to master if the page is not found on a replica.
 882                         // Since the caller supplied a revision ID, we are pretty sure the revision is
 883                         // supposed to exist, so we should try hard to find it.
 884                         $conds['rev_id'] = $revId;
 885                         return $this->newRevisionFromConds( $conds, $flags );
 886                 } else {
 887                         // Use a join to get the latest revision.
 888                         // Note that we don't use newRevisionFromConds here because we don't want to retry
 889                         // and fall back to master. The assumption is that we only want to force the fallback
 890                         // if we are quite sure the revision exists because the caller supplied a revision ID.
 891                         // If the page isn't found at all on a replica, it probably simply does not exist.
 892                         $db = $this->getDBConnection( ( $flags & self::READ_LATEST ) ? DB_MASTER : DB_REPLICA );
 893
 894                         $conds[] = 'rev_id=page_latest';
 895                         $rev = $this->loadRevisionFromConds( $db, $conds, $flags );
 896
 897                         $this->releaseDBConnection( $db );
 898                         return $rev;
 899                 }
 900         }
 901
 902         /**
 903          * Load either the current, or a specified, revision
 904          * that's attached to a given page ID.
 905          * Returns null if no such revision can be found.
 906          *
 907          * MCR migration note: this replaces Revision::newFromPageId
 908          *
 909          * $flags include:
 910          *      IDBAccessObject::READ_LATEST: Select the data from the master (since 1.20)
 911          *      IDBAccessObject::READ_LOCKING : Select & lock the data from the master
 912          *
 913          * @param int $pageId
 914          * @param int $revId (optional)
 915          * @param int $flags Bitfield (optional)
 916          * @return RevisionRecord|null
 917          */
 918         public function getRevisionByPageId( $pageId, $revId = 0, $flags = 0 ) {
 919                 $conds = [ 'page_id' => $pageId ];
 920                 if ( $revId ) {
 921                         // Use the specified revision ID.
 922                         // Note that we use newRevisionFromConds here because we want to retry
 923                         // and fall back to master if the page is not found on a replica.
 924                         // Since the caller supplied a revision ID, we are pretty sure the revision is
 925                         // supposed to exist, so we should try hard to find it.
 926                         $conds['rev_id'] = $revId;
 927                         return $this->newRevisionFromConds( $conds, $flags );
 928                 } else {
 929                         // Use a join to get the latest revision.
 930                         // Note that we don't use newRevisionFromConds here because we don't want to retry
 931                         // and fall back to master. The assumption is that we only want to force the fallback
 932                         // if we are quite sure the revision exists because the caller supplied a revision ID.
 933                         // If the page isn't found at all on a replica, it probably simply does not exist.
 934                         $db = $this->getDBConnection( ( $flags & self::READ_LATEST ) ? DB_MASTER : DB_REPLICA );
 935
 936                         $conds[] = 'rev_id=page_latest';
 937                         $rev = $this->loadRevisionFromConds( $db, $conds, $flags );
 938
 939                         $this->releaseDBConnection( $db );
 940                         return $rev;
 941                 }
 942         }
 943
 944         /**
 945          * Load the revision for the given title with the given timestamp.
 946          * WARNING: Timestamps may in some circumstances not be unique,
 947          * so this isn't the best key to use.
 948          *
 949          * MCR migration note: this replaces Revision::loadFromTimestamp
 950          *
 951          * @param Title $title
 952          * @param string $timestamp
 953          * @return RevisionRecord|null
 954          */
 955         public function getRevisionFromTimestamp( $title, $timestamp ) {
 956                 return $this->newRevisionFromConds(
 957                         [
 958                                 'rev_timestamp' => $timestamp,
 959                                 'page_namespace' => $title->getNamespace(),
 960                                 'page_title' => $title->getDBkey()
 961                         ],
 962                         0,
 963                         $title
 964                 );
 965         }
 966
 967         /**
 968          * Make a fake revision object from an archive table row. This is queried
 969          * for permissions or even inserted (as in Special:Undelete)
 970          *
 971          * MCR migration note: this replaces Revision::newFromArchiveRow
 972          *
 973          * @param object $row
 974          * @param int $queryFlags
 975          * @param Title|null $title
 976          * @param array $overrides associative array with fields of $row to override. This may be
 977          *   used e.g. to force the parent revision ID or page ID. Keys in the array are fields
 978          *   names from the archive table without the 'ar_' prefix, i.e. use 'parent_id' to
 979          *   override ar_parent_id.
 980          *
 981          * @return RevisionRecord
 982          * @throws MWException
 983          */
 984         public function newRevisionFromArchiveRow(
 985                 $row,
 986                 $queryFlags = 0,
 987                 Title $title = null,
 988                 array $overrides = []
 989         ) {
 990                 Assert::parameterType( 'object', $row, '$row' );
 991
 992                 // check second argument, since Revision::newFromArchiveRow had $overrides in that spot.
 993                 Assert::parameterType( 'integer', $queryFlags, '$queryFlags' );
 994
 995                 if ( !$title && isset( $overrides['title'] ) ) {
 996                         if ( !( $overrides['title'] instanceof Title ) ) {
 997                                 throw new MWException( 'title field override must contain a Title object.' );
 998                         }
 999
1000                         $title = $overrides['title'];
1001                 }
1002
1003                 if ( !isset( $title ) ) {
1004                         if ( isset( $row->ar_namespace ) && isset( $row->ar_title ) ) {
1005                                 $title = Title::makeTitle( $row->ar_namespace, $row->ar_title );
1006                         } else {
1007                                 throw new InvalidArgumentException(
1008                                         'A Title or ar_namespace and ar_title must be given'
1009                                 );
1010                         }
1011                 }
1012
1013                 foreach ( $overrides as $key => $value ) {
1014                         $field = "ar_$key";
1015                         $row->$field = $value;
1016                 }
1017
1018                 $user = $this->getUserIdentityFromRowObject( $row, 'ar_' );
1019
1020                 $comment = CommentStore::newKey( 'ar_comment' )
1021                         // Legacy because $row may have come from self::selectFields()
1022                         ->getCommentLegacy( $this->getDBConnection( DB_REPLICA ), $row, true );
1023
1024                 $mainSlot = $this->emulateMainSlot_1_29( $row, $queryFlags, $title );
1025                 $slots = new RevisionSlots( [ 'main' => $mainSlot ] );
1026
1027                 return new RevisionArchiveRecord( $title, $user, $comment, $row, $slots, $this->wikiId );
1028         }
1029
1030         /**
1031          * @param object $row
1032          * @param string $prefix Field prefix, such as 'rev_' or 'ar_'.
1033          *
1034          * @return UserIdentityValue
1035          */
1036         private function getUserIdentityFromRowObject( $row, $prefix = 'rev_' ) {
1037                 $idField = "{$prefix}user";
1038                 $nameField = "{$prefix}user_text";
1039
1040                 $userId = intval( $row->$idField );
1041
1042                 if ( isset( $row->user_name ) ) {
1043                         $userName = $row->user_name;
1044                 } elseif ( isset( $row->$nameField ) ) {
1045                         $userName = $row->$nameField;
1046                 } else {
1047                         $userName = User::whoIs( $userId );
1048                 }
1049
1050                 if ( $userName === false ) {
1051                         wfWarn( __METHOD__ . ': Cannot determine user name for user ID ' . $userId );
1052                         $userName = '';
1053                 }
1054
1055                 return new UserIdentityValue( $userId, $userName );
1056         }
1057
1058         /**
1059          * @see RevisionFactory::newRevisionFromRow_1_29
1060          *
1061          * MCR migration note: this replaces Revision::newFromRow
1062          *
1063          * @param object $row
1064          * @param int $queryFlags
1065          * @param Title|null $title
1066          *
1067          * @return RevisionRecord
1068          * @throws MWException
1069          * @throws RevisionAccessException
1070          */
1071         private function newRevisionFromRow_1_29( $row, $queryFlags = 0, Title $title = null ) {
1072                 Assert::parameterType( 'object', $row, '$row' );
1073
1074                 if ( !$title ) {
1075                         $pageId = isset( $row->rev_page ) ? $row->rev_page : 0; // XXX: also check page_id?
1076                         $revId = isset( $row->rev_id ) ? $row->rev_id : 0;
1077
1078                         $title = $this->getTitle( $pageId, $revId, $queryFlags );
1079                 }
1080
1081                 if ( !isset( $row->page_latest ) ) {
1082                         $row->page_latest = $title->getLatestRevID();
1083                         if ( $row->page_latest === 0 && $title->exists() ) {
1084                                 wfWarn( 'Encountered title object in limbo: ID ' . $title->getArticleID() );
1085                         }
1086                 }
1087
1088                 $user = $this->getUserIdentityFromRowObject( $row );
1089
1090                 $comment = CommentStore::newKey( 'rev_comment' )
1091                         // Legacy because $row may have come from self::selectFields()
1092                         ->getCommentLegacy( $this->getDBConnection( DB_REPLICA ), $row, true );
1093
1094                 $mainSlot = $this->emulateMainSlot_1_29( $row, $queryFlags, $title );
1095                 $slots = new RevisionSlots( [ 'main' => $mainSlot ] );
1096
1097                 return new RevisionStoreRecord( $title, $user, $comment, $row, $slots, $this->wikiId );
1098         }
1099
1100         /**
1101          * @see RevisionFactory::newRevisionFromRow
1102          *
1103          * MCR migration note: this replaces Revision::newFromRow
1104          *
1105          * @param object $row
1106          * @param int $queryFlags
1107          * @param Title|null $title
1108          *
1109          * @return RevisionRecord
1110          */
1111         public function newRevisionFromRow( $row, $queryFlags = 0, Title $title = null ) {
1112                 return $this->newRevisionFromRow_1_29( $row, $queryFlags, $title );
1113         }
1114
1115         /**
1116          * Constructs a new MutableRevisionRecord based on the given associative array following
1117          * the MW1.29 convention for the Revision constructor.
1118          *
1119          * MCR migration note: this replaces Revision::newFromRow
1120          *
1121          * @param array $fields
1122          * @param int $queryFlags
1123          * @param Title|null $title
1124          *
1125          * @return MutableRevisionRecord
1126          * @throws MWException
1127          * @throws RevisionAccessException
1128          */
1129         public function newMutableRevisionFromArray(
1130                 array $fields,
1131                 $queryFlags = 0,
1132                 Title $title = null
1133         ) {
1134                 if ( !$title && isset( $fields['title'] ) ) {
1135                         if ( !( $fields['title'] instanceof Title ) ) {
1136                                 throw new MWException( 'title field must contain a Title object.' );
1137                         }
1138
1139                         $title = $fields['title'];
1140                 }
1141
1142                 if ( !$title ) {
1143                         $pageId = isset( $fields['page'] ) ? $fields['page'] : 0;
1144                         $revId = isset( $fields['id'] ) ? $fields['id'] : 0;
1145
1146                         $title = $this->getTitle( $pageId, $revId, $queryFlags );
1147                 }
1148
1149                 if ( !isset( $fields['page'] ) ) {
1150                         $fields['page'] = $title->getArticleID( $queryFlags );
1151                 }
1152
1153                 // if we have a content object, use it to set the model and type
1154                 if ( !empty( $fields['content'] ) ) {
1155                         if ( !( $fields['content'] instanceof Content ) ) {
1156                                 throw new MWException( 'content field must contain a Content object.' );
1157                         }
1158
1159                         if ( !empty( $fields['text_id'] ) ) {
1160                                 throw new MWException(
1161                                         "Text already stored in external store (id {$fields['text_id']}), " .
1162                                         "can't serialize content object"
1163                                 );
1164                         }
1165                 }
1166
1167                 // Replaces old lazy loading logic in Revision::getUserText.
1168                 if ( !isset( $fields['user_text'] ) && isset( $fields['user'] ) ) {
1169                         if ( $fields['user'] instanceof UserIdentity ) {
1170                                 /** @var User $user */
1171                                 $user = $fields['user'];
1172                                 $fields['user_text'] = $user->getName();
1173                                 $fields['user'] = $user->getId();
1174                         } else {
1175                                 // TODO: wrap this in a callback to make it lazy again.
1176                                 $name = $fields['user'] === 0 ? false : User::whoIs( $fields['user'] );
1177
1178                                 if ( $name === false ) {
1179                                         throw new MWException(
1180                                                 'user_text not given, and unknown user ID ' . $fields['user']
1181                                         );
1182                                 }
1183
1184                                 $fields['user_text'] = $name;
1185                         }
1186                 }
1187
1188                 if (
1189                         isset( $fields['comment'] )
1190                         && !( $fields['comment'] instanceof CommentStoreComment )
1191                 ) {
1192                         $commentData = isset( $fields['comment_data'] ) ? $fields['comment_data'] : null;
1193
1194                         if ( $fields['comment'] instanceof Message ) {
1195                                 $fields['comment'] = CommentStoreComment::newUnsavedComment(
1196                                         $fields['comment'],
1197                                         $commentData
1198                                 );
1199                         } else {
1200                                 $commentText = trim( strval( $fields['comment'] ) );
1201                                 $fields['comment'] = CommentStoreComment::newUnsavedComment(
1202                                         $commentText,
1203                                         $commentData
1204                                 );
1205                         }
1206                 }
1207
1208                 $mainSlot = $this->emulateMainSlot_1_29( $fields, $queryFlags, $title );
1209
1210                 $revision = new MutableRevisionRecord( $title, $this->wikiId );
1211                 $this->initializeMutableRevisionFromArray( $revision, $fields );
1212                 $revision->setSlot( $mainSlot );
1213
1214                 return $revision;
1215         }
1216
1217         /**
1218          * @param MutableRevisionRecord $record
1219          * @param array $fields
1220          */
1221         private function initializeMutableRevisionFromArray(
1222                 MutableRevisionRecord $record,
1223                 array $fields
1224         ) {
1225                 /** @var UserIdentity $user */
1226                 $user = null;
1227
1228                 if ( isset( $fields['user'] ) && ( $fields['user'] instanceof UserIdentity ) ) {
1229                         $user = $fields['user'];
1230                 } elseif ( isset( $fields['user'] ) && isset( $fields['user_text'] ) ) {
1231                         $user = new UserIdentityValue( intval( $fields['user'] ), $fields['user_text'] );
1232                 } elseif ( isset( $fields['user'] ) ) {
1233                         $user = User::newFromId( intval( $fields['user'] ) );
1234                 } elseif ( isset( $fields['user_text'] ) ) {
1235                         $user = User::newFromName( $fields['user_text'] );
1236
1237                         // User::newFromName will return false for IP addresses (and invalid names)
1238                         if ( $user == false ) {
1239                                 $user = new UserIdentityValue( 0, $fields['user_text'] );
1240                         }
1241                 }
1242
1243                 if ( $user ) {
1244                         $record->setUser( $user );
1245                 }
1246
1247                 $timestamp = isset( $fields['timestamp'] )
1248                         ? strval( $fields['timestamp'] )
1249                         : wfTimestampNow(); // TODO: use a callback, so we can override it for testing.
1250
1251                 $record->setTimestamp( $timestamp );
1252
1253                 if ( isset( $fields['page'] ) ) {
1254                         $record->setPageId( intval( $fields['page'] ) );
1255                 }
1256
1257                 if ( isset( $fields['id'] ) ) {
1258                         $record->setId( intval( $fields['id'] ) );
1259                 }
1260                 if ( isset( $fields['parent_id'] ) ) {
1261                         $record->setParentId( intval( $fields['parent_id'] ) );
1262                 }
1263
1264                 if ( isset( $fields['sha1'] ) ) {
1265                         $record->setSha1( $fields['sha1'] );
1266                 }
1267                 if ( isset( $fields['size'] ) ) {
1268                         $record->setSize( intval( $fields['size'] ) );
1269                 }
1270
1271                 if ( isset( $fields['minor_edit'] ) ) {
1272                         $record->setMinorEdit( intval( $fields['minor_edit'] ) !== 0 );
1273                 }
1274                 if ( isset( $fields['deleted'] ) ) {
1275                         $record->setVisibility( intval( $fields['deleted'] ) );
1276                 }
1277
1278                 if ( isset( $fields['comment'] ) ) {
1279                         Assert::parameterType(
1280                                 CommentStoreComment::class,
1281                                 $fields['comment'],
1282                                 '$row[\'comment\']'
1283                         );
1284                         $record->setComment( $fields['comment'] );
1285                 }
1286         }
1287
1288         /**
1289          * Load a page revision from a given revision ID number.
1290          * Returns null if no such revision can be found.
1291          *
1292          * MCR migration note: this corresponds to Revision::loadFromId
1293          *
1294          * @note direct use is deprecated!
1295          * @todo remove when unused! there seem to be no callers of Revision::loadFromId
1296          *
1297          * @param IDatabase $db
1298          * @param int $id
1299          *
1300          * @return RevisionRecord|null
1301          */
1302         public function loadRevisionFromId( IDatabase $db, $id ) {
1303                 return $this->loadRevisionFromConds( $db, [ 'rev_id' => intval( $id ) ] );
1304         }
1305
1306         /**
1307          * Load either the current, or a specified, revision
1308          * that's attached to a given page. If not attached
1309          * to that page, will return null.
1310          *
1311          * MCR migration note: this replaces Revision::loadFromPageId
1312          *
1313          * @note direct use is deprecated!
1314          * @todo remove when unused!
1315          *
1316          * @param IDatabase $db
1317          * @param int $pageid
1318          * @param int $id
1319          * @return RevisionRecord|null
1320          */
1321         public function loadRevisionFromPageId( IDatabase $db, $pageid, $id = 0 ) {
1322                 $conds = [ 'rev_page' => intval( $pageid ), 'page_id' => intval( $pageid ) ];
1323                 if ( $id ) {
1324                         $conds['rev_id'] = intval( $id );
1325                 } else {
1326                         $conds[] = 'rev_id=page_latest';
1327                 }
1328                 return $this->loadRevisionFromConds( $db, $conds );
1329         }
1330
1331         /**
1332          * Load either the current, or a specified, revision
1333          * that's attached to a given page. If not attached
1334          * to that page, will return null.
1335          *
1336          * MCR migration note: this replaces Revision::loadFromTitle
1337          *
1338          * @note direct use is deprecated!
1339          * @todo remove when unused!
1340          *
1341          * @param IDatabase $db
1342          * @param Title $title
1343          * @param int $id
1344          *
1345          * @return RevisionRecord|null
1346          */
1347         public function loadRevisionFromTitle( IDatabase $db, $title, $id = 0 ) {
1348                 if ( $id ) {
1349                         $matchId = intval( $id );
1350                 } else {
1351                         $matchId = 'page_latest';
1352                 }
1353
1354                 return $this->loadRevisionFromConds(
1355                         $db,
1356                         [
1357                                 "rev_id=$matchId",
1358                                 'page_namespace' => $title->getNamespace(),
1359                                 'page_title' => $title->getDBkey()
1360                         ],
1361                         0,
1362                         $title
1363                 );
1364         }
1365
1366         /**
1367          * Load the revision for the given title with the given timestamp.
1368          * WARNING: Timestamps may in some circumstances not be unique,
1369          * so this isn't the best key to use.
1370          *
1371          * MCR migration note: this replaces Revision::loadFromTimestamp
1372          *
1373          * @note direct use is deprecated! Use getRevisionFromTimestamp instead!
1374          * @todo remove when unused!
1375          *
1376          * @param IDatabase $db
1377          * @param Title $title
1378          * @param string $timestamp
1379          * @return RevisionRecord|null
1380          */
1381         public function loadRevisionFromTimestamp( IDatabase $db, $title, $timestamp ) {
1382                 return $this->loadRevisionFromConds( $db,
1383                         [
1384                                 'rev_timestamp' => $db->timestamp( $timestamp ),
1385                                 'page_namespace' => $title->getNamespace(),
1386                                 'page_title' => $title->getDBkey()
1387                         ],
1388                         0,
1389                         $title
1390                 );
1391         }
1392
1393         /**
1394          * Given a set of conditions, fetch a revision
1395          *
1396          * This method should be used if we are pretty sure the revision exists.
1397          * Unless $flags has READ_LATEST set, this method will first try to find the revision
1398          * on a replica before hitting the master database.
1399          *
1400          * MCR migration note: this corresponds to Revision::newFromConds
1401          *
1402          * @param array $conditions
1403          * @param int $flags (optional)
1404          * @param Title $title
1405          *
1406          * @return RevisionRecord|null
1407          */
1408         private function newRevisionFromConds( $conditions, $flags = 0, Title $title = null ) {
1409                 $db = $this->getDBConnection( ( $flags & self::READ_LATEST ) ? DB_MASTER : DB_REPLICA );
1410                 $rev = $this->loadRevisionFromConds( $db, $conditions, $flags, $title );
1411                 $this->releaseDBConnection( $db );
1412
1413                 $lb = $this->getDBLoadBalancer();
1414
1415                 // Make sure new pending/committed revision are visibile later on
1416                 // within web requests to certain avoid bugs like T93866 and T94407.
1417                 if ( !$rev
1418                         && !( $flags & self::READ_LATEST )
1419                         && $lb->getServerCount() > 1
1420                         && $lb->hasOrMadeRecentMasterChanges()
1421                 ) {
1422                         $flags = self::READ_LATEST;
1423                         $db = $this->getDBConnection( DB_MASTER );
1424                         $rev = $this->loadRevisionFromConds( $db, $conditions, $flags, $title );
1425                         $this->releaseDBConnection( $db );
1426                 }
1427
1428                 return $rev;
1429         }
1430
1431         /**
1432          * Given a set of conditions, fetch a revision from
1433          * the given database connection.
1434          *
1435          * MCR migration note: this corresponds to Revision::loadFromConds
1436          *
1437          * @param IDatabase $db
1438          * @param array $conditions
1439          * @param int $flags (optional)
1440          * @param Title $title
1441          *
1442          * @return RevisionRecord|null
1443          */
1444         private function loadRevisionFromConds(
1445                 IDatabase $db,
1446                 $conditions,
1447                 $flags = 0,
1448                 Title $title = null
1449         ) {
1450                 $row = $this->fetchRevisionRowFromConds( $db, $conditions, $flags );
1451                 if ( $row ) {
1452                         $rev = $this->newRevisionFromRow( $row, $flags, $title );
1453
1454                         return $rev;
1455                 }
1456
1457                 return null;
1458         }
1459
1460         /**
1461          * Throws an exception if the given database connection does not belong to the wiki this
1462          * RevisionStore is bound to.
1463          *
1464          * @param IDatabase $db
1465          * @throws MWException
1466          */
1467         private function checkDatabaseWikiId( IDatabase $db ) {
1468                 $storeWiki = $this->wikiId;
1469                 $dbWiki = $db->getDomainID();
1470
1471                 if ( $dbWiki === $storeWiki ) {
1472                         return;
1473                 }
1474
1475                 // XXX: we really want the default database ID...
1476                 $storeWiki = $storeWiki ?: wfWikiID();
1477                 $dbWiki = $dbWiki ?: wfWikiID();
1478
1479                 if ( $dbWiki === $storeWiki ) {
1480                         return;
1481                 }
1482
1483                 // HACK: counteract encoding imposed by DatabaseDomain
1484                 $storeWiki = str_replace( '?h', '-', $storeWiki );
1485                 $dbWiki = str_replace( '?h', '-', $dbWiki );
1486
1487                 if ( $dbWiki === $storeWiki ) {
1488                         return;
1489                 }
1490
1491                 throw new MWException( "RevisionStore for $storeWiki "
1492                         . "cannot be used with a DB connection for $dbWiki" );
1493         }
1494
1495         /**
1496          * Given a set of conditions, return a row with the
1497          * fields necessary to build RevisionRecord objects.
1498          *
1499          * MCR migration note: this corresponds to Revision::fetchFromConds
1500          *
1501          * @param IDatabase $db
1502          * @param array $conditions
1503          * @param int $flags (optional)
1504          *
1505          * @return object|false data row as a raw object
1506          */
1507         private function fetchRevisionRowFromConds( IDatabase $db, $conditions, $flags = 0 ) {
1508                 $this->checkDatabaseWikiId( $db );
1509
1510                 $revQuery = self::getQueryInfo( [ 'page', 'user' ] );
1511                 $options = [];
1512                 if ( ( $flags & self::READ_LOCKING ) == self::READ_LOCKING ) {
1513                         $options[] = 'FOR UPDATE';
1514                 }
1515                 return $db->selectRow(
1516                         $revQuery['tables'],
1517                         $revQuery['fields'],
1518                         $conditions,
1519                         __METHOD__,
1520                         $options,
1521                         $revQuery['joins']
1522                 );
1523         }
1524
1525         /**
1526          * Return the tables, fields, and join conditions to be selected to create
1527          * a new revision object.
1528          *
1529          * MCR migration note: this replaces Revision::getQueryInfo
1530          *
1531          * @since 1.31
1532          *
1533          * @param array $options Any combination of the following strings
1534          *  - 'page': Join with the page table, and select fields to identify the page
1535          *  - 'user': Join with the user table, and select the user name
1536          *  - 'text': Join with the text table, and select fields to load page text
1537          *
1538          * @return array With three keys:
1539          *  - tables: (string[]) to include in the `$table` to `IDatabase->select()`
1540          *  - fields: (string[]) to include in the `$vars` to `IDatabase->select()`
1541          *  - joins: (array) to include in the `$join_conds` to `IDatabase->select()`
1542          */
1543         public function getQueryInfo( $options = [] ) {
1544                 $ret = [
1545                         'tables' => [],
1546                         'fields' => [],
1547                         'joins'  => [],
1548                 ];
1549
1550                 $ret['tables'][] = 'revision';
1551                 $ret['fields'] = array_merge( $ret['fields'], [
1552                         'rev_id',
1553                         'rev_page',
1554                         'rev_text_id',
1555                         'rev_timestamp',
1556                         'rev_user_text',
1557                         'rev_user',
1558                         'rev_minor_edit',
1559                         'rev_deleted',
1560                         'rev_len',
1561                         'rev_parent_id',
1562                         'rev_sha1',
1563                 ] );
1564
1565                 $commentQuery = CommentStore::newKey( 'rev_comment' )->getJoin();
1566                 $ret['tables'] = array_merge( $ret['tables'], $commentQuery['tables'] );
1567                 $ret['fields'] = array_merge( $ret['fields'], $commentQuery['fields'] );
1568                 $ret['joins'] = array_merge( $ret['joins'], $commentQuery['joins'] );
1569
1570                 if ( $this->contentHandlerUseDB ) {
1571                         $ret['fields'][] = 'rev_content_format';
1572                         $ret['fields'][] = 'rev_content_model';
1573                 }
1574
1575                 if ( in_array( 'page', $options, true ) ) {
1576                         $ret['tables'][] = 'page';
1577                         $ret['fields'] = array_merge( $ret['fields'], [
1578                                 'page_namespace',
1579                                 'page_title',
1580                                 'page_id',
1581                                 'page_latest',
1582                                 'page_is_redirect',
1583                                 'page_len',
1584                         ] );
1585                         $ret['joins']['page'] = [ 'INNER JOIN', [ 'page_id = rev_page' ] ];
1586                 }
1587
1588                 if ( in_array( 'user', $options, true ) ) {
1589                         $ret['tables'][] = 'user';
1590                         $ret['fields'] = array_merge( $ret['fields'], [
1591                                 'user_name',
1592                         ] );
1593                         $ret['joins']['user'] = [ 'LEFT JOIN', [ 'rev_user != 0', 'user_id = rev_user' ] ];
1594                 }
1595
1596                 if ( in_array( 'text', $options, true ) ) {
1597                         $ret['tables'][] = 'text';
1598                         $ret['fields'] = array_merge( $ret['fields'], [
1599                                 'old_text',
1600                                 'old_flags'
1601                         ] );
1602                         $ret['joins']['text'] = [ 'INNER JOIN', [ 'rev_text_id=old_id' ] ];
1603                 }
1604
1605                 return $ret;
1606         }
1607
1608         /**
1609          * Return the tables, fields, and join conditions to be selected to create
1610          * a new archived revision object.
1611          *
1612          * MCR migration note: this replaces Revision::getArchiveQueryInfo
1613          *
1614          * @since 1.31
1615          *
1616          * @return array With three keys:
1617          *   - tables: (string[]) to include in the `$table` to `IDatabase->select()`
1618          *   - fields: (string[]) to include in the `$vars` to `IDatabase->select()`
1619          *   - joins: (array) to include in the `$join_conds` to `IDatabase->select()`
1620          */
1621         public function getArchiveQueryInfo() {
1622                 $commentQuery = CommentStore::newKey( 'ar_comment' )->getJoin();
1623                 $ret = [
1624                         'tables' => [ 'archive' ] + $commentQuery['tables'],
1625                         'fields' => [
1626                                         'ar_id',
1627                                         'ar_page_id',
1628                                         'ar_namespace',
1629                                         'ar_title',
1630                                         'ar_rev_id',
1631                                         'ar_text',
1632                                         'ar_text_id',
1633                                         'ar_timestamp',
1634                                         'ar_user_text',
1635                                         'ar_user',
1636                                         'ar_minor_edit',
1637                                         'ar_deleted',
1638                                         'ar_len',
1639                                         'ar_parent_id',
1640                                         'ar_sha1',
1641                                 ] + $commentQuery['fields'],
1642                         'joins' => $commentQuery['joins'],
1643                 ];
1644
1645                 if ( $this->contentHandlerUseDB ) {
1646                         $ret['fields'][] = 'ar_content_format';
1647                         $ret['fields'][] = 'ar_content_model';
1648                 }
1649
1650                 return $ret;
1651         }
1652
1653         /**
1654          * Do a batched query for the sizes of a set of revisions.
1655          *
1656          * MCR migration note: this replaces Revision::getParentLengths
1657          *
1658          * @param IDatabase $db
1659          * @param int[] $revIds
1660          * @return int[] associative array mapping revision IDs from $revIds to the nominal size
1661          *         of the corresponding revision.
1662          */
1663         public function listRevisionSizes( IDatabase $db, array $revIds ) {
1664                 $this->checkDatabaseWikiId( $db );
1665
1666                 $revLens = [];
1667                 if ( !$revIds ) {
1668                         return $revLens; // empty
1669                 }
1670
1671                 $res = $db->select(
1672                         'revision',
1673                         [ 'rev_id', 'rev_len' ],
1674                         [ 'rev_id' => $revIds ],
1675                         __METHOD__
1676                 );
1677
1678                 foreach ( $res as $row ) {
1679                         $revLens[$row->rev_id] = intval( $row->rev_len );
1680                 }
1681
1682                 return $revLens;
1683         }
1684
1685         /**
1686          * Get previous revision for this title
1687          *
1688          * MCR migration note: this replaces Revision::getPrevious
1689          *
1690          * @param RevisionRecord $rev
1691          *
1692          * @return RevisionRecord|null
1693          */
1694         public function getPreviousRevision( RevisionRecord $rev ) {
1695                 $title = $this->getTitle( $rev->getPageId(), $rev->getId() );
1696                 $prev = $title->getPreviousRevisionID( $rev->getId() );
1697                 if ( $prev ) {
1698                         return $this->getRevisionByTitle( $title, $prev );
1699                 }
1700                 return null;
1701         }
1702
1703         /**
1704          * Get next revision for this title
1705          *
1706          * MCR migration note: this replaces Revision::getNext
1707          *
1708          * @param RevisionRecord $rev
1709          *
1710          * @return RevisionRecord|null
1711          */
1712         public function getNextRevision( RevisionRecord $rev ) {
1713                 $title = $this->getTitle( $rev->getPageId(), $rev->getId() );
1714                 $next = $title->getNextRevisionID( $rev->getId() );
1715                 if ( $next ) {
1716                         return $this->getRevisionByTitle( $title, $next );
1717                 }
1718                 return null;
1719         }
1720
1721         /**
1722          * Get previous revision Id for this page_id
1723          * This is used to populate rev_parent_id on save
1724          *
1725          * MCR migration note: this corresponds to Revision::getPreviousRevisionId
1726          *
1727          * @param IDatabase $db
1728          * @param RevisionRecord $rev
1729          *
1730          * @return int
1731          */
1732         private function getPreviousRevisionId( IDatabase $db, RevisionRecord $rev ) {
1733                 $this->checkDatabaseWikiId( $db );
1734
1735                 if ( $rev->getPageId() === null ) {
1736                         return 0;
1737                 }
1738                 # Use page_latest if ID is not given
1739                 if ( !$rev->getId() ) {
1740                         $prevId = $db->selectField(
1741                                 'page', 'page_latest',
1742                                 [ 'page_id' => $rev->getPageId() ],
1743                                 __METHOD__
1744                         );
1745                 } else {
1746                         $prevId = $db->selectField(
1747                                 'revision', 'rev_id',
1748                                 [ 'rev_page' => $rev->getPageId(), 'rev_id < ' . $rev->getId() ],
1749                                 __METHOD__,
1750                                 [ 'ORDER BY' => 'rev_id DESC' ]
1751                         );
1752                 }
1753                 return intval( $prevId );
1754         }
1755
1756         /**
1757          * Get rev_timestamp from rev_id, without loading the rest of the row
1758          *
1759          * MCR migration note: this replaces Revision::getTimestampFromId
1760          *
1761          * @param Title $title
1762          * @param int $id
1763          * @param int $flags
1764          * @return string|bool False if not found
1765          */
1766         public function getTimestampFromId( $title, $id, $flags = 0 ) {
1767                 $db = $this->getDBConnection(
1768                         ( $flags & IDBAccessObject::READ_LATEST ) ? DB_MASTER : DB_REPLICA
1769                 );
1770
1771                 $conds = [ 'rev_id' => $id ];
1772                 $conds['rev_page'] = $title->getArticleID();
1773                 $timestamp = $db->selectField( 'revision', 'rev_timestamp', $conds, __METHOD__ );
1774
1775                 $this->releaseDBConnection( $db );
1776                 return ( $timestamp !== false ) ? wfTimestamp( TS_MW, $timestamp ) : false;
1777         }
1778
1779         /**
1780          * Get count of revisions per page...not very efficient
1781          *
1782          * MCR migration note: this replaces Revision::countByPageId
1783          *
1784          * @param IDatabase $db
1785          * @param int $id Page id
1786          * @return int
1787          */
1788         public function countRevisionsByPageId( IDatabase $db, $id ) {
1789                 $this->checkDatabaseWikiId( $db );
1790
1791                 $row = $db->selectRow( 'revision',
1792                         [ 'revCount' => 'COUNT(*)' ],
1793                         [ 'rev_page' => $id ],
1794                         __METHOD__
1795                 );
1796                 if ( $row ) {
1797                         return intval( $row->revCount );
1798                 }
1799                 return 0;
1800         }
1801
1802         /**
1803          * Get count of revisions per page...not very efficient
1804          *
1805          * MCR migration note: this replaces Revision::countByTitle
1806          *
1807          * @param IDatabase $db
1808          * @param Title $title
1809          * @return int
1810          */
1811         public function countRevisionsByTitle( IDatabase $db, $title ) {
1812                 $id = $title->getArticleID();
1813                 if ( $id ) {
1814                         return $this->countRevisionsByPageId( $db, $id );
1815                 }
1816                 return 0;
1817         }
1818
1819         /**
1820          * Check if no edits were made by other users since
1821          * the time a user started editing the page. Limit to
1822          * 50 revisions for the sake of performance.
1823          *
1824          * MCR migration note: this replaces Revision::userWasLastToEdit
1825          *
1826          * @deprecated since 1.31; Can possibly be removed, since the self-conflict suppression
1827          *       logic in EditPage that uses this seems conceptually dubious. Revision::userWasLastToEdit
1828          *       has been deprecated since 1.24.
1829          *
1830          * @param IDatabase $db The Database to perform the check on.
1831          * @param int $pageId The ID of the page in question
1832          * @param int $userId The ID of the user in question
1833          * @param string $since Look at edits since this time
1834          *
1835          * @return bool True if the given user was the only one to edit since the given timestamp
1836          */
1837         public function userWasLastToEdit( IDatabase $db, $pageId, $userId, $since ) {
1838                 $this->checkDatabaseWikiId( $db );
1839
1840                 if ( !$userId ) {
1841                         return false;
1842                 }
1843
1844                 $res = $db->select(
1845                         'revision',
1846                         'rev_user',
1847                         [
1848                                 'rev_page' => $pageId,
1849                                 'rev_timestamp > ' . $db->addQuotes( $db->timestamp( $since ) )
1850                         ],
1851                         __METHOD__,
1852                         [ 'ORDER BY' => 'rev_timestamp ASC', 'LIMIT' => 50 ]
1853                 );
1854                 foreach ( $res as $row ) {
1855                         if ( $row->rev_user != $userId ) {
1856                                 return false;
1857                         }
1858                 }
1859                 return true;
1860         }
1861
1862         /**
1863          * Load a revision based on a known page ID and current revision ID from the DB
1864          *
1865          * This method allows for the use of caching, though accessing anything that normally
1866          * requires permission checks (aside from the text) will trigger a small DB lookup.
1867          *
1868          * MCR migration note: this replaces Revision::newKnownCurrent
1869          *
1870          * @param Title $title the associated page title
1871          * @param int $revId current revision of this page. Defaults to $title->getLatestRevID().
1872          *
1873          * @return RevisionRecord|bool Returns false if missing
1874          */
1875         public function getKnownCurrentRevision( Title $title, $revId ) {
1876                 $db = $this->getDBConnectionRef( DB_REPLICA );
1877
1878                 $pageId = $title->getArticleID();
1879
1880                 if ( !$pageId ) {
1881                         return false;
1882                 }
1883
1884                 if ( !$revId ) {
1885                         $revId = $title->getLatestRevID();
1886                 }
1887
1888                 if ( !$revId ) {
1889                         wfWarn(
1890                                 'No latest revision known for page ' . $title->getPrefixedDBkey()
1891                                 . ' even though it exists with page ID ' . $pageId
1892                         );
1893                         return false;
1894                 }
1895
1896                 $row = $this->cache->getWithSetCallback(
1897                         // Page/rev IDs passed in from DB to reflect history merges
1898                         $this->cache->makeGlobalKey( 'revision-row-1.29', $db->getDomainID(), $pageId, $revId ),
1899                         WANObjectCache::TTL_WEEK,
1900                         function ( $curValue, &$ttl, array &$setOpts ) use ( $db, $pageId, $revId ) {
1901                                 $setOpts += Database::getCacheSetOptions( $db );
1902
1903                                 $conds = [
1904                                         'rev_page' => intval( $pageId ),
1905                                         'page_id' => intval( $pageId ),
1906                                         'rev_id' => intval( $revId ),
1907                                 ];
1908
1909                                 $row = $this->fetchRevisionRowFromConds( $db, $conds );
1910                                 return $row ?: false; // don't cache negatives
1911                         }
1912                 );
1913
1914                 // Reflect revision deletion and user renames
1915                 if ( $row ) {
1916                         return $this->newRevisionFromRow( $row, 0, $title );
1917                 } else {
1918                         return false;
1919                 }
1920         }
1921
1922         // TODO: move relevant methods from Title here, e.g. getFirstRevision, isBigDeletion, etc.
1923
1924 }