Merge "Add cleanup for handlers cache."
[lhc/web/wiklou.git] / 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 ActorMigration;
30 use CommentStore;
31 use CommentStoreComment;
32 use Content;
33 use ContentHandler;
34 use DBAccessObjectUtils;
35 use Hooks;
36 use IDBAccessObject;
37 use InvalidArgumentException;
38 use IP;
39 use LogicException;
40 use MediaWiki\Linker\LinkTarget;
41 use MediaWiki\User\UserIdentity;
42 use MediaWiki\User\UserIdentityValue;
43 use Message;
44 use MWException;
45 use MWUnknownContentModelException;
46 use Psr\Log\LoggerAwareInterface;
47 use Psr\Log\LoggerInterface;
48 use Psr\Log\NullLogger;
49 use RecentChange;
50 use stdClass;
51 use Title;
52 use User;
53 use WANObjectCache;
54 use Wikimedia\Assert\Assert;
55 use Wikimedia\Rdbms\Database;
56 use Wikimedia\Rdbms\DBConnRef;
57 use Wikimedia\Rdbms\IDatabase;
58 use Wikimedia\Rdbms\LoadBalancer;
59
60 /**
61 * Service for looking up page revisions.
62 *
63 * @since 1.31
64 *
65 * @note This was written to act as a drop-in replacement for the corresponding
66 * static methods in Revision.
67 */
68 class RevisionStore
69 implements IDBAccessObject, RevisionFactory, RevisionLookup, LoggerAwareInterface {
70
71 /**
72 * @var SqlBlobStore
73 */
74 private $blobStore;
75
76 /**
77 * @var bool|string
78 */
79 private $wikiId;
80
81 /**
82 * @var boolean
83 */
84 private $contentHandlerUseDB = true;
85
86 /**
87 * @var LoadBalancer
88 */
89 private $loadBalancer;
90
91 /**
92 * @var WANObjectCache
93 */
94 private $cache;
95
96 /**
97 * @var CommentStore
98 */
99 private $commentStore;
100
101 /**
102 * @var ActorMigration
103 */
104 private $actorMigration;
105
106 /**
107 * @var LoggerInterface
108 */
109 private $logger;
110
111 /**
112 * @todo $blobStore should be allowed to be any BlobStore!
113 *
114 * @param LoadBalancer $loadBalancer
115 * @param SqlBlobStore $blobStore
116 * @param WANObjectCache $cache
117 * @param CommentStore $commentStore
118 * @param ActorMigration $actorMigration
119 * @param bool|string $wikiId
120 */
121 public function __construct(
122 LoadBalancer $loadBalancer,
123 SqlBlobStore $blobStore,
124 WANObjectCache $cache,
125 CommentStore $commentStore,
126 ActorMigration $actorMigration,
127 $wikiId = false
128 ) {
129 Assert::parameterType( 'string|boolean', $wikiId, '$wikiId' );
130
131 $this->loadBalancer = $loadBalancer;
132 $this->blobStore = $blobStore;
133 $this->cache = $cache;
134 $this->commentStore = $commentStore;
135 $this->actorMigration = $actorMigration;
136 $this->wikiId = $wikiId;
137 $this->logger = new NullLogger();
138 }
139
140 public function setLogger( LoggerInterface $logger ) {
141 $this->logger = $logger;
142 }
143
144 /**
145 * @return bool Whether the store is read-only
146 */
147 public function isReadOnly() {
148 return $this->blobStore->isReadOnly();
149 }
150
151 /**
152 * @return bool
153 */
154 public function getContentHandlerUseDB() {
155 return $this->contentHandlerUseDB;
156 }
157
158 /**
159 * @param bool $contentHandlerUseDB
160 */
161 public function setContentHandlerUseDB( $contentHandlerUseDB ) {
162 $this->contentHandlerUseDB = $contentHandlerUseDB;
163 }
164
165 /**
166 * @return LoadBalancer
167 */
168 private function getDBLoadBalancer() {
169 return $this->loadBalancer;
170 }
171
172 /**
173 * @param int $mode DB_MASTER or DB_REPLICA
174 *
175 * @return IDatabase
176 */
177 private function getDBConnection( $mode ) {
178 $lb = $this->getDBLoadBalancer();
179 return $lb->getConnection( $mode, [], $this->wikiId );
180 }
181
182 /**
183 * @param IDatabase $connection
184 */
185 private function releaseDBConnection( IDatabase $connection ) {
186 $lb = $this->getDBLoadBalancer();
187 $lb->reuseConnection( $connection );
188 }
189
190 /**
191 * @param int $mode DB_MASTER or DB_REPLICA
192 *
193 * @return DBConnRef
194 */
195 private function getDBConnectionRef( $mode ) {
196 $lb = $this->getDBLoadBalancer();
197 return $lb->getConnectionRef( $mode, [], $this->wikiId );
198 }
199
200 /**
201 * Determines the page Title based on the available information.
202 *
203 * MCR migration note: this corresponds to Revision::getTitle
204 *
205 * @note this method should be private, external use should be avoided!
206 *
207 * @param int|null $pageId
208 * @param int|null $revId
209 * @param int $queryFlags
210 *
211 * @return Title
212 * @throws RevisionAccessException
213 */
214 public function getTitle( $pageId, $revId, $queryFlags = self::READ_NORMAL ) {
215 if ( !$pageId && !$revId ) {
216 throw new InvalidArgumentException( '$pageId and $revId cannot both be 0 or null' );
217 }
218
219 // This method recalls itself with READ_LATEST if READ_NORMAL doesn't get us a Title
220 // So ignore READ_LATEST_IMMUTABLE flags and handle the fallback logic in this method
221 if ( DBAccessObjectUtils::hasFlags( $queryFlags, self::READ_LATEST_IMMUTABLE ) ) {
222 $queryFlags = self::READ_NORMAL;
223 }
224
225 $canUseTitleNewFromId = ( $pageId !== null && $pageId > 0 && $this->wikiId === false );
226 list( $dbMode, $dbOptions ) = DBAccessObjectUtils::getDBOptions( $queryFlags );
227 $titleFlags = ( $dbMode == DB_MASTER ? Title::GAID_FOR_UPDATE : 0 );
228
229 // Loading by ID is best, but Title::newFromID does not support that for foreign IDs.
230 if ( $canUseTitleNewFromId ) {
231 // TODO: better foreign title handling (introduce TitleFactory)
232 $title = Title::newFromID( $pageId, $titleFlags );
233 if ( $title ) {
234 return $title;
235 }
236 }
237
238 // rev_id is defined as NOT NULL, but this revision may not yet have been inserted.
239 $canUseRevId = ( $revId !== null && $revId > 0 );
240
241 if ( $canUseRevId ) {
242 $dbr = $this->getDBConnectionRef( $dbMode );
243 // @todo: Title::getSelectFields(), or Title::getQueryInfo(), or something like that
244 $row = $dbr->selectRow(
245 [ 'revision', 'page' ],
246 [
247 'page_namespace',
248 'page_title',
249 'page_id',
250 'page_latest',
251 'page_is_redirect',
252 'page_len',
253 ],
254 [ 'rev_id' => $revId ],
255 __METHOD__,
256 $dbOptions,
257 [ 'page' => [ 'JOIN', 'page_id=rev_page' ] ]
258 );
259 if ( $row ) {
260 // TODO: better foreign title handling (introduce TitleFactory)
261 return Title::newFromRow( $row );
262 }
263 }
264
265 // If we still don't have a title, fallback to master if that wasn't already happening.
266 if ( $dbMode !== DB_MASTER ) {
267 $title = $this->getTitle( $pageId, $revId, self::READ_LATEST );
268 if ( $title ) {
269 $this->logger->info(
270 __METHOD__ . ' fell back to READ_LATEST and got a Title.',
271 [ 'trace' => wfBacktrace() ]
272 );
273 return $title;
274 }
275 }
276
277 throw new RevisionAccessException(
278 "Could not determine title for page ID $pageId and revision ID $revId"
279 );
280 }
281
282 /**
283 * @param mixed $value
284 * @param string $name
285 *
286 * @throw IncompleteRevisionException if $value is null
287 * @return mixed $value, if $value is not null
288 */
289 private function failOnNull( $value, $name ) {
290 if ( $value === null ) {
291 throw new IncompleteRevisionException(
292 "$name must not be " . var_export( $value, true ) . "!"
293 );
294 }
295
296 return $value;
297 }
298
299 /**
300 * @param mixed $value
301 * @param string $name
302 *
303 * @throw IncompleteRevisionException if $value is empty
304 * @return mixed $value, if $value is not null
305 */
306 private function failOnEmpty( $value, $name ) {
307 if ( $value === null || $value === 0 || $value === '' ) {
308 throw new IncompleteRevisionException(
309 "$name must not be " . var_export( $value, true ) . "!"
310 );
311 }
312
313 return $value;
314 }
315
316 /**
317 * Insert a new revision into the database, returning the new revision record
318 * on success and dies horribly on failure.
319 *
320 * MCR migration note: this replaces Revision::insertOn
321 *
322 * @param RevisionRecord $rev
323 * @param IDatabase $dbw (master connection)
324 *
325 * @throws InvalidArgumentException
326 * @return RevisionRecord the new revision record.
327 */
328 public function insertRevisionOn( RevisionRecord $rev, IDatabase $dbw ) {
329 // TODO: pass in a DBTransactionContext instead of a database connection.
330 $this->checkDatabaseWikiId( $dbw );
331
332 if ( !$rev->getSlotRoles() ) {
333 throw new InvalidArgumentException( 'At least one slot needs to be defined!' );
334 }
335
336 if ( $rev->getSlotRoles() !== [ 'main' ] ) {
337 throw new InvalidArgumentException( 'Only the main slot is supported for now!' );
338 }
339
340 // TODO: we shouldn't need an actual Title here.
341 $title = Title::newFromLinkTarget( $rev->getPageAsLinkTarget() );
342 $pageId = $this->failOnEmpty( $rev->getPageId(), 'rev_page field' ); // check this early
343
344 $parentId = $rev->getParentId() === null
345 ? $this->getPreviousRevisionId( $dbw, $rev )
346 : $rev->getParentId();
347
348 // Record the text (or external storage URL) to the blob store
349 $slot = $rev->getSlot( 'main', RevisionRecord::RAW );
350
351 $size = $this->failOnNull( $rev->getSize(), 'size field' );
352 $sha1 = $this->failOnEmpty( $rev->getSha1(), 'sha1 field' );
353
354 if ( !$slot->hasAddress() ) {
355 $content = $slot->getContent();
356 $format = $content->getDefaultFormat();
357 $model = $content->getModel();
358
359 $this->checkContentModel( $content, $title );
360
361 $data = $content->serialize( $format );
362
363 // Hints allow the blob store to optimize by "leaking" application level information to it.
364 // TODO: with the new MCR storage schema, we rev_id have this before storing the blobs.
365 // When we have it, add rev_id as a hint. Can be used with rev_parent_id for
366 // differential storage or compression of subsequent revisions.
367 $blobHints = [
368 BlobStore::DESIGNATION_HINT => 'page-content', // BlobStore may be used for other things too.
369 BlobStore::PAGE_HINT => $pageId,
370 BlobStore::ROLE_HINT => $slot->getRole(),
371 BlobStore::PARENT_HINT => $parentId,
372 BlobStore::SHA1_HINT => $slot->getSha1(),
373 BlobStore::MODEL_HINT => $model,
374 BlobStore::FORMAT_HINT => $format,
375 ];
376
377 $blobAddress = $this->blobStore->storeBlob( $data, $blobHints );
378 } else {
379 $blobAddress = $slot->getAddress();
380 $model = $slot->getModel();
381 $format = $slot->getFormat();
382 }
383
384 $textId = $this->blobStore->getTextIdFromAddress( $blobAddress );
385
386 if ( !$textId ) {
387 throw new LogicException(
388 'Blob address not supported in 1.29 database schema: ' . $blobAddress
389 );
390 }
391
392 // getTextIdFromAddress() is free to insert something into the text table, so $textId
393 // may be a new value, not anything already contained in $blobAddress.
394 $blobAddress = 'tt:' . $textId;
395
396 $comment = $this->failOnNull( $rev->getComment( RevisionRecord::RAW ), 'comment' );
397 $user = $this->failOnNull( $rev->getUser( RevisionRecord::RAW ), 'user' );
398 $timestamp = $this->failOnEmpty( $rev->getTimestamp(), 'timestamp field' );
399
400 // Checks.
401 $this->failOnNull( $user->getId(), 'user field' );
402 $this->failOnEmpty( $user->getName(), 'user_text field' );
403
404 # Record the edit in revisions
405 $row = [
406 'rev_page' => $pageId,
407 'rev_parent_id' => $parentId,
408 'rev_text_id' => $textId,
409 'rev_minor_edit' => $rev->isMinor() ? 1 : 0,
410 'rev_timestamp' => $dbw->timestamp( $timestamp ),
411 'rev_deleted' => $rev->getVisibility(),
412 'rev_len' => $size,
413 'rev_sha1' => $sha1,
414 ];
415
416 if ( $rev->getId() !== null ) {
417 // Needed to restore revisions with their original ID
418 $row['rev_id'] = $rev->getId();
419 }
420
421 list( $commentFields, $commentCallback ) =
422 $this->commentStore->insertWithTempTable( $dbw, 'rev_comment', $comment );
423 $row += $commentFields;
424
425 list( $actorFields, $actorCallback ) =
426 $this->actorMigration->getInsertValuesWithTempTable( $dbw, 'rev_user', $user );
427 $row += $actorFields;
428
429 if ( $this->contentHandlerUseDB ) {
430 // MCR migration note: rev_content_model and rev_content_format will go away
431
432 $defaultModel = ContentHandler::getDefaultModelFor( $title );
433 $defaultFormat = ContentHandler::getForModelID( $defaultModel )->getDefaultFormat();
434
435 $row['rev_content_model'] = ( $model === $defaultModel ) ? null : $model;
436 $row['rev_content_format'] = ( $format === $defaultFormat ) ? null : $format;
437 }
438
439 $dbw->insert( 'revision', $row, __METHOD__ );
440
441 if ( !isset( $row['rev_id'] ) ) {
442 // only if auto-increment was used
443 $row['rev_id'] = intval( $dbw->insertId() );
444 }
445 $commentCallback( $row['rev_id'] );
446 $actorCallback( $row['rev_id'], $row );
447
448 // Insert IP revision into ip_changes for use when querying for a range.
449 if ( $user->getId() === 0 && IP::isValid( $user->getName() ) ) {
450 $ipcRow = [
451 'ipc_rev_id' => $row['rev_id'],
452 'ipc_rev_timestamp' => $row['rev_timestamp'],
453 'ipc_hex' => IP::toHex( $user->getName() ),
454 ];
455 $dbw->insert( 'ip_changes', $ipcRow, __METHOD__ );
456 }
457
458 $newSlot = SlotRecord::newSaved( $row['rev_id'], $textId, $blobAddress, $slot );
459 $slots = new RevisionSlots( [ 'main' => $newSlot ] );
460
461 $rev = new RevisionStoreRecord(
462 $title,
463 $user,
464 $comment,
465 (object)$row,
466 $slots,
467 $this->wikiId
468 );
469
470 $newSlot = $rev->getSlot( 'main', RevisionRecord::RAW );
471
472 // sanity checks
473 Assert::postcondition( $rev->getId() > 0, 'revision must have an ID' );
474 Assert::postcondition( $rev->getPageId() > 0, 'revision must have a page ID' );
475 Assert::postcondition(
476 $rev->getComment( RevisionRecord::RAW ) !== null,
477 'revision must have a comment'
478 );
479 Assert::postcondition(
480 $rev->getUser( RevisionRecord::RAW ) !== null,
481 'revision must have a user'
482 );
483
484 Assert::postcondition( $newSlot !== null, 'revision must have a main slot' );
485 Assert::postcondition(
486 $newSlot->getAddress() !== null,
487 'main slot must have an addess'
488 );
489
490 Hooks::run( 'RevisionRecordInserted', [ $rev ] );
491
492 return $rev;
493 }
494
495 /**
496 * MCR migration note: this corresponds to Revision::checkContentModel
497 *
498 * @param Content $content
499 * @param Title $title
500 *
501 * @throws MWException
502 * @throws MWUnknownContentModelException
503 */
504 private function checkContentModel( Content $content, Title $title ) {
505 // Note: may return null for revisions that have not yet been inserted
506
507 $model = $content->getModel();
508 $format = $content->getDefaultFormat();
509 $handler = $content->getContentHandler();
510
511 $name = "$title";
512
513 if ( !$handler->isSupportedFormat( $format ) ) {
514 throw new MWException( "Can't use format $format with content model $model on $name" );
515 }
516
517 if ( !$this->contentHandlerUseDB ) {
518 // if $wgContentHandlerUseDB is not set,
519 // all revisions must use the default content model and format.
520
521 $defaultModel = ContentHandler::getDefaultModelFor( $title );
522 $defaultHandler = ContentHandler::getForModelID( $defaultModel );
523 $defaultFormat = $defaultHandler->getDefaultFormat();
524
525 if ( $model != $defaultModel ) {
526 throw new MWException( "Can't save non-default content model with "
527 . "\$wgContentHandlerUseDB disabled: model is $model, "
528 . "default for $name is $defaultModel"
529 );
530 }
531
532 if ( $format != $defaultFormat ) {
533 throw new MWException( "Can't use non-default content format with "
534 . "\$wgContentHandlerUseDB disabled: format is $format, "
535 . "default for $name is $defaultFormat"
536 );
537 }
538 }
539
540 if ( !$content->isValid() ) {
541 throw new MWException(
542 "New content for $name is not valid! Content model is $model"
543 );
544 }
545 }
546
547 /**
548 * Create a new null-revision for insertion into a page's
549 * history. This will not re-save the text, but simply refer
550 * to the text from the previous version.
551 *
552 * Such revisions can for instance identify page rename
553 * operations and other such meta-modifications.
554 *
555 * MCR migration note: this replaces Revision::newNullRevision
556 *
557 * @todo Introduce newFromParentRevision(). newNullRevision can then be based on that
558 * (or go away).
559 *
560 * @param IDatabase $dbw
561 * @param Title $title Title of the page to read from
562 * @param CommentStoreComment $comment RevisionRecord's summary
563 * @param bool $minor Whether the revision should be considered as minor
564 * @param User $user The user to attribute the revision to
565 * @return RevisionRecord|null RevisionRecord or null on error
566 */
567 public function newNullRevision(
568 IDatabase $dbw,
569 Title $title,
570 CommentStoreComment $comment,
571 $minor,
572 User $user
573 ) {
574 $this->checkDatabaseWikiId( $dbw );
575
576 $fields = [ 'page_latest', 'page_namespace', 'page_title',
577 'rev_id', 'rev_text_id', 'rev_len', 'rev_sha1' ];
578
579 if ( $this->contentHandlerUseDB ) {
580 $fields[] = 'rev_content_model';
581 $fields[] = 'rev_content_format';
582 }
583
584 $current = $dbw->selectRow(
585 [ 'page', 'revision' ],
586 $fields,
587 [
588 'page_id' => $title->getArticleID(),
589 'page_latest=rev_id',
590 ],
591 __METHOD__,
592 [ 'FOR UPDATE' ] // T51581
593 );
594
595 if ( $current ) {
596 $fields = [
597 'page' => $title->getArticleID(),
598 'user_text' => $user->getName(),
599 'user' => $user->getId(),
600 'actor' => $user->getActorId(),
601 'comment' => $comment,
602 'minor_edit' => $minor,
603 'text_id' => $current->rev_text_id,
604 'parent_id' => $current->page_latest,
605 'len' => $current->rev_len,
606 'sha1' => $current->rev_sha1
607 ];
608
609 if ( $this->contentHandlerUseDB ) {
610 $fields['content_model'] = $current->rev_content_model;
611 $fields['content_format'] = $current->rev_content_format;
612 }
613
614 $fields['title'] = Title::makeTitle( $current->page_namespace, $current->page_title );
615
616 $mainSlot = $this->emulateMainSlot_1_29( $fields, self::READ_LATEST, $title );
617 $revision = new MutableRevisionRecord( $title, $this->wikiId );
618 $this->initializeMutableRevisionFromArray( $revision, $fields );
619 $revision->setSlot( $mainSlot );
620 } else {
621 $revision = null;
622 }
623
624 return $revision;
625 }
626
627 /**
628 * MCR migration note: this replaces Revision::isUnpatrolled
629 *
630 * @todo This is overly specific, so move or kill this method.
631 *
632 * @param RevisionRecord $rev
633 *
634 * @return int Rcid of the unpatrolled row, zero if there isn't one
635 */
636 public function getRcIdIfUnpatrolled( RevisionRecord $rev ) {
637 $rc = $this->getRecentChange( $rev );
638 if ( $rc && $rc->getAttribute( 'rc_patrolled' ) == 0 ) {
639 return $rc->getAttribute( 'rc_id' );
640 } else {
641 return 0;
642 }
643 }
644
645 /**
646 * Get the RC object belonging to the current revision, if there's one
647 *
648 * MCR migration note: this replaces Revision::getRecentChange
649 *
650 * @todo move this somewhere else?
651 *
652 * @param RevisionRecord $rev
653 * @param int $flags (optional) $flags include:
654 * IDBAccessObject::READ_LATEST: Select the data from the master
655 *
656 * @return null|RecentChange
657 */
658 public function getRecentChange( RevisionRecord $rev, $flags = 0 ) {
659 $dbr = $this->getDBConnection( DB_REPLICA );
660
661 list( $dbType, ) = DBAccessObjectUtils::getDBOptions( $flags );
662
663 $userIdentity = $rev->getUser( RevisionRecord::RAW );
664
665 if ( !$userIdentity ) {
666 // If the revision has no user identity, chances are it never went
667 // into the database, and doesn't have an RC entry.
668 return null;
669 }
670
671 // TODO: Select by rc_this_oldid alone - but as of Nov 2017, there is no index on that!
672 $actorWhere = $this->actorMigration->getWhere( $dbr, 'rc_user', $rev->getUser(), false );
673 $rc = RecentChange::newFromConds(
674 [
675 $actorWhere['conds'],
676 'rc_timestamp' => $dbr->timestamp( $rev->getTimestamp() ),
677 'rc_this_oldid' => $rev->getId()
678 ],
679 __METHOD__,
680 $dbType
681 );
682
683 $this->releaseDBConnection( $dbr );
684
685 // XXX: cache this locally? Glue it to the RevisionRecord?
686 return $rc;
687 }
688
689 /**
690 * Maps fields of the archive row to corresponding revision rows.
691 *
692 * @param object $archiveRow
693 *
694 * @return object a revision row object, corresponding to $archiveRow.
695 */
696 private static function mapArchiveFields( $archiveRow ) {
697 $fieldMap = [
698 // keep with ar prefix:
699 'ar_id' => 'ar_id',
700
701 // not the same suffix:
702 'ar_page_id' => 'rev_page',
703 'ar_rev_id' => 'rev_id',
704
705 // same suffix:
706 'ar_text_id' => 'rev_text_id',
707 'ar_timestamp' => 'rev_timestamp',
708 'ar_user_text' => 'rev_user_text',
709 'ar_user' => 'rev_user',
710 'ar_actor' => 'rev_actor',
711 'ar_minor_edit' => 'rev_minor_edit',
712 'ar_deleted' => 'rev_deleted',
713 'ar_len' => 'rev_len',
714 'ar_parent_id' => 'rev_parent_id',
715 'ar_sha1' => 'rev_sha1',
716 'ar_comment' => 'rev_comment',
717 'ar_comment_cid' => 'rev_comment_cid',
718 'ar_comment_id' => 'rev_comment_id',
719 'ar_comment_text' => 'rev_comment_text',
720 'ar_comment_data' => 'rev_comment_data',
721 'ar_comment_old' => 'rev_comment_old',
722 'ar_content_format' => 'rev_content_format',
723 'ar_content_model' => 'rev_content_model',
724 ];
725
726 if ( empty( $archiveRow->ar_text_id ) ) {
727 $fieldMap['ar_text'] = 'old_text';
728 $fieldMap['ar_flags'] = 'old_flags';
729 }
730
731 $revRow = new stdClass();
732 foreach ( $fieldMap as $arKey => $revKey ) {
733 if ( property_exists( $archiveRow, $arKey ) ) {
734 $revRow->$revKey = $archiveRow->$arKey;
735 }
736 }
737
738 return $revRow;
739 }
740
741 /**
742 * Constructs a RevisionRecord for the revisions main slot, based on the MW1.29 schema.
743 *
744 * @param object|array $row Either a database row or an array
745 * @param int $queryFlags for callbacks
746 * @param Title $title
747 *
748 * @return SlotRecord The main slot, extracted from the MW 1.29 style row.
749 * @throws MWException
750 */
751 private function emulateMainSlot_1_29( $row, $queryFlags, Title $title ) {
752 $mainSlotRow = new stdClass();
753 $mainSlotRow->role_name = 'main';
754 $mainSlotRow->model_name = null;
755 $mainSlotRow->slot_revision_id = null;
756 $mainSlotRow->content_address = null;
757 $mainSlotRow->slot_content_id = null;
758
759 $content = null;
760 $blobData = null;
761 $blobFlags = null;
762
763 if ( is_object( $row ) ) {
764 // archive row
765 if ( !isset( $row->rev_id ) && ( isset( $row->ar_user ) || isset( $row->ar_actor ) ) ) {
766 $row = $this->mapArchiveFields( $row );
767 }
768
769 if ( isset( $row->rev_text_id ) && $row->rev_text_id > 0 ) {
770 $mainSlotRow->slot_content_id = $row->rev_text_id;
771 $mainSlotRow->content_address = 'tt:' . $row->rev_text_id;
772 }
773
774 if ( isset( $row->old_text ) ) {
775 // this happens when the text-table gets joined directly, in the pre-1.30 schema
776 $blobData = isset( $row->old_text ) ? strval( $row->old_text ) : null;
777 // Check against selects that might have not included old_flags
778 if ( !property_exists( $row, 'old_flags' ) ) {
779 throw new InvalidArgumentException( 'old_flags was not set in $row' );
780 }
781 $blobFlags = ( $row->old_flags === null ) ? '' : $row->old_flags;
782 }
783
784 $mainSlotRow->slot_revision_id = intval( $row->rev_id );
785
786 $mainSlotRow->content_size = isset( $row->rev_len ) ? intval( $row->rev_len ) : null;
787 $mainSlotRow->content_sha1 = isset( $row->rev_sha1 ) ? strval( $row->rev_sha1 ) : null;
788 $mainSlotRow->model_name = isset( $row->rev_content_model )
789 ? strval( $row->rev_content_model )
790 : null;
791 // XXX: in the future, we'll probably always use the default format, and drop content_format
792 $mainSlotRow->format_name = isset( $row->rev_content_format )
793 ? strval( $row->rev_content_format )
794 : null;
795 } elseif ( is_array( $row ) ) {
796 $mainSlotRow->slot_revision_id = isset( $row['id'] ) ? intval( $row['id'] ) : null;
797
798 $mainSlotRow->slot_content_id = isset( $row['text_id'] )
799 ? intval( $row['text_id'] )
800 : null;
801 $mainSlotRow->content_address = isset( $row['text_id'] )
802 ? 'tt:' . intval( $row['text_id'] )
803 : null;
804 $mainSlotRow->content_size = isset( $row['len'] ) ? intval( $row['len'] ) : null;
805 $mainSlotRow->content_sha1 = isset( $row['sha1'] ) ? strval( $row['sha1'] ) : null;
806
807 $mainSlotRow->model_name = isset( $row['content_model'] )
808 ? strval( $row['content_model'] ) : null; // XXX: must be a string!
809 // XXX: in the future, we'll probably always use the default format, and drop content_format
810 $mainSlotRow->format_name = isset( $row['content_format'] )
811 ? strval( $row['content_format'] ) : null;
812 $blobData = isset( $row['text'] ) ? rtrim( strval( $row['text'] ) ) : null;
813 // XXX: If the flags field is not set then $blobFlags should be null so that no
814 // decoding will happen. An empty string will result in default decodings.
815 $blobFlags = isset( $row['flags'] ) ? trim( strval( $row['flags'] ) ) : null;
816
817 // if we have a Content object, override mText and mContentModel
818 if ( !empty( $row['content'] ) ) {
819 if ( !( $row['content'] instanceof Content ) ) {
820 throw new MWException( 'content field must contain a Content object.' );
821 }
822
823 /** @var Content $content */
824 $content = $row['content'];
825 $handler = $content->getContentHandler();
826
827 $mainSlotRow->model_name = $content->getModel();
828
829 // XXX: in the future, we'll probably always use the default format.
830 if ( $mainSlotRow->format_name === null ) {
831 $mainSlotRow->format_name = $handler->getDefaultFormat();
832 }
833 }
834 } else {
835 throw new MWException( 'Revision constructor passed invalid row format.' );
836 }
837
838 // With the old schema, the content changes with every revision.
839 // ...except for null-revisions. Would be nice if we could detect them.
840 $mainSlotRow->slot_inherited = 0;
841
842 if ( $mainSlotRow->model_name === null ) {
843 $mainSlotRow->model_name = function ( SlotRecord $slot ) use ( $title ) {
844 // TODO: MCR: consider slot role in getDefaultModelFor()! Use LinkTarget!
845 // TODO: MCR: deprecate $title->getModel().
846 return ContentHandler::getDefaultModelFor( $title );
847 };
848 }
849
850 if ( !$content ) {
851 $content = function ( SlotRecord $slot )
852 use ( $blobData, $blobFlags, $queryFlags, $mainSlotRow )
853 {
854 return $this->loadSlotContent(
855 $slot,
856 $blobData,
857 $blobFlags,
858 $mainSlotRow->format_name,
859 $queryFlags
860 );
861 };
862 }
863
864 $mainSlotRow->slot_id = $mainSlotRow->slot_revision_id;
865 return new SlotRecord( $mainSlotRow, $content );
866 }
867
868 /**
869 * Loads a Content object based on a slot row.
870 *
871 * This method does not call $slot->getContent(), and may be used as a callback
872 * called by $slot->getContent().
873 *
874 * MCR migration note: this roughly corresponds to Revision::getContentInternal
875 *
876 * @param SlotRecord $slot The SlotRecord to load content for
877 * @param string|null $blobData The content blob, in the form indicated by $blobFlags
878 * @param string|null $blobFlags Flags indicating how $blobData needs to be processed.
879 * Use null if no processing should happen. That is in constrast to the empty string,
880 * which causes the blob to be decoded according to the configured legacy encoding.
881 * @param string|null $blobFormat MIME type indicating how $dataBlob is encoded
882 * @param int $queryFlags
883 *
884 * @throw RevisionAccessException
885 * @return Content
886 */
887 private function loadSlotContent(
888 SlotRecord $slot,
889 $blobData = null,
890 $blobFlags = null,
891 $blobFormat = null,
892 $queryFlags = 0
893 ) {
894 if ( $blobData !== null ) {
895 Assert::parameterType( 'string', $blobData, '$blobData' );
896 Assert::parameterType( 'string|null', $blobFlags, '$blobFlags' );
897
898 $cacheKey = $slot->hasAddress() ? $slot->getAddress() : null;
899
900 if ( $blobFlags === null ) {
901 // No blob flags, so use the blob verbatim.
902 $data = $blobData;
903 } else {
904 $data = $this->blobStore->expandBlob( $blobData, $blobFlags, $cacheKey );
905 if ( $data === false ) {
906 throw new RevisionAccessException(
907 "Failed to expand blob data using flags $blobFlags (key: $cacheKey)"
908 );
909 }
910 }
911
912 } else {
913 $address = $slot->getAddress();
914 try {
915 $data = $this->blobStore->getBlob( $address, $queryFlags );
916 } catch ( BlobAccessException $e ) {
917 throw new RevisionAccessException(
918 "Failed to load data blob from $address: " . $e->getMessage(), 0, $e
919 );
920 }
921 }
922
923 // Unserialize content
924 $handler = ContentHandler::getForModelID( $slot->getModel() );
925
926 $content = $handler->unserializeContent( $data, $blobFormat );
927 return $content;
928 }
929
930 /**
931 * Load a page revision from a given revision ID number.
932 * Returns null if no such revision can be found.
933 *
934 * MCR migration note: this replaces Revision::newFromId
935 *
936 * $flags include:
937 * IDBAccessObject::READ_LATEST: Select the data from the master
938 * IDBAccessObject::READ_LOCKING : Select & lock the data from the master
939 *
940 * @param int $id
941 * @param int $flags (optional)
942 * @return RevisionRecord|null
943 */
944 public function getRevisionById( $id, $flags = 0 ) {
945 return $this->newRevisionFromConds( [ 'rev_id' => intval( $id ) ], $flags );
946 }
947
948 /**
949 * Load either the current, or a specified, revision
950 * that's attached to a given link target. If not attached
951 * to that link target, will return null.
952 *
953 * MCR migration note: this replaces Revision::newFromTitle
954 *
955 * $flags include:
956 * IDBAccessObject::READ_LATEST: Select the data from the master
957 * IDBAccessObject::READ_LOCKING : Select & lock the data from the master
958 *
959 * @param LinkTarget $linkTarget
960 * @param int $revId (optional)
961 * @param int $flags Bitfield (optional)
962 * @return RevisionRecord|null
963 */
964 public function getRevisionByTitle( LinkTarget $linkTarget, $revId = 0, $flags = 0 ) {
965 $conds = [
966 'page_namespace' => $linkTarget->getNamespace(),
967 'page_title' => $linkTarget->getDBkey()
968 ];
969 if ( $revId ) {
970 // Use the specified revision ID.
971 // Note that we use newRevisionFromConds here because we want to retry
972 // and fall back to master if the page is not found on a replica.
973 // Since the caller supplied a revision ID, we are pretty sure the revision is
974 // supposed to exist, so we should try hard to find it.
975 $conds['rev_id'] = $revId;
976 return $this->newRevisionFromConds( $conds, $flags );
977 } else {
978 // Use a join to get the latest revision.
979 // Note that we don't use newRevisionFromConds here because we don't want to retry
980 // and fall back to master. The assumption is that we only want to force the fallback
981 // if we are quite sure the revision exists because the caller supplied a revision ID.
982 // If the page isn't found at all on a replica, it probably simply does not exist.
983 $db = $this->getDBConnection( ( $flags & self::READ_LATEST ) ? DB_MASTER : DB_REPLICA );
984
985 $conds[] = 'rev_id=page_latest';
986 $rev = $this->loadRevisionFromConds( $db, $conds, $flags );
987
988 $this->releaseDBConnection( $db );
989 return $rev;
990 }
991 }
992
993 /**
994 * Load either the current, or a specified, revision
995 * that's attached to a given page ID.
996 * Returns null if no such revision can be found.
997 *
998 * MCR migration note: this replaces Revision::newFromPageId
999 *
1000 * $flags include:
1001 * IDBAccessObject::READ_LATEST: Select the data from the master (since 1.20)
1002 * IDBAccessObject::READ_LOCKING : Select & lock the data from the master
1003 *
1004 * @param int $pageId
1005 * @param int $revId (optional)
1006 * @param int $flags Bitfield (optional)
1007 * @return RevisionRecord|null
1008 */
1009 public function getRevisionByPageId( $pageId, $revId = 0, $flags = 0 ) {
1010 $conds = [ 'page_id' => $pageId ];
1011 if ( $revId ) {
1012 // Use the specified revision ID.
1013 // Note that we use newRevisionFromConds here because we want to retry
1014 // and fall back to master if the page is not found on a replica.
1015 // Since the caller supplied a revision ID, we are pretty sure the revision is
1016 // supposed to exist, so we should try hard to find it.
1017 $conds['rev_id'] = $revId;
1018 return $this->newRevisionFromConds( $conds, $flags );
1019 } else {
1020 // Use a join to get the latest revision.
1021 // Note that we don't use newRevisionFromConds here because we don't want to retry
1022 // and fall back to master. The assumption is that we only want to force the fallback
1023 // if we are quite sure the revision exists because the caller supplied a revision ID.
1024 // If the page isn't found at all on a replica, it probably simply does not exist.
1025 $db = $this->getDBConnection( ( $flags & self::READ_LATEST ) ? DB_MASTER : DB_REPLICA );
1026
1027 $conds[] = 'rev_id=page_latest';
1028 $rev = $this->loadRevisionFromConds( $db, $conds, $flags );
1029
1030 $this->releaseDBConnection( $db );
1031 return $rev;
1032 }
1033 }
1034
1035 /**
1036 * Load the revision for the given title with the given timestamp.
1037 * WARNING: Timestamps may in some circumstances not be unique,
1038 * so this isn't the best key to use.
1039 *
1040 * MCR migration note: this replaces Revision::loadFromTimestamp
1041 *
1042 * @param Title $title
1043 * @param string $timestamp
1044 * @return RevisionRecord|null
1045 */
1046 public function getRevisionByTimestamp( $title, $timestamp ) {
1047 return $this->newRevisionFromConds(
1048 [
1049 'rev_timestamp' => $timestamp,
1050 'page_namespace' => $title->getNamespace(),
1051 'page_title' => $title->getDBkey()
1052 ],
1053 0,
1054 $title
1055 );
1056 }
1057
1058 /**
1059 * Make a fake revision object from an archive table row. This is queried
1060 * for permissions or even inserted (as in Special:Undelete)
1061 *
1062 * MCR migration note: this replaces Revision::newFromArchiveRow
1063 *
1064 * @param object $row
1065 * @param int $queryFlags
1066 * @param Title|null $title
1067 * @param array $overrides associative array with fields of $row to override. This may be
1068 * used e.g. to force the parent revision ID or page ID. Keys in the array are fields
1069 * names from the archive table without the 'ar_' prefix, i.e. use 'parent_id' to
1070 * override ar_parent_id.
1071 *
1072 * @return RevisionRecord
1073 * @throws MWException
1074 */
1075 public function newRevisionFromArchiveRow(
1076 $row,
1077 $queryFlags = 0,
1078 Title $title = null,
1079 array $overrides = []
1080 ) {
1081 Assert::parameterType( 'object', $row, '$row' );
1082
1083 // check second argument, since Revision::newFromArchiveRow had $overrides in that spot.
1084 Assert::parameterType( 'integer', $queryFlags, '$queryFlags' );
1085
1086 if ( !$title && isset( $overrides['title'] ) ) {
1087 if ( !( $overrides['title'] instanceof Title ) ) {
1088 throw new MWException( 'title field override must contain a Title object.' );
1089 }
1090
1091 $title = $overrides['title'];
1092 }
1093
1094 if ( !isset( $title ) ) {
1095 if ( isset( $row->ar_namespace ) && isset( $row->ar_title ) ) {
1096 $title = Title::makeTitle( $row->ar_namespace, $row->ar_title );
1097 } else {
1098 throw new InvalidArgumentException(
1099 'A Title or ar_namespace and ar_title must be given'
1100 );
1101 }
1102 }
1103
1104 foreach ( $overrides as $key => $value ) {
1105 $field = "ar_$key";
1106 $row->$field = $value;
1107 }
1108
1109 try {
1110 $user = User::newFromAnyId(
1111 isset( $row->ar_user ) ? $row->ar_user : null,
1112 isset( $row->ar_user_text ) ? $row->ar_user_text : null,
1113 isset( $row->ar_actor ) ? $row->ar_actor : null
1114 );
1115 } catch ( InvalidArgumentException $ex ) {
1116 wfWarn( __METHOD__ . ': ' . $ex->getMessage() );
1117 $user = new UserIdentityValue( 0, '', 0 );
1118 }
1119
1120 $comment = $this->commentStore
1121 // Legacy because $row may have come from self::selectFields()
1122 ->getCommentLegacy( $this->getDBConnection( DB_REPLICA ), 'ar_comment', $row, true );
1123
1124 $mainSlot = $this->emulateMainSlot_1_29( $row, $queryFlags, $title );
1125 $slots = new RevisionSlots( [ 'main' => $mainSlot ] );
1126
1127 return new RevisionArchiveRecord( $title, $user, $comment, $row, $slots, $this->wikiId );
1128 }
1129
1130 /**
1131 * @see RevisionFactory::newRevisionFromRow_1_29
1132 *
1133 * MCR migration note: this replaces Revision::newFromRow
1134 *
1135 * @param object $row
1136 * @param int $queryFlags
1137 * @param Title|null $title
1138 *
1139 * @return RevisionRecord
1140 * @throws MWException
1141 * @throws RevisionAccessException
1142 */
1143 private function newRevisionFromRow_1_29( $row, $queryFlags = 0, Title $title = null ) {
1144 Assert::parameterType( 'object', $row, '$row' );
1145
1146 if ( !$title ) {
1147 $pageId = isset( $row->rev_page ) ? $row->rev_page : 0; // XXX: also check page_id?
1148 $revId = isset( $row->rev_id ) ? $row->rev_id : 0;
1149
1150 $title = $this->getTitle( $pageId, $revId, $queryFlags );
1151 }
1152
1153 if ( !isset( $row->page_latest ) ) {
1154 $row->page_latest = $title->getLatestRevID();
1155 if ( $row->page_latest === 0 && $title->exists() ) {
1156 wfWarn( 'Encountered title object in limbo: ID ' . $title->getArticleID() );
1157 }
1158 }
1159
1160 try {
1161 $user = User::newFromAnyId(
1162 isset( $row->rev_user ) ? $row->rev_user : null,
1163 isset( $row->rev_user_text ) ? $row->rev_user_text : null,
1164 isset( $row->rev_actor ) ? $row->rev_actor : null
1165 );
1166 } catch ( InvalidArgumentException $ex ) {
1167 wfWarn( __METHOD__ . ': ' . $ex->getMessage() );
1168 $user = new UserIdentityValue( 0, '', 0 );
1169 }
1170
1171 $comment = $this->commentStore
1172 // Legacy because $row may have come from self::selectFields()
1173 ->getCommentLegacy( $this->getDBConnection( DB_REPLICA ), 'rev_comment', $row, true );
1174
1175 $mainSlot = $this->emulateMainSlot_1_29( $row, $queryFlags, $title );
1176 $slots = new RevisionSlots( [ 'main' => $mainSlot ] );
1177
1178 return new RevisionStoreRecord( $title, $user, $comment, $row, $slots, $this->wikiId );
1179 }
1180
1181 /**
1182 * @see RevisionFactory::newRevisionFromRow
1183 *
1184 * MCR migration note: this replaces Revision::newFromRow
1185 *
1186 * @param object $row
1187 * @param int $queryFlags
1188 * @param Title|null $title
1189 *
1190 * @return RevisionRecord
1191 */
1192 public function newRevisionFromRow( $row, $queryFlags = 0, Title $title = null ) {
1193 return $this->newRevisionFromRow_1_29( $row, $queryFlags, $title );
1194 }
1195
1196 /**
1197 * Constructs a new MutableRevisionRecord based on the given associative array following
1198 * the MW1.29 convention for the Revision constructor.
1199 *
1200 * MCR migration note: this replaces Revision::newFromRow
1201 *
1202 * @param array $fields
1203 * @param int $queryFlags
1204 * @param Title|null $title
1205 *
1206 * @return MutableRevisionRecord
1207 * @throws MWException
1208 * @throws RevisionAccessException
1209 */
1210 public function newMutableRevisionFromArray(
1211 array $fields,
1212 $queryFlags = 0,
1213 Title $title = null
1214 ) {
1215 if ( !$title && isset( $fields['title'] ) ) {
1216 if ( !( $fields['title'] instanceof Title ) ) {
1217 throw new MWException( 'title field must contain a Title object.' );
1218 }
1219
1220 $title = $fields['title'];
1221 }
1222
1223 if ( !$title ) {
1224 $pageId = isset( $fields['page'] ) ? $fields['page'] : 0;
1225 $revId = isset( $fields['id'] ) ? $fields['id'] : 0;
1226
1227 $title = $this->getTitle( $pageId, $revId, $queryFlags );
1228 }
1229
1230 if ( !isset( $fields['page'] ) ) {
1231 $fields['page'] = $title->getArticleID( $queryFlags );
1232 }
1233
1234 // if we have a content object, use it to set the model and type
1235 if ( !empty( $fields['content'] ) ) {
1236 if ( !( $fields['content'] instanceof Content ) ) {
1237 throw new MWException( 'content field must contain a Content object.' );
1238 }
1239
1240 if ( !empty( $fields['text_id'] ) ) {
1241 throw new MWException(
1242 "Text already stored in external store (id {$fields['text_id']}), " .
1243 "can't serialize content object"
1244 );
1245 }
1246 }
1247
1248 if (
1249 isset( $fields['comment'] )
1250 && !( $fields['comment'] instanceof CommentStoreComment )
1251 ) {
1252 $commentData = isset( $fields['comment_data'] ) ? $fields['comment_data'] : null;
1253
1254 if ( $fields['comment'] instanceof Message ) {
1255 $fields['comment'] = CommentStoreComment::newUnsavedComment(
1256 $fields['comment'],
1257 $commentData
1258 );
1259 } else {
1260 $commentText = trim( strval( $fields['comment'] ) );
1261 $fields['comment'] = CommentStoreComment::newUnsavedComment(
1262 $commentText,
1263 $commentData
1264 );
1265 }
1266 }
1267
1268 $mainSlot = $this->emulateMainSlot_1_29( $fields, $queryFlags, $title );
1269
1270 $revision = new MutableRevisionRecord( $title, $this->wikiId );
1271 $this->initializeMutableRevisionFromArray( $revision, $fields );
1272 $revision->setSlot( $mainSlot );
1273
1274 return $revision;
1275 }
1276
1277 /**
1278 * @param MutableRevisionRecord $record
1279 * @param array $fields
1280 */
1281 private function initializeMutableRevisionFromArray(
1282 MutableRevisionRecord $record,
1283 array $fields
1284 ) {
1285 /** @var UserIdentity $user */
1286 $user = null;
1287
1288 if ( isset( $fields['user'] ) && ( $fields['user'] instanceof UserIdentity ) ) {
1289 $user = $fields['user'];
1290 } else {
1291 try {
1292 $user = User::newFromAnyId(
1293 isset( $fields['user'] ) ? $fields['user'] : null,
1294 isset( $fields['user_text'] ) ? $fields['user_text'] : null,
1295 isset( $fields['actor'] ) ? $fields['actor'] : null
1296 );
1297 } catch ( InvalidArgumentException $ex ) {
1298 $user = null;
1299 }
1300 }
1301
1302 if ( $user ) {
1303 $record->setUser( $user );
1304 }
1305
1306 $timestamp = isset( $fields['timestamp'] )
1307 ? strval( $fields['timestamp'] )
1308 : wfTimestampNow(); // TODO: use a callback, so we can override it for testing.
1309
1310 $record->setTimestamp( $timestamp );
1311
1312 if ( isset( $fields['page'] ) ) {
1313 $record->setPageId( intval( $fields['page'] ) );
1314 }
1315
1316 if ( isset( $fields['id'] ) ) {
1317 $record->setId( intval( $fields['id'] ) );
1318 }
1319 if ( isset( $fields['parent_id'] ) ) {
1320 $record->setParentId( intval( $fields['parent_id'] ) );
1321 }
1322
1323 if ( isset( $fields['sha1'] ) ) {
1324 $record->setSha1( $fields['sha1'] );
1325 }
1326 if ( isset( $fields['size'] ) ) {
1327 $record->setSize( intval( $fields['size'] ) );
1328 }
1329
1330 if ( isset( $fields['minor_edit'] ) ) {
1331 $record->setMinorEdit( intval( $fields['minor_edit'] ) !== 0 );
1332 }
1333 if ( isset( $fields['deleted'] ) ) {
1334 $record->setVisibility( intval( $fields['deleted'] ) );
1335 }
1336
1337 if ( isset( $fields['comment'] ) ) {
1338 Assert::parameterType(
1339 CommentStoreComment::class,
1340 $fields['comment'],
1341 '$row[\'comment\']'
1342 );
1343 $record->setComment( $fields['comment'] );
1344 }
1345 }
1346
1347 /**
1348 * Load a page revision from a given revision ID number.
1349 * Returns null if no such revision can be found.
1350 *
1351 * MCR migration note: this corresponds to Revision::loadFromId
1352 *
1353 * @note direct use is deprecated!
1354 * @todo remove when unused! there seem to be no callers of Revision::loadFromId
1355 *
1356 * @param IDatabase $db
1357 * @param int $id
1358 *
1359 * @return RevisionRecord|null
1360 */
1361 public function loadRevisionFromId( IDatabase $db, $id ) {
1362 return $this->loadRevisionFromConds( $db, [ 'rev_id' => intval( $id ) ] );
1363 }
1364
1365 /**
1366 * Load either the current, or a specified, revision
1367 * that's attached to a given page. If not attached
1368 * to that page, will return null.
1369 *
1370 * MCR migration note: this replaces Revision::loadFromPageId
1371 *
1372 * @note direct use is deprecated!
1373 * @todo remove when unused!
1374 *
1375 * @param IDatabase $db
1376 * @param int $pageid
1377 * @param int $id
1378 * @return RevisionRecord|null
1379 */
1380 public function loadRevisionFromPageId( IDatabase $db, $pageid, $id = 0 ) {
1381 $conds = [ 'rev_page' => intval( $pageid ), 'page_id' => intval( $pageid ) ];
1382 if ( $id ) {
1383 $conds['rev_id'] = intval( $id );
1384 } else {
1385 $conds[] = 'rev_id=page_latest';
1386 }
1387 return $this->loadRevisionFromConds( $db, $conds );
1388 }
1389
1390 /**
1391 * Load either the current, or a specified, revision
1392 * that's attached to a given page. If not attached
1393 * to that page, will return null.
1394 *
1395 * MCR migration note: this replaces Revision::loadFromTitle
1396 *
1397 * @note direct use is deprecated!
1398 * @todo remove when unused!
1399 *
1400 * @param IDatabase $db
1401 * @param Title $title
1402 * @param int $id
1403 *
1404 * @return RevisionRecord|null
1405 */
1406 public function loadRevisionFromTitle( IDatabase $db, $title, $id = 0 ) {
1407 if ( $id ) {
1408 $matchId = intval( $id );
1409 } else {
1410 $matchId = 'page_latest';
1411 }
1412
1413 return $this->loadRevisionFromConds(
1414 $db,
1415 [
1416 "rev_id=$matchId",
1417 'page_namespace' => $title->getNamespace(),
1418 'page_title' => $title->getDBkey()
1419 ],
1420 0,
1421 $title
1422 );
1423 }
1424
1425 /**
1426 * Load the revision for the given title with the given timestamp.
1427 * WARNING: Timestamps may in some circumstances not be unique,
1428 * so this isn't the best key to use.
1429 *
1430 * MCR migration note: this replaces Revision::loadFromTimestamp
1431 *
1432 * @note direct use is deprecated! Use getRevisionFromTimestamp instead!
1433 * @todo remove when unused!
1434 *
1435 * @param IDatabase $db
1436 * @param Title $title
1437 * @param string $timestamp
1438 * @return RevisionRecord|null
1439 */
1440 public function loadRevisionFromTimestamp( IDatabase $db, $title, $timestamp ) {
1441 return $this->loadRevisionFromConds( $db,
1442 [
1443 'rev_timestamp' => $db->timestamp( $timestamp ),
1444 'page_namespace' => $title->getNamespace(),
1445 'page_title' => $title->getDBkey()
1446 ],
1447 0,
1448 $title
1449 );
1450 }
1451
1452 /**
1453 * Given a set of conditions, fetch a revision
1454 *
1455 * This method should be used if we are pretty sure the revision exists.
1456 * Unless $flags has READ_LATEST set, this method will first try to find the revision
1457 * on a replica before hitting the master database.
1458 *
1459 * MCR migration note: this corresponds to Revision::newFromConds
1460 *
1461 * @param array $conditions
1462 * @param int $flags (optional)
1463 * @param Title $title
1464 *
1465 * @return RevisionRecord|null
1466 */
1467 private function newRevisionFromConds( $conditions, $flags = 0, Title $title = null ) {
1468 $db = $this->getDBConnection( ( $flags & self::READ_LATEST ) ? DB_MASTER : DB_REPLICA );
1469 $rev = $this->loadRevisionFromConds( $db, $conditions, $flags, $title );
1470 $this->releaseDBConnection( $db );
1471
1472 $lb = $this->getDBLoadBalancer();
1473
1474 // Make sure new pending/committed revision are visibile later on
1475 // within web requests to certain avoid bugs like T93866 and T94407.
1476 if ( !$rev
1477 && !( $flags & self::READ_LATEST )
1478 && $lb->getServerCount() > 1
1479 && $lb->hasOrMadeRecentMasterChanges()
1480 ) {
1481 $flags = self::READ_LATEST;
1482 $db = $this->getDBConnection( DB_MASTER );
1483 $rev = $this->loadRevisionFromConds( $db, $conditions, $flags, $title );
1484 $this->releaseDBConnection( $db );
1485 }
1486
1487 return $rev;
1488 }
1489
1490 /**
1491 * Given a set of conditions, fetch a revision from
1492 * the given database connection.
1493 *
1494 * MCR migration note: this corresponds to Revision::loadFromConds
1495 *
1496 * @param IDatabase $db
1497 * @param array $conditions
1498 * @param int $flags (optional)
1499 * @param Title $title
1500 *
1501 * @return RevisionRecord|null
1502 */
1503 private function loadRevisionFromConds(
1504 IDatabase $db,
1505 $conditions,
1506 $flags = 0,
1507 Title $title = null
1508 ) {
1509 $row = $this->fetchRevisionRowFromConds( $db, $conditions, $flags );
1510 if ( $row ) {
1511 $rev = $this->newRevisionFromRow( $row, $flags, $title );
1512
1513 return $rev;
1514 }
1515
1516 return null;
1517 }
1518
1519 /**
1520 * Throws an exception if the given database connection does not belong to the wiki this
1521 * RevisionStore is bound to.
1522 *
1523 * @param IDatabase $db
1524 * @throws MWException
1525 */
1526 private function checkDatabaseWikiId( IDatabase $db ) {
1527 $storeWiki = $this->wikiId;
1528 $dbWiki = $db->getDomainID();
1529
1530 if ( $dbWiki === $storeWiki ) {
1531 return;
1532 }
1533
1534 // XXX: we really want the default database ID...
1535 $storeWiki = $storeWiki ?: wfWikiID();
1536 $dbWiki = $dbWiki ?: wfWikiID();
1537
1538 if ( $dbWiki === $storeWiki ) {
1539 return;
1540 }
1541
1542 // HACK: counteract encoding imposed by DatabaseDomain
1543 $storeWiki = str_replace( '?h', '-', $storeWiki );
1544 $dbWiki = str_replace( '?h', '-', $dbWiki );
1545
1546 if ( $dbWiki === $storeWiki ) {
1547 return;
1548 }
1549
1550 throw new MWException( "RevisionStore for $storeWiki "
1551 . "cannot be used with a DB connection for $dbWiki" );
1552 }
1553
1554 /**
1555 * Given a set of conditions, return a row with the
1556 * fields necessary to build RevisionRecord objects.
1557 *
1558 * MCR migration note: this corresponds to Revision::fetchFromConds
1559 *
1560 * @param IDatabase $db
1561 * @param array $conditions
1562 * @param int $flags (optional)
1563 *
1564 * @return object|false data row as a raw object
1565 */
1566 private function fetchRevisionRowFromConds( IDatabase $db, $conditions, $flags = 0 ) {
1567 $this->checkDatabaseWikiId( $db );
1568
1569 $revQuery = self::getQueryInfo( [ 'page', 'user' ] );
1570 $options = [];
1571 if ( ( $flags & self::READ_LOCKING ) == self::READ_LOCKING ) {
1572 $options[] = 'FOR UPDATE';
1573 }
1574 return $db->selectRow(
1575 $revQuery['tables'],
1576 $revQuery['fields'],
1577 $conditions,
1578 __METHOD__,
1579 $options,
1580 $revQuery['joins']
1581 );
1582 }
1583
1584 /**
1585 * Return the tables, fields, and join conditions to be selected to create
1586 * a new revision object.
1587 *
1588 * MCR migration note: this replaces Revision::getQueryInfo
1589 *
1590 * @since 1.31
1591 *
1592 * @param array $options Any combination of the following strings
1593 * - 'page': Join with the page table, and select fields to identify the page
1594 * - 'user': Join with the user table, and select the user name
1595 * - 'text': Join with the text table, and select fields to load page text
1596 *
1597 * @return array With three keys:
1598 * - tables: (string[]) to include in the `$table` to `IDatabase->select()`
1599 * - fields: (string[]) to include in the `$vars` to `IDatabase->select()`
1600 * - joins: (array) to include in the `$join_conds` to `IDatabase->select()`
1601 */
1602 public function getQueryInfo( $options = [] ) {
1603 $ret = [
1604 'tables' => [],
1605 'fields' => [],
1606 'joins' => [],
1607 ];
1608
1609 $ret['tables'][] = 'revision';
1610 $ret['fields'] = array_merge( $ret['fields'], [
1611 'rev_id',
1612 'rev_page',
1613 'rev_text_id',
1614 'rev_timestamp',
1615 'rev_minor_edit',
1616 'rev_deleted',
1617 'rev_len',
1618 'rev_parent_id',
1619 'rev_sha1',
1620 ] );
1621
1622 $commentQuery = $this->commentStore->getJoin( 'rev_comment' );
1623 $ret['tables'] = array_merge( $ret['tables'], $commentQuery['tables'] );
1624 $ret['fields'] = array_merge( $ret['fields'], $commentQuery['fields'] );
1625 $ret['joins'] = array_merge( $ret['joins'], $commentQuery['joins'] );
1626
1627 $actorQuery = $this->actorMigration->getJoin( 'rev_user' );
1628 $ret['tables'] = array_merge( $ret['tables'], $actorQuery['tables'] );
1629 $ret['fields'] = array_merge( $ret['fields'], $actorQuery['fields'] );
1630 $ret['joins'] = array_merge( $ret['joins'], $actorQuery['joins'] );
1631
1632 if ( $this->contentHandlerUseDB ) {
1633 $ret['fields'][] = 'rev_content_format';
1634 $ret['fields'][] = 'rev_content_model';
1635 }
1636
1637 if ( in_array( 'page', $options, true ) ) {
1638 $ret['tables'][] = 'page';
1639 $ret['fields'] = array_merge( $ret['fields'], [
1640 'page_namespace',
1641 'page_title',
1642 'page_id',
1643 'page_latest',
1644 'page_is_redirect',
1645 'page_len',
1646 ] );
1647 $ret['joins']['page'] = [ 'INNER JOIN', [ 'page_id = rev_page' ] ];
1648 }
1649
1650 if ( in_array( 'user', $options, true ) ) {
1651 $ret['tables'][] = 'user';
1652 $ret['fields'] = array_merge( $ret['fields'], [
1653 'user_name',
1654 ] );
1655 $u = $actorQuery['fields']['rev_user'];
1656 $ret['joins']['user'] = [ 'LEFT JOIN', [ "$u != 0", "user_id = $u" ] ];
1657 }
1658
1659 if ( in_array( 'text', $options, true ) ) {
1660 $ret['tables'][] = 'text';
1661 $ret['fields'] = array_merge( $ret['fields'], [
1662 'old_text',
1663 'old_flags'
1664 ] );
1665 $ret['joins']['text'] = [ 'INNER JOIN', [ 'rev_text_id=old_id' ] ];
1666 }
1667
1668 return $ret;
1669 }
1670
1671 /**
1672 * Return the tables, fields, and join conditions to be selected to create
1673 * a new archived revision object.
1674 *
1675 * MCR migration note: this replaces Revision::getArchiveQueryInfo
1676 *
1677 * @since 1.31
1678 *
1679 * @return array With three keys:
1680 * - tables: (string[]) to include in the `$table` to `IDatabase->select()`
1681 * - fields: (string[]) to include in the `$vars` to `IDatabase->select()`
1682 * - joins: (array) to include in the `$join_conds` to `IDatabase->select()`
1683 */
1684 public function getArchiveQueryInfo() {
1685 $commentQuery = $this->commentStore->getJoin( 'ar_comment' );
1686 $actorQuery = $this->actorMigration->getJoin( 'ar_user' );
1687 $ret = [
1688 'tables' => [ 'archive' ] + $commentQuery['tables'] + $actorQuery['tables'],
1689 'fields' => [
1690 'ar_id',
1691 'ar_page_id',
1692 'ar_namespace',
1693 'ar_title',
1694 'ar_rev_id',
1695 'ar_text',
1696 'ar_text_id',
1697 'ar_timestamp',
1698 'ar_minor_edit',
1699 'ar_deleted',
1700 'ar_len',
1701 'ar_parent_id',
1702 'ar_sha1',
1703 ] + $commentQuery['fields'] + $actorQuery['fields'],
1704 'joins' => $commentQuery['joins'] + $actorQuery['joins'],
1705 ];
1706
1707 if ( $this->contentHandlerUseDB ) {
1708 $ret['fields'][] = 'ar_content_format';
1709 $ret['fields'][] = 'ar_content_model';
1710 }
1711
1712 return $ret;
1713 }
1714
1715 /**
1716 * Do a batched query for the sizes of a set of revisions.
1717 *
1718 * MCR migration note: this replaces Revision::getParentLengths
1719 *
1720 * @param int[] $revIds
1721 * @return int[] associative array mapping revision IDs from $revIds to the nominal size
1722 * of the corresponding revision.
1723 */
1724 public function getRevisionSizes( array $revIds ) {
1725 return $this->listRevisionSizes( $this->getDBConnection( DB_REPLICA ), $revIds );
1726 }
1727
1728 /**
1729 * Do a batched query for the sizes of a set of revisions.
1730 *
1731 * MCR migration note: this replaces Revision::getParentLengths
1732 *
1733 * @deprecated use RevisionStore::getRevisionSizes instead.
1734 *
1735 * @param IDatabase $db
1736 * @param int[] $revIds
1737 * @return int[] associative array mapping revision IDs from $revIds to the nominal size
1738 * of the corresponding revision.
1739 */
1740 public function listRevisionSizes( IDatabase $db, array $revIds ) {
1741 $this->checkDatabaseWikiId( $db );
1742
1743 $revLens = [];
1744 if ( !$revIds ) {
1745 return $revLens; // empty
1746 }
1747
1748 $res = $db->select(
1749 'revision',
1750 [ 'rev_id', 'rev_len' ],
1751 [ 'rev_id' => $revIds ],
1752 __METHOD__
1753 );
1754
1755 foreach ( $res as $row ) {
1756 $revLens[$row->rev_id] = intval( $row->rev_len );
1757 }
1758
1759 return $revLens;
1760 }
1761
1762 /**
1763 * Get previous revision for this title
1764 *
1765 * MCR migration note: this replaces Revision::getPrevious
1766 *
1767 * @param RevisionRecord $rev
1768 * @param Title $title if known (optional)
1769 *
1770 * @return RevisionRecord|null
1771 */
1772 public function getPreviousRevision( RevisionRecord $rev, Title $title = null ) {
1773 if ( $title === null ) {
1774 $title = $this->getTitle( $rev->getPageId(), $rev->getId() );
1775 }
1776 $prev = $title->getPreviousRevisionID( $rev->getId() );
1777 if ( $prev ) {
1778 return $this->getRevisionByTitle( $title, $prev );
1779 }
1780 return null;
1781 }
1782
1783 /**
1784 * Get next revision for this title
1785 *
1786 * MCR migration note: this replaces Revision::getNext
1787 *
1788 * @param RevisionRecord $rev
1789 * @param Title $title if known (optional)
1790 *
1791 * @return RevisionRecord|null
1792 */
1793 public function getNextRevision( RevisionRecord $rev, Title $title = null ) {
1794 if ( $title === null ) {
1795 $title = $this->getTitle( $rev->getPageId(), $rev->getId() );
1796 }
1797 $next = $title->getNextRevisionID( $rev->getId() );
1798 if ( $next ) {
1799 return $this->getRevisionByTitle( $title, $next );
1800 }
1801 return null;
1802 }
1803
1804 /**
1805 * Get previous revision Id for this page_id
1806 * This is used to populate rev_parent_id on save
1807 *
1808 * MCR migration note: this corresponds to Revision::getPreviousRevisionId
1809 *
1810 * @param IDatabase $db
1811 * @param RevisionRecord $rev
1812 *
1813 * @return int
1814 */
1815 private function getPreviousRevisionId( IDatabase $db, RevisionRecord $rev ) {
1816 $this->checkDatabaseWikiId( $db );
1817
1818 if ( $rev->getPageId() === null ) {
1819 return 0;
1820 }
1821 # Use page_latest if ID is not given
1822 if ( !$rev->getId() ) {
1823 $prevId = $db->selectField(
1824 'page', 'page_latest',
1825 [ 'page_id' => $rev->getPageId() ],
1826 __METHOD__
1827 );
1828 } else {
1829 $prevId = $db->selectField(
1830 'revision', 'rev_id',
1831 [ 'rev_page' => $rev->getPageId(), 'rev_id < ' . $rev->getId() ],
1832 __METHOD__,
1833 [ 'ORDER BY' => 'rev_id DESC' ]
1834 );
1835 }
1836 return intval( $prevId );
1837 }
1838
1839 /**
1840 * Get rev_timestamp from rev_id, without loading the rest of the row
1841 *
1842 * MCR migration note: this replaces Revision::getTimestampFromId
1843 *
1844 * @param Title $title
1845 * @param int $id
1846 * @param int $flags
1847 * @return string|bool False if not found
1848 */
1849 public function getTimestampFromId( $title, $id, $flags = 0 ) {
1850 $db = $this->getDBConnection(
1851 ( $flags & IDBAccessObject::READ_LATEST ) ? DB_MASTER : DB_REPLICA
1852 );
1853
1854 $conds = [ 'rev_id' => $id ];
1855 $conds['rev_page'] = $title->getArticleID();
1856 $timestamp = $db->selectField( 'revision', 'rev_timestamp', $conds, __METHOD__ );
1857
1858 $this->releaseDBConnection( $db );
1859 return ( $timestamp !== false ) ? wfTimestamp( TS_MW, $timestamp ) : false;
1860 }
1861
1862 /**
1863 * Get count of revisions per page...not very efficient
1864 *
1865 * MCR migration note: this replaces Revision::countByPageId
1866 *
1867 * @param IDatabase $db
1868 * @param int $id Page id
1869 * @return int
1870 */
1871 public function countRevisionsByPageId( IDatabase $db, $id ) {
1872 $this->checkDatabaseWikiId( $db );
1873
1874 $row = $db->selectRow( 'revision',
1875 [ 'revCount' => 'COUNT(*)' ],
1876 [ 'rev_page' => $id ],
1877 __METHOD__
1878 );
1879 if ( $row ) {
1880 return intval( $row->revCount );
1881 }
1882 return 0;
1883 }
1884
1885 /**
1886 * Get count of revisions per page...not very efficient
1887 *
1888 * MCR migration note: this replaces Revision::countByTitle
1889 *
1890 * @param IDatabase $db
1891 * @param Title $title
1892 * @return int
1893 */
1894 public function countRevisionsByTitle( IDatabase $db, $title ) {
1895 $id = $title->getArticleID();
1896 if ( $id ) {
1897 return $this->countRevisionsByPageId( $db, $id );
1898 }
1899 return 0;
1900 }
1901
1902 /**
1903 * Check if no edits were made by other users since
1904 * the time a user started editing the page. Limit to
1905 * 50 revisions for the sake of performance.
1906 *
1907 * MCR migration note: this replaces Revision::userWasLastToEdit
1908 *
1909 * @deprecated since 1.31; Can possibly be removed, since the self-conflict suppression
1910 * logic in EditPage that uses this seems conceptually dubious. Revision::userWasLastToEdit
1911 * has been deprecated since 1.24.
1912 *
1913 * @param IDatabase $db The Database to perform the check on.
1914 * @param int $pageId The ID of the page in question
1915 * @param int $userId The ID of the user in question
1916 * @param string $since Look at edits since this time
1917 *
1918 * @return bool True if the given user was the only one to edit since the given timestamp
1919 */
1920 public function userWasLastToEdit( IDatabase $db, $pageId, $userId, $since ) {
1921 $this->checkDatabaseWikiId( $db );
1922
1923 if ( !$userId ) {
1924 return false;
1925 }
1926
1927 $revQuery = self::getQueryInfo();
1928 $res = $db->select(
1929 $revQuery['tables'],
1930 [
1931 'rev_user' => $revQuery['fields']['rev_user'],
1932 ],
1933 [
1934 'rev_page' => $pageId,
1935 'rev_timestamp > ' . $db->addQuotes( $db->timestamp( $since ) )
1936 ],
1937 __METHOD__,
1938 [ 'ORDER BY' => 'rev_timestamp ASC', 'LIMIT' => 50 ],
1939 $revQuery['joins']
1940 );
1941 foreach ( $res as $row ) {
1942 if ( $row->rev_user != $userId ) {
1943 return false;
1944 }
1945 }
1946 return true;
1947 }
1948
1949 /**
1950 * Load a revision based on a known page ID and current revision ID from the DB
1951 *
1952 * This method allows for the use of caching, though accessing anything that normally
1953 * requires permission checks (aside from the text) will trigger a small DB lookup.
1954 *
1955 * MCR migration note: this replaces Revision::newKnownCurrent
1956 *
1957 * @param Title $title the associated page title
1958 * @param int $revId current revision of this page. Defaults to $title->getLatestRevID().
1959 *
1960 * @return RevisionRecord|bool Returns false if missing
1961 */
1962 public function getKnownCurrentRevision( Title $title, $revId ) {
1963 $db = $this->getDBConnectionRef( DB_REPLICA );
1964
1965 $pageId = $title->getArticleID();
1966
1967 if ( !$pageId ) {
1968 return false;
1969 }
1970
1971 if ( !$revId ) {
1972 $revId = $title->getLatestRevID();
1973 }
1974
1975 if ( !$revId ) {
1976 wfWarn(
1977 'No latest revision known for page ' . $title->getPrefixedDBkey()
1978 . ' even though it exists with page ID ' . $pageId
1979 );
1980 return false;
1981 }
1982
1983 $row = $this->cache->getWithSetCallback(
1984 // Page/rev IDs passed in from DB to reflect history merges
1985 $this->cache->makeGlobalKey( 'revision-row-1.29', $db->getDomainID(), $pageId, $revId ),
1986 WANObjectCache::TTL_WEEK,
1987 function ( $curValue, &$ttl, array &$setOpts ) use ( $db, $pageId, $revId ) {
1988 $setOpts += Database::getCacheSetOptions( $db );
1989
1990 $conds = [
1991 'rev_page' => intval( $pageId ),
1992 'page_id' => intval( $pageId ),
1993 'rev_id' => intval( $revId ),
1994 ];
1995
1996 $row = $this->fetchRevisionRowFromConds( $db, $conds );
1997 return $row ?: false; // don't cache negatives
1998 }
1999 );
2000
2001 // Reflect revision deletion and user renames
2002 if ( $row ) {
2003 return $this->newRevisionFromRow( $row, 0, $title );
2004 } else {
2005 return false;
2006 }
2007 }
2008
2009 // TODO: move relevant methods from Title here, e.g. getFirstRevision, isBigDeletion, etc.
2010
2011 }