MediaWiki master
ImportableOldRevisionImporter.php
Go to the documentation of this file.
1<?php
2
15use Psr\Log\LoggerInterface;
19
24
25 private bool $doUpdates;
26 private LoggerInterface $logger;
27 private IConnectionProvider $dbProvider;
28 private RevisionStore $revisionStore;
29 private SlotRoleRegistry $slotRoleRegistry;
30 private WikiPageFactory $wikiPageFactory;
31 private PageUpdaterFactory $pageUpdaterFactory;
32 private UserFactory $userFactory;
33
34 public function __construct(
35 $doUpdates,
36 LoggerInterface $logger,
37 IConnectionProvider $dbProvider,
38 RevisionStore $revisionStore,
39 SlotRoleRegistry $slotRoleRegistry,
40 ?WikiPageFactory $wikiPageFactory = null,
41 ?PageUpdaterFactory $pageUpdaterFactory = null,
42 ?UserFactory $userFactory = null
43 ) {
44 $this->doUpdates = $doUpdates;
45 $this->logger = $logger;
46 $this->dbProvider = $dbProvider;
47 $this->revisionStore = $revisionStore;
48 $this->slotRoleRegistry = $slotRoleRegistry;
49
50 $services = MediaWikiServices::getInstance();
51 // @todo: temporary - remove when FileImporter extension is updated
52 $this->wikiPageFactory = $wikiPageFactory ?? $services->getWikiPageFactory();
53 $this->pageUpdaterFactory = $pageUpdaterFactory ?? $services->getPageUpdaterFactory();
54 $this->userFactory = $userFactory ?? $services->getUserFactory();
55 }
56
58 public function import( ImportableOldRevision $importableRevision, $doUpdates = true ) {
59 $dbw = $this->dbProvider->getPrimaryDatabase();
60
61 # Sneak a single revision into place
62 $user = $importableRevision->getUserObj() ?: $this->userFactory->newFromName( $importableRevision->getUser() );
63 if ( $user ) {
64 $userId = $user->getId();
65 $userText = $user->getName();
66 } else {
67 $userId = 0;
68 $userText = $importableRevision->getUser();
69 $user = $this->userFactory->newAnonymous();
70 }
71
72 // avoid memory leak...?
73 Title::clearCaches();
74
75 $page = $this->wikiPageFactory->newFromTitle( $importableRevision->getTitle() );
76 $page->loadPageData( IDBAccessObject::READ_LATEST );
77 $mustCreatePage = !$page->exists();
78 if ( $mustCreatePage ) {
79 $pageId = $page->insertOn( $dbw );
80 } else {
81 $pageId = $page->getId();
82
83 // Note: sha1 has been in XML dumps since 2012. If you have an
84 // older dump, the duplicate detection here won't work.
85 if ( $importableRevision->getSha1Base36() !== false ) {
86 $prior = (bool)$dbw->newSelectQueryBuilder()
87 ->select( '1' )
88 ->from( 'revision' )
89 ->where( [
90 'rev_page' => $pageId,
91 'rev_timestamp' => $dbw->timestamp( $importableRevision->getTimestamp() ),
92 'rev_sha1' => $importableRevision->getSha1Base36()
93 ] )
94 ->caller( __METHOD__ )->fetchField();
95 if ( $prior ) {
96 // @todo FIXME: This could fail slightly for multiple matches :P
97 $this->logger->debug( __METHOD__ . ": skipping existing revision for [[" .
98 $importableRevision->getTitle()->getPrefixedText() . "]], timestamp " .
99 $importableRevision->getTimestamp() . "\n" );
100 return false;
101 }
102 }
103 }
104
105 if ( !$pageId ) {
106 // This seems to happen if two clients simultaneously try to import the
107 // same page
108 $this->logger->debug( __METHOD__ . ': got invalid $pageId when importing revision of [[' .
109 $importableRevision->getTitle()->getPrefixedText() . ']], timestamp ' .
110 $importableRevision->getTimestamp() . "\n" );
111 return false;
112 }
113
114 // Select previous version to make size diffs correct
115 // @todo This assumes that multiple revisions of the same page are imported
116 // in order from oldest to newest.
117 $queryBuilder = $this->revisionStore->newSelectQueryBuilder( $dbw )
118 ->joinComment()
119 ->where( [ 'rev_page' => $pageId ] )
120 ->andWhere( $dbw->expr(
121 'rev_timestamp', '<=', $dbw->timestamp( $importableRevision->getTimestamp() )
122 ) )
123 ->orderBy( [ 'rev_timestamp', 'rev_id' ], SelectQueryBuilder::SORT_DESC );
124 $prevRevRow = $queryBuilder->caller( __METHOD__ )->fetchRow();
125
126 # @todo FIXME: Use original rev_id optionally (better for backups)
127 # Insert the row
128 $revisionRecord = new MutableRevisionRecord( $importableRevision->getTitle() );
129 $revisionRecord->setParentId( $prevRevRow ? (int)$prevRevRow->rev_id : 0 );
130 $revisionRecord->setComment(
131 CommentStoreComment::newUnsavedComment( $importableRevision->getComment() )
132 );
133
134 try {
135 $revUser = $this->userFactory->newFromAnyId( $userId, $userText );
136 } catch ( InvalidArgumentException $ex ) {
137 $revUser = RequestContext::getMain()->getUser();
138 }
139 $revisionRecord->setUser( $revUser );
140
141 $originalRevision = $prevRevRow
142 ? $this->revisionStore->newRevisionFromRow(
143 $prevRevRow,
144 IDBAccessObject::READ_LATEST,
145 $importableRevision->getTitle()
146 )
147 : null;
148
149 foreach ( $importableRevision->getSlotRoles() as $role ) {
150 if ( !$this->slotRoleRegistry->isDefinedRole( $role ) ) {
151 throw new RuntimeException( "Undefined slot role $role" );
152 }
153
154 $newContent = $importableRevision->getContent( $role );
155 if ( !$originalRevision || !$originalRevision->hasSlot( $role ) ) {
156 $revisionRecord->setContent( $role, $newContent );
157 } else {
158 $originalSlot = $originalRevision->getSlot( $role );
159 if ( !$originalSlot->hasSameContent( $importableRevision->getSlot( $role ) ) ) {
160 $revisionRecord->setContent( $role, $newContent );
161 } else {
162 $revisionRecord->inheritSlot( $originalRevision->getSlot( $role ) );
163 }
164 }
165 }
166
167 $revisionRecord->setTimestamp( $importableRevision->getTimestamp() );
168 $revisionRecord->setMinorEdit( $importableRevision->getMinor() );
169 $revisionRecord->setPageId( $pageId );
170
171 $updater = $this->pageUpdaterFactory->newDerivedPageDataUpdater( $page );
172 $latestRev = $updater->grabCurrentRevision();
173 $latestRevId = $latestRev ? $latestRev->getId() : null;
174
175 $inserted = $this->revisionStore->insertRevisionOn( $revisionRecord, $dbw );
176 if ( $latestRev ) {
177 // If not found (false), cast to 0 so that the page is updated
178 // Just to be on the safe side, even though it should always be found
179 $latestRevTimestamp = $latestRev->getTimestamp();
180 } else {
181 $latestRevTimestamp = 0;
182 }
183 if ( $importableRevision->getTimestamp() >= $latestRevTimestamp ) {
184 $changed = $page->updateRevisionOn( $dbw, $inserted, $latestRevId );
185 } else {
186 $changed = false;
187 }
188
189 $tags = $importableRevision->getTags();
190 if ( $tags !== [] ) {
191 MediaWikiServices::getInstance()->getChangeTagsStore()->addTags( $tags, null, $inserted->getId() );
192 }
193
194 if ( $changed !== false && $this->doUpdates ) {
195 $this->logger->debug( __METHOD__ . ": running updates" );
196 // countable/oldcountable stuff is handled in WikiImporter::finishImportPage
197
198 $options = [
199 PageUpdatedEvent::FLAG_SILENT => true,
200 PageUpdatedEvent::FLAG_AUTOMATED => true,
201 'created' => $mustCreatePage,
202 'oldcountable' => 'no-change',
203 ];
204
205 $updater->setCause( PageUpdater::CAUSE_IMPORT );
206 $updater->setPerformer( $user ); // TODO: get the actual performer, not the revision author.
207 $updater->prepareUpdate( $inserted, $options );
208 $updater->doUpdates();
209 }
210
211 return true;
212 }
213
214}
__construct( $doUpdates, LoggerInterface $logger, IConnectionProvider $dbProvider, RevisionStore $revisionStore, SlotRoleRegistry $slotRoleRegistry, ?WikiPageFactory $wikiPageFactory=null, ?PageUpdaterFactory $pageUpdaterFactory=null, ?UserFactory $userFactory=null)
Value object for a comment stored by CommentStore.
Group all the pieces relevant to the context of a request into one instance.
Service locator for MediaWiki core services.
Domain event representing a page edit.
Service for creating WikiPage objects.
Service for looking up page revisions.
A registry service for SlotRoleHandlers, used to define which slot roles are available on which page.
A factory for PageUpdater and DerivedPageDataUpdater instances.
Controller-like object for creating and updating pages by creating new revisions.
Represents a title within MediaWiki.
Definition Title.php:78
Create User objects.
Build SELECT queries with a fluent interface.
getContent( $role=SlotRecord::MAIN)
Provide primary and replica IDatabase connections.
Interface for database access objects.