MediaWiki master
ImportableOldRevisionImporter.php
Go to the documentation of this file.
1<?php
2
13use Psr\Log\LoggerInterface;
16
21
22 private bool $doUpdates;
23 private LoggerInterface $logger;
24 private IConnectionProvider $dbProvider;
25 private RevisionStore $revisionStore;
26 private SlotRoleRegistry $slotRoleRegistry;
27 private WikiPageFactory $wikiPageFactory;
28 private PageUpdaterFactory $pageUpdaterFactory;
29 private UserFactory $userFactory;
30
31 public function __construct(
32 $doUpdates,
33 LoggerInterface $logger,
34 IConnectionProvider $dbProvider,
35 RevisionStore $revisionStore,
36 SlotRoleRegistry $slotRoleRegistry,
37 WikiPageFactory $wikiPageFactory = null,
38 PageUpdaterFactory $pageUpdaterFactory = null,
39 UserFactory $userFactory = null
40 ) {
41 $this->doUpdates = $doUpdates;
42 $this->logger = $logger;
43 $this->dbProvider = $dbProvider;
44 $this->revisionStore = $revisionStore;
45 $this->slotRoleRegistry = $slotRoleRegistry;
46
47 $services = MediaWikiServices::getInstance();
48 // @todo: temporary - remove when FileImporter extension is updated
49 $this->wikiPageFactory = $wikiPageFactory ?? $services->getWikiPageFactory();
50 $this->pageUpdaterFactory = $pageUpdaterFactory ?? $services->getPageUpdaterFactory();
51 $this->userFactory = $userFactory ?? $services->getUserFactory();
52 }
53
55 public function import( ImportableOldRevision $importableRevision, $doUpdates = true ) {
56 $dbw = $this->dbProvider->getPrimaryDatabase();
57
58 # Sneak a single revision into place
59 $user = $importableRevision->getUserObj() ?: $this->userFactory->newFromName( $importableRevision->getUser() );
60 if ( $user ) {
61 $userId = $user->getId();
62 $userText = $user->getName();
63 } else {
64 $userId = 0;
65 $userText = $importableRevision->getUser();
66 $user = $this->userFactory->newAnonymous();
67 }
68
69 // avoid memory leak...?
70 Title::clearCaches();
71
72 $page = $this->wikiPageFactory->newFromTitle( $importableRevision->getTitle() );
73 $page->loadPageData( IDBAccessObject::READ_LATEST );
74 $mustCreatePage = !$page->exists();
75 if ( $mustCreatePage ) {
76 $pageId = $page->insertOn( $dbw );
77 } else {
78 $pageId = $page->getId();
79
80 // Note: sha1 has been in XML dumps since 2012. If you have an
81 // older dump, the duplicate detection here won't work.
82 if ( $importableRevision->getSha1Base36() !== false ) {
83 $prior = (bool)$dbw->newSelectQueryBuilder()
84 ->select( '1' )
85 ->from( 'revision' )
86 ->where( [
87 'rev_page' => $pageId,
88 'rev_timestamp' => $dbw->timestamp( $importableRevision->getTimestamp() ),
89 'rev_sha1' => $importableRevision->getSha1Base36()
90 ] )
91 ->caller( __METHOD__ )->fetchField();
92 if ( $prior ) {
93 // @todo FIXME: This could fail slightly for multiple matches :P
94 $this->logger->debug( __METHOD__ . ": skipping existing revision for [[" .
95 $importableRevision->getTitle()->getPrefixedText() . "]], timestamp " .
96 $importableRevision->getTimestamp() . "\n" );
97 return false;
98 }
99 }
100 }
101
102 if ( !$pageId ) {
103 // This seems to happen if two clients simultaneously try to import the
104 // same page
105 $this->logger->debug( __METHOD__ . ': got invalid $pageId when importing revision of [[' .
106 $importableRevision->getTitle()->getPrefixedText() . ']], timestamp ' .
107 $importableRevision->getTimestamp() . "\n" );
108 return false;
109 }
110
111 // Select previous version to make size diffs correct
112 // @todo This assumes that multiple revisions of the same page are imported
113 // in order from oldest to newest.
114 $queryBuilder = $this->revisionStore->newSelectQueryBuilder( $dbw )
115 ->joinComment()
116 ->where( [ 'rev_page' => $pageId ] )
117 ->andWhere( $dbw->expr(
118 'rev_timestamp', '<=', $dbw->timestamp( $importableRevision->getTimestamp() )
119 ) )
120 ->orderBy( [ 'rev_timestamp', 'rev_id' ], SelectQueryBuilder::SORT_DESC );
121 $prevRevRow = $queryBuilder->caller( __METHOD__ )->fetchRow();
122
123 # @todo FIXME: Use original rev_id optionally (better for backups)
124 # Insert the row
125 $revisionRecord = new MutableRevisionRecord( $importableRevision->getTitle() );
126 $revisionRecord->setParentId( $prevRevRow ? (int)$prevRevRow->rev_id : 0 );
127 $revisionRecord->setComment(
128 CommentStoreComment::newUnsavedComment( $importableRevision->getComment() )
129 );
130
131 try {
132 $revUser = $this->userFactory->newFromAnyId( $userId, $userText );
133 } catch ( InvalidArgumentException $ex ) {
134 $revUser = RequestContext::getMain()->getUser();
135 }
136 $revisionRecord->setUser( $revUser );
137
138 $originalRevision = $prevRevRow
139 ? $this->revisionStore->newRevisionFromRow(
140 $prevRevRow,
141 IDBAccessObject::READ_LATEST,
142 $importableRevision->getTitle()
143 )
144 : null;
145
146 foreach ( $importableRevision->getSlotRoles() as $role ) {
147 if ( !$this->slotRoleRegistry->isDefinedRole( $role ) ) {
148 throw new RuntimeException( "Undefined slot role $role" );
149 }
150
151 $newContent = $importableRevision->getContent( $role );
152 if ( !$originalRevision || !$originalRevision->hasSlot( $role ) ) {
153 $revisionRecord->setContent( $role, $newContent );
154 } else {
155 $originalSlot = $originalRevision->getSlot( $role );
156 if ( !$originalSlot->hasSameContent( $importableRevision->getSlot( $role ) ) ) {
157 $revisionRecord->setContent( $role, $newContent );
158 } else {
159 $revisionRecord->inheritSlot( $originalRevision->getSlot( $role ) );
160 }
161 }
162 }
163
164 $revisionRecord->setTimestamp( $importableRevision->getTimestamp() );
165 $revisionRecord->setMinorEdit( $importableRevision->getMinor() );
166 $revisionRecord->setPageId( $pageId );
167
168 $latestRevId = $page->getLatest();
169
170 $inserted = $this->revisionStore->insertRevisionOn( $revisionRecord, $dbw );
171 if ( $latestRevId ) {
172 // If not found (false), cast to 0 so that the page is updated
173 // Just to be on the safe side, even though it should always be found
174 $latestRevTimestamp = (int)$this->revisionStore->getTimestampFromId(
175 $latestRevId,
176 IDBAccessObject::READ_LATEST
177 );
178 } else {
179 $latestRevTimestamp = 0;
180 }
181 if ( $importableRevision->getTimestamp() >= $latestRevTimestamp ) {
182 $changed = $page->updateRevisionOn( $dbw, $inserted, $latestRevId );
183 } else {
184 $changed = false;
185 }
186
187 $tags = $importableRevision->getTags();
188 if ( $tags !== [] ) {
189 ChangeTags::addTags( $tags, null, $inserted->getId() );
190 }
191
192 if ( $changed !== false && $this->doUpdates ) {
193 $this->logger->debug( __METHOD__ . ": running updates" );
194 // countable/oldcountable stuff is handled in WikiImporter::finishImportPage
195
196 $options = [
197 'created' => $mustCreatePage,
198 'oldcountable' => 'no-change',
199 'causeAction' => 'import-page',
200 'causeAgent' => $user->getName(),
201 ];
202
203 $updater = $this->pageUpdaterFactory->newDerivedPageDataUpdater( $page );
204 $updater->prepareUpdate( $inserted, $options );
205 $updater->doUpdates();
206 }
207
208 return true;
209 }
210
211}
static addTags( $tags, $rc_id=null, $rev_id=null, $log_id=null, $params=null, RecentChange $rc=null)
Add tags to a change given its rc_id, rev_id and/or log_id.
__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.
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.
Represents a title within MediaWiki.
Definition Title.php:78
Creates User objects.
Build SELECT queries with a fluent interface.
getContent( $role=SlotRecord::MAIN)
Provide primary and replica IDatabase connections.