MediaWiki REL1_30
CategoryMembershipChangeJob.php
Go to the documentation of this file.
1<?php
24
38 private $ticket;
39
41
42 public function __construct( Title $title, array $params ) {
43 parent::__construct( 'categoryMembershipChange', $title, $params );
44 // Only need one job per page. Note that ENQUEUE_FUDGE_SEC handles races where an
45 // older revision job gets inserted while the newer revision job is de-duplicated.
46 $this->removeDuplicates = true;
47 }
48
49 public function run() {
50 $lbFactory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
51 $lb = $lbFactory->getMainLB();
52 $dbw = $lb->getConnection( DB_MASTER );
53
54 $this->ticket = $lbFactory->getEmptyTransactionTicket( __METHOD__ );
55
56 $page = WikiPage::newFromID( $this->params['pageId'], WikiPage::READ_LATEST );
57 if ( !$page ) {
58 $this->setLastError( "Could not find page #{$this->params['pageId']}" );
59 return false; // deleted?
60 }
61
62 // Use a named lock so that jobs for this page see each others' changes
63 $lockKey = "CategoryMembershipUpdates:{$page->getId()}";
64 $scopedLock = $dbw->getScopedLockAndFlush( $lockKey, __METHOD__, 3 );
65 if ( !$scopedLock ) {
66 $this->setLastError( "Could not acquire lock '$lockKey'" );
67 return false;
68 }
69
70 $dbr = $lb->getConnection( DB_REPLICA, [ 'recentchanges' ] );
71 // Wait till the replica DB is caught up so that jobs for this page see each others' changes
72 if ( !$lb->safeWaitForMasterPos( $dbr ) ) {
73 $this->setLastError( "Timed out while waiting for replica DB to catch up" );
74 return false;
75 }
76 // Clear any stale REPEATABLE-READ snapshot
77 $dbr->flushSnapshot( __METHOD__ );
78
79 $cutoffUnix = wfTimestamp( TS_UNIX, $this->params['revTimestamp'] );
80 // Using ENQUEUE_FUDGE_SEC handles jobs inserted out of revision order due to the delay
81 // between COMMIT and actual enqueueing of the CategoryMembershipChangeJob job.
82 $cutoffUnix -= self::ENQUEUE_FUDGE_SEC;
83
84 // Get the newest revision that has a SRC_CATEGORIZE row...
85 $row = $dbr->selectRow(
86 [ 'revision', 'recentchanges' ],
87 [ 'rev_timestamp', 'rev_id' ],
88 [
89 'rev_page' => $page->getId(),
90 'rev_timestamp >= ' . $dbr->addQuotes( $dbr->timestamp( $cutoffUnix ) )
91 ],
92 __METHOD__,
93 [ 'ORDER BY' => 'rev_timestamp DESC, rev_id DESC' ],
94 [
95 'recentchanges' => [
96 'INNER JOIN',
97 [
98 'rc_this_oldid = rev_id',
99 'rc_source' => RecentChange::SRC_CATEGORIZE,
100 // Allow rc_cur_id or rc_timestamp index usage
101 'rc_cur_id = rev_page',
102 'rc_timestamp >= rev_timestamp'
103 ]
104 ]
105 ]
106 );
107 // Only consider revisions newer than any such revision
108 if ( $row ) {
109 $cutoffUnix = wfTimestamp( TS_UNIX, $row->rev_timestamp );
110 $lastRevId = (int)$row->rev_id;
111 } else {
112 $lastRevId = 0;
113 }
114
115 // Find revisions to this page made around and after this revision which lack category
116 // notifications in recent changes. This lets jobs pick up were the last one left off.
117 $encCutoff = $dbr->addQuotes( $dbr->timestamp( $cutoffUnix ) );
118 $res = $dbr->select(
119 'revision',
121 [
122 'rev_page' => $page->getId(),
123 "rev_timestamp > $encCutoff" .
124 " OR (rev_timestamp = $encCutoff AND rev_id > $lastRevId)"
125 ],
126 __METHOD__,
127 [ 'ORDER BY' => 'rev_timestamp ASC, rev_id ASC' ]
128 );
129
130 // Apply all category updates in revision timestamp order
131 foreach ( $res as $row ) {
132 $this->notifyUpdatesForRevision( $lbFactory, $page, Revision::newFromRow( $row ) );
133 }
134
135 return true;
136 }
137
144 protected function notifyUpdatesForRevision(
145 LBFactory $lbFactory, WikiPage $page, Revision $newRev
146 ) {
147 $config = RequestContext::getMain()->getConfig();
148 $title = $page->getTitle();
149
150 // Get the new revision
151 if ( !$newRev->getContent() ) {
152 return; // deleted?
153 }
154
155 // Get the prior revision (the same for null edits)
156 if ( $newRev->getParentId() ) {
157 $oldRev = Revision::newFromId( $newRev->getParentId(), Revision::READ_LATEST );
158 if ( !$oldRev->getContent() ) {
159 return; // deleted?
160 }
161 } else {
162 $oldRev = null;
163 }
164
165 // Parse the new revision and get the categories
166 $categoryChanges = $this->getExplicitCategoriesChanges( $title, $newRev, $oldRev );
167 list( $categoryInserts, $categoryDeletes ) = $categoryChanges;
168 if ( !$categoryInserts && !$categoryDeletes ) {
169 return; // nothing to do
170 }
171
172 $catMembChange = new CategoryMembershipChange( $title, $newRev );
173 $catMembChange->checkTemplateLinks();
174
175 $batchSize = $config->get( 'UpdateRowsPerQuery' );
176 $insertCount = 0;
177
178 foreach ( $categoryInserts as $categoryName ) {
179 $categoryTitle = Title::makeTitle( NS_CATEGORY, $categoryName );
180 $catMembChange->triggerCategoryAddedNotification( $categoryTitle );
181 if ( $insertCount++ && ( $insertCount % $batchSize ) == 0 ) {
182 $lbFactory->commitAndWaitForReplication( __METHOD__, $this->ticket );
183 }
184 }
185
186 foreach ( $categoryDeletes as $categoryName ) {
187 $categoryTitle = Title::makeTitle( NS_CATEGORY, $categoryName );
188 $catMembChange->triggerCategoryRemovedNotification( $categoryTitle );
189 if ( $insertCount++ && ( $insertCount++ % $batchSize ) == 0 ) {
190 $lbFactory->commitAndWaitForReplication( __METHOD__, $this->ticket );
191 }
192 }
193 }
194
196 Title $title, Revision $newRev, Revision $oldRev = null
197 ) {
198 // Inject the same timestamp for both revision parses to avoid seeing category changes
199 // due to time-based parser functions. Inject the same page title for the parses too.
200 // Note that REPEATABLE-READ makes template/file pages appear unchanged between parses.
201 $parseTimestamp = $newRev->getTimestamp();
202 // Parse the old rev and get the categories. Do not use link tables as that
203 // assumes these updates are perfectly FIFO and that link tables are always
204 // up to date, neither of which are true.
205 $oldCategories = $oldRev
206 ? $this->getCategoriesAtRev( $title, $oldRev, $parseTimestamp )
207 : [];
208 // Parse the new revision and get the categories
209 $newCategories = $this->getCategoriesAtRev( $title, $newRev, $parseTimestamp );
210
211 $categoryInserts = array_values( array_diff( $newCategories, $oldCategories ) );
212 $categoryDeletes = array_values( array_diff( $oldCategories, $newCategories ) );
213
214 return [ $categoryInserts, $categoryDeletes ];
215 }
216
224 private function getCategoriesAtRev( Title $title, Revision $rev, $parseTimestamp ) {
225 $content = $rev->getContent();
226 $options = $content->getContentHandler()->makeParserOptions( 'canonical' );
227 $options->setTimestamp( $parseTimestamp );
228 // This could possibly use the parser cache if it checked the revision ID,
229 // but that's more complicated than it's worth.
230 $output = $content->getParserOutput( $title, $rev->getId(), $options );
231
232 // array keys will cast numeric category names to ints
233 // so we need to cast them back to strings to avoid breaking things!
234 return array_map( 'strval', array_keys( $output->getCategories() ) );
235 }
236
237 public function getDeduplicationInfo() {
238 $info = parent::getDeduplicationInfo();
239 unset( $info['params']['revTimestamp'] ); // first job wins
240
241 return $info;
242 }
243}
Apache License January AND DISTRIBUTION Definitions License shall mean the terms and conditions for use
wfTimestamp( $outputtype=TS_UNIX, $ts=0)
Get a timestamp string in one of various formats.
Job to add recent change entries mentioning category membership changes.
getExplicitCategoriesChanges(Title $title, Revision $newRev, Revision $oldRev=null)
__construct(Title $title, array $params)
getDeduplicationInfo()
Subclasses may need to override this to make duplication detection work.
notifyUpdatesForRevision(LBFactory $lbFactory, WikiPage $page, Revision $newRev)
getCategoriesAtRev(Title $title, Revision $rev, $parseTimestamp)
Class to both describe a background job and handle jobs.
Definition Job.php:31
Title $title
Definition Job.php:42
setLastError( $error)
Definition Job.php:403
array $params
Array of job parameters.
Definition Job.php:36
MediaWikiServices is the service locator for the application scope of MediaWiki.
const SRC_CATEGORIZE
static getMain()
Static methods.
static selectFields()
Return the list of revision fields that should be selected to create a new revision.
Definition Revision.php:452
static newFromRow( $row)
Definition Revision.php:238
getContent( $audience=self::FOR_PUBLIC, User $user=null)
Fetch revision content if it's available to the specified audience.
getParentId()
Get parent revision ID (the original previous page revision)
Definition Revision.php:788
static newFromId( $id, $flags=0)
Load a page revision from a given revision ID number.
Definition Revision.php:116
Represents a title within MediaWiki.
Definition Title.php:39
Class representing a MediaWiki article and history.
Definition WikiPage.php:37
static newFromID( $id, $from='fromdb')
Constructor from a page id.
Definition WikiPage.php:159
getTitle()
Get the title object of the article.
Definition WikiPage.php:239
An interface for generating database load balancers.
Definition LBFactory.php:38
commitAndWaitForReplication( $fname, $ticket, array $opts=[])
Convenience method for safely running commitMasterChanges()/waitForReplication()
$res
Definition database.txt:21
deferred txt A few of the database updates required by various functions here can be deferred until after the result page is displayed to the user For updating the view updating the linked to tables after a etc PHP does not yet have any way to tell the server to actually return and disconnect while still running these but it might have such a feature in the future We handle these by creating a deferred update object and putting those objects on a global list
Definition deferred.txt:11
This document is intended to provide useful advice for parties seeking to redistribute MediaWiki to end users It s targeted particularly at maintainers for Linux since it s been observed that distribution packages of MediaWiki often break We ve consistently had to recommend that users seeking support use official tarballs instead of their distribution s and this often solves whatever problem the user is having It would be nice if this could such as
const NS_CATEGORY
Definition Defines.php:79
the array() calling protocol came about after MediaWiki 1.4rc1.
static configuration should be added through ResourceLoaderGetConfigVars instead can be used to get the real title after the basic globals have been set but before ordinary actions take place $output
Definition hooks.txt:2225
null means default in associative array with keys and values unescaped Should be merged with default with a value of false meaning to suppress the attribute in associative array with keys and values unescaped & $options
Definition hooks.txt:1971
presenting them properly to the user as errors is done by the caller return true use this to change the list i e etc $rev
Definition hooks.txt:1760
injection txt This is an overview of how MediaWiki makes use of dependency injection The design described here grew from the discussion of RFC T384 The term dependency this means that anything an object needs to operate should be injected from the the object itself should only know narrow no concrete implementation of the logic it relies on The requirement to inject everything typically results in an architecture that based on two main types of and essentially stateless service objects that use other service objects to operate on the value objects As of the beginning MediaWiki is only starting to use the DI approach Much of the code still relies on global state or direct resulting in a highly cyclical dependency which acts as the top level factory for services in MediaWiki which can be used to gain access to default instances of various services MediaWikiServices however also allows new services to be defined and default services to be redefined Services are defined or redefined by providing a callback the instantiator that will return a new instance of the service When it will create an instance of MediaWikiServices and populate it with the services defined in the files listed by thereby bootstrapping the DI framework Per $wgServiceWiringFiles lists includes ServiceWiring php
Definition injection.txt:37
const DB_REPLICA
Definition defines.php:25
const DB_MASTER
Definition defines.php:26