MediaWiki  master
CategoryMembershipChangeJob.php
Go to the documentation of this file.
1 <?php
29 
45  private $ticket;
46 
47  private const ENQUEUE_FUDGE_SEC = 60;
48 
54  public static function newSpec( PageIdentity $page, $revisionTimestamp ) {
55  return new JobSpecification(
56  'categoryMembershipChange',
57  [
58  'pageId' => $page->getId(),
59  'revTimestamp' => $revisionTimestamp,
60  ],
61  [
62  'removeDuplicates' => true,
63  'removeDuplicatesIgnoreParams' => [ 'revTimestamp' ]
64  ],
65  $page
66  );
67  }
68 
75  public function __construct( PageIdentity $page, array $params ) {
76  parent::__construct( 'categoryMembershipChange', $page, $params );
77  // Only need one job per page. Note that ENQUEUE_FUDGE_SEC handles races where an
78  // older revision job gets inserted while the newer revision job is de-duplicated.
79  $this->removeDuplicates = true;
80  }
81 
82  public function run() {
83  $services = MediaWikiServices::getInstance();
84  $lbFactory = $services->getDBLoadBalancerFactory();
85  $lb = $lbFactory->getMainLB();
86  $dbw = $lb->getConnectionRef( DB_PRIMARY );
87 
88  $this->ticket = $lbFactory->getEmptyTransactionTicket( __METHOD__ );
89 
90  $page = $services->getWikiPageFactory()->newFromID( $this->params['pageId'], WikiPage::READ_LATEST );
91  if ( !$page ) {
92  $this->setLastError( "Could not find page #{$this->params['pageId']}" );
93  return false; // deleted?
94  }
95 
96  // Cut down on the time spent in waitForPrimaryPos() in the critical section
97  $dbr = $lb->getConnectionRef( DB_REPLICA, [ 'recentchanges' ] );
98  if ( !$lb->waitForPrimaryPos( $dbr ) ) {
99  $this->setLastError( "Timed out while pre-waiting for replica DB to catch up" );
100  return false;
101  }
102 
103  // Use a named lock so that jobs for this page see each others' changes
104  $lockKey = "{$dbw->getDomainID()}:CategoryMembershipChange:{$page->getId()}"; // per-wiki
105  $scopedLock = $dbw->getScopedLockAndFlush( $lockKey, __METHOD__, 3 );
106  if ( !$scopedLock ) {
107  $this->setLastError( "Could not acquire lock '$lockKey'" );
108  return false;
109  }
110 
111  // Wait till replica DB is caught up so that jobs for this page see each others' changes
112  if ( !$lb->waitForPrimaryPos( $dbr ) ) {
113  $this->setLastError( "Timed out while waiting for replica DB to catch up" );
114  return false;
115  }
116  // Clear any stale REPEATABLE-READ snapshot
117  $dbr->flushSnapshot( __METHOD__ );
118 
119  $cutoffUnix = wfTimestamp( TS_UNIX, $this->params['revTimestamp'] );
120  // Using ENQUEUE_FUDGE_SEC handles jobs inserted out of revision order due to the delay
121  // between COMMIT and actual enqueueing of the CategoryMembershipChangeJob job.
122  $cutoffUnix -= self::ENQUEUE_FUDGE_SEC;
123 
124  // Get the newest page revision that has a SRC_CATEGORIZE row.
125  // Assume that category changes before it were already handled.
126  $row = $dbr->selectRow(
127  'revision',
128  [ 'rev_timestamp', 'rev_id' ],
129  [
130  'rev_page' => $page->getId(),
131  'rev_timestamp >= ' . $dbr->addQuotes( $dbr->timestamp( $cutoffUnix ) ),
132  'EXISTS (' . $dbr->selectSQLText(
133  'recentchanges',
134  '1',
135  [
136  'rc_this_oldid = rev_id',
137  'rc_source' => RecentChange::SRC_CATEGORIZE,
138  ],
139  __METHOD__
140  ) . ')'
141  ],
142  __METHOD__,
143  [ 'ORDER BY' => [ 'rev_timestamp DESC', 'rev_id DESC' ] ]
144  );
145  // Only consider revisions newer than any such revision
146  if ( $row ) {
147  $cutoffUnix = wfTimestamp( TS_UNIX, $row->rev_timestamp );
148  $lastRevId = (int)$row->rev_id;
149  } else {
150  $lastRevId = 0;
151  }
152 
153  // Find revisions to this page made around and after this revision which lack category
154  // notifications in recent changes. This lets jobs pick up were the last one left off.
155  $encCutoff = $dbr->addQuotes( $dbr->timestamp( $cutoffUnix ) );
156  $revisionStore = $services->getRevisionStore();
157  $revQuery = $revisionStore->getQueryInfo();
158  $res = $dbr->select(
159  $revQuery['tables'],
160  $revQuery['fields'],
161  [
162  'rev_page' => $page->getId(),
163  "rev_timestamp > $encCutoff" .
164  " OR (rev_timestamp = $encCutoff AND rev_id > $lastRevId)"
165  ],
166  __METHOD__,
167  [ 'ORDER BY' => [ 'rev_timestamp ASC', 'rev_id ASC' ] ],
168  $revQuery['joins']
169  );
170 
171  // Apply all category updates in revision timestamp order
172  foreach ( $res as $row ) {
173  $this->notifyUpdatesForRevision( $lbFactory, $page, $revisionStore->newRevisionFromRow( $row ) );
174  }
175 
176  return true;
177  }
178 
185  protected function notifyUpdatesForRevision(
186  LBFactory $lbFactory, WikiPage $page, RevisionRecord $newRev
187  ) {
188  $title = $page->getTitle();
189 
190  // Get the new revision
191  if ( $newRev->isDeleted( RevisionRecord::DELETED_TEXT ) ) {
192  return;
193  }
194 
195  $services = MediaWikiServices::getInstance();
196  // Get the prior revision (the same for null edits)
197  if ( $newRev->getParentId() ) {
198  $oldRev = $services->getRevisionLookup()
199  ->getRevisionById( $newRev->getParentId(), RevisionLookup::READ_LATEST );
200  if ( !$oldRev || $oldRev->isDeleted( RevisionRecord::DELETED_TEXT ) ) {
201  return;
202  }
203  } else {
204  $oldRev = null;
205  }
206 
207  // Parse the new revision and get the categories
208  $categoryChanges = $this->getExplicitCategoriesChanges( $page, $newRev, $oldRev );
209  list( $categoryInserts, $categoryDeletes ) = $categoryChanges;
210  if ( !$categoryInserts && !$categoryDeletes ) {
211  return; // nothing to do
212  }
213 
214  $blc = $services->getBacklinkCacheFactory()->getBacklinkCache( $title );
215  $catMembChange = new CategoryMembershipChange( $title, $blc, $newRev );
216  $catMembChange->checkTemplateLinks();
217 
218  $batchSize = $services->getMainConfig()->get( MainConfigNames::UpdateRowsPerQuery );
219  $insertCount = 0;
220 
221  foreach ( $categoryInserts as $categoryName ) {
222  $categoryTitle = Title::makeTitle( NS_CATEGORY, $categoryName );
223  $catMembChange->triggerCategoryAddedNotification( $categoryTitle );
224  if ( $insertCount++ && ( $insertCount % $batchSize ) == 0 ) {
225  $lbFactory->commitAndWaitForReplication( __METHOD__, $this->ticket );
226  }
227  }
228 
229  foreach ( $categoryDeletes as $categoryName ) {
230  $categoryTitle = Title::makeTitle( NS_CATEGORY, $categoryName );
231  $catMembChange->triggerCategoryRemovedNotification( $categoryTitle );
232  if ( $insertCount++ && ( $insertCount++ % $batchSize ) == 0 ) {
233  $lbFactory->commitAndWaitForReplication( __METHOD__, $this->ticket );
234  }
235  }
236  }
237 
239  WikiPage $page, RevisionRecord $newRev, RevisionRecord $oldRev = null
240  ) {
241  // Inject the same timestamp for both revision parses to avoid seeing category changes
242  // due to time-based parser functions. Inject the same page title for the parses too.
243  // Note that REPEATABLE-READ makes template/file pages appear unchanged between parses.
244  $parseTimestamp = $newRev->getTimestamp();
245  // Parse the old rev and get the categories. Do not use link tables as that
246  // assumes these updates are perfectly FIFO and that link tables are always
247  // up to date, neither of which are true.
248  $oldCategories = $oldRev
249  ? $this->getCategoriesAtRev( $page, $oldRev, $parseTimestamp )
250  : [];
251  // Parse the new revision and get the categories
252  $newCategories = $this->getCategoriesAtRev( $page, $newRev, $parseTimestamp );
253 
254  $categoryInserts = array_values( array_diff( $newCategories, $oldCategories ) );
255  $categoryDeletes = array_values( array_diff( $oldCategories, $newCategories ) );
256 
257  return [ $categoryInserts, $categoryDeletes ];
258  }
259 
267  private function getCategoriesAtRev( WikiPage $page, RevisionRecord $rev, $parseTimestamp ) {
268  $services = MediaWikiServices::getInstance();
269  $options = $page->makeParserOptions( 'canonical' );
270  $options->setTimestamp( $parseTimestamp );
271 
272  $output = $rev instanceof RevisionStoreRecord && $rev->isCurrent()
273  ? $services->getParserCache()->get( $page, $options )
274  : null;
275 
276  if ( !$output || $output->getCacheRevisionId() !== $rev->getId() ) {
277  $output = $services->getRevisionRenderer()->getRenderedRevision( $rev, $options )
278  ->getRevisionParserOutput();
279  }
280 
281  // array keys will cast numeric category names to ints
282  // so we need to cast them back to strings to avoid breaking things!
283  return array_map( 'strval', array_keys( $output->getCategories() ) );
284  }
285 
286  public function getDeduplicationInfo() {
287  $info = parent::getDeduplicationInfo();
288  unset( $info['params']['revTimestamp'] ); // first job wins
289 
290  return $info;
291  }
292 }
const NS_CATEGORY
Definition: Defines.php:78
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.
getCategoriesAtRev(WikiPage $page, RevisionRecord $rev, $parseTimestamp)
getExplicitCategoriesChanges(WikiPage $page, RevisionRecord $newRev, RevisionRecord $oldRev=null)
static newSpec(PageIdentity $page, $revisionTimestamp)
__construct(PageIdentity $page, array $params)
Constructor for use by the Job Queue infrastructure.
getDeduplicationInfo()
Subclasses may need to override this to make duplication detection work.
notifyUpdatesForRevision(LBFactory $lbFactory, WikiPage $page, RevisionRecord $newRev)
Job queue task description base code.
Class to both describe a background job and handle jobs.
Definition: Job.php:39
Title $title
Definition: Job.php:50
setLastError( $error)
Definition: Job.php:469
array $params
Array of job parameters.
Definition: Job.php:44
A class containing constants representing the names of configuration variables.
Service locator for MediaWiki core services.
Page revision base class.
getParentId( $wikiId=self::LOCAL)
Get parent revision ID (the original previous page revision).
isCurrent()
Checks whether the revision record is a stored current revision.
getTimestamp()
MCR migration note: this replaced Revision::getTimestamp.
isDeleted( $field)
MCR migration note: this replaced Revision::isDeleted.
getId( $wikiId=self::LOCAL)
Get revision ID.
A RevisionRecord representing an existing revision persisted in the revision table.
const SRC_CATEGORIZE
static makeTitle( $ns, $title, $fragment='', $interwiki='')
Create a new Title from a namespace index and a DB key.
Definition: Title.php:638
Base representation for an editable wiki page.
Definition: WikiPage.php:62
makeParserOptions( $context)
Get parser options suitable for rendering the primary article wikitext.
Definition: WikiPage.php:1934
getTitle()
Get the title object of the article.
Definition: WikiPage.php:303
An interface for generating database load balancers.
Definition: LBFactory.php:43
commitAndWaitForReplication( $fname, $ticket, array $opts=[])
Call commitPrimaryChanges() and waitForReplication() if $ticket indicates it is safe.
Definition: LBFactory.php:536
Interface for objects (potentially) representing an editable wiki page.
getId( $wikiId=self::LOCAL)
Returns the page ID.
Service for looking up page revisions.
const DB_REPLICA
Definition: defines.php:26
const DB_PRIMARY
Definition: defines.php:28
$revQuery