includes/deferred/LinksUpdate.php

   1 <?php
   2 /**
   3  * Updater for link tracking tables after a page edit.
   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  * @file
  21  */
  22
  23 use Wikimedia\Rdbms\IDatabase;
  24 use MediaWiki\Logger\LoggerFactory;
  25 use MediaWiki\MediaWikiServices;
  26 use Wikimedia\ScopedCallback;
  27
  28 /**
  29  * Class the manages updates of *_link tables as well as similar extension-managed tables
  30  *
  31  * @note: LinksUpdate is managed by DeferredUpdates::execute(). Do not run this in a transaction.
  32  *
  33  * See docs/deferred.txt
  34  */
  35 class LinksUpdate extends DataUpdate implements EnqueueableDataUpdate {
  36         // @todo make members protected, but make sure extensions don't break
  37
  38         /** @var int Page ID of the article linked from */
  39         public $mId;
  40
  41         /** @var Title Title object of the article linked from */
  42         public $mTitle;
  43
  44         /** @var ParserOutput */
  45         public $mParserOutput;
  46
  47         /** @var array Map of title strings to IDs for the links in the document */
  48         public $mLinks;
  49
  50         /** @var array DB keys of the images used, in the array key only */
  51         public $mImages;
  52
  53         /** @var array Map of title strings to IDs for the template references, including broken ones */
  54         public $mTemplates;
  55
  56         /** @var array URLs of external links, array key only */
  57         public $mExternals;
  58
  59         /** @var array Map of category names to sort keys */
  60         public $mCategories;
  61
  62         /** @var array Map of language codes to titles */
  63         public $mInterlangs;
  64
  65         /** @var array 2-D map of (prefix => DBK => 1) */
  66         public $mInterwikis;
  67
  68         /** @var array Map of arbitrary name to value */
  69         public $mProperties;
  70
  71         /** @var bool Whether to queue jobs for recursive updates */
  72         public $mRecursive;
  73
  74         /** @var Revision Revision for which this update has been triggered */
  75         private $mRevision;
  76
  77         /**
  78          * @var null|array Added links if calculated.
  79          */
  80         private $linkInsertions = null;
  81
  82         /**
  83          * @var null|array Deleted links if calculated.
  84          */
  85         private $linkDeletions = null;
  86
  87         /**
  88          * @var null|array Added external links if calculated.
  89          */
  90         private $externalLinkInsertions = null;
  91
  92         /**
  93          * @var null|array Deleted external links if calculated.
  94          */
  95         private $externalLinkDeletions = null;
  96
  97         /**
  98          * @var null|array Added properties if calculated.
  99          */
 100         private $propertyInsertions = null;
 101
 102         /**
 103          * @var null|array Deleted properties if calculated.
 104          */
 105         private $propertyDeletions = null;
 106
 107         /**
 108          * @var User|null
 109          */
 110         private $user;
 111
 112         /** @var IDatabase */
 113         private $db;
 114
 115         /**
 116          * @param Title $title Title of the page we're updating
 117          * @param ParserOutput $parserOutput Output from a full parse of this page
 118          * @param bool $recursive Queue jobs for recursive updates?
 119          * @throws MWException
 120          */
 121         function __construct( Title $title, ParserOutput $parserOutput, $recursive = true ) {
 122                 parent::__construct();
 123
 124                 $this->mTitle = $title;
 125                 $this->mId = $title->getArticleID( Title::GAID_FOR_UPDATE );
 126
 127                 if ( !$this->mId ) {
 128                         throw new InvalidArgumentException(
 129                                 "The Title object yields no ID. Perhaps the page doesn't exist?"
 130                         );
 131                 }
 132
 133                 $this->mParserOutput = $parserOutput;
 134
 135                 $this->mLinks = $parserOutput->getLinks();
 136                 $this->mImages = $parserOutput->getImages();
 137                 $this->mTemplates = $parserOutput->getTemplates();
 138                 $this->mExternals = $parserOutput->getExternalLinks();
 139                 $this->mCategories = $parserOutput->getCategories();
 140                 $this->mProperties = $parserOutput->getProperties();
 141                 $this->mInterwikis = $parserOutput->getInterwikiLinks();
 142
 143                 # Convert the format of the interlanguage links
 144                 # I didn't want to change it in the ParserOutput, because that array is passed all
 145                 # the way back to the skin, so either a skin API break would be required, or an
 146                 # inefficient back-conversion.
 147                 $ill = $parserOutput->getLanguageLinks();
 148                 $this->mInterlangs = [];
 149                 foreach ( $ill as $link ) {
 150                         list( $key, $title ) = explode( ':', $link, 2 );
 151                         $this->mInterlangs[$key] = $title;
 152                 }
 153
 154                 foreach ( $this->mCategories as &$sortkey ) {
 155                         # If the sortkey is longer then 255 bytes, it is truncated by DB, and then doesn't match
 156                         # when comparing existing vs current categories, causing T27254.
 157                         $sortkey = mb_strcut( $sortkey, 0, 255 );
 158                 }
 159
 160                 $this->mRecursive = $recursive;
 161
 162                 // Avoid PHP 7.1 warning from passing $this by reference
 163                 $linksUpdate = $this;
 164                 Hooks::run( 'LinksUpdateConstructed', [ &$linksUpdate ] );
 165         }
 166
 167         /**
 168          * Update link tables with outgoing links from an updated article
 169          *
 170          * @note this is managed by DeferredUpdates::execute(). Do not run this in a transaction.
 171          */
 172         public function doUpdate() {
 173                 if ( $this->ticket ) {
 174                         // Make sure all links update threads see the changes of each other.
 175                         // This handles the case when updates have to batched into several COMMITs.
 176                         $scopedLock = self::acquirePageLock( $this->getDB(), $this->mId );
 177                         if ( !$scopedLock ) {
 178                                 throw new RuntimeException( "Could not acquire lock for page ID '{$this->mId}'." );
 179                         }
 180                 }
 181
 182                 // Avoid PHP 7.1 warning from passing $this by reference
 183                 $linksUpdate = $this;
 184                 Hooks::run( 'LinksUpdate', [ &$linksUpdate ] );
 185                 $this->doIncrementalUpdate();
 186
 187                 // Commit and release the lock (if set)
 188                 ScopedCallback::consume( $scopedLock );
 189                 // Run post-commit hook handlers without DBO_TRX
 190                 DeferredUpdates::addUpdate( new AutoCommitUpdate(
 191                         $this->getDB(),
 192                         __METHOD__,
 193                         function () {
 194                                 // Avoid PHP 7.1 warning from passing $this by reference
 195                                 $linksUpdate = $this;
 196                                 Hooks::run( 'LinksUpdateComplete', [ &$linksUpdate, $this->ticket ] );
 197                         }
 198                 ) );
 199         }
 200
 201         /**
 202          * Acquire a lock for performing link table updates for a page on a DB
 203          *
 204          * @param IDatabase $dbw
 205          * @param int $pageId
 206          * @param string $why One of (job, atomicity)
 207          * @return ScopedCallback|null
 208          * @since 1.27
 209          */
 210         public static function acquirePageLock( IDatabase $dbw, $pageId, $why = 'atomicity' ) {
 211                 $key = "LinksUpdate:$why:pageid:$pageId";
 212                 $scopedLock = $dbw->getScopedLockAndFlush( $key, __METHOD__, 15 );
 213                 if ( !$scopedLock ) {
 214                         $logger = LoggerFactory::getInstance( 'SecondaryDataUpdate' );
 215                         $logger->info( "Could not acquire lock '{key}' for page ID '{page_id}'.", [
 216                                 'key' => $key,
 217                                 'page_id' => $pageId,
 218                         ] );
 219                         return null;
 220                 }
 221
 222                 return $scopedLock;
 223         }
 224
 225         protected function doIncrementalUpdate() {
 226                 # Page links
 227                 $existingPL = $this->getExistingLinks();
 228                 $this->linkDeletions = $this->getLinkDeletions( $existingPL );
 229                 $this->linkInsertions = $this->getLinkInsertions( $existingPL );
 230                 $this->incrTableUpdate( 'pagelinks', 'pl', $this->linkDeletions, $this->linkInsertions );
 231
 232                 # Image links
 233                 $existingIL = $this->getExistingImages();
 234                 $imageDeletes = $this->getImageDeletions( $existingIL );
 235                 $this->incrTableUpdate(
 236                         'imagelinks',
 237                         'il',
 238                         $imageDeletes,
 239                         $this->getImageInsertions( $existingIL ) );
 240
 241                 # Invalidate all image description pages which had links added or removed
 242                 $imageUpdates = $imageDeletes + array_diff_key( $this->mImages, $existingIL );
 243                 $this->invalidateImageDescriptions( $imageUpdates );
 244
 245                 # External links
 246                 $existingEL = $this->getExistingExternals();
 247                 $this->externalLinkDeletions = $this->getExternalDeletions( $existingEL );
 248                 $this->externalLinkInsertions = $this->getExternalInsertions(
 249                         $existingEL );
 250                 $this->incrTableUpdate(
 251                         'externallinks',
 252                         'el',
 253                         $this->externalLinkDeletions,
 254                         $this->externalLinkInsertions );
 255
 256                 # Language links
 257                 $existingLL = $this->getExistingInterlangs();
 258                 $this->incrTableUpdate(
 259                         'langlinks',
 260                         'll',
 261                         $this->getInterlangDeletions( $existingLL ),
 262                         $this->getInterlangInsertions( $existingLL ) );
 263
 264                 # Inline interwiki links
 265                 $existingIW = $this->getExistingInterwikis();
 266                 $this->incrTableUpdate(
 267                         'iwlinks',
 268                         'iwl',
 269                         $this->getInterwikiDeletions( $existingIW ),
 270                         $this->getInterwikiInsertions( $existingIW ) );
 271
 272                 # Template links
 273                 $existingTL = $this->getExistingTemplates();
 274                 $this->incrTableUpdate(
 275                         'templatelinks',
 276                         'tl',
 277                         $this->getTemplateDeletions( $existingTL ),
 278                         $this->getTemplateInsertions( $existingTL ) );
 279
 280                 # Category links
 281                 $existingCL = $this->getExistingCategories();
 282                 $categoryDeletes = $this->getCategoryDeletions( $existingCL );
 283                 $this->incrTableUpdate(
 284                         'categorylinks',
 285                         'cl',
 286                         $categoryDeletes,
 287                         $this->getCategoryInsertions( $existingCL ) );
 288                 $categoryInserts = array_diff_assoc( $this->mCategories, $existingCL );
 289                 $categoryUpdates = $categoryInserts + $categoryDeletes;
 290
 291                 # Page properties
 292                 $existingPP = $this->getExistingProperties();
 293                 $this->propertyDeletions = $this->getPropertyDeletions( $existingPP );
 294                 $this->incrTableUpdate(
 295                         'page_props',
 296                         'pp',
 297                         $this->propertyDeletions,
 298                         $this->getPropertyInsertions( $existingPP ) );
 299
 300                 # Invalidate the necessary pages
 301                 $this->propertyInsertions = array_diff_assoc( $this->mProperties, $existingPP );
 302                 $changed = $this->propertyDeletions + $this->propertyInsertions;
 303                 $this->invalidateProperties( $changed );
 304
 305                 # Invalidate all categories which were added, deleted or changed (set symmetric difference)
 306                 $this->invalidateCategories( $categoryUpdates );
 307                 $this->updateCategoryCounts( $categoryInserts, $categoryDeletes );
 308
 309                 # Refresh links of all pages including this page
 310                 # This will be in a separate transaction
 311                 if ( $this->mRecursive ) {
 312                         $this->queueRecursiveJobs();
 313                 }
 314
 315                 # Update the links table freshness for this title
 316                 $this->updateLinksTimestamp();
 317         }
 318
 319         /**
 320          * Queue recursive jobs for this page
 321          *
 322          * Which means do LinksUpdate on all pages that include the current page,
 323          * using the job queue.
 324          */
 325         protected function queueRecursiveJobs() {
 326                 $action = $this->getCauseAction();
 327                 $agent = $this->getCauseAgent();
 328
 329                 self::queueRecursiveJobsForTable( $this->mTitle, 'templatelinks', $action, $agent );
 330                 if ( $this->mTitle->getNamespace() == NS_FILE ) {
 331                         // Process imagelinks in case the title is or was a redirect
 332                         self::queueRecursiveJobsForTable( $this->mTitle, 'imagelinks', $action, $agent );
 333                 }
 334
 335                 $bc = $this->mTitle->getBacklinkCache();
 336                 // Get jobs for cascade-protected backlinks for a high priority queue.
 337                 // If meta-templates change to using a new template, the new template
 338                 // should be implicitly protected as soon as possible, if applicable.
 339                 // These jobs duplicate a subset of the above ones, but can run sooner.
 340                 // Which ever runs first generally no-ops the other one.
 341                 $jobs = [];
 342                 foreach ( $bc->getCascadeProtectedLinks() as $title ) {
 343                         $jobs[] = RefreshLinksJob::newPrioritized(
 344                                 $title,
 345                                 [
 346                                         'causeAction' => $action,
 347                                         'causeAgent' => $agent
 348                                 ]
 349                         );
 350                 }
 351                 JobQueueGroup::singleton()->push( $jobs );
 352         }
 353
 354         /**
 355          * Queue a RefreshLinks job for any table.
 356          *
 357          * @param Title $title Title to do job for
 358          * @param string $table Table to use (e.g. 'templatelinks')
 359          * @param string $action Triggering action
 360          * @param string $userName Triggering user name
 361          */
 362         public static function queueRecursiveJobsForTable(
 363                 Title $title, $table, $action = 'unknown', $userName = 'unknown'
 364         ) {
 365                 if ( $title->getBacklinkCache()->hasLinks( $table ) ) {
 366                         $job = new RefreshLinksJob(
 367                                 $title,
 368                                 [
 369                                         'table' => $table,
 370                                         'recursive' => true,
 371                                 ] + Job::newRootJobParams( // "overall" refresh links job info
 372                                         "refreshlinks:{$table}:{$title->getPrefixedText()}"
 373                                 ) + [ 'causeAction' => $action, 'causeAgent' => $userName ]
 374                         );
 375
 376                         JobQueueGroup::singleton()->push( $job );
 377                 }
 378         }
 379
 380         /**
 381          * @param array $cats
 382          */
 383         private function invalidateCategories( $cats ) {
 384                 PurgeJobUtils::invalidatePages( $this->getDB(), NS_CATEGORY, array_keys( $cats ) );
 385         }
 386
 387         /**
 388          * Update all the appropriate counts in the category table.
 389          * @param array $added Associative array of category name => sort key
 390          * @param array $deleted Associative array of category name => sort key
 391          */
 392         private function updateCategoryCounts( array $added, array $deleted ) {
 393                 global $wgUpdateRowsPerQuery;
 394
 395                 if ( !$added && !$deleted ) {
 396                         return;
 397                 }
 398
 399                 $domainId = $this->getDB()->getDomainID();
 400                 $wp = WikiPage::factory( $this->mTitle );
 401                 $lbf = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
 402                 // T163801: try to release any row locks to reduce contention
 403                 $lbf->commitAndWaitForReplication( __METHOD__, $this->ticket, [ 'domain' => $domainId ] );
 404
 405                 foreach ( array_chunk( array_keys( $added ), $wgUpdateRowsPerQuery ) as $addBatch ) {
 406                         $wp->updateCategoryCounts( $addBatch, [], $this->mId );
 407                         $lbf->commitAndWaitForReplication(
 408                                 __METHOD__, $this->ticket, [ 'domain' => $domainId ] );
 409                 }
 410
 411                 foreach ( array_chunk( array_keys( $deleted ), $wgUpdateRowsPerQuery ) as $deleteBatch ) {
 412                         $wp->updateCategoryCounts( [], $deleteBatch, $this->mId );
 413                         $lbf->commitAndWaitForReplication(
 414                                 __METHOD__, $this->ticket, [ 'domain' => $domainId ] );
 415                 }
 416         }
 417
 418         /**
 419          * @param array $images
 420          */
 421         private function invalidateImageDescriptions( array $images ) {
 422                 PurgeJobUtils::invalidatePages( $this->getDB(), NS_FILE, array_keys( $images ) );
 423         }
 424
 425         /**
 426          * Update a table by doing a delete query then an insert query
 427          * @param string $table Table name
 428          * @param string $prefix Field name prefix
 429          * @param array $deletions
 430          * @param array $insertions Rows to insert
 431          */
 432         private function incrTableUpdate( $table, $prefix, $deletions, $insertions ) {
 433                 $services = MediaWikiServices::getInstance();
 434                 $bSize = $services->getMainConfig()->get( 'UpdateRowsPerQuery' );
 435                 $lbf = $services->getDBLoadBalancerFactory();
 436
 437                 if ( $table === 'page_props' ) {
 438                         $fromField = 'pp_page';
 439                 } else {
 440                         $fromField = "{$prefix}_from";
 441                 }
 442
 443                 $deleteWheres = []; // list of WHERE clause arrays for each DB delete() call
 444                 if ( $table === 'pagelinks' || $table === 'templatelinks' || $table === 'iwlinks' ) {
 445                         $baseKey = ( $table === 'iwlinks' ) ? 'iwl_prefix' : "{$prefix}_namespace";
 446
 447                         $curBatchSize = 0;
 448                         $curDeletionBatch = [];
 449                         $deletionBatches = [];
 450                         foreach ( $deletions as $ns => $dbKeys ) {
 451                                 foreach ( $dbKeys as $dbKey => $unused ) {
 452                                         $curDeletionBatch[$ns][$dbKey] = 1;
 453                                         if ( ++$curBatchSize >= $bSize ) {
 454                                                 $deletionBatches[] = $curDeletionBatch;
 455                                                 $curDeletionBatch = [];
 456                                                 $curBatchSize = 0;
 457                                         }
 458                                 }
 459                         }
 460                         if ( $curDeletionBatch ) {
 461                                 $deletionBatches[] = $curDeletionBatch;
 462                         }
 463
 464                         foreach ( $deletionBatches as $deletionBatch ) {
 465                                 $deleteWheres[] = [
 466                                         $fromField => $this->mId,
 467                                         $this->getDB()->makeWhereFrom2d( $deletionBatch, $baseKey, "{$prefix}_title" )
 468                                 ];
 469                         }
 470                 } else {
 471                         if ( $table === 'langlinks' ) {
 472                                 $toField = 'll_lang';
 473                         } elseif ( $table === 'page_props' ) {
 474                                 $toField = 'pp_propname';
 475                         } else {
 476                                 $toField = $prefix . '_to';
 477                         }
 478
 479                         $deletionBatches = array_chunk( array_keys( $deletions ), $bSize );
 480                         foreach ( $deletionBatches as $deletionBatch ) {
 481                                 $deleteWheres[] = [ $fromField => $this->mId, $toField => $deletionBatch ];
 482                         }
 483                 }
 484
 485                 $domainId = $this->getDB()->getDomainID();
 486
 487                 foreach ( $deleteWheres as $deleteWhere ) {
 488                         $this->getDB()->delete( $table, $deleteWhere, __METHOD__ );
 489                         $lbf->commitAndWaitForReplication(
 490                                 __METHOD__, $this->ticket, [ 'domain' => $domainId ]
 491                         );
 492                 }
 493
 494                 $insertBatches = array_chunk( $insertions, $bSize );
 495                 foreach ( $insertBatches as $insertBatch ) {
 496                         $this->getDB()->insert( $table, $insertBatch, __METHOD__, 'IGNORE' );
 497                         $lbf->commitAndWaitForReplication(
 498                                 __METHOD__, $this->ticket, [ 'domain' => $domainId ]
 499                         );
 500                 }
 501
 502                 if ( count( $insertions ) ) {
 503                         Hooks::run( 'LinksUpdateAfterInsert', [ $this, $table, $insertions ] );
 504                 }
 505         }
 506
 507         /**
 508          * Get an array of pagelinks insertions for passing to the DB
 509          * Skips the titles specified by the 2-D array $existing
 510          * @param array $existing
 511          * @return array
 512          */
 513         private function getLinkInsertions( $existing = [] ) {
 514                 $arr = [];
 515                 foreach ( $this->mLinks as $ns => $dbkeys ) {
 516                         $diffs = isset( $existing[$ns] )
 517                                 ? array_diff_key( $dbkeys, $existing[$ns] )
 518                                 : $dbkeys;
 519                         foreach ( $diffs as $dbk => $id ) {
 520                                 $arr[] = [
 521                                         'pl_from' => $this->mId,
 522                                         'pl_from_namespace' => $this->mTitle->getNamespace(),
 523                                         'pl_namespace' => $ns,
 524                                         'pl_title' => $dbk
 525                                 ];
 526                         }
 527                 }
 528
 529                 return $arr;
 530         }
 531
 532         /**
 533          * Get an array of template insertions. Like getLinkInsertions()
 534          * @param array $existing
 535          * @return array
 536          */
 537         private function getTemplateInsertions( $existing = [] ) {
 538                 $arr = [];
 539                 foreach ( $this->mTemplates as $ns => $dbkeys ) {
 540                         $diffs = isset( $existing[$ns] ) ? array_diff_key( $dbkeys, $existing[$ns] ) : $dbkeys;
 541                         foreach ( $diffs as $dbk => $id ) {
 542                                 $arr[] = [
 543                                         'tl_from' => $this->mId,
 544                                         'tl_from_namespace' => $this->mTitle->getNamespace(),
 545                                         'tl_namespace' => $ns,
 546                                         'tl_title' => $dbk
 547                                 ];
 548                         }
 549                 }
 550
 551                 return $arr;
 552         }
 553
 554         /**
 555          * Get an array of image insertions
 556          * Skips the names specified in $existing
 557          * @param array $existing
 558          * @return array
 559          */
 560         private function getImageInsertions( $existing = [] ) {
 561                 $arr = [];
 562                 $diffs = array_diff_key( $this->mImages, $existing );
 563                 foreach ( $diffs as $iname => $dummy ) {
 564                         $arr[] = [
 565                                 'il_from' => $this->mId,
 566                                 'il_from_namespace' => $this->mTitle->getNamespace(),
 567                                 'il_to' => $iname
 568                         ];
 569                 }
 570
 571                 return $arr;
 572         }
 573
 574         /**
 575          * Get an array of externallinks insertions. Skips the names specified in $existing
 576          * @param array $existing
 577          * @return array
 578          */
 579         private function getExternalInsertions( $existing = [] ) {
 580                 $arr = [];
 581                 $diffs = array_diff_key( $this->mExternals, $existing );
 582                 foreach ( $diffs as $url => $dummy ) {
 583                         foreach ( LinkFilter::makeIndexes( $url ) as $index ) {
 584                                 $arr[] = [
 585                                         'el_from' => $this->mId,
 586                                         'el_to' => $url,
 587                                         'el_index' => $index,
 588                                         'el_index_60' => substr( $index, 0, 60 ),
 589                                 ];
 590                         }
 591                 }
 592
 593                 return $arr;
 594         }
 595
 596         /**
 597          * Get an array of category insertions
 598          *
 599          * @param array $existing Mapping existing category names to sort keys. If both
 600          * match a link in $this, the link will be omitted from the output
 601          *
 602          * @return array
 603          */
 604         private function getCategoryInsertions( $existing = [] ) {
 605                 global $wgCategoryCollation;
 606                 $diffs = array_diff_assoc( $this->mCategories, $existing );
 607                 $arr = [];
 608                 $contLang = MediaWikiServices::getInstance()->getContentLanguage();
 609                 $collation = Collation::singleton();
 610                 foreach ( $diffs as $name => $prefix ) {
 611                         $nt = Title::makeTitleSafe( NS_CATEGORY, $name );
 612                         $contLang->findVariantLink( $name, $nt, true );
 613
 614                         $type = MWNamespace::getCategoryLinkType( $this->mTitle->getNamespace() );
 615
 616                         # Treat custom sortkeys as a prefix, so that if multiple
 617                         # things are forced to sort as '*' or something, they'll
 618                         # sort properly in the category rather than in page_id
 619                         # order or such.
 620                         $sortkey = $collation->getSortKey( $this->mTitle->getCategorySortkey( $prefix ) );
 621
 622                         $arr[] = [
 623                                 'cl_from' => $this->mId,
 624                                 'cl_to' => $name,
 625                                 'cl_sortkey' => $sortkey,
 626                                 'cl_timestamp' => $this->getDB()->timestamp(),
 627                                 'cl_sortkey_prefix' => $prefix,
 628                                 'cl_collation' => $wgCategoryCollation,
 629                                 'cl_type' => $type,
 630                         ];
 631                 }
 632
 633                 return $arr;
 634         }
 635
 636         /**
 637          * Get an array of interlanguage link insertions
 638          *
 639          * @param array $existing Mapping existing language codes to titles
 640          *
 641          * @return array
 642          */
 643         private function getInterlangInsertions( $existing = [] ) {
 644                 $diffs = array_diff_assoc( $this->mInterlangs, $existing );
 645                 $arr = [];
 646                 foreach ( $diffs as $lang => $title ) {
 647                         $arr[] = [
 648                                 'll_from' => $this->mId,
 649                                 'll_lang' => $lang,
 650                                 'll_title' => $title
 651                         ];
 652                 }
 653
 654                 return $arr;
 655         }
 656
 657         /**
 658          * Get an array of page property insertions
 659          * @param array $existing
 660          * @return array
 661          */
 662         function getPropertyInsertions( $existing = [] ) {
 663                 $diffs = array_diff_assoc( $this->mProperties, $existing );
 664
 665                 $arr = [];
 666                 foreach ( array_keys( $diffs ) as $name ) {
 667                         $arr[] = $this->getPagePropRowData( $name );
 668                 }
 669
 670                 return $arr;
 671         }
 672
 673         /**
 674          * Returns an associative array to be used for inserting a row into
 675          * the page_props table. Besides the given property name, this will
 676          * include the page id from $this->mId and any property value from
 677          * $this->mProperties.
 678          *
 679          * The array returned will include the pp_sortkey field if this
 680          * is present in the database (as indicated by $wgPagePropsHaveSortkey).
 681          * The sortkey value is currently determined by getPropertySortKeyValue().
 682          *
 683          * @note this assumes that $this->mProperties[$prop] is defined.
 684          *
 685          * @param string $prop The name of the property.
 686          *
 687          * @return array
 688          */
 689         private function getPagePropRowData( $prop ) {
 690                 global $wgPagePropsHaveSortkey;
 691
 692                 $value = $this->mProperties[$prop];
 693
 694                 $row = [
 695                         'pp_page' => $this->mId,
 696                         'pp_propname' => $prop,
 697                         'pp_value' => $value,
 698                 ];
 699
 700                 if ( $wgPagePropsHaveSortkey ) {
 701                         $row['pp_sortkey'] = $this->getPropertySortKeyValue( $value );
 702                 }
 703
 704                 return $row;
 705         }
 706
 707         /**
 708          * Determines the sort key for the given property value.
 709          * This will return $value if it is a float or int,
 710          * 1 or resp. 0 if it is a bool, and null otherwise.
 711          *
 712          * @note In the future, we may allow the sortkey to be specified explicitly
 713          *       in ParserOutput::setProperty.
 714          *
 715          * @param mixed $value
 716          *
 717          * @return float|null
 718          */
 719         private function getPropertySortKeyValue( $value ) {
 720                 if ( is_int( $value ) || is_float( $value ) || is_bool( $value ) ) {
 721                         return floatval( $value );
 722                 }
 723
 724                 return null;
 725         }
 726
 727         /**
 728          * Get an array of interwiki insertions for passing to the DB
 729          * Skips the titles specified by the 2-D array $existing
 730          * @param array $existing
 731          * @return array
 732          */
 733         private function getInterwikiInsertions( $existing = [] ) {
 734                 $arr = [];
 735                 foreach ( $this->mInterwikis as $prefix => $dbkeys ) {
 736                         $diffs = isset( $existing[$prefix] )
 737                                 ? array_diff_key( $dbkeys, $existing[$prefix] )
 738                                 : $dbkeys;
 739
 740                         foreach ( $diffs as $dbk => $id ) {
 741                                 $arr[] = [
 742                                         'iwl_from' => $this->mId,
 743                                         'iwl_prefix' => $prefix,
 744                                         'iwl_title' => $dbk
 745                                 ];
 746                         }
 747                 }
 748
 749                 return $arr;
 750         }
 751
 752         /**
 753          * Given an array of existing links, returns those links which are not in $this
 754          * and thus should be deleted.
 755          * @param array $existing
 756          * @return array
 757          */
 758         private function getLinkDeletions( $existing ) {
 759                 $del = [];
 760                 foreach ( $existing as $ns => $dbkeys ) {
 761                         if ( isset( $this->mLinks[$ns] ) ) {
 762                                 $del[$ns] = array_diff_key( $existing[$ns], $this->mLinks[$ns] );
 763                         } else {
 764                                 $del[$ns] = $existing[$ns];
 765                         }
 766                 }
 767
 768                 return $del;
 769         }
 770
 771         /**
 772          * Given an array of existing templates, returns those templates which are not in $this
 773          * and thus should be deleted.
 774          * @param array $existing
 775          * @return array
 776          */
 777         private function getTemplateDeletions( $existing ) {
 778                 $del = [];
 779                 foreach ( $existing as $ns => $dbkeys ) {
 780                         if ( isset( $this->mTemplates[$ns] ) ) {
 781                                 $del[$ns] = array_diff_key( $existing[$ns], $this->mTemplates[$ns] );
 782                         } else {
 783                                 $del[$ns] = $existing[$ns];
 784                         }
 785                 }
 786
 787                 return $del;
 788         }
 789
 790         /**
 791          * Given an array of existing images, returns those images which are not in $this
 792          * and thus should be deleted.
 793          * @param array $existing
 794          * @return array
 795          */
 796         private function getImageDeletions( $existing ) {
 797                 return array_diff_key( $existing, $this->mImages );
 798         }
 799
 800         /**
 801          * Given an array of existing external links, returns those links which are not
 802          * in $this and thus should be deleted.
 803          * @param array $existing
 804          * @return array
 805          */
 806         private function getExternalDeletions( $existing ) {
 807                 return array_diff_key( $existing, $this->mExternals );
 808         }
 809
 810         /**
 811          * Given an array of existing categories, returns those categories which are not in $this
 812          * and thus should be deleted.
 813          * @param array $existing
 814          * @return array
 815          */
 816         private function getCategoryDeletions( $existing ) {
 817                 return array_diff_assoc( $existing, $this->mCategories );
 818         }
 819
 820         /**
 821          * Given an array of existing interlanguage links, returns those links which are not
 822          * in $this and thus should be deleted.
 823          * @param array $existing
 824          * @return array
 825          */
 826         private function getInterlangDeletions( $existing ) {
 827                 return array_diff_assoc( $existing, $this->mInterlangs );
 828         }
 829
 830         /**
 831          * Get array of properties which should be deleted.
 832          * @param array $existing
 833          * @return array
 834          */
 835         private function getPropertyDeletions( $existing ) {
 836                 return array_diff_assoc( $existing, $this->mProperties );
 837         }
 838
 839         /**
 840          * Given an array of existing interwiki links, returns those links which are not in $this
 841          * and thus should be deleted.
 842          * @param array $existing
 843          * @return array
 844          */
 845         private function getInterwikiDeletions( $existing ) {
 846                 $del = [];
 847                 foreach ( $existing as $prefix => $dbkeys ) {
 848                         if ( isset( $this->mInterwikis[$prefix] ) ) {
 849                                 $del[$prefix] = array_diff_key( $existing[$prefix], $this->mInterwikis[$prefix] );
 850                         } else {
 851                                 $del[$prefix] = $existing[$prefix];
 852                         }
 853                 }
 854
 855                 return $del;
 856         }
 857
 858         /**
 859          * Get an array of existing links, as a 2-D array
 860          *
 861          * @return array
 862          */
 863         private function getExistingLinks() {
 864                 $res = $this->getDB()->select( 'pagelinks', [ 'pl_namespace', 'pl_title' ],
 865                         [ 'pl_from' => $this->mId ], __METHOD__ );
 866                 $arr = [];
 867                 foreach ( $res as $row ) {
 868                         if ( !isset( $arr[$row->pl_namespace] ) ) {
 869                                 $arr[$row->pl_namespace] = [];
 870                         }
 871                         $arr[$row->pl_namespace][$row->pl_title] = 1;
 872                 }
 873
 874                 return $arr;
 875         }
 876
 877         /**
 878          * Get an array of existing templates, as a 2-D array
 879          *
 880          * @return array
 881          */
 882         private function getExistingTemplates() {
 883                 $res = $this->getDB()->select( 'templatelinks', [ 'tl_namespace', 'tl_title' ],
 884                         [ 'tl_from' => $this->mId ], __METHOD__ );
 885                 $arr = [];
 886                 foreach ( $res as $row ) {
 887                         if ( !isset( $arr[$row->tl_namespace] ) ) {
 888                                 $arr[$row->tl_namespace] = [];
 889                         }
 890                         $arr[$row->tl_namespace][$row->tl_title] = 1;
 891                 }
 892
 893                 return $arr;
 894         }
 895
 896         /**
 897          * Get an array of existing images, image names in the keys
 898          *
 899          * @return array
 900          */
 901         private function getExistingImages() {
 902                 $res = $this->getDB()->select( 'imagelinks', [ 'il_to' ],
 903                         [ 'il_from' => $this->mId ], __METHOD__ );
 904                 $arr = [];
 905                 foreach ( $res as $row ) {
 906                         $arr[$row->il_to] = 1;
 907                 }
 908
 909                 return $arr;
 910         }
 911
 912         /**
 913          * Get an array of existing external links, URLs in the keys
 914          *
 915          * @return array
 916          */
 917         private function getExistingExternals() {
 918                 $res = $this->getDB()->select( 'externallinks', [ 'el_to' ],
 919                         [ 'el_from' => $this->mId ], __METHOD__ );
 920                 $arr = [];
 921                 foreach ( $res as $row ) {
 922                         $arr[$row->el_to] = 1;
 923                 }
 924
 925                 return $arr;
 926         }
 927
 928         /**
 929          * Get an array of existing categories, with the name in the key and sort key in the value.
 930          *
 931          * @return array
 932          */
 933         private function getExistingCategories() {
 934                 $res = $this->getDB()->select( 'categorylinks', [ 'cl_to', 'cl_sortkey_prefix' ],
 935                         [ 'cl_from' => $this->mId ], __METHOD__ );
 936                 $arr = [];
 937                 foreach ( $res as $row ) {
 938                         $arr[$row->cl_to] = $row->cl_sortkey_prefix;
 939                 }
 940
 941                 return $arr;
 942         }
 943
 944         /**
 945          * Get an array of existing interlanguage links, with the language code in the key and the
 946          * title in the value.
 947          *
 948          * @return array
 949          */
 950         private function getExistingInterlangs() {
 951                 $res = $this->getDB()->select( 'langlinks', [ 'll_lang', 'll_title' ],
 952                         [ 'll_from' => $this->mId ], __METHOD__ );
 953                 $arr = [];
 954                 foreach ( $res as $row ) {
 955                         $arr[$row->ll_lang] = $row->ll_title;
 956                 }
 957
 958                 return $arr;
 959         }
 960
 961         /**
 962          * Get an array of existing inline interwiki links, as a 2-D array
 963          * @return array (prefix => array(dbkey => 1))
 964          */
 965         private function getExistingInterwikis() {
 966                 $res = $this->getDB()->select( 'iwlinks', [ 'iwl_prefix', 'iwl_title' ],
 967                         [ 'iwl_from' => $this->mId ], __METHOD__ );
 968                 $arr = [];
 969                 foreach ( $res as $row ) {
 970                         if ( !isset( $arr[$row->iwl_prefix] ) ) {
 971                                 $arr[$row->iwl_prefix] = [];
 972                         }
 973                         $arr[$row->iwl_prefix][$row->iwl_title] = 1;
 974                 }
 975
 976                 return $arr;
 977         }
 978
 979         /**
 980          * Get an array of existing categories, with the name in the key and sort key in the value.
 981          *
 982          * @return array Array of property names and values
 983          */
 984         private function getExistingProperties() {
 985                 $res = $this->getDB()->select( 'page_props', [ 'pp_propname', 'pp_value' ],
 986                         [ 'pp_page' => $this->mId ], __METHOD__ );
 987                 $arr = [];
 988                 foreach ( $res as $row ) {
 989                         $arr[$row->pp_propname] = $row->pp_value;
 990                 }
 991
 992                 return $arr;
 993         }
 994
 995         /**
 996          * Return the title object of the page being updated
 997          * @return Title
 998          */
 999         public function getTitle() {
1000                 return $this->mTitle;
1001         }
1002
1003         /**
1004          * Returns parser output
1005          * @since 1.19
1006          * @return ParserOutput
1007          */
1008         public function getParserOutput() {
1009                 return $this->mParserOutput;
1010         }
1011
1012         /**
1013          * Return the list of images used as generated by the parser
1014          * @return array
1015          */
1016         public function getImages() {
1017                 return $this->mImages;
1018         }
1019
1020         /**
1021          * Set the revision corresponding to this LinksUpdate
1022          *
1023          * @since 1.27
1024          *
1025          * @param Revision $revision
1026          */
1027         public function setRevision( Revision $revision ) {
1028                 $this->mRevision = $revision;
1029         }
1030
1031         /**
1032          * @since 1.28
1033          * @return null|Revision
1034          */
1035         public function getRevision() {
1036                 return $this->mRevision;
1037         }
1038
1039         /**
1040          * Set the User who triggered this LinksUpdate
1041          *
1042          * @since 1.27
1043          * @param User $user
1044          */
1045         public function setTriggeringUser( User $user ) {
1046                 $this->user = $user;
1047         }
1048
1049         /**
1050          * @since 1.27
1051          * @return null|User
1052          */
1053         public function getTriggeringUser() {
1054                 return $this->user;
1055         }
1056
1057         /**
1058          * Invalidate any necessary link lists related to page property changes
1059          * @param array $changed
1060          */
1061         private function invalidateProperties( $changed ) {
1062                 global $wgPagePropLinkInvalidations;
1063
1064                 foreach ( $changed as $name => $value ) {
1065                         if ( isset( $wgPagePropLinkInvalidations[$name] ) ) {
1066                                 $inv = $wgPagePropLinkInvalidations[$name];
1067                                 if ( !is_array( $inv ) ) {
1068                                         $inv = [ $inv ];
1069                                 }
1070                                 foreach ( $inv as $table ) {
1071                                         DeferredUpdates::addUpdate(
1072                                                 new HTMLCacheUpdate( $this->mTitle, $table, 'page-props' )
1073                                         );
1074                                 }
1075                         }
1076                 }
1077         }
1078
1079         /**
1080          * Fetch page links added by this LinksUpdate.  Only available after the update is complete.
1081          * @since 1.22
1082          * @return null|array Array of Titles
1083          */
1084         public function getAddedLinks() {
1085                 if ( $this->linkInsertions === null ) {
1086                         return null;
1087                 }
1088                 $result = [];
1089                 foreach ( $this->linkInsertions as $insertion ) {
1090                         $result[] = Title::makeTitle( $insertion['pl_namespace'], $insertion['pl_title'] );
1091                 }
1092
1093                 return $result;
1094         }
1095
1096         /**
1097          * Fetch page links removed by this LinksUpdate.  Only available after the update is complete.
1098          * @since 1.22
1099          * @return null|array Array of Titles
1100          */
1101         public function getRemovedLinks() {
1102                 if ( $this->linkDeletions === null ) {
1103                         return null;
1104                 }
1105                 $result = [];
1106                 foreach ( $this->linkDeletions as $ns => $titles ) {
1107                         foreach ( $titles as $title => $unused ) {
1108                                 $result[] = Title::makeTitle( $ns, $title );
1109                         }
1110                 }
1111
1112                 return $result;
1113         }
1114
1115         /**
1116          * Fetch external links added by this LinksUpdate. Only available after
1117          * the update is complete.
1118          * @since 1.33
1119          * @return null|array Array of Strings
1120          */
1121         public function getAddedExternalLinks() {
1122                 if ( $this->externalLinkInsertions === null ) {
1123                         return null;
1124                 }
1125                 $result = [];
1126                 foreach ( $this->externalLinkInsertions as $key => $value ) {
1127                         $result[] = $value['el_to'];
1128                 }
1129                 return $result;
1130         }
1131
1132         /**
1133          * Fetch external links removed by this LinksUpdate. Only available after
1134          * the update is complete.
1135          * @since 1.33
1136          * @return null|array Array of Strings
1137          */
1138         public function getRemovedExternalLinks() {
1139                 if ( $this->externalLinkDeletions === null ) {
1140                         return null;
1141                 }
1142                 return array_keys( $this->externalLinkDeletions );
1143         }
1144
1145         /**
1146          * Fetch page properties added by this LinksUpdate.
1147          * Only available after the update is complete.
1148          * @since 1.28
1149          * @return null|array
1150          */
1151         public function getAddedProperties() {
1152                 return $this->propertyInsertions;
1153         }
1154
1155         /**
1156          * Fetch page properties removed by this LinksUpdate.
1157          * Only available after the update is complete.
1158          * @since 1.28
1159          * @return null|array
1160          */
1161         public function getRemovedProperties() {
1162                 return $this->propertyDeletions;
1163         }
1164
1165         /**
1166          * Update links table freshness
1167          */
1168         private function updateLinksTimestamp() {
1169                 if ( $this->mId ) {
1170                         // The link updates made here only reflect the freshness of the parser output
1171                         $timestamp = $this->mParserOutput->getCacheTime();
1172                         $this->getDB()->update( 'page',
1173                                 [ 'page_links_updated' => $this->getDB()->timestamp( $timestamp ) ],
1174                                 [ 'page_id' => $this->mId ],
1175                                 __METHOD__
1176                         );
1177                 }
1178         }
1179
1180         /**
1181          * @return IDatabase
1182          */
1183         private function getDB() {
1184                 if ( !$this->db ) {
1185                         $this->db = wfGetDB( DB_MASTER );
1186                 }
1187
1188                 return $this->db;
1189         }
1190
1191         public function getAsJobSpecification() {
1192                 if ( $this->user ) {
1193                         $userInfo = [
1194                                 'userId' => $this->user->getId(),
1195                                 'userName' => $this->user->getName(),
1196                         ];
1197                 } else {
1198                         $userInfo = false;
1199                 }
1200
1201                 if ( $this->mRevision ) {
1202                         $triggeringRevisionId = $this->mRevision->getId();
1203                 } else {
1204                         $triggeringRevisionId = false;
1205                 }
1206
1207                 return [
1208                         'wiki' => WikiMap::getWikiIdFromDbDomain( $this->getDB()->getDomainID() ),
1209                         'job'  => new JobSpecification(
1210                                 'refreshLinksPrioritized',
1211                                 [
1212                                         // Reuse the parser cache if it was saved
1213                                         'rootJobTimestamp' => $this->mParserOutput->getCacheTime(),
1214                                         'useRecursiveLinksUpdate' => $this->mRecursive,
1215                                         'triggeringUser' => $userInfo,
1216                                         'triggeringRevisionId' => $triggeringRevisionId,
1217                                         'causeAction' => $this->getCauseAction(),
1218                                         'causeAgent' => $this->getCauseAgent()
1219                                 ],
1220                                 [ 'removeDuplicates' => true ],
1221                                 $this->getTitle()
1222                         )
1223                 ];
1224         }
1225 }