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