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 /**
  24  * Class the manages updates of *_link tables as well as similar extension-managed tables
  25  *
  26  * @note: LinksUpdate is managed by DeferredUpdates::execute(). Do not run this in a transaction.
  27  *
  28  * See docs/deferred.txt
  29  */
  30 class LinksUpdate extends SqlDataUpdate implements EnqueueableDataUpdate {
  31         // @todo make members protected, but make sure extensions don't break
  32
  33         /** @var int Page ID of the article linked from */
  34         public $mId;
  35
  36         /** @var Title Title object of the article linked from */
  37         public $mTitle;
  38
  39         /** @var ParserOutput */
  40         public $mParserOutput;
  41
  42         /** @var array Map of title strings to IDs for the links in the document */
  43         public $mLinks;
  44
  45         /** @var array DB keys of the images used, in the array key only */
  46         public $mImages;
  47
  48         /** @var array Map of title strings to IDs for the template references, including broken ones */
  49         public $mTemplates;
  50
  51         /** @var array URLs of external links, array key only */
  52         public $mExternals;
  53
  54         /** @var array Map of category names to sort keys */
  55         public $mCategories;
  56
  57         /** @var array Map of language codes to titles */
  58         public $mInterlangs;
  59
  60         /** @var array 2-D map of (prefix => DBK => 1) */
  61         public $mInterwikis;
  62
  63         /** @var array Map of arbitrary name to value */
  64         public $mProperties;
  65
  66         /** @var bool Whether to queue jobs for recursive updates */
  67         public $mRecursive;
  68
  69         /** @var Revision Revision for which this update has been triggered */
  70         private $mRevision;
  71
  72         /**
  73          * @var null|array Added links if calculated.
  74          */
  75         private $linkInsertions = null;
  76
  77         /**
  78          * @var null|array Deleted links if calculated.
  79          */
  80         private $linkDeletions = null;
  81
  82         /**
  83          * @var User|null
  84          */
  85         private $user;
  86
  87         const BATCH_SIZE = 500; // try to keep typical updates in a single transaction
  88
  89         /**
  90          * Constructor
  91          *
  92          * @param Title $title Title of the page we're updating
  93          * @param ParserOutput $parserOutput Output from a full parse of this page
  94          * @param bool $recursive Queue jobs for recursive updates?
  95          * @throws MWException
  96          */
  97         function __construct( Title $title, ParserOutput $parserOutput, $recursive = true ) {
  98                 // Implicit transactions are disabled as they interfere with batching
  99                 parent::__construct( false );
 100
 101                 $this->mTitle = $title;
 102                 $this->mId = $title->getArticleID( Title::GAID_FOR_UPDATE );
 103
 104                 if ( !$this->mId ) {
 105                         throw new InvalidArgumentException(
 106                                 "The Title object yields no ID. Perhaps the page doesn't exist?"
 107                         );
 108                 }
 109
 110                 $this->mParserOutput = $parserOutput;
 111
 112                 $this->mLinks = $parserOutput->getLinks();
 113                 $this->mImages = $parserOutput->getImages();
 114                 $this->mTemplates = $parserOutput->getTemplates();
 115                 $this->mExternals = $parserOutput->getExternalLinks();
 116                 $this->mCategories = $parserOutput->getCategories();
 117                 $this->mProperties = $parserOutput->getProperties();
 118                 $this->mInterwikis = $parserOutput->getInterwikiLinks();
 119
 120                 # Convert the format of the interlanguage links
 121                 # I didn't want to change it in the ParserOutput, because that array is passed all
 122                 # the way back to the skin, so either a skin API break would be required, or an
 123                 # inefficient back-conversion.
 124                 $ill = $parserOutput->getLanguageLinks();
 125                 $this->mInterlangs = [];
 126                 foreach ( $ill as $link ) {
 127                         list( $key, $title ) = explode( ':', $link, 2 );
 128                         $this->mInterlangs[$key] = $title;
 129                 }
 130
 131                 foreach ( $this->mCategories as &$sortkey ) {
 132                         # If the sortkey is longer then 255 bytes,
 133                         # it truncated by DB, and then doesn't get
 134                         # matched when comparing existing vs current
 135                         # categories, causing bug 25254.
 136                         # Also. substr behaves weird when given "".
 137                         if ( $sortkey !== '' ) {
 138                                 $sortkey = substr( $sortkey, 0, 255 );
 139                         }
 140                 }
 141
 142                 $this->mRecursive = $recursive;
 143
 144                 Hooks::run( 'LinksUpdateConstructed', [ &$this ] );
 145         }
 146
 147         /**
 148          * Update link tables with outgoing links from an updated article
 149          *
 150          * @note: this is managed by DeferredUpdates::execute(). Do not run this in a transaction.
 151          */
 152         public function doUpdate() {
 153                 // Make sure all links update threads see the changes of each other.
 154                 // This handles the case when updates have to batched into several COMMITs.
 155                 $scopedLock = self::acquirePageLock( $this->mDb, $this->mId );
 156
 157                 Hooks::run( 'LinksUpdate', [ &$this ] );
 158                 $this->doIncrementalUpdate();
 159
 160                 $this->mDb->onTransactionIdle( function() use ( &$scopedLock ) {
 161                         Hooks::run( 'LinksUpdateComplete', [ &$this ] );
 162                         // Release the lock *after* the final COMMIT for correctness
 163                         ScopedCallback::consume( $scopedLock );
 164                 } );
 165         }
 166
 167         /**
 168          * Acquire a lock for performing link table updates for a page on a DB
 169          *
 170          * @param IDatabase $dbw
 171          * @param integer $pageId
 172          * @return ScopedCallback|null Returns null on failure
 173          * @throws RuntimeException
 174          * @since 1.27
 175          */
 176         public static function acquirePageLock( IDatabase $dbw, $pageId ) {
 177                 $scopedLock = $dbw->getScopedLockAndFlush(
 178                         "LinksUpdate:pageid:$pageId",
 179                         __METHOD__,
 180                         15
 181                 );
 182                 if ( !$scopedLock ) {
 183                         throw new RuntimeException( "Could not acquire lock on page #$pageId." );
 184                 }
 185
 186                 return $scopedLock;
 187         }
 188
 189         protected function doIncrementalUpdate() {
 190                 # Page links
 191                 $existing = $this->getExistingLinks();
 192                 $this->linkDeletions = $this->getLinkDeletions( $existing );
 193                 $this->linkInsertions = $this->getLinkInsertions( $existing );
 194                 $this->incrTableUpdate( 'pagelinks', 'pl', $this->linkDeletions, $this->linkInsertions );
 195
 196                 # Image links
 197                 $existing = $this->getExistingImages();
 198                 $imageDeletes = $this->getImageDeletions( $existing );
 199                 $this->incrTableUpdate( 'imagelinks', 'il', $imageDeletes,
 200                         $this->getImageInsertions( $existing ) );
 201
 202                 # Invalidate all image description pages which had links added or removed
 203                 $imageUpdates = $imageDeletes + array_diff_key( $this->mImages, $existing );
 204                 $this->invalidateImageDescriptions( $imageUpdates );
 205
 206                 # External links
 207                 $existing = $this->getExistingExternals();
 208                 $this->incrTableUpdate( 'externallinks', 'el', $this->getExternalDeletions( $existing ),
 209                         $this->getExternalInsertions( $existing ) );
 210
 211                 # Language links
 212                 $existing = $this->getExistingInterlangs();
 213                 $this->incrTableUpdate( 'langlinks', 'll', $this->getInterlangDeletions( $existing ),
 214                         $this->getInterlangInsertions( $existing ) );
 215
 216                 # Inline interwiki links
 217                 $existing = $this->getExistingInterwikis();
 218                 $this->incrTableUpdate( 'iwlinks', 'iwl', $this->getInterwikiDeletions( $existing ),
 219                         $this->getInterwikiInsertions( $existing ) );
 220
 221                 # Template links
 222                 $existing = $this->getExistingTemplates();
 223                 $this->incrTableUpdate( 'templatelinks', 'tl', $this->getTemplateDeletions( $existing ),
 224                         $this->getTemplateInsertions( $existing ) );
 225
 226                 # Category links
 227                 $existing = $this->getExistingCategories();
 228                 $categoryDeletes = $this->getCategoryDeletions( $existing );
 229                 $this->incrTableUpdate( 'categorylinks', 'cl', $categoryDeletes,
 230                         $this->getCategoryInsertions( $existing ) );
 231
 232                 # Invalidate all categories which were added, deleted or changed (set symmetric difference)
 233                 $categoryInserts = array_diff_assoc( $this->mCategories, $existing );
 234                 $categoryUpdates = $categoryInserts + $categoryDeletes;
 235                 $this->invalidateCategories( $categoryUpdates );
 236                 $this->updateCategoryCounts( $categoryInserts, $categoryDeletes );
 237
 238                 # Page properties
 239                 $existing = $this->getExistingProperties();
 240                 $propertiesDeletes = $this->getPropertyDeletions( $existing );
 241                 $this->incrTableUpdate( 'page_props', 'pp', $propertiesDeletes,
 242                         $this->getPropertyInsertions( $existing ) );
 243
 244                 # Invalidate the necessary pages
 245                 $changed = $propertiesDeletes + array_diff_assoc( $this->mProperties, $existing );
 246                 $this->invalidateProperties( $changed );
 247
 248                 # Update the links table freshness for this title
 249                 $this->updateLinksTimestamp();
 250
 251                 # Refresh links of all pages including this page
 252                 # This will be in a separate transaction
 253                 if ( $this->mRecursive ) {
 254                         $this->queueRecursiveJobs();
 255                 }
 256
 257         }
 258
 259         /**
 260          * Queue recursive jobs for this page
 261          *
 262          * Which means do LinksUpdate on all pages that include the current page,
 263          * using the job queue.
 264          */
 265         protected function queueRecursiveJobs() {
 266                 self::queueRecursiveJobsForTable( $this->mTitle, 'templatelinks' );
 267                 if ( $this->mTitle->getNamespace() == NS_FILE ) {
 268                         // Process imagelinks in case the title is or was a redirect
 269                         self::queueRecursiveJobsForTable( $this->mTitle, 'imagelinks' );
 270                 }
 271
 272                 $bc = $this->mTitle->getBacklinkCache();
 273                 // Get jobs for cascade-protected backlinks for a high priority queue.
 274                 // If meta-templates change to using a new template, the new template
 275                 // should be implicitly protected as soon as possible, if applicable.
 276                 // These jobs duplicate a subset of the above ones, but can run sooner.
 277                 // Which ever runs first generally no-ops the other one.
 278                 $jobs = [];
 279                 foreach ( $bc->getCascadeProtectedLinks() as $title ) {
 280                         $jobs[] = RefreshLinksJob::newPrioritized( $title, [] );
 281                 }
 282                 JobQueueGroup::singleton()->push( $jobs );
 283         }
 284
 285         /**
 286          * Queue a RefreshLinks job for any table.
 287          *
 288          * @param Title $title Title to do job for
 289          * @param string $table Table to use (e.g. 'templatelinks')
 290          */
 291         public static function queueRecursiveJobsForTable( Title $title, $table ) {
 292                 if ( $title->getBacklinkCache()->hasLinks( $table ) ) {
 293                         $job = new RefreshLinksJob(
 294                                 $title,
 295                                 [
 296                                         'table' => $table,
 297                                         'recursive' => true,
 298                                 ] + Job::newRootJobParams( // "overall" refresh links job info
 299                                         "refreshlinks:{$table}:{$title->getPrefixedText()}"
 300                                 )
 301                         );
 302
 303                         JobQueueGroup::singleton()->push( $job );
 304                 }
 305         }
 306
 307         /**
 308          * @param array $cats
 309          */
 310         function invalidateCategories( $cats ) {
 311                 $this->invalidatePages( NS_CATEGORY, array_keys( $cats ) );
 312         }
 313
 314         /**
 315          * Update all the appropriate counts in the category table.
 316          * @param array $added Associative array of category name => sort key
 317          * @param array $deleted Associative array of category name => sort key
 318          */
 319         function updateCategoryCounts( $added, $deleted ) {
 320                 $a = WikiPage::factory( $this->mTitle );
 321                 $a->updateCategoryCounts(
 322                         array_keys( $added ), array_keys( $deleted )
 323                 );
 324         }
 325
 326         /**
 327          * @param array $images
 328          */
 329         function invalidateImageDescriptions( $images ) {
 330                 $this->invalidatePages( NS_FILE, array_keys( $images ) );
 331         }
 332
 333         /**
 334          * Update a table by doing a delete query then an insert query
 335          * @param string $table Table name
 336          * @param string $prefix Field name prefix
 337          * @param array $deletions
 338          * @param array $insertions Rows to insert
 339          */
 340         private function incrTableUpdate( $table, $prefix, $deletions, $insertions ) {
 341                 if ( $table === 'page_props' ) {
 342                         $fromField = 'pp_page';
 343                 } else {
 344                         $fromField = "{$prefix}_from";
 345                 }
 346
 347                 $deleteWheres = []; // list of WHERE clause arrays for each DB delete() call
 348                 if ( $table === 'pagelinks' || $table === 'templatelinks' || $table === 'iwlinks' ) {
 349                         $baseKey =  ( $table === 'iwlinks' ) ? 'iwl_prefix' : "{$prefix}_namespace";
 350
 351                         $curBatchSize = 0;
 352                         $curDeletionBatch = [];
 353                         $deletionBatches = [];
 354                         foreach ( $deletions as $ns => $dbKeys ) {
 355                                 foreach ( $dbKeys as $dbKey => $unused ) {
 356                                         $curDeletionBatch[$ns][$dbKey] = 1;
 357                                         if ( ++$curBatchSize >= self::BATCH_SIZE ) {
 358                                                 $deletionBatches[] = $curDeletionBatch;
 359                                                 $curDeletionBatch = [];
 360                                                 $curBatchSize = 0;
 361                                         }
 362                                 }
 363                         }
 364                         if ( $curDeletionBatch ) {
 365                                 $deletionBatches[] = $curDeletionBatch;
 366                         }
 367
 368                         foreach ( $deletionBatches as $deletionBatch ) {
 369                                 $deleteWheres[] = [
 370                                         $fromField => $this->mId,
 371                                         $this->mDb->makeWhereFrom2d( $deletionBatch, $baseKey, "{$prefix}_title" )
 372                                 ];
 373                         }
 374                 } else {
 375                         if ( $table === 'langlinks' ) {
 376                                 $toField = 'll_lang';
 377                         } elseif ( $table === 'page_props' ) {
 378                                 $toField = 'pp_propname';
 379                         } else {
 380                                 $toField = $prefix . '_to';
 381                         }
 382
 383                         $deletionBatches = array_chunk( array_keys( $deletions ), self::BATCH_SIZE );
 384                         foreach ( $deletionBatches as $deletionBatch ) {
 385                                 $deleteWheres[] = [ $fromField => $this->mId, $toField => $deletionBatch ];
 386                         }
 387                 }
 388
 389                 foreach ( $deleteWheres as $deleteWhere ) {
 390                         $this->mDb->delete( $table, $deleteWhere, __METHOD__ );
 391                         $this->mDb->commit( __METHOD__, 'flush' );
 392                         wfGetLBFactory()->waitForReplication( [ 'wiki' => $this->mDb->getWikiID() ] );
 393                 }
 394
 395                 $insertBatches = array_chunk( $insertions, self::BATCH_SIZE );
 396                 foreach ( $insertBatches as $insertBatch ) {
 397                         $this->mDb->insert( $table, $insertBatch, __METHOD__, 'IGNORE' );
 398                         $this->mDb->commit( __METHOD__, 'flush' );
 399                         wfGetLBFactory()->waitForReplication( [ 'wiki' => $this->mDb->getWikiID() ] );
 400                 }
 401
 402                 if ( count( $insertions ) ) {
 403                         Hooks::run( 'LinksUpdateAfterInsert', [ $this, $table, $insertions ] );
 404                 }
 405         }
 406
 407         /**
 408          * Get an array of pagelinks insertions for passing to the DB
 409          * Skips the titles specified by the 2-D array $existing
 410          * @param array $existing
 411          * @return array
 412          */
 413         private function getLinkInsertions( $existing = [] ) {
 414                 $arr = [];
 415                 foreach ( $this->mLinks as $ns => $dbkeys ) {
 416                         $diffs = isset( $existing[$ns] )
 417                                 ? array_diff_key( $dbkeys, $existing[$ns] )
 418                                 : $dbkeys;
 419                         foreach ( $diffs as $dbk => $id ) {
 420                                 $arr[] = [
 421                                         'pl_from' => $this->mId,
 422                                         'pl_from_namespace' => $this->mTitle->getNamespace(),
 423                                         'pl_namespace' => $ns,
 424                                         'pl_title' => $dbk
 425                                 ];
 426                         }
 427                 }
 428
 429                 return $arr;
 430         }
 431
 432         /**
 433          * Get an array of template insertions. Like getLinkInsertions()
 434          * @param array $existing
 435          * @return array
 436          */
 437         private function getTemplateInsertions( $existing = [] ) {
 438                 $arr = [];
 439                 foreach ( $this->mTemplates as $ns => $dbkeys ) {
 440                         $diffs = isset( $existing[$ns] ) ? array_diff_key( $dbkeys, $existing[$ns] ) : $dbkeys;
 441                         foreach ( $diffs as $dbk => $id ) {
 442                                 $arr[] = [
 443                                         'tl_from' => $this->mId,
 444                                         'tl_from_namespace' => $this->mTitle->getNamespace(),
 445                                         'tl_namespace' => $ns,
 446                                         'tl_title' => $dbk
 447                                 ];
 448                         }
 449                 }
 450
 451                 return $arr;
 452         }
 453
 454         /**
 455          * Get an array of image insertions
 456          * Skips the names specified in $existing
 457          * @param array $existing
 458          * @return array
 459          */
 460         private function getImageInsertions( $existing = [] ) {
 461                 $arr = [];
 462                 $diffs = array_diff_key( $this->mImages, $existing );
 463                 foreach ( $diffs as $iname => $dummy ) {
 464                         $arr[] = [
 465                                 'il_from' => $this->mId,
 466                                 'il_from_namespace' => $this->mTitle->getNamespace(),
 467                                 'il_to' => $iname
 468                         ];
 469                 }
 470
 471                 return $arr;
 472         }
 473
 474         /**
 475          * Get an array of externallinks insertions. Skips the names specified in $existing
 476          * @param array $existing
 477          * @return array
 478          */
 479         private function getExternalInsertions( $existing = [] ) {
 480                 $arr = [];
 481                 $diffs = array_diff_key( $this->mExternals, $existing );
 482                 foreach ( $diffs as $url => $dummy ) {
 483                         foreach ( wfMakeUrlIndexes( $url ) as $index ) {
 484                                 $arr[] = [
 485                                         'el_id' => $this->mDb->nextSequenceValue( 'externallinks_el_id_seq' ),
 486                                         'el_from' => $this->mId,
 487                                         'el_to' => $url,
 488                                         'el_index' => $index,
 489                                 ];
 490                         }
 491                 }
 492
 493                 return $arr;
 494         }
 495
 496         /**
 497          * Get an array of category insertions
 498          *
 499          * @param array $existing Mapping existing category names to sort keys. If both
 500          * match a link in $this, the link will be omitted from the output
 501          *
 502          * @return array
 503          */
 504         private function getCategoryInsertions( $existing = [] ) {
 505                 global $wgContLang, $wgCategoryCollation;
 506                 $diffs = array_diff_assoc( $this->mCategories, $existing );
 507                 $arr = [];
 508                 foreach ( $diffs as $name => $prefix ) {
 509                         $nt = Title::makeTitleSafe( NS_CATEGORY, $name );
 510                         $wgContLang->findVariantLink( $name, $nt, true );
 511
 512                         if ( $this->mTitle->getNamespace() == NS_CATEGORY ) {
 513                                 $type = 'subcat';
 514                         } elseif ( $this->mTitle->getNamespace() == NS_FILE ) {
 515                                 $type = 'file';
 516                         } else {
 517                                 $type = 'page';
 518                         }
 519
 520                         # Treat custom sortkeys as a prefix, so that if multiple
 521                         # things are forced to sort as '*' or something, they'll
 522                         # sort properly in the category rather than in page_id
 523                         # order or such.
 524                         $sortkey = Collation::singleton()->getSortKey(
 525                                 $this->mTitle->getCategorySortkey( $prefix ) );
 526
 527                         $arr[] = [
 528                                 'cl_from' => $this->mId,
 529                                 'cl_to' => $name,
 530                                 'cl_sortkey' => $sortkey,
 531                                 'cl_timestamp' => $this->mDb->timestamp(),
 532                                 'cl_sortkey_prefix' => $prefix,
 533                                 'cl_collation' => $wgCategoryCollation,
 534                                 'cl_type' => $type,
 535                         ];
 536                 }
 537
 538                 return $arr;
 539         }
 540
 541         /**
 542          * Get an array of interlanguage link insertions
 543          *
 544          * @param array $existing Mapping existing language codes to titles
 545          *
 546          * @return array
 547          */
 548         private function getInterlangInsertions( $existing = [] ) {
 549                 $diffs = array_diff_assoc( $this->mInterlangs, $existing );
 550                 $arr = [];
 551                 foreach ( $diffs as $lang => $title ) {
 552                         $arr[] = [
 553                                 'll_from' => $this->mId,
 554                                 'll_lang' => $lang,
 555                                 'll_title' => $title
 556                         ];
 557                 }
 558
 559                 return $arr;
 560         }
 561
 562         /**
 563          * Get an array of page property insertions
 564          * @param array $existing
 565          * @return array
 566          */
 567         function getPropertyInsertions( $existing = [] ) {
 568                 $diffs = array_diff_assoc( $this->mProperties, $existing );
 569
 570                 $arr = [];
 571                 foreach ( array_keys( $diffs ) as $name ) {
 572                         $arr[] = $this->getPagePropRowData( $name );
 573                 }
 574
 575                 return $arr;
 576         }
 577
 578         /**
 579          * Returns an associative array to be used for inserting a row into
 580          * the page_props table. Besides the given property name, this will
 581          * include the page id from $this->mId and any property value from
 582          * $this->mProperties.
 583          *
 584          * The array returned will include the pp_sortkey field if this
 585          * is present in the database (as indicated by $wgPagePropsHaveSortkey).
 586          * The sortkey value is currently determined by getPropertySortKeyValue().
 587          *
 588          * @note this assumes that $this->mProperties[$prop] is defined.
 589          *
 590          * @param string $prop The name of the property.
 591          *
 592          * @return array
 593          */
 594         private function getPagePropRowData( $prop ) {
 595                 global $wgPagePropsHaveSortkey;
 596
 597                 $value = $this->mProperties[$prop];
 598
 599                 $row = [
 600                         'pp_page' => $this->mId,
 601                         'pp_propname' => $prop,
 602                         'pp_value' => $value,
 603                 ];
 604
 605                 if ( $wgPagePropsHaveSortkey ) {
 606                         $row['pp_sortkey'] = $this->getPropertySortKeyValue( $value );
 607                 }
 608
 609                 return $row;
 610         }
 611
 612         /**
 613          * Determines the sort key for the given property value.
 614          * This will return $value if it is a float or int,
 615          * 1 or resp. 0 if it is a bool, and null otherwise.
 616          *
 617          * @note In the future, we may allow the sortkey to be specified explicitly
 618          *       in ParserOutput::setProperty.
 619          *
 620          * @param mixed $value
 621          *
 622          * @return float|null
 623          */
 624         private function getPropertySortKeyValue( $value ) {
 625                 if ( is_int( $value ) || is_float( $value ) || is_bool( $value ) ) {
 626                         return floatval( $value );
 627                 }
 628
 629                 return null;
 630         }
 631
 632         /**
 633          * Get an array of interwiki insertions for passing to the DB
 634          * Skips the titles specified by the 2-D array $existing
 635          * @param array $existing
 636          * @return array
 637          */
 638         private function getInterwikiInsertions( $existing = [] ) {
 639                 $arr = [];
 640                 foreach ( $this->mInterwikis as $prefix => $dbkeys ) {
 641                         $diffs = isset( $existing[$prefix] )
 642                                 ? array_diff_key( $dbkeys, $existing[$prefix] )
 643                                 : $dbkeys;
 644
 645                         foreach ( $diffs as $dbk => $id ) {
 646                                 $arr[] = [
 647                                         'iwl_from' => $this->mId,
 648                                         'iwl_prefix' => $prefix,
 649                                         'iwl_title' => $dbk
 650                                 ];
 651                         }
 652                 }
 653
 654                 return $arr;
 655         }
 656
 657         /**
 658          * Given an array of existing links, returns those links which are not in $this
 659          * and thus should be deleted.
 660          * @param array $existing
 661          * @return array
 662          */
 663         private function getLinkDeletions( $existing ) {
 664                 $del = [];
 665                 foreach ( $existing as $ns => $dbkeys ) {
 666                         if ( isset( $this->mLinks[$ns] ) ) {
 667                                 $del[$ns] = array_diff_key( $existing[$ns], $this->mLinks[$ns] );
 668                         } else {
 669                                 $del[$ns] = $existing[$ns];
 670                         }
 671                 }
 672
 673                 return $del;
 674         }
 675
 676         /**
 677          * Given an array of existing templates, returns those templates which are not in $this
 678          * and thus should be deleted.
 679          * @param array $existing
 680          * @return array
 681          */
 682         private function getTemplateDeletions( $existing ) {
 683                 $del = [];
 684                 foreach ( $existing as $ns => $dbkeys ) {
 685                         if ( isset( $this->mTemplates[$ns] ) ) {
 686                                 $del[$ns] = array_diff_key( $existing[$ns], $this->mTemplates[$ns] );
 687                         } else {
 688                                 $del[$ns] = $existing[$ns];
 689                         }
 690                 }
 691
 692                 return $del;
 693         }
 694
 695         /**
 696          * Given an array of existing images, returns those images which are not in $this
 697          * and thus should be deleted.
 698          * @param array $existing
 699          * @return array
 700          */
 701         private function getImageDeletions( $existing ) {
 702                 return array_diff_key( $existing, $this->mImages );
 703         }
 704
 705         /**
 706          * Given an array of existing external links, returns those links which are not
 707          * in $this and thus should be deleted.
 708          * @param array $existing
 709          * @return array
 710          */
 711         private function getExternalDeletions( $existing ) {
 712                 return array_diff_key( $existing, $this->mExternals );
 713         }
 714
 715         /**
 716          * Given an array of existing categories, returns those categories which are not in $this
 717          * and thus should be deleted.
 718          * @param array $existing
 719          * @return array
 720          */
 721         private function getCategoryDeletions( $existing ) {
 722                 return array_diff_assoc( $existing, $this->mCategories );
 723         }
 724
 725         /**
 726          * Given an array of existing interlanguage links, returns those links which are not
 727          * in $this and thus should be deleted.
 728          * @param array $existing
 729          * @return array
 730          */
 731         private function getInterlangDeletions( $existing ) {
 732                 return array_diff_assoc( $existing, $this->mInterlangs );
 733         }
 734
 735         /**
 736          * Get array of properties which should be deleted.
 737          * @param array $existing
 738          * @return array
 739          */
 740         function getPropertyDeletions( $existing ) {
 741                 return array_diff_assoc( $existing, $this->mProperties );
 742         }
 743
 744         /**
 745          * Given an array of existing interwiki links, returns those links which are not in $this
 746          * and thus should be deleted.
 747          * @param array $existing
 748          * @return array
 749          */
 750         private function getInterwikiDeletions( $existing ) {
 751                 $del = [];
 752                 foreach ( $existing as $prefix => $dbkeys ) {
 753                         if ( isset( $this->mInterwikis[$prefix] ) ) {
 754                                 $del[$prefix] = array_diff_key( $existing[$prefix], $this->mInterwikis[$prefix] );
 755                         } else {
 756                                 $del[$prefix] = $existing[$prefix];
 757                         }
 758                 }
 759
 760                 return $del;
 761         }
 762
 763         /**
 764          * Get an array of existing links, as a 2-D array
 765          *
 766          * @return array
 767          */
 768         private function getExistingLinks() {
 769                 $res = $this->mDb->select( 'pagelinks', [ 'pl_namespace', 'pl_title' ],
 770                         [ 'pl_from' => $this->mId ], __METHOD__, $this->mOptions );
 771                 $arr = [];
 772                 foreach ( $res as $row ) {
 773                         if ( !isset( $arr[$row->pl_namespace] ) ) {
 774                                 $arr[$row->pl_namespace] = [];
 775                         }
 776                         $arr[$row->pl_namespace][$row->pl_title] = 1;
 777                 }
 778
 779                 return $arr;
 780         }
 781
 782         /**
 783          * Get an array of existing templates, as a 2-D array
 784          *
 785          * @return array
 786          */
 787         private function getExistingTemplates() {
 788                 $res = $this->mDb->select( 'templatelinks', [ 'tl_namespace', 'tl_title' ],
 789                         [ 'tl_from' => $this->mId ], __METHOD__, $this->mOptions );
 790                 $arr = [];
 791                 foreach ( $res as $row ) {
 792                         if ( !isset( $arr[$row->tl_namespace] ) ) {
 793                                 $arr[$row->tl_namespace] = [];
 794                         }
 795                         $arr[$row->tl_namespace][$row->tl_title] = 1;
 796                 }
 797
 798                 return $arr;
 799         }
 800
 801         /**
 802          * Get an array of existing images, image names in the keys
 803          *
 804          * @return array
 805          */
 806         private function getExistingImages() {
 807                 $res = $this->mDb->select( 'imagelinks', [ 'il_to' ],
 808                         [ 'il_from' => $this->mId ], __METHOD__, $this->mOptions );
 809                 $arr = [];
 810                 foreach ( $res as $row ) {
 811                         $arr[$row->il_to] = 1;
 812                 }
 813
 814                 return $arr;
 815         }
 816
 817         /**
 818          * Get an array of existing external links, URLs in the keys
 819          *
 820          * @return array
 821          */
 822         private function getExistingExternals() {
 823                 $res = $this->mDb->select( 'externallinks', [ 'el_to' ],
 824                         [ 'el_from' => $this->mId ], __METHOD__, $this->mOptions );
 825                 $arr = [];
 826                 foreach ( $res as $row ) {
 827                         $arr[$row->el_to] = 1;
 828                 }
 829
 830                 return $arr;
 831         }
 832
 833         /**
 834          * Get an array of existing categories, with the name in the key and sort key in the value.
 835          *
 836          * @return array
 837          */
 838         private function getExistingCategories() {
 839                 $res = $this->mDb->select( 'categorylinks', [ 'cl_to', 'cl_sortkey_prefix' ],
 840                         [ 'cl_from' => $this->mId ], __METHOD__, $this->mOptions );
 841                 $arr = [];
 842                 foreach ( $res as $row ) {
 843                         $arr[$row->cl_to] = $row->cl_sortkey_prefix;
 844                 }
 845
 846                 return $arr;
 847         }
 848
 849         /**
 850          * Get an array of existing interlanguage links, with the language code in the key and the
 851          * title in the value.
 852          *
 853          * @return array
 854          */
 855         private function getExistingInterlangs() {
 856                 $res = $this->mDb->select( 'langlinks', [ 'll_lang', 'll_title' ],
 857                         [ 'll_from' => $this->mId ], __METHOD__, $this->mOptions );
 858                 $arr = [];
 859                 foreach ( $res as $row ) {
 860                         $arr[$row->ll_lang] = $row->ll_title;
 861                 }
 862
 863                 return $arr;
 864         }
 865
 866         /**
 867          * Get an array of existing inline interwiki links, as a 2-D array
 868          * @return array (prefix => array(dbkey => 1))
 869          */
 870         protected function getExistingInterwikis() {
 871                 $res = $this->mDb->select( 'iwlinks', [ 'iwl_prefix', 'iwl_title' ],
 872                         [ 'iwl_from' => $this->mId ], __METHOD__, $this->mOptions );
 873                 $arr = [];
 874                 foreach ( $res as $row ) {
 875                         if ( !isset( $arr[$row->iwl_prefix] ) ) {
 876                                 $arr[$row->iwl_prefix] = [];
 877                         }
 878                         $arr[$row->iwl_prefix][$row->iwl_title] = 1;
 879                 }
 880
 881                 return $arr;
 882         }
 883
 884         /**
 885          * Get an array of existing categories, with the name in the key and sort key in the value.
 886          *
 887          * @return array Array of property names and values
 888          */
 889         private function getExistingProperties() {
 890                 $res = $this->mDb->select( 'page_props', [ 'pp_propname', 'pp_value' ],
 891                         [ 'pp_page' => $this->mId ], __METHOD__, $this->mOptions );
 892                 $arr = [];
 893                 foreach ( $res as $row ) {
 894                         $arr[$row->pp_propname] = $row->pp_value;
 895                 }
 896
 897                 return $arr;
 898         }
 899
 900         /**
 901          * Return the title object of the page being updated
 902          * @return Title
 903          */
 904         public function getTitle() {
 905                 return $this->mTitle;
 906         }
 907
 908         /**
 909          * Returns parser output
 910          * @since 1.19
 911          * @return ParserOutput
 912          */
 913         public function getParserOutput() {
 914                 return $this->mParserOutput;
 915         }
 916
 917         /**
 918          * Return the list of images used as generated by the parser
 919          * @return array
 920          */
 921         public function getImages() {
 922                 return $this->mImages;
 923         }
 924
 925         /**
 926          * Set the revision corresponding to this LinksUpdate
 927          *
 928          * @since 1.27
 929          *
 930          * @param Revision $revision
 931          */
 932         public function setRevision( Revision $revision ) {
 933                 $this->mRevision = $revision;
 934         }
 935
 936         /**
 937          * Set the User who triggered this LinksUpdate
 938          *
 939          * @since 1.27
 940          * @param User $user
 941          */
 942         public function setTriggeringUser( User $user ) {
 943                 $this->user = $user;
 944         }
 945
 946         /**
 947          * @since 1.27
 948          * @return null|User
 949          */
 950         public function getTriggeringUser() {
 951                 return $this->user;
 952         }
 953
 954         /**
 955          * Invalidate any necessary link lists related to page property changes
 956          * @param array $changed
 957          */
 958         private function invalidateProperties( $changed ) {
 959                 global $wgPagePropLinkInvalidations;
 960
 961                 foreach ( $changed as $name => $value ) {
 962                         if ( isset( $wgPagePropLinkInvalidations[$name] ) ) {
 963                                 $inv = $wgPagePropLinkInvalidations[$name];
 964                                 if ( !is_array( $inv ) ) {
 965                                         $inv = [ $inv ];
 966                                 }
 967                                 foreach ( $inv as $table ) {
 968                                         DeferredUpdates::addUpdate( new HTMLCacheUpdate( $this->mTitle, $table ) );
 969                                 }
 970                         }
 971                 }
 972         }
 973
 974         /**
 975          * Fetch page links added by this LinksUpdate.  Only available after the update is complete.
 976          * @since 1.22
 977          * @return null|array Array of Titles
 978          */
 979         public function getAddedLinks() {
 980                 if ( $this->linkInsertions === null ) {
 981                         return null;
 982                 }
 983                 $result = [];
 984                 foreach ( $this->linkInsertions as $insertion ) {
 985                         $result[] = Title::makeTitle( $insertion['pl_namespace'], $insertion['pl_title'] );
 986                 }
 987
 988                 return $result;
 989         }
 990
 991         /**
 992          * Fetch page links removed by this LinksUpdate.  Only available after the update is complete.
 993          * @since 1.22
 994          * @return null|array Array of Titles
 995          */
 996         public function getRemovedLinks() {
 997                 if ( $this->linkDeletions === null ) {
 998                         return null;
 999                 }
1000                 $result = [];
1001                 foreach ( $this->linkDeletions as $ns => $titles ) {
1002                         foreach ( $titles as $title => $unused ) {
1003                                 $result[] = Title::makeTitle( $ns, $title );
1004                         }
1005                 }
1006
1007                 return $result;
1008         }
1009
1010         /**
1011          * Update links table freshness
1012          */
1013         protected function updateLinksTimestamp() {
1014                 if ( $this->mId ) {
1015                         // The link updates made here only reflect the freshness of the parser output
1016                         $timestamp = $this->mParserOutput->getCacheTime();
1017                         $this->mDb->update( 'page',
1018                                 [ 'page_links_updated' => $this->mDb->timestamp( $timestamp ) ],
1019                                 [ 'page_id' => $this->mId ],
1020                                 __METHOD__
1021                         );
1022                 }
1023         }
1024
1025         public function getAsJobSpecification() {
1026                 if ( $this->user ) {
1027                         $userInfo = [
1028                                 'userId' => $this->user->getId(),
1029                                 'userName' => $this->user->getName(),
1030                         ];
1031                 } else {
1032                         $userInfo = false;
1033                 }
1034
1035                 if ( $this->mRevision ) {
1036                         $triggeringRevisionId = $this->mRevision->getId();
1037                 } else {
1038                         $triggeringRevisionId = false;
1039                 }
1040
1041                 return [
1042                         'wiki' => $this->mDb->getWikiID(),
1043                         'job'  => new JobSpecification(
1044                                 'refreshLinksPrioritized',
1045                                 [
1046                                         // Reuse the parser cache if it was saved
1047                                         'rootJobTimestamp' => $this->mParserOutput->getCacheTime(),
1048                                         'useRecursiveLinksUpdate' => $this->mRecursive,
1049                                         'triggeringUser' => $userInfo,
1050                                         'triggeringRevisionId' => $triggeringRevisionId,
1051                                 ],
1052                                 [ 'removeDuplicates' => true ],
1053                                 $this->getTitle()
1054                         )
1055                 ];
1056         }
1057 }