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