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