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