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\Logger\LoggerFactory;
  25 use MediaWiki\MediaWikiServices;
  26 use Wikimedia\ScopedCallback;
  27
  28 /**
  29  * Class the manages updates of *_link tables as well as similar extension-managed tables
  30  *
  31  * @note: LinksUpdate is managed by DeferredUpdates::execute(). Do not run this in a transaction.
  32  *
  33  * See docs/deferred.txt
  34  */
  35 class LinksUpdate extends DataUpdate implements EnqueueableDataUpdate {
  36         // @todo make members protected, but make sure extensions don't break
  37
  38         /** @var int Page ID of the article linked from */
  39         public $mId;
  40
  41         /** @var Title Title object of the article linked from */
  42         public $mTitle;
  43
  44         /** @var ParserOutput */
  45         public $mParserOutput;
  46
  47         /** @var array Map of title strings to IDs for the links in the document */
  48         public $mLinks;
  49
  50         /** @var array DB keys of the images used, in the array key only */
  51         public $mImages;
  52
  53         /** @var array Map of title strings to IDs for the template references, including broken ones */
  54         public $mTemplates;
  55
  56         /** @var array URLs of external links, array key only */
  57         public $mExternals;
  58
  59         /** @var array Map of category names to sort keys */
  60         public $mCategories;
  61
  62         /** @var array Map of language codes to titles */
  63         public $mInterlangs;
  64
  65         /** @var array 2-D map of (prefix => DBK => 1) */
  66         public $mInterwikis;
  67
  68         /** @var array Map of arbitrary name to value */
  69         public $mProperties;
  70
  71         /** @var bool Whether to queue jobs for recursive updates */
  72         public $mRecursive;
  73
  74         /** @var Revision Revision for which this update has been triggered */
  75         private $mRevision;
  76
  77         /**
  78          * @var null|array Added links if calculated.
  79          */
  80         private $linkInsertions = null;
  81
  82         /**
  83          * @var null|array Deleted links if calculated.
  84          */
  85         private $linkDeletions = null;
  86
  87         /**
  88          * @var null|array Added properties if calculated.
  89          */
  90         private $propertyInsertions = null;
  91
  92         /**
  93          * @var null|array Deleted properties if calculated.
  94          */
  95         private $propertyDeletions = null;
  96
  97         /**
  98          * @var User|null
  99          */
 100         private $user;
 101
 102         /** @var IDatabase */
 103         private $db;
 104
 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, it is truncated by DB, and then doesn't match
 146                         # when comparing existing vs current categories, causing T27254.
 147                         $sortkey = mb_strcut( $sortkey, 0, 255 );
 148                 }
 149
 150                 $this->mRecursive = $recursive;
 151
 152                 // Avoid PHP 7.1 warning from passing $this by reference
 153                 $linksUpdate = $this;
 154                 Hooks::run( 'LinksUpdateConstructed', [ &$linksUpdate ] );
 155         }
 156
 157         /**
 158          * Update link tables with outgoing links from an updated article
 159          *
 160          * @note this is managed by DeferredUpdates::execute(). Do not run this in a transaction.
 161          */
 162         public function doUpdate() {
 163                 if ( $this->ticket ) {
 164                         // Make sure all links update threads see the changes of each other.
 165                         // This handles the case when updates have to batched into several COMMITs.
 166                         $scopedLock = self::acquirePageLock( $this->getDB(), $this->mId );
 167                         if ( !$scopedLock ) {
 168                                 throw new RuntimeException( "Could not acquire lock for page ID '{$this->mId}'." );
 169                         }
 170                 }
 171
 172                 // Avoid PHP 7.1 warning from passing $this by reference
 173                 $linksUpdate = $this;
 174                 Hooks::run( 'LinksUpdate', [ &$linksUpdate ] );
 175                 $this->doIncrementalUpdate();
 176
 177                 // Commit and release the lock (if set)
 178                 ScopedCallback::consume( $scopedLock );
 179                 // Run post-commit hook handlers without DBO_TRX
 180                 DeferredUpdates::addUpdate( new AutoCommitUpdate(
 181                         $this->getDB(),
 182                         __METHOD__,
 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                 ) );
 189         }
 190
 191         /**
 192          * Acquire a lock for performing link table updates for a page on a DB
 193          *
 194          * @param IDatabase $dbw
 195          * @param int $pageId
 196          * @param string $why One of (job, atomicity)
 197          * @return ScopedCallback|null
 198          * @since 1.27
 199          */
 200         public static function acquirePageLock( IDatabase $dbw, $pageId, $why = 'atomicity' ) {
 201                 $key = "LinksUpdate:$why:pageid:$pageId";
 202                 $scopedLock = $dbw->getScopedLockAndFlush( $key, __METHOD__, 15 );
 203                 if ( !$scopedLock ) {
 204                         $logger = LoggerFactory::getInstance( 'SecondaryDataUpdate' );
 205                         $logger->info( "Could not acquire lock '{key}' for page ID '{page_id}'.", [
 206                                 'key' => $key,
 207                                 'page_id' => $pageId,
 208                         ] );
 209                         return null;
 210                 }
 211
 212                 return $scopedLock;
 213         }
 214
 215         protected function doIncrementalUpdate() {
 216                 # Page links
 217                 $existingPL = $this->getExistingLinks();
 218                 $this->linkDeletions = $this->getLinkDeletions( $existingPL );
 219                 $this->linkInsertions = $this->getLinkInsertions( $existingPL );
 220                 $this->incrTableUpdate( 'pagelinks', 'pl', $this->linkDeletions, $this->linkInsertions );
 221
 222                 # Image links
 223                 $existingIL = $this->getExistingImages();
 224                 $imageDeletes = $this->getImageDeletions( $existingIL );
 225                 $this->incrTableUpdate(
 226                         'imagelinks',
 227                         'il',
 228                         $imageDeletes,
 229                         $this->getImageInsertions( $existingIL ) );
 230
 231                 # Invalidate all image description pages which had links added or removed
 232                 $imageUpdates = $imageDeletes + array_diff_key( $this->mImages, $existingIL );
 233                 $this->invalidateImageDescriptions( $imageUpdates );
 234
 235                 # External links
 236                 $existingEL = $this->getExistingExternals();
 237                 $this->incrTableUpdate(
 238                         'externallinks',
 239                         'el',
 240                         $this->getExternalDeletions( $existingEL ),
 241                         $this->getExternalInsertions( $existingEL ) );
 242
 243                 # Language links
 244                 $existingLL = $this->getExistingInterlangs();
 245                 $this->incrTableUpdate(
 246                         'langlinks',
 247                         'll',
 248                         $this->getInterlangDeletions( $existingLL ),
 249                         $this->getInterlangInsertions( $existingLL ) );
 250
 251                 # Inline interwiki links
 252                 $existingIW = $this->getExistingInterwikis();
 253                 $this->incrTableUpdate(
 254                         'iwlinks',
 255                         'iwl',
 256                         $this->getInterwikiDeletions( $existingIW ),
 257                         $this->getInterwikiInsertions( $existingIW ) );
 258
 259                 # Template links
 260                 $existingTL = $this->getExistingTemplates();
 261                 $this->incrTableUpdate(
 262                         'templatelinks',
 263                         'tl',
 264                         $this->getTemplateDeletions( $existingTL ),
 265                         $this->getTemplateInsertions( $existingTL ) );
 266
 267                 # Category links
 268                 $existingCL = $this->getExistingCategories();
 269                 $categoryDeletes = $this->getCategoryDeletions( $existingCL );
 270                 $this->incrTableUpdate(
 271                         'categorylinks',
 272                         'cl',
 273                         $categoryDeletes,
 274                         $this->getCategoryInsertions( $existingCL ) );
 275                 $categoryInserts = array_diff_assoc( $this->mCategories, $existingCL );
 276                 $categoryUpdates = $categoryInserts + $categoryDeletes;
 277
 278                 # Page properties
 279                 $existingPP = $this->getExistingProperties();
 280                 $this->propertyDeletions = $this->getPropertyDeletions( $existingPP );
 281                 $this->incrTableUpdate(
 282                         'page_props',
 283                         'pp',
 284                         $this->propertyDeletions,
 285                         $this->getPropertyInsertions( $existingPP ) );
 286
 287                 # Invalidate the necessary pages
 288                 $this->propertyInsertions = array_diff_assoc( $this->mProperties, $existingPP );
 289                 $changed = $this->propertyDeletions + $this->propertyInsertions;
 290                 $this->invalidateProperties( $changed );
 291
 292                 # Invalidate all categories which were added, deleted or changed (set symmetric difference)
 293                 $this->invalidateCategories( $categoryUpdates );
 294                 $this->updateCategoryCounts( $categoryInserts, $categoryDeletes );
 295
 296                 # Refresh links of all pages including this page
 297                 # This will be in a separate transaction
 298                 if ( $this->mRecursive ) {
 299                         $this->queueRecursiveJobs();
 300                 }
 301
 302                 # Update the links table freshness for this title
 303                 $this->updateLinksTimestamp();
 304         }
 305
 306         /**
 307          * Queue recursive jobs for this page
 308          *
 309          * Which means do LinksUpdate on all pages that include the current page,
 310          * using the job queue.
 311          */
 312         protected function queueRecursiveJobs() {
 313                 $action = $this->getCauseAction();
 314                 $agent = $this->getCauseAgent();
 315
 316                 self::queueRecursiveJobsForTable( $this->mTitle, 'templatelinks', $action, $agent );
 317                 if ( $this->mTitle->getNamespace() == NS_FILE ) {
 318                         // Process imagelinks in case the title is or was a redirect
 319                         self::queueRecursiveJobsForTable( $this->mTitle, 'imagelinks', $action, $agent );
 320                 }
 321
 322                 $bc = $this->mTitle->getBacklinkCache();
 323                 // Get jobs for cascade-protected backlinks for a high priority queue.
 324                 // If meta-templates change to using a new template, the new template
 325                 // should be implicitly protected as soon as possible, if applicable.
 326                 // These jobs duplicate a subset of the above ones, but can run sooner.
 327                 // Which ever runs first generally no-ops the other one.
 328                 $jobs = [];
 329                 foreach ( $bc->getCascadeProtectedLinks() as $title ) {
 330                         $jobs[] = RefreshLinksJob::newPrioritized(
 331                                 $title,
 332                                 [
 333                                         'causeAction' => $action,
 334                                         'causeAgent' => $agent
 335                                 ]
 336                         );
 337                 }
 338                 JobQueueGroup::singleton()->push( $jobs );
 339         }
 340
 341         /**
 342          * Queue a RefreshLinks job for any table.
 343          *
 344          * @param Title $title Title to do job for
 345          * @param string $table Table to use (e.g. 'templatelinks')
 346          * @param string $action Triggering action
 347          * @param string $userName Triggering user name
 348          */
 349         public static function queueRecursiveJobsForTable(
 350                 Title $title, $table, $action = 'unknown', $userName = 'unknown'
 351         ) {
 352                 if ( $title->getBacklinkCache()->hasLinks( $table ) ) {
 353                         $job = new RefreshLinksJob(
 354                                 $title,
 355                                 [
 356                                         'table' => $table,
 357                                         'recursive' => true,
 358                                 ] + Job::newRootJobParams( // "overall" refresh links job info
 359                                         "refreshlinks:{$table}:{$title->getPrefixedText()}"
 360                                 ) + [ 'causeAction' => $action, 'causeAgent' => $userName ]
 361                         );
 362
 363                         JobQueueGroup::singleton()->push( $job );
 364                 }
 365         }
 366
 367         /**
 368          * @param array $cats
 369          */
 370         private function invalidateCategories( $cats ) {
 371                 PurgeJobUtils::invalidatePages( $this->getDB(), NS_CATEGORY, array_keys( $cats ) );
 372         }
 373
 374         /**
 375          * Update all the appropriate counts in the category table.
 376          * @param array $added Associative array of category name => sort key
 377          * @param array $deleted Associative array of category name => sort key
 378          */
 379         private function updateCategoryCounts( array $added, array $deleted ) {
 380                 global $wgUpdateRowsPerQuery;
 381
 382                 if ( !$added && !$deleted ) {
 383                         return;
 384                 }
 385
 386                 $domainId = $this->getDB()->getDomainID();
 387                 $wp = WikiPage::factory( $this->mTitle );
 388                 $lbf = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
 389                 // T163801: try to release any row locks to reduce contention
 390                 $lbf->commitAndWaitForReplication( __METHOD__, $this->ticket, [ 'domain' => $domainId ] );
 391
 392                 foreach ( array_chunk( array_keys( $added ), $wgUpdateRowsPerQuery ) as $addBatch ) {
 393                         $wp->updateCategoryCounts( $addBatch, [], $this->mId );
 394                         $lbf->commitAndWaitForReplication(
 395                                 __METHOD__, $this->ticket, [ 'domain' => $domainId ] );
 396                 }
 397
 398                 foreach ( array_chunk( array_keys( $deleted ), $wgUpdateRowsPerQuery ) as $deleteBatch ) {
 399                         $wp->updateCategoryCounts( [], $deleteBatch, $this->mId );
 400                         $lbf->commitAndWaitForReplication(
 401                                 __METHOD__, $this->ticket, [ 'domain' => $domainId ] );
 402                 }
 403         }
 404
 405         /**
 406          * @param array $images
 407          */
 408         private function invalidateImageDescriptions( $images ) {
 409                 PurgeJobUtils::invalidatePages( $this->getDB(), NS_FILE, array_keys( $images ) );
 410         }
 411
 412         /**
 413          * Update a table by doing a delete query then an insert query
 414          * @param string $table Table name
 415          * @param string $prefix Field name prefix
 416          * @param array $deletions
 417          * @param array $insertions Rows to insert
 418          */
 419         private function incrTableUpdate( $table, $prefix, $deletions, $insertions ) {
 420                 $services = MediaWikiServices::getInstance();
 421                 $bSize = $services->getMainConfig()->get( 'UpdateRowsPerQuery' );
 422                 $lbf = $services->getDBLoadBalancerFactory();
 423
 424                 if ( $table === 'page_props' ) {
 425                         $fromField = 'pp_page';
 426                 } else {
 427                         $fromField = "{$prefix}_from";
 428                 }
 429
 430                 $deleteWheres = []; // list of WHERE clause arrays for each DB delete() call
 431                 if ( $table === 'pagelinks' || $table === 'templatelinks' || $table === 'iwlinks' ) {
 432                         $baseKey = ( $table === 'iwlinks' ) ? 'iwl_prefix' : "{$prefix}_namespace";
 433
 434                         $curBatchSize = 0;
 435                         $curDeletionBatch = [];
 436                         $deletionBatches = [];
 437                         foreach ( $deletions as $ns => $dbKeys ) {
 438                                 foreach ( $dbKeys as $dbKey => $unused ) {
 439                                         $curDeletionBatch[$ns][$dbKey] = 1;
 440                                         if ( ++$curBatchSize >= $bSize ) {
 441                                                 $deletionBatches[] = $curDeletionBatch;
 442                                                 $curDeletionBatch = [];
 443                                                 $curBatchSize = 0;
 444                                         }
 445                                 }
 446                         }
 447                         if ( $curDeletionBatch ) {
 448                                 $deletionBatches[] = $curDeletionBatch;
 449                         }
 450
 451                         foreach ( $deletionBatches as $deletionBatch ) {
 452                                 $deleteWheres[] = [
 453                                         $fromField => $this->mId,
 454                                         $this->getDB()->makeWhereFrom2d( $deletionBatch, $baseKey, "{$prefix}_title" )
 455                                 ];
 456                         }
 457                 } else {
 458                         if ( $table === 'langlinks' ) {
 459                                 $toField = 'll_lang';
 460                         } elseif ( $table === 'page_props' ) {
 461                                 $toField = 'pp_propname';
 462                         } else {
 463                                 $toField = $prefix . '_to';
 464                         }
 465
 466                         $deletionBatches = array_chunk( array_keys( $deletions ), $bSize );
 467                         foreach ( $deletionBatches as $deletionBatch ) {
 468                                 $deleteWheres[] = [ $fromField => $this->mId, $toField => $deletionBatch ];
 469                         }
 470                 }
 471
 472                 $domainId = $this->getDB()->getDomainID();
 473
 474                 foreach ( $deleteWheres as $deleteWhere ) {
 475                         $this->getDB()->delete( $table, $deleteWhere, __METHOD__ );
 476                         $lbf->commitAndWaitForReplication(
 477                                 __METHOD__, $this->ticket, [ 'domain' => $domainId ]
 478                         );
 479                 }
 480
 481                 $insertBatches = array_chunk( $insertions, $bSize );
 482                 foreach ( $insertBatches as $insertBatch ) {
 483                         $this->getDB()->insert( $table, $insertBatch, __METHOD__, 'IGNORE' );
 484                         $lbf->commitAndWaitForReplication(
 485                                 __METHOD__, $this->ticket, [ 'domain' => $domainId ]
 486                         );
 487                 }
 488
 489                 if ( count( $insertions ) ) {
 490                         Hooks::run( 'LinksUpdateAfterInsert', [ $this, $table, $insertions ] );
 491                 }
 492         }
 493
 494         /**
 495          * Get an array of pagelinks insertions for passing to the DB
 496          * Skips the titles specified by the 2-D array $existing
 497          * @param array $existing
 498          * @return array
 499          */
 500         private function getLinkInsertions( $existing = [] ) {
 501                 $arr = [];
 502                 foreach ( $this->mLinks as $ns => $dbkeys ) {
 503                         $diffs = isset( $existing[$ns] )
 504                                 ? array_diff_key( $dbkeys, $existing[$ns] )
 505                                 : $dbkeys;
 506                         foreach ( $diffs as $dbk => $id ) {
 507                                 $arr[] = [
 508                                         'pl_from' => $this->mId,
 509                                         'pl_from_namespace' => $this->mTitle->getNamespace(),
 510                                         'pl_namespace' => $ns,
 511                                         'pl_title' => $dbk
 512                                 ];
 513                         }
 514                 }
 515
 516                 return $arr;
 517         }
 518
 519         /**
 520          * Get an array of template insertions. Like getLinkInsertions()
 521          * @param array $existing
 522          * @return array
 523          */
 524         private function getTemplateInsertions( $existing = [] ) {
 525                 $arr = [];
 526                 foreach ( $this->mTemplates as $ns => $dbkeys ) {
 527                         $diffs = isset( $existing[$ns] ) ? array_diff_key( $dbkeys, $existing[$ns] ) : $dbkeys;
 528                         foreach ( $diffs as $dbk => $id ) {
 529                                 $arr[] = [
 530                                         'tl_from' => $this->mId,
 531                                         'tl_from_namespace' => $this->mTitle->getNamespace(),
 532                                         'tl_namespace' => $ns,
 533                                         'tl_title' => $dbk
 534                                 ];
 535                         }
 536                 }
 537
 538                 return $arr;
 539         }
 540
 541         /**
 542          * Get an array of image insertions
 543          * Skips the names specified in $existing
 544          * @param array $existing
 545          * @return array
 546          */
 547         private function getImageInsertions( $existing = [] ) {
 548                 $arr = [];
 549                 $diffs = array_diff_key( $this->mImages, $existing );
 550                 foreach ( $diffs as $iname => $dummy ) {
 551                         $arr[] = [
 552                                 'il_from' => $this->mId,
 553                                 'il_from_namespace' => $this->mTitle->getNamespace(),
 554                                 'il_to' => $iname
 555                         ];
 556                 }
 557
 558                 return $arr;
 559         }
 560
 561         /**
 562          * Get an array of externallinks insertions. Skips the names specified in $existing
 563          * @param array $existing
 564          * @return array
 565          */
 566         private function getExternalInsertions( $existing = [] ) {
 567                 $arr = [];
 568                 $diffs = array_diff_key( $this->mExternals, $existing );
 569                 foreach ( $diffs as $url => $dummy ) {
 570                         foreach ( wfMakeUrlIndexes( $url ) as $index ) {
 571                                 $arr[] = [
 572                                         'el_from' => $this->mId,
 573                                         'el_to' => $url,
 574                                         'el_index' => $index,
 575                                         'el_index_60' => substr( $index, 0, 60 ),
 576                                 ];
 577                         }
 578                 }
 579
 580                 return $arr;
 581         }
 582
 583         /**
 584          * Get an array of category insertions
 585          *
 586          * @param array $existing Mapping existing category names to sort keys. If both
 587          * match a link in $this, the link will be omitted from the output
 588          *
 589          * @return array
 590          */
 591         private function getCategoryInsertions( $existing = [] ) {
 592                 global $wgCategoryCollation;
 593                 $diffs = array_diff_assoc( $this->mCategories, $existing );
 594                 $arr = [];
 595                 $contLang = MediaWikiServices::getInstance()->getContentLanguage();
 596                 $collation = Collation::singleton();
 597                 foreach ( $diffs as $name => $prefix ) {
 598                         $nt = Title::makeTitleSafe( NS_CATEGORY, $name );
 599                         $contLang->findVariantLink( $name, $nt, true );
 600
 601                         $type = MWNamespace::getCategoryLinkType( $this->mTitle->getNamespace() );
 602
 603                         # Treat custom sortkeys as a prefix, so that if multiple
 604                         # things are forced to sort as '*' or something, they'll
 605                         # sort properly in the category rather than in page_id
 606                         # order or such.
 607                         $sortkey = $collation->getSortKey( $this->mTitle->getCategorySortkey( $prefix ) );
 608
 609                         $arr[] = [
 610                                 'cl_from' => $this->mId,
 611                                 'cl_to' => $name,
 612                                 'cl_sortkey' => $sortkey,
 613                                 'cl_timestamp' => $this->getDB()->timestamp(),
 614                                 'cl_sortkey_prefix' => $prefix,
 615                                 'cl_collation' => $wgCategoryCollation,
 616                                 'cl_type' => $type,
 617                         ];
 618                 }
 619
 620                 return $arr;
 621         }
 622
 623         /**
 624          * Get an array of interlanguage link insertions
 625          *
 626          * @param array $existing Mapping existing language codes to titles
 627          *
 628          * @return array
 629          */
 630         private function getInterlangInsertions( $existing = [] ) {
 631                 $diffs = array_diff_assoc( $this->mInterlangs, $existing );
 632                 $arr = [];
 633                 foreach ( $diffs as $lang => $title ) {
 634                         $arr[] = [
 635                                 'll_from' => $this->mId,
 636                                 'll_lang' => $lang,
 637                                 'll_title' => $title
 638                         ];
 639                 }
 640
 641                 return $arr;
 642         }
 643
 644         /**
 645          * Get an array of page property insertions
 646          * @param array $existing
 647          * @return array
 648          */
 649         function getPropertyInsertions( $existing = [] ) {
 650                 $diffs = array_diff_assoc( $this->mProperties, $existing );
 651
 652                 $arr = [];
 653                 foreach ( array_keys( $diffs ) as $name ) {
 654                         $arr[] = $this->getPagePropRowData( $name );
 655                 }
 656
 657                 return $arr;
 658         }
 659
 660         /**
 661          * Returns an associative array to be used for inserting a row into
 662          * the page_props table. Besides the given property name, this will
 663          * include the page id from $this->mId and any property value from
 664          * $this->mProperties.
 665          *
 666          * The array returned will include the pp_sortkey field if this
 667          * is present in the database (as indicated by $wgPagePropsHaveSortkey).
 668          * The sortkey value is currently determined by getPropertySortKeyValue().
 669          *
 670          * @note this assumes that $this->mProperties[$prop] is defined.
 671          *
 672          * @param string $prop The name of the property.
 673          *
 674          * @return array
 675          */
 676         private function getPagePropRowData( $prop ) {
 677                 global $wgPagePropsHaveSortkey;
 678
 679                 $value = $this->mProperties[$prop];
 680
 681                 $row = [
 682                         'pp_page' => $this->mId,
 683                         'pp_propname' => $prop,
 684                         'pp_value' => $value,
 685                 ];
 686
 687                 if ( $wgPagePropsHaveSortkey ) {
 688                         $row['pp_sortkey'] = $this->getPropertySortKeyValue( $value );
 689                 }
 690
 691                 return $row;
 692         }
 693
 694         /**
 695          * Determines the sort key for the given property value.
 696          * This will return $value if it is a float or int,
 697          * 1 or resp. 0 if it is a bool, and null otherwise.
 698          *
 699          * @note In the future, we may allow the sortkey to be specified explicitly
 700          *       in ParserOutput::setProperty.
 701          *
 702          * @param mixed $value
 703          *
 704          * @return float|null
 705          */
 706         private function getPropertySortKeyValue( $value ) {
 707                 if ( is_int( $value ) || is_float( $value ) || is_bool( $value ) ) {
 708                         return floatval( $value );
 709                 }
 710
 711                 return null;
 712         }
 713
 714         /**
 715          * Get an array of interwiki insertions for passing to the DB
 716          * Skips the titles specified by the 2-D array $existing
 717          * @param array $existing
 718          * @return array
 719          */
 720         private function getInterwikiInsertions( $existing = [] ) {
 721                 $arr = [];
 722                 foreach ( $this->mInterwikis as $prefix => $dbkeys ) {
 723                         $diffs = isset( $existing[$prefix] )
 724                                 ? array_diff_key( $dbkeys, $existing[$prefix] )
 725                                 : $dbkeys;
 726
 727                         foreach ( $diffs as $dbk => $id ) {
 728                                 $arr[] = [
 729                                         'iwl_from' => $this->mId,
 730                                         'iwl_prefix' => $prefix,
 731                                         'iwl_title' => $dbk
 732                                 ];
 733                         }
 734                 }
 735
 736                 return $arr;
 737         }
 738
 739         /**
 740          * Given an array of existing links, returns those links which are not in $this
 741          * and thus should be deleted.
 742          * @param array $existing
 743          * @return array
 744          */
 745         private function getLinkDeletions( $existing ) {
 746                 $del = [];
 747                 foreach ( $existing as $ns => $dbkeys ) {
 748                         if ( isset( $this->mLinks[$ns] ) ) {
 749                                 $del[$ns] = array_diff_key( $existing[$ns], $this->mLinks[$ns] );
 750                         } else {
 751                                 $del[$ns] = $existing[$ns];
 752                         }
 753                 }
 754
 755                 return $del;
 756         }
 757
 758         /**
 759          * Given an array of existing templates, returns those templates which are not in $this
 760          * and thus should be deleted.
 761          * @param array $existing
 762          * @return array
 763          */
 764         private function getTemplateDeletions( $existing ) {
 765                 $del = [];
 766                 foreach ( $existing as $ns => $dbkeys ) {
 767                         if ( isset( $this->mTemplates[$ns] ) ) {
 768                                 $del[$ns] = array_diff_key( $existing[$ns], $this->mTemplates[$ns] );
 769                         } else {
 770                                 $del[$ns] = $existing[$ns];
 771                         }
 772                 }
 773
 774                 return $del;
 775         }
 776
 777         /**
 778          * Given an array of existing images, returns those images which are not in $this
 779          * and thus should be deleted.
 780          * @param array $existing
 781          * @return array
 782          */
 783         private function getImageDeletions( $existing ) {
 784                 return array_diff_key( $existing, $this->mImages );
 785         }
 786
 787         /**
 788          * Given an array of existing external links, returns those links which are not
 789          * in $this and thus should be deleted.
 790          * @param array $existing
 791          * @return array
 792          */
 793         private function getExternalDeletions( $existing ) {
 794                 return array_diff_key( $existing, $this->mExternals );
 795         }
 796
 797         /**
 798          * Given an array of existing categories, returns those categories which are not in $this
 799          * and thus should be deleted.
 800          * @param array $existing
 801          * @return array
 802          */
 803         private function getCategoryDeletions( $existing ) {
 804                 return array_diff_assoc( $existing, $this->mCategories );
 805         }
 806
 807         /**
 808          * Given an array of existing interlanguage links, returns those links which are not
 809          * in $this and thus should be deleted.
 810          * @param array $existing
 811          * @return array
 812          */
 813         private function getInterlangDeletions( $existing ) {
 814                 return array_diff_assoc( $existing, $this->mInterlangs );
 815         }
 816
 817         /**
 818          * Get array of properties which should be deleted.
 819          * @param array $existing
 820          * @return array
 821          */
 822         function getPropertyDeletions( $existing ) {
 823                 return array_diff_assoc( $existing, $this->mProperties );
 824         }
 825
 826         /**
 827          * Given an array of existing interwiki links, returns those links which are not in $this
 828          * and thus should be deleted.
 829          * @param array $existing
 830          * @return array
 831          */
 832         private function getInterwikiDeletions( $existing ) {
 833                 $del = [];
 834                 foreach ( $existing as $prefix => $dbkeys ) {
 835                         if ( isset( $this->mInterwikis[$prefix] ) ) {
 836                                 $del[$prefix] = array_diff_key( $existing[$prefix], $this->mInterwikis[$prefix] );
 837                         } else {
 838                                 $del[$prefix] = $existing[$prefix];
 839                         }
 840                 }
 841
 842                 return $del;
 843         }
 844
 845         /**
 846          * Get an array of existing links, as a 2-D array
 847          *
 848          * @return array
 849          */
 850         private function getExistingLinks() {
 851                 $res = $this->getDB()->select( 'pagelinks', [ 'pl_namespace', 'pl_title' ],
 852                         [ 'pl_from' => $this->mId ], __METHOD__ );
 853                 $arr = [];
 854                 foreach ( $res as $row ) {
 855                         if ( !isset( $arr[$row->pl_namespace] ) ) {
 856                                 $arr[$row->pl_namespace] = [];
 857                         }
 858                         $arr[$row->pl_namespace][$row->pl_title] = 1;
 859                 }
 860
 861                 return $arr;
 862         }
 863
 864         /**
 865          * Get an array of existing templates, as a 2-D array
 866          *
 867          * @return array
 868          */
 869         private function getExistingTemplates() {
 870                 $res = $this->getDB()->select( 'templatelinks', [ 'tl_namespace', 'tl_title' ],
 871                         [ 'tl_from' => $this->mId ], __METHOD__ );
 872                 $arr = [];
 873                 foreach ( $res as $row ) {
 874                         if ( !isset( $arr[$row->tl_namespace] ) ) {
 875                                 $arr[$row->tl_namespace] = [];
 876                         }
 877                         $arr[$row->tl_namespace][$row->tl_title] = 1;
 878                 }
 879
 880                 return $arr;
 881         }
 882
 883         /**
 884          * Get an array of existing images, image names in the keys
 885          *
 886          * @return array
 887          */
 888         private function getExistingImages() {
 889                 $res = $this->getDB()->select( 'imagelinks', [ 'il_to' ],
 890                         [ 'il_from' => $this->mId ], __METHOD__ );
 891                 $arr = [];
 892                 foreach ( $res as $row ) {
 893                         $arr[$row->il_to] = 1;
 894                 }
 895
 896                 return $arr;
 897         }
 898
 899         /**
 900          * Get an array of existing external links, URLs in the keys
 901          *
 902          * @return array
 903          */
 904         private function getExistingExternals() {
 905                 $res = $this->getDB()->select( 'externallinks', [ 'el_to' ],
 906                         [ 'el_from' => $this->mId ], __METHOD__ );
 907                 $arr = [];
 908                 foreach ( $res as $row ) {
 909                         $arr[$row->el_to] = 1;
 910                 }
 911
 912                 return $arr;
 913         }
 914
 915         /**
 916          * Get an array of existing categories, with the name in the key and sort key in the value.
 917          *
 918          * @return array
 919          */
 920         private function getExistingCategories() {
 921                 $res = $this->getDB()->select( 'categorylinks', [ 'cl_to', 'cl_sortkey_prefix' ],
 922                         [ 'cl_from' => $this->mId ], __METHOD__ );
 923                 $arr = [];
 924                 foreach ( $res as $row ) {
 925                         $arr[$row->cl_to] = $row->cl_sortkey_prefix;
 926                 }
 927
 928                 return $arr;
 929         }
 930
 931         /**
 932          * Get an array of existing interlanguage links, with the language code in the key and the
 933          * title in the value.
 934          *
 935          * @return array
 936          */
 937         private function getExistingInterlangs() {
 938                 $res = $this->getDB()->select( 'langlinks', [ 'll_lang', 'll_title' ],
 939                         [ 'll_from' => $this->mId ], __METHOD__ );
 940                 $arr = [];
 941                 foreach ( $res as $row ) {
 942                         $arr[$row->ll_lang] = $row->ll_title;
 943                 }
 944
 945                 return $arr;
 946         }
 947
 948         /**
 949          * Get an array of existing inline interwiki links, as a 2-D array
 950          * @return array (prefix => array(dbkey => 1))
 951          */
 952         private function getExistingInterwikis() {
 953                 $res = $this->getDB()->select( 'iwlinks', [ 'iwl_prefix', 'iwl_title' ],
 954                         [ 'iwl_from' => $this->mId ], __METHOD__ );
 955                 $arr = [];
 956                 foreach ( $res as $row ) {
 957                         if ( !isset( $arr[$row->iwl_prefix] ) ) {
 958                                 $arr[$row->iwl_prefix] = [];
 959                         }
 960                         $arr[$row->iwl_prefix][$row->iwl_title] = 1;
 961                 }
 962
 963                 return $arr;
 964         }
 965
 966         /**
 967          * Get an array of existing categories, with the name in the key and sort key in the value.
 968          *
 969          * @return array Array of property names and values
 970          */
 971         private function getExistingProperties() {
 972                 $res = $this->getDB()->select( 'page_props', [ 'pp_propname', 'pp_value' ],
 973                         [ 'pp_page' => $this->mId ], __METHOD__ );
 974                 $arr = [];
 975                 foreach ( $res as $row ) {
 976                         $arr[$row->pp_propname] = $row->pp_value;
 977                 }
 978
 979                 return $arr;
 980         }
 981
 982         /**
 983          * Return the title object of the page being updated
 984          * @return Title
 985          */
 986         public function getTitle() {
 987                 return $this->mTitle;
 988         }
 989
 990         /**
 991          * Returns parser output
 992          * @since 1.19
 993          * @return ParserOutput
 994          */
 995         public function getParserOutput() {
 996                 return $this->mParserOutput;
 997         }
 998
 999         /**
1000          * Return the list of images used as generated by the parser
1001          * @return array
1002          */
1003         public function getImages() {
1004                 return $this->mImages;
1005         }
1006
1007         /**
1008          * Set the revision corresponding to this LinksUpdate
1009          *
1010          * @since 1.27
1011          *
1012          * @param Revision $revision
1013          */
1014         public function setRevision( Revision $revision ) {
1015                 $this->mRevision = $revision;
1016         }
1017
1018         /**
1019          * @since 1.28
1020          * @return null|Revision
1021          */
1022         public function getRevision() {
1023                 return $this->mRevision;
1024         }
1025
1026         /**
1027          * Set the User who triggered this LinksUpdate
1028          *
1029          * @since 1.27
1030          * @param User $user
1031          */
1032         public function setTriggeringUser( User $user ) {
1033                 $this->user = $user;
1034         }
1035
1036         /**
1037          * @since 1.27
1038          * @return null|User
1039          */
1040         public function getTriggeringUser() {
1041                 return $this->user;
1042         }
1043
1044         /**
1045          * Invalidate any necessary link lists related to page property changes
1046          * @param array $changed
1047          */
1048         private function invalidateProperties( $changed ) {
1049                 global $wgPagePropLinkInvalidations;
1050
1051                 foreach ( $changed as $name => $value ) {
1052                         if ( isset( $wgPagePropLinkInvalidations[$name] ) ) {
1053                                 $inv = $wgPagePropLinkInvalidations[$name];
1054                                 if ( !is_array( $inv ) ) {
1055                                         $inv = [ $inv ];
1056                                 }
1057                                 foreach ( $inv as $table ) {
1058                                         DeferredUpdates::addUpdate(
1059                                                 new HTMLCacheUpdate( $this->mTitle, $table, 'page-props' )
1060                                         );
1061                                 }
1062                         }
1063                 }
1064         }
1065
1066         /**
1067          * Fetch page links added by this LinksUpdate.  Only available after the update is complete.
1068          * @since 1.22
1069          * @return null|array Array of Titles
1070          */
1071         public function getAddedLinks() {
1072                 if ( $this->linkInsertions === null ) {
1073                         return null;
1074                 }
1075                 $result = [];
1076                 foreach ( $this->linkInsertions as $insertion ) {
1077                         $result[] = Title::makeTitle( $insertion['pl_namespace'], $insertion['pl_title'] );
1078                 }
1079
1080                 return $result;
1081         }
1082
1083         /**
1084          * Fetch page links removed by this LinksUpdate.  Only available after the update is complete.
1085          * @since 1.22
1086          * @return null|array Array of Titles
1087          */
1088         public function getRemovedLinks() {
1089                 if ( $this->linkDeletions === null ) {
1090                         return null;
1091                 }
1092                 $result = [];
1093                 foreach ( $this->linkDeletions as $ns => $titles ) {
1094                         foreach ( $titles as $title => $unused ) {
1095                                 $result[] = Title::makeTitle( $ns, $title );
1096                         }
1097                 }
1098
1099                 return $result;
1100         }
1101
1102         /**
1103          * Fetch page properties added by this LinksUpdate.
1104          * Only available after the update is complete.
1105          * @since 1.28
1106          * @return null|array
1107          */
1108         public function getAddedProperties() {
1109                 return $this->propertyInsertions;
1110         }
1111
1112         /**
1113          * Fetch page properties removed by this LinksUpdate.
1114          * Only available after the update is complete.
1115          * @since 1.28
1116          * @return null|array
1117          */
1118         public function getRemovedProperties() {
1119                 return $this->propertyDeletions;
1120         }
1121
1122         /**
1123          * Update links table freshness
1124          */
1125         private function updateLinksTimestamp() {
1126                 if ( $this->mId ) {
1127                         // The link updates made here only reflect the freshness of the parser output
1128                         $timestamp = $this->mParserOutput->getCacheTime();
1129                         $this->getDB()->update( 'page',
1130                                 [ 'page_links_updated' => $this->getDB()->timestamp( $timestamp ) ],
1131                                 [ 'page_id' => $this->mId ],
1132                                 __METHOD__
1133                         );
1134                 }
1135         }
1136
1137         /**
1138          * @return IDatabase
1139          */
1140         private function getDB() {
1141                 if ( !$this->db ) {
1142                         $this->db = wfGetDB( DB_MASTER );
1143                 }
1144
1145                 return $this->db;
1146         }
1147
1148         public function getAsJobSpecification() {
1149                 if ( $this->user ) {
1150                         $userInfo = [
1151                                 'userId' => $this->user->getId(),
1152                                 'userName' => $this->user->getName(),
1153                         ];
1154                 } else {
1155                         $userInfo = false;
1156                 }
1157
1158                 if ( $this->mRevision ) {
1159                         $triggeringRevisionId = $this->mRevision->getId();
1160                 } else {
1161                         $triggeringRevisionId = false;
1162                 }
1163
1164                 return [
1165                         'wiki' => WikiMap::getWikiIdFromDomain( $this->getDB()->getDomainID() ),
1166                         'job'  => new JobSpecification(
1167                                 'refreshLinksPrioritized',
1168                                 [
1169                                         // Reuse the parser cache if it was saved
1170                                         'rootJobTimestamp' => $this->mParserOutput->getCacheTime(),
1171                                         'useRecursiveLinksUpdate' => $this->mRecursive,
1172                                         'triggeringUser' => $userInfo,
1173                                         'triggeringRevisionId' => $triggeringRevisionId,
1174                                         'causeAction' => $this->getCauseAction(),
1175                                         'causeAgent' => $this->getCauseAgent()
1176                                 ],
1177                                 [ 'removeDuplicates' => true ],
1178                                 $this->getTitle()
1179                         )
1180                 ];
1181         }
1182 }