includes/deferred/LinksUpdate.php

   1 <?php
   2 /**
   3  * Updater for link tracking tables after a page edit.
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  */
  22
  23 use Wikimedia\Rdbms\IDatabase;
  24 use MediaWiki\MediaWikiServices;
  25 use Wikimedia\ScopedCallback;
  26
  27 /**
  28  * Class the manages updates of *_link tables as well as similar extension-managed tables
  29  *
  30  * @note: LinksUpdate is managed by DeferredUpdates::execute(). Do not run this in a transaction.
  31  *
  32  * See docs/deferred.txt
  33  */
  34 class LinksUpdate extends DataUpdate implements EnqueueableDataUpdate {
  35         // @todo make members protected, but make sure extensions don't break
  36
  37         /** @var int Page ID of the article linked from */
  38         public $mId;
  39
  40         /** @var Title Title object of the article linked from */
  41         public $mTitle;
  42
  43         /** @var ParserOutput */
  44         public $mParserOutput;
  45
  46         /** @var array Map of title strings to IDs for the links in the document */
  47         public $mLinks;
  48
  49         /** @var array DB keys of the images used, in the array key only */
  50         public $mImages;
  51
  52         /** @var array Map of title strings to IDs for the template references, including broken ones */
  53         public $mTemplates;
  54
  55         /** @var array URLs of external links, array key only */
  56         public $mExternals;
  57
  58         /** @var array Map of category names to sort keys */
  59         public $mCategories;
  60
  61         /** @var array Map of language codes to titles */
  62         public $mInterlangs;
  63
  64         /** @var array 2-D map of (prefix => DBK => 1) */
  65         public $mInterwikis;
  66
  67         /** @var array Map of arbitrary name to value */
  68         public $mProperties;
  69
  70         /** @var bool Whether to queue jobs for recursive updates */
  71         public $mRecursive;
  72
  73         /** @var Revision Revision for which this update has been triggered */
  74         private $mRevision;
  75
  76         /**
  77          * @var null|array Added links if calculated.
  78          */
  79         private $linkInsertions = null;
  80
  81         /**
  82          * @var null|array Deleted links if calculated.
  83          */
  84         private $linkDeletions = null;
  85
  86         /**
  87          * @var null|array Added properties if calculated.
  88          */
  89         private $propertyInsertions = null;
  90
  91         /**
  92          * @var null|array Deleted properties if calculated.
  93          */
  94         private $propertyDeletions = null;
  95
  96         /**
  97          * @var User|null
  98          */
  99         private $user;
 100
 101         /** @var IDatabase */
 102         private $db;
 103
 104         /**
 105          * Constructor
 106          *
 107          * @param Title $title Title of the page we're updating
 108          * @param ParserOutput $parserOutput Output from a full parse of this page
 109          * @param bool $recursive Queue jobs for recursive updates?
 110          * @throws MWException
 111          */
 112         function __construct( Title $title, ParserOutput $parserOutput, $recursive = true ) {
 113                 parent::__construct();
 114
 115                 $this->mTitle = $title;
 116                 $this->mId = $title->getArticleID( Title::GAID_FOR_UPDATE );
 117
 118                 if ( !$this->mId ) {
 119                         throw new InvalidArgumentException(
 120                                 "The Title object yields no ID. Perhaps the page doesn't exist?"
 121                         );
 122                 }
 123
 124                 $this->mParserOutput = $parserOutput;
 125
 126                 $this->mLinks = $parserOutput->getLinks();
 127                 $this->mImages = $parserOutput->getImages();
 128                 $this->mTemplates = $parserOutput->getTemplates();
 129                 $this->mExternals = $parserOutput->getExternalLinks();
 130                 $this->mCategories = $parserOutput->getCategories();
 131                 $this->mProperties = $parserOutput->getProperties();
 132                 $this->mInterwikis = $parserOutput->getInterwikiLinks();
 133
 134                 # Convert the format of the interlanguage links
 135                 # I didn't want to change it in the ParserOutput, because that array is passed all
 136                 # the way back to the skin, so either a skin API break would be required, or an
 137                 # inefficient back-conversion.
 138                 $ill = $parserOutput->getLanguageLinks();
 139                 $this->mInterlangs = [];
 140                 foreach ( $ill as $link ) {
 141                         list( $key, $title ) = explode( ':', $link, 2 );
 142                         $this->mInterlangs[$key] = $title;
 143                 }
 144
 145                 foreach ( $this->mCategories as &$sortkey ) {
 146                         # If the sortkey is longer then 255 bytes,
 147                         # it truncated by DB, and then doesn't get
 148                         # matched when comparing existing vs current
 149                         # categories, causing T27254.
 150                         # Also. substr behaves weird when given "".
 151                         if ( $sortkey !== '' ) {
 152                                 $sortkey = substr( $sortkey, 0, 255 );
 153                         }
 154                 }
 155
 156                 $this->mRecursive = $recursive;
 157
 158                 // Avoid PHP 7.1 warning from passing $this by reference
 159                 $linksUpdate = $this;
 160                 Hooks::run( 'LinksUpdateConstructed', [ &$linksUpdate ] );
 161         }
 162
 163         /**
 164          * Update link tables with outgoing links from an updated article
 165          *
 166          * @note: this is managed by DeferredUpdates::execute(). Do not run this in a transaction.
 167          */
 168         public function doUpdate() {
 169                 if ( $this->ticket ) {
 170                         // Make sure all links update threads see the changes of each other.
 171                         // This handles the case when updates have to batched into several COMMITs.
 172                         $scopedLock = self::acquirePageLock( $this->getDB(), $this->mId );
 173                 }
 174
 175                 // Avoid PHP 7.1 warning from passing $this by reference
 176                 $linksUpdate = $this;
 177                 Hooks::run( 'LinksUpdate', [ &$linksUpdate ] );
 178                 $this->doIncrementalUpdate();
 179
 180                 // Commit and release the lock (if set)
 181                 ScopedCallback::consume( $scopedLock );
 182                 // Run post-commit hooks without DBO_TRX
 183                 $this->getDB()->onTransactionIdle(
 184                         function () {
 185                                 // Avoid PHP 7.1 warning from passing $this by reference
 186                                 $linksUpdate = $this;
 187                                 Hooks::run( 'LinksUpdateComplete', [ &$linksUpdate, $this->ticket ] );
 188                         },
 189                         __METHOD__
 190                 );
 191         }
 192
 193         /**
 194          * Acquire a lock for performing link table updates for a page on a DB
 195          *
 196          * @param IDatabase $dbw
 197          * @param integer $pageId
 198          * @param string $why One of (job, atomicity)
 199          * @return ScopedCallback
 200          * @throws RuntimeException
 201          * @since 1.27
 202          */
 203         public static function acquirePageLock( IDatabase $dbw, $pageId, $why = 'atomicity' ) {
 204                 $key = "LinksUpdate:$why:pageid:$pageId";
 205                 $scopedLock = $dbw->getScopedLockAndFlush( $key, __METHOD__, 15 );
 206                 if ( !$scopedLock ) {
 207                         throw new RuntimeException( "Could not acquire lock '$key'." );
 208                 }
 209
 210                 return $scopedLock;
 211         }
 212
 213         protected function doIncrementalUpdate() {
 214                 # Page links
 215                 $existingPL = $this->getExistingLinks();
 216                 $this->linkDeletions = $this->getLinkDeletions( $existingPL );
 217                 $this->linkInsertions = $this->getLinkInsertions( $existingPL );
 218                 $this->incrTableUpdate( 'pagelinks', 'pl', $this->linkDeletions, $this->linkInsertions );
 219
 220                 # Image links
 221                 $existingIL = $this->getExistingImages();
 222                 $imageDeletes = $this->getImageDeletions( $existingIL );
 223                 $this->incrTableUpdate(
 224                         'imagelinks',
 225                         'il',
 226                         $imageDeletes,
 227                         $this->getImageInsertions( $existingIL ) );
 228
 229                 # Invalidate all image description pages which had links added or removed
 230                 $imageUpdates = $imageDeletes + array_diff_key( $this->mImages, $existingIL );
 231                 $this->invalidateImageDescriptions( $imageUpdates );
 232
 233                 # External links
 234                 $existingEL = $this->getExistingExternals();
 235                 $this->incrTableUpdate(
 236                         'externallinks',
 237                         'el',
 238                         $this->getExternalDeletions( $existingEL ),
 239                         $this->getExternalInsertions( $existingEL ) );
 240
 241                 # Language links
 242                 $existingLL = $this->getExistingInterlangs();
 243                 $this->incrTableUpdate(
 244                         'langlinks',
 245                         'll',
 246                         $this->getInterlangDeletions( $existingLL ),
 247                         $this->getInterlangInsertions( $existingLL ) );
 248
 249                 # Inline interwiki links
 250                 $existingIW = $this->getExistingInterwikis();
 251                 $this->incrTableUpdate(
 252                         'iwlinks',
 253                         'iwl',
 254                         $this->getInterwikiDeletions( $existingIW ),
 255                         $this->getInterwikiInsertions( $existingIW ) );
 256
 257                 # Template links
 258                 $existingTL = $this->getExistingTemplates();
 259                 $this->incrTableUpdate(
 260                         'templatelinks',
 261                         'tl',
 262                         $this->getTemplateDeletions( $existingTL ),
 263                         $this->getTemplateInsertions( $existingTL ) );
 264
 265                 # Category links
 266                 $existingCL = $this->getExistingCategories();
 267                 $categoryDeletes = $this->getCategoryDeletions( $existingCL );
 268                 $this->incrTableUpdate(
 269                         'categorylinks',
 270                         'cl',
 271                         $categoryDeletes,
 272                         $this->getCategoryInsertions( $existingCL ) );
 273                 $categoryInserts = array_diff_assoc( $this->mCategories, $existingCL );
 274                 $categoryUpdates = $categoryInserts + $categoryDeletes;
 275
 276                 # Page properties
 277                 $existingPP = $this->getExistingProperties();
 278                 $this->propertyDeletions = $this->getPropertyDeletions( $existingPP );
 279                 $this->incrTableUpdate(
 280                         'page_props',
 281                         'pp',
 282                         $this->propertyDeletions,
 283                         $this->getPropertyInsertions( $existingPP ) );
 284
 285                 # Invalidate the necessary pages
 286                 $this->propertyInsertions = array_diff_assoc( $this->mProperties, $existingPP );
 287                 $changed = $this->propertyDeletions + $this->propertyInsertions;
 288                 $this->invalidateProperties( $changed );
 289
 290                 # Invalidate all categories which were added, deleted or changed (set symmetric difference)
 291                 $this->invalidateCategories( $categoryUpdates );
 292                 $this->updateCategoryCounts( $categoryInserts, $categoryDeletes );
 293
 294                 # Refresh links of all pages including this page
 295                 # This will be in a separate transaction
 296                 if ( $this->mRecursive ) {
 297                         $this->queueRecursiveJobs();
 298                 }
 299
 300                 # Update the links table freshness for this title
 301                 $this->updateLinksTimestamp();
 302         }
 303
 304         /**
 305          * Queue recursive jobs for this page
 306          *
 307          * Which means do LinksUpdate on all pages that include the current page,
 308          * using the job queue.
 309          */
 310         protected function queueRecursiveJobs() {
 311                 self::queueRecursiveJobsForTable( $this->mTitle, 'templatelinks' );
 312                 if ( $this->mTitle->getNamespace() == NS_FILE ) {
 313                         // Process imagelinks in case the title is or was a redirect
 314                         self::queueRecursiveJobsForTable( $this->mTitle, 'imagelinks' );
 315                 }
 316
 317                 $bc = $this->mTitle->getBacklinkCache();
 318                 // Get jobs for cascade-protected backlinks for a high priority queue.
 319                 // If meta-templates change to using a new template, the new template
 320                 // should be implicitly protected as soon as possible, if applicable.
 321                 // These jobs duplicate a subset of the above ones, but can run sooner.
 322                 // Which ever runs first generally no-ops the other one.
 323                 $jobs = [];
 324                 foreach ( $bc->getCascadeProtectedLinks() as $title ) {
 325                         $jobs[] = RefreshLinksJob::newPrioritized( $title, [] );
 326                 }
 327                 JobQueueGroup::singleton()->push( $jobs );
 328         }
 329
 330         /**
 331          * Queue a RefreshLinks job for any table.
 332          *
 333          * @param Title $title Title to do job for
 334          * @param string $table Table to use (e.g. 'templatelinks')
 335          */
 336         public static function queueRecursiveJobsForTable( Title $title, $table ) {
 337                 if ( $title->getBacklinkCache()->hasLinks( $table ) ) {
 338                         $job = new RefreshLinksJob(
 339                                 $title,
 340                                 [
 341                                         'table' => $table,
 342                                         'recursive' => true,
 343                                 ] + Job::newRootJobParams( // "overall" refresh links job info
 344                                         "refreshlinks:{$table}:{$title->getPrefixedText()}"
 345                                 )
 346                         );
 347
 348                         JobQueueGroup::singleton()->push( $job );
 349                 }
 350         }
 351
 352         /**
 353          * @param array $cats
 354          */
 355         private function invalidateCategories( $cats ) {
 356                 PurgeJobUtils::invalidatePages( $this->getDB(), NS_CATEGORY, array_keys( $cats ) );
 357         }
 358
 359         /**
 360          * Update all the appropriate counts in the category table.
 361          * @param array $added Associative array of category name => sort key
 362          * @param array $deleted Associative array of category name => sort key
 363          */
 364         private function updateCategoryCounts( array $added, array $deleted ) {
 365                 global $wgUpdateRowsPerQuery;
 366
 367                 if ( !$added && !$deleted ) {
 368                         return;
 369                 }
 370
 371                 $wikiId = $this->getDB()->getWikiID();
 372                 $wp = WikiPage::factory( $this->mTitle );
 373                 $lbf = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
 374                 // T163801: try to release any row locks to reduce contention
 375                 $lbf->commitAndWaitForReplication( __METHOD__, $this->ticket, [ 'wiki' => $wikiId ] );
 376
 377                 foreach ( array_chunk( array_keys( $added ), $wgUpdateRowsPerQuery ) as $addBatch ) {
 378                         $wp->updateCategoryCounts( $addBatch, [], $this->mId );
 379                         $lbf->commitAndWaitForReplication( __METHOD__, $this->ticket, [ 'wiki' => $wikiId ] );
 380                 }
 381
 382                 foreach ( array_chunk( array_keys( $deleted ), $wgUpdateRowsPerQuery ) as $deleteBatch ) {
 383                         $wp->updateCategoryCounts( [], $deleteBatch, $this->mId );
 384                         $lbf->commitAndWaitForReplication( __METHOD__, $this->ticket, [ 'wiki' => $wikiId ] );
 385                 }
 386         }
 387
 388         /**
 389          * @param array $images
 390          */
 391         private function invalidateImageDescriptions( $images ) {
 392                 PurgeJobUtils::invalidatePages( $this->getDB(), NS_FILE, array_keys( $images ) );
 393         }
 394
 395         /**
 396          * Update a table by doing a delete query then an insert query
 397          * @param string $table Table name
 398          * @param string $prefix Field name prefix
 399          * @param array $deletions
 400          * @param array $insertions Rows to insert
 401          */
 402         private function incrTableUpdate( $table, $prefix, $deletions, $insertions ) {
 403                 $services = MediaWikiServices::getInstance();
 404                 $bSize = $services->getMainConfig()->get( 'UpdateRowsPerQuery' );
 405                 $lbf = $services->getDBLoadBalancerFactory();
 406
 407                 if ( $table === 'page_props' ) {
 408                         $fromField = 'pp_page';
 409                 } else {
 410                         $fromField = "{$prefix}_from";
 411                 }
 412
 413                 $deleteWheres = []; // list of WHERE clause arrays for each DB delete() call
 414                 if ( $table === 'pagelinks' || $table === 'templatelinks' || $table === 'iwlinks' ) {
 415                         $baseKey = ( $table === 'iwlinks' ) ? 'iwl_prefix' : "{$prefix}_namespace";
 416
 417                         $curBatchSize = 0;
 418                         $curDeletionBatch = [];
 419                         $deletionBatches = [];
 420                         foreach ( $deletions as $ns => $dbKeys ) {
 421                                 foreach ( $dbKeys as $dbKey => $unused ) {
 422                                         $curDeletionBatch[$ns][$dbKey] = 1;
 423                                         if ( ++$curBatchSize >= $bSize ) {
 424                                                 $deletionBatches[] = $curDeletionBatch;
 425                                                 $curDeletionBatch = [];
 426                                                 $curBatchSize = 0;
 427                                         }
 428                                 }
 429                         }
 430                         if ( $curDeletionBatch ) {
 431                                 $deletionBatches[] = $curDeletionBatch;
 432                         }
 433
 434                         foreach ( $deletionBatches as $deletionBatch ) {
 435                                 $deleteWheres[] = [
 436                                         $fromField => $this->mId,
 437                                         $this->getDB()->makeWhereFrom2d( $deletionBatch, $baseKey, "{$prefix}_title" )
 438                                 ];
 439                         }
 440                 } else {
 441                         if ( $table === 'langlinks' ) {
 442                                 $toField = 'll_lang';
 443                         } elseif ( $table === 'page_props' ) {
 444                                 $toField = 'pp_propname';
 445                         } else {
 446                                 $toField = $prefix . '_to';
 447                         }
 448
 449                         $deletionBatches = array_chunk( array_keys( $deletions ), $bSize );
 450                         foreach ( $deletionBatches as $deletionBatch ) {
 451                                 $deleteWheres[] = [ $fromField => $this->mId, $toField => $deletionBatch ];
 452                         }
 453                 }
 454
 455                 foreach ( $deleteWheres as $deleteWhere ) {
 456                         $this->getDB()->delete( $table, $deleteWhere, __METHOD__ );
 457                         $lbf->commitAndWaitForReplication(
 458                                 __METHOD__, $this->ticket, [ 'wiki' => $this->getDB()->getWikiID() ]
 459                         );
 460                 }
 461
 462                 $insertBatches = array_chunk( $insertions, $bSize );
 463                 foreach ( $insertBatches as $insertBatch ) {
 464                         $this->getDB()->insert( $table, $insertBatch, __METHOD__, 'IGNORE' );
 465                         $lbf->commitAndWaitForReplication(
 466                                 __METHOD__, $this->ticket, [ 'wiki' => $this->getDB()->getWikiID() ]
 467                         );
 468                 }
 469
 470                 if ( count( $insertions ) ) {
 471                         Hooks::run( 'LinksUpdateAfterInsert', [ $this, $table, $insertions ] );
 472                 }
 473         }
 474
 475         /**
 476          * Get an array of pagelinks insertions for passing to the DB
 477          * Skips the titles specified by the 2-D array $existing
 478          * @param array $existing
 479          * @return array
 480          */
 481         private function getLinkInsertions( $existing = [] ) {
 482                 $arr = [];
 483                 foreach ( $this->mLinks as $ns => $dbkeys ) {
 484                         $diffs = isset( $existing[$ns] )
 485                                 ? array_diff_key( $dbkeys, $existing[$ns] )
 486                                 : $dbkeys;
 487                         foreach ( $diffs as $dbk => $id ) {
 488                                 $arr[] = [
 489                                         'pl_from' => $this->mId,
 490                                         'pl_from_namespace' => $this->mTitle->getNamespace(),
 491                                         'pl_namespace' => $ns,
 492                                         'pl_title' => $dbk
 493                                 ];
 494                         }
 495                 }
 496
 497                 return $arr;
 498         }
 499
 500         /**
 501          * Get an array of template insertions. Like getLinkInsertions()
 502          * @param array $existing
 503          * @return array
 504          */
 505         private function getTemplateInsertions( $existing = [] ) {
 506                 $arr = [];
 507                 foreach ( $this->mTemplates as $ns => $dbkeys ) {
 508                         $diffs = isset( $existing[$ns] ) ? array_diff_key( $dbkeys, $existing[$ns] ) : $dbkeys;
 509                         foreach ( $diffs as $dbk => $id ) {
 510                                 $arr[] = [
 511                                         'tl_from' => $this->mId,
 512                                         'tl_from_namespace' => $this->mTitle->getNamespace(),
 513                                         'tl_namespace' => $ns,
 514                                         'tl_title' => $dbk
 515                                 ];
 516                         }
 517                 }
 518
 519                 return $arr;
 520         }
 521
 522         /**
 523          * Get an array of image insertions
 524          * Skips the names specified in $existing
 525          * @param array $existing
 526          * @return array
 527          */
 528         private function getImageInsertions( $existing = [] ) {
 529                 $arr = [];
 530                 $diffs = array_diff_key( $this->mImages, $existing );
 531                 foreach ( $diffs as $iname => $dummy ) {
 532                         $arr[] = [
 533                                 'il_from' => $this->mId,
 534                                 'il_from_namespace' => $this->mTitle->getNamespace(),
 535                                 'il_to' => $iname
 536                         ];
 537                 }
 538
 539                 return $arr;
 540         }
 541
 542         /**
 543          * Get an array of externallinks insertions. Skips the names specified in $existing
 544          * @param array $existing
 545          * @return array
 546          */
 547         private function getExternalInsertions( $existing = [] ) {
 548                 $arr = [];
 549                 $diffs = array_diff_key( $this->mExternals, $existing );
 550                 foreach ( $diffs as $url => $dummy ) {
 551                         foreach ( wfMakeUrlIndexes( $url ) as $index ) {
 552                                 $arr[] = [
 553                                         'el_id' => $this->getDB()->nextSequenceValue( 'externallinks_el_id_seq' ),
 554                                         'el_from' => $this->mId,
 555                                         'el_to' => $url,
 556                                         'el_index' => $index,
 557                                 ];
 558                         }
 559                 }
 560
 561                 return $arr;
 562         }
 563
 564         /**
 565          * Get an array of category insertions
 566          *
 567          * @param array $existing Mapping existing category names to sort keys. If both
 568          * match a link in $this, the link will be omitted from the output
 569          *
 570          * @return array
 571          */
 572         private function getCategoryInsertions( $existing = [] ) {
 573                 global $wgContLang, $wgCategoryCollation;
 574                 $diffs = array_diff_assoc( $this->mCategories, $existing );
 575                 $arr = [];
 576                 foreach ( $diffs as $name => $prefix ) {
 577                         $nt = Title::makeTitleSafe( NS_CATEGORY, $name );
 578                         $wgContLang->findVariantLink( $name, $nt, true );
 579
 580                         if ( $this->mTitle->getNamespace() == NS_CATEGORY ) {
 581                                 $type = 'subcat';
 582                         } elseif ( $this->mTitle->getNamespace() == NS_FILE ) {
 583                                 $type = 'file';
 584                         } else {
 585                                 $type = 'page';
 586                         }
 587
 588                         # Treat custom sortkeys as a prefix, so that if multiple
 589                         # things are forced to sort as '*' or something, they'll
 590                         # sort properly in the category rather than in page_id
 591                         # order or such.
 592                         $sortkey = Collation::singleton()->getSortKey(
 593                                 $this->mTitle->getCategorySortkey( $prefix ) );
 594
 595                         $arr[] = [
 596                                 'cl_from' => $this->mId,
 597                                 'cl_to' => $name,
 598                                 'cl_sortkey' => $sortkey,
 599                                 'cl_timestamp' => $this->getDB()->timestamp(),
 600                                 'cl_sortkey_prefix' => $prefix,
 601                                 'cl_collation' => $wgCategoryCollation,
 602                                 'cl_type' => $type,
 603                         ];
 604                 }
 605
 606                 return $arr;
 607         }
 608
 609         /**
 610          * Get an array of interlanguage link insertions
 611          *
 612          * @param array $existing Mapping existing language codes to titles
 613          *
 614          * @return array
 615          */
 616         private function getInterlangInsertions( $existing = [] ) {
 617                 $diffs = array_diff_assoc( $this->mInterlangs, $existing );
 618                 $arr = [];
 619                 foreach ( $diffs as $lang => $title ) {
 620                         $arr[] = [
 621                                 'll_from' => $this->mId,
 622                                 'll_lang' => $lang,
 623                                 'll_title' => $title
 624                         ];
 625                 }
 626
 627                 return $arr;
 628         }
 629
 630         /**
 631          * Get an array of page property insertions
 632          * @param array $existing
 633          * @return array
 634          */
 635         function getPropertyInsertions( $existing = [] ) {
 636                 $diffs = array_diff_assoc( $this->mProperties, $existing );
 637
 638                 $arr = [];
 639                 foreach ( array_keys( $diffs ) as $name ) {
 640                         $arr[] = $this->getPagePropRowData( $name );
 641                 }
 642
 643                 return $arr;
 644         }
 645
 646         /**
 647          * Returns an associative array to be used for inserting a row into
 648          * the page_props table. Besides the given property name, this will
 649          * include the page id from $this->mId and any property value from
 650          * $this->mProperties.
 651          *
 652          * The array returned will include the pp_sortkey field if this
 653          * is present in the database (as indicated by $wgPagePropsHaveSortkey).
 654          * The sortkey value is currently determined by getPropertySortKeyValue().
 655          *
 656          * @note this assumes that $this->mProperties[$prop] is defined.
 657          *
 658          * @param string $prop The name of the property.
 659          *
 660          * @return array
 661          */
 662         private function getPagePropRowData( $prop ) {
 663                 global $wgPagePropsHaveSortkey;
 664
 665                 $value = $this->mProperties[$prop];
 666
 667                 $row = [
 668                         'pp_page' => $this->mId,
 669                         'pp_propname' => $prop,
 670                         'pp_value' => $value,
 671                 ];
 672
 673                 if ( $wgPagePropsHaveSortkey ) {
 674                         $row['pp_sortkey'] = $this->getPropertySortKeyValue( $value );
 675                 }
 676
 677                 return $row;
 678         }
 679
 680         /**
 681          * Determines the sort key for the given property value.
 682          * This will return $value if it is a float or int,
 683          * 1 or resp. 0 if it is a bool, and null otherwise.
 684          *
 685          * @note In the future, we may allow the sortkey to be specified explicitly
 686          *       in ParserOutput::setProperty.
 687          *
 688          * @param mixed $value
 689          *
 690          * @return float|null
 691          */
 692         private function getPropertySortKeyValue( $value ) {
 693                 if ( is_int( $value ) || is_float( $value ) || is_bool( $value ) ) {
 694                         return floatval( $value );
 695                 }
 696
 697                 return null;
 698         }
 699
 700         /**
 701          * Get an array of interwiki insertions for passing to the DB
 702          * Skips the titles specified by the 2-D array $existing
 703          * @param array $existing
 704          * @return array
 705          */
 706         private function getInterwikiInsertions( $existing = [] ) {
 707                 $arr = [];
 708                 foreach ( $this->mInterwikis as $prefix => $dbkeys ) {
 709                         $diffs = isset( $existing[$prefix] )
 710                                 ? array_diff_key( $dbkeys, $existing[$prefix] )
 711                                 : $dbkeys;
 712
 713                         foreach ( $diffs as $dbk => $id ) {
 714                                 $arr[] = [
 715                                         'iwl_from' => $this->mId,
 716                                         'iwl_prefix' => $prefix,
 717                                         'iwl_title' => $dbk
 718                                 ];
 719                         }
 720                 }
 721
 722                 return $arr;
 723         }
 724
 725         /**
 726          * Given an array of existing links, returns those links which are not in $this
 727          * and thus should be deleted.
 728          * @param array $existing
 729          * @return array
 730          */
 731         private function getLinkDeletions( $existing ) {
 732                 $del = [];
 733                 foreach ( $existing as $ns => $dbkeys ) {
 734                         if ( isset( $this->mLinks[$ns] ) ) {
 735                                 $del[$ns] = array_diff_key( $existing[$ns], $this->mLinks[$ns] );
 736                         } else {
 737                                 $del[$ns] = $existing[$ns];
 738                         }
 739                 }
 740
 741                 return $del;
 742         }
 743
 744         /**
 745          * Given an array of existing templates, returns those templates which are not in $this
 746          * and thus should be deleted.
 747          * @param array $existing
 748          * @return array
 749          */
 750         private function getTemplateDeletions( $existing ) {
 751                 $del = [];
 752                 foreach ( $existing as $ns => $dbkeys ) {
 753                         if ( isset( $this->mTemplates[$ns] ) ) {
 754                                 $del[$ns] = array_diff_key( $existing[$ns], $this->mTemplates[$ns] );
 755                         } else {
 756                                 $del[$ns] = $existing[$ns];
 757                         }
 758                 }
 759
 760                 return $del;
 761         }
 762
 763         /**
 764          * Given an array of existing images, returns those images which are not in $this
 765          * and thus should be deleted.
 766          * @param array $existing
 767          * @return array
 768          */
 769         private function getImageDeletions( $existing ) {
 770                 return array_diff_key( $existing, $this->mImages );
 771         }
 772
 773         /**
 774          * Given an array of existing external links, returns those links which are not
 775          * in $this and thus should be deleted.
 776          * @param array $existing
 777          * @return array
 778          */
 779         private function getExternalDeletions( $existing ) {
 780                 return array_diff_key( $existing, $this->mExternals );
 781         }
 782
 783         /**
 784          * Given an array of existing categories, returns those categories which are not in $this
 785          * and thus should be deleted.
 786          * @param array $existing
 787          * @return array
 788          */
 789         private function getCategoryDeletions( $existing ) {
 790                 return array_diff_assoc( $existing, $this->mCategories );
 791         }
 792
 793         /**
 794          * Given an array of existing interlanguage links, returns those links which are not
 795          * in $this and thus should be deleted.
 796          * @param array $existing
 797          * @return array
 798          */
 799         private function getInterlangDeletions( $existing ) {
 800                 return array_diff_assoc( $existing, $this->mInterlangs );
 801         }
 802
 803         /**
 804          * Get array of properties which should be deleted.
 805          * @param array $existing
 806          * @return array
 807          */
 808         function getPropertyDeletions( $existing ) {
 809                 return array_diff_assoc( $existing, $this->mProperties );
 810         }
 811
 812         /**
 813          * Given an array of existing interwiki links, returns those links which are not in $this
 814          * and thus should be deleted.
 815          * @param array $existing
 816          * @return array
 817          */
 818         private function getInterwikiDeletions( $existing ) {
 819                 $del = [];
 820                 foreach ( $existing as $prefix => $dbkeys ) {
 821                         if ( isset( $this->mInterwikis[$prefix] ) ) {
 822                                 $del[$prefix] = array_diff_key( $existing[$prefix], $this->mInterwikis[$prefix] );
 823                         } else {
 824                                 $del[$prefix] = $existing[$prefix];
 825                         }
 826                 }
 827
 828                 return $del;
 829         }
 830
 831         /**
 832          * Get an array of existing links, as a 2-D array
 833          *
 834          * @return array
 835          */
 836         private function getExistingLinks() {
 837                 $res = $this->getDB()->select( 'pagelinks', [ 'pl_namespace', 'pl_title' ],
 838                         [ 'pl_from' => $this->mId ], __METHOD__ );
 839                 $arr = [];
 840                 foreach ( $res as $row ) {
 841                         if ( !isset( $arr[$row->pl_namespace] ) ) {
 842                                 $arr[$row->pl_namespace] = [];
 843                         }
 844                         $arr[$row->pl_namespace][$row->pl_title] = 1;
 845                 }
 846
 847                 return $arr;
 848         }
 849
 850         /**
 851          * Get an array of existing templates, as a 2-D array
 852          *
 853          * @return array
 854          */
 855         private function getExistingTemplates() {
 856                 $res = $this->getDB()->select( 'templatelinks', [ 'tl_namespace', 'tl_title' ],
 857                         [ 'tl_from' => $this->mId ], __METHOD__ );
 858                 $arr = [];
 859                 foreach ( $res as $row ) {
 860                         if ( !isset( $arr[$row->tl_namespace] ) ) {
 861                                 $arr[$row->tl_namespace] = [];
 862                         }
 863                         $arr[$row->tl_namespace][$row->tl_title] = 1;
 864                 }
 865
 866                 return $arr;
 867         }
 868
 869         /**
 870          * Get an array of existing images, image names in the keys
 871          *
 872          * @return array
 873          */
 874         private function getExistingImages() {
 875                 $res = $this->getDB()->select( 'imagelinks', [ 'il_to' ],
 876                         [ 'il_from' => $this->mId ], __METHOD__ );
 877                 $arr = [];
 878                 foreach ( $res as $row ) {
 879                         $arr[$row->il_to] = 1;
 880                 }
 881
 882                 return $arr;
 883         }
 884
 885         /**
 886          * Get an array of existing external links, URLs in the keys
 887          *
 888          * @return array
 889          */
 890         private function getExistingExternals() {
 891                 $res = $this->getDB()->select( 'externallinks', [ 'el_to' ],
 892                         [ 'el_from' => $this->mId ], __METHOD__ );
 893                 $arr = [];
 894                 foreach ( $res as $row ) {
 895                         $arr[$row->el_to] = 1;
 896                 }
 897
 898                 return $arr;
 899         }
 900
 901         /**
 902          * Get an array of existing categories, with the name in the key and sort key in the value.
 903          *
 904          * @return array
 905          */
 906         private function getExistingCategories() {
 907                 $res = $this->getDB()->select( 'categorylinks', [ 'cl_to', 'cl_sortkey_prefix' ],
 908                         [ 'cl_from' => $this->mId ], __METHOD__ );
 909                 $arr = [];
 910                 foreach ( $res as $row ) {
 911                         $arr[$row->cl_to] = $row->cl_sortkey_prefix;
 912                 }
 913
 914                 return $arr;
 915         }
 916
 917         /**
 918          * Get an array of existing interlanguage links, with the language code in the key and the
 919          * title in the value.
 920          *
 921          * @return array
 922          */
 923         private function getExistingInterlangs() {
 924                 $res = $this->getDB()->select( 'langlinks', [ 'll_lang', 'll_title' ],
 925                         [ 'll_from' => $this->mId ], __METHOD__ );
 926                 $arr = [];
 927                 foreach ( $res as $row ) {
 928                         $arr[$row->ll_lang] = $row->ll_title;
 929                 }
 930
 931                 return $arr;
 932         }
 933
 934         /**
 935          * Get an array of existing inline interwiki links, as a 2-D array
 936          * @return array (prefix => array(dbkey => 1))
 937          */
 938         private function getExistingInterwikis() {
 939                 $res = $this->getDB()->select( 'iwlinks', [ 'iwl_prefix', 'iwl_title' ],
 940                         [ 'iwl_from' => $this->mId ], __METHOD__ );
 941                 $arr = [];
 942                 foreach ( $res as $row ) {
 943                         if ( !isset( $arr[$row->iwl_prefix] ) ) {
 944                                 $arr[$row->iwl_prefix] = [];
 945                         }
 946                         $arr[$row->iwl_prefix][$row->iwl_title] = 1;
 947                 }
 948
 949                 return $arr;
 950         }
 951
 952         /**
 953          * Get an array of existing categories, with the name in the key and sort key in the value.
 954          *
 955          * @return array Array of property names and values
 956          */
 957         private function getExistingProperties() {
 958                 $res = $this->getDB()->select( 'page_props', [ 'pp_propname', 'pp_value' ],
 959                         [ 'pp_page' => $this->mId ], __METHOD__ );
 960                 $arr = [];
 961                 foreach ( $res as $row ) {
 962                         $arr[$row->pp_propname] = $row->pp_value;
 963                 }
 964
 965                 return $arr;
 966         }
 967
 968         /**
 969          * Return the title object of the page being updated
 970          * @return Title
 971          */
 972         public function getTitle() {
 973                 return $this->mTitle;
 974         }
 975
 976         /**
 977          * Returns parser output
 978          * @since 1.19
 979          * @return ParserOutput
 980          */
 981         public function getParserOutput() {
 982                 return $this->mParserOutput;
 983         }
 984
 985         /**
 986          * Return the list of images used as generated by the parser
 987          * @return array
 988          */
 989         public function getImages() {
 990                 return $this->mImages;
 991         }
 992
 993         /**
 994          * Set the revision corresponding to this LinksUpdate
 995          *
 996          * @since 1.27
 997          *
 998          * @param Revision $revision
 999          */
1000         public function setRevision( Revision $revision ) {
1001                 $this->mRevision = $revision;
1002         }
1003
1004         /**
1005          * @since 1.28
1006          * @return null|Revision
1007          */
1008         public function getRevision() {
1009                 return $this->mRevision;
1010         }
1011
1012         /**
1013          * Set the User who triggered this LinksUpdate
1014          *
1015          * @since 1.27
1016          * @param User $user
1017          */
1018         public function setTriggeringUser( User $user ) {
1019                 $this->user = $user;
1020         }
1021
1022         /**
1023          * @since 1.27
1024          * @return null|User
1025          */
1026         public function getTriggeringUser() {
1027                 return $this->user;
1028         }
1029
1030         /**
1031          * Invalidate any necessary link lists related to page property changes
1032          * @param array $changed
1033          */
1034         private function invalidateProperties( $changed ) {
1035                 global $wgPagePropLinkInvalidations;
1036
1037                 foreach ( $changed as $name => $value ) {
1038                         if ( isset( $wgPagePropLinkInvalidations[$name] ) ) {
1039                                 $inv = $wgPagePropLinkInvalidations[$name];
1040                                 if ( !is_array( $inv ) ) {
1041                                         $inv = [ $inv ];
1042                                 }
1043                                 foreach ( $inv as $table ) {
1044                                         DeferredUpdates::addUpdate( new HTMLCacheUpdate( $this->mTitle, $table ) );
1045                                 }
1046                         }
1047                 }
1048         }
1049
1050         /**
1051          * Fetch page links added by this LinksUpdate.  Only available after the update is complete.
1052          * @since 1.22
1053          * @return null|array Array of Titles
1054          */
1055         public function getAddedLinks() {
1056                 if ( $this->linkInsertions === null ) {
1057                         return null;
1058                 }
1059                 $result = [];
1060                 foreach ( $this->linkInsertions as $insertion ) {
1061                         $result[] = Title::makeTitle( $insertion['pl_namespace'], $insertion['pl_title'] );
1062                 }
1063
1064                 return $result;
1065         }
1066
1067         /**
1068          * Fetch page links removed by this LinksUpdate.  Only available after the update is complete.
1069          * @since 1.22
1070          * @return null|array Array of Titles
1071          */
1072         public function getRemovedLinks() {
1073                 if ( $this->linkDeletions === null ) {
1074                         return null;
1075                 }
1076                 $result = [];
1077                 foreach ( $this->linkDeletions as $ns => $titles ) {
1078                         foreach ( $titles as $title => $unused ) {
1079                                 $result[] = Title::makeTitle( $ns, $title );
1080                         }
1081                 }
1082
1083                 return $result;
1084         }
1085
1086         /**
1087          * Fetch page properties added by this LinksUpdate.
1088          * Only available after the update is complete.
1089          * @since 1.28
1090          * @return null|array
1091          */
1092         public function getAddedProperties() {
1093                 return $this->propertyInsertions;
1094         }
1095
1096         /**
1097          * Fetch page properties removed by this LinksUpdate.
1098          * Only available after the update is complete.
1099          * @since 1.28
1100          * @return null|array
1101          */
1102         public function getRemovedProperties() {
1103                 return $this->propertyDeletions;
1104         }
1105
1106         /**
1107          * Update links table freshness
1108          */
1109         private function updateLinksTimestamp() {
1110                 if ( $this->mId ) {
1111                         // The link updates made here only reflect the freshness of the parser output
1112                         $timestamp = $this->mParserOutput->getCacheTime();
1113                         $this->getDB()->update( 'page',
1114                                 [ 'page_links_updated' => $this->getDB()->timestamp( $timestamp ) ],
1115                                 [ 'page_id' => $this->mId ],
1116                                 __METHOD__
1117                         );
1118                 }
1119         }
1120
1121         /**
1122          * @return IDatabase
1123          */
1124         private function getDB() {
1125                 if ( !$this->db ) {
1126                         $this->db = wfGetDB( DB_MASTER );
1127                 }
1128
1129                 return $this->db;
1130         }
1131
1132         public function getAsJobSpecification() {
1133                 if ( $this->user ) {
1134                         $userInfo = [
1135                                 'userId' => $this->user->getId(),
1136                                 'userName' => $this->user->getName(),
1137                         ];
1138                 } else {
1139                         $userInfo = false;
1140                 }
1141
1142                 if ( $this->mRevision ) {
1143                         $triggeringRevisionId = $this->mRevision->getId();
1144                 } else {
1145                         $triggeringRevisionId = false;
1146                 }
1147
1148                 return [
1149                         'wiki' => $this->getDB()->getWikiID(),
1150                         'job'  => new JobSpecification(
1151                                 'refreshLinksPrioritized',
1152                                 [
1153                                         // Reuse the parser cache if it was saved
1154                                         'rootJobTimestamp' => $this->mParserOutput->getCacheTime(),
1155                                         'useRecursiveLinksUpdate' => $this->mRecursive,
1156                                         'triggeringUser' => $userInfo,
1157                                         'triggeringRevisionId' => $triggeringRevisionId,
1158                                 ],
1159                                 [ 'removeDuplicates' => true ],
1160                                 $this->getTitle()
1161                         )
1162                 ];
1163         }
1164 }