includes/deferred/LinksUpdate.php

   1 <?php
   2 /**
   3  * Updater for link tracking tables after a page edit.
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  */
  22
  23 use MediaWiki\MediaWikiServices;
  24 use Wikimedia\ScopedCallback;
  25
  26 /**
  27  * Class the manages updates of *_link tables as well as similar extension-managed tables
  28  *
  29  * @note: LinksUpdate is managed by DeferredUpdates::execute(). Do not run this in a transaction.
  30  *
  31  * See docs/deferred.txt
  32  */
  33 class LinksUpdate extends DataUpdate implements EnqueueableDataUpdate {
  34         // @todo make members protected, but make sure extensions don't break
  35
  36         /** @var int Page ID of the article linked from */
  37         public $mId;
  38
  39         /** @var Title Title object of the article linked from */
  40         public $mTitle;
  41
  42         /** @var ParserOutput */
  43         public $mParserOutput;
  44
  45         /** @var array Map of title strings to IDs for the links in the document */
  46         public $mLinks;
  47
  48         /** @var array DB keys of the images used, in the array key only */
  49         public $mImages;
  50
  51         /** @var array Map of title strings to IDs for the template references, including broken ones */
  52         public $mTemplates;
  53
  54         /** @var array URLs of external links, array key only */
  55         public $mExternals;
  56
  57         /** @var array Map of category names to sort keys */
  58         public $mCategories;
  59
  60         /** @var array Map of language codes to titles */
  61         public $mInterlangs;
  62
  63         /** @var array 2-D map of (prefix => DBK => 1) */
  64         public $mInterwikis;
  65
  66         /** @var array Map of arbitrary name to value */
  67         public $mProperties;
  68
  69         /** @var bool Whether to queue jobs for recursive updates */
  70         public $mRecursive;
  71
  72         /** @var Revision Revision for which this update has been triggered */
  73         private $mRevision;
  74
  75         /**
  76          * @var null|array Added links if calculated.
  77          */
  78         private $linkInsertions = null;
  79
  80         /**
  81          * @var null|array Deleted links if calculated.
  82          */
  83         private $linkDeletions = null;
  84
  85         /**
  86          * @var null|array Added properties if calculated.
  87          */
  88         private $propertyInsertions = null;
  89
  90         /**
  91          * @var null|array Deleted properties if calculated.
  92          */
  93         private $propertyDeletions = null;
  94
  95         /**
  96          * @var User|null
  97          */
  98         private $user;
  99
 100         /** @var IDatabase */
 101         private $db;
 102
 103         /**
 104          * Constructor
 105          *
 106          * @param Title $title Title of the page we're updating
 107          * @param ParserOutput $parserOutput Output from a full parse of this page
 108          * @param bool $recursive Queue jobs for recursive updates?
 109          * @throws MWException
 110          */
 111         function __construct( Title $title, ParserOutput $parserOutput, $recursive = true ) {
 112                 parent::__construct();
 113
 114                 $this->mTitle = $title;
 115                 $this->mId = $title->getArticleID( Title::GAID_FOR_UPDATE );
 116
 117                 if ( !$this->mId ) {
 118                         throw new InvalidArgumentException(
 119                                 "The Title object yields no ID. Perhaps the page doesn't exist?"
 120                         );
 121                 }
 122
 123                 $this->mParserOutput = $parserOutput;
 124
 125                 $this->mLinks = $parserOutput->getLinks();
 126                 $this->mImages = $parserOutput->getImages();
 127                 $this->mTemplates = $parserOutput->getTemplates();
 128                 $this->mExternals = $parserOutput->getExternalLinks();
 129                 $this->mCategories = $parserOutput->getCategories();
 130                 $this->mProperties = $parserOutput->getProperties();
 131                 $this->mInterwikis = $parserOutput->getInterwikiLinks();
 132
 133                 # Convert the format of the interlanguage links
 134                 # I didn't want to change it in the ParserOutput, because that array is passed all
 135                 # the way back to the skin, so either a skin API break would be required, or an
 136                 # inefficient back-conversion.
 137                 $ill = $parserOutput->getLanguageLinks();
 138                 $this->mInterlangs = [];
 139                 foreach ( $ill as $link ) {
 140                         list( $key, $title ) = explode( ':', $link, 2 );
 141                         $this->mInterlangs[$key] = $title;
 142                 }
 143
 144                 foreach ( $this->mCategories as &$sortkey ) {
 145                         # If the sortkey is longer then 255 bytes,
 146                         # it truncated by DB, and then doesn't get
 147                         # matched when comparing existing vs current
 148                         # categories, causing bug 25254.
 149                         # Also. substr behaves weird when given "".
 150                         if ( $sortkey !== '' ) {
 151                                 $sortkey = substr( $sortkey, 0, 255 );
 152                         }
 153                 }
 154
 155                 $this->mRecursive = $recursive;
 156
 157                 Hooks::run( 'LinksUpdateConstructed', [ &$this ] );
 158         }
 159
 160         /**
 161          * Update link tables with outgoing links from an updated article
 162          *
 163          * @note: this is managed by DeferredUpdates::execute(). Do not run this in a transaction.
 164          */
 165         public function doUpdate() {
 166                 if ( $this->ticket ) {
 167                         // Make sure all links update threads see the changes of each other.
 168                         // This handles the case when updates have to batched into several COMMITs.
 169                         $scopedLock = self::acquirePageLock( $this->getDB(), $this->mId );
 170                 }
 171
 172                 Hooks::run( 'LinksUpdate', [ &$this ] );
 173                 $this->doIncrementalUpdate();
 174
 175                 // Commit and release the lock (if set)
 176                 ScopedCallback::consume( $scopedLock );
 177                 // Run post-commit hooks without DBO_TRX
 178                 $this->getDB()->onTransactionIdle(
 179                         function () {
 180                                 Hooks::run( 'LinksUpdateComplete', [ &$this, $this->ticket ] );
 181                         },
 182                         __METHOD__
 183                 );
 184         }
 185
 186         /**
 187          * Acquire a lock for performing link table updates for a page on a DB
 188          *
 189          * @param IDatabase $dbw
 190          * @param integer $pageId
 191          * @param string $why One of (job, atomicity)
 192          * @return ScopedCallback
 193          * @throws RuntimeException
 194          * @since 1.27
 195          */
 196         public static function acquirePageLock( IDatabase $dbw, $pageId, $why = 'atomicity' ) {
 197                 $key = "LinksUpdate:$why:pageid:$pageId";
 198                 $scopedLock = $dbw->getScopedLockAndFlush( $key, __METHOD__, 15 );
 199                 if ( !$scopedLock ) {
 200                         throw new RuntimeException( "Could not acquire lock '$key'." );
 201                 }
 202
 203                 return $scopedLock;
 204         }
 205
 206         protected function doIncrementalUpdate() {
 207                 # Page links
 208                 $existing = $this->getExistingLinks();
 209                 $this->linkDeletions = $this->getLinkDeletions( $existing );
 210                 $this->linkInsertions = $this->getLinkInsertions( $existing );
 211                 $this->incrTableUpdate( 'pagelinks', 'pl', $this->linkDeletions, $this->linkInsertions );
 212
 213                 # Image links
 214                 $existing = $this->getExistingImages();
 215                 $imageDeletes = $this->getImageDeletions( $existing );
 216                 $this->incrTableUpdate( 'imagelinks', 'il', $imageDeletes,
 217                         $this->getImageInsertions( $existing ) );
 218
 219                 # Invalidate all image description pages which had links added or removed
 220                 $imageUpdates = $imageDeletes + array_diff_key( $this->mImages, $existing );
 221                 $this->invalidateImageDescriptions( $imageUpdates );
 222
 223                 # External links
 224                 $existing = $this->getExistingExternals();
 225                 $this->incrTableUpdate( 'externallinks', 'el', $this->getExternalDeletions( $existing ),
 226                         $this->getExternalInsertions( $existing ) );
 227
 228                 # Language links
 229                 $existing = $this->getExistingInterlangs();
 230                 $this->incrTableUpdate( 'langlinks', 'll', $this->getInterlangDeletions( $existing ),
 231                         $this->getInterlangInsertions( $existing ) );
 232
 233                 # Inline interwiki links
 234                 $existing = $this->getExistingInterwikis();
 235                 $this->incrTableUpdate( 'iwlinks', 'iwl', $this->getInterwikiDeletions( $existing ),
 236                         $this->getInterwikiInsertions( $existing ) );
 237
 238                 # Template links
 239                 $existing = $this->getExistingTemplates();
 240                 $this->incrTableUpdate( 'templatelinks', 'tl', $this->getTemplateDeletions( $existing ),
 241                         $this->getTemplateInsertions( $existing ) );
 242
 243                 # Category links
 244                 $existing = $this->getExistingCategories();
 245                 $categoryDeletes = $this->getCategoryDeletions( $existing );
 246                 $this->incrTableUpdate( 'categorylinks', 'cl', $categoryDeletes,
 247                         $this->getCategoryInsertions( $existing ) );
 248
 249                 # Invalidate all categories which were added, deleted or changed (set symmetric difference)
 250                 $categoryInserts = array_diff_assoc( $this->mCategories, $existing );
 251                 $categoryUpdates = $categoryInserts + $categoryDeletes;
 252                 $this->invalidateCategories( $categoryUpdates );
 253                 $this->updateCategoryCounts( $categoryInserts, $categoryDeletes );
 254
 255                 # Page properties
 256                 $existing = $this->getExistingProperties();
 257                 $this->propertyDeletions = $this->getPropertyDeletions( $existing );
 258                 $this->incrTableUpdate( 'page_props', 'pp', $this->propertyDeletions,
 259                         $this->getPropertyInsertions( $existing ) );
 260
 261                 # Invalidate the necessary pages
 262                 $this->propertyInsertions = array_diff_assoc( $this->mProperties, $existing );
 263                 $changed = $this->propertyDeletions + $this->propertyInsertions;
 264                 $this->invalidateProperties( $changed );
 265
 266                 # Refresh links of all pages including this page
 267                 # This will be in a separate transaction
 268                 if ( $this->mRecursive ) {
 269                         $this->queueRecursiveJobs();
 270                 }
 271
 272                 # Update the links table freshness for this title
 273                 $this->updateLinksTimestamp();
 274         }
 275
 276         /**
 277          * Queue recursive jobs for this page
 278          *
 279          * Which means do LinksUpdate on all pages that include the current page,
 280          * using the job queue.
 281          */
 282         protected function queueRecursiveJobs() {
 283                 self::queueRecursiveJobsForTable( $this->mTitle, 'templatelinks' );
 284                 if ( $this->mTitle->getNamespace() == NS_FILE ) {
 285                         // Process imagelinks in case the title is or was a redirect
 286                         self::queueRecursiveJobsForTable( $this->mTitle, 'imagelinks' );
 287                 }
 288
 289                 $bc = $this->mTitle->getBacklinkCache();
 290                 // Get jobs for cascade-protected backlinks for a high priority queue.
 291                 // If meta-templates change to using a new template, the new template
 292                 // should be implicitly protected as soon as possible, if applicable.
 293                 // These jobs duplicate a subset of the above ones, but can run sooner.
 294                 // Which ever runs first generally no-ops the other one.
 295                 $jobs = [];
 296                 foreach ( $bc->getCascadeProtectedLinks() as $title ) {
 297                         $jobs[] = RefreshLinksJob::newPrioritized( $title, [] );
 298                 }
 299                 JobQueueGroup::singleton()->push( $jobs );
 300         }
 301
 302         /**
 303          * Queue a RefreshLinks job for any table.
 304          *
 305          * @param Title $title Title to do job for
 306          * @param string $table Table to use (e.g. 'templatelinks')
 307          */
 308         public static function queueRecursiveJobsForTable( Title $title, $table ) {
 309                 if ( $title->getBacklinkCache()->hasLinks( $table ) ) {
 310                         $job = new RefreshLinksJob(
 311                                 $title,
 312                                 [
 313                                         'table' => $table,
 314                                         'recursive' => true,
 315                                 ] + Job::newRootJobParams( // "overall" refresh links job info
 316                                         "refreshlinks:{$table}:{$title->getPrefixedText()}"
 317                                 )
 318                         );
 319
 320                         JobQueueGroup::singleton()->push( $job );
 321                 }
 322         }
 323
 324         /**
 325          * @param array $cats
 326          */
 327         function invalidateCategories( $cats ) {
 328                 PurgeJobUtils::invalidatePages( $this->getDB(), NS_CATEGORY, array_keys( $cats ) );
 329         }
 330
 331         /**
 332          * Update all the appropriate counts in the category table.
 333          * @param array $added Associative array of category name => sort key
 334          * @param array $deleted Associative array of category name => sort key
 335          */
 336         function updateCategoryCounts( $added, $deleted ) {
 337                 $a = WikiPage::factory( $this->mTitle );
 338                 $a->updateCategoryCounts(
 339                         array_keys( $added ), array_keys( $deleted )
 340                 );
 341         }
 342
 343         /**
 344          * @param array $images
 345          */
 346         function invalidateImageDescriptions( $images ) {
 347                 PurgeJobUtils::invalidatePages( $this->getDB(), NS_FILE, array_keys( $images ) );
 348         }
 349
 350         /**
 351          * Update a table by doing a delete query then an insert query
 352          * @param string $table Table name
 353          * @param string $prefix Field name prefix
 354          * @param array $deletions
 355          * @param array $insertions Rows to insert
 356          */
 357         private function incrTableUpdate( $table, $prefix, $deletions, $insertions ) {
 358                 $services = MediaWikiServices::getInstance();
 359                 $bSize = $services->getMainConfig()->get( 'UpdateRowsPerQuery' );
 360                 $factory = $services->getDBLoadBalancerFactory();
 361
 362                 if ( $table === 'page_props' ) {
 363                         $fromField = 'pp_page';
 364                 } else {
 365                         $fromField = "{$prefix}_from";
 366                 }
 367
 368                 $deleteWheres = []; // list of WHERE clause arrays for each DB delete() call
 369                 if ( $table === 'pagelinks' || $table === 'templatelinks' || $table === 'iwlinks' ) {
 370                         $baseKey =  ( $table === 'iwlinks' ) ? 'iwl_prefix' : "{$prefix}_namespace";
 371
 372                         $curBatchSize = 0;
 373                         $curDeletionBatch = [];
 374                         $deletionBatches = [];
 375                         foreach ( $deletions as $ns => $dbKeys ) {
 376                                 foreach ( $dbKeys as $dbKey => $unused ) {
 377                                         $curDeletionBatch[$ns][$dbKey] = 1;
 378                                         if ( ++$curBatchSize >= $bSize ) {
 379                                                 $deletionBatches[] = $curDeletionBatch;
 380                                                 $curDeletionBatch = [];
 381                                                 $curBatchSize = 0;
 382                                         }
 383                                 }
 384                         }
 385                         if ( $curDeletionBatch ) {
 386                                 $deletionBatches[] = $curDeletionBatch;
 387                         }
 388
 389                         foreach ( $deletionBatches as $deletionBatch ) {
 390                                 $deleteWheres[] = [
 391                                         $fromField => $this->mId,
 392                                         $this->getDB()->makeWhereFrom2d( $deletionBatch, $baseKey, "{$prefix}_title" )
 393                                 ];
 394                         }
 395                 } else {
 396                         if ( $table === 'langlinks' ) {
 397                                 $toField = 'll_lang';
 398                         } elseif ( $table === 'page_props' ) {
 399                                 $toField = 'pp_propname';
 400                         } else {
 401                                 $toField = $prefix . '_to';
 402                         }
 403
 404                         $deletionBatches = array_chunk( array_keys( $deletions ), $bSize );
 405                         foreach ( $deletionBatches as $deletionBatch ) {
 406                                 $deleteWheres[] = [ $fromField => $this->mId, $toField => $deletionBatch ];
 407                         }
 408                 }
 409
 410                 foreach ( $deleteWheres as $deleteWhere ) {
 411                         $this->getDB()->delete( $table, $deleteWhere, __METHOD__ );
 412                         $factory->commitAndWaitForReplication(
 413                                 __METHOD__, $this->ticket, [ 'wiki' => $this->getDB()->getWikiID() ]
 414                         );
 415                 }
 416
 417                 $insertBatches = array_chunk( $insertions, $bSize );
 418                 foreach ( $insertBatches as $insertBatch ) {
 419                         $this->getDB()->insert( $table, $insertBatch, __METHOD__, 'IGNORE' );
 420                         $factory->commitAndWaitForReplication(
 421                                 __METHOD__, $this->ticket, [ 'wiki' => $this->getDB()->getWikiID() ]
 422                         );
 423                 }
 424
 425                 if ( count( $insertions ) ) {
 426                         Hooks::run( 'LinksUpdateAfterInsert', [ $this, $table, $insertions ] );
 427                 }
 428         }
 429
 430         /**
 431          * Get an array of pagelinks insertions for passing to the DB
 432          * Skips the titles specified by the 2-D array $existing
 433          * @param array $existing
 434          * @return array
 435          */
 436         private function getLinkInsertions( $existing = [] ) {
 437                 $arr = [];
 438                 foreach ( $this->mLinks as $ns => $dbkeys ) {
 439                         $diffs = isset( $existing[$ns] )
 440                                 ? array_diff_key( $dbkeys, $existing[$ns] )
 441                                 : $dbkeys;
 442                         foreach ( $diffs as $dbk => $id ) {
 443                                 $arr[] = [
 444                                         'pl_from' => $this->mId,
 445                                         'pl_from_namespace' => $this->mTitle->getNamespace(),
 446                                         'pl_namespace' => $ns,
 447                                         'pl_title' => $dbk
 448                                 ];
 449                         }
 450                 }
 451
 452                 return $arr;
 453         }
 454
 455         /**
 456          * Get an array of template insertions. Like getLinkInsertions()
 457          * @param array $existing
 458          * @return array
 459          */
 460         private function getTemplateInsertions( $existing = [] ) {
 461                 $arr = [];
 462                 foreach ( $this->mTemplates as $ns => $dbkeys ) {
 463                         $diffs = isset( $existing[$ns] ) ? array_diff_key( $dbkeys, $existing[$ns] ) : $dbkeys;
 464                         foreach ( $diffs as $dbk => $id ) {
 465                                 $arr[] = [
 466                                         'tl_from' => $this->mId,
 467                                         'tl_from_namespace' => $this->mTitle->getNamespace(),
 468                                         'tl_namespace' => $ns,
 469                                         'tl_title' => $dbk
 470                                 ];
 471                         }
 472                 }
 473
 474                 return $arr;
 475         }
 476
 477         /**
 478          * Get an array of image insertions
 479          * Skips the names specified in $existing
 480          * @param array $existing
 481          * @return array
 482          */
 483         private function getImageInsertions( $existing = [] ) {
 484                 $arr = [];
 485                 $diffs = array_diff_key( $this->mImages, $existing );
 486                 foreach ( $diffs as $iname => $dummy ) {
 487                         $arr[] = [
 488                                 'il_from' => $this->mId,
 489                                 'il_from_namespace' => $this->mTitle->getNamespace(),
 490                                 'il_to' => $iname
 491                         ];
 492                 }
 493
 494                 return $arr;
 495         }
 496
 497         /**
 498          * Get an array of externallinks insertions. Skips the names specified in $existing
 499          * @param array $existing
 500          * @return array
 501          */
 502         private function getExternalInsertions( $existing = [] ) {
 503                 $arr = [];
 504                 $diffs = array_diff_key( $this->mExternals, $existing );
 505                 foreach ( $diffs as $url => $dummy ) {
 506                         foreach ( wfMakeUrlIndexes( $url ) as $index ) {
 507                                 $arr[] = [
 508                                         'el_id' => $this->getDB()->nextSequenceValue( 'externallinks_el_id_seq' ),
 509                                         'el_from' => $this->mId,
 510                                         'el_to' => $url,
 511                                         'el_index' => $index,
 512                                 ];
 513                         }
 514                 }
 515
 516                 return $arr;
 517         }
 518
 519         /**
 520          * Get an array of category insertions
 521          *
 522          * @param array $existing Mapping existing category names to sort keys. If both
 523          * match a link in $this, the link will be omitted from the output
 524          *
 525          * @return array
 526          */
 527         private function getCategoryInsertions( $existing = [] ) {
 528                 global $wgContLang, $wgCategoryCollation;
 529                 $diffs = array_diff_assoc( $this->mCategories, $existing );
 530                 $arr = [];
 531                 foreach ( $diffs as $name => $prefix ) {
 532                         $nt = Title::makeTitleSafe( NS_CATEGORY, $name );
 533                         $wgContLang->findVariantLink( $name, $nt, true );
 534
 535                         if ( $this->mTitle->getNamespace() == NS_CATEGORY ) {
 536                                 $type = 'subcat';
 537                         } elseif ( $this->mTitle->getNamespace() == NS_FILE ) {
 538                                 $type = 'file';
 539                         } else {
 540                                 $type = 'page';
 541                         }
 542
 543                         # Treat custom sortkeys as a prefix, so that if multiple
 544                         # things are forced to sort as '*' or something, they'll
 545                         # sort properly in the category rather than in page_id
 546                         # order or such.
 547                         $sortkey = Collation::singleton()->getSortKey(
 548                                 $this->mTitle->getCategorySortkey( $prefix ) );
 549
 550                         $arr[] = [
 551                                 'cl_from' => $this->mId,
 552                                 'cl_to' => $name,
 553                                 'cl_sortkey' => $sortkey,
 554                                 'cl_timestamp' => $this->getDB()->timestamp(),
 555                                 'cl_sortkey_prefix' => $prefix,
 556                                 'cl_collation' => $wgCategoryCollation,
 557                                 'cl_type' => $type,
 558                         ];
 559                 }
 560
 561                 return $arr;
 562         }
 563
 564         /**
 565          * Get an array of interlanguage link insertions
 566          *
 567          * @param array $existing Mapping existing language codes to titles
 568          *
 569          * @return array
 570          */
 571         private function getInterlangInsertions( $existing = [] ) {
 572                 $diffs = array_diff_assoc( $this->mInterlangs, $existing );
 573                 $arr = [];
 574                 foreach ( $diffs as $lang => $title ) {
 575                         $arr[] = [
 576                                 'll_from' => $this->mId,
 577                                 'll_lang' => $lang,
 578                                 'll_title' => $title
 579                         ];
 580                 }
 581
 582                 return $arr;
 583         }
 584
 585         /**
 586          * Get an array of page property insertions
 587          * @param array $existing
 588          * @return array
 589          */
 590         function getPropertyInsertions( $existing = [] ) {
 591                 $diffs = array_diff_assoc( $this->mProperties, $existing );
 592
 593                 $arr = [];
 594                 foreach ( array_keys( $diffs ) as $name ) {
 595                         $arr[] = $this->getPagePropRowData( $name );
 596                 }
 597
 598                 return $arr;
 599         }
 600
 601         /**
 602          * Returns an associative array to be used for inserting a row into
 603          * the page_props table. Besides the given property name, this will
 604          * include the page id from $this->mId and any property value from
 605          * $this->mProperties.
 606          *
 607          * The array returned will include the pp_sortkey field if this
 608          * is present in the database (as indicated by $wgPagePropsHaveSortkey).
 609          * The sortkey value is currently determined by getPropertySortKeyValue().
 610          *
 611          * @note this assumes that $this->mProperties[$prop] is defined.
 612          *
 613          * @param string $prop The name of the property.
 614          *
 615          * @return array
 616          */
 617         private function getPagePropRowData( $prop ) {
 618                 global $wgPagePropsHaveSortkey;
 619
 620                 $value = $this->mProperties[$prop];
 621
 622                 $row = [
 623                         'pp_page' => $this->mId,
 624                         'pp_propname' => $prop,
 625                         'pp_value' => $value,
 626                 ];
 627
 628                 if ( $wgPagePropsHaveSortkey ) {
 629                         $row['pp_sortkey'] = $this->getPropertySortKeyValue( $value );
 630                 }
 631
 632                 return $row;
 633         }
 634
 635         /**
 636          * Determines the sort key for the given property value.
 637          * This will return $value if it is a float or int,
 638          * 1 or resp. 0 if it is a bool, and null otherwise.
 639          *
 640          * @note In the future, we may allow the sortkey to be specified explicitly
 641          *       in ParserOutput::setProperty.
 642          *
 643          * @param mixed $value
 644          *
 645          * @return float|null
 646          */
 647         private function getPropertySortKeyValue( $value ) {
 648                 if ( is_int( $value ) || is_float( $value ) || is_bool( $value ) ) {
 649                         return floatval( $value );
 650                 }
 651
 652                 return null;
 653         }
 654
 655         /**
 656          * Get an array of interwiki insertions for passing to the DB
 657          * Skips the titles specified by the 2-D array $existing
 658          * @param array $existing
 659          * @return array
 660          */
 661         private function getInterwikiInsertions( $existing = [] ) {
 662                 $arr = [];
 663                 foreach ( $this->mInterwikis as $prefix => $dbkeys ) {
 664                         $diffs = isset( $existing[$prefix] )
 665                                 ? array_diff_key( $dbkeys, $existing[$prefix] )
 666                                 : $dbkeys;
 667
 668                         foreach ( $diffs as $dbk => $id ) {
 669                                 $arr[] = [
 670                                         'iwl_from' => $this->mId,
 671                                         'iwl_prefix' => $prefix,
 672                                         'iwl_title' => $dbk
 673                                 ];
 674                         }
 675                 }
 676
 677                 return $arr;
 678         }
 679
 680         /**
 681          * Given an array of existing links, returns those links which are not in $this
 682          * and thus should be deleted.
 683          * @param array $existing
 684          * @return array
 685          */
 686         private function getLinkDeletions( $existing ) {
 687                 $del = [];
 688                 foreach ( $existing as $ns => $dbkeys ) {
 689                         if ( isset( $this->mLinks[$ns] ) ) {
 690                                 $del[$ns] = array_diff_key( $existing[$ns], $this->mLinks[$ns] );
 691                         } else {
 692                                 $del[$ns] = $existing[$ns];
 693                         }
 694                 }
 695
 696                 return $del;
 697         }
 698
 699         /**
 700          * Given an array of existing templates, returns those templates which are not in $this
 701          * and thus should be deleted.
 702          * @param array $existing
 703          * @return array
 704          */
 705         private function getTemplateDeletions( $existing ) {
 706                 $del = [];
 707                 foreach ( $existing as $ns => $dbkeys ) {
 708                         if ( isset( $this->mTemplates[$ns] ) ) {
 709                                 $del[$ns] = array_diff_key( $existing[$ns], $this->mTemplates[$ns] );
 710                         } else {
 711                                 $del[$ns] = $existing[$ns];
 712                         }
 713                 }
 714
 715                 return $del;
 716         }
 717
 718         /**
 719          * Given an array of existing images, returns those images which are not in $this
 720          * and thus should be deleted.
 721          * @param array $existing
 722          * @return array
 723          */
 724         private function getImageDeletions( $existing ) {
 725                 return array_diff_key( $existing, $this->mImages );
 726         }
 727
 728         /**
 729          * Given an array of existing external links, returns those links which are not
 730          * in $this and thus should be deleted.
 731          * @param array $existing
 732          * @return array
 733          */
 734         private function getExternalDeletions( $existing ) {
 735                 return array_diff_key( $existing, $this->mExternals );
 736         }
 737
 738         /**
 739          * Given an array of existing categories, returns those categories which are not in $this
 740          * and thus should be deleted.
 741          * @param array $existing
 742          * @return array
 743          */
 744         private function getCategoryDeletions( $existing ) {
 745                 return array_diff_assoc( $existing, $this->mCategories );
 746         }
 747
 748         /**
 749          * Given an array of existing interlanguage links, returns those links which are not
 750          * in $this and thus should be deleted.
 751          * @param array $existing
 752          * @return array
 753          */
 754         private function getInterlangDeletions( $existing ) {
 755                 return array_diff_assoc( $existing, $this->mInterlangs );
 756         }
 757
 758         /**
 759          * Get array of properties which should be deleted.
 760          * @param array $existing
 761          * @return array
 762          */
 763         function getPropertyDeletions( $existing ) {
 764                 return array_diff_assoc( $existing, $this->mProperties );
 765         }
 766
 767         /**
 768          * Given an array of existing interwiki links, returns those links which are not in $this
 769          * and thus should be deleted.
 770          * @param array $existing
 771          * @return array
 772          */
 773         private function getInterwikiDeletions( $existing ) {
 774                 $del = [];
 775                 foreach ( $existing as $prefix => $dbkeys ) {
 776                         if ( isset( $this->mInterwikis[$prefix] ) ) {
 777                                 $del[$prefix] = array_diff_key( $existing[$prefix], $this->mInterwikis[$prefix] );
 778                         } else {
 779                                 $del[$prefix] = $existing[$prefix];
 780                         }
 781                 }
 782
 783                 return $del;
 784         }
 785
 786         /**
 787          * Get an array of existing links, as a 2-D array
 788          *
 789          * @return array
 790          */
 791         private function getExistingLinks() {
 792                 $res = $this->getDB()->select( 'pagelinks', [ 'pl_namespace', 'pl_title' ],
 793                         [ 'pl_from' => $this->mId ], __METHOD__ );
 794                 $arr = [];
 795                 foreach ( $res as $row ) {
 796                         if ( !isset( $arr[$row->pl_namespace] ) ) {
 797                                 $arr[$row->pl_namespace] = [];
 798                         }
 799                         $arr[$row->pl_namespace][$row->pl_title] = 1;
 800                 }
 801
 802                 return $arr;
 803         }
 804
 805         /**
 806          * Get an array of existing templates, as a 2-D array
 807          *
 808          * @return array
 809          */
 810         private function getExistingTemplates() {
 811                 $res = $this->getDB()->select( 'templatelinks', [ 'tl_namespace', 'tl_title' ],
 812                         [ 'tl_from' => $this->mId ], __METHOD__ );
 813                 $arr = [];
 814                 foreach ( $res as $row ) {
 815                         if ( !isset( $arr[$row->tl_namespace] ) ) {
 816                                 $arr[$row->tl_namespace] = [];
 817                         }
 818                         $arr[$row->tl_namespace][$row->tl_title] = 1;
 819                 }
 820
 821                 return $arr;
 822         }
 823
 824         /**
 825          * Get an array of existing images, image names in the keys
 826          *
 827          * @return array
 828          */
 829         private function getExistingImages() {
 830                 $res = $this->getDB()->select( 'imagelinks', [ 'il_to' ],
 831                         [ 'il_from' => $this->mId ], __METHOD__ );
 832                 $arr = [];
 833                 foreach ( $res as $row ) {
 834                         $arr[$row->il_to] = 1;
 835                 }
 836
 837                 return $arr;
 838         }
 839
 840         /**
 841          * Get an array of existing external links, URLs in the keys
 842          *
 843          * @return array
 844          */
 845         private function getExistingExternals() {
 846                 $res = $this->getDB()->select( 'externallinks', [ 'el_to' ],
 847                         [ 'el_from' => $this->mId ], __METHOD__ );
 848                 $arr = [];
 849                 foreach ( $res as $row ) {
 850                         $arr[$row->el_to] = 1;
 851                 }
 852
 853                 return $arr;
 854         }
 855
 856         /**
 857          * Get an array of existing categories, with the name in the key and sort key in the value.
 858          *
 859          * @return array
 860          */
 861         private function getExistingCategories() {
 862                 $res = $this->getDB()->select( 'categorylinks', [ 'cl_to', 'cl_sortkey_prefix' ],
 863                         [ 'cl_from' => $this->mId ], __METHOD__ );
 864                 $arr = [];
 865                 foreach ( $res as $row ) {
 866                         $arr[$row->cl_to] = $row->cl_sortkey_prefix;
 867                 }
 868
 869                 return $arr;
 870         }
 871
 872         /**
 873          * Get an array of existing interlanguage links, with the language code in the key and the
 874          * title in the value.
 875          *
 876          * @return array
 877          */
 878         private function getExistingInterlangs() {
 879                 $res = $this->getDB()->select( 'langlinks', [ 'll_lang', 'll_title' ],
 880                         [ 'll_from' => $this->mId ], __METHOD__ );
 881                 $arr = [];
 882                 foreach ( $res as $row ) {
 883                         $arr[$row->ll_lang] = $row->ll_title;
 884                 }
 885
 886                 return $arr;
 887         }
 888
 889         /**
 890          * Get an array of existing inline interwiki links, as a 2-D array
 891          * @return array (prefix => array(dbkey => 1))
 892          */
 893         private function getExistingInterwikis() {
 894                 $res = $this->getDB()->select( 'iwlinks', [ 'iwl_prefix', 'iwl_title' ],
 895                         [ 'iwl_from' => $this->mId ], __METHOD__ );
 896                 $arr = [];
 897                 foreach ( $res as $row ) {
 898                         if ( !isset( $arr[$row->iwl_prefix] ) ) {
 899                                 $arr[$row->iwl_prefix] = [];
 900                         }
 901                         $arr[$row->iwl_prefix][$row->iwl_title] = 1;
 902                 }
 903
 904                 return $arr;
 905         }
 906
 907         /**
 908          * Get an array of existing categories, with the name in the key and sort key in the value.
 909          *
 910          * @return array Array of property names and values
 911          */
 912         private function getExistingProperties() {
 913                 $res = $this->getDB()->select( 'page_props', [ 'pp_propname', 'pp_value' ],
 914                         [ 'pp_page' => $this->mId ], __METHOD__ );
 915                 $arr = [];
 916                 foreach ( $res as $row ) {
 917                         $arr[$row->pp_propname] = $row->pp_value;
 918                 }
 919
 920                 return $arr;
 921         }
 922
 923         /**
 924          * Return the title object of the page being updated
 925          * @return Title
 926          */
 927         public function getTitle() {
 928                 return $this->mTitle;
 929         }
 930
 931         /**
 932          * Returns parser output
 933          * @since 1.19
 934          * @return ParserOutput
 935          */
 936         public function getParserOutput() {
 937                 return $this->mParserOutput;
 938         }
 939
 940         /**
 941          * Return the list of images used as generated by the parser
 942          * @return array
 943          */
 944         public function getImages() {
 945                 return $this->mImages;
 946         }
 947
 948         /**
 949          * Set the revision corresponding to this LinksUpdate
 950          *
 951          * @since 1.27
 952          *
 953          * @param Revision $revision
 954          */
 955         public function setRevision( Revision $revision ) {
 956                 $this->mRevision = $revision;
 957         }
 958
 959         /**
 960          * @since 1.28
 961          * @return null|Revision
 962          */
 963         public function getRevision() {
 964                 return $this->mRevision;
 965         }
 966
 967         /**
 968          * Set the User who triggered this LinksUpdate
 969          *
 970          * @since 1.27
 971          * @param User $user
 972          */
 973         public function setTriggeringUser( User $user ) {
 974                 $this->user = $user;
 975         }
 976
 977         /**
 978          * @since 1.27
 979          * @return null|User
 980          */
 981         public function getTriggeringUser() {
 982                 return $this->user;
 983         }
 984
 985         /**
 986          * Invalidate any necessary link lists related to page property changes
 987          * @param array $changed
 988          */
 989         private function invalidateProperties( $changed ) {
 990                 global $wgPagePropLinkInvalidations;
 991
 992                 foreach ( $changed as $name => $value ) {
 993                         if ( isset( $wgPagePropLinkInvalidations[$name] ) ) {
 994                                 $inv = $wgPagePropLinkInvalidations[$name];
 995                                 if ( !is_array( $inv ) ) {
 996                                         $inv = [ $inv ];
 997                                 }
 998                                 foreach ( $inv as $table ) {
 999                                         DeferredUpdates::addUpdate( new HTMLCacheUpdate( $this->mTitle, $table ) );
1000                                 }
1001                         }
1002                 }
1003         }
1004
1005         /**
1006          * Fetch page links added by this LinksUpdate.  Only available after the update is complete.
1007          * @since 1.22
1008          * @return null|array Array of Titles
1009          */
1010         public function getAddedLinks() {
1011                 if ( $this->linkInsertions === null ) {
1012                         return null;
1013                 }
1014                 $result = [];
1015                 foreach ( $this->linkInsertions as $insertion ) {
1016                         $result[] = Title::makeTitle( $insertion['pl_namespace'], $insertion['pl_title'] );
1017                 }
1018
1019                 return $result;
1020         }
1021
1022         /**
1023          * Fetch page links removed by this LinksUpdate.  Only available after the update is complete.
1024          * @since 1.22
1025          * @return null|array Array of Titles
1026          */
1027         public function getRemovedLinks() {
1028                 if ( $this->linkDeletions === null ) {
1029                         return null;
1030                 }
1031                 $result = [];
1032                 foreach ( $this->linkDeletions as $ns => $titles ) {
1033                         foreach ( $titles as $title => $unused ) {
1034                                 $result[] = Title::makeTitle( $ns, $title );
1035                         }
1036                 }
1037
1038                 return $result;
1039         }
1040
1041         /**
1042          * Fetch page properties added by this LinksUpdate.
1043          * Only available after the update is complete.
1044          * @since 1.28
1045          * @return null|array
1046          */
1047         public function getAddedProperties() {
1048                 return $this->propertyInsertions;
1049         }
1050
1051         /**
1052          * Fetch page properties removed by this LinksUpdate.
1053          * Only available after the update is complete.
1054          * @since 1.28
1055          * @return null|array
1056          */
1057         public function getRemovedProperties() {
1058                 return $this->propertyDeletions;
1059         }
1060
1061         /**
1062          * Update links table freshness
1063          */
1064         private function updateLinksTimestamp() {
1065                 if ( $this->mId ) {
1066                         // The link updates made here only reflect the freshness of the parser output
1067                         $timestamp = $this->mParserOutput->getCacheTime();
1068                         $this->getDB()->update( 'page',
1069                                 [ 'page_links_updated' => $this->getDB()->timestamp( $timestamp ) ],
1070                                 [ 'page_id' => $this->mId ],
1071                                 __METHOD__
1072                         );
1073                 }
1074         }
1075
1076         /**
1077          * @return IDatabase
1078          */
1079         private function getDB() {
1080                 if ( !$this->db ) {
1081                         $this->db = wfGetDB( DB_MASTER );
1082                 }
1083
1084                 return $this->db;
1085         }
1086
1087         public function getAsJobSpecification() {
1088                 if ( $this->user ) {
1089                         $userInfo = [
1090                                 'userId' => $this->user->getId(),
1091                                 'userName' => $this->user->getName(),
1092                         ];
1093                 } else {
1094                         $userInfo = false;
1095                 }
1096
1097                 if ( $this->mRevision ) {
1098                         $triggeringRevisionId = $this->mRevision->getId();
1099                 } else {
1100                         $triggeringRevisionId = false;
1101                 }
1102
1103                 return [
1104                         'wiki' => $this->getDB()->getWikiID(),
1105                         'job'  => new JobSpecification(
1106                                 'refreshLinksPrioritized',
1107                                 [
1108                                         // Reuse the parser cache if it was saved
1109                                         'rootJobTimestamp' => $this->mParserOutput->getCacheTime(),
1110                                         'useRecursiveLinksUpdate' => $this->mRecursive,
1111                                         'triggeringUser' => $userInfo,
1112                                         'triggeringRevisionId' => $triggeringRevisionId,
1113                                 ],
1114                                 [ 'removeDuplicates' => true ],
1115                                 $this->getTitle()
1116                         )
1117                 ];
1118         }
1119 }