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