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