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                 PurgeJobUtils::invalidatePages( $this->mDb, 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                 PurgeJobUtils::invalidatePages( $this->mDb, 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->commitAndWaitForReplication(
 391                                 __METHOD__, $this->ticket, [ 'wiki' => $this->mDb->getWikiID() ]
 392                         );
 393                 }
 394
 395                 $insertBatches = array_chunk( $insertions, $bSize );
 396                 foreach ( $insertBatches as $insertBatch ) {
 397                         $this->mDb->insert( $table, $insertBatch, __METHOD__, 'IGNORE' );
 398                         $factory->commitAndWaitForReplication(
 399                                 __METHOD__, $this->ticket, [ 'wiki' => $this->mDb->getWikiID() ]
 400                         );
 401                 }
 402
 403                 if ( count( $insertions ) ) {
 404                         Hooks::run( 'LinksUpdateAfterInsert', [ $this, $table, $insertions ] );
 405                 }
 406         }
 407
 408         /**
 409          * Get an array of pagelinks insertions for passing to the DB
 410          * Skips the titles specified by the 2-D array $existing
 411          * @param array $existing
 412          * @return array
 413          */
 414         private function getLinkInsertions( $existing = [] ) {
 415                 $arr = [];
 416                 foreach ( $this->mLinks as $ns => $dbkeys ) {
 417                         $diffs = isset( $existing[$ns] )
 418                                 ? array_diff_key( $dbkeys, $existing[$ns] )
 419                                 : $dbkeys;
 420                         foreach ( $diffs as $dbk => $id ) {
 421                                 $arr[] = [
 422                                         'pl_from' => $this->mId,
 423                                         'pl_from_namespace' => $this->mTitle->getNamespace(),
 424                                         'pl_namespace' => $ns,
 425                                         'pl_title' => $dbk
 426                                 ];
 427                         }
 428                 }
 429
 430                 return $arr;
 431         }
 432
 433         /**
 434          * Get an array of template insertions. Like getLinkInsertions()
 435          * @param array $existing
 436          * @return array
 437          */
 438         private function getTemplateInsertions( $existing = [] ) {
 439                 $arr = [];
 440                 foreach ( $this->mTemplates as $ns => $dbkeys ) {
 441                         $diffs = isset( $existing[$ns] ) ? array_diff_key( $dbkeys, $existing[$ns] ) : $dbkeys;
 442                         foreach ( $diffs as $dbk => $id ) {
 443                                 $arr[] = [
 444                                         'tl_from' => $this->mId,
 445                                         'tl_from_namespace' => $this->mTitle->getNamespace(),
 446                                         'tl_namespace' => $ns,
 447                                         'tl_title' => $dbk
 448                                 ];
 449                         }
 450                 }
 451
 452                 return $arr;
 453         }
 454
 455         /**
 456          * Get an array of image insertions
 457          * Skips the names specified in $existing
 458          * @param array $existing
 459          * @return array
 460          */
 461         private function getImageInsertions( $existing = [] ) {
 462                 $arr = [];
 463                 $diffs = array_diff_key( $this->mImages, $existing );
 464                 foreach ( $diffs as $iname => $dummy ) {
 465                         $arr[] = [
 466                                 'il_from' => $this->mId,
 467                                 'il_from_namespace' => $this->mTitle->getNamespace(),
 468                                 'il_to' => $iname
 469                         ];
 470                 }
 471
 472                 return $arr;
 473         }
 474
 475         /**
 476          * Get an array of externallinks insertions. Skips the names specified in $existing
 477          * @param array $existing
 478          * @return array
 479          */
 480         private function getExternalInsertions( $existing = [] ) {
 481                 $arr = [];
 482                 $diffs = array_diff_key( $this->mExternals, $existing );
 483                 foreach ( $diffs as $url => $dummy ) {
 484                         foreach ( wfMakeUrlIndexes( $url ) as $index ) {
 485                                 $arr[] = [
 486                                         'el_id' => $this->mDb->nextSequenceValue( 'externallinks_el_id_seq' ),
 487                                         'el_from' => $this->mId,
 488                                         'el_to' => $url,
 489                                         'el_index' => $index,
 490                                 ];
 491                         }
 492                 }
 493
 494                 return $arr;
 495         }
 496
 497         /**
 498          * Get an array of category insertions
 499          *
 500          * @param array $existing Mapping existing category names to sort keys. If both
 501          * match a link in $this, the link will be omitted from the output
 502          *
 503          * @return array
 504          */
 505         private function getCategoryInsertions( $existing = [] ) {
 506                 global $wgContLang, $wgCategoryCollation;
 507                 $diffs = array_diff_assoc( $this->mCategories, $existing );
 508                 $arr = [];
 509                 foreach ( $diffs as $name => $prefix ) {
 510                         $nt = Title::makeTitleSafe( NS_CATEGORY, $name );
 511                         $wgContLang->findVariantLink( $name, $nt, true );
 512
 513                         if ( $this->mTitle->getNamespace() == NS_CATEGORY ) {
 514                                 $type = 'subcat';
 515                         } elseif ( $this->mTitle->getNamespace() == NS_FILE ) {
 516                                 $type = 'file';
 517                         } else {
 518                                 $type = 'page';
 519                         }
 520
 521                         # Treat custom sortkeys as a prefix, so that if multiple
 522                         # things are forced to sort as '*' or something, they'll
 523                         # sort properly in the category rather than in page_id
 524                         # order or such.
 525                         $sortkey = Collation::singleton()->getSortKey(
 526                                 $this->mTitle->getCategorySortkey( $prefix ) );
 527
 528                         $arr[] = [
 529                                 'cl_from' => $this->mId,
 530                                 'cl_to' => $name,
 531                                 'cl_sortkey' => $sortkey,
 532                                 'cl_timestamp' => $this->mDb->timestamp(),
 533                                 'cl_sortkey_prefix' => $prefix,
 534                                 'cl_collation' => $wgCategoryCollation,
 535                                 'cl_type' => $type,
 536                         ];
 537                 }
 538
 539                 return $arr;
 540         }
 541
 542         /**
 543          * Get an array of interlanguage link insertions
 544          *
 545          * @param array $existing Mapping existing language codes to titles
 546          *
 547          * @return array
 548          */
 549         private function getInterlangInsertions( $existing = [] ) {
 550                 $diffs = array_diff_assoc( $this->mInterlangs, $existing );
 551                 $arr = [];
 552                 foreach ( $diffs as $lang => $title ) {
 553                         $arr[] = [
 554                                 'll_from' => $this->mId,
 555                                 'll_lang' => $lang,
 556                                 'll_title' => $title
 557                         ];
 558                 }
 559
 560                 return $arr;
 561         }
 562
 563         /**
 564          * Get an array of page property insertions
 565          * @param array $existing
 566          * @return array
 567          */
 568         function getPropertyInsertions( $existing = [] ) {
 569                 $diffs = array_diff_assoc( $this->mProperties, $existing );
 570
 571                 $arr = [];
 572                 foreach ( array_keys( $diffs ) as $name ) {
 573                         $arr[] = $this->getPagePropRowData( $name );
 574                 }
 575
 576                 return $arr;
 577         }
 578
 579         /**
 580          * Returns an associative array to be used for inserting a row into
 581          * the page_props table. Besides the given property name, this will
 582          * include the page id from $this->mId and any property value from
 583          * $this->mProperties.
 584          *
 585          * The array returned will include the pp_sortkey field if this
 586          * is present in the database (as indicated by $wgPagePropsHaveSortkey).
 587          * The sortkey value is currently determined by getPropertySortKeyValue().
 588          *
 589          * @note this assumes that $this->mProperties[$prop] is defined.
 590          *
 591          * @param string $prop The name of the property.
 592          *
 593          * @return array
 594          */
 595         private function getPagePropRowData( $prop ) {
 596                 global $wgPagePropsHaveSortkey;
 597
 598                 $value = $this->mProperties[$prop];
 599
 600                 $row = [
 601                         'pp_page' => $this->mId,
 602                         'pp_propname' => $prop,
 603                         'pp_value' => $value,
 604                 ];
 605
 606                 if ( $wgPagePropsHaveSortkey ) {
 607                         $row['pp_sortkey'] = $this->getPropertySortKeyValue( $value );
 608                 }
 609
 610                 return $row;
 611         }
 612
 613         /**
 614          * Determines the sort key for the given property value.
 615          * This will return $value if it is a float or int,
 616          * 1 or resp. 0 if it is a bool, and null otherwise.
 617          *
 618          * @note In the future, we may allow the sortkey to be specified explicitly
 619          *       in ParserOutput::setProperty.
 620          *
 621          * @param mixed $value
 622          *
 623          * @return float|null
 624          */
 625         private function getPropertySortKeyValue( $value ) {
 626                 if ( is_int( $value ) || is_float( $value ) || is_bool( $value ) ) {
 627                         return floatval( $value );
 628                 }
 629
 630                 return null;
 631         }
 632
 633         /**
 634          * Get an array of interwiki insertions for passing to the DB
 635          * Skips the titles specified by the 2-D array $existing
 636          * @param array $existing
 637          * @return array
 638          */
 639         private function getInterwikiInsertions( $existing = [] ) {
 640                 $arr = [];
 641                 foreach ( $this->mInterwikis as $prefix => $dbkeys ) {
 642                         $diffs = isset( $existing[$prefix] )
 643                                 ? array_diff_key( $dbkeys, $existing[$prefix] )
 644                                 : $dbkeys;
 645
 646                         foreach ( $diffs as $dbk => $id ) {
 647                                 $arr[] = [
 648                                         'iwl_from' => $this->mId,
 649                                         'iwl_prefix' => $prefix,
 650                                         'iwl_title' => $dbk
 651                                 ];
 652                         }
 653                 }
 654
 655                 return $arr;
 656         }
 657
 658         /**
 659          * Given an array of existing links, returns those links which are not in $this
 660          * and thus should be deleted.
 661          * @param array $existing
 662          * @return array
 663          */
 664         private function getLinkDeletions( $existing ) {
 665                 $del = [];
 666                 foreach ( $existing as $ns => $dbkeys ) {
 667                         if ( isset( $this->mLinks[$ns] ) ) {
 668                                 $del[$ns] = array_diff_key( $existing[$ns], $this->mLinks[$ns] );
 669                         } else {
 670                                 $del[$ns] = $existing[$ns];
 671                         }
 672                 }
 673
 674                 return $del;
 675         }
 676
 677         /**
 678          * Given an array of existing templates, returns those templates which are not in $this
 679          * and thus should be deleted.
 680          * @param array $existing
 681          * @return array
 682          */
 683         private function getTemplateDeletions( $existing ) {
 684                 $del = [];
 685                 foreach ( $existing as $ns => $dbkeys ) {
 686                         if ( isset( $this->mTemplates[$ns] ) ) {
 687                                 $del[$ns] = array_diff_key( $existing[$ns], $this->mTemplates[$ns] );
 688                         } else {
 689                                 $del[$ns] = $existing[$ns];
 690                         }
 691                 }
 692
 693                 return $del;
 694         }
 695
 696         /**
 697          * Given an array of existing images, returns those images which are not in $this
 698          * and thus should be deleted.
 699          * @param array $existing
 700          * @return array
 701          */
 702         private function getImageDeletions( $existing ) {
 703                 return array_diff_key( $existing, $this->mImages );
 704         }
 705
 706         /**
 707          * Given an array of existing external links, returns those links which are not
 708          * in $this and thus should be deleted.
 709          * @param array $existing
 710          * @return array
 711          */
 712         private function getExternalDeletions( $existing ) {
 713                 return array_diff_key( $existing, $this->mExternals );
 714         }
 715
 716         /**
 717          * Given an array of existing categories, returns those categories which are not in $this
 718          * and thus should be deleted.
 719          * @param array $existing
 720          * @return array
 721          */
 722         private function getCategoryDeletions( $existing ) {
 723                 return array_diff_assoc( $existing, $this->mCategories );
 724         }
 725
 726         /**
 727          * Given an array of existing interlanguage links, returns those links which are not
 728          * in $this and thus should be deleted.
 729          * @param array $existing
 730          * @return array
 731          */
 732         private function getInterlangDeletions( $existing ) {
 733                 return array_diff_assoc( $existing, $this->mInterlangs );
 734         }
 735
 736         /**
 737          * Get array of properties which should be deleted.
 738          * @param array $existing
 739          * @return array
 740          */
 741         function getPropertyDeletions( $existing ) {
 742                 return array_diff_assoc( $existing, $this->mProperties );
 743         }
 744
 745         /**
 746          * Given an array of existing interwiki links, returns those links which are not in $this
 747          * and thus should be deleted.
 748          * @param array $existing
 749          * @return array
 750          */
 751         private function getInterwikiDeletions( $existing ) {
 752                 $del = [];
 753                 foreach ( $existing as $prefix => $dbkeys ) {
 754                         if ( isset( $this->mInterwikis[$prefix] ) ) {
 755                                 $del[$prefix] = array_diff_key( $existing[$prefix], $this->mInterwikis[$prefix] );
 756                         } else {
 757                                 $del[$prefix] = $existing[$prefix];
 758                         }
 759                 }
 760
 761                 return $del;
 762         }
 763
 764         /**
 765          * Get an array of existing links, as a 2-D array
 766          *
 767          * @return array
 768          */
 769         private function getExistingLinks() {
 770                 $res = $this->mDb->select( 'pagelinks', [ 'pl_namespace', 'pl_title' ],
 771                         [ 'pl_from' => $this->mId ], __METHOD__, $this->mOptions );
 772                 $arr = [];
 773                 foreach ( $res as $row ) {
 774                         if ( !isset( $arr[$row->pl_namespace] ) ) {
 775                                 $arr[$row->pl_namespace] = [];
 776                         }
 777                         $arr[$row->pl_namespace][$row->pl_title] = 1;
 778                 }
 779
 780                 return $arr;
 781         }
 782
 783         /**
 784          * Get an array of existing templates, as a 2-D array
 785          *
 786          * @return array
 787          */
 788         private function getExistingTemplates() {
 789                 $res = $this->mDb->select( 'templatelinks', [ 'tl_namespace', 'tl_title' ],
 790                         [ 'tl_from' => $this->mId ], __METHOD__, $this->mOptions );
 791                 $arr = [];
 792                 foreach ( $res as $row ) {
 793                         if ( !isset( $arr[$row->tl_namespace] ) ) {
 794                                 $arr[$row->tl_namespace] = [];
 795                         }
 796                         $arr[$row->tl_namespace][$row->tl_title] = 1;
 797                 }
 798
 799                 return $arr;
 800         }
 801
 802         /**
 803          * Get an array of existing images, image names in the keys
 804          *
 805          * @return array
 806          */
 807         private function getExistingImages() {
 808                 $res = $this->mDb->select( 'imagelinks', [ 'il_to' ],
 809                         [ 'il_from' => $this->mId ], __METHOD__, $this->mOptions );
 810                 $arr = [];
 811                 foreach ( $res as $row ) {
 812                         $arr[$row->il_to] = 1;
 813                 }
 814
 815                 return $arr;
 816         }
 817
 818         /**
 819          * Get an array of existing external links, URLs in the keys
 820          *
 821          * @return array
 822          */
 823         private function getExistingExternals() {
 824                 $res = $this->mDb->select( 'externallinks', [ 'el_to' ],
 825                         [ 'el_from' => $this->mId ], __METHOD__, $this->mOptions );
 826                 $arr = [];
 827                 foreach ( $res as $row ) {
 828                         $arr[$row->el_to] = 1;
 829                 }
 830
 831                 return $arr;
 832         }
 833
 834         /**
 835          * Get an array of existing categories, with the name in the key and sort key in the value.
 836          *
 837          * @return array
 838          */
 839         private function getExistingCategories() {
 840                 $res = $this->mDb->select( 'categorylinks', [ 'cl_to', 'cl_sortkey_prefix' ],
 841                         [ 'cl_from' => $this->mId ], __METHOD__, $this->mOptions );
 842                 $arr = [];
 843                 foreach ( $res as $row ) {
 844                         $arr[$row->cl_to] = $row->cl_sortkey_prefix;
 845                 }
 846
 847                 return $arr;
 848         }
 849
 850         /**
 851          * Get an array of existing interlanguage links, with the language code in the key and the
 852          * title in the value.
 853          *
 854          * @return array
 855          */
 856         private function getExistingInterlangs() {
 857                 $res = $this->mDb->select( 'langlinks', [ 'll_lang', 'll_title' ],
 858                         [ 'll_from' => $this->mId ], __METHOD__, $this->mOptions );
 859                 $arr = [];
 860                 foreach ( $res as $row ) {
 861                         $arr[$row->ll_lang] = $row->ll_title;
 862                 }
 863
 864                 return $arr;
 865         }
 866
 867         /**
 868          * Get an array of existing inline interwiki links, as a 2-D array
 869          * @return array (prefix => array(dbkey => 1))
 870          */
 871         protected function getExistingInterwikis() {
 872                 $res = $this->mDb->select( 'iwlinks', [ 'iwl_prefix', 'iwl_title' ],
 873                         [ 'iwl_from' => $this->mId ], __METHOD__, $this->mOptions );
 874                 $arr = [];
 875                 foreach ( $res as $row ) {
 876                         if ( !isset( $arr[$row->iwl_prefix] ) ) {
 877                                 $arr[$row->iwl_prefix] = [];
 878                         }
 879                         $arr[$row->iwl_prefix][$row->iwl_title] = 1;
 880                 }
 881
 882                 return $arr;
 883         }
 884
 885         /**
 886          * Get an array of existing categories, with the name in the key and sort key in the value.
 887          *
 888          * @return array Array of property names and values
 889          */
 890         private function getExistingProperties() {
 891                 $res = $this->mDb->select( 'page_props', [ 'pp_propname', 'pp_value' ],
 892                         [ 'pp_page' => $this->mId ], __METHOD__, $this->mOptions );
 893                 $arr = [];
 894                 foreach ( $res as $row ) {
 895                         $arr[$row->pp_propname] = $row->pp_value;
 896                 }
 897
 898                 return $arr;
 899         }
 900
 901         /**
 902          * Return the title object of the page being updated
 903          * @return Title
 904          */
 905         public function getTitle() {
 906                 return $this->mTitle;
 907         }
 908
 909         /**
 910          * Returns parser output
 911          * @since 1.19
 912          * @return ParserOutput
 913          */
 914         public function getParserOutput() {
 915                 return $this->mParserOutput;
 916         }
 917
 918         /**
 919          * Return the list of images used as generated by the parser
 920          * @return array
 921          */
 922         public function getImages() {
 923                 return $this->mImages;
 924         }
 925
 926         /**
 927          * Set the revision corresponding to this LinksUpdate
 928          *
 929          * @since 1.27
 930          *
 931          * @param Revision $revision
 932          */
 933         public function setRevision( Revision $revision ) {
 934                 $this->mRevision = $revision;
 935         }
 936
 937         /**
 938          * @since 1.28
 939          * @return null|Revision
 940          */
 941         public function getRevision() {
 942                 return $this->mRevision;
 943         }
 944
 945         /**
 946          * Set the User who triggered this LinksUpdate
 947          *
 948          * @since 1.27
 949          * @param User $user
 950          */
 951         public function setTriggeringUser( User $user ) {
 952                 $this->user = $user;
 953         }
 954
 955         /**
 956          * @since 1.27
 957          * @return null|User
 958          */
 959         public function getTriggeringUser() {
 960                 return $this->user;
 961         }
 962
 963         /**
 964          * Invalidate any necessary link lists related to page property changes
 965          * @param array $changed
 966          */
 967         private function invalidateProperties( $changed ) {
 968                 global $wgPagePropLinkInvalidations;
 969
 970                 foreach ( $changed as $name => $value ) {
 971                         if ( isset( $wgPagePropLinkInvalidations[$name] ) ) {
 972                                 $inv = $wgPagePropLinkInvalidations[$name];
 973                                 if ( !is_array( $inv ) ) {
 974                                         $inv = [ $inv ];
 975                                 }
 976                                 foreach ( $inv as $table ) {
 977                                         DeferredUpdates::addUpdate( new HTMLCacheUpdate( $this->mTitle, $table ) );
 978                                 }
 979                         }
 980                 }
 981         }
 982
 983         /**
 984          * Fetch page links added by this LinksUpdate.  Only available after the update is complete.
 985          * @since 1.22
 986          * @return null|array Array of Titles
 987          */
 988         public function getAddedLinks() {
 989                 if ( $this->linkInsertions === null ) {
 990                         return null;
 991                 }
 992                 $result = [];
 993                 foreach ( $this->linkInsertions as $insertion ) {
 994                         $result[] = Title::makeTitle( $insertion['pl_namespace'], $insertion['pl_title'] );
 995                 }
 996
 997                 return $result;
 998         }
 999
1000         /**
1001          * Fetch page links removed by this LinksUpdate.  Only available after the update is complete.
1002          * @since 1.22
1003          * @return null|array Array of Titles
1004          */
1005         public function getRemovedLinks() {
1006                 if ( $this->linkDeletions === null ) {
1007                         return null;
1008                 }
1009                 $result = [];
1010                 foreach ( $this->linkDeletions as $ns => $titles ) {
1011                         foreach ( $titles as $title => $unused ) {
1012                                 $result[] = Title::makeTitle( $ns, $title );
1013                         }
1014                 }
1015
1016                 return $result;
1017         }
1018
1019         /**
1020          * Update links table freshness
1021          */
1022         protected function updateLinksTimestamp() {
1023                 if ( $this->mId ) {
1024                         // The link updates made here only reflect the freshness of the parser output
1025                         $timestamp = $this->mParserOutput->getCacheTime();
1026                         $this->mDb->update( 'page',
1027                                 [ 'page_links_updated' => $this->mDb->timestamp( $timestamp ) ],
1028                                 [ 'page_id' => $this->mId ],
1029                                 __METHOD__
1030                         );
1031                 }
1032         }
1033
1034         public function getAsJobSpecification() {
1035                 if ( $this->user ) {
1036                         $userInfo = [
1037                                 'userId' => $this->user->getId(),
1038                                 'userName' => $this->user->getName(),
1039                         ];
1040                 } else {
1041                         $userInfo = false;
1042                 }
1043
1044                 if ( $this->mRevision ) {
1045                         $triggeringRevisionId = $this->mRevision->getId();
1046                 } else {
1047                         $triggeringRevisionId = false;
1048                 }
1049
1050                 return [
1051                         'wiki' => $this->mDb->getWikiID(),
1052                         'job'  => new JobSpecification(
1053                                 'refreshLinksPrioritized',
1054                                 [
1055                                         // Reuse the parser cache if it was saved
1056                                         'rootJobTimestamp' => $this->mParserOutput->getCacheTime(),
1057                                         'useRecursiveLinksUpdate' => $this->mRecursive,
1058                                         'triggeringUser' => $userInfo,
1059                                         'triggeringRevisionId' => $triggeringRevisionId,
1060                                 ],
1061                                 [ 'removeDuplicates' => true ],
1062                                 $this->getTitle()
1063                         )
1064                 ];
1065         }
1066 }