includes/deferred/LinksUpdate.php

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