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