includes/deferred/LinksUpdate.php

   1 <?php
   2 /**
   3  * Updater for link tracking tables after a page edit.
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  */
  22
  23 use Wikimedia\Rdbms\IDatabase;
  24 use MediaWiki\Logger\LoggerFactory;
  25 use MediaWiki\MediaWikiServices;
  26 use Wikimedia\ScopedCallback;
  27
  28 /**
  29  * Class the manages updates of *_link tables as well as similar extension-managed tables
  30  *
  31  * @note: LinksUpdate is managed by DeferredUpdates::execute(). Do not run this in a transaction.
  32  *
  33  * See docs/deferred.txt
  34  */
  35 class LinksUpdate extends DataUpdate implements EnqueueableDataUpdate {
  36         // @todo make members protected, but make sure extensions don't break
  37
  38         /** @var int Page ID of the article linked from */
  39         public $mId;
  40
  41         /** @var Title Title object of the article linked from */
  42         public $mTitle;
  43
  44         /** @var ParserOutput */
  45         public $mParserOutput;
  46
  47         /** @var array Map of title strings to IDs for the links in the document */
  48         public $mLinks;
  49
  50         /** @var array DB keys of the images used, in the array key only */
  51         public $mImages;
  52
  53         /** @var array Map of title strings to IDs for the template references, including broken ones */
  54         public $mTemplates;
  55
  56         /** @var array URLs of external links, array key only */
  57         public $mExternals;
  58
  59         /** @var array Map of category names to sort keys */
  60         public $mCategories;
  61
  62         /** @var array Map of language codes to titles */
  63         public $mInterlangs;
  64
  65         /** @var array 2-D map of (prefix => DBK => 1) */
  66         public $mInterwikis;
  67
  68         /** @var array Map of arbitrary name to value */
  69         public $mProperties;
  70
  71         /** @var bool Whether to queue jobs for recursive updates */
  72         public $mRecursive;
  73
  74         /** @var Revision Revision for which this update has been triggered */
  75         private $mRevision;
  76
  77         /**
  78          * @var null|array Added links if calculated.
  79          */
  80         private $linkInsertions = null;
  81
  82         /**
  83          * @var null|array Deleted links if calculated.
  84          */
  85         private $linkDeletions = null;
  86
  87         /**
  88          * @var null|array Added external links if calculated.
  89          */
  90         private $externalLinkInsertions = null;
  91
  92         /**
  93          * @var null|array Deleted external links if calculated.
  94          */
  95         private $externalLinkDeletions = null;
  96
  97         /**
  98          * @var null|array Added properties if calculated.
  99          */
 100         private $propertyInsertions = null;
 101
 102         /**
 103          * @var null|array Deleted properties if calculated.
 104          */
 105         private $propertyDeletions = null;
 106
 107         /**
 108          * @var User|null
 109          */
 110         private $user;
 111
 112         /** @var IDatabase */
 113         private $db;
 114
 115         /**
 116          * @param Title $title Title of the page we're updating
 117          * @param ParserOutput $parserOutput Output from a full parse of this page
 118          * @param bool $recursive Queue jobs for recursive updates?
 119          * @throws MWException
 120          */
 121         function __construct( Title $title, ParserOutput $parserOutput, $recursive = true ) {
 122                 parent::__construct();
 123
 124                 $this->mTitle = $title;
 125
 126                 if ( !$this->mId ) {
 127                         // NOTE: subclasses may initialize mId before calling this constructor!
 128                         $this->mId = $title->getArticleID( Title::GAID_FOR_UPDATE );
 129                 }
 130
 131                 if ( !$this->mId ) {
 132                         throw new InvalidArgumentException(
 133                                 "The Title object yields no ID. Perhaps the page doesn't exist?"
 134                         );
 135                 }
 136
 137                 $this->mParserOutput = $parserOutput;
 138
 139                 $this->mLinks = $parserOutput->getLinks();
 140                 $this->mImages = $parserOutput->getImages();
 141                 $this->mTemplates = $parserOutput->getTemplates();
 142                 $this->mExternals = $parserOutput->getExternalLinks();
 143                 $this->mCategories = $parserOutput->getCategories();
 144                 $this->mProperties = $parserOutput->getProperties();
 145                 $this->mInterwikis = $parserOutput->getInterwikiLinks();
 146
 147                 # Convert the format of the interlanguage links
 148                 # I didn't want to change it in the ParserOutput, because that array is passed all
 149                 # the way back to the skin, so either a skin API break would be required, or an
 150                 # inefficient back-conversion.
 151                 $ill = $parserOutput->getLanguageLinks();
 152                 $this->mInterlangs = [];
 153                 foreach ( $ill as $link ) {
 154                         list( $key, $title ) = explode( ':', $link, 2 );
 155                         $this->mInterlangs[$key] = $title;
 156                 }
 157
 158                 foreach ( $this->mCategories as &$sortkey ) {
 159                         # If the sortkey is longer then 255 bytes, it is truncated by DB, and then doesn't match
 160                         # when comparing existing vs current categories, causing T27254.
 161                         $sortkey = mb_strcut( $sortkey, 0, 255 );
 162                 }
 163
 164                 $this->mRecursive = $recursive;
 165
 166                 // Avoid PHP 7.1 warning from passing $this by reference
 167                 $linksUpdate = $this;
 168                 Hooks::run( 'LinksUpdateConstructed', [ &$linksUpdate ] );
 169         }
 170
 171         /**
 172          * Update link tables with outgoing links from an updated article
 173          *
 174          * @note this is managed by DeferredUpdates::execute(). Do not run this in a transaction.
 175          */
 176         public function doUpdate() {
 177                 if ( $this->ticket ) {
 178                         // Make sure all links update threads see the changes of each other.
 179                         // This handles the case when updates have to batched into several COMMITs.
 180                         $scopedLock = self::acquirePageLock( $this->getDB(), $this->mId );
 181                         if ( !$scopedLock ) {
 182                                 throw new RuntimeException( "Could not acquire lock for page ID '{$this->mId}'." );
 183                         }
 184                 }
 185
 186                 // Avoid PHP 7.1 warning from passing $this by reference
 187                 $linksUpdate = $this;
 188                 Hooks::run( 'LinksUpdate', [ &$linksUpdate ] );
 189                 $this->doIncrementalUpdate();
 190
 191                 // Commit and release the lock (if set)
 192                 ScopedCallback::consume( $scopedLock );
 193                 // Run post-commit hook handlers without DBO_TRX
 194                 DeferredUpdates::addUpdate( new AutoCommitUpdate(
 195                         $this->getDB(),
 196                         __METHOD__,
 197                         function () {
 198                                 // Avoid PHP 7.1 warning from passing $this by reference
 199                                 $linksUpdate = $this;
 200                                 Hooks::run( 'LinksUpdateComplete', [ &$linksUpdate, $this->ticket ] );
 201                         }
 202                 ) );
 203         }
 204
 205         /**
 206          * Acquire a session-level lock for performing link table updates for a page on a DB
 207          *
 208          * @param IDatabase $dbw
 209          * @param int $pageId
 210          * @param string $why One of (job, atomicity)
 211          * @return ScopedCallback|null
 212          * @since 1.27
 213          */
 214         public static function acquirePageLock( IDatabase $dbw, $pageId, $why = 'atomicity' ) {
 215                 $key = "{$dbw->getDomainID()}:LinksUpdate:$why:pageid:$pageId"; // per-wiki
 216                 $scopedLock = $dbw->getScopedLockAndFlush( $key, __METHOD__, 15 );
 217                 if ( !$scopedLock ) {
 218                         $logger = LoggerFactory::getInstance( 'SecondaryDataUpdate' );
 219                         $logger->info( "Could not acquire lock '{key}' for page ID '{page_id}'.", [
 220                                 'key' => $key,
 221                                 'page_id' => $pageId,
 222                         ] );
 223                         return null;
 224                 }
 225
 226                 return $scopedLock;
 227         }
 228
 229         protected function doIncrementalUpdate() {
 230                 # Page links
 231                 $existingPL = $this->getExistingLinks();
 232                 $this->linkDeletions = $this->getLinkDeletions( $existingPL );
 233                 $this->linkInsertions = $this->getLinkInsertions( $existingPL );
 234                 $this->incrTableUpdate( 'pagelinks', 'pl', $this->linkDeletions, $this->linkInsertions );
 235
 236                 # Image links
 237                 $existingIL = $this->getExistingImages();
 238                 $imageDeletes = $this->getImageDeletions( $existingIL );
 239                 $this->incrTableUpdate(
 240                         'imagelinks',
 241                         'il',
 242                         $imageDeletes,
 243                         $this->getImageInsertions( $existingIL ) );
 244
 245                 # Invalidate all image description pages which had links added or removed
 246                 $imageUpdates = $imageDeletes + array_diff_key( $this->mImages, $existingIL );
 247                 $this->invalidateImageDescriptions( $imageUpdates );
 248
 249                 # External links
 250                 $existingEL = $this->getExistingExternals();
 251                 $this->externalLinkDeletions = $this->getExternalDeletions( $existingEL );
 252                 $this->externalLinkInsertions = $this->getExternalInsertions(
 253                         $existingEL );
 254                 $this->incrTableUpdate(
 255                         'externallinks',
 256                         'el',
 257                         $this->externalLinkDeletions,
 258                         $this->externalLinkInsertions );
 259
 260                 # Language links
 261                 $existingLL = $this->getExistingInterlangs();
 262                 $this->incrTableUpdate(
 263                         'langlinks',
 264                         'll',
 265                         $this->getInterlangDeletions( $existingLL ),
 266                         $this->getInterlangInsertions( $existingLL ) );
 267
 268                 # Inline interwiki links
 269                 $existingIW = $this->getExistingInterwikis();
 270                 $this->incrTableUpdate(
 271                         'iwlinks',
 272                         'iwl',
 273                         $this->getInterwikiDeletions( $existingIW ),
 274                         $this->getInterwikiInsertions( $existingIW ) );
 275
 276                 # Template links
 277                 $existingTL = $this->getExistingTemplates();
 278                 $this->incrTableUpdate(
 279                         'templatelinks',
 280                         'tl',
 281                         $this->getTemplateDeletions( $existingTL ),
 282                         $this->getTemplateInsertions( $existingTL ) );
 283
 284                 # Category links
 285                 $existingCL = $this->getExistingCategories();
 286                 $categoryDeletes = $this->getCategoryDeletions( $existingCL );
 287                 $this->incrTableUpdate(
 288                         'categorylinks',
 289                         'cl',
 290                         $categoryDeletes,
 291                         $this->getCategoryInsertions( $existingCL ) );
 292                 $categoryInserts = array_diff_assoc( $this->mCategories, $existingCL );
 293                 $categoryUpdates = $categoryInserts + $categoryDeletes;
 294
 295                 # Page properties
 296                 $existingPP = $this->getExistingProperties();
 297                 $this->propertyDeletions = $this->getPropertyDeletions( $existingPP );
 298                 $this->incrTableUpdate(
 299                         'page_props',
 300                         'pp',
 301                         $this->propertyDeletions,
 302                         $this->getPropertyInsertions( $existingPP ) );
 303
 304                 # Invalidate the necessary pages
 305                 $this->propertyInsertions = array_diff_assoc( $this->mProperties, $existingPP );
 306                 $changed = $this->propertyDeletions + $this->propertyInsertions;
 307                 $this->invalidateProperties( $changed );
 308
 309                 # Invalidate all categories which were added, deleted or changed (set symmetric difference)
 310                 $this->invalidateCategories( $categoryUpdates );
 311                 $this->updateCategoryCounts( $categoryInserts, $categoryDeletes );
 312
 313                 # Refresh links of all pages including this page
 314                 # This will be in a separate transaction
 315                 if ( $this->mRecursive ) {
 316                         $this->queueRecursiveJobs();
 317                 }
 318
 319                 # Update the links table freshness for this title
 320                 $this->updateLinksTimestamp();
 321         }
 322
 323         /**
 324          * Queue recursive jobs for this page
 325          *
 326          * Which means do LinksUpdate on all pages that include the current page,
 327          * using the job queue.
 328          */
 329         protected function queueRecursiveJobs() {
 330                 $action = $this->getCauseAction();
 331                 $agent = $this->getCauseAgent();
 332
 333                 self::queueRecursiveJobsForTable( $this->mTitle, 'templatelinks', $action, $agent );
 334                 if ( $this->mTitle->getNamespace() == NS_FILE ) {
 335                         // Process imagelinks in case the title is or was a redirect
 336                         self::queueRecursiveJobsForTable( $this->mTitle, 'imagelinks', $action, $agent );
 337                 }
 338
 339                 $bc = $this->mTitle->getBacklinkCache();
 340                 // Get jobs for cascade-protected backlinks for a high priority queue.
 341                 // If meta-templates change to using a new template, the new template
 342                 // should be implicitly protected as soon as possible, if applicable.
 343                 // These jobs duplicate a subset of the above ones, but can run sooner.
 344                 // Which ever runs first generally no-ops the other one.
 345                 $jobs = [];
 346                 foreach ( $bc->getCascadeProtectedLinks() as $title ) {
 347                         $jobs[] = RefreshLinksJob::newPrioritized(
 348                                 $title,
 349                                 [
 350                                         'causeAction' => $action,
 351                                         'causeAgent' => $agent
 352                                 ]
 353                         );
 354                 }
 355                 JobQueueGroup::singleton()->push( $jobs );
 356         }
 357
 358         /**
 359          * Queue a RefreshLinks job for any table.
 360          *
 361          * @param Title $title Title to do job for
 362          * @param string $table Table to use (e.g. 'templatelinks')
 363          * @param string $action Triggering action
 364          * @param string $userName Triggering user name
 365          */
 366         public static function queueRecursiveJobsForTable(
 367                 Title $title, $table, $action = 'unknown', $userName = 'unknown'
 368         ) {
 369                 if ( $title->getBacklinkCache()->hasLinks( $table ) ) {
 370                         $job = new RefreshLinksJob(
 371                                 $title,
 372                                 [
 373                                         'table' => $table,
 374                                         'recursive' => true,
 375                                 ] + Job::newRootJobParams( // "overall" refresh links job info
 376                                         "refreshlinks:{$table}:{$title->getPrefixedText()}"
 377                                 ) + [ 'causeAction' => $action, 'causeAgent' => $userName ]
 378                         );
 379
 380                         JobQueueGroup::singleton()->push( $job );
 381                 }
 382         }
 383
 384         /**
 385          * @param array $cats
 386          */
 387         private function invalidateCategories( $cats ) {
 388                 PurgeJobUtils::invalidatePages( $this->getDB(), NS_CATEGORY, array_keys( $cats ) );
 389         }
 390
 391         /**
 392          * Update all the appropriate counts in the category table.
 393          * @param array $added Associative array of category name => sort key
 394          * @param array $deleted Associative array of category name => sort key
 395          */
 396         private function updateCategoryCounts( array $added, array $deleted ) {
 397                 global $wgUpdateRowsPerQuery;
 398
 399                 if ( !$added && !$deleted ) {
 400                         return;
 401                 }
 402
 403                 $domainId = $this->getDB()->getDomainID();
 404                 $wp = WikiPage::factory( $this->mTitle );
 405                 $lbf = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
 406                 // T163801: try to release any row locks to reduce contention
 407                 $lbf->commitAndWaitForReplication( __METHOD__, $this->ticket, [ 'domain' => $domainId ] );
 408
 409                 foreach ( array_chunk( array_keys( $added ), $wgUpdateRowsPerQuery ) as $addBatch ) {
 410                         $wp->updateCategoryCounts( $addBatch, [], $this->mId );
 411                         $lbf->commitAndWaitForReplication(
 412                                 __METHOD__, $this->ticket, [ 'domain' => $domainId ] );
 413                 }
 414
 415                 foreach ( array_chunk( array_keys( $deleted ), $wgUpdateRowsPerQuery ) as $deleteBatch ) {
 416                         $wp->updateCategoryCounts( [], $deleteBatch, $this->mId );
 417                         $lbf->commitAndWaitForReplication(
 418                                 __METHOD__, $this->ticket, [ 'domain' => $domainId ] );
 419                 }
 420         }
 421
 422         /**
 423          * @param array $images
 424          */
 425         private function invalidateImageDescriptions( array $images ) {
 426                 PurgeJobUtils::invalidatePages( $this->getDB(), NS_FILE, array_keys( $images ) );
 427         }
 428
 429         /**
 430          * Update a table by doing a delete query then an insert query
 431          * @param string $table Table name
 432          * @param string $prefix Field name prefix
 433          * @param array $deletions
 434          * @param array $insertions Rows to insert
 435          */
 436         private function incrTableUpdate( $table, $prefix, $deletions, $insertions ) {
 437                 $services = MediaWikiServices::getInstance();
 438                 $bSize = $services->getMainConfig()->get( 'UpdateRowsPerQuery' );
 439                 $lbf = $services->getDBLoadBalancerFactory();
 440
 441                 if ( $table === 'page_props' ) {
 442                         $fromField = 'pp_page';
 443                 } else {
 444                         $fromField = "{$prefix}_from";
 445                 }
 446
 447                 $deleteWheres = []; // list of WHERE clause arrays for each DB delete() call
 448                 if ( $table === 'pagelinks' || $table === 'templatelinks' || $table === 'iwlinks' ) {
 449                         $baseKey = ( $table === 'iwlinks' ) ? 'iwl_prefix' : "{$prefix}_namespace";
 450
 451                         $curBatchSize = 0;
 452                         $curDeletionBatch = [];
 453                         $deletionBatches = [];
 454                         foreach ( $deletions as $ns => $dbKeys ) {
 455                                 foreach ( $dbKeys as $dbKey => $unused ) {
 456                                         $curDeletionBatch[$ns][$dbKey] = 1;
 457                                         if ( ++$curBatchSize >= $bSize ) {
 458                                                 $deletionBatches[] = $curDeletionBatch;
 459                                                 $curDeletionBatch = [];
 460                                                 $curBatchSize = 0;
 461                                         }
 462                                 }
 463                         }
 464                         if ( $curDeletionBatch ) {
 465                                 $deletionBatches[] = $curDeletionBatch;
 466                         }
 467
 468                         foreach ( $deletionBatches as $deletionBatch ) {
 469                                 $deleteWheres[] = [
 470                                         $fromField => $this->mId,
 471                                         $this->getDB()->makeWhereFrom2d( $deletionBatch, $baseKey, "{$prefix}_title" )
 472                                 ];
 473                         }
 474                 } else {
 475                         if ( $table === 'langlinks' ) {
 476                                 $toField = 'll_lang';
 477                         } elseif ( $table === 'page_props' ) {
 478                                 $toField = 'pp_propname';
 479                         } else {
 480                                 $toField = $prefix . '_to';
 481                         }
 482
 483                         $deletionBatches = array_chunk( array_keys( $deletions ), $bSize );
 484                         foreach ( $deletionBatches as $deletionBatch ) {
 485                                 $deleteWheres[] = [ $fromField => $this->mId, $toField => $deletionBatch ];
 486                         }
 487                 }
 488
 489                 $domainId = $this->getDB()->getDomainID();
 490
 491                 foreach ( $deleteWheres as $deleteWhere ) {
 492                         $this->getDB()->delete( $table, $deleteWhere, __METHOD__ );
 493                         $lbf->commitAndWaitForReplication(
 494                                 __METHOD__, $this->ticket, [ 'domain' => $domainId ]
 495                         );
 496                 }
 497
 498                 $insertBatches = array_chunk( $insertions, $bSize );
 499                 foreach ( $insertBatches as $insertBatch ) {
 500                         $this->getDB()->insert( $table, $insertBatch, __METHOD__, [ 'IGNORE' ] );
 501                         $lbf->commitAndWaitForReplication(
 502                                 __METHOD__, $this->ticket, [ 'domain' => $domainId ]
 503                         );
 504                 }
 505
 506                 if ( count( $insertions ) ) {
 507                         Hooks::run( 'LinksUpdateAfterInsert', [ $this, $table, $insertions ] );
 508                 }
 509         }
 510
 511         /**
 512          * Get an array of pagelinks insertions for passing to the DB
 513          * Skips the titles specified by the 2-D array $existing
 514          * @param array $existing
 515          * @return array
 516          */
 517         private function getLinkInsertions( $existing = [] ) {
 518                 $arr = [];
 519                 foreach ( $this->mLinks as $ns => $dbkeys ) {
 520                         $diffs = isset( $existing[$ns] )
 521                                 ? array_diff_key( $dbkeys, $existing[$ns] )
 522                                 : $dbkeys;
 523                         foreach ( $diffs as $dbk => $id ) {
 524                                 $arr[] = [
 525                                         'pl_from' => $this->mId,
 526                                         'pl_from_namespace' => $this->mTitle->getNamespace(),
 527                                         'pl_namespace' => $ns,
 528                                         'pl_title' => $dbk
 529                                 ];
 530                         }
 531                 }
 532
 533                 return $arr;
 534         }
 535
 536         /**
 537          * Get an array of template insertions. Like getLinkInsertions()
 538          * @param array $existing
 539          * @return array
 540          */
 541         private function getTemplateInsertions( $existing = [] ) {
 542                 $arr = [];
 543                 foreach ( $this->mTemplates as $ns => $dbkeys ) {
 544                         $diffs = isset( $existing[$ns] ) ? array_diff_key( $dbkeys, $existing[$ns] ) : $dbkeys;
 545                         foreach ( $diffs as $dbk => $id ) {
 546                                 $arr[] = [
 547                                         'tl_from' => $this->mId,
 548                                         'tl_from_namespace' => $this->mTitle->getNamespace(),
 549                                         'tl_namespace' => $ns,
 550                                         'tl_title' => $dbk
 551                                 ];
 552                         }
 553                 }
 554
 555                 return $arr;
 556         }
 557
 558         /**
 559          * Get an array of image insertions
 560          * Skips the names specified in $existing
 561          * @param array $existing
 562          * @return array
 563          */
 564         private function getImageInsertions( $existing = [] ) {
 565                 $arr = [];
 566                 $diffs = array_diff_key( $this->mImages, $existing );
 567                 foreach ( $diffs as $iname => $dummy ) {
 568                         $arr[] = [
 569                                 'il_from' => $this->mId,
 570                                 'il_from_namespace' => $this->mTitle->getNamespace(),
 571                                 'il_to' => $iname
 572                         ];
 573                 }
 574
 575                 return $arr;
 576         }
 577
 578         /**
 579          * Get an array of externallinks insertions. Skips the names specified in $existing
 580          * @param array $existing
 581          * @return array
 582          */
 583         private function getExternalInsertions( $existing = [] ) {
 584                 $arr = [];
 585                 $diffs = array_diff_key( $this->mExternals, $existing );
 586                 foreach ( $diffs as $url => $dummy ) {
 587                         foreach ( LinkFilter::makeIndexes( $url ) as $index ) {
 588                                 $arr[] = [
 589                                         'el_from' => $this->mId,
 590                                         'el_to' => $url,
 591                                         'el_index' => $index,
 592                                         'el_index_60' => substr( $index, 0, 60 ),
 593                                 ];
 594                         }
 595                 }
 596
 597                 return $arr;
 598         }
 599
 600         /**
 601          * Get an array of category insertions
 602          *
 603          * @param array $existing Mapping existing category names to sort keys. If both
 604          * match a link in $this, the link will be omitted from the output
 605          *
 606          * @return array
 607          */
 608         private function getCategoryInsertions( $existing = [] ) {
 609                 global $wgCategoryCollation;
 610                 $diffs = array_diff_assoc( $this->mCategories, $existing );
 611                 $arr = [];
 612                 $contLang = MediaWikiServices::getInstance()->getContentLanguage();
 613                 $collation = Collation::singleton();
 614                 foreach ( $diffs as $name => $prefix ) {
 615                         $nt = Title::makeTitleSafe( NS_CATEGORY, $name );
 616                         $contLang->findVariantLink( $name, $nt, true );
 617
 618                         $type = MediaWikiServices::getInstance()->getNamespaceInfo()->
 619                                 getCategoryLinkType( $this->mTitle->getNamespace() );
 620
 621                         # Treat custom sortkeys as a prefix, so that if multiple
 622                         # things are forced to sort as '*' or something, they'll
 623                         # sort properly in the category rather than in page_id
 624                         # order or such.
 625                         $sortkey = $collation->getSortKey( $this->mTitle->getCategorySortkey( $prefix ) );
 626
 627                         $arr[] = [
 628                                 'cl_from' => $this->mId,
 629                                 'cl_to' => $name,
 630                                 'cl_sortkey' => $sortkey,
 631                                 'cl_timestamp' => $this->getDB()->timestamp(),
 632                                 'cl_sortkey_prefix' => $prefix,
 633                                 'cl_collation' => $wgCategoryCollation,
 634                                 'cl_type' => $type,
 635                         ];
 636                 }
 637
 638                 return $arr;
 639         }
 640
 641         /**
 642          * Get an array of interlanguage link insertions
 643          *
 644          * @param array $existing Mapping existing language codes to titles
 645          *
 646          * @return array
 647          */
 648         private function getInterlangInsertions( $existing = [] ) {
 649                 $diffs = array_diff_assoc( $this->mInterlangs, $existing );
 650                 $arr = [];
 651                 foreach ( $diffs as $lang => $title ) {
 652                         $arr[] = [
 653                                 'll_from' => $this->mId,
 654                                 'll_lang' => $lang,
 655                                 'll_title' => $title
 656                         ];
 657                 }
 658
 659                 return $arr;
 660         }
 661
 662         /**
 663          * Get an array of page property insertions
 664          * @param array $existing
 665          * @return array
 666          */
 667         function getPropertyInsertions( $existing = [] ) {
 668                 $diffs = array_diff_assoc( $this->mProperties, $existing );
 669
 670                 $arr = [];
 671                 foreach ( array_keys( $diffs ) as $name ) {
 672                         $arr[] = $this->getPagePropRowData( $name );
 673                 }
 674
 675                 return $arr;
 676         }
 677
 678         /**
 679          * Returns an associative array to be used for inserting a row into
 680          * the page_props table. Besides the given property name, this will
 681          * include the page id from $this->mId and any property value from
 682          * $this->mProperties.
 683          *
 684          * The array returned will include the pp_sortkey field if this
 685          * is present in the database (as indicated by $wgPagePropsHaveSortkey).
 686          * The sortkey value is currently determined by getPropertySortKeyValue().
 687          *
 688          * @note this assumes that $this->mProperties[$prop] is defined.
 689          *
 690          * @param string $prop The name of the property.
 691          *
 692          * @return array
 693          */
 694         private function getPagePropRowData( $prop ) {
 695                 global $wgPagePropsHaveSortkey;
 696
 697                 $value = $this->mProperties[$prop];
 698
 699                 $row = [
 700                         'pp_page' => $this->mId,
 701                         'pp_propname' => $prop,
 702                         'pp_value' => $value,
 703                 ];
 704
 705                 if ( $wgPagePropsHaveSortkey ) {
 706                         $row['pp_sortkey'] = $this->getPropertySortKeyValue( $value );
 707                 }
 708
 709                 return $row;
 710         }
 711
 712         /**
 713          * Determines the sort key for the given property value.
 714          * This will return $value if it is a float or int,
 715          * 1 or resp. 0 if it is a bool, and null otherwise.
 716          *
 717          * @note In the future, we may allow the sortkey to be specified explicitly
 718          *       in ParserOutput::setProperty.
 719          *
 720          * @param mixed $value
 721          *
 722          * @return float|null
 723          */
 724         private function getPropertySortKeyValue( $value ) {
 725                 if ( is_int( $value ) || is_float( $value ) || is_bool( $value ) ) {
 726                         return floatval( $value );
 727                 }
 728
 729                 return null;
 730         }
 731
 732         /**
 733          * Get an array of interwiki insertions for passing to the DB
 734          * Skips the titles specified by the 2-D array $existing
 735          * @param array $existing
 736          * @return array
 737          */
 738         private function getInterwikiInsertions( $existing = [] ) {
 739                 $arr = [];
 740                 foreach ( $this->mInterwikis as $prefix => $dbkeys ) {
 741                         $diffs = isset( $existing[$prefix] )
 742                                 ? array_diff_key( $dbkeys, $existing[$prefix] )
 743                                 : $dbkeys;
 744
 745                         foreach ( $diffs as $dbk => $id ) {
 746                                 $arr[] = [
 747                                         'iwl_from' => $this->mId,
 748                                         'iwl_prefix' => $prefix,
 749                                         'iwl_title' => $dbk
 750                                 ];
 751                         }
 752                 }
 753
 754                 return $arr;
 755         }
 756
 757         /**
 758          * Given an array of existing links, returns those links which are not in $this
 759          * and thus should be deleted.
 760          * @param array $existing
 761          * @return array
 762          */
 763         private function getLinkDeletions( $existing ) {
 764                 $del = [];
 765                 foreach ( $existing as $ns => $dbkeys ) {
 766                         if ( isset( $this->mLinks[$ns] ) ) {
 767                                 $del[$ns] = array_diff_key( $existing[$ns], $this->mLinks[$ns] );
 768                         } else {
 769                                 $del[$ns] = $existing[$ns];
 770                         }
 771                 }
 772
 773                 return $del;
 774         }
 775
 776         /**
 777          * Given an array of existing templates, returns those templates which are not in $this
 778          * and thus should be deleted.
 779          * @param array $existing
 780          * @return array
 781          */
 782         private function getTemplateDeletions( $existing ) {
 783                 $del = [];
 784                 foreach ( $existing as $ns => $dbkeys ) {
 785                         if ( isset( $this->mTemplates[$ns] ) ) {
 786                                 $del[$ns] = array_diff_key( $existing[$ns], $this->mTemplates[$ns] );
 787                         } else {
 788                                 $del[$ns] = $existing[$ns];
 789                         }
 790                 }
 791
 792                 return $del;
 793         }
 794
 795         /**
 796          * Given an array of existing images, returns those images which are not in $this
 797          * and thus should be deleted.
 798          * @param array $existing
 799          * @return array
 800          */
 801         private function getImageDeletions( $existing ) {
 802                 return array_diff_key( $existing, $this->mImages );
 803         }
 804
 805         /**
 806          * Given an array of existing external links, returns those links which are not
 807          * in $this and thus should be deleted.
 808          * @param array $existing
 809          * @return array
 810          */
 811         private function getExternalDeletions( $existing ) {
 812                 return array_diff_key( $existing, $this->mExternals );
 813         }
 814
 815         /**
 816          * Given an array of existing categories, returns those categories which are not in $this
 817          * and thus should be deleted.
 818          * @param array $existing
 819          * @return array
 820          */
 821         private function getCategoryDeletions( $existing ) {
 822                 return array_diff_assoc( $existing, $this->mCategories );
 823         }
 824
 825         /**
 826          * Given an array of existing interlanguage links, returns those links which are not
 827          * in $this and thus should be deleted.
 828          * @param array $existing
 829          * @return array
 830          */
 831         private function getInterlangDeletions( $existing ) {
 832                 return array_diff_assoc( $existing, $this->mInterlangs );
 833         }
 834
 835         /**
 836          * Get array of properties which should be deleted.
 837          * @param array $existing
 838          * @return array
 839          */
 840         private function getPropertyDeletions( $existing ) {
 841                 return array_diff_assoc( $existing, $this->mProperties );
 842         }
 843
 844         /**
 845          * Given an array of existing interwiki links, returns those links which are not in $this
 846          * and thus should be deleted.
 847          * @param array $existing
 848          * @return array
 849          */
 850         private function getInterwikiDeletions( $existing ) {
 851                 $del = [];
 852                 foreach ( $existing as $prefix => $dbkeys ) {
 853                         if ( isset( $this->mInterwikis[$prefix] ) ) {
 854                                 $del[$prefix] = array_diff_key( $existing[$prefix], $this->mInterwikis[$prefix] );
 855                         } else {
 856                                 $del[$prefix] = $existing[$prefix];
 857                         }
 858                 }
 859
 860                 return $del;
 861         }
 862
 863         /**
 864          * Get an array of existing links, as a 2-D array
 865          *
 866          * @return array
 867          */
 868         private function getExistingLinks() {
 869                 $res = $this->getDB()->select( 'pagelinks', [ 'pl_namespace', 'pl_title' ],
 870                         [ 'pl_from' => $this->mId ], __METHOD__ );
 871                 $arr = [];
 872                 foreach ( $res as $row ) {
 873                         if ( !isset( $arr[$row->pl_namespace] ) ) {
 874                                 $arr[$row->pl_namespace] = [];
 875                         }
 876                         $arr[$row->pl_namespace][$row->pl_title] = 1;
 877                 }
 878
 879                 return $arr;
 880         }
 881
 882         /**
 883          * Get an array of existing templates, as a 2-D array
 884          *
 885          * @return array
 886          */
 887         private function getExistingTemplates() {
 888                 $res = $this->getDB()->select( 'templatelinks', [ 'tl_namespace', 'tl_title' ],
 889                         [ 'tl_from' => $this->mId ], __METHOD__ );
 890                 $arr = [];
 891                 foreach ( $res as $row ) {
 892                         if ( !isset( $arr[$row->tl_namespace] ) ) {
 893                                 $arr[$row->tl_namespace] = [];
 894                         }
 895                         $arr[$row->tl_namespace][$row->tl_title] = 1;
 896                 }
 897
 898                 return $arr;
 899         }
 900
 901         /**
 902          * Get an array of existing images, image names in the keys
 903          *
 904          * @return array
 905          */
 906         private function getExistingImages() {
 907                 $res = $this->getDB()->select( 'imagelinks', [ 'il_to' ],
 908                         [ 'il_from' => $this->mId ], __METHOD__ );
 909                 $arr = [];
 910                 foreach ( $res as $row ) {
 911                         $arr[$row->il_to] = 1;
 912                 }
 913
 914                 return $arr;
 915         }
 916
 917         /**
 918          * Get an array of existing external links, URLs in the keys
 919          *
 920          * @return array
 921          */
 922         private function getExistingExternals() {
 923                 $res = $this->getDB()->select( 'externallinks', [ 'el_to' ],
 924                         [ 'el_from' => $this->mId ], __METHOD__ );
 925                 $arr = [];
 926                 foreach ( $res as $row ) {
 927                         $arr[$row->el_to] = 1;
 928                 }
 929
 930                 return $arr;
 931         }
 932
 933         /**
 934          * Get an array of existing categories, with the name in the key and sort key in the value.
 935          *
 936          * @return array
 937          */
 938         private function getExistingCategories() {
 939                 $res = $this->getDB()->select( 'categorylinks', [ 'cl_to', 'cl_sortkey_prefix' ],
 940                         [ 'cl_from' => $this->mId ], __METHOD__ );
 941                 $arr = [];
 942                 foreach ( $res as $row ) {
 943                         $arr[$row->cl_to] = $row->cl_sortkey_prefix;
 944                 }
 945
 946                 return $arr;
 947         }
 948
 949         /**
 950          * Get an array of existing interlanguage links, with the language code in the key and the
 951          * title in the value.
 952          *
 953          * @return array
 954          */
 955         private function getExistingInterlangs() {
 956                 $res = $this->getDB()->select( 'langlinks', [ 'll_lang', 'll_title' ],
 957                         [ 'll_from' => $this->mId ], __METHOD__ );
 958                 $arr = [];
 959                 foreach ( $res as $row ) {
 960                         $arr[$row->ll_lang] = $row->ll_title;
 961                 }
 962
 963                 return $arr;
 964         }
 965
 966         /**
 967          * Get an array of existing inline interwiki links, as a 2-D array
 968          * @return array [ prefix => [ dbkey => 1 ] ]
 969          */
 970         private function getExistingInterwikis() {
 971                 $res = $this->getDB()->select( 'iwlinks', [ 'iwl_prefix', 'iwl_title' ],
 972                         [ 'iwl_from' => $this->mId ], __METHOD__ );
 973                 $arr = [];
 974                 foreach ( $res as $row ) {
 975                         if ( !isset( $arr[$row->iwl_prefix] ) ) {
 976                                 $arr[$row->iwl_prefix] = [];
 977                         }
 978                         $arr[$row->iwl_prefix][$row->iwl_title] = 1;
 979                 }
 980
 981                 return $arr;
 982         }
 983
 984         /**
 985          * Get an array of existing categories, with the name in the key and sort key in the value.
 986          *
 987          * @return array Array of property names and values
 988          */
 989         private function getExistingProperties() {
 990                 $res = $this->getDB()->select( 'page_props', [ 'pp_propname', 'pp_value' ],
 991                         [ 'pp_page' => $this->mId ], __METHOD__ );
 992                 $arr = [];
 993                 foreach ( $res as $row ) {
 994                         $arr[$row->pp_propname] = $row->pp_value;
 995                 }
 996
 997                 return $arr;
 998         }
 999
1000         /**
1001          * Return the title object of the page being updated
1002          * @return Title
1003          */
1004         public function getTitle() {
1005                 return $this->mTitle;
1006         }
1007
1008         /**
1009          * Returns parser output
1010          * @since 1.19
1011          * @return ParserOutput
1012          */
1013         public function getParserOutput() {
1014                 return $this->mParserOutput;
1015         }
1016
1017         /**
1018          * Return the list of images used as generated by the parser
1019          * @return array
1020          */
1021         public function getImages() {
1022                 return $this->mImages;
1023         }
1024
1025         /**
1026          * Set the revision corresponding to this LinksUpdate
1027          *
1028          * @since 1.27
1029          *
1030          * @param Revision $revision
1031          */
1032         public function setRevision( Revision $revision ) {
1033                 $this->mRevision = $revision;
1034         }
1035
1036         /**
1037          * @since 1.28
1038          * @return null|Revision
1039          */
1040         public function getRevision() {
1041                 return $this->mRevision;
1042         }
1043
1044         /**
1045          * Set the User who triggered this LinksUpdate
1046          *
1047          * @since 1.27
1048          * @param User $user
1049          */
1050         public function setTriggeringUser( User $user ) {
1051                 $this->user = $user;
1052         }
1053
1054         /**
1055          * @since 1.27
1056          * @return null|User
1057          */
1058         public function getTriggeringUser() {
1059                 return $this->user;
1060         }
1061
1062         /**
1063          * Invalidate any necessary link lists related to page property changes
1064          * @param array $changed
1065          */
1066         private function invalidateProperties( $changed ) {
1067                 global $wgPagePropLinkInvalidations;
1068
1069                 foreach ( $changed as $name => $value ) {
1070                         if ( isset( $wgPagePropLinkInvalidations[$name] ) ) {
1071                                 $inv = $wgPagePropLinkInvalidations[$name];
1072                                 if ( !is_array( $inv ) ) {
1073                                         $inv = [ $inv ];
1074                                 }
1075                                 foreach ( $inv as $table ) {
1076                                         DeferredUpdates::addUpdate(
1077                                                 new HTMLCacheUpdate( $this->mTitle, $table, 'page-props' )
1078                                         );
1079                                 }
1080                         }
1081                 }
1082         }
1083
1084         /**
1085          * Fetch page links added by this LinksUpdate.  Only available after the update is complete.
1086          * @since 1.22
1087          * @return null|array Array of Titles
1088          */
1089         public function getAddedLinks() {
1090                 if ( $this->linkInsertions === null ) {
1091                         return null;
1092                 }
1093                 $result = [];
1094                 foreach ( $this->linkInsertions as $insertion ) {
1095                         $result[] = Title::makeTitle( $insertion['pl_namespace'], $insertion['pl_title'] );
1096                 }
1097
1098                 return $result;
1099         }
1100
1101         /**
1102          * Fetch page links removed by this LinksUpdate.  Only available after the update is complete.
1103          * @since 1.22
1104          * @return null|array Array of Titles
1105          */
1106         public function getRemovedLinks() {
1107                 if ( $this->linkDeletions === null ) {
1108                         return null;
1109                 }
1110                 $result = [];
1111                 foreach ( $this->linkDeletions as $ns => $titles ) {
1112                         foreach ( $titles as $title => $unused ) {
1113                                 $result[] = Title::makeTitle( $ns, $title );
1114                         }
1115                 }
1116
1117                 return $result;
1118         }
1119
1120         /**
1121          * Fetch external links added by this LinksUpdate. Only available after
1122          * the update is complete.
1123          * @since 1.33
1124          * @return null|array Array of Strings
1125          */
1126         public function getAddedExternalLinks() {
1127                 if ( $this->externalLinkInsertions === null ) {
1128                         return null;
1129                 }
1130                 $result = [];
1131                 foreach ( $this->externalLinkInsertions as $key => $value ) {
1132                         $result[] = $value['el_to'];
1133                 }
1134                 return $result;
1135         }
1136
1137         /**
1138          * Fetch external links removed by this LinksUpdate. Only available after
1139          * the update is complete.
1140          * @since 1.33
1141          * @return null|array Array of Strings
1142          */
1143         public function getRemovedExternalLinks() {
1144                 if ( $this->externalLinkDeletions === null ) {
1145                         return null;
1146                 }
1147                 return array_keys( $this->externalLinkDeletions );
1148         }
1149
1150         /**
1151          * Fetch page properties added by this LinksUpdate.
1152          * Only available after the update is complete.
1153          * @since 1.28
1154          * @return null|array
1155          */
1156         public function getAddedProperties() {
1157                 return $this->propertyInsertions;
1158         }
1159
1160         /**
1161          * Fetch page properties removed by this LinksUpdate.
1162          * Only available after the update is complete.
1163          * @since 1.28
1164          * @return null|array
1165          */
1166         public function getRemovedProperties() {
1167                 return $this->propertyDeletions;
1168         }
1169
1170         /**
1171          * Update links table freshness
1172          */
1173         private function updateLinksTimestamp() {
1174                 if ( $this->mId ) {
1175                         // The link updates made here only reflect the freshness of the parser output
1176                         $timestamp = $this->mParserOutput->getCacheTime();
1177                         $this->getDB()->update( 'page',
1178                                 [ 'page_links_updated' => $this->getDB()->timestamp( $timestamp ) ],
1179                                 [ 'page_id' => $this->mId ],
1180                                 __METHOD__
1181                         );
1182                 }
1183         }
1184
1185         /**
1186          * @return IDatabase
1187          */
1188         protected function getDB() {
1189                 if ( !$this->db ) {
1190                         $this->db = wfGetDB( DB_MASTER );
1191                 }
1192
1193                 return $this->db;
1194         }
1195
1196         public function getAsJobSpecification() {
1197                 if ( $this->user ) {
1198                         $userInfo = [
1199                                 'userId' => $this->user->getId(),
1200                                 'userName' => $this->user->getName(),
1201                         ];
1202                 } else {
1203                         $userInfo = false;
1204                 }
1205
1206                 if ( $this->mRevision ) {
1207                         $triggeringRevisionId = $this->mRevision->getId();
1208                 } else {
1209                         $triggeringRevisionId = false;
1210                 }
1211
1212                 return [
1213                         'domain' => $this->getDB()->getDomainID(),
1214                         'job'  => new JobSpecification(
1215                                 'refreshLinksPrioritized',
1216                                 [
1217                                         // Reuse the parser cache if it was saved
1218                                         'rootJobTimestamp' => $this->mParserOutput->getCacheTime(),
1219                                         'useRecursiveLinksUpdate' => $this->mRecursive,
1220                                         'triggeringUser' => $userInfo,
1221                                         'triggeringRevisionId' => $triggeringRevisionId,
1222                                         'causeAction' => $this->getCauseAction(),
1223                                         'causeAgent' => $this->getCauseAgent()
1224                                 ],
1225                                 [ 'removeDuplicates' => true ],
1226                                 $this->getTitle()
1227                         )
1228                 ];
1229         }
1230 }