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