includes/deferred/LinksUpdate.php

   1 <?php
   2 /**
   3  * Updater for link tracking tables after a page edit.
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  */
  22
  23 use MediaWiki\MediaWikiServices;
  24
  25 /**
  26  * Class the manages updates of *_link tables as well as similar extension-managed tables
  27  *
  28  * @note: LinksUpdate is managed by DeferredUpdates::execute(). Do not run this in a transaction.
  29  *
  30  * See docs/deferred.txt
  31  */
  32 class LinksUpdate extends DataUpdate implements EnqueueableDataUpdate {
  33         // @todo make members protected, but make sure extensions don't break
  34
  35         /** @var int Page ID of the article linked from */
  36         public $mId;
  37
  38         /** @var Title Title object of the article linked from */
  39         public $mTitle;
  40
  41         /** @var ParserOutput */
  42         public $mParserOutput;
  43
  44         /** @var array Map of title strings to IDs for the links in the document */
  45         public $mLinks;
  46
  47         /** @var array DB keys of the images used, in the array key only */
  48         public $mImages;
  49
  50         /** @var array Map of title strings to IDs for the template references, including broken ones */
  51         public $mTemplates;
  52
  53         /** @var array URLs of external links, array key only */
  54         public $mExternals;
  55
  56         /** @var array Map of category names to sort keys */
  57         public $mCategories;
  58
  59         /** @var array Map of language codes to titles */
  60         public $mInterlangs;
  61
  62         /** @var array 2-D map of (prefix => DBK => 1) */
  63         public $mInterwikis;
  64
  65         /** @var array Map of arbitrary name to value */
  66         public $mProperties;
  67
  68         /** @var bool Whether to queue jobs for recursive updates */
  69         public $mRecursive;
  70
  71         /** @var Revision Revision for which this update has been triggered */
  72         private $mRevision;
  73
  74         /**
  75          * @var null|array Added links if calculated.
  76          */
  77         private $linkInsertions = null;
  78
  79         /**
  80          * @var null|array Deleted links if calculated.
  81          */
  82         private $linkDeletions = null;
  83
  84         /**
  85          * @var null|array Added properties if calculated.
  86          */
  87         private $propertyInsertions = null;
  88
  89         /**
  90          * @var null|array Deleted properties if calculated.
  91          */
  92         private $propertyDeletions = null;
  93
  94         /**
  95          * @var User|null
  96          */
  97         private $user;
  98
  99         /** @var IDatabase */
 100         private $db;
 101
 102         /**
 103          * Constructor
 104          *
 105          * @param Title $title Title of the page we're updating
 106          * @param ParserOutput $parserOutput Output from a full parse of this page
 107          * @param bool $recursive Queue jobs for recursive updates?
 108          * @throws MWException
 109          */
 110         function __construct( Title $title, ParserOutput $parserOutput, $recursive = true ) {
 111                 parent::__construct();
 112
 113                 $this->mTitle = $title;
 114                 $this->mId = $title->getArticleID( Title::GAID_FOR_UPDATE );
 115
 116                 if ( !$this->mId ) {
 117                         throw new InvalidArgumentException(
 118                                 "The Title object yields no ID. Perhaps the page doesn't exist?"
 119                         );
 120                 }
 121
 122                 $this->mParserOutput = $parserOutput;
 123
 124                 $this->mLinks = $parserOutput->getLinks();
 125                 $this->mImages = $parserOutput->getImages();
 126                 $this->mTemplates = $parserOutput->getTemplates();
 127                 $this->mExternals = $parserOutput->getExternalLinks();
 128                 $this->mCategories = $parserOutput->getCategories();
 129                 $this->mProperties = $parserOutput->getProperties();
 130                 $this->mInterwikis = $parserOutput->getInterwikiLinks();
 131
 132                 # Convert the format of the interlanguage links
 133                 # I didn't want to change it in the ParserOutput, because that array is passed all
 134                 # the way back to the skin, so either a skin API break would be required, or an
 135                 # inefficient back-conversion.
 136                 $ill = $parserOutput->getLanguageLinks();
 137                 $this->mInterlangs = [];
 138                 foreach ( $ill as $link ) {
 139                         list( $key, $title ) = explode( ':', $link, 2 );
 140                         $this->mInterlangs[$key] = $title;
 141                 }
 142
 143                 foreach ( $this->mCategories as &$sortkey ) {
 144                         # If the sortkey is longer then 255 bytes,
 145                         # it truncated by DB, and then doesn't get
 146                         # matched when comparing existing vs current
 147                         # categories, causing bug 25254.
 148                         # Also. substr behaves weird when given "".
 149                         if ( $sortkey !== '' ) {
 150                                 $sortkey = substr( $sortkey, 0, 255 );
 151                         }
 152                 }
 153
 154                 $this->mRecursive = $recursive;
 155
 156                 Hooks::run( 'LinksUpdateConstructed', [ &$this ] );
 157         }
 158
 159         /**
 160          * Update link tables with outgoing links from an updated article
 161          *
 162          * @note: this is managed by DeferredUpdates::execute(). Do not run this in a transaction.
 163          */
 164         public function doUpdate() {
 165                 if ( $this->ticket ) {
 166                         // Make sure all links update threads see the changes of each other.
 167                         // This handles the case when updates have to batched into several COMMITs.
 168                         $scopedLock = self::acquirePageLock( $this->getDB(), $this->mId );
 169                 }
 170
 171                 Hooks::run( 'LinksUpdate', [ &$this ] );
 172                 $this->doIncrementalUpdate();
 173
 174                 // Commit and release the lock (if set)
 175                 ScopedCallback::consume( $scopedLock );
 176                 // Run post-commit hooks without DBO_TRX
 177                 $this->getDB()->onTransactionIdle(
 178                         function () {
 179                                 Hooks::run( 'LinksUpdateComplete', [ &$this, $this->ticket ] );
 180                         },
 181                         __METHOD__
 182                 );
 183         }
 184
 185         /**
 186          * Acquire a lock for performing link table updates for a page on a DB
 187          *
 188          * @param IDatabase $dbw
 189          * @param integer $pageId
 190          * @param string $why One of (job, atomicity)
 191          * @return ScopedCallback
 192          * @throws RuntimeException
 193          * @since 1.27
 194          */
 195         public static function acquirePageLock( IDatabase $dbw, $pageId, $why = 'atomicity' ) {
 196                 $key = "LinksUpdate:$why:pageid:$pageId";
 197                 $scopedLock = $dbw->getScopedLockAndFlush( $key, __METHOD__, 15 );
 198                 if ( !$scopedLock ) {
 199                         throw new RuntimeException( "Could not acquire lock '$key'." );
 200                 }
 201
 202                 return $scopedLock;
 203         }
 204
 205         protected function doIncrementalUpdate() {
 206                 # Page links
 207                 $existing = $this->getExistingLinks();
 208                 $this->linkDeletions = $this->getLinkDeletions( $existing );
 209                 $this->linkInsertions = $this->getLinkInsertions( $existing );
 210                 $this->incrTableUpdate( 'pagelinks', 'pl', $this->linkDeletions, $this->linkInsertions );
 211
 212                 # Image links
 213                 $existing = $this->getExistingImages();
 214                 $imageDeletes = $this->getImageDeletions( $existing );
 215                 $this->incrTableUpdate( 'imagelinks', 'il', $imageDeletes,
 216                         $this->getImageInsertions( $existing ) );
 217
 218                 # Invalidate all image description pages which had links added or removed
 219                 $imageUpdates = $imageDeletes + array_diff_key( $this->mImages, $existing );
 220                 $this->invalidateImageDescriptions( $imageUpdates );
 221
 222                 # External links
 223                 $existing = $this->getExistingExternals();
 224                 $this->incrTableUpdate( 'externallinks', 'el', $this->getExternalDeletions( $existing ),
 225                         $this->getExternalInsertions( $existing ) );
 226
 227                 # Language links
 228                 $existing = $this->getExistingInterlangs();
 229                 $this->incrTableUpdate( 'langlinks', 'll', $this->getInterlangDeletions( $existing ),
 230                         $this->getInterlangInsertions( $existing ) );
 231
 232                 # Inline interwiki links
 233                 $existing = $this->getExistingInterwikis();
 234                 $this->incrTableUpdate( 'iwlinks', 'iwl', $this->getInterwikiDeletions( $existing ),
 235                         $this->getInterwikiInsertions( $existing ) );
 236
 237                 # Template links
 238                 $existing = $this->getExistingTemplates();
 239                 $this->incrTableUpdate( 'templatelinks', 'tl', $this->getTemplateDeletions( $existing ),
 240                         $this->getTemplateInsertions( $existing ) );
 241
 242                 # Category links
 243                 $existing = $this->getExistingCategories();
 244                 $categoryDeletes = $this->getCategoryDeletions( $existing );
 245                 $this->incrTableUpdate( 'categorylinks', 'cl', $categoryDeletes,
 246                         $this->getCategoryInsertions( $existing ) );
 247
 248                 # Invalidate all categories which were added, deleted or changed (set symmetric difference)
 249                 $categoryInserts = array_diff_assoc( $this->mCategories, $existing );
 250                 $categoryUpdates = $categoryInserts + $categoryDeletes;
 251                 $this->invalidateCategories( $categoryUpdates );
 252                 $this->updateCategoryCounts( $categoryInserts, $categoryDeletes );
 253
 254                 # Page properties
 255                 $existing = $this->getExistingProperties();
 256                 $this->propertyDeletions = $this->getPropertyDeletions( $existing );
 257                 $this->incrTableUpdate( 'page_props', 'pp', $this->propertyDeletions,
 258                         $this->getPropertyInsertions( $existing ) );
 259
 260                 # Invalidate the necessary pages
 261                 $this->propertyInsertions = array_diff_assoc( $this->mProperties, $existing );
 262                 $changed = $this->propertyDeletions + $this->propertyInsertions;
 263                 $this->invalidateProperties( $changed );
 264
 265                 # Refresh links of all pages including this page
 266                 # This will be in a separate transaction
 267                 if ( $this->mRecursive ) {
 268                         $this->queueRecursiveJobs();
 269                 }
 270
 271                 # Update the links table freshness for this title
 272                 $this->updateLinksTimestamp();
 273         }
 274
 275         /**
 276          * Queue recursive jobs for this page
 277          *
 278          * Which means do LinksUpdate on all pages that include the current page,
 279          * using the job queue.
 280          */
 281         protected function queueRecursiveJobs() {
 282                 self::queueRecursiveJobsForTable( $this->mTitle, 'templatelinks' );
 283                 if ( $this->mTitle->getNamespace() == NS_FILE ) {
 284                         // Process imagelinks in case the title is or was a redirect
 285                         self::queueRecursiveJobsForTable( $this->mTitle, 'imagelinks' );
 286                 }
 287
 288                 $bc = $this->mTitle->getBacklinkCache();
 289                 // Get jobs for cascade-protected backlinks for a high priority queue.
 290                 // If meta-templates change to using a new template, the new template
 291                 // should be implicitly protected as soon as possible, if applicable.
 292                 // These jobs duplicate a subset of the above ones, but can run sooner.
 293                 // Which ever runs first generally no-ops the other one.
 294                 $jobs = [];
 295                 foreach ( $bc->getCascadeProtectedLinks() as $title ) {
 296                         $jobs[] = RefreshLinksJob::newPrioritized( $title, [] );
 297                 }
 298                 JobQueueGroup::singleton()->push( $jobs );
 299         }
 300
 301         /**
 302          * Queue a RefreshLinks job for any table.
 303          *
 304          * @param Title $title Title to do job for
 305          * @param string $table Table to use (e.g. 'templatelinks')
 306          */
 307         public static function queueRecursiveJobsForTable( Title $title, $table ) {
 308                 if ( $title->getBacklinkCache()->hasLinks( $table ) ) {
 309                         $job = new RefreshLinksJob(
 310                                 $title,
 311                                 [
 312                                         'table' => $table,
 313                                         'recursive' => true,
 314                                 ] + Job::newRootJobParams( // "overall" refresh links job info
 315                                         "refreshlinks:{$table}:{$title->getPrefixedText()}"
 316                                 )
 317                         );
 318
 319                         JobQueueGroup::singleton()->push( $job );
 320                 }
 321         }
 322
 323         /**
 324          * @param array $cats
 325          */
 326         function invalidateCategories( $cats ) {
 327                 PurgeJobUtils::invalidatePages( $this->getDB(), NS_CATEGORY, array_keys( $cats ) );
 328         }
 329
 330         /**
 331          * Update all the appropriate counts in the category table.
 332          * @param array $added Associative array of category name => sort key
 333          * @param array $deleted Associative array of category name => sort key
 334          */
 335         function updateCategoryCounts( $added, $deleted ) {
 336                 $a = WikiPage::factory( $this->mTitle );
 337                 $a->updateCategoryCounts(
 338                         array_keys( $added ), array_keys( $deleted )
 339                 );
 340         }
 341
 342         /**
 343          * @param array $images
 344          */
 345         function invalidateImageDescriptions( $images ) {
 346                 PurgeJobUtils::invalidatePages( $this->getDB(), NS_FILE, array_keys( $images ) );
 347         }
 348
 349         /**
 350          * Update a table by doing a delete query then an insert query
 351          * @param string $table Table name
 352          * @param string $prefix Field name prefix
 353          * @param array $deletions
 354          * @param array $insertions Rows to insert
 355          */
 356         private function incrTableUpdate( $table, $prefix, $deletions, $insertions ) {
 357                 $services = MediaWikiServices::getInstance();
 358                 $bSize = $services->getMainConfig()->get( 'UpdateRowsPerQuery' );
 359                 $factory = $services->getDBLoadBalancerFactory();
 360
 361                 if ( $table === 'page_props' ) {
 362                         $fromField = 'pp_page';
 363                 } else {
 364                         $fromField = "{$prefix}_from";
 365                 }
 366
 367                 $deleteWheres = []; // list of WHERE clause arrays for each DB delete() call
 368                 if ( $table === 'pagelinks' || $table === 'templatelinks' || $table === 'iwlinks' ) {
 369                         $baseKey =  ( $table === 'iwlinks' ) ? 'iwl_prefix' : "{$prefix}_namespace";
 370
 371                         $curBatchSize = 0;
 372                         $curDeletionBatch = [];
 373                         $deletionBatches = [];
 374                         foreach ( $deletions as $ns => $dbKeys ) {
 375                                 foreach ( $dbKeys as $dbKey => $unused ) {
 376                                         $curDeletionBatch[$ns][$dbKey] = 1;
 377                                         if ( ++$curBatchSize >= $bSize ) {
 378                                                 $deletionBatches[] = $curDeletionBatch;
 379                                                 $curDeletionBatch = [];
 380                                                 $curBatchSize = 0;
 381                                         }
 382                                 }
 383                         }
 384                         if ( $curDeletionBatch ) {
 385                                 $deletionBatches[] = $curDeletionBatch;
 386                         }
 387
 388                         foreach ( $deletionBatches as $deletionBatch ) {
 389                                 $deleteWheres[] = [
 390                                         $fromField => $this->mId,
 391                                         $this->getDB()->makeWhereFrom2d( $deletionBatch, $baseKey, "{$prefix}_title" )
 392                                 ];
 393                         }
 394                 } else {
 395                         if ( $table === 'langlinks' ) {
 396                                 $toField = 'll_lang';
 397                         } elseif ( $table === 'page_props' ) {
 398                                 $toField = 'pp_propname';
 399                         } else {
 400                                 $toField = $prefix . '_to';
 401                         }
 402
 403                         $deletionBatches = array_chunk( array_keys( $deletions ), $bSize );
 404                         foreach ( $deletionBatches as $deletionBatch ) {
 405                                 $deleteWheres[] = [ $fromField => $this->mId, $toField => $deletionBatch ];
 406                         }
 407                 }
 408
 409                 foreach ( $deleteWheres as $deleteWhere ) {
 410                         $this->getDB()->delete( $table, $deleteWhere, __METHOD__ );
 411                         $factory->commitAndWaitForReplication(
 412                                 __METHOD__, $this->ticket, [ 'wiki' => $this->getDB()->getWikiID() ]
 413                         );
 414                 }
 415
 416                 $insertBatches = array_chunk( $insertions, $bSize );
 417                 foreach ( $insertBatches as $insertBatch ) {
 418                         $this->getDB()->insert( $table, $insertBatch, __METHOD__, 'IGNORE' );
 419                         $factory->commitAndWaitForReplication(
 420                                 __METHOD__, $this->ticket, [ 'wiki' => $this->getDB()->getWikiID() ]
 421                         );
 422                 }
 423
 424                 if ( count( $insertions ) ) {
 425                         Hooks::run( 'LinksUpdateAfterInsert', [ $this, $table, $insertions ] );
 426                 }
 427         }
 428
 429         /**
 430          * Get an array of pagelinks insertions for passing to the DB
 431          * Skips the titles specified by the 2-D array $existing
 432          * @param array $existing
 433          * @return array
 434          */
 435         private function getLinkInsertions( $existing = [] ) {
 436                 $arr = [];
 437                 foreach ( $this->mLinks as $ns => $dbkeys ) {
 438                         $diffs = isset( $existing[$ns] )
 439                                 ? array_diff_key( $dbkeys, $existing[$ns] )
 440                                 : $dbkeys;
 441                         foreach ( $diffs as $dbk => $id ) {
 442                                 $arr[] = [
 443                                         'pl_from' => $this->mId,
 444                                         'pl_from_namespace' => $this->mTitle->getNamespace(),
 445                                         'pl_namespace' => $ns,
 446                                         'pl_title' => $dbk
 447                                 ];
 448                         }
 449                 }
 450
 451                 return $arr;
 452         }
 453
 454         /**
 455          * Get an array of template insertions. Like getLinkInsertions()
 456          * @param array $existing
 457          * @return array
 458          */
 459         private function getTemplateInsertions( $existing = [] ) {
 460                 $arr = [];
 461                 foreach ( $this->mTemplates as $ns => $dbkeys ) {
 462                         $diffs = isset( $existing[$ns] ) ? array_diff_key( $dbkeys, $existing[$ns] ) : $dbkeys;
 463                         foreach ( $diffs as $dbk => $id ) {
 464                                 $arr[] = [
 465                                         'tl_from' => $this->mId,
 466                                         'tl_from_namespace' => $this->mTitle->getNamespace(),
 467                                         'tl_namespace' => $ns,
 468                                         'tl_title' => $dbk
 469                                 ];
 470                         }
 471                 }
 472
 473                 return $arr;
 474         }
 475
 476         /**
 477          * Get an array of image insertions
 478          * Skips the names specified in $existing
 479          * @param array $existing
 480          * @return array
 481          */
 482         private function getImageInsertions( $existing = [] ) {
 483                 $arr = [];
 484                 $diffs = array_diff_key( $this->mImages, $existing );
 485                 foreach ( $diffs as $iname => $dummy ) {
 486                         $arr[] = [
 487                                 'il_from' => $this->mId,
 488                                 'il_from_namespace' => $this->mTitle->getNamespace(),
 489                                 'il_to' => $iname
 490                         ];
 491                 }
 492
 493                 return $arr;
 494         }
 495
 496         /**
 497          * Get an array of externallinks insertions. Skips the names specified in $existing
 498          * @param array $existing
 499          * @return array
 500          */
 501         private function getExternalInsertions( $existing = [] ) {
 502                 $arr = [];
 503                 $diffs = array_diff_key( $this->mExternals, $existing );
 504                 foreach ( $diffs as $url => $dummy ) {
 505                         foreach ( wfMakeUrlIndexes( $url ) as $index ) {
 506                                 $arr[] = [
 507                                         'el_id' => $this->getDB()->nextSequenceValue( 'externallinks_el_id_seq' ),
 508                                         'el_from' => $this->mId,
 509                                         'el_to' => $url,
 510                                         'el_index' => $index,
 511                                 ];
 512                         }
 513                 }
 514
 515                 return $arr;
 516         }
 517
 518         /**
 519          * Get an array of category insertions
 520          *
 521          * @param array $existing Mapping existing category names to sort keys. If both
 522          * match a link in $this, the link will be omitted from the output
 523          *
 524          * @return array
 525          */
 526         private function getCategoryInsertions( $existing = [] ) {
 527                 global $wgContLang, $wgCategoryCollation;
 528                 $diffs = array_diff_assoc( $this->mCategories, $existing );
 529                 $arr = [];
 530                 foreach ( $diffs as $name => $prefix ) {
 531                         $nt = Title::makeTitleSafe( NS_CATEGORY, $name );
 532                         $wgContLang->findVariantLink( $name, $nt, true );
 533
 534                         if ( $this->mTitle->getNamespace() == NS_CATEGORY ) {
 535                                 $type = 'subcat';
 536                         } elseif ( $this->mTitle->getNamespace() == NS_FILE ) {
 537                                 $type = 'file';
 538                         } else {
 539                                 $type = 'page';
 540                         }
 541
 542                         # Treat custom sortkeys as a prefix, so that if multiple
 543                         # things are forced to sort as '*' or something, they'll
 544                         # sort properly in the category rather than in page_id
 545                         # order or such.
 546                         $sortkey = Collation::singleton()->getSortKey(
 547                                 $this->mTitle->getCategorySortkey( $prefix ) );
 548
 549                         $arr[] = [
 550                                 'cl_from' => $this->mId,
 551                                 'cl_to' => $name,
 552                                 'cl_sortkey' => $sortkey,
 553                                 'cl_timestamp' => $this->getDB()->timestamp(),
 554                                 'cl_sortkey_prefix' => $prefix,
 555                                 'cl_collation' => $wgCategoryCollation,
 556                                 'cl_type' => $type,
 557                         ];
 558                 }
 559
 560                 return $arr;
 561         }
 562
 563         /**
 564          * Get an array of interlanguage link insertions
 565          *
 566          * @param array $existing Mapping existing language codes to titles
 567          *
 568          * @return array
 569          */
 570         private function getInterlangInsertions( $existing = [] ) {
 571                 $diffs = array_diff_assoc( $this->mInterlangs, $existing );
 572                 $arr = [];
 573                 foreach ( $diffs as $lang => $title ) {
 574                         $arr[] = [
 575                                 'll_from' => $this->mId,
 576                                 'll_lang' => $lang,
 577                                 'll_title' => $title
 578                         ];
 579                 }
 580
 581                 return $arr;
 582         }
 583
 584         /**
 585          * Get an array of page property insertions
 586          * @param array $existing
 587          * @return array
 588          */
 589         function getPropertyInsertions( $existing = [] ) {
 590                 $diffs = array_diff_assoc( $this->mProperties, $existing );
 591
 592                 $arr = [];
 593                 foreach ( array_keys( $diffs ) as $name ) {
 594                         $arr[] = $this->getPagePropRowData( $name );
 595                 }
 596
 597                 return $arr;
 598         }
 599
 600         /**
 601          * Returns an associative array to be used for inserting a row into
 602          * the page_props table. Besides the given property name, this will
 603          * include the page id from $this->mId and any property value from
 604          * $this->mProperties.
 605          *
 606          * The array returned will include the pp_sortkey field if this
 607          * is present in the database (as indicated by $wgPagePropsHaveSortkey).
 608          * The sortkey value is currently determined by getPropertySortKeyValue().
 609          *
 610          * @note this assumes that $this->mProperties[$prop] is defined.
 611          *
 612          * @param string $prop The name of the property.
 613          *
 614          * @return array
 615          */
 616         private function getPagePropRowData( $prop ) {
 617                 global $wgPagePropsHaveSortkey;
 618
 619                 $value = $this->mProperties[$prop];
 620
 621                 $row = [
 622                         'pp_page' => $this->mId,
 623                         'pp_propname' => $prop,
 624                         'pp_value' => $value,
 625                 ];
 626
 627                 if ( $wgPagePropsHaveSortkey ) {
 628                         $row['pp_sortkey'] = $this->getPropertySortKeyValue( $value );
 629                 }
 630
 631                 return $row;
 632         }
 633
 634         /**
 635          * Determines the sort key for the given property value.
 636          * This will return $value if it is a float or int,
 637          * 1 or resp. 0 if it is a bool, and null otherwise.
 638          *
 639          * @note In the future, we may allow the sortkey to be specified explicitly
 640          *       in ParserOutput::setProperty.
 641          *
 642          * @param mixed $value
 643          *
 644          * @return float|null
 645          */
 646         private function getPropertySortKeyValue( $value ) {
 647                 if ( is_int( $value ) || is_float( $value ) || is_bool( $value ) ) {
 648                         return floatval( $value );
 649                 }
 650
 651                 return null;
 652         }
 653
 654         /**
 655          * Get an array of interwiki insertions for passing to the DB
 656          * Skips the titles specified by the 2-D array $existing
 657          * @param array $existing
 658          * @return array
 659          */
 660         private function getInterwikiInsertions( $existing = [] ) {
 661                 $arr = [];
 662                 foreach ( $this->mInterwikis as $prefix => $dbkeys ) {
 663                         $diffs = isset( $existing[$prefix] )
 664                                 ? array_diff_key( $dbkeys, $existing[$prefix] )
 665                                 : $dbkeys;
 666
 667                         foreach ( $diffs as $dbk => $id ) {
 668                                 $arr[] = [
 669                                         'iwl_from' => $this->mId,
 670                                         'iwl_prefix' => $prefix,
 671                                         'iwl_title' => $dbk
 672                                 ];
 673                         }
 674                 }
 675
 676                 return $arr;
 677         }
 678
 679         /**
 680          * Given an array of existing links, returns those links which are not in $this
 681          * and thus should be deleted.
 682          * @param array $existing
 683          * @return array
 684          */
 685         private function getLinkDeletions( $existing ) {
 686                 $del = [];
 687                 foreach ( $existing as $ns => $dbkeys ) {
 688                         if ( isset( $this->mLinks[$ns] ) ) {
 689                                 $del[$ns] = array_diff_key( $existing[$ns], $this->mLinks[$ns] );
 690                         } else {
 691                                 $del[$ns] = $existing[$ns];
 692                         }
 693                 }
 694
 695                 return $del;
 696         }
 697
 698         /**
 699          * Given an array of existing templates, returns those templates which are not in $this
 700          * and thus should be deleted.
 701          * @param array $existing
 702          * @return array
 703          */
 704         private function getTemplateDeletions( $existing ) {
 705                 $del = [];
 706                 foreach ( $existing as $ns => $dbkeys ) {
 707                         if ( isset( $this->mTemplates[$ns] ) ) {
 708                                 $del[$ns] = array_diff_key( $existing[$ns], $this->mTemplates[$ns] );
 709                         } else {
 710                                 $del[$ns] = $existing[$ns];
 711                         }
 712                 }
 713
 714                 return $del;
 715         }
 716
 717         /**
 718          * Given an array of existing images, returns those images which are not in $this
 719          * and thus should be deleted.
 720          * @param array $existing
 721          * @return array
 722          */
 723         private function getImageDeletions( $existing ) {
 724                 return array_diff_key( $existing, $this->mImages );
 725         }
 726
 727         /**
 728          * Given an array of existing external links, returns those links which are not
 729          * in $this and thus should be deleted.
 730          * @param array $existing
 731          * @return array
 732          */
 733         private function getExternalDeletions( $existing ) {
 734                 return array_diff_key( $existing, $this->mExternals );
 735         }
 736
 737         /**
 738          * Given an array of existing categories, returns those categories which are not in $this
 739          * and thus should be deleted.
 740          * @param array $existing
 741          * @return array
 742          */
 743         private function getCategoryDeletions( $existing ) {
 744                 return array_diff_assoc( $existing, $this->mCategories );
 745         }
 746
 747         /**
 748          * Given an array of existing interlanguage links, returns those links which are not
 749          * in $this and thus should be deleted.
 750          * @param array $existing
 751          * @return array
 752          */
 753         private function getInterlangDeletions( $existing ) {
 754                 return array_diff_assoc( $existing, $this->mInterlangs );
 755         }
 756
 757         /**
 758          * Get array of properties which should be deleted.
 759          * @param array $existing
 760          * @return array
 761          */
 762         function getPropertyDeletions( $existing ) {
 763                 return array_diff_assoc( $existing, $this->mProperties );
 764         }
 765
 766         /**
 767          * Given an array of existing interwiki links, returns those links which are not in $this
 768          * and thus should be deleted.
 769          * @param array $existing
 770          * @return array
 771          */
 772         private function getInterwikiDeletions( $existing ) {
 773                 $del = [];
 774                 foreach ( $existing as $prefix => $dbkeys ) {
 775                         if ( isset( $this->mInterwikis[$prefix] ) ) {
 776                                 $del[$prefix] = array_diff_key( $existing[$prefix], $this->mInterwikis[$prefix] );
 777                         } else {
 778                                 $del[$prefix] = $existing[$prefix];
 779                         }
 780                 }
 781
 782                 return $del;
 783         }
 784
 785         /**
 786          * Get an array of existing links, as a 2-D array
 787          *
 788          * @return array
 789          */
 790         private function getExistingLinks() {
 791                 $res = $this->getDB()->select( 'pagelinks', [ 'pl_namespace', 'pl_title' ],
 792                         [ 'pl_from' => $this->mId ], __METHOD__ );
 793                 $arr = [];
 794                 foreach ( $res as $row ) {
 795                         if ( !isset( $arr[$row->pl_namespace] ) ) {
 796                                 $arr[$row->pl_namespace] = [];
 797                         }
 798                         $arr[$row->pl_namespace][$row->pl_title] = 1;
 799                 }
 800
 801                 return $arr;
 802         }
 803
 804         /**
 805          * Get an array of existing templates, as a 2-D array
 806          *
 807          * @return array
 808          */
 809         private function getExistingTemplates() {
 810                 $res = $this->getDB()->select( 'templatelinks', [ 'tl_namespace', 'tl_title' ],
 811                         [ 'tl_from' => $this->mId ], __METHOD__ );
 812                 $arr = [];
 813                 foreach ( $res as $row ) {
 814                         if ( !isset( $arr[$row->tl_namespace] ) ) {
 815                                 $arr[$row->tl_namespace] = [];
 816                         }
 817                         $arr[$row->tl_namespace][$row->tl_title] = 1;
 818                 }
 819
 820                 return $arr;
 821         }
 822
 823         /**
 824          * Get an array of existing images, image names in the keys
 825          *
 826          * @return array
 827          */
 828         private function getExistingImages() {
 829                 $res = $this->getDB()->select( 'imagelinks', [ 'il_to' ],
 830                         [ 'il_from' => $this->mId ], __METHOD__ );
 831                 $arr = [];
 832                 foreach ( $res as $row ) {
 833                         $arr[$row->il_to] = 1;
 834                 }
 835
 836                 return $arr;
 837         }
 838
 839         /**
 840          * Get an array of existing external links, URLs in the keys
 841          *
 842          * @return array
 843          */
 844         private function getExistingExternals() {
 845                 $res = $this->getDB()->select( 'externallinks', [ 'el_to' ],
 846                         [ 'el_from' => $this->mId ], __METHOD__ );
 847                 $arr = [];
 848                 foreach ( $res as $row ) {
 849                         $arr[$row->el_to] = 1;
 850                 }
 851
 852                 return $arr;
 853         }
 854
 855         /**
 856          * Get an array of existing categories, with the name in the key and sort key in the value.
 857          *
 858          * @return array
 859          */
 860         private function getExistingCategories() {
 861                 $res = $this->getDB()->select( 'categorylinks', [ 'cl_to', 'cl_sortkey_prefix' ],
 862                         [ 'cl_from' => $this->mId ], __METHOD__ );
 863                 $arr = [];
 864                 foreach ( $res as $row ) {
 865                         $arr[$row->cl_to] = $row->cl_sortkey_prefix;
 866                 }
 867
 868                 return $arr;
 869         }
 870
 871         /**
 872          * Get an array of existing interlanguage links, with the language code in the key and the
 873          * title in the value.
 874          *
 875          * @return array
 876          */
 877         private function getExistingInterlangs() {
 878                 $res = $this->getDB()->select( 'langlinks', [ 'll_lang', 'll_title' ],
 879                         [ 'll_from' => $this->mId ], __METHOD__ );
 880                 $arr = [];
 881                 foreach ( $res as $row ) {
 882                         $arr[$row->ll_lang] = $row->ll_title;
 883                 }
 884
 885                 return $arr;
 886         }
 887
 888         /**
 889          * Get an array of existing inline interwiki links, as a 2-D array
 890          * @return array (prefix => array(dbkey => 1))
 891          */
 892         private function getExistingInterwikis() {
 893                 $res = $this->getDB()->select( 'iwlinks', [ 'iwl_prefix', 'iwl_title' ],
 894                         [ 'iwl_from' => $this->mId ], __METHOD__ );
 895                 $arr = [];
 896                 foreach ( $res as $row ) {
 897                         if ( !isset( $arr[$row->iwl_prefix] ) ) {
 898                                 $arr[$row->iwl_prefix] = [];
 899                         }
 900                         $arr[$row->iwl_prefix][$row->iwl_title] = 1;
 901                 }
 902
 903                 return $arr;
 904         }
 905
 906         /**
 907          * Get an array of existing categories, with the name in the key and sort key in the value.
 908          *
 909          * @return array Array of property names and values
 910          */
 911         private function getExistingProperties() {
 912                 $res = $this->getDB()->select( 'page_props', [ 'pp_propname', 'pp_value' ],
 913                         [ 'pp_page' => $this->mId ], __METHOD__ );
 914                 $arr = [];
 915                 foreach ( $res as $row ) {
 916                         $arr[$row->pp_propname] = $row->pp_value;
 917                 }
 918
 919                 return $arr;
 920         }
 921
 922         /**
 923          * Return the title object of the page being updated
 924          * @return Title
 925          */
 926         public function getTitle() {
 927                 return $this->mTitle;
 928         }
 929
 930         /**
 931          * Returns parser output
 932          * @since 1.19
 933          * @return ParserOutput
 934          */
 935         public function getParserOutput() {
 936                 return $this->mParserOutput;
 937         }
 938
 939         /**
 940          * Return the list of images used as generated by the parser
 941          * @return array
 942          */
 943         public function getImages() {
 944                 return $this->mImages;
 945         }
 946
 947         /**
 948          * Set the revision corresponding to this LinksUpdate
 949          *
 950          * @since 1.27
 951          *
 952          * @param Revision $revision
 953          */
 954         public function setRevision( Revision $revision ) {
 955                 $this->mRevision = $revision;
 956         }
 957
 958         /**
 959          * @since 1.28
 960          * @return null|Revision
 961          */
 962         public function getRevision() {
 963                 return $this->mRevision;
 964         }
 965
 966         /**
 967          * Set the User who triggered this LinksUpdate
 968          *
 969          * @since 1.27
 970          * @param User $user
 971          */
 972         public function setTriggeringUser( User $user ) {
 973                 $this->user = $user;
 974         }
 975
 976         /**
 977          * @since 1.27
 978          * @return null|User
 979          */
 980         public function getTriggeringUser() {
 981                 return $this->user;
 982         }
 983
 984         /**
 985          * Invalidate any necessary link lists related to page property changes
 986          * @param array $changed
 987          */
 988         private function invalidateProperties( $changed ) {
 989                 global $wgPagePropLinkInvalidations;
 990
 991                 foreach ( $changed as $name => $value ) {
 992                         if ( isset( $wgPagePropLinkInvalidations[$name] ) ) {
 993                                 $inv = $wgPagePropLinkInvalidations[$name];
 994                                 if ( !is_array( $inv ) ) {
 995                                         $inv = [ $inv ];
 996                                 }
 997                                 foreach ( $inv as $table ) {
 998                                         DeferredUpdates::addUpdate( new HTMLCacheUpdate( $this->mTitle, $table ) );
 999                                 }
1000                         }
1001                 }
1002         }
1003
1004         /**
1005          * Fetch page links added by this LinksUpdate.  Only available after the update is complete.
1006          * @since 1.22
1007          * @return null|array Array of Titles
1008          */
1009         public function getAddedLinks() {
1010                 if ( $this->linkInsertions === null ) {
1011                         return null;
1012                 }
1013                 $result = [];
1014                 foreach ( $this->linkInsertions as $insertion ) {
1015                         $result[] = Title::makeTitle( $insertion['pl_namespace'], $insertion['pl_title'] );
1016                 }
1017
1018                 return $result;
1019         }
1020
1021         /**
1022          * Fetch page links removed by this LinksUpdate.  Only available after the update is complete.
1023          * @since 1.22
1024          * @return null|array Array of Titles
1025          */
1026         public function getRemovedLinks() {
1027                 if ( $this->linkDeletions === null ) {
1028                         return null;
1029                 }
1030                 $result = [];
1031                 foreach ( $this->linkDeletions as $ns => $titles ) {
1032                         foreach ( $titles as $title => $unused ) {
1033                                 $result[] = Title::makeTitle( $ns, $title );
1034                         }
1035                 }
1036
1037                 return $result;
1038         }
1039
1040         /**
1041          * Fetch page properties added by this LinksUpdate.
1042          * Only available after the update is complete.
1043          * @since 1.28
1044          * @return null|array
1045          */
1046         public function getAddedProperties() {
1047                 return $this->propertyInsertions;
1048         }
1049
1050         /**
1051          * Fetch page properties removed by this LinksUpdate.
1052          * Only available after the update is complete.
1053          * @since 1.28
1054          * @return null|array
1055          */
1056         public function getRemovedProperties() {
1057                 return $this->propertyDeletions;
1058         }
1059
1060         /**
1061          * Update links table freshness
1062          */
1063         private function updateLinksTimestamp() {
1064                 if ( $this->mId ) {
1065                         // The link updates made here only reflect the freshness of the parser output
1066                         $timestamp = $this->mParserOutput->getCacheTime();
1067                         $this->getDB()->update( 'page',
1068                                 [ 'page_links_updated' => $this->getDB()->timestamp( $timestamp ) ],
1069                                 [ 'page_id' => $this->mId ],
1070                                 __METHOD__
1071                         );
1072                 }
1073         }
1074
1075         /**
1076          * @return IDatabase
1077          */
1078         private function getDB() {
1079                 if ( !$this->db ) {
1080                         $this->db = wfGetDB( DB_MASTER );
1081                 }
1082
1083                 return $this->db;
1084         }
1085
1086         public function getAsJobSpecification() {
1087                 if ( $this->user ) {
1088                         $userInfo = [
1089                                 'userId' => $this->user->getId(),
1090                                 'userName' => $this->user->getName(),
1091                         ];
1092                 } else {
1093                         $userInfo = false;
1094                 }
1095
1096                 if ( $this->mRevision ) {
1097                         $triggeringRevisionId = $this->mRevision->getId();
1098                 } else {
1099                         $triggeringRevisionId = false;
1100                 }
1101
1102                 return [
1103                         'wiki' => $this->getDB()->getWikiID(),
1104                         'job'  => new JobSpecification(
1105                                 'refreshLinksPrioritized',
1106                                 [
1107                                         // Reuse the parser cache if it was saved
1108                                         'rootJobTimestamp' => $this->mParserOutput->getCacheTime(),
1109                                         'useRecursiveLinksUpdate' => $this->mRecursive,
1110                                         'triggeringUser' => $userInfo,
1111                                         'triggeringRevisionId' => $triggeringRevisionId,
1112                                 ],
1113                                 [ 'removeDuplicates' => true ],
1114                                 $this->getTitle()
1115                         )
1116                 ];
1117         }
1118 }