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