includes/deferred/LinksUpdate.php

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