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 use Wikimedia\ScopedCallback;
  25
  26 /**
  27  * Class the manages updates of *_link tables as well as similar extension-managed tables
  28  *
  29  * @note: LinksUpdate is managed by DeferredUpdates::execute(). Do not run this in a transaction.
  30  *
  31  * See docs/deferred.txt
  32  */
  33 class LinksUpdate extends DataUpdate implements EnqueueableDataUpdate {
  34         // @todo make members protected, but make sure extensions don't break
  35
  36         /** @var int Page ID of the article linked from */
  37         public $mId;
  38
  39         /** @var Title Title object of the article linked from */
  40         public $mTitle;
  41
  42         /** @var ParserOutput */
  43         public $mParserOutput;
  44
  45         /** @var array Map of title strings to IDs for the links in the document */
  46         public $mLinks;
  47
  48         /** @var array DB keys of the images used, in the array key only */
  49         public $mImages;
  50
  51         /** @var array Map of title strings to IDs for the template references, including broken ones */
  52         public $mTemplates;
  53
  54         /** @var array URLs of external links, array key only */
  55         public $mExternals;
  56
  57         /** @var array Map of category names to sort keys */
  58         public $mCategories;
  59
  60         /** @var array Map of language codes to titles */
  61         public $mInterlangs;
  62
  63         /** @var array 2-D map of (prefix => DBK => 1) */
  64         public $mInterwikis;
  65
  66         /** @var array Map of arbitrary name to value */
  67         public $mProperties;
  68
  69         /** @var bool Whether to queue jobs for recursive updates */
  70         public $mRecursive;
  71
  72         /** @var Revision Revision for which this update has been triggered */
  73         private $mRevision;
  74
  75         /**
  76          * @var null|array Added links if calculated.
  77          */
  78         private $linkInsertions = null;
  79
  80         /**
  81          * @var null|array Deleted links if calculated.
  82          */
  83         private $linkDeletions = null;
  84
  85         /**
  86          * @var null|array Added properties if calculated.
  87          */
  88         private $propertyInsertions = null;
  89
  90         /**
  91          * @var null|array Deleted properties if calculated.
  92          */
  93         private $propertyDeletions = null;
  94
  95         /**
  96          * @var User|null
  97          */
  98         private $user;
  99
 100         /** @var IDatabase */
 101         private $db;
 102
 103         /**
 104          * Constructor
 105          *
 106          * @param Title $title Title of the page we're updating
 107          * @param ParserOutput $parserOutput Output from a full parse of this page
 108          * @param bool $recursive Queue jobs for recursive updates?
 109          * @throws MWException
 110          */
 111         function __construct( Title $title, ParserOutput $parserOutput, $recursive = true ) {
 112                 parent::__construct();
 113
 114                 $this->mTitle = $title;
 115                 $this->mId = $title->getArticleID( Title::GAID_FOR_UPDATE );
 116
 117                 if ( !$this->mId ) {
 118                         throw new InvalidArgumentException(
 119                                 "The Title object yields no ID. Perhaps the page doesn't exist?"
 120                         );
 121                 }
 122
 123                 $this->mParserOutput = $parserOutput;
 124
 125                 $this->mLinks = $parserOutput->getLinks();
 126                 $this->mImages = $parserOutput->getImages();
 127                 $this->mTemplates = $parserOutput->getTemplates();
 128                 $this->mExternals = $parserOutput->getExternalLinks();
 129                 $this->mCategories = $parserOutput->getCategories();
 130                 $this->mProperties = $parserOutput->getProperties();
 131                 $this->mInterwikis = $parserOutput->getInterwikiLinks();
 132
 133                 # Convert the format of the interlanguage links
 134                 # I didn't want to change it in the ParserOutput, because that array is passed all
 135                 # the way back to the skin, so either a skin API break would be required, or an
 136                 # inefficient back-conversion.
 137                 $ill = $parserOutput->getLanguageLinks();
 138                 $this->mInterlangs = [];
 139                 foreach ( $ill as $link ) {
 140                         list( $key, $title ) = explode( ':', $link, 2 );
 141                         $this->mInterlangs[$key] = $title;
 142                 }
 143
 144                 foreach ( $this->mCategories as &$sortkey ) {
 145                         # If the sortkey is longer then 255 bytes,
 146                         # it truncated by DB, and then doesn't get
 147                         # matched when comparing existing vs current
 148                         # categories, causing bug 25254.
 149                         # Also. substr behaves weird when given "".
 150                         if ( $sortkey !== '' ) {
 151                                 $sortkey = substr( $sortkey, 0, 255 );
 152                         }
 153                 }
 154
 155                 $this->mRecursive = $recursive;
 156
 157                 Hooks::run( 'LinksUpdateConstructed', [ &$this ] );
 158         }
 159
 160         /**
 161          * Update link tables with outgoing links from an updated article
 162          *
 163          * @note: this is managed by DeferredUpdates::execute(). Do not run this in a transaction.
 164          */
 165         public function doUpdate() {
 166                 if ( $this->ticket ) {
 167                         // Make sure all links update threads see the changes of each other.
 168                         // This handles the case when updates have to batched into several COMMITs.
 169                         $scopedLock = self::acquirePageLock( $this->getDB(), $this->mId );
 170                 }
 171
 172                 Hooks::run( 'LinksUpdate', [ &$this ] );
 173                 $this->doIncrementalUpdate();
 174
 175                 // Commit and release the lock (if set)
 176                 ScopedCallback::consume( $scopedLock );
 177                 // Run post-commit hooks without DBO_TRX
 178                 $this->getDB()->onTransactionIdle(
 179                         function () {
 180                                 Hooks::run( 'LinksUpdateComplete', [ &$this, $this->ticket ] );
 181                         },
 182                         __METHOD__
 183                 );
 184         }
 185
 186         /**
 187          * Acquire a lock for performing link table updates for a page on a DB
 188          *
 189          * @param IDatabase $dbw
 190          * @param integer $pageId
 191          * @param string $why One of (job, atomicity)
 192          * @return ScopedCallback
 193          * @throws RuntimeException
 194          * @since 1.27
 195          */
 196         public static function acquirePageLock( IDatabase $dbw, $pageId, $why = 'atomicity' ) {
 197                 $key = "LinksUpdate:$why:pageid:$pageId";
 198                 $scopedLock = $dbw->getScopedLockAndFlush( $key, __METHOD__, 15 );
 199                 if ( !$scopedLock ) {
 200                         throw new RuntimeException( "Could not acquire lock '$key'." );
 201                 }
 202
 203                 return $scopedLock;
 204         }
 205
 206         protected function doIncrementalUpdate() {
 207                 # Page links
 208                 $existingPL = $this->getExistingLinks();
 209                 $this->linkDeletions = $this->getLinkDeletions( $existingPL );
 210                 $this->linkInsertions = $this->getLinkInsertions( $existingPL );
 211                 $this->incrTableUpdate( 'pagelinks', 'pl', $this->linkDeletions, $this->linkInsertions );
 212
 213                 # Image links
 214                 $existingIL = $this->getExistingImages();
 215                 $imageDeletes = $this->getImageDeletions( $existingIL );
 216                 $this->incrTableUpdate(
 217                         'imagelinks',
 218                         'il',
 219                         $imageDeletes,
 220                         $this->getImageInsertions( $existingIL ) );
 221
 222                 # Invalidate all image description pages which had links added or removed
 223                 $imageUpdates = $imageDeletes + array_diff_key( $this->mImages, $existingIL );
 224                 $this->invalidateImageDescriptions( $imageUpdates );
 225
 226                 # External links
 227                 $existingEL = $this->getExistingExternals();
 228                 $this->incrTableUpdate(
 229                         'externallinks',
 230                         'el',
 231                         $this->getExternalDeletions( $existingEL ),
 232                         $this->getExternalInsertions( $existingEL ) );
 233
 234                 # Language links
 235                 $existingLL = $this->getExistingInterlangs();
 236                 $this->incrTableUpdate(
 237                         'langlinks',
 238                         'll',
 239                         $this->getInterlangDeletions( $existingLL ),
 240                         $this->getInterlangInsertions( $existingLL ) );
 241
 242                 # Inline interwiki links
 243                 $existingIW = $this->getExistingInterwikis();
 244                 $this->incrTableUpdate(
 245                         'iwlinks',
 246                         'iwl',
 247                         $this->getInterwikiDeletions( $existingIW ),
 248                         $this->getInterwikiInsertions( $existingIW ) );
 249
 250                 # Template links
 251                 $existingTL = $this->getExistingTemplates();
 252                 $this->incrTableUpdate(
 253                         'templatelinks',
 254                         'tl',
 255                         $this->getTemplateDeletions( $existingTL ),
 256                         $this->getTemplateInsertions( $existingTL ) );
 257
 258                 # Category links
 259                 $existingCL = $this->getExistingCategories();
 260                 $categoryDeletes = $this->getCategoryDeletions( $existingCL );
 261                 $this->incrTableUpdate(
 262                         'categorylinks',
 263                         'cl',
 264                         $categoryDeletes,
 265                         $this->getCategoryInsertions( $existingCL ) );
 266                 $categoryInserts = array_diff_assoc( $this->mCategories, $existingCL );
 267                 $categoryUpdates = $categoryInserts + $categoryDeletes;
 268
 269                 # Page properties
 270                 $existingPP = $this->getExistingProperties();
 271                 $this->propertyDeletions = $this->getPropertyDeletions( $existingPP );
 272                 $this->incrTableUpdate(
 273                         'page_props',
 274                         'pp',
 275                         $this->propertyDeletions,
 276                         $this->getPropertyInsertions( $existingPP ) );
 277
 278                 # Invalidate the necessary pages
 279                 $this->propertyInsertions = array_diff_assoc( $this->mProperties, $existingPP );
 280                 $changed = $this->propertyDeletions + $this->propertyInsertions;
 281                 $this->invalidateProperties( $changed );
 282
 283                 # Invalidate all categories which were added, deleted or changed (set symmetric difference)
 284                 $this->invalidateCategories( $categoryUpdates );
 285                 $this->updateCategoryCounts( $categoryInserts, $categoryDeletes );
 286
 287                 # Refresh links of all pages including this page
 288                 # This will be in a separate transaction
 289                 if ( $this->mRecursive ) {
 290                         $this->queueRecursiveJobs();
 291                 }
 292
 293                 # Update the links table freshness for this title
 294                 $this->updateLinksTimestamp();
 295         }
 296
 297         /**
 298          * Queue recursive jobs for this page
 299          *
 300          * Which means do LinksUpdate on all pages that include the current page,
 301          * using the job queue.
 302          */
 303         protected function queueRecursiveJobs() {
 304                 self::queueRecursiveJobsForTable( $this->mTitle, 'templatelinks' );
 305                 if ( $this->mTitle->getNamespace() == NS_FILE ) {
 306                         // Process imagelinks in case the title is or was a redirect
 307                         self::queueRecursiveJobsForTable( $this->mTitle, 'imagelinks' );
 308                 }
 309
 310                 $bc = $this->mTitle->getBacklinkCache();
 311                 // Get jobs for cascade-protected backlinks for a high priority queue.
 312                 // If meta-templates change to using a new template, the new template
 313                 // should be implicitly protected as soon as possible, if applicable.
 314                 // These jobs duplicate a subset of the above ones, but can run sooner.
 315                 // Which ever runs first generally no-ops the other one.
 316                 $jobs = [];
 317                 foreach ( $bc->getCascadeProtectedLinks() as $title ) {
 318                         $jobs[] = RefreshLinksJob::newPrioritized( $title, [] );
 319                 }
 320                 JobQueueGroup::singleton()->push( $jobs );
 321         }
 322
 323         /**
 324          * Queue a RefreshLinks job for any table.
 325          *
 326          * @param Title $title Title to do job for
 327          * @param string $table Table to use (e.g. 'templatelinks')
 328          */
 329         public static function queueRecursiveJobsForTable( Title $title, $table ) {
 330                 if ( $title->getBacklinkCache()->hasLinks( $table ) ) {
 331                         $job = new RefreshLinksJob(
 332                                 $title,
 333                                 [
 334                                         'table' => $table,
 335                                         'recursive' => true,
 336                                 ] + Job::newRootJobParams( // "overall" refresh links job info
 337                                         "refreshlinks:{$table}:{$title->getPrefixedText()}"
 338                                 )
 339                         );
 340
 341                         JobQueueGroup::singleton()->push( $job );
 342                 }
 343         }
 344
 345         /**
 346          * @param array $cats
 347          */
 348         private function invalidateCategories( $cats ) {
 349                 PurgeJobUtils::invalidatePages( $this->getDB(), NS_CATEGORY, array_keys( $cats ) );
 350         }
 351
 352         /**
 353          * Update all the appropriate counts in the category table.
 354          * @param array $added Associative array of category name => sort key
 355          * @param array $deleted Associative array of category name => sort key
 356          */
 357         private function updateCategoryCounts( array $added, array $deleted ) {
 358                 global $wgUpdateRowsPerQuery;
 359
 360                 $wp = WikiPage::factory( $this->mTitle );
 361                 $factory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
 362
 363                 foreach ( array_chunk( array_keys( $added ), $wgUpdateRowsPerQuery ) as $addBatch ) {
 364                         $wp->updateCategoryCounts( $addBatch, [], $this->mId );
 365                         $factory->commitAndWaitForReplication(
 366                                 __METHOD__, $this->ticket, [ 'wiki' => $this->getDB()->getWikiID() ]
 367                         );
 368                 }
 369
 370                 foreach ( array_chunk( array_keys( $deleted ), $wgUpdateRowsPerQuery ) as $deleteBatch ) {
 371                         $wp->updateCategoryCounts( [], $deleteBatch, $this->mId );
 372                         $factory->commitAndWaitForReplication(
 373                                 __METHOD__, $this->ticket, [ 'wiki' => $this->getDB()->getWikiID() ]
 374                         );
 375                 }
 376         }
 377
 378         /**
 379          * @param array $images
 380          */
 381         private function invalidateImageDescriptions( $images ) {
 382                 PurgeJobUtils::invalidatePages( $this->getDB(), NS_FILE, array_keys( $images ) );
 383         }
 384
 385         /**
 386          * Update a table by doing a delete query then an insert query
 387          * @param string $table Table name
 388          * @param string $prefix Field name prefix
 389          * @param array $deletions
 390          * @param array $insertions Rows to insert
 391          */
 392         private function incrTableUpdate( $table, $prefix, $deletions, $insertions ) {
 393                 $services = MediaWikiServices::getInstance();
 394                 $bSize = $services->getMainConfig()->get( 'UpdateRowsPerQuery' );
 395                 $factory = $services->getDBLoadBalancerFactory();
 396
 397                 if ( $table === 'page_props' ) {
 398                         $fromField = 'pp_page';
 399                 } else {
 400                         $fromField = "{$prefix}_from";
 401                 }
 402
 403                 $deleteWheres = []; // list of WHERE clause arrays for each DB delete() call
 404                 if ( $table === 'pagelinks' || $table === 'templatelinks' || $table === 'iwlinks' ) {
 405                         $baseKey =  ( $table === 'iwlinks' ) ? 'iwl_prefix' : "{$prefix}_namespace";
 406
 407                         $curBatchSize = 0;
 408                         $curDeletionBatch = [];
 409                         $deletionBatches = [];
 410                         foreach ( $deletions as $ns => $dbKeys ) {
 411                                 foreach ( $dbKeys as $dbKey => $unused ) {
 412                                         $curDeletionBatch[$ns][$dbKey] = 1;
 413                                         if ( ++$curBatchSize >= $bSize ) {
 414                                                 $deletionBatches[] = $curDeletionBatch;
 415                                                 $curDeletionBatch = [];
 416                                                 $curBatchSize = 0;
 417                                         }
 418                                 }
 419                         }
 420                         if ( $curDeletionBatch ) {
 421                                 $deletionBatches[] = $curDeletionBatch;
 422                         }
 423
 424                         foreach ( $deletionBatches as $deletionBatch ) {
 425                                 $deleteWheres[] = [
 426                                         $fromField => $this->mId,
 427                                         $this->getDB()->makeWhereFrom2d( $deletionBatch, $baseKey, "{$prefix}_title" )
 428                                 ];
 429                         }
 430                 } else {
 431                         if ( $table === 'langlinks' ) {
 432                                 $toField = 'll_lang';
 433                         } elseif ( $table === 'page_props' ) {
 434                                 $toField = 'pp_propname';
 435                         } else {
 436                                 $toField = $prefix . '_to';
 437                         }
 438
 439                         $deletionBatches = array_chunk( array_keys( $deletions ), $bSize );
 440                         foreach ( $deletionBatches as $deletionBatch ) {
 441                                 $deleteWheres[] = [ $fromField => $this->mId, $toField => $deletionBatch ];
 442                         }
 443                 }
 444
 445                 foreach ( $deleteWheres as $deleteWhere ) {
 446                         $this->getDB()->delete( $table, $deleteWhere, __METHOD__ );
 447                         $factory->commitAndWaitForReplication(
 448                                 __METHOD__, $this->ticket, [ 'wiki' => $this->getDB()->getWikiID() ]
 449                         );
 450                 }
 451
 452                 $insertBatches = array_chunk( $insertions, $bSize );
 453                 foreach ( $insertBatches as $insertBatch ) {
 454                         $this->getDB()->insert( $table, $insertBatch, __METHOD__, 'IGNORE' );
 455                         $factory->commitAndWaitForReplication(
 456                                 __METHOD__, $this->ticket, [ 'wiki' => $this->getDB()->getWikiID() ]
 457                         );
 458                 }
 459
 460                 if ( count( $insertions ) ) {
 461                         Hooks::run( 'LinksUpdateAfterInsert', [ $this, $table, $insertions ] );
 462                 }
 463         }
 464
 465         /**
 466          * Get an array of pagelinks insertions for passing to the DB
 467          * Skips the titles specified by the 2-D array $existing
 468          * @param array $existing
 469          * @return array
 470          */
 471         private function getLinkInsertions( $existing = [] ) {
 472                 $arr = [];
 473                 foreach ( $this->mLinks as $ns => $dbkeys ) {
 474                         $diffs = isset( $existing[$ns] )
 475                                 ? array_diff_key( $dbkeys, $existing[$ns] )
 476                                 : $dbkeys;
 477                         foreach ( $diffs as $dbk => $id ) {
 478                                 $arr[] = [
 479                                         'pl_from' => $this->mId,
 480                                         'pl_from_namespace' => $this->mTitle->getNamespace(),
 481                                         'pl_namespace' => $ns,
 482                                         'pl_title' => $dbk
 483                                 ];
 484                         }
 485                 }
 486
 487                 return $arr;
 488         }
 489
 490         /**
 491          * Get an array of template insertions. Like getLinkInsertions()
 492          * @param array $existing
 493          * @return array
 494          */
 495         private function getTemplateInsertions( $existing = [] ) {
 496                 $arr = [];
 497                 foreach ( $this->mTemplates as $ns => $dbkeys ) {
 498                         $diffs = isset( $existing[$ns] ) ? array_diff_key( $dbkeys, $existing[$ns] ) : $dbkeys;
 499                         foreach ( $diffs as $dbk => $id ) {
 500                                 $arr[] = [
 501                                         'tl_from' => $this->mId,
 502                                         'tl_from_namespace' => $this->mTitle->getNamespace(),
 503                                         'tl_namespace' => $ns,
 504                                         'tl_title' => $dbk
 505                                 ];
 506                         }
 507                 }
 508
 509                 return $arr;
 510         }
 511
 512         /**
 513          * Get an array of image insertions
 514          * Skips the names specified in $existing
 515          * @param array $existing
 516          * @return array
 517          */
 518         private function getImageInsertions( $existing = [] ) {
 519                 $arr = [];
 520                 $diffs = array_diff_key( $this->mImages, $existing );
 521                 foreach ( $diffs as $iname => $dummy ) {
 522                         $arr[] = [
 523                                 'il_from' => $this->mId,
 524                                 'il_from_namespace' => $this->mTitle->getNamespace(),
 525                                 'il_to' => $iname
 526                         ];
 527                 }
 528
 529                 return $arr;
 530         }
 531
 532         /**
 533          * Get an array of externallinks insertions. Skips the names specified in $existing
 534          * @param array $existing
 535          * @return array
 536          */
 537         private function getExternalInsertions( $existing = [] ) {
 538                 $arr = [];
 539                 $diffs = array_diff_key( $this->mExternals, $existing );
 540                 foreach ( $diffs as $url => $dummy ) {
 541                         foreach ( wfMakeUrlIndexes( $url ) as $index ) {
 542                                 $arr[] = [
 543                                         'el_id' => $this->getDB()->nextSequenceValue( 'externallinks_el_id_seq' ),
 544                                         'el_from' => $this->mId,
 545                                         'el_to' => $url,
 546                                         'el_index' => $index,
 547                                 ];
 548                         }
 549                 }
 550
 551                 return $arr;
 552         }
 553
 554         /**
 555          * Get an array of category insertions
 556          *
 557          * @param array $existing Mapping existing category names to sort keys. If both
 558          * match a link in $this, the link will be omitted from the output
 559          *
 560          * @return array
 561          */
 562         private function getCategoryInsertions( $existing = [] ) {
 563                 global $wgContLang, $wgCategoryCollation;
 564                 $diffs = array_diff_assoc( $this->mCategories, $existing );
 565                 $arr = [];
 566                 foreach ( $diffs as $name => $prefix ) {
 567                         $nt = Title::makeTitleSafe( NS_CATEGORY, $name );
 568                         $wgContLang->findVariantLink( $name, $nt, true );
 569
 570                         if ( $this->mTitle->getNamespace() == NS_CATEGORY ) {
 571                                 $type = 'subcat';
 572                         } elseif ( $this->mTitle->getNamespace() == NS_FILE ) {
 573                                 $type = 'file';
 574                         } else {
 575                                 $type = 'page';
 576                         }
 577
 578                         # Treat custom sortkeys as a prefix, so that if multiple
 579                         # things are forced to sort as '*' or something, they'll
 580                         # sort properly in the category rather than in page_id
 581                         # order or such.
 582                         $sortkey = Collation::singleton()->getSortKey(
 583                                 $this->mTitle->getCategorySortkey( $prefix ) );
 584
 585                         $arr[] = [
 586                                 'cl_from' => $this->mId,
 587                                 'cl_to' => $name,
 588                                 'cl_sortkey' => $sortkey,
 589                                 'cl_timestamp' => $this->getDB()->timestamp(),
 590                                 'cl_sortkey_prefix' => $prefix,
 591                                 'cl_collation' => $wgCategoryCollation,
 592                                 'cl_type' => $type,
 593                         ];
 594                 }
 595
 596                 return $arr;
 597         }
 598
 599         /**
 600          * Get an array of interlanguage link insertions
 601          *
 602          * @param array $existing Mapping existing language codes to titles
 603          *
 604          * @return array
 605          */
 606         private function getInterlangInsertions( $existing = [] ) {
 607                 $diffs = array_diff_assoc( $this->mInterlangs, $existing );
 608                 $arr = [];
 609                 foreach ( $diffs as $lang => $title ) {
 610                         $arr[] = [
 611                                 'll_from' => $this->mId,
 612                                 'll_lang' => $lang,
 613                                 'll_title' => $title
 614                         ];
 615                 }
 616
 617                 return $arr;
 618         }
 619
 620         /**
 621          * Get an array of page property insertions
 622          * @param array $existing
 623          * @return array
 624          */
 625         function getPropertyInsertions( $existing = [] ) {
 626                 $diffs = array_diff_assoc( $this->mProperties, $existing );
 627
 628                 $arr = [];
 629                 foreach ( array_keys( $diffs ) as $name ) {
 630                         $arr[] = $this->getPagePropRowData( $name );
 631                 }
 632
 633                 return $arr;
 634         }
 635
 636         /**
 637          * Returns an associative array to be used for inserting a row into
 638          * the page_props table. Besides the given property name, this will
 639          * include the page id from $this->mId and any property value from
 640          * $this->mProperties.
 641          *
 642          * The array returned will include the pp_sortkey field if this
 643          * is present in the database (as indicated by $wgPagePropsHaveSortkey).
 644          * The sortkey value is currently determined by getPropertySortKeyValue().
 645          *
 646          * @note this assumes that $this->mProperties[$prop] is defined.
 647          *
 648          * @param string $prop The name of the property.
 649          *
 650          * @return array
 651          */
 652         private function getPagePropRowData( $prop ) {
 653                 global $wgPagePropsHaveSortkey;
 654
 655                 $value = $this->mProperties[$prop];
 656
 657                 $row = [
 658                         'pp_page' => $this->mId,
 659                         'pp_propname' => $prop,
 660                         'pp_value' => $value,
 661                 ];
 662
 663                 if ( $wgPagePropsHaveSortkey ) {
 664                         $row['pp_sortkey'] = $this->getPropertySortKeyValue( $value );
 665                 }
 666
 667                 return $row;
 668         }
 669
 670         /**
 671          * Determines the sort key for the given property value.
 672          * This will return $value if it is a float or int,
 673          * 1 or resp. 0 if it is a bool, and null otherwise.
 674          *
 675          * @note In the future, we may allow the sortkey to be specified explicitly
 676          *       in ParserOutput::setProperty.
 677          *
 678          * @param mixed $value
 679          *
 680          * @return float|null
 681          */
 682         private function getPropertySortKeyValue( $value ) {
 683                 if ( is_int( $value ) || is_float( $value ) || is_bool( $value ) ) {
 684                         return floatval( $value );
 685                 }
 686
 687                 return null;
 688         }
 689
 690         /**
 691          * Get an array of interwiki insertions for passing to the DB
 692          * Skips the titles specified by the 2-D array $existing
 693          * @param array $existing
 694          * @return array
 695          */
 696         private function getInterwikiInsertions( $existing = [] ) {
 697                 $arr = [];
 698                 foreach ( $this->mInterwikis as $prefix => $dbkeys ) {
 699                         $diffs = isset( $existing[$prefix] )
 700                                 ? array_diff_key( $dbkeys, $existing[$prefix] )
 701                                 : $dbkeys;
 702
 703                         foreach ( $diffs as $dbk => $id ) {
 704                                 $arr[] = [
 705                                         'iwl_from' => $this->mId,
 706                                         'iwl_prefix' => $prefix,
 707                                         'iwl_title' => $dbk
 708                                 ];
 709                         }
 710                 }
 711
 712                 return $arr;
 713         }
 714
 715         /**
 716          * Given an array of existing links, returns those links which are not in $this
 717          * and thus should be deleted.
 718          * @param array $existing
 719          * @return array
 720          */
 721         private function getLinkDeletions( $existing ) {
 722                 $del = [];
 723                 foreach ( $existing as $ns => $dbkeys ) {
 724                         if ( isset( $this->mLinks[$ns] ) ) {
 725                                 $del[$ns] = array_diff_key( $existing[$ns], $this->mLinks[$ns] );
 726                         } else {
 727                                 $del[$ns] = $existing[$ns];
 728                         }
 729                 }
 730
 731                 return $del;
 732         }
 733
 734         /**
 735          * Given an array of existing templates, returns those templates which are not in $this
 736          * and thus should be deleted.
 737          * @param array $existing
 738          * @return array
 739          */
 740         private function getTemplateDeletions( $existing ) {
 741                 $del = [];
 742                 foreach ( $existing as $ns => $dbkeys ) {
 743                         if ( isset( $this->mTemplates[$ns] ) ) {
 744                                 $del[$ns] = array_diff_key( $existing[$ns], $this->mTemplates[$ns] );
 745                         } else {
 746                                 $del[$ns] = $existing[$ns];
 747                         }
 748                 }
 749
 750                 return $del;
 751         }
 752
 753         /**
 754          * Given an array of existing images, returns those images which are not in $this
 755          * and thus should be deleted.
 756          * @param array $existing
 757          * @return array
 758          */
 759         private function getImageDeletions( $existing ) {
 760                 return array_diff_key( $existing, $this->mImages );
 761         }
 762
 763         /**
 764          * Given an array of existing external links, returns those links which are not
 765          * in $this and thus should be deleted.
 766          * @param array $existing
 767          * @return array
 768          */
 769         private function getExternalDeletions( $existing ) {
 770                 return array_diff_key( $existing, $this->mExternals );
 771         }
 772
 773         /**
 774          * Given an array of existing categories, returns those categories which are not in $this
 775          * and thus should be deleted.
 776          * @param array $existing
 777          * @return array
 778          */
 779         private function getCategoryDeletions( $existing ) {
 780                 return array_diff_assoc( $existing, $this->mCategories );
 781         }
 782
 783         /**
 784          * Given an array of existing interlanguage links, returns those links which are not
 785          * in $this and thus should be deleted.
 786          * @param array $existing
 787          * @return array
 788          */
 789         private function getInterlangDeletions( $existing ) {
 790                 return array_diff_assoc( $existing, $this->mInterlangs );
 791         }
 792
 793         /**
 794          * Get array of properties which should be deleted.
 795          * @param array $existing
 796          * @return array
 797          */
 798         function getPropertyDeletions( $existing ) {
 799                 return array_diff_assoc( $existing, $this->mProperties );
 800         }
 801
 802         /**
 803          * Given an array of existing interwiki links, returns those links which are not in $this
 804          * and thus should be deleted.
 805          * @param array $existing
 806          * @return array
 807          */
 808         private function getInterwikiDeletions( $existing ) {
 809                 $del = [];
 810                 foreach ( $existing as $prefix => $dbkeys ) {
 811                         if ( isset( $this->mInterwikis[$prefix] ) ) {
 812                                 $del[$prefix] = array_diff_key( $existing[$prefix], $this->mInterwikis[$prefix] );
 813                         } else {
 814                                 $del[$prefix] = $existing[$prefix];
 815                         }
 816                 }
 817
 818                 return $del;
 819         }
 820
 821         /**
 822          * Get an array of existing links, as a 2-D array
 823          *
 824          * @return array
 825          */
 826         private function getExistingLinks() {
 827                 $res = $this->getDB()->select( 'pagelinks', [ 'pl_namespace', 'pl_title' ],
 828                         [ 'pl_from' => $this->mId ], __METHOD__ );
 829                 $arr = [];
 830                 foreach ( $res as $row ) {
 831                         if ( !isset( $arr[$row->pl_namespace] ) ) {
 832                                 $arr[$row->pl_namespace] = [];
 833                         }
 834                         $arr[$row->pl_namespace][$row->pl_title] = 1;
 835                 }
 836
 837                 return $arr;
 838         }
 839
 840         /**
 841          * Get an array of existing templates, as a 2-D array
 842          *
 843          * @return array
 844          */
 845         private function getExistingTemplates() {
 846                 $res = $this->getDB()->select( 'templatelinks', [ 'tl_namespace', 'tl_title' ],
 847                         [ 'tl_from' => $this->mId ], __METHOD__ );
 848                 $arr = [];
 849                 foreach ( $res as $row ) {
 850                         if ( !isset( $arr[$row->tl_namespace] ) ) {
 851                                 $arr[$row->tl_namespace] = [];
 852                         }
 853                         $arr[$row->tl_namespace][$row->tl_title] = 1;
 854                 }
 855
 856                 return $arr;
 857         }
 858
 859         /**
 860          * Get an array of existing images, image names in the keys
 861          *
 862          * @return array
 863          */
 864         private function getExistingImages() {
 865                 $res = $this->getDB()->select( 'imagelinks', [ 'il_to' ],
 866                         [ 'il_from' => $this->mId ], __METHOD__ );
 867                 $arr = [];
 868                 foreach ( $res as $row ) {
 869                         $arr[$row->il_to] = 1;
 870                 }
 871
 872                 return $arr;
 873         }
 874
 875         /**
 876          * Get an array of existing external links, URLs in the keys
 877          *
 878          * @return array
 879          */
 880         private function getExistingExternals() {
 881                 $res = $this->getDB()->select( 'externallinks', [ 'el_to' ],
 882                         [ 'el_from' => $this->mId ], __METHOD__ );
 883                 $arr = [];
 884                 foreach ( $res as $row ) {
 885                         $arr[$row->el_to] = 1;
 886                 }
 887
 888                 return $arr;
 889         }
 890
 891         /**
 892          * Get an array of existing categories, with the name in the key and sort key in the value.
 893          *
 894          * @return array
 895          */
 896         private function getExistingCategories() {
 897                 $res = $this->getDB()->select( 'categorylinks', [ 'cl_to', 'cl_sortkey_prefix' ],
 898                         [ 'cl_from' => $this->mId ], __METHOD__ );
 899                 $arr = [];
 900                 foreach ( $res as $row ) {
 901                         $arr[$row->cl_to] = $row->cl_sortkey_prefix;
 902                 }
 903
 904                 return $arr;
 905         }
 906
 907         /**
 908          * Get an array of existing interlanguage links, with the language code in the key and the
 909          * title in the value.
 910          *
 911          * @return array
 912          */
 913         private function getExistingInterlangs() {
 914                 $res = $this->getDB()->select( 'langlinks', [ 'll_lang', 'll_title' ],
 915                         [ 'll_from' => $this->mId ], __METHOD__ );
 916                 $arr = [];
 917                 foreach ( $res as $row ) {
 918                         $arr[$row->ll_lang] = $row->ll_title;
 919                 }
 920
 921                 return $arr;
 922         }
 923
 924         /**
 925          * Get an array of existing inline interwiki links, as a 2-D array
 926          * @return array (prefix => array(dbkey => 1))
 927          */
 928         private function getExistingInterwikis() {
 929                 $res = $this->getDB()->select( 'iwlinks', [ 'iwl_prefix', 'iwl_title' ],
 930                         [ 'iwl_from' => $this->mId ], __METHOD__ );
 931                 $arr = [];
 932                 foreach ( $res as $row ) {
 933                         if ( !isset( $arr[$row->iwl_prefix] ) ) {
 934                                 $arr[$row->iwl_prefix] = [];
 935                         }
 936                         $arr[$row->iwl_prefix][$row->iwl_title] = 1;
 937                 }
 938
 939                 return $arr;
 940         }
 941
 942         /**
 943          * Get an array of existing categories, with the name in the key and sort key in the value.
 944          *
 945          * @return array Array of property names and values
 946          */
 947         private function getExistingProperties() {
 948                 $res = $this->getDB()->select( 'page_props', [ 'pp_propname', 'pp_value' ],
 949                         [ 'pp_page' => $this->mId ], __METHOD__ );
 950                 $arr = [];
 951                 foreach ( $res as $row ) {
 952                         $arr[$row->pp_propname] = $row->pp_value;
 953                 }
 954
 955                 return $arr;
 956         }
 957
 958         /**
 959          * Return the title object of the page being updated
 960          * @return Title
 961          */
 962         public function getTitle() {
 963                 return $this->mTitle;
 964         }
 965
 966         /**
 967          * Returns parser output
 968          * @since 1.19
 969          * @return ParserOutput
 970          */
 971         public function getParserOutput() {
 972                 return $this->mParserOutput;
 973         }
 974
 975         /**
 976          * Return the list of images used as generated by the parser
 977          * @return array
 978          */
 979         public function getImages() {
 980                 return $this->mImages;
 981         }
 982
 983         /**
 984          * Set the revision corresponding to this LinksUpdate
 985          *
 986          * @since 1.27
 987          *
 988          * @param Revision $revision
 989          */
 990         public function setRevision( Revision $revision ) {
 991                 $this->mRevision = $revision;
 992         }
 993
 994         /**
 995          * @since 1.28
 996          * @return null|Revision
 997          */
 998         public function getRevision() {
 999                 return $this->mRevision;
1000         }
1001
1002         /**
1003          * Set the User who triggered this LinksUpdate
1004          *
1005          * @since 1.27
1006          * @param User $user
1007          */
1008         public function setTriggeringUser( User $user ) {
1009                 $this->user = $user;
1010         }
1011
1012         /**
1013          * @since 1.27
1014          * @return null|User
1015          */
1016         public function getTriggeringUser() {
1017                 return $this->user;
1018         }
1019
1020         /**
1021          * Invalidate any necessary link lists related to page property changes
1022          * @param array $changed
1023          */
1024         private function invalidateProperties( $changed ) {
1025                 global $wgPagePropLinkInvalidations;
1026
1027                 foreach ( $changed as $name => $value ) {
1028                         if ( isset( $wgPagePropLinkInvalidations[$name] ) ) {
1029                                 $inv = $wgPagePropLinkInvalidations[$name];
1030                                 if ( !is_array( $inv ) ) {
1031                                         $inv = [ $inv ];
1032                                 }
1033                                 foreach ( $inv as $table ) {
1034                                         DeferredUpdates::addUpdate( new HTMLCacheUpdate( $this->mTitle, $table ) );
1035                                 }
1036                         }
1037                 }
1038         }
1039
1040         /**
1041          * Fetch page links added by this LinksUpdate.  Only available after the update is complete.
1042          * @since 1.22
1043          * @return null|array Array of Titles
1044          */
1045         public function getAddedLinks() {
1046                 if ( $this->linkInsertions === null ) {
1047                         return null;
1048                 }
1049                 $result = [];
1050                 foreach ( $this->linkInsertions as $insertion ) {
1051                         $result[] = Title::makeTitle( $insertion['pl_namespace'], $insertion['pl_title'] );
1052                 }
1053
1054                 return $result;
1055         }
1056
1057         /**
1058          * Fetch page links removed by this LinksUpdate.  Only available after the update is complete.
1059          * @since 1.22
1060          * @return null|array Array of Titles
1061          */
1062         public function getRemovedLinks() {
1063                 if ( $this->linkDeletions === null ) {
1064                         return null;
1065                 }
1066                 $result = [];
1067                 foreach ( $this->linkDeletions as $ns => $titles ) {
1068                         foreach ( $titles as $title => $unused ) {
1069                                 $result[] = Title::makeTitle( $ns, $title );
1070                         }
1071                 }
1072
1073                 return $result;
1074         }
1075
1076         /**
1077          * Fetch page properties added by this LinksUpdate.
1078          * Only available after the update is complete.
1079          * @since 1.28
1080          * @return null|array
1081          */
1082         public function getAddedProperties() {
1083                 return $this->propertyInsertions;
1084         }
1085
1086         /**
1087          * Fetch page properties removed by this LinksUpdate.
1088          * Only available after the update is complete.
1089          * @since 1.28
1090          * @return null|array
1091          */
1092         public function getRemovedProperties() {
1093                 return $this->propertyDeletions;
1094         }
1095
1096         /**
1097          * Update links table freshness
1098          */
1099         private function updateLinksTimestamp() {
1100                 if ( $this->mId ) {
1101                         // The link updates made here only reflect the freshness of the parser output
1102                         $timestamp = $this->mParserOutput->getCacheTime();
1103                         $this->getDB()->update( 'page',
1104                                 [ 'page_links_updated' => $this->getDB()->timestamp( $timestamp ) ],
1105                                 [ 'page_id' => $this->mId ],
1106                                 __METHOD__
1107                         );
1108                 }
1109         }
1110
1111         /**
1112          * @return IDatabase
1113          */
1114         private function getDB() {
1115                 if ( !$this->db ) {
1116                         $this->db = wfGetDB( DB_MASTER );
1117                 }
1118
1119                 return $this->db;
1120         }
1121
1122         public function getAsJobSpecification() {
1123                 if ( $this->user ) {
1124                         $userInfo = [
1125                                 'userId' => $this->user->getId(),
1126                                 'userName' => $this->user->getName(),
1127                         ];
1128                 } else {
1129                         $userInfo = false;
1130                 }
1131
1132                 if ( $this->mRevision ) {
1133                         $triggeringRevisionId = $this->mRevision->getId();
1134                 } else {
1135                         $triggeringRevisionId = false;
1136                 }
1137
1138                 return [
1139                         'wiki' => $this->getDB()->getWikiID(),
1140                         'job'  => new JobSpecification(
1141                                 'refreshLinksPrioritized',
1142                                 [
1143                                         // Reuse the parser cache if it was saved
1144                                         'rootJobTimestamp' => $this->mParserOutput->getCacheTime(),
1145                                         'useRecursiveLinksUpdate' => $this->mRecursive,
1146                                         'triggeringUser' => $userInfo,
1147                                         'triggeringRevisionId' => $triggeringRevisionId,
1148                                 ],
1149                                 [ 'removeDuplicates' => true ],
1150                                 $this->getTitle()
1151                         )
1152                 ];
1153         }
1154 }