includes/cache/BacklinkCache.php

   1 <?php
   2 /**
   3  * Class for fetching backlink lists, approximate backlink counts and
   4  * partitions.
   5  *
   6  * This program is free software; you can redistribute it and/or modify
   7  * it under the terms of the GNU General Public License as published by
   8  * the Free Software Foundation; either version 2 of the License, or
   9  * (at your option) any later version.
  10  *
  11  * This program is distributed in the hope that it will be useful,
  12  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  13  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  14  * GNU General Public License for more details.
  15  *
  16  * You should have received a copy of the GNU General Public License along
  17  * with this program; if not, write to the Free Software Foundation, Inc.,
  18  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  19  * http://www.gnu.org/copyleft/gpl.html
  20  *
  21  * @file
  22  * @author Tim Starling
  23  * @copyright © 2009, Tim Starling, Domas Mituzas
  24  * @copyright © 2010, Max Sem
  25  * @copyright © 2011, Antoine Musso
  26  */
  27
  28 use Wikimedia\Rdbms\ResultWrapper;
  29 use Wikimedia\Rdbms\FakeResultWrapper;
  30 use Wikimedia\Rdbms\IDatabase;
  31
  32 /**
  33  * Class for fetching backlink lists, approximate backlink counts and
  34  * partitions. This is a shared cache.
  35  *
  36  * Instances of this class should typically be fetched with the method
  37  * $title->getBacklinkCache().
  38  *
  39  * Ideally you should only get your backlinks from here when you think
  40  * there is some advantage in caching them. Otherwise it's just a waste
  41  * of memory.
  42  *
  43  * Introduced by r47317
  44  */
  45 class BacklinkCache {
  46         /** @var BacklinkCache */
  47         protected static $instance;
  48
  49         /**
  50          * Multi dimensions array representing batches. Keys are:
  51          *  > (string) links table name
  52          *   > (int) batch size
  53          *    > 'numRows' : Number of rows for this link table
  54          *    > 'batches' : [ $start, $end ]
  55          *
  56          * @see BacklinkCache::partitionResult()
  57          *
  58          * Cleared with BacklinkCache::clear()
  59          * @var array[]
  60          */
  61         protected $partitionCache = [];
  62
  63         /**
  64          * Contains the whole links from a database result.
  65          * This is raw data that will be partitioned in $partitionCache
  66          *
  67          * Initialized with BacklinkCache::getLinks()
  68          * Cleared with BacklinkCache::clear()
  69          * @var ResultWrapper[]
  70          */
  71         protected $fullResultCache = [];
  72
  73         /**
  74          * Local copy of a database object.
  75          *
  76          * Accessor: BacklinkCache::getDB()
  77          * Mutator : BacklinkCache::setDB()
  78          * Cleared with BacklinkCache::clear()
  79          */
  80         protected $db;
  81
  82         /**
  83          * Local copy of a Title object
  84          */
  85         protected $title;
  86
  87         const CACHE_EXPIRY = 3600;
  88
  89         /**
  90          * Create a new BacklinkCache
  91          *
  92          * @param Title $title : Title object to create a backlink cache for
  93          */
  94         public function __construct( Title $title ) {
  95                 $this->title = $title;
  96         }
  97
  98         /**
  99          * Create a new BacklinkCache or reuse any existing one.
 100          * Currently, only one cache instance can exist; callers that
 101          * need multiple backlink cache objects should keep them in scope.
 102          *
 103          * @param Title $title Title object to get a backlink cache for
 104          * @return BacklinkCache
 105          */
 106         public static function get( Title $title ) {
 107                 if ( !self::$instance || !self::$instance->title->equals( $title ) ) {
 108                         self::$instance = new self( $title );
 109                 }
 110                 return self::$instance;
 111         }
 112
 113         /**
 114          * Serialization handler, diasallows to serialize the database to prevent
 115          * failures after this class is deserialized from cache with dead DB
 116          * connection.
 117          *
 118          * @return array
 119          */
 120         function __sleep() {
 121                 return [ 'partitionCache', 'fullResultCache', 'title' ];
 122         }
 123
 124         /**
 125          * Clear locally stored data and database object.
 126          */
 127         public function clear() {
 128                 $this->partitionCache = [];
 129                 $this->fullResultCache = [];
 130                 unset( $this->db );
 131         }
 132
 133         /**
 134          * Set the Database object to use
 135          *
 136          * @param IDatabase $db
 137          */
 138         public function setDB( $db ) {
 139                 $this->db = $db;
 140         }
 141
 142         /**
 143          * Get the replica DB connection to the database
 144          * When non existing, will initialize the connection.
 145          * @return IDatabase
 146          */
 147         protected function getDB() {
 148                 if ( !isset( $this->db ) ) {
 149                         $this->db = wfGetDB( DB_REPLICA );
 150                 }
 151
 152                 return $this->db;
 153         }
 154
 155         /**
 156          * Get the backlinks for a given table. Cached in process memory only.
 157          * @param string $table
 158          * @param int|bool $startId
 159          * @param int|bool $endId
 160          * @param int $max
 161          * @return TitleArrayFromResult
 162          */
 163         public function getLinks( $table, $startId = false, $endId = false, $max = INF ) {
 164                 return TitleArray::newFromResult( $this->queryLinks( $table, $startId, $endId, $max ) );
 165         }
 166
 167         /**
 168          * Get the backlinks for a given table. Cached in process memory only.
 169          * @param string $table
 170          * @param int|bool $startId
 171          * @param int|bool $endId
 172          * @param int $max
 173          * @param string $select 'all' or 'ids'
 174          * @return ResultWrapper
 175          */
 176         protected function queryLinks( $table, $startId, $endId, $max, $select = 'all' ) {
 177                 $fromField = $this->getPrefix( $table ) . '_from';
 178
 179                 if ( !$startId && !$endId && is_infinite( $max )
 180                         && isset( $this->fullResultCache[$table] )
 181                 ) {
 182                         wfDebug( __METHOD__ . ": got results from cache\n" );
 183                         $res = $this->fullResultCache[$table];
 184                 } else {
 185                         wfDebug( __METHOD__ . ": got results from DB\n" );
 186                         $conds = $this->getConditions( $table );
 187                         // Use the from field in the condition rather than the joined page_id,
 188                         // because databases are stupid and don't necessarily propagate indexes.
 189                         if ( $startId ) {
 190                                 $conds[] = "$fromField >= " . intval( $startId );
 191                         }
 192                         if ( $endId ) {
 193                                 $conds[] = "$fromField <= " . intval( $endId );
 194                         }
 195                         $options = [ 'ORDER BY' => $fromField ];
 196                         if ( is_finite( $max ) && $max > 0 ) {
 197                                 $options['LIMIT'] = $max;
 198                         }
 199
 200                         if ( $select === 'ids' ) {
 201                                 // Just select from the backlink table and ignore the page JOIN
 202                                 $res = $this->getDB()->select(
 203                                         $table,
 204                                         [ $this->getPrefix( $table ) . '_from AS page_id' ],
 205                                         array_filter( $conds, function ( $clause ) { // kind of janky
 206                                                 return !preg_match( '/(\b|=)page_id(\b|=)/', $clause );
 207                                         } ),
 208                                         __METHOD__,
 209                                         $options
 210                                 );
 211                         } else {
 212                                 // Select from the backlink table and JOIN with page title information
 213                                 $res = $this->getDB()->select(
 214                                         [ $table, 'page' ],
 215                                         [ 'page_namespace', 'page_title', 'page_id' ],
 216                                         $conds,
 217                                         __METHOD__,
 218                                         array_merge( [ 'STRAIGHT_JOIN' ], $options )
 219                                 );
 220                         }
 221
 222                         if ( $select === 'all' && !$startId && !$endId && $res->numRows() < $max ) {
 223                                 // The full results fit within the limit, so cache them
 224                                 $this->fullResultCache[$table] = $res;
 225                         } else {
 226                                 wfDebug( __METHOD__ . ": results from DB were uncacheable\n" );
 227                         }
 228                 }
 229
 230                 return $res;
 231         }
 232
 233         /**
 234          * Get the field name prefix for a given table
 235          * @param string $table
 236          * @throws MWException
 237          * @return null|string
 238          */
 239         protected function getPrefix( $table ) {
 240                 static $prefixes = [
 241                         'pagelinks' => 'pl',
 242                         'imagelinks' => 'il',
 243                         'categorylinks' => 'cl',
 244                         'templatelinks' => 'tl',
 245                         'redirect' => 'rd',
 246                 ];
 247
 248                 if ( isset( $prefixes[$table] ) ) {
 249                         return $prefixes[$table];
 250                 } else {
 251                         $prefix = null;
 252                         Hooks::run( 'BacklinkCacheGetPrefix', [ $table, &$prefix ] );
 253                         if ( $prefix ) {
 254                                 return $prefix;
 255                         } else {
 256                                 throw new MWException( "Invalid table \"$table\" in " . __CLASS__ );
 257                         }
 258                 }
 259         }
 260
 261         /**
 262          * Get the SQL condition array for selecting backlinks, with a join
 263          * on the page table.
 264          * @param string $table
 265          * @throws MWException
 266          * @return array|null
 267          */
 268         protected function getConditions( $table ) {
 269                 $prefix = $this->getPrefix( $table );
 270
 271                 switch ( $table ) {
 272                         case 'pagelinks':
 273                         case 'templatelinks':
 274                                 $conds = [
 275                                         "{$prefix}_namespace" => $this->title->getNamespace(),
 276                                         "{$prefix}_title" => $this->title->getDBkey(),
 277                                         "page_id={$prefix}_from"
 278                                 ];
 279                                 break;
 280                         case 'redirect':
 281                                 $conds = [
 282                                         "{$prefix}_namespace" => $this->title->getNamespace(),
 283                                         "{$prefix}_title" => $this->title->getDBkey(),
 284                                         $this->getDB()->makeList( [
 285                                                 "{$prefix}_interwiki" => '',
 286                                                 "{$prefix}_interwiki IS NULL",
 287                                         ], LIST_OR ),
 288                                         "page_id={$prefix}_from"
 289                                 ];
 290                                 break;
 291                         case 'imagelinks':
 292                         case 'categorylinks':
 293                                 $conds = [
 294                                         "{$prefix}_to" => $this->title->getDBkey(),
 295                                         "page_id={$prefix}_from"
 296                                 ];
 297                                 break;
 298                         default:
 299                                 $conds = null;
 300                                 Hooks::run( 'BacklinkCacheGetConditions', [ $table, $this->title, &$conds ] );
 301                                 if ( !$conds ) {
 302                                         throw new MWException( "Invalid table \"$table\" in " . __CLASS__ );
 303                                 }
 304                 }
 305
 306                 return $conds;
 307         }
 308
 309         /**
 310          * Check if there are any backlinks
 311          * @param string $table
 312          * @return bool
 313          */
 314         public function hasLinks( $table ) {
 315                 return ( $this->getNumLinks( $table, 1 ) > 0 );
 316         }
 317
 318         /**
 319          * Get the approximate number of backlinks
 320          * @param string $table
 321          * @param int $max Only count up to this many backlinks
 322          * @return int
 323          */
 324         public function getNumLinks( $table, $max = INF ) {
 325                 global $wgUpdateRowsPerJob;
 326
 327                 $cache = ObjectCache::getMainWANInstance();
 328                 // 1) try partition cache ...
 329                 if ( isset( $this->partitionCache[$table] ) ) {
 330                         $entry = reset( $this->partitionCache[$table] );
 331
 332                         return min( $max, $entry['numRows'] );
 333                 }
 334
 335                 // 2) ... then try full result cache ...
 336                 if ( isset( $this->fullResultCache[$table] ) ) {
 337                         return min( $max, $this->fullResultCache[$table]->numRows() );
 338                 }
 339
 340                 $memcKey = $cache->makeKey(
 341                         'numbacklinks',
 342                         md5( $this->title->getPrefixedDBkey() ),
 343                         $table
 344                 );
 345
 346                 // 3) ... fallback to memcached ...
 347                 $count = $cache->get( $memcKey );
 348                 if ( $count ) {
 349                         return min( $max, $count );
 350                 }
 351
 352                 // 4) fetch from the database ...
 353                 if ( is_infinite( $max ) ) { // no limit at all
 354                         // Use partition() since it will batch the query and skip the JOIN.
 355                         // Use $wgUpdateRowsPerJob just to encourage cache reuse for jobs.
 356                         $this->partition( $table, $wgUpdateRowsPerJob ); // updates $this->partitionCache
 357                         return $this->partitionCache[$table][$wgUpdateRowsPerJob]['numRows'];
 358                 } else { // probably some sane limit
 359                         // Fetch the full title info, since the caller will likely need it next
 360                         $count = $this->getLinks( $table, false, false, $max )->count();
 361                         if ( $count < $max ) { // full count
 362                                 $cache->set( $memcKey, $count, self::CACHE_EXPIRY );
 363                         }
 364                 }
 365
 366                 return min( $max, $count );
 367         }
 368
 369         /**
 370          * Partition the backlinks into batches.
 371          * Returns an array giving the start and end of each range. The first
 372          * batch has a start of false, and the last batch has an end of false.
 373          *
 374          * @param string $table The links table name
 375          * @param int $batchSize
 376          * @return array
 377          */
 378         public function partition( $table, $batchSize ) {
 379                 // 1) try partition cache ...
 380                 if ( isset( $this->partitionCache[$table][$batchSize] ) ) {
 381                         wfDebug( __METHOD__ . ": got from partition cache\n" );
 382
 383                         return $this->partitionCache[$table][$batchSize]['batches'];
 384                 }
 385
 386                 $cache = ObjectCache::getMainWANInstance();
 387                 $this->partitionCache[$table][$batchSize] = false;
 388                 $cacheEntry =& $this->partitionCache[$table][$batchSize];
 389
 390                 // 2) ... then try full result cache ...
 391                 if ( isset( $this->fullResultCache[$table] ) ) {
 392                         $cacheEntry = $this->partitionResult( $this->fullResultCache[$table], $batchSize );
 393                         wfDebug( __METHOD__ . ": got from full result cache\n" );
 394
 395                         return $cacheEntry['batches'];
 396                 }
 397
 398                 $memcKey = $cache->makeKey(
 399                         'backlinks',
 400                         md5( $this->title->getPrefixedDBkey() ),
 401                         $table,
 402                         $batchSize
 403                 );
 404
 405                 // 3) ... fallback to memcached ...
 406                 $memcValue = $cache->get( $memcKey );
 407                 if ( is_array( $memcValue ) ) {
 408                         $cacheEntry = $memcValue;
 409                         wfDebug( __METHOD__ . ": got from memcached $memcKey\n" );
 410
 411                         return $cacheEntry['batches'];
 412                 }
 413
 414                 // 4) ... finally fetch from the slow database :(
 415                 $cacheEntry = [ 'numRows' => 0, 'batches' => [] ]; // final result
 416                 // Do the selects in batches to avoid client-side OOMs (T45452).
 417                 // Use a LIMIT that plays well with $batchSize to keep equal sized partitions.
 418                 $selectSize = max( $batchSize, 200000 - ( 200000 % $batchSize ) );
 419                 $start = false;
 420                 do {
 421                         $res = $this->queryLinks( $table, $start, false, $selectSize, 'ids' );
 422                         $partitions = $this->partitionResult( $res, $batchSize, false );
 423                         // Merge the link count and range partitions for this chunk
 424                         $cacheEntry['numRows'] += $partitions['numRows'];
 425                         $cacheEntry['batches'] = array_merge( $cacheEntry['batches'], $partitions['batches'] );
 426                         if ( count( $partitions['batches'] ) ) {
 427                                 list( , $lEnd ) = end( $partitions['batches'] );
 428                                 $start = $lEnd + 1; // pick up after this inclusive range
 429                         }
 430                 } while ( $partitions['numRows'] >= $selectSize );
 431                 // Make sure the first range has start=false and the last one has end=false
 432                 if ( count( $cacheEntry['batches'] ) ) {
 433                         $cacheEntry['batches'][0][0] = false;
 434                         $cacheEntry['batches'][count( $cacheEntry['batches'] ) - 1][1] = false;
 435                 }
 436
 437                 // Save partitions to memcached
 438                 $cache->set( $memcKey, $cacheEntry, self::CACHE_EXPIRY );
 439
 440                 // Save backlink count to memcached
 441                 $memcKey = $cache->makeKey(
 442                         'numbacklinks',
 443                         md5( $this->title->getPrefixedDBkey() ),
 444                         $table
 445                 );
 446                 $cache->set( $memcKey, $cacheEntry['numRows'], self::CACHE_EXPIRY );
 447
 448                 wfDebug( __METHOD__ . ": got from database\n" );
 449
 450                 return $cacheEntry['batches'];
 451         }
 452
 453         /**
 454          * Partition a DB result with backlinks in it into batches
 455          * @param ResultWrapper $res Database result
 456          * @param int $batchSize
 457          * @param bool $isComplete Whether $res includes all the backlinks
 458          * @throws MWException
 459          * @return array
 460          */
 461         protected function partitionResult( $res, $batchSize, $isComplete = true ) {
 462                 $batches = [];
 463                 $numRows = $res->numRows();
 464                 $numBatches = ceil( $numRows / $batchSize );
 465
 466                 for ( $i = 0; $i < $numBatches; $i++ ) {
 467                         if ( $i == 0 && $isComplete ) {
 468                                 $start = false;
 469                         } else {
 470                                 $rowNum = $i * $batchSize;
 471                                 $res->seek( $rowNum );
 472                                 $row = $res->fetchObject();
 473                                 $start = (int)$row->page_id;
 474                         }
 475
 476                         if ( $i == ( $numBatches - 1 ) && $isComplete ) {
 477                                 $end = false;
 478                         } else {
 479                                 $rowNum = min( $numRows - 1, ( $i + 1 ) * $batchSize - 1 );
 480                                 $res->seek( $rowNum );
 481                                 $row = $res->fetchObject();
 482                                 $end = (int)$row->page_id;
 483                         }
 484
 485                         # Sanity check order
 486                         if ( $start && $end && $start > $end ) {
 487                                 throw new MWException( __METHOD__ . ': Internal error: query result out of order' );
 488                         }
 489
 490                         $batches[] = [ $start, $end ];
 491                 }
 492
 493                 return [ 'numRows' => $numRows, 'batches' => $batches ];
 494         }
 495
 496         /**
 497          * Get a Title iterator for cascade-protected template/file use backlinks
 498          *
 499          * @return TitleArray
 500          * @since 1.25
 501          */
 502         public function getCascadeProtectedLinks() {
 503                 $dbr = $this->getDB();
 504
 505                 // @todo: use UNION without breaking tests that use temp tables
 506                 $resSets = [];
 507                 $resSets[] = $dbr->select(
 508                         [ 'templatelinks', 'page_restrictions', 'page' ],
 509                         [ 'page_namespace', 'page_title', 'page_id' ],
 510                         [
 511                                 'tl_namespace' => $this->title->getNamespace(),
 512                                 'tl_title' => $this->title->getDBkey(),
 513                                 'tl_from = pr_page',
 514                                 'pr_cascade' => 1,
 515                                 'page_id = tl_from'
 516                         ],
 517                         __METHOD__,
 518                         [ 'DISTINCT' ]
 519                 );
 520                 if ( $this->title->getNamespace() == NS_FILE ) {
 521                         $resSets[] = $dbr->select(
 522                                 [ 'imagelinks', 'page_restrictions', 'page' ],
 523                                 [ 'page_namespace', 'page_title', 'page_id' ],
 524                                 [
 525                                         'il_to' => $this->title->getDBkey(),
 526                                         'il_from = pr_page',
 527                                         'pr_cascade' => 1,
 528                                         'page_id = il_from'
 529                                 ],
 530                                 __METHOD__,
 531                                 [ 'DISTINCT' ]
 532                         );
 533                 }
 534
 535                 // Combine and de-duplicate the results
 536                 $mergedRes = [];
 537                 foreach ( $resSets as $res ) {
 538                         foreach ( $res as $row ) {
 539                                 $mergedRes[$row->page_id] = $row;
 540                         }
 541                 }
 542
 543                 return TitleArray::newFromResult(
 544                         new FakeResultWrapper( array_values( $mergedRes ) ) );
 545         }
 546 }