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