includes/BacklinkCache.php

   1 <?php
   2 /**
   3  * File for BacklinkCache class
   4  * @file
   5  */
   6
   7 /**
   8  * Class for fetching backlink lists, approximate backlink counts and
   9  * partitions. This is a shared cache.
  10  *
  11  * Instances of this class should typically be fetched with the method
  12  * $title->getBacklinkCache().
  13  *
  14  * Ideally you should only get your backlinks from here when you think
  15  * there is some advantage in caching them. Otherwise it's just a waste
  16  * of memory.
  17  *
  18  * Introduced by r47317
  19  *
  20  * @internal documentation reviewed on 18 Mar 2011 by hashar
  21  *
  22  * @author Tim Starling
  23  * @copyright © 2009, Tim Starling, Domas Mituzas
  24  * @copyright © 2010, Max Sem
  25  * @copyright © 2011, Ashar Voultoiz
  26  */
  27 class BacklinkCache {
  28
  29         /**
  30          * Multi dimensions array representing batches. Keys are:
  31          *  > (string) links table name
  32          *    > 'numRows' : Number of rows for this link table
  33          *    > 'batches' : array( $start, $end )
  34          *
  35          * @see BacklinkCache::partitionResult()
  36          *
  37          * Cleared with BacklinkCache::clear()
  38          */
  39         protected $partitionCache = array();
  40
  41         /**
  42          * Contains the whole links from a database result.
  43          * This is raw data that will be partitioned in $partitionCache
  44          *
  45          * Initialized with BacklinkCache::getLinks()
  46          * Cleared with BacklinkCache::clear()
  47          */
  48         protected $fullResultCache = array();
  49
  50         /**
  51          * Local copy of a database object.
  52          *
  53          * Accessor: BacklinkCache::getDB()
  54          * Mutator : BacklinkCache::setDB()
  55          * Cleared with BacklinkCache::clear()
  56          */
  57         protected $db;
  58
  59         /**
  60          * Local copy of a Title object
  61          */
  62         protected $title;
  63
  64         const CACHE_EXPIRY = 3600;
  65
  66         /**
  67          * Create a new BacklinkCache
  68          * @param Title $title : Title object to create a backlink cache for.
  69          */
  70         function __construct( $title ) {
  71                 $this->title = $title;
  72         }
  73
  74         /**
  75          * Serialization handler, diasallows to serialize the database to prevent
  76          * failures after this class is deserialized from cache with dead DB
  77          * connection.
  78          */
  79         function __sleep() {
  80                 return array( 'partitionCache', 'fullResultCache', 'title' );
  81         }
  82
  83         /**
  84          * Clear locally stored data and database object.
  85          */
  86         public function clear() {
  87                 $this->partitionCache = array();
  88                 $this->fullResultCache = array();
  89                 unset( $this->db );
  90         }
  91
  92         /**
  93          * Set the Database object to use
  94          *
  95          * @param $db DatabaseBase
  96          */
  97         public function setDB( $db ) {
  98                 $this->db = $db;
  99         }
 100
 101         /**
 102          * Get the slave connection to the database
 103          * When non existing, will initialize the connection.
 104          * @return Database object
 105          */
 106         protected function getDB() {
 107                 if ( !isset( $this->db ) ) {
 108                         $this->db = wfGetDB( DB_SLAVE );
 109                 }
 110
 111                 return $this->db;
 112         }
 113
 114         /**
 115          * Get the backlinks for a given table. Cached in process memory only.
 116          * @param $table String
 117          * @param $startId Integer or false
 118          * @param $endId Integer or false
 119          * @return TitleArrayFromResult
 120          */
 121         public function getLinks( $table, $startId = false, $endId = false ) {
 122                 wfProfileIn( __METHOD__ );
 123
 124                 $fromField = $this->getPrefix( $table ) . '_from';
 125
 126                 if ( $startId || $endId ) {
 127                         // Partial range, not cached
 128                         wfDebug( __METHOD__ . ": from DB (uncacheable range)\n" );
 129                         $conds = $this->getConditions( $table );
 130
 131                         // Use the from field in the condition rather than the joined page_id,
 132                         // because databases are stupid and don't necessarily propagate indexes.
 133                         if ( $startId ) {
 134                                 $conds[] = "$fromField >= " . intval( $startId );
 135                         }
 136
 137                         if ( $endId ) {
 138                                 $conds[] = "$fromField <= " . intval( $endId );
 139                         }
 140
 141                         $res = $this->getDB()->select(
 142                                 array( $table, 'page' ),
 143                                 array( 'page_namespace', 'page_title', 'page_id' ),
 144                                 $conds,
 145                                 __METHOD__,
 146                                 array(
 147                                         'STRAIGHT_JOIN',
 148                                         'ORDER BY' => $fromField
 149                                 ) );
 150                         $ta = TitleArray::newFromResult( $res );
 151
 152                         wfProfileOut( __METHOD__ );
 153                         return $ta;
 154                 }
 155
 156                 // @todo FIXME: Make this a function?
 157                 if ( !isset( $this->fullResultCache[$table] ) ) {
 158                         wfDebug( __METHOD__ . ": from DB\n" );
 159                         $res = $this->getDB()->select(
 160                                 array( $table, 'page' ),
 161                                 array( 'page_namespace', 'page_title', 'page_id' ),
 162                                 $this->getConditions( $table ),
 163                                 __METHOD__,
 164                                 array(
 165                                         'STRAIGHT_JOIN',
 166                                         'ORDER BY' => $fromField,
 167                                 ) );
 168                         $this->fullResultCache[$table] = $res;
 169                 }
 170
 171                 $ta = TitleArray::newFromResult( $this->fullResultCache[$table] );
 172
 173                 wfProfileOut( __METHOD__ );
 174                 return $ta;
 175         }
 176
 177         /**
 178          * Get the field name prefix for a given table
 179          * @param $table String
 180          */
 181         protected function getPrefix( $table ) {
 182                 static $prefixes = array(
 183                         'pagelinks'     => 'pl',
 184                         'imagelinks'    => 'il',
 185                         'categorylinks' => 'cl',
 186                         'templatelinks' => 'tl',
 187                         'redirect'      => 'rd',
 188                 );
 189
 190                 if ( isset( $prefixes[$table] ) ) {
 191                         return $prefixes[$table];
 192                 } else {
 193                         throw new MWException( "Invalid table \"$table\" in " . __CLASS__ );
 194                 }
 195         }
 196
 197         /**
 198          * Get the SQL condition array for selecting backlinks, with a join
 199          * on the page table.
 200          * @param $table String
 201          */
 202         protected function getConditions( $table ) {
 203                 $prefix = $this->getPrefix( $table );
 204
 205                 // @todo FIXME: imagelinks and categorylinks do not rely on getNamespace,
 206                 // they could be moved up for nicer case statements
 207                 switch ( $table ) {
 208                         case 'pagelinks':
 209                         case 'templatelinks':
 210                         case 'redirect':
 211                                 $conds = array(
 212                                         "{$prefix}_namespace" => $this->title->getNamespace(),
 213                                         "{$prefix}_title"     => $this->title->getDBkey(),
 214                                         "page_id={$prefix}_from"
 215                                 );
 216                                 break;
 217                         case 'imagelinks':
 218                                 $conds = array(
 219                                         'il_to' => $this->title->getDBkey(),
 220                                         'page_id=il_from'
 221                                 );
 222                                 break;
 223                         case 'categorylinks':
 224                                 $conds = array(
 225                                         'cl_to' => $this->title->getDBkey(),
 226                                         'page_id=cl_from',
 227                                 );
 228                                 break;
 229                         default:
 230                                 throw new MWException( "Invalid table \"$table\" in " . __CLASS__ );
 231                 }
 232
 233                 return $conds;
 234         }
 235
 236         /**
 237          * Get the approximate number of backlinks
 238          * @param $table String
 239          * @return integer
 240          */
 241         public function getNumLinks( $table ) {
 242                 if ( isset( $this->fullResultCache[$table] ) ) {
 243                         return $this->fullResultCache[$table]->numRows();
 244                 }
 245
 246                 if ( isset( $this->partitionCache[$table] ) ) {
 247                         $entry = reset( $this->partitionCache[$table] );
 248                         return $entry['numRows'];
 249                 }
 250
 251                 $titleArray = $this->getLinks( $table );
 252
 253                 return $titleArray->count();
 254         }
 255
 256         /**
 257          * Partition the backlinks into batches.
 258          * Returns an array giving the start and end of each range. The firsti
 259          * batch has a start of false, and the last batch has an end of false.
 260          *
 261          * @param $table String: the links table name
 262          * @param $batchSize Integer
 263          * @return Array
 264          */
 265         public function partition( $table, $batchSize ) {
 266
 267                 // 1) try this per process cache first
 268
 269                 if ( isset( $this->partitionCache[$table][$batchSize] ) ) {
 270                         wfDebug( __METHOD__ . ": got from partition cache\n" );
 271                         return $this->partitionCache[$table][$batchSize]['batches'];
 272                 }
 273
 274                 $this->partitionCache[$table][$batchSize] = false;
 275                 $cacheEntry =& $this->partitionCache[$table][$batchSize];
 276
 277                 // 2) try full result cache
 278
 279                 if ( isset( $this->fullResultCache[$table] ) ) {
 280                         $cacheEntry = $this->partitionResult( $this->fullResultCache[$table], $batchSize );
 281                         wfDebug( __METHOD__ . ": got from full result cache\n" );
 282
 283                         return $cacheEntry['batches'];
 284                 }
 285
 286                 // 3) ... fallback to memcached ...
 287
 288                 global $wgMemc;
 289
 290                 $memcKey = wfMemcKey(
 291                         'backlinks',
 292                         md5( $this->title->getPrefixedDBkey() ),
 293                         $table,
 294                         $batchSize
 295                 );
 296
 297                 $memcValue = $wgMemc->get( $memcKey );
 298
 299                 if ( is_array( $memcValue ) ) {
 300                         $cacheEntry = $memcValue;
 301                         wfDebug( __METHOD__ . ": got from memcached $memcKey\n" );
 302
 303                         return $cacheEntry['batches'];
 304                 }
 305
 306
 307                 // 4) ... finally fetch from the slow database :(
 308
 309                 $this->getLinks( $table );
 310                 $cacheEntry = $this->partitionResult( $this->fullResultCache[$table], $batchSize );
 311                 // Save to memcached
 312                 $wgMemc->set( $memcKey, $cacheEntry, self::CACHE_EXPIRY );
 313
 314                 wfDebug( __METHOD__ . ": got from database\n" );
 315                 return $cacheEntry['batches'];
 316         }
 317
 318         /**
 319          * Partition a DB result with backlinks in it into batches
 320          * @param $res ResultWrapper database result
 321          * @param $batchSize integer
 322          * @return array @see
 323          */
 324         protected function partitionResult( $res, $batchSize ) {
 325                 $batches = array();
 326                 $numRows = $res->numRows();
 327                 $numBatches = ceil( $numRows / $batchSize );
 328
 329                 for ( $i = 0; $i < $numBatches; $i++ ) {
 330                         if ( $i == 0  ) {
 331                                 $start = false;
 332                         } else {
 333                                 $rowNum = intval( $numRows * $i / $numBatches );
 334                                 $res->seek( $rowNum );
 335                                 $row = $res->fetchObject();
 336                                 $start = $row->page_id;
 337                         }
 338
 339                         if ( $i == $numBatches - 1 ) {
 340                                 $end = false;
 341                         } else {
 342                                 $rowNum = intval( $numRows * ( $i + 1 ) / $numBatches );
 343                                 $res->seek( $rowNum );
 344                                 $row = $res->fetchObject();
 345                                 $end = $row->page_id - 1;
 346                         }
 347
 348                         # Sanity check order
 349                         if ( $start && $end && $start > $end ) {
 350                                 throw new MWException( __METHOD__ . ': Internal error: query result out of order' );
 351                         }
 352
 353                         $batches[] = array( $start, $end );
 354                 }
 355
 356                 return array( 'numRows' => $numRows, 'batches' => $batches );
 357         }
 358 }