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, Antoine Musso
  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          * @return array
  80          */
  81         function __sleep() {
  82                 return array( 'partitionCache', 'fullResultCache', 'title' );
  83         }
  84
  85         /**
  86          * Clear locally stored data and database object.
  87          */
  88         public function clear() {
  89                 $this->partitionCache = array();
  90                 $this->fullResultCache = array();
  91                 unset( $this->db );
  92         }
  93
  94         /**
  95          * Set the Database object to use
  96          *
  97          * @param $db DatabaseBase
  98          */
  99         public function setDB( $db ) {
 100                 $this->db = $db;
 101         }
 102
 103         /**
 104          * Get the slave connection to the database
 105          * When non existing, will initialize the connection.
 106          * @return DatabaseBase object
 107          */
 108         protected function getDB() {
 109                 if ( !isset( $this->db ) ) {
 110                         $this->db = wfGetDB( DB_SLAVE );
 111                 }
 112
 113                 return $this->db;
 114         }
 115
 116         /**
 117          * Get the backlinks for a given table. Cached in process memory only.
 118          * @param $table String
 119          * @param $startId Integer or false
 120          * @param $endId Integer or false
 121          * @return TitleArrayFromResult
 122          */
 123         public function getLinks( $table, $startId = false, $endId = false ) {
 124                 wfProfileIn( __METHOD__ );
 125
 126                 $fromField = $this->getPrefix( $table ) . '_from';
 127
 128                 if ( $startId || $endId ) {
 129                         // Partial range, not cached
 130                         wfDebug( __METHOD__ . ": from DB (uncacheable range)\n" );
 131                         $conds = $this->getConditions( $table );
 132
 133                         // Use the from field in the condition rather than the joined page_id,
 134                         // because databases are stupid and don't necessarily propagate indexes.
 135                         if ( $startId ) {
 136                                 $conds[] = "$fromField >= " . intval( $startId );
 137                         }
 138
 139                         if ( $endId ) {
 140                                 $conds[] = "$fromField <= " . intval( $endId );
 141                         }
 142
 143                         $res = $this->getDB()->select(
 144                                 array( $table, 'page' ),
 145                                 array( 'page_namespace', 'page_title', 'page_id' ),
 146                                 $conds,
 147                                 __METHOD__,
 148                                 array(
 149                                         'STRAIGHT_JOIN',
 150                                         'ORDER BY' => $fromField
 151                                 ) );
 152                         $ta = TitleArray::newFromResult( $res );
 153
 154                         wfProfileOut( __METHOD__ );
 155                         return $ta;
 156                 }
 157
 158                 // @todo FIXME: Make this a function?
 159                 if ( !isset( $this->fullResultCache[$table] ) ) {
 160                         wfDebug( __METHOD__ . ": from DB\n" );
 161                         $res = $this->getDB()->select(
 162                                 array( $table, 'page' ),
 163                                 array( 'page_namespace', 'page_title', 'page_id' ),
 164                                 $this->getConditions( $table ),
 165                                 __METHOD__,
 166                                 array(
 167                                         'STRAIGHT_JOIN',
 168                                         'ORDER BY' => $fromField,
 169                                 ) );
 170                         $this->fullResultCache[$table] = $res;
 171                 }
 172
 173                 $ta = TitleArray::newFromResult( $this->fullResultCache[$table] );
 174
 175                 wfProfileOut( __METHOD__ );
 176                 return $ta;
 177         }
 178
 179         /**
 180          * Get the field name prefix for a given table
 181          * @param $table String
 182          * @return null|string
 183          */
 184         protected function getPrefix( $table ) {
 185                 static $prefixes = array(
 186                         'pagelinks'     => 'pl',
 187                         'imagelinks'    => 'il',
 188                         'categorylinks' => 'cl',
 189                         'templatelinks' => 'tl',
 190                         'redirect'      => 'rd',
 191                 );
 192
 193                 if ( isset( $prefixes[$table] ) ) {
 194                         return $prefixes[$table];
 195                 } else {
 196                         $prefix = null;
 197                         wfRunHooks( 'BacklinkCacheGetPrefix', array( $table, &$prefix ) );
 198                         if( $prefix ) {
 199                                 return $prefix;
 200                         } else {
 201                                 throw new MWException( "Invalid table \"$table\" in " . __CLASS__ );
 202                         }
 203                 }
 204         }
 205
 206         /**
 207          * Get the SQL condition array for selecting backlinks, with a join
 208          * on the page table.
 209          * @param $table String
 210          * @return array|null
 211          */
 212         protected function getConditions( $table ) {
 213                 $prefix = $this->getPrefix( $table );
 214
 215                 // @todo FIXME: imagelinks and categorylinks do not rely on getNamespace,
 216                 // they could be moved up for nicer case statements
 217                 switch ( $table ) {
 218                         case 'pagelinks':
 219                         case 'templatelinks':
 220                                 $conds = array(
 221                                         "{$prefix}_namespace" => $this->title->getNamespace(),
 222                                         "{$prefix}_title"     => $this->title->getDBkey(),
 223                                         "page_id={$prefix}_from"
 224                                 );
 225                                 break;
 226                         case 'redirect':
 227                                 $conds = array(
 228                                         "{$prefix}_namespace" => $this->title->getNamespace(),
 229                                         "{$prefix}_title"     => $this->title->getDBkey(),
 230                                         $this->getDb()->makeList( array(
 231                                                 "{$prefix}_interwiki = ''",
 232                                                 "{$prefix}_interwiki is null",
 233                                         ), LIST_OR ),
 234                                         "page_id={$prefix}_from"
 235                                 );
 236                                 break;
 237                         case 'imagelinks':
 238                                 $conds = array(
 239                                         'il_to' => $this->title->getDBkey(),
 240                                         'page_id=il_from'
 241                                 );
 242                                 break;
 243                         case 'categorylinks':
 244                                 $conds = array(
 245                                         'cl_to' => $this->title->getDBkey(),
 246                                         'page_id=cl_from',
 247                                 );
 248                                 break;
 249                         default:
 250                                 $conds = null;
 251                                 wfRunHooks( 'BacklinkCacheGetConditions', array( $table, $this->title, &$conds ) );
 252                                 if( !$conds )
 253                                         throw new MWException( "Invalid table \"$table\" in " . __CLASS__ );
 254                 }
 255
 256                 return $conds;
 257         }
 258
 259         /**
 260          * Get the approximate number of backlinks
 261          * @param $table String
 262          * @return integer
 263          */
 264         public function getNumLinks( $table ) {
 265                 if ( isset( $this->fullResultCache[$table] ) ) {
 266                         return $this->fullResultCache[$table]->numRows();
 267                 }
 268
 269                 if ( isset( $this->partitionCache[$table] ) ) {
 270                         $entry = reset( $this->partitionCache[$table] );
 271                         return $entry['numRows'];
 272                 }
 273
 274                 $titleArray = $this->getLinks( $table );
 275
 276                 return $titleArray->count();
 277         }
 278
 279         /**
 280          * Partition the backlinks into batches.
 281          * Returns an array giving the start and end of each range. The first
 282          * batch has a start of false, and the last batch has an end of false.
 283          *
 284          * @param $table String: the links table name
 285          * @param $batchSize Integer
 286          * @return Array
 287          */
 288         public function partition( $table, $batchSize ) {
 289
 290                 // 1) try partition cache ...
 291
 292                 if ( isset( $this->partitionCache[$table][$batchSize] ) ) {
 293                         wfDebug( __METHOD__ . ": got from partition cache\n" );
 294                         return $this->partitionCache[$table][$batchSize]['batches'];
 295                 }
 296
 297                 $this->partitionCache[$table][$batchSize] = false;
 298                 $cacheEntry =& $this->partitionCache[$table][$batchSize];
 299
 300                 // 2) ... then try full result cache ...
 301
 302                 if ( isset( $this->fullResultCache[$table] ) ) {
 303                         $cacheEntry = $this->partitionResult( $this->fullResultCache[$table], $batchSize );
 304                         wfDebug( __METHOD__ . ": got from full result cache\n" );
 305
 306                         return $cacheEntry['batches'];
 307                 }
 308
 309                 // 3) ... fallback to memcached ...
 310
 311                 global $wgMemc;
 312
 313                 $memcKey = wfMemcKey(
 314                         'backlinks',
 315                         md5( $this->title->getPrefixedDBkey() ),
 316                         $table,
 317                         $batchSize
 318                 );
 319
 320                 $memcValue = $wgMemc->get( $memcKey );
 321
 322                 if ( is_array( $memcValue ) ) {
 323                         $cacheEntry = $memcValue;
 324                         wfDebug( __METHOD__ . ": got from memcached $memcKey\n" );
 325
 326                         return $cacheEntry['batches'];
 327                 }
 328
 329
 330                 // 4) ... finally fetch from the slow database :(
 331
 332                 $this->getLinks( $table );
 333                 $cacheEntry = $this->partitionResult( $this->fullResultCache[$table], $batchSize );
 334                 // Save to memcached
 335                 $wgMemc->set( $memcKey, $cacheEntry, self::CACHE_EXPIRY );
 336
 337                 wfDebug( __METHOD__ . ": got from database\n" );
 338                 return $cacheEntry['batches'];
 339         }
 340
 341         /**
 342          * Partition a DB result with backlinks in it into batches
 343          * @param $res ResultWrapper database result
 344          * @param $batchSize integer
 345          * @return array @see
 346          */
 347         protected function partitionResult( $res, $batchSize ) {
 348                 $batches = array();
 349                 $numRows = $res->numRows();
 350                 $numBatches = ceil( $numRows / $batchSize );
 351
 352                 for ( $i = 0; $i < $numBatches; $i++ ) {
 353                         if ( $i == 0  ) {
 354                                 $start = false;
 355                         } else {
 356                                 $rowNum = intval( $numRows * $i / $numBatches );
 357                                 $res->seek( $rowNum );
 358                                 $row = $res->fetchObject();
 359                                 $start = $row->page_id;
 360                         }
 361
 362                         if ( $i == $numBatches - 1 ) {
 363                                 $end = false;
 364                         } else {
 365                                 $rowNum = intval( $numRows * ( $i + 1 ) / $numBatches );
 366                                 $res->seek( $rowNum );
 367                                 $row = $res->fetchObject();
 368                                 $end = $row->page_id - 1;
 369                         }
 370
 371                         # Sanity check order
 372                         if ( $start && $end && $start > $end ) {
 373                                 throw new MWException( __METHOD__ . ': Internal error: query result out of order' );
 374                         }
 375
 376                         $batches[] = array( $start, $end );
 377                 }
 378
 379                 return array( 'numRows' => $numRows, 'batches' => $batches );
 380         }
 381 }