includes/BacklinkCache.php

   1 <?php
   2 /**
   3  * Class for fetching backlink lists, approximate backlink counts and partitions.
   4  * Instances of this class should typically be fetched with $title->getBacklinkCache().
   5  *
   6  * Ideally you should only get your backlinks from here when you think there is some
   7  * advantage in caching them. Otherwise it's just a waste of memory.
   8  */
   9 class BacklinkCache {
  10         var $partitionCache = array();
  11         var $fullResultCache = array();
  12         var $title;
  13         var $db;
  14
  15         const CACHE_EXPIRY = 3600;
  16
  17         /**
  18          * Create a new BacklinkCache
  19          */
  20         function __construct( $title ) {
  21                 $this->title = $title;
  22         }
  23
  24         /**
  25          * Clear locally stored data
  26          */
  27         function clear() {
  28                 $this->partitionCache = array();
  29                 $this->fullResultCache = array();
  30                 unset( $this->db );
  31         }
  32
  33         /**
  34          * Set the Database object to use
  35          */
  36         public function setDB( $db ) {
  37                 $this->db = $db;
  38         }
  39
  40         protected function getDB() {
  41                 if ( !isset( $this->db ) ) {
  42                         $this->db = wfGetDB( DB_SLAVE );
  43                 }
  44                 return $this->db;
  45         }
  46
  47         /**
  48          * Get the backlinks for a given table. Cached in process memory only.
  49          * @param $table String
  50          * @param $startId Integer or false
  51          * @param $endId Integer or false
  52          * @return TitleArray
  53          */
  54         public function getLinks( $table, $startId = false, $endId = false ) {
  55                 wfProfileIn( __METHOD__ );
  56
  57                 $fromField = $this->getPrefix( $table ) . '_from';
  58
  59                 if ( $startId || $endId ) {
  60                         // Partial range, not cached
  61                         wfDebug( __METHOD__ . ": from DB (uncacheable range)\n" );
  62                         $conds = $this->getConditions( $table );
  63                         // Use the from field in the condition rather than the joined page_id,
  64                         // because databases are stupid and don't necessarily propagate indexes.
  65                         if ( $startId ) {
  66                                 $conds[] = "$fromField >= " . intval( $startId );
  67                         }
  68                         if ( $endId ) {
  69                                 $conds[] = "$fromField <= " . intval( $endId );
  70                         }
  71                         $res = $this->getDB()->select(
  72                                 array( $table, 'page' ),
  73                                 array( 'page_namespace', 'page_title', 'page_id' ),
  74                                 $conds,
  75                                 __METHOD__,
  76                                 array(
  77                                         'STRAIGHT_JOIN',
  78                                         'ORDER BY' => $fromField
  79                                 ) );
  80                         $ta = TitleArray::newFromResult( $res );
  81                         wfProfileOut( __METHOD__ );
  82                         return $ta;
  83                 }
  84
  85                 if ( !isset( $this->fullResultCache[$table] ) ) {
  86                         wfDebug( __METHOD__ . ": from DB\n" );
  87                         $res = $this->getDB()->select(
  88                                 array( $table, 'page' ),
  89                                 array( 'page_namespace', 'page_title', 'page_id' ),
  90                                 $this->getConditions( $table ),
  91                                 __METHOD__,
  92                                 array(
  93                                         'STRAIGHT_JOIN',
  94                                         'ORDER BY' => $fromField,
  95                                 ) );
  96                         $this->fullResultCache[$table] = $res;
  97                 }
  98                 $ta = TitleArray::newFromResult( $this->fullResultCache[$table] );
  99                 wfProfileOut( __METHOD__ );
 100                 return $ta;
 101         }
 102
 103         /**
 104          * Get the field name prefix for a given table
 105          */
 106         protected function getPrefix( $table ) {
 107                 static $prefixes = array(
 108                         'pagelinks' => 'pl',
 109                         'imagelinks' => 'il',
 110                         'categorylinks' => 'cl',
 111                         'templatelinks' => 'tl',
 112                         'redirect' => 'rd',
 113                 );
 114
 115                 if ( isset( $prefixes[$table] ) ) {
 116                         return $prefixes[$table];
 117                 } else {
 118                         throw new MWException( "Invalid table \"$table\" in " . __CLASS__ );
 119                 }
 120         }
 121
 122         /**
 123          * Get the SQL condition array for selecting backlinks, with a join on the page table
 124          */
 125         protected function getConditions( $table ) {
 126                 $prefix = $this->getPrefix( $table );
 127
 128                 switch ( $table ) {
 129                         case 'pagelinks':
 130                         case 'templatelinks':
 131                         case 'redirect':
 132                                 $conds = array(
 133                                         "{$prefix}_namespace" => $this->title->getNamespace(),
 134                                         "{$prefix}_title" => $this->title->getDBkey(),
 135                                         "page_id={$prefix}_from"
 136                                 );
 137                                 break;
 138                         case 'imagelinks':
 139                                 $conds = array(
 140                                         'il_to' => $this->title->getDBkey(),
 141                                         'page_id=il_from'
 142                                 );
 143                                 break;
 144                         case 'categorylinks':
 145                                 $conds = array(
 146                                         'cl_to' => $this->title->getDBkey(),
 147                                         'page_id=cl_from',
 148                                 );
 149                                 break;
 150                         default:
 151                                 throw new MWException( "Invalid table \"$table\" in " . __CLASS__ );
 152                 }
 153                 return $conds;
 154         }
 155
 156         /**
 157          * Get the approximate number of backlinks
 158          */
 159         public function getNumLinks( $table ) {
 160                 if ( isset( $this->fullResultCache[$table] ) ) {
 161                         return $this->fullResultCache[$table]->numRows();
 162                 }
 163
 164                 if ( isset( $this->partitionCache[$table] ) ) {
 165                         $entry = reset( $this->partitionCache[$table] );
 166                         return $entry['numRows'];
 167                 }
 168
 169                 $titleArray = $this->getLinks( $table );
 170                 return $titleArray->count();
 171         }
 172
 173         /**
 174          * Partition the backlinks into batches.
 175          * Returns an array giving the start and end of each range. The first batch has
 176          * a start of false, and the last batch has an end of false.
 177          *
 178          * @param $table String: the links table name
 179          * @param $batchSize Integer
 180          * @return Array
 181          */
 182         public function partition( $table, $batchSize ) {
 183                 // Try cache
 184                 if ( isset( $this->partitionCache[$table][$batchSize] ) ) {
 185                         wfDebug( __METHOD__ . ": got from partition cache\n" );
 186                         return $this->partitionCache[$table][$batchSize]['batches'];
 187                 }
 188
 189                 $this->partitionCache[$table][$batchSize] = false;
 190                 $cacheEntry =& $this->partitionCache[$table][$batchSize];
 191
 192                 // Try full result cache
 193                 if ( isset( $this->fullResultCache[$table] ) ) {
 194                         $cacheEntry = $this->partitionResult( $this->fullResultCache[$table], $batchSize );
 195                         wfDebug( __METHOD__ . ": got from full result cache\n" );
 196                         return $cacheEntry['batches'];
 197                 }
 198
 199                 // Try memcached
 200                 global $wgMemc;
 201                 $memcKey = wfMemcKey(
 202                         'backlinks',
 203                         md5( $this->title->getPrefixedDBkey() ),
 204                         $table,
 205                         $batchSize
 206                 );
 207                 $memcValue = $wgMemc->get( $memcKey );
 208
 209                 if ( is_array( $memcValue ) ) {
 210                         $cacheEntry = $memcValue;
 211                         wfDebug( __METHOD__ . ": got from memcached $memcKey\n" );
 212                         return $cacheEntry['batches'];
 213                 }
 214                 // Fetch from database
 215                 $this->getLinks( $table );
 216                 $cacheEntry = $this->partitionResult( $this->fullResultCache[$table], $batchSize );
 217                 // Save to memcached
 218                 $wgMemc->set( $memcKey, $cacheEntry, self::CACHE_EXPIRY );
 219                 wfDebug( __METHOD__ . ": got from database\n" );
 220                 return $cacheEntry['batches'];
 221         }
 222
 223         /**
 224          * Partition a DB result with backlinks in it into batches
 225          */
 226         protected function partitionResult( $res, $batchSize ) {
 227                 $batches = array();
 228                 $numRows = $res->numRows();
 229                 $numBatches = ceil( $numRows / $batchSize );
 230
 231                 for ( $i = 0; $i < $numBatches; $i++ ) {
 232                         if ( $i == 0  ) {
 233                                 $start = false;
 234                         } else {
 235                                 $rowNum = intval( $numRows * $i / $numBatches );
 236                                 $res->seek( $rowNum );
 237                                 $row = $res->fetchObject();
 238                                 $start = $row->page_id;
 239                         }
 240
 241                         if ( $i == $numBatches - 1 ) {
 242                                 $end = false;
 243                         } else {
 244                                 $rowNum = intval( $numRows * ( $i + 1 ) / $numBatches );
 245                                 $res->seek( $rowNum );
 246                                 $row = $res->fetchObject();
 247                                 $end = $row->page_id - 1;
 248                         }
 249
 250                         # Sanity check order
 251                         if ( $start && $end && $start > $end ) {
 252                                 throw new MWException( __METHOD__ . ': Internal error: query result out of order' );
 253                         }
 254
 255                         $batches[] = array( $start, $end );
 256                 }
 257                 return array( 'numRows' => $numRows, 'batches' => $batches );
 258         }
 259 }