includes/BacklinkCache.php

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