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( $table, 'page' ),
  71                                 array( 'page_namespace', 'page_title', 'page_id'),
  72                                 $conds,
  73                                 __METHOD__,
  74                                 array('STRAIGHT_JOIN') );
  75                         $ta = TitleArray::newFromResult( $res );
  76                         wfProfileOut( __METHOD__ );
  77                         return $ta;
  78                 }
  79
  80                 if ( !isset( $this->fullResultCache[$table] ) ) {
  81                         wfDebug( __METHOD__.": from DB\n" );
  82                         $res = $this->getDB()->select(
  83                                 array( $table, 'page' ),
  84                                 array( 'page_namespace', 'page_title', 'page_id' ),
  85                                 $this->getConditions( $table ),
  86                                 __METHOD__,
  87                                 array('STRAIGHT_JOIN') );
  88                         $this->fullResultCache[$table] = $res;
  89                 }
  90                 $ta = TitleArray::newFromResult( $this->fullResultCache[$table] );
  91                 wfProfileOut( __METHOD__ );
  92                 return $ta;
  93         }
  94
  95         /**
  96          * Get the field name prefix for a given table
  97          */
  98         protected function getPrefix( $table ) {
  99                 static $prefixes = array(
 100                         'pagelinks' => 'pl',
 101                         'imagelinks' => 'il',
 102                         'categorylinks' => 'cl',
 103                         'templatelinks' => 'tl',
 104                         'redirect' => 'rd',
 105                 );
 106                 if ( isset( $prefixes[$table] ) ) {
 107                         return $prefixes[$table];
 108                 } else {
 109                         throw new MWException( "Invalid table \"$table\" in " . __CLASS__ );
 110                 }
 111         }
 112
 113         /**
 114          * Get the SQL condition array for selecting backlinks, with a join on the page table
 115          */
 116         protected function getConditions( $table ) {
 117                 $prefix = $this->getPrefix( $table );
 118                 switch ( $table ) {
 119                         case 'pagelinks':
 120                         case 'templatelinks':
 121                         case 'redirect':
 122                                 $conds = array(
 123                                         "{$prefix}_namespace" => $this->title->getNamespace(),
 124                                         "{$prefix}_title" => $this->title->getDBkey(),
 125                                         "page_id={$prefix}_from"
 126                                 );
 127                                 break;
 128                         case 'imagelinks':
 129                                 $conds = array(
 130                                         'il_to' => $this->title->getDBkey(),
 131                                         'page_id=il_from'
 132                                 );
 133                                 break;
 134                         case 'categorylinks':
 135                                 $conds = array(
 136                                         'cl_to' => $this->title->getDBkey(),
 137                                         'page_id=cl_from',
 138                                 );
 139                                 break;
 140                         default:
 141                                 throw new MWException( "Invalid table \"$table\" in " . __CLASS__ );
 142                 }
 143                 return $conds;
 144         }
 145
 146         /**
 147          * Get the approximate number of backlinks
 148          */
 149         public function getNumLinks( $table ) {
 150                 if ( isset( $this->fullResultCache[$table] ) ) {
 151                         return $this->fullResultCache[$table]->numRows();
 152                 }
 153                 if ( isset( $this->partitionCache[$table] ) ) {
 154                         $entry = reset( $this->partitionCache[$table] );
 155                         return $entry['numRows'];
 156                 }
 157                 $titleArray = $this->getLinks( $table );
 158                 return $titleArray->count();
 159         }
 160
 161         /**
 162          * Partition the backlinks into batches.
 163          * Returns an array giving the start and end of each range. The first batch has
 164          * a start of false, and the last batch has an end of false.
 165          *
 166          * @param string $table The links table name
 167          * @param integer $batchSize
 168          * @return array
 169          */
 170         public function partition( $table, $batchSize ) {
 171                 // Try cache
 172                 if ( isset( $this->partitionCache[$table][$batchSize] ) ) {
 173                         wfDebug( __METHOD__.": got from partition cache\n" );
 174                         return $this->partitionCache[$table][$batchSize]['batches'];
 175                 }
 176                 $this->partitionCache[$table][$batchSize] = false;
 177                 $cacheEntry =& $this->partitionCache[$table][$batchSize];
 178
 179                 // Try full result cache
 180                 if ( isset( $this->fullResultCache[$table] ) ) {
 181                         $cacheEntry = $this->partitionResult( $this->fullResultCache[$table], $batchSize );
 182                         wfDebug( __METHOD__.": got from full result cache\n" );
 183                         return $cacheEntry['batches'];
 184                 }
 185                 // Try memcached
 186                 global $wgMemc;
 187                 $memcKey = wfMemcKey( 'backlinks', md5( $this->title->getPrefixedDBkey() ),
 188                         $table, $batchSize );
 189                 $memcValue = $wgMemc->get( $memcKey );
 190                 if ( is_array( $memcValue ) ) {
 191                         $cacheEntry = $memcValue;
 192                         wfDebug( __METHOD__.": got from memcached $memcKey\n" );
 193                         return $cacheEntry['batches'];
 194                 }
 195                 // Fetch from database
 196                 $this->getLinks( $table );
 197                 $cacheEntry = $this->partitionResult( $this->fullResultCache[$table], $batchSize );
 198                 // Save to memcached
 199                 $wgMemc->set( $memcKey, $cacheEntry, self::CACHE_EXPIRY );
 200                 wfDebug( __METHOD__.": got from database\n" );
 201                 return $cacheEntry['batches'];
 202         }
 203
 204         /**
 205          * Partition a DB result with backlinks in it into batches
 206          */
 207         protected function partitionResult( $res, $batchSize ) {
 208                 $batches = array();
 209                 $numRows = $res->numRows();
 210                 $numBatches = ceil( $numRows / $batchSize );
 211                 for ( $i = 0; $i < $numBatches; $i++ ) {
 212                         if ( $i == 0  ) {
 213                                 $start = false;
 214                         } else {
 215                                 $rowNum = intval( $numRows * $i / $numBatches );
 216                                 $res->seek( $rowNum );
 217                                 $row = $res->fetchObject();
 218                                 $start = $row->page_id;
 219                         }
 220                         if ( $i == $numBatches - 1 ) {
 221                                 $end = false;
 222                         } else {
 223                                 $rowNum = intval( $numRows * ( $i + 1 ) / $numBatches );
 224                                 $res->seek( $rowNum );
 225                                 $row = $res->fetchObject();
 226                                 $end = $row->page_id - 1;
 227                         }
 228                         $batches[] = array( $start, $end );
 229                 }
 230                 return array( 'numRows' => $numRows, 'batches' => $batches );
 231         }
 232 }