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 $table String
  51          * @param $startId Integer or false
  52          * @param $endId Integer or false
  53          * @return TitleArray
  54          */
  55         public function getLinks( $table, $startId = false, $endId = false ) {
  56                 wfProfileIn( __METHOD__ );
  57
  58                 $fromField = $this->getPrefix( $table ) . '_from';
  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                 if ( isset( $prefixes[$table] ) ) {
 115                         return $prefixes[$table];
 116                 } else {
 117                         throw new MWException( "Invalid table \"$table\" in " . __CLASS__ );
 118                 }
 119         }
 120
 121         /**
 122          * Get the SQL condition array for selecting backlinks, with a join on the page table
 123          */
 124         protected function getConditions( $table ) {
 125                 $prefix = $this->getPrefix( $table );
 126                 switch ( $table ) {
 127                         case 'pagelinks':
 128                         case 'templatelinks':
 129                         case 'redirect':
 130                                 $conds = array(
 131                                         "{$prefix}_namespace" => $this->title->getNamespace(),
 132                                         "{$prefix}_title" => $this->title->getDBkey(),
 133                                         "page_id={$prefix}_from"
 134                                 );
 135                                 break;
 136                         case 'imagelinks':
 137                                 $conds = array(
 138                                         'il_to' => $this->title->getDBkey(),
 139                                         'page_id=il_from'
 140                                 );
 141                                 break;
 142                         case 'categorylinks':
 143                                 $conds = array(
 144                                         'cl_to' => $this->title->getDBkey(),
 145                                         'page_id=cl_from',
 146                                 );
 147                                 break;
 148                         default:
 149                                 throw new MWException( "Invalid table \"$table\" in " . __CLASS__ );
 150                 }
 151                 return $conds;
 152         }
 153
 154         /**
 155          * Get the approximate number of backlinks
 156          */
 157         public function getNumLinks( $table ) {
 158                 if ( isset( $this->fullResultCache[$table] ) ) {
 159                         return $this->fullResultCache[$table]->numRows();
 160                 }
 161                 if ( isset( $this->partitionCache[$table] ) ) {
 162                         $entry = reset( $this->partitionCache[$table] );
 163                         return $entry['numRows'];
 164                 }
 165                 $titleArray = $this->getLinks( $table );
 166                 return $titleArray->count();
 167         }
 168
 169         /**
 170          * Partition the backlinks into batches.
 171          * Returns an array giving the start and end of each range. The first batch has
 172          * a start of false, and the last batch has an end of false.
 173          *
 174          * @param $table String: the links table name
 175          * @param $batchSize Integer
 176          * @return Array
 177          */
 178         public function partition( $table, $batchSize ) {
 179                 // Try cache
 180                 if ( isset( $this->partitionCache[$table][$batchSize] ) ) {
 181                         wfDebug( __METHOD__.": got from partition cache\n" );
 182                         return $this->partitionCache[$table][$batchSize]['batches'];
 183                 }
 184                 $this->partitionCache[$table][$batchSize] = false;
 185                 $cacheEntry =& $this->partitionCache[$table][$batchSize];
 186
 187                 // Try full result cache
 188                 if ( isset( $this->fullResultCache[$table] ) ) {
 189                         $cacheEntry = $this->partitionResult( $this->fullResultCache[$table], $batchSize );
 190                         wfDebug( __METHOD__.": got from full result cache\n" );
 191                         return $cacheEntry['batches'];
 192                 }
 193                 // Try memcached
 194                 global $wgMemc;
 195                 $memcKey = wfMemcKey( 'backlinks', md5( $this->title->getPrefixedDBkey() ),
 196                         $table, $batchSize );
 197                 $memcValue = $wgMemc->get( $memcKey );
 198                 if ( is_array( $memcValue ) ) {
 199                         $cacheEntry = $memcValue;
 200                         wfDebug( __METHOD__.": got from memcached $memcKey\n" );
 201                         return $cacheEntry['batches'];
 202                 }
 203                 // Fetch from database
 204                 $this->getLinks( $table );
 205                 $cacheEntry = $this->partitionResult( $this->fullResultCache[$table], $batchSize );
 206                 // Save to memcached
 207                 $wgMemc->set( $memcKey, $cacheEntry, self::CACHE_EXPIRY );
 208                 wfDebug( __METHOD__.": got from database\n" );
 209                 return $cacheEntry['batches'];
 210         }
 211
 212         /**
 213          * Partition a DB result with backlinks in it into batches
 214          */
 215         protected function partitionResult( $res, $batchSize ) {
 216                 $batches = array();
 217                 $numRows = $res->numRows();
 218                 $numBatches = ceil( $numRows / $batchSize );
 219                 for ( $i = 0; $i < $numBatches; $i++ ) {
 220                         if ( $i == 0  ) {
 221                                 $start = false;
 222                         } else {
 223                                 $rowNum = intval( $numRows * $i / $numBatches );
 224                                 $res->seek( $rowNum );
 225                                 $row = $res->fetchObject();
 226                                 $start = $row->page_id;
 227                         }
 228                         if ( $i == $numBatches - 1 ) {
 229                                 $end = false;
 230                         } else {
 231                                 $rowNum = intval( $numRows * ( $i + 1 ) / $numBatches );
 232                                 $res->seek( $rowNum );
 233                                 $row = $res->fetchObject();
 234                                 $end = $row->page_id - 1;
 235                         }
 236
 237                         # Sanity check order
 238                         if ( $start && $end && $start > $end ) {
 239                                 throw new MWException( __METHOD__.': Internal error: query result out of order' );
 240                         }
 241
 242                         $batches[] = array( $start, $end );
 243                 }
 244                 return array( 'numRows' => $numRows, 'batches' => $batches );
 245         }
 246 }