includes/HTMLCacheUpdate.php

   1 <?php
   2
   3 /**
   4  * Class to invalidate the HTML cache of all the pages linking to a given title.
   5  * Small numbers of links will be done immediately, large numbers are pushed onto
   6  * the job queue.
   7  *
   8  * This class is designed to work efficiently with small numbers of links, and
   9  * to work reasonably well with up to ~10^5 links. Above ~10^6 links, the memory
  10  * and time requirements of loading all backlinked IDs in doUpdate() might become
  11  * prohibitive. The requirements measured at Wikimedia are approximately:
  12  *
  13  *   memory: 48 bytes per row
  14  *   time: 16us per row for the query plus processing
  15  *
  16  * The reason this query is done is to support partitioning of the job
  17  * by backlinked ID. The memory issue could be allieviated by doing this query in
  18  * batches, but of course LIMIT with an offset is inefficient on the DB side.
  19  *
  20  * The class is nevertheless a vast improvement on the previous method of using
  21  * Image::getLinksTo() and Title::touchArray(), which uses about 2KB of memory per
  22  * link.
  23  *
  24  * @ingroup Cache
  25  */
  26 class HTMLCacheUpdate
  27 {
  28         public $mTitle, $mTable, $mPrefix;
  29         public $mRowsPerJob, $mRowsPerQuery;
  30
  31         function __construct( $titleTo, $table ) {
  32                 global $wgUpdateRowsPerJob, $wgUpdateRowsPerQuery;
  33
  34                 $this->mTitle = $titleTo;
  35                 $this->mTable = $table;
  36                 $this->mRowsPerJob = $wgUpdateRowsPerJob;
  37                 $this->mRowsPerQuery = $wgUpdateRowsPerQuery;
  38                 $this->mCache = $this->mTitle->getBacklinkCache();
  39         }
  40
  41         public function doUpdate() {
  42                 # Fetch the IDs
  43                 $numRows = $this->mCache->getNumLinks( $this->mTable );
  44
  45                 if ( $numRows != 0 ) {
  46                         if ( $numRows > $this->mRowsPerJob ) {
  47                                 $this->insertJobs();
  48                         } else {
  49                                 $this->invalidate();
  50                         }
  51                 }
  52                 wfRunHooks( 'HTMLCacheUpdate::doUpdate', array($this->mTitle) );
  53         }
  54
  55         protected function insertJobs() {
  56                 $batches = $this->mCache->partition( $this->mTable, $this->mRowsPerJob );
  57                 if ( !$batches ) {
  58                         return;
  59                 }
  60                 foreach ( $batches as $batch ) {
  61                         $params = array(
  62                                 'table' => $this->mTable,
  63                                 'start' => $batch[0],
  64                                 'end' => $batch[1],
  65                         );
  66                         $jobs[] = new HTMLCacheUpdateJob( $this->mTitle, $params );
  67                 }
  68                 Job::batchInsert( $jobs );
  69         }
  70
  71
  72         /**
  73          * Invalidate a set of pages, right now
  74          */
  75         public function invalidate( $startId = false, $endId = false ) {
  76                 global $wgUseFileCache, $wgUseSquid;
  77
  78                 $titleArray = $this->mCache->getLinks( $this->mTable, $startId, $endId );
  79                 if ( $titleArray->count() == 0 ) {
  80                         return;
  81                 }
  82
  83                 $dbw = wfGetDB( DB_MASTER );
  84                 $timestamp = $dbw->timestamp();
  85
  86                 # Get all IDs in this query into an array
  87                 $ids = array();
  88                 foreach ( $titleArray as $title ) {
  89                         $ids[] = $title->getArticleID();
  90                 }
  91                 # Update page_touched
  92                 $dbw->update( 'page',
  93                         array( 'page_touched' => $timestamp ),
  94                         array( 'page_id IN (' . $dbw->makeList( $ids ) . ')' ),
  95                         __METHOD__
  96                 );
  97
  98                 # Update squid
  99                 if ( $wgUseSquid ) {
 100                         $u = SquidUpdate::newFromTitles( $titleArray );
 101                         $u->doUpdate();
 102                 }
 103
 104                 # Update file cache
 105                 if  ( $wgUseFileCache ) {
 106                         foreach ( $titleArray as $title ) {
 107                                 HTMLFileCache::clearFileCache( $title );
 108                         }
 109                 }
 110         }
 111 }
 112
 113 /**
 114  * Job wrapper for HTMLCacheUpdate. Gets run whenever a related
 115  * job gets called from the queue.
 116  *
 117  * @ingroup JobQueue
 118  */
 119 class HTMLCacheUpdateJob extends Job {
 120         var $table, $start, $end;
 121
 122         /**
 123          * Construct a job
 124          * @param Title $title The title linked to
 125          * @param array $params Job parameters (table, start and end page_ids)
 126          * @param integer $id job_id
 127          */
 128         function __construct( $title, $params, $id = 0 ) {
 129                 parent::__construct( 'htmlCacheUpdate', $title, $params, $id );
 130                 $this->table = $params['table'];
 131                 $this->start = $params['start'];
 132                 $this->end = $params['end'];
 133         }
 134
 135         public function run() {
 136                 $update = new HTMLCacheUpdate( $this->title, $this->table );
 137                 $update->invalidate( $this->start, $this->end );
 138                 return true;
 139         }
 140 }