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, $mStart, $mEnd;
  29         public $mRowsPerJob, $mRowsPerQuery;
  30
  31         function __construct( $titleTo, $table, $start = false, $end = false ) {
  32                 global $wgUpdateRowsPerJob, $wgUpdateRowsPerQuery;
  33
  34                 $this->mTitle = $titleTo;
  35                 $this->mTable = $table;
  36                 $this->mStart = $start;
  37                 $this->mEnd = $end;
  38                 $this->mRowsPerJob = $wgUpdateRowsPerJob;
  39                 $this->mRowsPerQuery = $wgUpdateRowsPerQuery;
  40                 $this->mCache = $this->mTitle->getBacklinkCache();
  41         }
  42
  43         public function doUpdate() {
  44                 if ( $this->mStart || $this->mEnd ) {
  45                         $this->doPartialUpdate();
  46                         return;
  47                 }
  48
  49                 # Get an estimate of the number of rows from the BacklinkCache
  50                 $numRows = $this->mCache->getNumLinks( $this->mTable );
  51                 if ( $numRows > $this->mRowsPerJob * 2 ) {
  52                         # Do fast cached partition
  53                         $this->insertJobs();
  54                 } else {
  55                         # Get the links from the DB
  56                         $titleArray = $this->mCache->getLinks( $this->mTable );
  57                         # Check if the row count estimate was correct
  58                         if ( $titleArray->count() > $this->mRowsPerJob * 2 ) {
  59                                 # Not correct, do accurate partition
  60                                 wfDebug( __METHOD__.": row count estimate was incorrect, repartitioning\n" );
  61                                 $this->insertJobsFromTitles( $titleArray );
  62                         } else {
  63                                 $this->invalidateTitles( $titleArray );
  64                         }
  65                 }
  66                 wfRunHooks( 'HTMLCacheUpdate::doUpdate', array($this->mTitle) );
  67         }
  68
  69         /**
  70          * Update some of the backlinks, defined by a page ID range
  71          */
  72         protected function doPartialUpdate() {
  73                 $titleArray = $this->mCache->getLinks( $this->mTable, $this->mStart, $this->mEnd );
  74                 if ( $titleArray->count() <= $this->mRowsPerJob * 2 ) {
  75                         # This partition is small enough, do the update
  76                         $this->invalidateTitles( $titleArray );
  77                 } else {
  78                         # Partitioning was excessively inaccurate. Divide the job further.
  79                         # This can occur when a large number of links are added in a short
  80                         # period of time, say by updating a heavily-used template.
  81                         $this->insertJobsFromTitles( $titleArray );
  82                 }
  83         }
  84
  85         /**
  86          * Partition the current range given by $this->mStart and $this->mEnd,
  87          * using a pre-calculated title array which gives the links in that range.
  88          * Queue the resulting jobs.
  89          */
  90         protected function insertJobsFromTitles( $titleArray ) {
  91                 # We make subpartitions in the sense that the start of the first job
  92                 # will be the start of the parent partition, and the end of the last
  93                 # job will be the end of the parent partition.
  94                 $jobs = array();
  95                 $start = $this->mStart; # start of the current job
  96                 $numTitles = 0;
  97                 foreach ( $titleArray as $title ) {
  98                         $id = $title->getArticleID();
  99                         # $numTitles is now the number of titles in the current job not
 100                         # including the current ID
 101                         if ( $numTitles >= $this->mRowsPerJob ) {
 102                                 # Add a job up to but not including the current ID
 103                                 $params = array(
 104                                         'table' => $this->mTable,
 105                                         'start' => $start,
 106                                         'end' => $id - 1
 107                                 );
 108                                 $jobs[] = new HTMLCacheUpdateJob( $this->mTitle, $params );
 109                                 $start = $id;
 110                                 $numTitles = 0;
 111                         }
 112                         $numTitles++;
 113                 }
 114                 # Last job
 115                 $params = array(
 116                         'table' => $this->mTable,
 117                         'start' => $start,
 118                         'end' => $this->mEnd
 119                 );
 120                 $jobs[] = new HTMLCacheUpdateJob( $this->mTitle, $params );
 121                 wfDebug( __METHOD__.": repartitioning into " . count( $jobs ) . " jobs\n" );
 122
 123                 if ( count( $jobs ) < 2 ) {
 124                         # I don't think this is possible at present, but handling this case
 125                         # makes the code a bit more robust against future code updates and
 126                         # avoids a potential infinite loop of repartitioning
 127                         wfDebug( __METHOD__.": repartitioning failed!\n" );
 128                         $this->invalidateTitles( $titleArray );
 129                         return;
 130                 }
 131
 132                 Job::batchInsert( $jobs );
 133         }
 134
 135         protected function insertJobs() {
 136                 $batches = $this->mCache->partition( $this->mTable, $this->mRowsPerJob );
 137                 if ( !$batches ) {
 138                         return;
 139                 }
 140                 $jobs = array();
 141                 foreach ( $batches as $batch ) {
 142                         $params = array(
 143                                 'table' => $this->mTable,
 144                                 'start' => $batch[0],
 145                                 'end' => $batch[1],
 146                         );
 147                         $jobs[] = new HTMLCacheUpdateJob( $this->mTitle, $params );
 148                 }
 149                 Job::batchInsert( $jobs );
 150         }
 151
 152         /**
 153          * Invalidate a range of pages, right now
 154          * @deprecated
 155          */
 156         public function invalidate( $startId = false, $endId = false ) {
 157                 $titleArray = $this->mCache->getLinks( $this->mTable, $startId, $endId );
 158                 $this->invalidateTitles( $titleArray );
 159         }
 160
 161         /**
 162          * Invalidate an array (or iterator) of Title objects, right now
 163          */
 164         protected function invalidateTitles( $titleArray ) {
 165                 global $wgUseFileCache, $wgUseSquid;
 166
 167                 $dbw = wfGetDB( DB_MASTER );
 168                 $timestamp = $dbw->timestamp();
 169
 170                 # Get all IDs in this query into an array
 171                 $ids = array();
 172                 foreach ( $titleArray as $title ) {
 173                         $ids[] = $title->getArticleID();
 174                 }
 175
 176                 if ( !$ids ) {
 177                         return;
 178                 }
 179
 180                 # Update page_touched
 181                 $batches = array_chunk( $ids, $this->mRowsPerQuery );
 182                 foreach ( $batches as $batch ) {
 183                         $dbw->update( 'page',
 184                                 array( 'page_touched' => $timestamp ),
 185                                 array( 'page_id IN (' . $dbw->makeList( $batch ) . ')' ),
 186                                 __METHOD__
 187                         );
 188                 }
 189
 190                 # Update squid
 191                 if ( $wgUseSquid ) {
 192                         $u = SquidUpdate::newFromTitles( $titleArray );
 193                         $u->doUpdate();
 194                 }
 195
 196                 # Update file cache
 197                 if  ( $wgUseFileCache ) {
 198                         foreach ( $titleArray as $title ) {
 199                                 HTMLFileCache::clearFileCache( $title );
 200                         }
 201                 }
 202         }
 203
 204 }
 205
 206 /**
 207  * Job wrapper for HTMLCacheUpdate. Gets run whenever a related
 208  * job gets called from the queue.
 209  *
 210  * @ingroup JobQueue
 211  */
 212 class HTMLCacheUpdateJob extends Job {
 213         var $table, $start, $end;
 214
 215         /**
 216          * Construct a job
 217          * @param $title Title: the title linked to
 218          * @param $params Array: job parameters (table, start and end page_ids)
 219          * @param $id Integer: job id
 220          */
 221         function __construct( $title, $params, $id = 0 ) {
 222                 parent::__construct( 'htmlCacheUpdate', $title, $params, $id );
 223                 $this->table = $params['table'];
 224                 $this->start = $params['start'];
 225                 $this->end = $params['end'];
 226         }
 227
 228         public function run() {
 229                 $update = new HTMLCacheUpdate( $this->title, $this->table, $this->start, $this->end );
 230                 $update->doUpdate();
 231                 return true;
 232         }
 233 }