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