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  * File::getLinksTo() and Title::touchArray(), which uses about 2KB of memory per
  22  * link.
  23  *
  24  * @ingroup Cache
  25  */
  26 class HTMLCacheUpdate
  27 {
  28         /**
  29          * @var Title
  30          */
  31         public $mTitle;
  32
  33         public $mTable, $mPrefix, $mStart, $mEnd;
  34         public $mRowsPerJob, $mRowsPerQuery;
  35
  36         function __construct( $titleTo, $table, $start = false, $end = false ) {
  37                 global $wgUpdateRowsPerJob, $wgUpdateRowsPerQuery;
  38
  39                 $this->mTitle = $titleTo;
  40                 $this->mTable = $table;
  41                 $this->mStart = $start;
  42                 $this->mEnd = $end;
  43                 $this->mRowsPerJob = $wgUpdateRowsPerJob;
  44                 $this->mRowsPerQuery = $wgUpdateRowsPerQuery;
  45                 $this->mCache = $this->mTitle->getBacklinkCache();
  46         }
  47
  48         public function doUpdate() {
  49                 if ( $this->mStart || $this->mEnd ) {
  50                         $this->doPartialUpdate();
  51                         return;
  52                 }
  53
  54                 # Get an estimate of the number of rows from the BacklinkCache
  55                 $numRows = $this->mCache->getNumLinks( $this->mTable );
  56                 if ( $numRows > $this->mRowsPerJob * 2 ) {
  57                         # Do fast cached partition
  58                         $this->insertJobs();
  59                 } else {
  60                         # Get the links from the DB
  61                         $titleArray = $this->mCache->getLinks( $this->mTable );
  62                         # Check if the row count estimate was correct
  63                         if ( $titleArray->count() > $this->mRowsPerJob * 2 ) {
  64                                 # Not correct, do accurate partition
  65                                 wfDebug( __METHOD__.": row count estimate was incorrect, repartitioning\n" );
  66                                 $this->insertJobsFromTitles( $titleArray );
  67                         } else {
  68                                 $this->invalidateTitles( $titleArray );
  69                         }
  70                 }
  71         }
  72
  73         /**
  74          * Update some of the backlinks, defined by a page ID range
  75          */
  76         protected function doPartialUpdate() {
  77                 $titleArray = $this->mCache->getLinks( $this->mTable, $this->mStart, $this->mEnd );
  78                 if ( $titleArray->count() <= $this->mRowsPerJob * 2 ) {
  79                         # This partition is small enough, do the update
  80                         $this->invalidateTitles( $titleArray );
  81                 } else {
  82                         # Partitioning was excessively inaccurate. Divide the job further.
  83                         # This can occur when a large number of links are added in a short
  84                         # period of time, say by updating a heavily-used template.
  85                         $this->insertJobsFromTitles( $titleArray );
  86                 }
  87         }
  88
  89         /**
  90          * Partition the current range given by $this->mStart and $this->mEnd,
  91          * using a pre-calculated title array which gives the links in that range.
  92          * Queue the resulting jobs.
  93          */
  94         protected function insertJobsFromTitles( $titleArray ) {
  95                 # We make subpartitions in the sense that the start of the first job
  96                 # will be the start of the parent partition, and the end of the last
  97                 # job will be the end of the parent partition.
  98                 $jobs = array();
  99                 $start = $this->mStart; # start of the current job
 100                 $numTitles = 0;
 101                 foreach ( $titleArray as $title ) {
 102                         $id = $title->getArticleID();
 103                         # $numTitles is now the number of titles in the current job not
 104                         # including the current ID
 105                         if ( $numTitles >= $this->mRowsPerJob ) {
 106                                 # Add a job up to but not including the current ID
 107                                 $params = array(
 108                                         'table' => $this->mTable,
 109                                         'start' => $start,
 110                                         'end' => $id - 1
 111                                 );
 112                                 $jobs[] = new HTMLCacheUpdateJob( $this->mTitle, $params );
 113                                 $start = $id;
 114                                 $numTitles = 0;
 115                         }
 116                         $numTitles++;
 117                 }
 118                 # Last job
 119                 $params = array(
 120                         'table' => $this->mTable,
 121                         'start' => $start,
 122                         'end' => $this->mEnd
 123                 );
 124                 $jobs[] = new HTMLCacheUpdateJob( $this->mTitle, $params );
 125                 wfDebug( __METHOD__.": repartitioning into " . count( $jobs ) . " jobs\n" );
 126
 127                 if ( count( $jobs ) < 2 ) {
 128                         # I don't think this is possible at present, but handling this case
 129                         # makes the code a bit more robust against future code updates and
 130                         # avoids a potential infinite loop of repartitioning
 131                         wfDebug( __METHOD__.": repartitioning failed!\n" );
 132                         $this->invalidateTitles( $titleArray );
 133                         return;
 134                 }
 135
 136                 Job::batchInsert( $jobs );
 137         }
 138
 139         protected function insertJobs() {
 140                 $batches = $this->mCache->partition( $this->mTable, $this->mRowsPerJob );
 141                 if ( !$batches ) {
 142                         return;
 143                 }
 144                 $jobs = array();
 145                 foreach ( $batches as $batch ) {
 146                         $params = array(
 147                                 'table' => $this->mTable,
 148                                 'start' => $batch[0],
 149                                 'end' => $batch[1],
 150                         );
 151                         $jobs[] = new HTMLCacheUpdateJob( $this->mTitle, $params );
 152                 }
 153                 Job::batchInsert( $jobs );
 154         }
 155
 156         /**
 157          * Invalidate a range of pages, right now
 158          * @deprecated
 159          */
 160         public function invalidate( $startId = false, $endId = false ) {
 161                 $titleArray = $this->mCache->getLinks( $this->mTable, $startId, $endId );
 162                 $this->invalidateTitles( $titleArray );
 163         }
 164
 165         /**
 166          * Invalidate an array (or iterator) of Title objects, right now
 167          */
 168         protected function invalidateTitles( $titleArray ) {
 169                 global $wgUseFileCache, $wgUseSquid;
 170
 171                 $dbw = wfGetDB( DB_MASTER );
 172                 $timestamp = $dbw->timestamp();
 173
 174                 # Get all IDs in this query into an array
 175                 $ids = array();
 176                 foreach ( $titleArray as $title ) {
 177                         $ids[] = $title->getArticleID();
 178                 }
 179
 180                 if ( !$ids ) {
 181                         return;
 182                 }
 183
 184                 # Update page_touched
 185                 $batches = array_chunk( $ids, $this->mRowsPerQuery );
 186                 foreach ( $batches as $batch ) {
 187                         $dbw->update( 'page',
 188                                 array( 'page_touched' => $timestamp ),
 189                                 array( 'page_id IN (' . $dbw->makeList( $batch ) . ')' ),
 190                                 __METHOD__
 191                         );
 192                 }
 193
 194                 # Update squid
 195                 if ( $wgUseSquid ) {
 196                         $u = SquidUpdate::newFromTitles( $titleArray );
 197                         $u->doUpdate();
 198                 }
 199
 200                 # Update file cache
 201                 if  ( $wgUseFileCache ) {
 202                         foreach ( $titleArray as $title ) {
 203                                 HTMLFileCache::clearFileCache( $title );
 204                         }
 205                 }
 206         }
 207
 208 }
 209
 210 /**
 211  * Job wrapper for HTMLCacheUpdate. Gets run whenever a related
 212  * job gets called from the queue.
 213  *
 214  * @ingroup JobQueue
 215  */
 216 class HTMLCacheUpdateJob extends Job {
 217         var $table, $start, $end;
 218
 219         /**
 220          * Construct a job
 221          * @param $title Title: the title linked to
 222          * @param $params Array: job parameters (table, start and end page_ids)
 223          * @param $id Integer: job id
 224          */
 225         function __construct( $title, $params, $id = 0 ) {
 226                 parent::__construct( 'htmlCacheUpdate', $title, $params, $id );
 227                 $this->table = $params['table'];
 228                 $this->start = $params['start'];
 229                 $this->end = $params['end'];
 230         }
 231
 232         public function run() {
 233                 $update = new HTMLCacheUpdate( $this->title, $this->table, $this->start, $this->end );
 234                 $update->doUpdate();
 235                 return true;
 236         }
 237 }