includes/cache/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                 if ( $this->mTable === 'globaltemplatelinks' ) {
  55                         global $wgEnableInterwikiTemplatesTracking;
  56
  57                         if ( $wgEnableInterwikiTemplatesTracking ) {
  58                                 $distantPageArray = $this->mCache->getDistantTemplateLinks( 'globaltemplatelinks' );
  59                                 $this->invalidateDistantTitles( $distantPageArray );
  60                         }
  61                         return;
  62                 }
  63
  64                 # Get an estimate of the number of rows from the BacklinkCache
  65                 $numRows = $this->mCache->getNumLinks( $this->mTable );
  66                 if ( $numRows > $this->mRowsPerJob * 2 ) {
  67                         # Do fast cached partition
  68                         $this->insertJobs();
  69                 } else {
  70                         # Get the links from the DB
  71                         $titleArray = $this->mCache->getLinks( $this->mTable );
  72                         # Check if the row count estimate was correct
  73                         if ( $titleArray->count() > $this->mRowsPerJob * 2 ) {
  74                                 # Not correct, do accurate partition
  75                                 wfDebug( __METHOD__.": row count estimate was incorrect, repartitioning\n" );
  76                                 $this->insertJobsFromTitles( $titleArray );
  77                         } else {
  78                                 $this->invalidateTitles( $titleArray );
  79                         }
  80                 }
  81                 wfRunHooks( 'HTMLCacheUpdate::doUpdate', array($this->mTitle) );
  82         }
  83
  84         /**
  85          * Update some of the backlinks, defined by a page ID range
  86          */
  87         protected function doPartialUpdate() {
  88                 $titleArray = $this->mCache->getLinks( $this->mTable, $this->mStart, $this->mEnd );
  89                 if ( $titleArray->count() <= $this->mRowsPerJob * 2 ) {
  90                         # This partition is small enough, do the update
  91                         $this->invalidateTitles( $titleArray );
  92                 } else {
  93                         # Partitioning was excessively inaccurate. Divide the job further.
  94                         # This can occur when a large number of links are added in a short
  95                         # period of time, say by updating a heavily-used template.
  96                         $this->insertJobsFromTitles( $titleArray );
  97                 }
  98         }
  99
 100         /**
 101          * Partition the current range given by $this->mStart and $this->mEnd,
 102          * using a pre-calculated title array which gives the links in that range.
 103          * Queue the resulting jobs.
 104          *
 105          * @param $titleArray array
 106          */
 107         protected function insertJobsFromTitles( $titleArray ) {
 108                 # We make subpartitions in the sense that the start of the first job
 109                 # will be the start of the parent partition, and the end of the last
 110                 # job will be the end of the parent partition.
 111                 $jobs = array();
 112                 $start = $this->mStart; # start of the current job
 113                 $numTitles = 0;
 114                 foreach ( $titleArray as $title ) {
 115                         $id = $title->getArticleID();
 116                         # $numTitles is now the number of titles in the current job not
 117                         # including the current ID
 118                         if ( $numTitles >= $this->mRowsPerJob ) {
 119                                 # Add a job up to but not including the current ID
 120                                 $params = array(
 121                                         'table' => $this->mTable,
 122                                         'start' => $start,
 123                                         'end' => $id - 1
 124                                 );
 125                                 $jobs[] = new HTMLCacheUpdateJob( $this->mTitle, $params );
 126                                 $start = $id;
 127                                 $numTitles = 0;
 128                         }
 129                         $numTitles++;
 130                 }
 131                 # Last job
 132                 $params = array(
 133                         'table' => $this->mTable,
 134                         'start' => $start,
 135                         'end' => $this->mEnd
 136                 );
 137                 $jobs[] = new HTMLCacheUpdateJob( $this->mTitle, $params );
 138                 wfDebug( __METHOD__.": repartitioning into " . count( $jobs ) . " jobs\n" );
 139
 140                 if ( count( $jobs ) < 2 ) {
 141                         # I don't think this is possible at present, but handling this case
 142                         # makes the code a bit more robust against future code updates and
 143                         # avoids a potential infinite loop of repartitioning
 144                         wfDebug( __METHOD__.": repartitioning failed!\n" );
 145                         $this->invalidateTitles( $titleArray );
 146                         return;
 147                 }
 148
 149                 Job::batchInsert( $jobs );
 150         }
 151
 152         protected function insertJobs() {
 153                 $batches = $this->mCache->partition( $this->mTable, $this->mRowsPerJob );
 154                 if ( !$batches ) {
 155                         return;
 156                 }
 157                 $jobs = array();
 158                 foreach ( $batches as $batch ) {
 159                         $params = array(
 160                                 'table' => $this->mTable,
 161                                 'start' => $batch[0],
 162                                 'end' => $batch[1],
 163                         );
 164                         $jobs[] = new HTMLCacheUpdateJob( $this->mTitle, $params );
 165                 }
 166                 Job::batchInsert( $jobs );
 167         }
 168
 169         /**
 170          * Invalidate an array (or iterator) of Title objects, right now
 171          */
 172         protected function invalidateTitles( $titleArray ) {
 173                 global $wgUseFileCache, $wgUseSquid;
 174
 175                 $dbw = wfGetDB( DB_MASTER );
 176                 $timestamp = $dbw->timestamp();
 177
 178                 # Get all IDs in this query into an array
 179                 $ids = array();
 180                 foreach ( $titleArray as $title ) {
 181                         $ids[] = $title->getArticleID();
 182                 }
 183
 184                 if ( !$ids ) {
 185                         return;
 186                 }
 187
 188                 # Update page_touched
 189                 $batches = array_chunk( $ids, $this->mRowsPerQuery );
 190                 foreach ( $batches as $batch ) {
 191                         $dbw->update( 'page',
 192                                 array( 'page_touched' => $timestamp ),
 193                                 array( 'page_id IN (' . $dbw->makeList( $batch ) . ')' ),
 194                                 __METHOD__
 195                         );
 196                 }
 197
 198                 # Update squid
 199                 if ( $wgUseSquid ) {
 200                         $u = SquidUpdate::newFromTitles( $titleArray );
 201                         $u->doUpdate();
 202                 }
 203
 204                 # Update file cache
 205                 if  ( $wgUseFileCache ) {
 206                         foreach ( $titleArray as $title ) {
 207                                 HTMLFileCache::clearFileCache( $title );
 208                         }
 209                 }
 210         }
 211
 212         /**
 213          * Invalidate an array of distant pages, given the wiki ID and page ID of those pages
 214          */
 215         protected function invalidateDistantTitles( $distantPageArray ) {
 216                 global $wgUseSquid;
 217
 218                 $pagesByWiki = array();
 219                 $titleArray = array();
 220                 # Sort by WikiID in $pagesByWiki
 221                 # Create the distant titles for Squid in $titleArray
 222                 foreach ( $distantPageArray as $row ) {
 223                         $wikiid = $row->gtl_from_wiki;
 224                         if( !isset( $pagesByWiki[$wikiid] ) ) {
 225                                 $pagesByWiki[$wikiid] = array();
 226 }
 227                         $pagesByWiki[$wikiid][] = $row->gtl_from_page;
 228                         $titleArray[] = Title::makeTitle( $row->gtl_from_namespace, $row->gtl_from_title, '', $row->gil_interwiki );
 229                 }
 230
 231                 foreach ( $pagesByWiki as $wikiid => $pages ) {
 232                         $dbw = wfGetDB( DB_MASTER, array( ), $wikiid );
 233                         $timestamp = $dbw->timestamp();
 234                         $batches = array_chunk( $pages, $this->mRowsPerQuery );
 235                         foreach ( $batches as $batch ) {
 236                                 $dbw->update( 'page',
 237                                         array( 'page_touched' => $timestamp ),
 238                                         array( 'page_id IN (' . $dbw->makeList( $batch ) . ')' ),
 239                                         __METHOD__
 240                                 );
 241                         }
 242                 }
 243
 244                 # Update squid
 245                 if ( $wgUseSquid ) {
 246                         $u = SquidUpdate::newFromTitles( $titleArray );
 247                         $u->doUpdate();
 248                 }
 249         }
 250 }
 251
 252 /**
 253  * Job wrapper for HTMLCacheUpdate. Gets run whenever a related
 254  * job gets called from the queue.
 255  *
 256  * @ingroup JobQueue
 257  */
 258 class HTMLCacheUpdateJob extends Job {
 259         var $table, $start, $end;
 260
 261         /**
 262          * Construct a job
 263          * @param $title Title: the title linked to
 264          * @param $params Array: job parameters (table, start and end page_ids)
 265          * @param $id Integer: job id
 266          */
 267         function __construct( $title, $params, $id = 0 ) {
 268                 parent::__construct( 'htmlCacheUpdate', $title, $params, $id );
 269                 $this->table = $params['table'];
 270                 $this->start = $params['start'];
 271                 $this->end = $params['end'];
 272         }
 273
 274         public function run() {
 275                 $update = new HTMLCacheUpdate( $this->title, $this->table, $this->start, $this->end );
 276                 $update->doUpdate();
 277                 return true;
 278         }
 279 }