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 implements DeferrableUpdate {
  27         /**
  28          * @var Title
  29          */
  30         public $mTitle;
  31
  32         public $mTable, $mPrefix, $mStart, $mEnd;
  33         public $mRowsPerJob, $mRowsPerQuery;
  34
  35         function __construct( $titleTo, $table, $start = false, $end = false ) {
  36                 global $wgUpdateRowsPerJob, $wgUpdateRowsPerQuery;
  37
  38                 $this->mTitle = $titleTo;
  39                 $this->mTable = $table;
  40                 $this->mStart = $start;
  41                 $this->mEnd = $end;
  42                 $this->mRowsPerJob = $wgUpdateRowsPerJob;
  43                 $this->mRowsPerQuery = $wgUpdateRowsPerQuery;
  44                 $this->mCache = $this->mTitle->getBacklinkCache();
  45         }
  46
  47         public function doUpdate() {
  48                 if ( $this->mStart || $this->mEnd ) {
  49                         $this->doPartialUpdate();
  50                         return;
  51                 }
  52
  53                 if ( $this->mTable === 'globaltemplatelinks' ) {
  54                         global $wgEnableInterwikiTemplatesTracking;
  55
  56                         if ( $wgEnableInterwikiTemplatesTracking ) {
  57                                 $distantPageArray = $this->mCache->getDistantTemplateLinks( 'globaltemplatelinks' );
  58                                 $this->invalidateDistantTitles( $distantPageArray );
  59                         }
  60                         return;
  61                 }
  62
  63                 # Get an estimate of the number of rows from the BacklinkCache
  64                 $numRows = $this->mCache->getNumLinks( $this->mTable );
  65                 if ( $numRows > $this->mRowsPerJob * 2 ) {
  66                         # Do fast cached partition
  67                         $this->insertJobs();
  68                 } else {
  69                         # Get the links from the DB
  70                         $titleArray = $this->mCache->getLinks( $this->mTable );
  71                         # Check if the row count estimate was correct
  72                         if ( $titleArray->count() > $this->mRowsPerJob * 2 ) {
  73                                 # Not correct, do accurate partition
  74                                 wfDebug( __METHOD__.": row count estimate was incorrect, repartitioning\n" );
  75                                 $this->insertJobsFromTitles( $titleArray );
  76                         } else {
  77                                 $this->invalidateTitles( $titleArray );
  78                         }
  79                 }
  80                 wfRunHooks( 'HTMLCacheUpdate::doUpdate', array($this->mTitle) );
  81         }
  82
  83         /**
  84          * Update some of the backlinks, defined by a page ID range
  85          */
  86         protected function doPartialUpdate() {
  87                 $titleArray = $this->mCache->getLinks( $this->mTable, $this->mStart, $this->mEnd );
  88                 if ( $titleArray->count() <= $this->mRowsPerJob * 2 ) {
  89                         # This partition is small enough, do the update
  90                         $this->invalidateTitles( $titleArray );
  91                 } else {
  92                         # Partitioning was excessively inaccurate. Divide the job further.
  93                         # This can occur when a large number of links are added in a short
  94                         # period of time, say by updating a heavily-used template.
  95                         $this->insertJobsFromTitles( $titleArray );
  96                 }
  97         }
  98
  99         /**
 100          * Partition the current range given by $this->mStart and $this->mEnd,
 101          * using a pre-calculated title array which gives the links in that range.
 102          * Queue the resulting jobs.
 103          *
 104          * @param $titleArray array
 105          */
 106         protected function insertJobsFromTitles( $titleArray ) {
 107                 # We make subpartitions in the sense that the start of the first job
 108                 # will be the start of the parent partition, and the end of the last
 109                 # job will be the end of the parent partition.
 110                 $jobs = array();
 111                 $start = $this->mStart; # start of the current job
 112                 $numTitles = 0;
 113                 foreach ( $titleArray as $title ) {
 114                         $id = $title->getArticleID();
 115                         # $numTitles is now the number of titles in the current job not
 116                         # including the current ID
 117                         if ( $numTitles >= $this->mRowsPerJob ) {
 118                                 # Add a job up to but not including the current ID
 119                                 $params = array(
 120                                         'table' => $this->mTable,
 121                                         'start' => $start,
 122                                         'end' => $id - 1
 123                                 );
 124                                 $jobs[] = new HTMLCacheUpdateJob( $this->mTitle, $params );
 125                                 $start = $id;
 126                                 $numTitles = 0;
 127                         }
 128                         $numTitles++;
 129                 }
 130                 # Last job
 131                 $params = array(
 132                         'table' => $this->mTable,
 133                         'start' => $start,
 134                         'end' => $this->mEnd
 135                 );
 136                 $jobs[] = new HTMLCacheUpdateJob( $this->mTitle, $params );
 137                 wfDebug( __METHOD__.": repartitioning into " . count( $jobs ) . " jobs\n" );
 138
 139                 if ( count( $jobs ) < 2 ) {
 140                         # I don't think this is possible at present, but handling this case
 141                         # makes the code a bit more robust against future code updates and
 142                         # avoids a potential infinite loop of repartitioning
 143                         wfDebug( __METHOD__.": repartitioning failed!\n" );
 144                         $this->invalidateTitles( $titleArray );
 145                         return;
 146                 }
 147
 148                 Job::batchInsert( $jobs );
 149         }
 150
 151         protected function insertJobs() {
 152                 $batches = $this->mCache->partition( $this->mTable, $this->mRowsPerJob );
 153                 if ( !$batches ) {
 154                         return;
 155                 }
 156                 $jobs = array();
 157                 foreach ( $batches as $batch ) {
 158                         $params = array(
 159                                 'table' => $this->mTable,
 160                                 'start' => $batch[0],
 161                                 'end' => $batch[1],
 162                         );
 163                         $jobs[] = new HTMLCacheUpdateJob( $this->mTitle, $params );
 164                 }
 165                 Job::batchInsert( $jobs );
 166         }
 167
 168         /**
 169          * Invalidate an array (or iterator) of Title objects, right now
 170          */
 171         protected function invalidateTitles( $titleArray ) {
 172                 global $wgUseFileCache, $wgUseSquid;
 173
 174                 $dbw = wfGetDB( DB_MASTER );
 175                 $timestamp = $dbw->timestamp();
 176
 177                 # Get all IDs in this query into an array
 178                 $ids = array();
 179                 foreach ( $titleArray as $title ) {
 180                         $ids[] = $title->getArticleID();
 181                 }
 182
 183                 if ( !$ids ) {
 184                         return;
 185                 }
 186
 187                 # Update page_touched
 188                 $batches = array_chunk( $ids, $this->mRowsPerQuery );
 189                 foreach ( $batches as $batch ) {
 190                         $dbw->update( 'page',
 191                                 array( 'page_touched' => $timestamp ),
 192                                 array( 'page_id IN (' . $dbw->makeList( $batch ) . ')' ),
 193                                 __METHOD__
 194                         );
 195                 }
 196
 197                 # Update squid
 198                 if ( $wgUseSquid ) {
 199                         $u = SquidUpdate::newFromTitles( $titleArray );
 200                         $u->doUpdate();
 201                 }
 202
 203                 # Update file cache
 204                 if  ( $wgUseFileCache ) {
 205                         foreach ( $titleArray as $title ) {
 206                                 HTMLFileCache::clearFileCache( $title );
 207                         }
 208                 }
 209         }
 210
 211         /**
 212          * Invalidate an array of distant pages, given the wiki ID and page ID of those pages
 213          */
 214         protected function invalidateDistantTitles( $distantPageArray ) {
 215                 global $wgUseSquid;
 216
 217                 $pagesByWiki = array();
 218                 $titleArray = array();
 219                 # Sort by WikiID in $pagesByWiki
 220                 # Create the distant titles for Squid in $titleArray
 221                 foreach ( $distantPageArray as $row ) {
 222                         $wikiid = $row->gtl_from_wiki;
 223                         if( !isset( $pagesByWiki[$wikiid] ) ) {
 224                                 $pagesByWiki[$wikiid] = array();
 225 }
 226                         $pagesByWiki[$wikiid][] = $row->gtl_from_page;
 227                         $titleArray[] = Title::makeTitle( $row->gtl_from_namespace, $row->gtl_from_title, '', $row->gil_interwiki );
 228                 }
 229
 230                 foreach ( $pagesByWiki as $wikiid => $pages ) {
 231                         $dbw = wfGetDB( DB_MASTER, array( ), $wikiid );
 232                         $timestamp = $dbw->timestamp();
 233                         $batches = array_chunk( $pages, $this->mRowsPerQuery );
 234                         foreach ( $batches as $batch ) {
 235                                 $dbw->update( 'page',
 236                                         array( 'page_touched' => $timestamp ),
 237                                         array( 'page_id IN (' . $dbw->makeList( $batch ) . ')' ),
 238                                         __METHOD__
 239                                 );
 240                         }
 241                 }
 242
 243                 # Update squid
 244                 if ( $wgUseSquid ) {
 245                         $u = SquidUpdate::newFromTitles( $titleArray );
 246                         $u->doUpdate();
 247                 }
 248         }
 249 }
 250
 251 /**
 252  * Job wrapper for HTMLCacheUpdate. Gets run whenever a related
 253  * job gets called from the queue.
 254  *
 255  * @ingroup JobQueue
 256  */
 257 class HTMLCacheUpdateJob extends Job {
 258         var $table, $start, $end;
 259
 260         /**
 261          * Construct a job
 262          * @param $title Title: the title linked to
 263          * @param $params Array: job parameters (table, start and end page_ids)
 264          * @param $id Integer: job id
 265          */
 266         function __construct( $title, $params, $id = 0 ) {
 267                 parent::__construct( 'htmlCacheUpdate', $title, $params, $id );
 268                 $this->table = $params['table'];
 269                 $this->start = $params['start'];
 270                 $this->end = $params['end'];
 271         }
 272
 273         public function run() {
 274                 $update = new HTMLCacheUpdate( $this->title, $this->table, $this->start, $this->end );
 275                 $update->doUpdate();
 276                 return true;
 277         }
 278 }