X-Git-Url: https://scripts.mit.edu/gitweb/autoinstalls/mediawiki.git/blobdiff_plain/19e297c21b10b1b8a3acad5e73fc71dcb35db44a..6932310fd58ebef145fa01eb76edf7150284d8ea:/includes/SiteStats.php diff --git a/includes/SiteStats.php b/includes/SiteStats.php index 11cef56f..ce87596a 100644 --- a/includes/SiteStats.php +++ b/includes/SiteStats.php @@ -1,18 +1,53 @@ doUpdate(); - $dbr = wfGetDB( DB_SLAVE ); - self::$row = $dbr->selectRow( 'site_stats', '*', false, __METHOD__ ); + self::$row = self::doLoad( wfGetDB( DB_REPLICA ) ); } self::$loaded = true; } + /** + * @return bool|stdClass + */ static function loadAndLazyInit() { - wfDebug( __METHOD__ . ": reading site_stats from slave\n" ); - $row = self::doLoad( wfGetDB( DB_SLAVE ) ); + global $wgMiserMode; - if( !self::isSane( $row ) ) { - // Might have just been initialized during this request? Underflow? - wfDebug( __METHOD__ . ": site_stats damaged or missing on slave\n" ); - $row = self::doLoad( wfGetDB( DB_MASTER ) ); + wfDebug( __METHOD__ . ": reading site_stats from replica DB\n" ); + $row = self::doLoad( wfGetDB( DB_REPLICA ) ); + + if ( !self::isSane( $row ) ) { + $lb = MediaWikiServices::getInstance()->getDBLoadBalancer(); + if ( $lb->hasOrMadeRecentMasterChanges() ) { + // Might have just been initialized during this request? Underflow? + wfDebug( __METHOD__ . ": site_stats damaged or missing on replica DB\n" ); + $row = self::doLoad( wfGetDB( DB_MASTER ) ); + } } - if( !self::isSane( $row ) ) { + if ( !$wgMiserMode && !self::isSane( $row ) ) { // Normally the site_stats table is initialized at install time. // Some manual construction scenarios may leave the table empty or // broken, however, for instance when importing from a dump into a // clean schema with mwdumper. wfDebug( __METHOD__ . ": initializing damaged or missing site_stats\n" ); - SiteStatsInit::doAllAndCommit( false ); + SiteStatsInit::doAllAndCommit( wfGetDB( DB_REPLICA ) ); $row = self::doLoad( wfGetDB( DB_MASTER ) ); } - if( !self::isSane( $row ) ) { + if ( !self::isSane( $row ) ) { wfDebug( __METHOD__ . ": site_stats persistently nonsensical o_O\n" ); } + return $row; } + /** + * @param IDatabase $db + * @return bool|stdClass + */ static function doLoad( $db ) { - return $db->selectRow( 'site_stats', '*', false, __METHOD__ ); + return $db->selectRow( 'site_stats', [ + 'ss_row_id', + 'ss_total_edits', + 'ss_good_articles', + 'ss_total_pages', + 'ss_users', + 'ss_active_users', + 'ss_images', + ], [], __METHOD__ ); } + /** + * Return the total number of page views. Except we don't track those anymore. + * Stop calling this function, it will be removed some time in the future. It's + * kept here simply to prevent fatal errors. + * + * @deprecated since 1.25 + * @return int + */ static function views() { - self::load(); - return self::$row->ss_total_views; + wfDeprecated( __METHOD__, '1.25' ); + return 0; } + /** + * @return int + */ static function edits() { self::load(); return self::$row->ss_total_edits; } + /** + * @return int + */ static function articles() { self::load(); return self::$row->ss_good_articles; } + /** + * @return int + */ static function pages() { self::load(); return self::$row->ss_total_pages; } + /** + * @return int + */ static function users() { self::load(); return self::$row->ss_users; } + /** + * @return int + */ static function activeUsers() { self::load(); return self::$row->ss_active_users; } + /** + * @return int + */ static function images() { self::load(); return self::$row->ss_images; } - /** - * @deprecated Use self::numberingroup('sysop') instead - */ - static function admins() { - wfDeprecated(__METHOD__); - return self::numberingroup( 'sysop' ); - } - /** * Find the number of users in a given user group. - * @param $group String: name of group - * @return Integer + * @param string $group Name of group + * @return int */ static function numberingroup( $group ) { - if ( !isset( self::$groupMemberCounts[$group] ) ) { - global $wgMemc; - $key = wfMemcKey( 'SiteStats', 'groupcounts', $group ); - $hit = $wgMemc->get( $key ); - if ( !$hit ) { - $dbr = wfGetDB( DB_SLAVE ); - $hit = $dbr->selectField( + $cache = MediaWikiServices::getInstance()->getMainWANObjectCache(); + return $cache->getWithSetCallback( + $cache->makeKey( 'SiteStats', 'groupcounts', $group ), + $cache::TTL_HOUR, + function ( $oldValue, &$ttl, array &$setOpts ) use ( $group ) { + $dbr = wfGetDB( DB_REPLICA ); + + $setOpts += Database::getCacheSetOptions( $dbr ); + + return $dbr->selectField( 'user_groups', 'COUNT(*)', - array( 'ug_group' => $group ), + [ + 'ug_group' => $group, + 'ug_expiry IS NULL OR ug_expiry >= ' . $dbr->addQuotes( $dbr->timestamp() ) + ], __METHOD__ ); - $wgMemc->set( $key, $hit, 3600 ); - } - self::$groupMemberCounts[$group] = $hit; - } - return self::$groupMemberCounts[$group]; + }, + [ 'pcTTL' => $cache::TTL_PROC_LONG ] + ); } + /** + * Total number of jobs in the job queue. + * @return int + */ static function jobs() { - if ( !isset( self::$jobs ) ) { - $dbr = wfGetDB( DB_SLAVE ); - self::$jobs = $dbr->estimateRowCount( 'job' ); - /* Zero rows still do single row read for row that doesn't exist, but people are annoyed by that */ - if ( self::$jobs == 1 ) { - self::$jobs = 0; - } - } - return self::$jobs; + $cache = MediaWikiServices::getInstance()->getMainWANObjectCache(); + return $cache->getWithSetCallback( + $cache->makeKey( 'SiteStats', 'jobscount' ), + $cache::TTL_MINUTE, + function ( $oldValue, &$ttl, array &$setOpts ) { + try{ + $jobs = array_sum( JobQueueGroup::singleton()->getQueueSizes() ); + } catch ( JobQueueError $e ) { + $jobs = 0; + } + return $jobs; + }, + [ 'pcTTL' => $cache::TTL_PROC_LONG ] + ); } + /** + * @param int $ns + * + * @return int + */ static function pagesInNs( $ns ) { - wfProfileIn( __METHOD__ ); - if( !isset( self::$pageCount[$ns] ) ) { - $dbr = wfGetDB( DB_SLAVE ); - $pageCount[$ns] = (int)$dbr->selectField( + if ( !isset( self::$pageCount[$ns] ) ) { + $dbr = wfGetDB( DB_REPLICA ); + self::$pageCount[$ns] = (int)$dbr->selectField( 'page', 'COUNT(*)', - array( 'page_namespace' => $ns ), + [ 'page_namespace' => $ns ], __METHOD__ ); } - wfProfileOut( __METHOD__ ); - return $pageCount[$ns]; + return self::$pageCount[$ns]; } - /** Is the provided row of site stats sane, or should it be regenerated? */ + /** + * Is the provided row of site stats sane, or should it be regenerated? + * + * Checks only fields which are filled by SiteStatsInit::refresh. + * + * @param bool|object $row + * + * @return bool + */ private static function isSane( $row ) { - if( - $row === false - or $row->ss_total_pages < $row->ss_good_articles - or $row->ss_total_edits < $row->ss_total_pages + if ( $row === false + || $row->ss_total_pages < $row->ss_good_articles + || $row->ss_total_edits < $row->ss_total_pages ) { return false; } // Now check for underflow/overflow - foreach( array( 'total_views', 'total_edits', 'good_articles', - 'total_pages', 'users', 'admins', 'images' ) as $member ) { - if( - $row->{"ss_$member"} > 2000000000 - or $row->{"ss_$member"} < 0 - ) { + foreach ( [ + 'ss_total_edits', + 'ss_good_articles', + 'ss_total_pages', + 'ss_users', + 'ss_images', + ] as $member ) { + if ( $row->$member > 2000000000 || $row->$member < 0 ) { return false; } } @@ -182,83 +280,6 @@ class SiteStats { } } - -/** - * - */ -class SiteStatsUpdate { - - var $mViews, $mEdits, $mGood, $mPages, $mUsers; - - function __construct( $views, $edits, $good, $pages = 0, $users = 0 ) { - $this->mViews = $views; - $this->mEdits = $edits; - $this->mGood = $good; - $this->mPages = $pages; - $this->mUsers = $users; - } - - function appendUpdate( &$sql, $field, $delta ) { - if ( $delta ) { - if ( $sql ) { - $sql .= ','; - } - if ( $delta < 0 ) { - $sql .= "$field=$field-1"; - } else { - $sql .= "$field=$field+1"; - } - } - } - - function doUpdate() { - $dbw = wfGetDB( DB_MASTER ); - - $updates = ''; - - $this->appendUpdate( $updates, 'ss_total_views', $this->mViews ); - $this->appendUpdate( $updates, 'ss_total_edits', $this->mEdits ); - $this->appendUpdate( $updates, 'ss_good_articles', $this->mGood ); - $this->appendUpdate( $updates, 'ss_total_pages', $this->mPages ); - $this->appendUpdate( $updates, 'ss_users', $this->mUsers ); - - if ( $updates ) { - $site_stats = $dbw->tableName( 'site_stats' ); - $sql = "UPDATE $site_stats SET $updates"; - - # Need a separate transaction because this a global lock - $dbw->begin(); - $dbw->query( $sql, __METHOD__ ); - $dbw->commit(); - } - } - - public static function cacheUpdate( $dbw ) { - global $wgActiveUserDays; - $dbr = wfGetDB( DB_SLAVE, array( 'SpecialStatistics', 'vslow' ) ); - # Get non-bot users than did some recent action other than making accounts. - # If account creation is included, the number gets inflated ~20+ fold on enwiki. - $activeUsers = $dbr->selectField( - 'recentchanges', - 'COUNT( DISTINCT rc_user_text )', - array( - 'rc_user != 0', - 'rc_bot' => 0, - "rc_log_type != 'newusers' OR rc_log_type IS NULL", - "rc_timestamp >= '{$dbw->timestamp( wfTimestamp( TS_UNIX ) - $wgActiveUserDays*24*3600 )}'", - ), - __METHOD__ - ); - $dbw->update( - 'site_stats', - array( 'ss_active_users' => intval( $activeUsers ) ), - array( 'ss_row_id' => 1 ), - __METHOD__ - ); - return $activeUsers; - } -} - /** * Class designed for counting of stats. */ @@ -268,19 +289,27 @@ class SiteStatsInit { private $db; // Various stats - private $mEdits, $mArticles, $mPages, $mUsers, $mViews, $mFiles = 0; + private $mEdits = null, $mArticles = null, $mPages = null; + private $mUsers = null, $mFiles = null; /** - * Constructor - * @param $useMaster Boolean: whether to use the master DB + * @param bool|IDatabase $database + * - bool: Whether to use the master DB + * - IDatabase: Database connection to use */ - public function __construct( $useMaster = false ) { - $this->db = wfGetDB( $useMaster ? DB_MASTER : DB_SLAVE ); + public function __construct( $database = false ) { + if ( $database instanceof IDatabase ) { + $this->db = $database; + } elseif ( $database ) { + $this->db = wfGetDB( DB_MASTER ); + } else { + $this->db = wfGetDB( DB_REPLICA, 'vslow' ); + } } /** * Count the total number of edits - * @return Integer + * @return int */ public function edits() { $this->mEdits = $this->db->selectField( 'revision', 'COUNT(*)', '', __METHOD__ ); @@ -290,53 +319,55 @@ class SiteStatsInit { /** * Count pages in article space(s) - * @return Integer + * @return int */ public function articles() { - global $wgContentNamespaces; - $this->mArticles = $this->db->selectField( - 'page', - 'COUNT(*)', - array( - 'page_namespace' => $wgContentNamespaces, - 'page_is_redirect' => 0, - 'page_len > 0' - ), - __METHOD__ - ); + global $wgArticleCountMethod; + + $tables = [ 'page' ]; + $conds = [ + 'page_namespace' => MWNamespace::getContentNamespaces(), + 'page_is_redirect' => 0, + ]; + + if ( $wgArticleCountMethod == 'link' ) { + $tables[] = 'pagelinks'; + $conds[] = 'pl_from=page_id'; + } elseif ( $wgArticleCountMethod == 'comma' ) { + // To make a correct check for this, we would need, for each page, + // to load the text, maybe uncompress it, maybe decode it and then + // check if there's one comma. + // But one thing we are sure is that if the page is empty, it can't + // contain a comma :) + $conds[] = 'page_len > 0'; + } + + $this->mArticles = $this->db->selectField( $tables, 'COUNT(DISTINCT page_id)', + $conds, __METHOD__ ); return $this->mArticles; } /** * Count total pages - * @return Integer + * @return int */ public function pages() { $this->mPages = $this->db->selectField( 'page', 'COUNT(*)', '', __METHOD__ ); return $this->mPages; } - + /** * Count total users - * @return Integer + * @return int */ public function users() { $this->mUsers = $this->db->selectField( 'user', 'COUNT(*)', '', __METHOD__ ); return $this->mUsers; } - - /** - * Count views - * @return Integer - */ - public function views() { - $this->mViews = $this->db->selectField( 'page', 'SUM(page_counter)', '', __METHOD__ ); - return $this->mViews; - } /** * Count total files - * @return Integer + * @return int */ public function files() { $this->mFiles = $this->db->selectField( 'image', 'COUNT(*)', '', __METHOD__ ); @@ -345,72 +376,48 @@ class SiteStatsInit { /** * Do all updates and commit them. More or less a replacement - * for the original initStats, but without the calls to wfOut() - * @param $update Boolean: whether to update the current stats or write fresh - * @param $noViews Boolean: when true, do not update the number of page views - * @param $activeUsers Boolean: whether to update the number of active users + * for the original initStats, but without output. + * + * @param IDatabase|bool $database + * - bool: Whether to use the master DB + * - IDatabase: Database connection to use + * @param array $options Array of options, may contain the following values + * - activeUsers bool: Whether to update the number of active users (default: false) */ - public static function doAllAndCommit( $update, $noViews = false, $activeUsers = false ) { + public static function doAllAndCommit( $database, array $options = [] ) { + $options += [ 'update' => false, 'activeUsers' => false ]; + // Grab the object and count everything - $counter = new SiteStatsInit( false ); + $counter = new SiteStatsInit( $database ); + $counter->edits(); $counter->articles(); $counter->pages(); $counter->users(); $counter->files(); - // Only do views if we don't want to not count them - if( !$noViews ) { - $counter->views(); - } - - // Update/refresh - if( $update ) { - $counter->update(); - } else { - $counter->refresh(); - } + $counter->refresh(); // Count active users if need be - if( $activeUsers ) { + if ( $options['activeUsers'] ) { SiteStatsUpdate::cacheUpdate( wfGetDB( DB_MASTER ) ); } } /** - * Update the current row with the selected values - */ - public function update() { - list( $values, $conds ) = $this->getDbParams(); - $dbw = wfGetDB( DB_MASTER ); - $dbw->update( 'site_stats', $values, $conds, __METHOD__ ); - } - - /** - * Refresh site_stats. Erase the current record and save all - * the new values. + * Refresh site_stats */ public function refresh() { - list( $values, $conds, $views ) = $this->getDbParams(); - $dbw = wfGetDB( DB_MASTER ); - $dbw->delete( 'site_stats', $conds, __METHOD__ ); - $dbw->insert( 'site_stats', array_merge( $values, $conds, $views ), __METHOD__ ); - } + $values = [ + 'ss_row_id' => 1, + 'ss_total_edits' => ( $this->mEdits === null ? $this->edits() : $this->mEdits ), + 'ss_good_articles' => ( $this->mArticles === null ? $this->articles() : $this->mArticles ), + 'ss_total_pages' => ( $this->mPages === null ? $this->pages() : $this->mPages ), + 'ss_users' => ( $this->mUsers === null ? $this->users() : $this->mUsers ), + 'ss_images' => ( $this->mFiles === null ? $this->files() : $this->mFiles ), + ]; - /** - * Return three arrays of params for the db queries - * @return Array - */ - private function getDbParams() { - $values = array( - 'ss_total_edits' => $this->mEdits, - 'ss_good_articles' => $this->mArticles, - 'ss_total_pages' => $this->mPages, - 'ss_users' => $this->mUsers, - 'ss_images' => $this->mFiles - ); - $conds = array( 'ss_row_id' => 1 ); - $views = array( 'ss_total_views' => $this->mViews ); - return array( $values, $conds, $views ); + $dbw = wfGetDB( DB_MASTER ); + $dbw->upsert( 'site_stats', $values, [ 'ss_row_id' ], $values, __METHOD__ ); } }