*/
class SiteStats {
static $row, $loaded = false;
- static $admins;
+ static $admins, $jobs;
static $pageCount = array();
+ static $groupMemberCounts = array();
static function recache() {
self::load( true );
}
+ /**
+ * @param $recache bool
+ */
static function load( $recache = false ) {
if ( self::$loaded && !$recache ) {
return;
$dbr = wfGetDB( DB_SLAVE );
self::$row = $dbr->selectRow( 'site_stats', '*', false, __METHOD__ );
}
+
+ self::$loaded = true;
}
-
+
+ /**
+ * @return Bool|ResultWrapper
+ */
static function loadAndLazyInit() {
wfDebug( __METHOD__ . ": reading site_stats from slave\n" );
$row = self::doLoad( wfGetDB( DB_SLAVE ) );
-
- if( $row === false ) {
- // Might have just been initialzed during this request?
- wfDebug( __METHOD__ . ": site_stats missing on slave\n" );
+
+ 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 ) );
}
-
- if( $row === false ) {
+
+ if( !self::isSane( $row ) ) {
// Normally the site_stats table is initialized at install time.
- // Some manual construction scenarios may leave the table empty,
- // however, for instance when importing from a dump into a clean
- // schema with mwdumper.
- wfDebug( __METHOD__ . ": initializing empty site_stats\n" );
-
- global $IP;
- require_once "$IP/maintenance/initStats.inc";
-
- ob_start();
- wfInitStats();
- ob_end_clean();
-
+ // 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( wfGetDB( DB_SLAVE ) );
+
$row = self::doLoad( wfGetDB( DB_MASTER ) );
}
-
- if( $row === false ) {
- wfDebug( __METHOD__ . ": init of site_stats failed o_O\n" );
+
+ if( !self::isSane( $row ) ) {
+ wfDebug( __METHOD__ . ": site_stats persistently nonsensical o_O\n" );
}
return $row;
}
+ /**
+ * @param $db DatabaseBase
+ * @return Bool|ResultWrapper
+ */
static function doLoad( $db ) {
return $db->selectRow( 'site_stats', '*', false, __METHOD__ );
}
+ /**
+ * @return int
+ */
static function views() {
self::load();
return self::$row->ss_total_views;
}
+ /**
+ * @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;
}
- static function admins() {
- if ( !isset( self::$admins ) ) {
+ /**
+ * Find the number of users in a given user group.
+ * @param $group String: name of group
+ * @return Integer
+ */
+ 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(
+ 'user_groups',
+ 'COUNT(*)',
+ array( 'ug_group' => $group ),
+ __METHOD__
+ );
+ $wgMemc->set( $key, $hit, 3600 );
+ }
+ self::$groupMemberCounts[$group] = $hit;
+ }
+ return self::$groupMemberCounts[$group];
+ }
+
+ /**
+ * @return int
+ */
+ static function jobs() {
+ if ( !isset( self::$jobs ) ) {
$dbr = wfGetDB( DB_SLAVE );
- self::$admins = $dbr->selectField( 'user_groups', 'COUNT(*)', array( 'ug_group' => 'sysop' ), __METHOD__ );
+ self::$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::$jobs = 0;
+ }
}
- return self::$admins;
+ return self::$jobs;
}
+ /**
+ * @param $ns int
+ *
+ * @return int
+ */
static function pagesInNs( $ns ) {
wfProfileIn( __METHOD__ );
if( !isset( self::$pageCount[$ns] ) ) {
$dbr = wfGetDB( DB_SLAVE );
- $pageCount[$ns] = (int)$dbr->selectField( 'page', 'COUNT(*)', array( 'page_namespace' => $ns ), __METHOD__ );
+ self::$pageCount[$ns] = (int)$dbr->selectField(
+ 'page',
+ 'COUNT(*)',
+ array( '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?
+ *
+ * @param $row
+ *
+ * @return bool
+ */
+ private static function isSane( $row ) {
+ 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
+ || $row->{"ss_$member"} < 0
+ ) {
+ return false;
+ }
+ }
+ return true;
+ }
}
-
/**
- *
+ * Class for handling updates to the site_stats table
*/
-class SiteStatsUpdate {
+class SiteStatsUpdate implements DeferrableUpdate {
var $mViews, $mEdits, $mGood, $mPages, $mUsers;
$this->mUsers = $users;
}
+ /**
+ * @param $sql
+ * @param $field
+ * @param $delta
+ */
function appendUpdate( &$sql, $field, $delta ) {
if ( $delta ) {
if ( $sql ) {
}
function doUpdate() {
- $fname = 'SiteStatsUpdate::doUpdate';
$dbw = wfGetDB( DB_MASTER );
- # First retrieve the row just to find out which schema we're in
- $row = $dbw->selectRow( 'site_stats', '*', false, $fname );
-
$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 ( isset( $row->ss_total_pages ) ) {
- # Update schema if required
- if ( $row->ss_total_pages == -1 && !$this->mViews ) {
- $dbr = wfGetDB( DB_SLAVE, array( 'SpecialStatistics', 'vslow') );
- list( $page, $user ) = $dbr->tableNamesN( 'page', 'user' );
-
- $sql = "SELECT COUNT(page_namespace) AS total FROM $page";
- $res = $dbr->query( $sql, $fname );
- $pageRow = $dbr->fetchObject( $res );
- $pages = $pageRow->total + $this->mPages;
-
- $sql = "SELECT COUNT(user_id) AS total FROM $user";
- $res = $dbr->query( $sql, $fname );
- $userRow = $dbr->fetchObject( $res );
- $users = $userRow->total + $this->mUsers;
-
- if ( $updates ) {
- $updates .= ',';
- }
- $updates .= "ss_total_pages=$pages, ss_users=$users";
- } else {
- $this->appendUpdate( $updates, 'ss_total_pages', $this->mPages );
- $this->appendUpdate( $updates, 'ss_users', $this->mUsers );
- }
- }
if ( $updates ) {
$site_stats = $dbw->tableName( 'site_stats' );
- $sql = $dbw->limitResultForUpdate("UPDATE $site_stats SET $updates", 1);
+ $sql = "UPDATE $site_stats SET $updates";
+
+ # Need a separate transaction because this a global lock
$dbw->begin();
- $dbw->query( $sql, $fname );
+ $dbw->query( $sql, __METHOD__ );
$dbw->commit();
}
+ }
- /*
- global $wgDBname, $wgTitle;
- if ( $this->mGood && $wgDBname == 'enwiki' ) {
- $good = $dbw->selectField( 'site_stats', 'ss_good_articles', '', $fname );
- error_log( $good . ' ' . $wgTitle->getPrefixedDBkey() . "\n", 3, '/home/wikipedia/logs/million.log' );
- }
- */
+ /**
+ * @param $dbw DatabaseBase
+ * @return bool|mixed
+ */
+ 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.
+ */
+class SiteStatsInit {
+
+ // Database connection
+ private $db;
+
+ // Various stats
+ private $mEdits, $mArticles, $mPages, $mUsers, $mViews, $mFiles = 0;
+
+ /**
+ * Constructor
+ * @param $database Boolean or DatabaseBase:
+ * - Boolean: whether to use the master DB
+ * - DatabaseBase: database connection to use
+ */
+ public function __construct( $database = false ) {
+ if ( $database instanceof DatabaseBase ) {
+ $this->db = $database;
+ } else {
+ $this->db = wfGetDB( $database ? DB_MASTER : DB_SLAVE );
+ }
+ }
+
+ /**
+ * Count the total number of edits
+ * @return Integer
+ */
+ public function edits() {
+ $this->mEdits = $this->db->selectField( 'revision', 'COUNT(*)', '', __METHOD__ );
+ $this->mEdits += $this->db->selectField( 'archive', 'COUNT(*)', '', __METHOD__ );
+ return $this->mEdits;
+ }
+
+ /**
+ * Count pages in article space(s)
+ * @return Integer
+ */
+ public function articles() {
+ global $wgArticleCountMethod;
+
+ $tables = array( 'page' );
+ $conds = array(
+ '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
+ */
+ public function pages() {
+ $this->mPages = $this->db->selectField( 'page', 'COUNT(*)', '', __METHOD__ );
+ return $this->mPages;
+ }
+
+ /**
+ * Count total users
+ * @return Integer
+ */
+ 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
+ */
+ public function files() {
+ $this->mFiles = $this->db->selectField( 'image', 'COUNT(*)', '', __METHOD__ );
+ return $this->mFiles;
+ }
+
+ /**
+ * Do all updates and commit them. More or less a replacement
+ * for the original initStats, but without output.
+ *
+ * @param $database Boolean or DatabaseBase:
+ * - Boolean: whether to use the master DB
+ * - DatabaseBase: database connection to use
+ * @param $options Array of options, may contain the following values
+ * - update Boolean: whether to update the current stats (true) or write fresh (false) (default: false)
+ * - views Boolean: when true, do not update the number of page views (default: true)
+ * - activeUsers Boolean: whether to update the number of active users (default: false)
+ */
+ public static function doAllAndCommit( $database, array $options = array() ) {
+ $options += array( 'update' => false, 'views' => true, 'activeUsers' => false );
+
+ // Grab the object and count everything
+ $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( $options['views'] ) {
+ $counter->views();
+ }
+
+ // Update/refresh
+ if( $options['update'] ) {
+ $counter->update();
+ } else {
+ $counter->refresh();
+ }
+
+ // Count active users if need be
+ 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.
+ */
+ 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__ );
+ }
+
+ /**
+ * 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 );
+ }
+}