includes/SiteStats.php

   1 <?php
   2
   3 /**
   4  * Static accessor class for site_stats and related things
   5  */
   6 class SiteStats {
   7         static $row, $loaded = false;
   8         static $admins, $jobs;
   9         static $pageCount = array();
  10         static $groupMemberCounts = array();
  11
  12         static function recache() {
  13                 self::load( true );
  14         }
  15
  16         static function load( $recache = false ) {
  17                 if ( self::$loaded && !$recache ) {
  18                         return;
  19                 }
  20
  21                 self::$row = self::loadAndLazyInit();
  22
  23                 # This code is somewhat schema-agnostic, because I'm changing it in a minor release -- TS
  24                 if ( !isset( self::$row->ss_total_pages ) && self::$row->ss_total_pages == -1 ) {
  25                         # Update schema
  26                         $u = new SiteStatsUpdate( 0, 0, 0 );
  27                         $u->doUpdate();
  28                         $dbr = wfGetDB( DB_SLAVE );
  29                         self::$row = $dbr->selectRow( 'site_stats', '*', false, __METHOD__ );
  30                 }
  31
  32                 self::$loaded = true;
  33         }
  34
  35         static function loadAndLazyInit() {
  36                 wfDebug( __METHOD__ . ": reading site_stats from slave\n" );
  37                 $row = self::doLoad( wfGetDB( DB_SLAVE ) );
  38
  39                 if( !self::isSane( $row ) ) {
  40                         // Might have just been initialized during this request? Underflow?
  41                         wfDebug( __METHOD__ . ": site_stats damaged or missing on slave\n" );
  42                         $row = self::doLoad( wfGetDB( DB_MASTER ) );
  43                 }
  44
  45                 if( !self::isSane( $row ) ) {
  46                         // Normally the site_stats table is initialized at install time.
  47                         // Some manual construction scenarios may leave the table empty or
  48                         // broken, however, for instance when importing from a dump into a
  49                         // clean schema with mwdumper.
  50                         wfDebug( __METHOD__ . ": initializing damaged or missing site_stats\n" );
  51
  52                         SiteStatsInit::doAllAndCommit( false );
  53
  54                         $row = self::doLoad( wfGetDB( DB_MASTER ) );
  55                 }
  56
  57                 if( !self::isSane( $row ) ) {
  58                         wfDebug( __METHOD__ . ": site_stats persistently nonsensical o_O\n" );
  59                 }
  60                 return $row;
  61         }
  62
  63         static function doLoad( $db ) {
  64                 return $db->selectRow( 'site_stats', '*', false, __METHOD__ );
  65         }
  66
  67         static function views() {
  68                 self::load();
  69                 return self::$row->ss_total_views;
  70         }
  71
  72         static function edits() {
  73                 self::load();
  74                 return self::$row->ss_total_edits;
  75         }
  76
  77         static function articles() {
  78                 self::load();
  79                 return self::$row->ss_good_articles;
  80         }
  81
  82         static function pages() {
  83                 self::load();
  84                 return self::$row->ss_total_pages;
  85         }
  86
  87         static function users() {
  88                 self::load();
  89                 return self::$row->ss_users;
  90         }
  91
  92         static function activeUsers() {
  93                 self::load();
  94                 return self::$row->ss_active_users;
  95         }
  96
  97         static function images() {
  98                 self::load();
  99                 return self::$row->ss_images;
 100         }
 101
 102         /**
 103          * @deprecated Use self::numberingroup('sysop') instead
 104          */
 105         static function admins() {
 106                 wfDeprecated(__METHOD__);
 107                 return self::numberingroup('sysop');
 108         }
 109
 110         /**
 111          * Find the number of users in a given user group.
 112          * @param string $group Name of group
 113          * @return int
 114          */
 115         static function numberingroup($group) {
 116                 if ( !isset( self::$groupMemberCounts[$group] ) ) {
 117                         global $wgMemc;
 118                         $key = wfMemcKey( 'SiteStats', 'groupcounts', $group );
 119                         $hit = $wgMemc->get( $key );
 120                         if ( !$hit ) {
 121                                 $dbr = wfGetDB( DB_SLAVE );
 122                                 $hit = $dbr->selectField( 'user_groups', 'COUNT(*)',
 123                                         array( 'ug_group' => $group ), __METHOD__ );
 124                                 $wgMemc->set( $key, $hit, 3600 );
 125                         }
 126                         self::$groupMemberCounts[$group] = $hit;
 127                 }
 128                 return self::$groupMemberCounts[$group];
 129         }
 130
 131         static function jobs() {
 132                 if ( !isset( self::$jobs ) ) {
 133                         $dbr = wfGetDB( DB_SLAVE );
 134                         self::$jobs = $dbr->estimateRowCount('job');
 135                         /* Zero rows still do single row read for row that doesn't exist, but people are annoyed by that */
 136                         if (self::$jobs == 1) {
 137                                 self::$jobs = 0;
 138                         }
 139                 }
 140                 return self::$jobs;
 141         }
 142
 143         static function pagesInNs( $ns ) {
 144                 wfProfileIn( __METHOD__ );
 145                 if( !isset( self::$pageCount[$ns] ) ) {
 146                         $dbr = wfGetDB( DB_SLAVE );
 147                         $pageCount[$ns] = (int)$dbr->selectField( 'page', 'COUNT(*)', array( 'page_namespace' => $ns ), __METHOD__ );
 148                 }
 149                 wfProfileOut( __METHOD__ );
 150                 return $pageCount[$ns];
 151         }
 152
 153         /** Is the provided row of site stats sane, or should it be regenerated? */
 154         private static function isSane( $row ) {
 155                 if(
 156                         $row === false
 157                         or $row->ss_total_pages < $row->ss_good_articles
 158                         or $row->ss_total_edits < $row->ss_total_pages
 159                         or $row->ss_users       < $row->ss_admins
 160                 ) {
 161                         return false;
 162                 }
 163                 // Now check for underflow/overflow
 164                 foreach( array( 'total_views', 'total_edits', 'good_articles',
 165                 'total_pages', 'users', 'admins', 'images' ) as $member ) {
 166                         if(
 167                                    $row->{"ss_$member"} > 2000000000
 168                                 or $row->{"ss_$member"} < 0
 169                         ) {
 170                                 return false;
 171                         }
 172                 }
 173                 return true;
 174         }
 175 }
 176
 177
 178 /**
 179  *
 180  */
 181 class SiteStatsUpdate {
 182
 183         var $mViews, $mEdits, $mGood, $mPages, $mUsers;
 184
 185         function __construct( $views, $edits, $good, $pages = 0, $users = 0 ) {
 186                 $this->mViews = $views;
 187                 $this->mEdits = $edits;
 188                 $this->mGood = $good;
 189                 $this->mPages = $pages;
 190                 $this->mUsers = $users;
 191         }
 192
 193         function appendUpdate( &$sql, $field, $delta ) {
 194                 if ( $delta ) {
 195                         if ( $sql ) {
 196                                 $sql .= ',';
 197                         }
 198                         if ( $delta < 0 ) {
 199                                 $sql .= "$field=$field-1";
 200                         } else {
 201                                 $sql .= "$field=$field+1";
 202                         }
 203                 }
 204         }
 205
 206         function doUpdate() {
 207                 $fname = 'SiteStatsUpdate::doUpdate';
 208                 $dbw = wfGetDB( DB_MASTER );
 209
 210                 $updates = '';
 211
 212                 $this->appendUpdate( $updates, 'ss_total_views', $this->mViews );
 213                 $this->appendUpdate( $updates, 'ss_total_edits', $this->mEdits );
 214                 $this->appendUpdate( $updates, 'ss_good_articles', $this->mGood );
 215                 $this->appendUpdate( $updates, 'ss_total_pages', $this->mPages );
 216                 $this->appendUpdate( $updates, 'ss_users', $this->mUsers );
 217
 218                 if ( $updates ) {
 219                         $site_stats = $dbw->tableName( 'site_stats' );
 220                         $sql = "UPDATE $site_stats SET $updates";
 221
 222                         # Need a separate transaction because this a global lock
 223                         $dbw->begin();
 224                         $dbw->query( $sql, $fname );
 225                         $dbw->commit();
 226                 }
 227         }
 228
 229         public static function cacheUpdate( $dbw ) {
 230                 $dbr = wfGetDB( DB_SLAVE, array( 'SpecialStatistics', 'vslow') );
 231                 # Get non-bot users than did some recent action other than making accounts.
 232                 # If account creation is included, the number gets inflated ~20+ fold on enwiki.
 233                 $activeUsers = $dbr->selectField( 'recentchanges', 'COUNT( DISTINCT rc_user_text )',
 234                         array( 'rc_user != 0', 'rc_bot' => 0, "rc_log_type != 'newusers' OR rc_log_type IS NULL" ),
 235                         __METHOD__ );
 236                 $dbw->update( 'site_stats',
 237                         array( 'ss_active_users' => intval($activeUsers) ),
 238                         array( 'ss_row_id' => 1 ), __METHOD__
 239                 );
 240                 return $activeUsers;
 241         }
 242 }
 243
 244 /**
 245  * Class designed for counting of stats.
 246  */
 247 class SiteStatsInit {
 248
 249         // Db connection
 250         private $db;
 251
 252         // Various stats
 253         private $mEdits, $mArticles, $mPages, $mUsers, $mViews, $mFiles = 0;
 254
 255         /**
 256          * Constructor
 257          * @param $useMaster bool Whether to use the master db
 258          */
 259         public function __construct( $useMaster = false ) {
 260                 $this->db = wfGetDB( $useMaster ? DB_MASTER : DB_SLAVE );
 261         }
 262
 263         /**
 264          * Count the total number of edits
 265          * @return int
 266          */
 267         public function edits() {
 268                 $this->mEdits = $this->db->selectField( 'revision', 'COUNT(*)', '', __METHOD__ );
 269                 $this->mEdits += $this->db->selectField( 'archive', 'COUNT(*)', '', __METHOD__ );
 270                 return $this->mEdits;
 271         }
 272
 273         /**
 274          * Count pages in article space
 275          * @return int
 276          */
 277         public function articles() {
 278                 global $wgContentNamespaces;
 279                 $this->mArticles = $this->db->selectField( 'page', 'COUNT(*)', array( 'page_namespace' => $wgContentNamespaces, 'page_is_redirect' => 0, 'page_len > 0' ), __METHOD__ );
 280                 return $this->mArticles;
 281         }
 282
 283         /**
 284          * Count total pages
 285          * @return int
 286          */
 287         public function pages() {
 288                 $this->mPages = $this->db->selectField( 'page', 'COUNT(*)', '', __METHOD__ );
 289                 return $this->mPages;
 290         }
 291
 292         /**
 293          * Count total users
 294          * @return int
 295          */
 296         public function users() {
 297                 $this->mUsers = $this->db->selectField( 'user', 'COUNT(*)', '', __METHOD__ );
 298                 return $this->mUsers;
 299         }
 300
 301         /**
 302          * Count views
 303          * @return int
 304          */
 305         public function views() {
 306                 $this->mViews = $this->db->selectField( 'page', 'SUM(page_counter)', '', __METHOD__ );
 307                 return $this->mViews;
 308         }
 309
 310         /**
 311          * Count total files
 312          * @return int
 313          */
 314         public function files() {
 315                 $this->mFiles = $this->db->selectField( 'image', 'COUNT(*)', '', __METHOD__ );
 316                 return $this->mFiles;
 317         }
 318
 319         /**
 320          * Do all updates and commit them. More or less a replacement
 321          * for the original initStats, but without the calls to wfOut()
 322          * @param $update bool Whether to update the current stats or write fresh
 323          * @param $noViews bool When true, do not update the number of page views
 324          * @param $activeUsers Whether to update the number of active users
 325          */
 326         public static function doAllAndCommit( $update, $noViews = false, $activeUsers = false ) {
 327                 // Grab the object and count everything
 328                 $counter = new SiteStatsInit( false );
 329                 $counter->edits();
 330                 $counter->articles();
 331                 $counter->pages();
 332                 $counter->users();
 333                 $counter->files();
 334
 335                 // Only do views if we don't want to not count them
 336                 if( !$noViews )
 337                         $counter->views();
 338
 339                 // Update/refresh
 340                 if( $update )
 341                         $counter->update();
 342                 else
 343                         $counter->refresh();
 344
 345                 // Count active users if need be
 346                 if( $activeUsers )
 347                         SiteStatsUpdate::cacheUpdate( wfGetDB( DB_MASTER ) );
 348         }
 349
 350         /**
 351          * Update the current row with the selected values
 352          */
 353         public function update() {
 354                 list( $values, $conds ) = $this->getDbParams();
 355                 $dbw = wfGetDB( DB_MASTER );
 356                 $dbw->update( 'site_stats', $values, $conds, __METHOD__ );
 357         }
 358
 359         /**
 360          * Refresh site_stats. Erase the current record and save all
 361          * the new values.
 362          */
 363         public function refresh() {
 364                 list( $values, $conds, $views ) = $this->getDbParams();
 365                 $dbw = wfGetDB( DB_MASTER );
 366                 $dbw->delete( 'site_stats', $conds, __METHOD__ );
 367                 $dbw->insert( 'site_stats', array_merge( $values, $conds, $views ), __METHOD__ );
 368         }
 369
 370         /**
 371          * Return three arrays of params for the db queries
 372          * @return array
 373          */
 374         private function getDbParams() {
 375                 $values = array( 'ss_total_edits' => $this->mEdits,
 376                                                 'ss_good_articles' => $this->mArticles,
 377                                                 'ss_total_pages' => $this->mPages,
 378                                                 'ss_users' => $this->mUsers,
 379                                                 'ss_admins' => SiteStats::numberingroup( 'sysop' ), // @todo make this go away
 380                                                 'ss_images' => $this->mFiles );
 381                 $conds = array( 'ss_row_id' => 1 );
 382                 $views = array( 'ss_total_views' => $this->mViews );
 383                 return array( $values, $conds, $views );
 384         }
 385 }