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 $group String: name of group
 113          * @return Integer
 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(
 123                                         'user_groups',
 124                                         'COUNT(*)',
 125                                         array( 'ug_group' => $group ),
 126                                         __METHOD__
 127                                 );
 128                                 $wgMemc->set( $key, $hit, 3600 );
 129                         }
 130                         self::$groupMemberCounts[$group] = $hit;
 131                 }
 132                 return self::$groupMemberCounts[$group];
 133         }
 134
 135         static function jobs() {
 136                 if ( !isset( self::$jobs ) ) {
 137                         $dbr = wfGetDB( DB_SLAVE );
 138                         self::$jobs = $dbr->estimateRowCount( 'job' );
 139                         /* Zero rows still do single row read for row that doesn't exist, but people are annoyed by that */
 140                         if ( self::$jobs == 1 ) {
 141                                 self::$jobs = 0;
 142                         }
 143                 }
 144                 return self::$jobs;
 145         }
 146
 147         static function pagesInNs( $ns ) {
 148                 wfProfileIn( __METHOD__ );
 149                 if( !isset( self::$pageCount[$ns] ) ) {
 150                         $dbr = wfGetDB( DB_SLAVE );
 151                         $pageCount[$ns] = (int)$dbr->selectField(
 152                                 'page',
 153                                 'COUNT(*)',
 154                                 array( 'page_namespace' => $ns ),
 155                                 __METHOD__
 156                         );
 157                 }
 158                 wfProfileOut( __METHOD__ );
 159                 return $pageCount[$ns];
 160         }
 161
 162         /** Is the provided row of site stats sane, or should it be regenerated? */
 163         private static function isSane( $row ) {
 164                 if(
 165                         $row === false
 166                         or $row->ss_total_pages < $row->ss_good_articles
 167                         or $row->ss_total_edits < $row->ss_total_pages
 168                 ) {
 169                         return false;
 170                 }
 171                 // Now check for underflow/overflow
 172                 foreach( array( 'total_views', 'total_edits', 'good_articles',
 173                 'total_pages', 'users', 'admins', 'images' ) as $member ) {
 174                         if(
 175                                 $row->{"ss_$member"} > 2000000000
 176                                 or $row->{"ss_$member"} < 0
 177                         ) {
 178                                 return false;
 179                         }
 180                 }
 181                 return true;
 182         }
 183 }
 184
 185
 186 /**
 187  *
 188  */
 189 class SiteStatsUpdate {
 190
 191         var $mViews, $mEdits, $mGood, $mPages, $mUsers;
 192
 193         function __construct( $views, $edits, $good, $pages = 0, $users = 0 ) {
 194                 $this->mViews = $views;
 195                 $this->mEdits = $edits;
 196                 $this->mGood = $good;
 197                 $this->mPages = $pages;
 198                 $this->mUsers = $users;
 199         }
 200
 201         function appendUpdate( &$sql, $field, $delta ) {
 202                 if ( $delta ) {
 203                         if ( $sql ) {
 204                                 $sql .= ',';
 205                         }
 206                         if ( $delta < 0 ) {
 207                                 $sql .= "$field=$field-1";
 208                         } else {
 209                                 $sql .= "$field=$field+1";
 210                         }
 211                 }
 212         }
 213
 214         function doUpdate() {
 215                 $dbw = wfGetDB( DB_MASTER );
 216
 217                 $updates = '';
 218
 219                 $this->appendUpdate( $updates, 'ss_total_views', $this->mViews );
 220                 $this->appendUpdate( $updates, 'ss_total_edits', $this->mEdits );
 221                 $this->appendUpdate( $updates, 'ss_good_articles', $this->mGood );
 222                 $this->appendUpdate( $updates, 'ss_total_pages', $this->mPages );
 223                 $this->appendUpdate( $updates, 'ss_users', $this->mUsers );
 224
 225                 if ( $updates ) {
 226                         $site_stats = $dbw->tableName( 'site_stats' );
 227                         $sql = "UPDATE $site_stats SET $updates";
 228
 229                         # Need a separate transaction because this a global lock
 230                         $dbw->begin();
 231                         $dbw->query( $sql, __METHOD__ );
 232                         $dbw->commit();
 233                 }
 234         }
 235
 236         public static function cacheUpdate( $dbw ) {
 237                 global $wgActiveUserDays;
 238                 $dbr = wfGetDB( DB_SLAVE, array( 'SpecialStatistics', 'vslow' ) );
 239                 # Get non-bot users than did some recent action other than making accounts.
 240                 # If account creation is included, the number gets inflated ~20+ fold on enwiki.
 241                 $activeUsers = $dbr->selectField(
 242                         'recentchanges',
 243                         'COUNT( DISTINCT rc_user_text )',
 244                         array(
 245                                 'rc_user != 0',
 246                                 'rc_bot' => 0,
 247                                 "rc_log_type != 'newusers' OR rc_log_type IS NULL",
 248                                 "rc_timestamp >= '{$dbw->timestamp( wfTimestamp( TS_UNIX ) - $wgActiveUserDays*24*3600 )}'",
 249                         ),
 250                         __METHOD__
 251                 );
 252                 $dbw->update(
 253                         'site_stats',
 254                         array( 'ss_active_users' => intval( $activeUsers ) ),
 255                         array( 'ss_row_id' => 1 ),
 256                         __METHOD__
 257                 );
 258                 return $activeUsers;
 259         }
 260 }
 261
 262 /**
 263  * Class designed for counting of stats.
 264  */
 265 class SiteStatsInit {
 266
 267         // Database connection
 268         private $db;
 269
 270         // Various stats
 271         private $mEdits, $mArticles, $mPages, $mUsers, $mViews, $mFiles = 0;
 272
 273         /**
 274          * Constructor
 275          * @param $useMaster Boolean: whether to use the master DB
 276          */
 277         public function __construct( $useMaster = false ) {
 278                 $this->db = wfGetDB( $useMaster ? DB_MASTER : DB_SLAVE );
 279         }
 280
 281         /**
 282          * Count the total number of edits
 283          * @return Integer
 284          */
 285         public function edits() {
 286                 $this->mEdits = $this->db->selectField( 'revision', 'COUNT(*)', '', __METHOD__ );
 287                 $this->mEdits += $this->db->selectField( 'archive', 'COUNT(*)', '', __METHOD__ );
 288                 return $this->mEdits;
 289         }
 290
 291         /**
 292          * Count pages in article space(s)
 293          * @return Integer
 294          */
 295         public function articles() {
 296                 global $wgContentNamespaces;
 297                 $this->mArticles = $this->db->selectField(
 298                         'page',
 299                         'COUNT(*)',
 300                         array(
 301                                 'page_namespace' => $wgContentNamespaces,
 302                                 'page_is_redirect' => 0,
 303                                 'page_len > 0'
 304                         ),
 305                         __METHOD__
 306                 );
 307                 return $this->mArticles;
 308         }
 309
 310         /**
 311          * Count total pages
 312          * @return Integer
 313          */
 314         public function pages() {
 315                 $this->mPages = $this->db->selectField( 'page', 'COUNT(*)', '', __METHOD__ );
 316                 return $this->mPages;
 317         }
 318
 319         /**
 320          * Count total users
 321          * @return Integer
 322          */
 323         public function users() {
 324                 $this->mUsers = $this->db->selectField( 'user', 'COUNT(*)', '', __METHOD__ );
 325                 return $this->mUsers;
 326         }
 327
 328         /**
 329          * Count views
 330          * @return Integer
 331          */
 332         public function views() {
 333                 $this->mViews = $this->db->selectField( 'page', 'SUM(page_counter)', '', __METHOD__ );
 334                 return $this->mViews;
 335         }
 336
 337         /**
 338          * Count total files
 339          * @return Integer
 340          */
 341         public function files() {
 342                 $this->mFiles = $this->db->selectField( 'image', 'COUNT(*)', '', __METHOD__ );
 343                 return $this->mFiles;
 344         }
 345
 346         /**
 347          * Do all updates and commit them. More or less a replacement
 348          * for the original initStats, but without the calls to wfOut()
 349          * @param $update Boolean: whether to update the current stats or write fresh
 350          * @param $noViews Boolean: when true, do not update the number of page views
 351          * @param $activeUsers Boolean: whether to update the number of active users
 352          */
 353         public static function doAllAndCommit( $update, $noViews = false, $activeUsers = false ) {
 354                 // Grab the object and count everything
 355                 $counter = new SiteStatsInit( false );
 356                 $counter->edits();
 357                 $counter->articles();
 358                 $counter->pages();
 359                 $counter->users();
 360                 $counter->files();
 361
 362                 // Only do views if we don't want to not count them
 363                 if( !$noViews ) {
 364                         $counter->views();
 365                 }
 366
 367                 // Update/refresh
 368                 if( $update ) {
 369                         $counter->update();
 370                 } else {
 371                         $counter->refresh();
 372                 }
 373
 374                 // Count active users if need be
 375                 if( $activeUsers ) {
 376                         SiteStatsUpdate::cacheUpdate( wfGetDB( DB_MASTER ) );
 377                 }
 378         }
 379
 380         /**
 381          * Update the current row with the selected values
 382          */
 383         public function update() {
 384                 list( $values, $conds ) = $this->getDbParams();
 385                 $dbw = wfGetDB( DB_MASTER );
 386                 $dbw->update( 'site_stats', $values, $conds, __METHOD__ );
 387         }
 388
 389         /**
 390          * Refresh site_stats. Erase the current record and save all
 391          * the new values.
 392          */
 393         public function refresh() {
 394                 list( $values, $conds, $views ) = $this->getDbParams();
 395                 $dbw = wfGetDB( DB_MASTER );
 396                 $dbw->delete( 'site_stats', $conds, __METHOD__ );
 397                 $dbw->insert( 'site_stats', array_merge( $values, $conds, $views ), __METHOD__ );
 398         }
 399
 400         /**
 401          * Return three arrays of params for the db queries
 402          * @return Array
 403          */
 404         private function getDbParams() {
 405                 $values = array(
 406                         'ss_total_edits' => $this->mEdits,
 407                         'ss_good_articles' => $this->mArticles,
 408                         'ss_total_pages' => $this->mPages,
 409                         'ss_users' => $this->mUsers,
 410                         'ss_images' => $this->mFiles
 411                 );
 412                 $conds = array( 'ss_row_id' => 1 );
 413                 $views = array( 'ss_total_views' => $this->mViews );
 414                 return array( $values, $conds, $views );
 415         }
 416 }