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         /**
  17          * @param $recache bool
  18          */
  19         static function load( $recache = false ) {
  20                 if ( self::$loaded && !$recache ) {
  21                         return;
  22                 }
  23
  24                 self::$row = self::loadAndLazyInit();
  25
  26                 # This code is somewhat schema-agnostic, because I'm changing it in a minor release -- TS
  27                 if ( !isset( self::$row->ss_total_pages ) && self::$row->ss_total_pages == -1 ) {
  28                         # Update schema
  29                         $u = new SiteStatsUpdate( 0, 0, 0 );
  30                         $u->doUpdate();
  31                         $dbr = wfGetDB( DB_SLAVE );
  32                         self::$row = $dbr->selectRow( 'site_stats', '*', false, __METHOD__ );
  33                 }
  34
  35                 self::$loaded = true;
  36         }
  37
  38         /**
  39          * @return Bool|ResultWrapper
  40          */
  41         static function loadAndLazyInit() {
  42                 wfDebug( __METHOD__ . ": reading site_stats from slave\n" );
  43                 $row = self::doLoad( wfGetDB( DB_SLAVE ) );
  44
  45                 if( !self::isSane( $row ) ) {
  46                         // Might have just been initialized during this request? Underflow?
  47                         wfDebug( __METHOD__ . ": site_stats damaged or missing on slave\n" );
  48                         $row = self::doLoad( wfGetDB( DB_MASTER ) );
  49                 }
  50
  51                 if( !self::isSane( $row ) ) {
  52                         // Normally the site_stats table is initialized at install time.
  53                         // Some manual construction scenarios may leave the table empty or
  54                         // broken, however, for instance when importing from a dump into a
  55                         // clean schema with mwdumper.
  56                         wfDebug( __METHOD__ . ": initializing damaged or missing site_stats\n" );
  57
  58                         SiteStatsInit::doAllAndCommit( wfGetDB( DB_SLAVE ) );
  59
  60                         $row = self::doLoad( wfGetDB( DB_MASTER ) );
  61                 }
  62
  63                 if( !self::isSane( $row ) ) {
  64                         wfDebug( __METHOD__ . ": site_stats persistently nonsensical o_O\n" );
  65                 }
  66                 return $row;
  67         }
  68
  69         /**
  70          * @param $db DatabaseBase
  71          * @return Bool|ResultWrapper
  72          */
  73         static function doLoad( $db ) {
  74                 return $db->selectRow( 'site_stats', '*', false, __METHOD__ );
  75         }
  76
  77         /**
  78          * @return int
  79          */
  80         static function views() {
  81                 self::load();
  82                 return self::$row->ss_total_views;
  83         }
  84
  85         /**
  86          * @return int
  87          */
  88         static function edits() {
  89                 self::load();
  90                 return self::$row->ss_total_edits;
  91         }
  92
  93         /**
  94          * @return int
  95          */
  96         static function articles() {
  97                 self::load();
  98                 return self::$row->ss_good_articles;
  99         }
 100
 101         /**
 102          * @return int
 103          */
 104         static function pages() {
 105                 self::load();
 106                 return self::$row->ss_total_pages;
 107         }
 108
 109         /**
 110          * @return int
 111          */
 112         static function users() {
 113                 self::load();
 114                 return self::$row->ss_users;
 115         }
 116
 117         /**
 118          * @return int
 119          */
 120         static function activeUsers() {
 121                 self::load();
 122                 return self::$row->ss_active_users;
 123         }
 124
 125         /**
 126          * @return int
 127          */
 128         static function images() {
 129                 self::load();
 130                 return self::$row->ss_images;
 131         }
 132
 133         /**
 134          * Find the number of users in a given user group.
 135          * @param $group String: name of group
 136          * @return Integer
 137          */
 138         static function numberingroup( $group ) {
 139                 if ( !isset( self::$groupMemberCounts[$group] ) ) {
 140                         global $wgMemc;
 141                         $key = wfMemcKey( 'SiteStats', 'groupcounts', $group );
 142                         $hit = $wgMemc->get( $key );
 143                         if ( !$hit ) {
 144                                 $dbr = wfGetDB( DB_SLAVE );
 145                                 $hit = $dbr->selectField(
 146                                         'user_groups',
 147                                         'COUNT(*)',
 148                                         array( 'ug_group' => $group ),
 149                                         __METHOD__
 150                                 );
 151                                 $wgMemc->set( $key, $hit, 3600 );
 152                         }
 153                         self::$groupMemberCounts[$group] = $hit;
 154                 }
 155                 return self::$groupMemberCounts[$group];
 156         }
 157
 158         /**
 159          * @return int
 160          */
 161         static function jobs() {
 162                 if ( !isset( self::$jobs ) ) {
 163                         $dbr = wfGetDB( DB_SLAVE );
 164                         self::$jobs = $dbr->estimateRowCount( 'job' );
 165                         /* Zero rows still do single row read for row that doesn't exist, but people are annoyed by that */
 166                         if ( self::$jobs == 1 ) {
 167                                 self::$jobs = 0;
 168                         }
 169                 }
 170                 return self::$jobs;
 171         }
 172
 173         /**
 174          * @param $ns int
 175          *
 176          * @return int
 177          */
 178         static function pagesInNs( $ns ) {
 179                 wfProfileIn( __METHOD__ );
 180                 if( !isset( self::$pageCount[$ns] ) ) {
 181                         $dbr = wfGetDB( DB_SLAVE );
 182                         self::$pageCount[$ns] = (int)$dbr->selectField(
 183                                 'page',
 184                                 'COUNT(*)',
 185                                 array( 'page_namespace' => $ns ),
 186                                 __METHOD__
 187                         );
 188                 }
 189                 wfProfileOut( __METHOD__ );
 190                 return self::$pageCount[$ns];
 191         }
 192
 193         /**
 194          * Is the provided row of site stats sane, or should it be regenerated?
 195          *
 196          * @param $row
 197          *
 198          * @return bool
 199          */
 200         private static function isSane( $row ) {
 201                 if(
 202                         $row === false
 203                         || $row->ss_total_pages < $row->ss_good_articles
 204                         || $row->ss_total_edits < $row->ss_total_pages
 205                 ) {
 206                         return false;
 207                 }
 208                 // Now check for underflow/overflow
 209                 foreach( array( 'total_views', 'total_edits', 'good_articles',
 210                 'total_pages', 'users', 'admins', 'images' ) as $member ) {
 211                         if(
 212                                 $row->{"ss_$member"} > 2000000000
 213                                 || $row->{"ss_$member"} < 0
 214                         ) {
 215                                 return false;
 216                         }
 217                 }
 218                 return true;
 219         }
 220 }
 221
 222 /**
 223  * Class for handling updates to the site_stats table
 224  */
 225 class SiteStatsUpdate implements DeferrableUpdate {
 226
 227         var $mViews, $mEdits, $mGood, $mPages, $mUsers;
 228
 229         function __construct( $views, $edits, $good, $pages = 0, $users = 0 ) {
 230                 $this->mViews = $views;
 231                 $this->mEdits = $edits;
 232                 $this->mGood = $good;
 233                 $this->mPages = $pages;
 234                 $this->mUsers = $users;
 235         }
 236
 237         /**
 238          * @param $sql
 239          * @param $field
 240          * @param $delta
 241          */
 242         function appendUpdate( &$sql, $field, $delta ) {
 243                 if ( $delta ) {
 244                         if ( $sql ) {
 245                                 $sql .= ',';
 246                         }
 247                         if ( $delta < 0 ) {
 248                                 $sql .= "$field=$field-1";
 249                         } else {
 250                                 $sql .= "$field=$field+1";
 251                         }
 252                 }
 253         }
 254
 255         function doUpdate() {
 256                 $dbw = wfGetDB( DB_MASTER );
 257
 258                 $updates = '';
 259
 260                 $this->appendUpdate( $updates, 'ss_total_views', $this->mViews );
 261                 $this->appendUpdate( $updates, 'ss_total_edits', $this->mEdits );
 262                 $this->appendUpdate( $updates, 'ss_good_articles', $this->mGood );
 263                 $this->appendUpdate( $updates, 'ss_total_pages', $this->mPages );
 264                 $this->appendUpdate( $updates, 'ss_users', $this->mUsers );
 265
 266                 if ( $updates ) {
 267                         $site_stats = $dbw->tableName( 'site_stats' );
 268                         $sql = "UPDATE $site_stats SET $updates";
 269
 270                         # Need a separate transaction because this a global lock
 271                         $dbw->begin( __METHOD__ );
 272                         $dbw->query( $sql, __METHOD__ );
 273                         $dbw->commit( __METHOD__ );
 274                 }
 275         }
 276
 277         /**
 278          * @param $dbw DatabaseBase
 279          * @return bool|mixed
 280          */
 281         public static function cacheUpdate( $dbw ) {
 282                 global $wgActiveUserDays;
 283                 $dbr = wfGetDB( DB_SLAVE, array( 'SpecialStatistics', 'vslow' ) );
 284                 # Get non-bot users than did some recent action other than making accounts.
 285                 # If account creation is included, the number gets inflated ~20+ fold on enwiki.
 286                 $activeUsers = $dbr->selectField(
 287                         'recentchanges',
 288                         'COUNT( DISTINCT rc_user_text )',
 289                         array(
 290                                 'rc_user != 0',
 291                                 'rc_bot' => 0,
 292                                 "rc_log_type != 'newusers' OR rc_log_type IS NULL",
 293                                 "rc_timestamp >= '{$dbw->timestamp( wfTimestamp( TS_UNIX ) - $wgActiveUserDays*24*3600 )}'",
 294                         ),
 295                         __METHOD__
 296                 );
 297                 $dbw->update(
 298                         'site_stats',
 299                         array( 'ss_active_users' => intval( $activeUsers ) ),
 300                         array( 'ss_row_id' => 1 ),
 301                         __METHOD__
 302                 );
 303                 return $activeUsers;
 304         }
 305 }
 306
 307 /**
 308  * Class designed for counting of stats.
 309  */
 310 class SiteStatsInit {
 311
 312         // Database connection
 313         private $db;
 314
 315         // Various stats
 316         private $mEdits, $mArticles, $mPages, $mUsers, $mViews, $mFiles = 0;
 317
 318         /**
 319          * Constructor
 320          * @param $database Boolean or DatabaseBase:
 321          * - Boolean: whether to use the master DB
 322          * - DatabaseBase: database connection to use
 323          */
 324         public function __construct( $database = false ) {
 325                 if ( $database instanceof DatabaseBase ) {
 326                         $this->db = $database;
 327                 } else {
 328                         $this->db = wfGetDB( $database ? DB_MASTER : DB_SLAVE );
 329                 }
 330         }
 331
 332         /**
 333          * Count the total number of edits
 334          * @return Integer
 335          */
 336         public function edits() {
 337                 $this->mEdits = $this->db->selectField( 'revision', 'COUNT(*)', '', __METHOD__ );
 338                 $this->mEdits += $this->db->selectField( 'archive', 'COUNT(*)', '', __METHOD__ );
 339                 return $this->mEdits;
 340         }
 341
 342         /**
 343          * Count pages in article space(s)
 344          * @return Integer
 345          */
 346         public function articles() {
 347                 global $wgArticleCountMethod;
 348
 349                 $tables = array( 'page' );
 350                 $conds = array(
 351                         'page_namespace' => MWNamespace::getContentNamespaces(),
 352                         'page_is_redirect' => 0,
 353                 );
 354
 355                 if ( $wgArticleCountMethod == 'link' ) {
 356                         $tables[] = 'pagelinks';
 357                         $conds[] = 'pl_from=page_id';
 358                 } elseif ( $wgArticleCountMethod == 'comma' ) {
 359                         // To make a correct check for this, we would need, for each page,
 360                         // to load the text, maybe uncompress it, maybe decode it and then
 361                         // check if there's one comma.
 362                         // But one thing we are sure is that if the page is empty, it can't
 363                         // contain a comma :)
 364                         $conds[] = 'page_len > 0';
 365                 }
 366
 367                 $this->mArticles = $this->db->selectField( $tables, 'COUNT(DISTINCT page_id)',
 368                         $conds, __METHOD__ );
 369                 return $this->mArticles;
 370         }
 371
 372         /**
 373          * Count total pages
 374          * @return Integer
 375          */
 376         public function pages() {
 377                 $this->mPages = $this->db->selectField( 'page', 'COUNT(*)', '', __METHOD__ );
 378                 return $this->mPages;
 379         }
 380
 381         /**
 382          * Count total users
 383          * @return Integer
 384          */
 385         public function users() {
 386                 $this->mUsers = $this->db->selectField( 'user', 'COUNT(*)', '', __METHOD__ );
 387                 return $this->mUsers;
 388         }
 389
 390         /**
 391          * Count views
 392          * @return Integer
 393          */
 394         public function views() {
 395                 $this->mViews = $this->db->selectField( 'page', 'SUM(page_counter)', '', __METHOD__ );
 396                 return $this->mViews;
 397         }
 398
 399         /**
 400          * Count total files
 401          * @return Integer
 402          */
 403         public function files() {
 404                 $this->mFiles = $this->db->selectField( 'image', 'COUNT(*)', '', __METHOD__ );
 405                 return $this->mFiles;
 406         }
 407
 408         /**
 409          * Do all updates and commit them. More or less a replacement
 410          * for the original initStats, but without output.
 411          *
 412          * @param $database DatabaseBase|bool
 413          * - Boolean: whether to use the master DB
 414          * - DatabaseBase: database connection to use
 415          * @param $options Array of options, may contain the following values
 416          * - update Boolean: whether to update the current stats (true) or write fresh (false) (default: false)
 417          * - views Boolean: when true, do not update the number of page views (default: true)
 418          * - activeUsers Boolean: whether to update the number of active users (default: false)
 419          */
 420         public static function doAllAndCommit( $database, array $options = array() ) {
 421                 $options += array( 'update' => false, 'views' => true, 'activeUsers' => false );
 422
 423                 // Grab the object and count everything
 424                 $counter = new SiteStatsInit( $database );
 425
 426                 $counter->edits();
 427                 $counter->articles();
 428                 $counter->pages();
 429                 $counter->users();
 430                 $counter->files();
 431
 432                 // Only do views if we don't want to not count them
 433                 if( $options['views'] ) {
 434                         $counter->views();
 435                 }
 436
 437                 // Update/refresh
 438                 if( $options['update'] ) {
 439                         $counter->update();
 440                 } else {
 441                         $counter->refresh();
 442                 }
 443
 444                 // Count active users if need be
 445                 if( $options['activeUsers'] ) {
 446                         SiteStatsUpdate::cacheUpdate( wfGetDB( DB_MASTER ) );
 447                 }
 448         }
 449
 450         /**
 451          * Update the current row with the selected values
 452          */
 453         public function update() {
 454                 list( $values, $conds ) = $this->getDbParams();
 455                 $dbw = wfGetDB( DB_MASTER );
 456                 $dbw->update( 'site_stats', $values, $conds, __METHOD__ );
 457         }
 458
 459         /**
 460          * Refresh site_stats. Erase the current record and save all
 461          * the new values.
 462          */
 463         public function refresh() {
 464                 list( $values, $conds, $views ) = $this->getDbParams();
 465                 $dbw = wfGetDB( DB_MASTER );
 466                 $dbw->delete( 'site_stats', $conds, __METHOD__ );
 467                 $dbw->insert( 'site_stats', array_merge( $values, $conds, $views ), __METHOD__ );
 468         }
 469
 470         /**
 471          * Return three arrays of params for the db queries
 472          * @return Array
 473          */
 474         private function getDbParams() {
 475                 $values = array(
 476                         'ss_total_edits' => $this->mEdits,
 477                         'ss_good_articles' => $this->mArticles,
 478                         'ss_total_pages' => $this->mPages,
 479                         'ss_users' => $this->mUsers,
 480                         'ss_images' => $this->mFiles
 481                 );
 482                 $conds = array( 'ss_row_id' => 1 );
 483                 $views = array( 'ss_total_views' => $this->mViews );
 484                 return array( $values, $conds, $views );
 485         }
 486 }