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 $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', '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         protected $views = 0;
 227         protected $edits = 0;
 228         protected $pages = 0;
 229         protected $goodPages = 0;
 230         protected $users = 0;
 231         protected $images = 0;
 232
 233         // @TODO: deprecate this
 234         function __construct( $views, $edits, $good, $pages = 0, $users = 0 ) {
 235                 $this->views = $views;
 236                 $this->edits = $edits;
 237                 $this->goodPages = $good;
 238                 $this->pages = $pages;
 239                 $this->users = $users;
 240         }
 241
 242         /**
 243          * @param $deltas Array
 244          * @return SiteStatsUpdate
 245          */
 246         public static function factory( array $deltas ) {
 247                 $update = new self( 0, 0, 0 );
 248
 249                 $fields = array( 'views', 'edits', 'pages', 'goodPages', 'users', 'images' );
 250                 foreach ( $fields as $field ) {
 251                         if ( isset( $deltas[$field] ) && $deltas[$field] ) {
 252                                 $update->$field = $deltas[$field];
 253                         }
 254                 }
 255
 256                 return $update;
 257         }
 258
 259         /**
 260          * @param $sql
 261          * @param $field
 262          * @param $delta
 263          */
 264         function appendUpdate( &$sql, $field, $delta ) {
 265                 if ( $delta ) {
 266                         if ( $sql ) {
 267                                 $sql .= ',';
 268                         }
 269                         if ( $delta < 0 ) {
 270                                 $sql .= "$field=$field-1";
 271                         } else {
 272                                 $sql .= "$field=$field+1";
 273                         }
 274                 }
 275         }
 276
 277         public function doUpdate() {
 278                 $dbw = wfGetDB( DB_MASTER );
 279
 280                 $updates = '';
 281
 282                 $this->appendUpdate( $updates, 'ss_total_views', $this->views );
 283                 $this->appendUpdate( $updates, 'ss_total_edits', $this->edits );
 284                 $this->appendUpdate( $updates, 'ss_good_articles', $this->goodPages );
 285                 $this->appendUpdate( $updates, 'ss_total_pages', $this->pages );
 286                 $this->appendUpdate( $updates, 'ss_users', $this->users );
 287                 $this->appendUpdate( $updates, 'ss_images', $this->images );
 288
 289                 if ( $updates ) {
 290                         $site_stats = $dbw->tableName( 'site_stats' );
 291                         $sql = "UPDATE $site_stats SET $updates";
 292
 293                         # Need a separate transaction because this a global lock
 294                         $dbw->begin( __METHOD__ );
 295                         $dbw->query( $sql, __METHOD__ );
 296                         $dbw->commit( __METHOD__ );
 297                 }
 298         }
 299
 300         /**
 301          * @param $dbw DatabaseBase
 302          * @return bool|mixed
 303          */
 304         public static function cacheUpdate( $dbw ) {
 305                 global $wgActiveUserDays;
 306                 $dbr = wfGetDB( DB_SLAVE, array( 'SpecialStatistics', 'vslow' ) );
 307                 # Get non-bot users than did some recent action other than making accounts.
 308                 # If account creation is included, the number gets inflated ~20+ fold on enwiki.
 309                 $activeUsers = $dbr->selectField(
 310                         'recentchanges',
 311                         'COUNT( DISTINCT rc_user_text )',
 312                         array(
 313                                 'rc_user != 0',
 314                                 'rc_bot' => 0,
 315                                 "rc_log_type != 'newusers' OR rc_log_type IS NULL",
 316                                 "rc_timestamp >= '{$dbw->timestamp( wfTimestamp( TS_UNIX ) - $wgActiveUserDays*24*3600 )}'",
 317                         ),
 318                         __METHOD__
 319                 );
 320                 $dbw->update(
 321                         'site_stats',
 322                         array( 'ss_active_users' => intval( $activeUsers ) ),
 323                         array( 'ss_row_id' => 1 ),
 324                         __METHOD__
 325                 );
 326                 return $activeUsers;
 327         }
 328 }
 329
 330 /**
 331  * Class designed for counting of stats.
 332  */
 333 class SiteStatsInit {
 334
 335         // Database connection
 336         private $db;
 337
 338         // Various stats
 339         private $mEdits, $mArticles, $mPages, $mUsers, $mViews, $mFiles = 0;
 340
 341         /**
 342          * Constructor
 343          * @param $database Boolean or DatabaseBase:
 344          * - Boolean: whether to use the master DB
 345          * - DatabaseBase: database connection to use
 346          */
 347         public function __construct( $database = false ) {
 348                 if ( $database instanceof DatabaseBase ) {
 349                         $this->db = $database;
 350                 } else {
 351                         $this->db = wfGetDB( $database ? DB_MASTER : DB_SLAVE );
 352                 }
 353         }
 354
 355         /**
 356          * Count the total number of edits
 357          * @return Integer
 358          */
 359         public function edits() {
 360                 $this->mEdits = $this->db->selectField( 'revision', 'COUNT(*)', '', __METHOD__ );
 361                 $this->mEdits += $this->db->selectField( 'archive', 'COUNT(*)', '', __METHOD__ );
 362                 return $this->mEdits;
 363         }
 364
 365         /**
 366          * Count pages in article space(s)
 367          * @return Integer
 368          */
 369         public function articles() {
 370                 global $wgArticleCountMethod;
 371
 372                 $tables = array( 'page' );
 373                 $conds = array(
 374                         'page_namespace' => MWNamespace::getContentNamespaces(),
 375                         'page_is_redirect' => 0,
 376                 );
 377
 378                 if ( $wgArticleCountMethod == 'link' ) {
 379                         $tables[] = 'pagelinks';
 380                         $conds[] = 'pl_from=page_id';
 381                 } elseif ( $wgArticleCountMethod == 'comma' ) {
 382                         // To make a correct check for this, we would need, for each page,
 383                         // to load the text, maybe uncompress it, maybe decode it and then
 384                         // check if there's one comma.
 385                         // But one thing we are sure is that if the page is empty, it can't
 386                         // contain a comma :)
 387                         $conds[] = 'page_len > 0';
 388                 }
 389
 390                 $this->mArticles = $this->db->selectField( $tables, 'COUNT(DISTINCT page_id)',
 391                         $conds, __METHOD__ );
 392                 return $this->mArticles;
 393         }
 394
 395         /**
 396          * Count total pages
 397          * @return Integer
 398          */
 399         public function pages() {
 400                 $this->mPages = $this->db->selectField( 'page', 'COUNT(*)', '', __METHOD__ );
 401                 return $this->mPages;
 402         }
 403
 404         /**
 405          * Count total users
 406          * @return Integer
 407          */
 408         public function users() {
 409                 $this->mUsers = $this->db->selectField( 'user', 'COUNT(*)', '', __METHOD__ );
 410                 return $this->mUsers;
 411         }
 412
 413         /**
 414          * Count views
 415          * @return Integer
 416          */
 417         public function views() {
 418                 $this->mViews = $this->db->selectField( 'page', 'SUM(page_counter)', '', __METHOD__ );
 419                 return $this->mViews;
 420         }
 421
 422         /**
 423          * Count total files
 424          * @return Integer
 425          */
 426         public function files() {
 427                 $this->mFiles = $this->db->selectField( 'image', 'COUNT(*)', '', __METHOD__ );
 428                 return $this->mFiles;
 429         }
 430
 431         /**
 432          * Do all updates and commit them. More or less a replacement
 433          * for the original initStats, but without output.
 434          *
 435          * @param $database DatabaseBase|bool
 436          * - Boolean: whether to use the master DB
 437          * - DatabaseBase: database connection to use
 438          * @param $options Array of options, may contain the following values
 439          * - update Boolean: whether to update the current stats (true) or write fresh (false) (default: false)
 440          * - views Boolean: when true, do not update the number of page views (default: true)
 441          * - activeUsers Boolean: whether to update the number of active users (default: false)
 442          */
 443         public static function doAllAndCommit( $database, array $options = array() ) {
 444                 $options += array( 'update' => false, 'views' => true, 'activeUsers' => false );
 445
 446                 // Grab the object and count everything
 447                 $counter = new SiteStatsInit( $database );
 448
 449                 $counter->edits();
 450                 $counter->articles();
 451                 $counter->pages();
 452                 $counter->users();
 453                 $counter->files();
 454
 455                 // Only do views if we don't want to not count them
 456                 if( $options['views'] ) {
 457                         $counter->views();
 458                 }
 459
 460                 // Update/refresh
 461                 if( $options['update'] ) {
 462                         $counter->update();
 463                 } else {
 464                         $counter->refresh();
 465                 }
 466
 467                 // Count active users if need be
 468                 if( $options['activeUsers'] ) {
 469                         SiteStatsUpdate::cacheUpdate( wfGetDB( DB_MASTER ) );
 470                 }
 471         }
 472
 473         /**
 474          * Update the current row with the selected values
 475          */
 476         public function update() {
 477                 list( $values, $conds ) = $this->getDbParams();
 478                 $dbw = wfGetDB( DB_MASTER );
 479                 $dbw->update( 'site_stats', $values, $conds, __METHOD__ );
 480         }
 481
 482         /**
 483          * Refresh site_stats. Erase the current record and save all
 484          * the new values.
 485          */
 486         public function refresh() {
 487                 list( $values, $conds, $views ) = $this->getDbParams();
 488                 $dbw = wfGetDB( DB_MASTER );
 489                 $dbw->delete( 'site_stats', $conds, __METHOD__ );
 490                 $dbw->insert( 'site_stats', array_merge( $values, $conds, $views ), __METHOD__ );
 491         }
 492
 493         /**
 494          * Return three arrays of params for the db queries
 495          * @return Array
 496          */
 497         private function getDbParams() {
 498                 $values = array(
 499                         'ss_total_edits' => $this->mEdits,
 500                         'ss_good_articles' => $this->mArticles,
 501                         'ss_total_pages' => $this->mPages,
 502                         'ss_users' => $this->mUsers,
 503                         'ss_images' => $this->mFiles
 504                 );
 505                 $conds = array( 'ss_row_id' => 1 );
 506                 $views = array( 'ss_total_views' => $this->mViews );
 507                 return array( $values, $conds, $views );
 508         }
 509 }