includes/SiteStats.php

   1 <?php
   2 /**
   3  * Accessors and mutators for the site-wide statistics.
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  */
  22
  23 /**
  24  * Static accessor class for site_stats and related things
  25  */
  26 class SiteStats {
  27         static $row, $loaded = false;
  28         static $jobs;
  29         static $pageCount = array();
  30         static $groupMemberCounts = array();
  31
  32         static function recache() {
  33                 self::load( true );
  34         }
  35
  36         /**
  37          * @param $recache bool
  38          */
  39         static function load( $recache = false ) {
  40                 if ( self::$loaded && !$recache ) {
  41                         return;
  42                 }
  43
  44                 self::$row = self::loadAndLazyInit();
  45
  46                 # This code is somewhat schema-agnostic, because I'm changing it in a minor release -- TS
  47                 if ( !isset( self::$row->ss_total_pages ) && self::$row->ss_total_pages == -1 ) {
  48                         # Update schema
  49                         $u = new SiteStatsUpdate( 0, 0, 0 );
  50                         $u->doUpdate();
  51                         self::$row = self::doLoad( wfGetDB( DB_SLAVE ) );
  52                 }
  53
  54                 self::$loaded = true;
  55         }
  56
  57         /**
  58          * @return Bool|ResultWrapper
  59          */
  60         static function loadAndLazyInit() {
  61                 wfDebug( __METHOD__ . ": reading site_stats from slave\n" );
  62                 $row = self::doLoad( wfGetDB( DB_SLAVE ) );
  63
  64                 if( !self::isSane( $row ) ) {
  65                         // Might have just been initialized during this request? Underflow?
  66                         wfDebug( __METHOD__ . ": site_stats damaged or missing on slave\n" );
  67                         $row = self::doLoad( wfGetDB( DB_MASTER ) );
  68                 }
  69
  70                 if( !self::isSane( $row ) ) {
  71                         // Normally the site_stats table is initialized at install time.
  72                         // Some manual construction scenarios may leave the table empty or
  73                         // broken, however, for instance when importing from a dump into a
  74                         // clean schema with mwdumper.
  75                         wfDebug( __METHOD__ . ": initializing damaged or missing site_stats\n" );
  76
  77                         SiteStatsInit::doAllAndCommit( wfGetDB( DB_SLAVE ) );
  78
  79                         $row = self::doLoad( wfGetDB( DB_MASTER ) );
  80                 }
  81
  82                 if( !self::isSane( $row ) ) {
  83                         wfDebug( __METHOD__ . ": site_stats persistently nonsensical o_O\n" );
  84                 }
  85                 return $row;
  86         }
  87
  88         /**
  89          * @param $db DatabaseBase
  90          * @return Bool|ResultWrapper
  91          */
  92         static function doLoad( $db ) {
  93                 return $db->selectRow( 'site_stats', array(
  94                                 'ss_row_id',
  95                                 'ss_total_views',
  96                                 'ss_total_edits',
  97                                 'ss_good_articles',
  98                                 'ss_total_pages',
  99                                 'ss_users',
 100                                 'ss_active_users',
 101                                 'ss_images',
 102                         ), false, __METHOD__ );
 103         }
 104
 105         /**
 106          * @return int
 107          */
 108         static function views() {
 109                 self::load();
 110                 return self::$row->ss_total_views;
 111         }
 112
 113         /**
 114          * @return int
 115          */
 116         static function edits() {
 117                 self::load();
 118                 return self::$row->ss_total_edits;
 119         }
 120
 121         /**
 122          * @return int
 123          */
 124         static function articles() {
 125                 self::load();
 126                 return self::$row->ss_good_articles;
 127         }
 128
 129         /**
 130          * @return int
 131          */
 132         static function pages() {
 133                 self::load();
 134                 return self::$row->ss_total_pages;
 135         }
 136
 137         /**
 138          * @return int
 139          */
 140         static function users() {
 141                 self::load();
 142                 return self::$row->ss_users;
 143         }
 144
 145         /**
 146          * @return int
 147          */
 148         static function activeUsers() {
 149                 self::load();
 150                 return self::$row->ss_active_users;
 151         }
 152
 153         /**
 154          * @return int
 155          */
 156         static function images() {
 157                 self::load();
 158                 return self::$row->ss_images;
 159         }
 160
 161         /**
 162          * Find the number of users in a given user group.
 163          * @param $group String: name of group
 164          * @return Integer
 165          */
 166         static function numberingroup( $group ) {
 167                 if ( !isset( self::$groupMemberCounts[$group] ) ) {
 168                         global $wgMemc;
 169                         $key = wfMemcKey( 'SiteStats', 'groupcounts', $group );
 170                         $hit = $wgMemc->get( $key );
 171                         if ( !$hit ) {
 172                                 $dbr = wfGetDB( DB_SLAVE );
 173                                 $hit = $dbr->selectField(
 174                                         'user_groups',
 175                                         'COUNT(*)',
 176                                         array( 'ug_group' => $group ),
 177                                         __METHOD__
 178                                 );
 179                                 $wgMemc->set( $key, $hit, 3600 );
 180                         }
 181                         self::$groupMemberCounts[$group] = $hit;
 182                 }
 183                 return self::$groupMemberCounts[$group];
 184         }
 185
 186         /**
 187          * @return int
 188          */
 189         static function jobs() {
 190                 if ( !isset( self::$jobs ) ) {
 191                         $dbr = wfGetDB( DB_SLAVE );
 192                         self::$jobs = $dbr->estimateRowCount( 'job' );
 193                         /* Zero rows still do single row read for row that doesn't exist, but people are annoyed by that */
 194                         if ( self::$jobs == 1 ) {
 195                                 self::$jobs = 0;
 196                         }
 197                 }
 198                 return self::$jobs;
 199         }
 200
 201         /**
 202          * @param $ns int
 203          *
 204          * @return int
 205          */
 206         static function pagesInNs( $ns ) {
 207                 wfProfileIn( __METHOD__ );
 208                 if( !isset( self::$pageCount[$ns] ) ) {
 209                         $dbr = wfGetDB( DB_SLAVE );
 210                         self::$pageCount[$ns] = (int)$dbr->selectField(
 211                                 'page',
 212                                 'COUNT(*)',
 213                                 array( 'page_namespace' => $ns ),
 214                                 __METHOD__
 215                         );
 216                 }
 217                 wfProfileOut( __METHOD__ );
 218                 return self::$pageCount[$ns];
 219         }
 220
 221         /**
 222          * Is the provided row of site stats sane, or should it be regenerated?
 223          *
 224          * @param $row
 225          *
 226          * @return bool
 227          */
 228         private static function isSane( $row ) {
 229                 if(
 230                         $row === false
 231                         || $row->ss_total_pages < $row->ss_good_articles
 232                         || $row->ss_total_edits < $row->ss_total_pages
 233                 ) {
 234                         return false;
 235                 }
 236                 // Now check for underflow/overflow
 237                 foreach( array( 'total_views', 'total_edits', 'good_articles',
 238                 'total_pages', 'users', 'images' ) as $member ) {
 239                         if(
 240                                 $row->{"ss_$member"} > 2000000000
 241                                 || $row->{"ss_$member"} < 0
 242                         ) {
 243                                 return false;
 244                         }
 245                 }
 246                 return true;
 247         }
 248 }
 249
 250 /**
 251  * Class for handling updates to the site_stats table
 252  */
 253 class SiteStatsUpdate implements DeferrableUpdate {
 254         protected $views = 0;
 255         protected $edits = 0;
 256         protected $pages = 0;
 257         protected $articles = 0;
 258         protected $users = 0;
 259         protected $images = 0;
 260
 261         // @TODO: deprecate this constructor
 262         function __construct( $views, $edits, $good, $pages = 0, $users = 0 ) {
 263                 $this->views = $views;
 264                 $this->edits = $edits;
 265                 $this->articles = $good;
 266                 $this->pages = $pages;
 267                 $this->users = $users;
 268         }
 269
 270         /**
 271          * @param $deltas Array
 272          * @return SiteStatsUpdate
 273          */
 274         public static function factory( array $deltas ) {
 275                 $update = new self( 0, 0, 0 );
 276
 277                 $fields = array( 'views', 'edits', 'pages', 'articles', 'users', 'images' );
 278                 foreach ( $fields as $field ) {
 279                         if ( isset( $deltas[$field] ) && $deltas[$field] ) {
 280                                 $update->$field = $deltas[$field];
 281                         }
 282                 }
 283
 284                 return $update;
 285         }
 286
 287         public function doUpdate() {
 288                 global $wgSiteStatsAsyncFactor;
 289
 290                 $rate = $wgSiteStatsAsyncFactor; // convenience
 291                 // If set to do so, only do actual DB updates 1 every $rate times.
 292                 // The other times, just update "pending delta" values in memcached.
 293                 if ( $rate && ( $rate < 0 || mt_rand( 0, $rate - 1 ) != 0 ) ) {
 294                         $this->doUpdatePendingDeltas();
 295                 } else {
 296                         $dbw = wfGetDB( DB_MASTER );
 297                         // Need a separate transaction because this a global lock
 298                         $dbw->begin( __METHOD__ );
 299
 300                         $lockKey = wfMemcKey( 'site_stats' ); // prepend wiki ID
 301                         if ( $rate ) {
 302                                 // Lock the table so we don't have double DB/memcached updates
 303                                 if ( !$dbw->lockIsFree( $lockKey, __METHOD__ )
 304                                         || !$dbw->lock( $lockKey, __METHOD__, 1 ) // 1 sec timeout
 305                                 ) {
 306                                         $this->doUpdatePendingDeltas();
 307                                         return;
 308                                 }
 309                                 $pd = $this->getPendingDeltas();
 310                                 // Piggy-back the async deltas onto those of this stats update....
 311                                 $this->views    += ( $pd['ss_total_views']['+'] - $pd['ss_total_views']['-'] );
 312                                 $this->edits    += ( $pd['ss_total_edits']['+'] - $pd['ss_total_edits']['-'] );
 313                                 $this->articles += ( $pd['ss_good_articles']['+'] - $pd['ss_good_articles']['-'] );
 314                                 $this->pages    += ( $pd['ss_total_pages']['+'] - $pd['ss_total_pages']['-'] );
 315                                 $this->users    += ( $pd['ss_users']['+'] - $pd['ss_users']['-'] );
 316                                 $this->images   += ( $pd['ss_images']['+'] - $pd['ss_images']['-'] );
 317                         }
 318
 319                         // Build up an SQL query of deltas and apply them...
 320                         $updates = '';
 321                         $this->appendUpdate( $updates, 'ss_total_views', $this->views );
 322                         $this->appendUpdate( $updates, 'ss_total_edits', $this->edits );
 323                         $this->appendUpdate( $updates, 'ss_good_articles', $this->articles );
 324                         $this->appendUpdate( $updates, 'ss_total_pages', $this->pages );
 325                         $this->appendUpdate( $updates, 'ss_users', $this->users );
 326                         $this->appendUpdate( $updates, 'ss_images', $this->images );
 327                         if ( $updates != '' ) {
 328                                 $dbw->update( 'site_stats', array( $updates ), array(), __METHOD__ );
 329                         }
 330
 331                         if ( $rate ) {
 332                                 // Decrement the async deltas now that we applied them
 333                                 $this->removePendingDeltas( $pd );
 334                                 // Commit the updates and unlock the table
 335                                 $dbw->unlock( $lockKey, __METHOD__ );
 336                         }
 337
 338                         $dbw->commit( __METHOD__ );
 339                 }
 340         }
 341
 342         /**
 343          * @param $dbw DatabaseBase
 344          * @return bool|mixed
 345          */
 346         public static function cacheUpdate( $dbw ) {
 347                 global $wgActiveUserDays;
 348                 $dbr = wfGetDB( DB_SLAVE, array( 'SpecialStatistics', 'vslow' ) );
 349                 # Get non-bot users than did some recent action other than making accounts.
 350                 # If account creation is included, the number gets inflated ~20+ fold on enwiki.
 351                 $activeUsers = $dbr->selectField(
 352                         'recentchanges',
 353                         'COUNT( DISTINCT rc_user_text )',
 354                         array(
 355                                 'rc_user != 0',
 356                                 'rc_bot' => 0,
 357                                 'rc_log_type != ' . $dbr->addQuotes( 'newusers' ) . ' OR rc_log_type IS NULL',
 358                                 'rc_timestamp >= ' . $dbr->addQuotes( $dbr->timestamp( wfTimestamp( TS_UNIX ) - $wgActiveUserDays*24*3600 ) ),
 359                         ),
 360                         __METHOD__
 361                 );
 362                 $dbw->update(
 363                         'site_stats',
 364                         array( 'ss_active_users' => intval( $activeUsers ) ),
 365                         array( 'ss_row_id' => 1 ),
 366                         __METHOD__
 367                 );
 368                 return $activeUsers;
 369         }
 370
 371         protected function doUpdatePendingDeltas() {
 372                 $this->adjustPending( 'ss_total_views', $this->views );
 373                 $this->adjustPending( 'ss_total_edits', $this->edits );
 374                 $this->adjustPending( 'ss_good_articles', $this->articles );
 375                 $this->adjustPending( 'ss_total_pages', $this->pages );
 376                 $this->adjustPending( 'ss_users', $this->users );
 377                 $this->adjustPending( 'ss_images', $this->images );
 378         }
 379
 380         /**
 381          * @param $sql string
 382          * @param $field string
 383          * @param $delta integer
 384          */
 385         protected function appendUpdate( &$sql, $field, $delta ) {
 386                 if ( $delta ) {
 387                         if ( $sql ) {
 388                                 $sql .= ',';
 389                         }
 390                         if ( $delta < 0 ) {
 391                                 $sql .= "$field=$field-" . abs( $delta );
 392                         } else {
 393                                 $sql .= "$field=$field+" . abs( $delta );
 394                         }
 395                 }
 396         }
 397
 398         /**
 399          * @param $type string
 400          * @param $sign string ('+' or '-')
 401          * @return string
 402          */
 403         private function getTypeCacheKey( $type, $sign ) {
 404                 return wfMemcKey( 'sitestatsupdate', 'pendingdelta', $type, $sign );
 405         }
 406
 407         /**
 408          * Adjust the pending deltas for a stat type.
 409          * Each stat type has two pending counters, one for increments and decrements
 410          * @param $type string
 411          * @param $delta integer Delta (positive or negative)
 412          * @return void
 413          */
 414         protected function adjustPending( $type, $delta ) {
 415                 global $wgMemc;
 416
 417                 if ( $delta < 0 ) { // decrement
 418                         $key = $this->getTypeCacheKey( $type, '-' );
 419                 } else { // increment
 420                         $key = $this->getTypeCacheKey( $type, '+' );
 421                 }
 422
 423                 $magnitude = abs( $delta );
 424                 if ( !$wgMemc->incr( $key, $magnitude ) ) { // not there?
 425                         if ( !$wgMemc->add( $key, $magnitude ) ) { // race?
 426                                 $wgMemc->incr( $key, $magnitude );
 427                         }
 428                 }
 429         }
 430
 431         /**
 432          * Get pending delta counters for each stat type
 433          * @return Array Positive and negative deltas for each type
 434          * @return void
 435          */
 436         protected function getPendingDeltas() {
 437                 global $wgMemc;
 438
 439                 $pending = array();
 440                 foreach ( array( 'ss_total_views', 'ss_total_edits',
 441                         'ss_good_articles', 'ss_total_pages', 'ss_users', 'ss_images' ) as $type )
 442                 {
 443                         // Get pending increments and pending decrements
 444                         $pending[$type]['+'] = (int)$wgMemc->get( $this->getTypeCacheKey( $type, '+' ) );
 445                         $pending[$type]['-'] = (int)$wgMemc->get( $this->getTypeCacheKey( $type, '-' ) );
 446                 }
 447
 448                 return $pending;
 449         }
 450
 451         /**
 452          * Reduce pending delta counters after updates have been applied
 453          * @param Array $pd Result of getPendingDeltas(), used for DB update
 454          * @return void
 455          */
 456         protected function removePendingDeltas( array $pd ) {
 457                 global $wgMemc;
 458
 459                 foreach ( $pd as $type => $deltas ) {
 460                         foreach ( $deltas as $sign => $magnitude ) {
 461                                 // Lower the pending counter now that we applied these changes
 462                                 $wgMemc->decr( $this->getTypeCacheKey( $type, $sign ), $magnitude );
 463                         }
 464                 }
 465         }
 466 }
 467
 468 /**
 469  * Class designed for counting of stats.
 470  */
 471 class SiteStatsInit {
 472
 473         // Database connection
 474         private $db;
 475
 476         // Various stats
 477         private $mEdits, $mArticles, $mPages, $mUsers, $mViews, $mFiles = 0;
 478
 479         /**
 480          * Constructor
 481          * @param $database Boolean or DatabaseBase:
 482          * - Boolean: whether to use the master DB
 483          * - DatabaseBase: database connection to use
 484          */
 485         public function __construct( $database = false ) {
 486                 if ( $database instanceof DatabaseBase ) {
 487                         $this->db = $database;
 488                 } else {
 489                         $this->db = wfGetDB( $database ? DB_MASTER : DB_SLAVE );
 490                 }
 491         }
 492
 493         /**
 494          * Count the total number of edits
 495          * @return Integer
 496          */
 497         public function edits() {
 498                 $this->mEdits = $this->db->selectField( 'revision', 'COUNT(*)', '', __METHOD__ );
 499                 $this->mEdits += $this->db->selectField( 'archive', 'COUNT(*)', '', __METHOD__ );
 500                 return $this->mEdits;
 501         }
 502
 503         /**
 504          * Count pages in article space(s)
 505          * @return Integer
 506          */
 507         public function articles() {
 508                 global $wgArticleCountMethod;
 509
 510                 $tables = array( 'page' );
 511                 $conds = array(
 512                         'page_namespace' => MWNamespace::getContentNamespaces(),
 513                         'page_is_redirect' => 0,
 514                 );
 515
 516                 if ( $wgArticleCountMethod == 'link' ) {
 517                         $tables[] = 'pagelinks';
 518                         $conds[] = 'pl_from=page_id';
 519                 } elseif ( $wgArticleCountMethod == 'comma' ) {
 520                         // To make a correct check for this, we would need, for each page,
 521                         // to load the text, maybe uncompress it, maybe decode it and then
 522                         // check if there's one comma.
 523                         // But one thing we are sure is that if the page is empty, it can't
 524                         // contain a comma :)
 525                         $conds[] = 'page_len > 0';
 526                 }
 527
 528                 $this->mArticles = $this->db->selectField( $tables, 'COUNT(DISTINCT page_id)',
 529                         $conds, __METHOD__ );
 530                 return $this->mArticles;
 531         }
 532
 533         /**
 534          * Count total pages
 535          * @return Integer
 536          */
 537         public function pages() {
 538                 $this->mPages = $this->db->selectField( 'page', 'COUNT(*)', '', __METHOD__ );
 539                 return $this->mPages;
 540         }
 541
 542         /**
 543          * Count total users
 544          * @return Integer
 545          */
 546         public function users() {
 547                 $this->mUsers = $this->db->selectField( 'user', 'COUNT(*)', '', __METHOD__ );
 548                 return $this->mUsers;
 549         }
 550
 551         /**
 552          * Count views
 553          * @return Integer
 554          */
 555         public function views() {
 556                 $this->mViews = $this->db->selectField( 'page', 'SUM(page_counter)', '', __METHOD__ );
 557                 return $this->mViews;
 558         }
 559
 560         /**
 561          * Count total files
 562          * @return Integer
 563          */
 564         public function files() {
 565                 $this->mFiles = $this->db->selectField( 'image', 'COUNT(*)', '', __METHOD__ );
 566                 return $this->mFiles;
 567         }
 568
 569         /**
 570          * Do all updates and commit them. More or less a replacement
 571          * for the original initStats, but without output.
 572          *
 573          * @param $database DatabaseBase|bool
 574          * - Boolean: whether to use the master DB
 575          * - DatabaseBase: database connection to use
 576          * @param $options Array of options, may contain the following values
 577          * - update Boolean: whether to update the current stats (true) or write fresh (false) (default: false)
 578          * - views Boolean: when true, do not update the number of page views (default: true)
 579          * - activeUsers Boolean: whether to update the number of active users (default: false)
 580          */
 581         public static function doAllAndCommit( $database, array $options = array() ) {
 582                 $options += array( 'update' => false, 'views' => true, 'activeUsers' => false );
 583
 584                 // Grab the object and count everything
 585                 $counter = new SiteStatsInit( $database );
 586
 587                 $counter->edits();
 588                 $counter->articles();
 589                 $counter->pages();
 590                 $counter->users();
 591                 $counter->files();
 592
 593                 // Only do views if we don't want to not count them
 594                 if( $options['views'] ) {
 595                         $counter->views();
 596                 }
 597
 598                 // Update/refresh
 599                 if( $options['update'] ) {
 600                         $counter->update();
 601                 } else {
 602                         $counter->refresh();
 603                 }
 604
 605                 // Count active users if need be
 606                 if( $options['activeUsers'] ) {
 607                         SiteStatsUpdate::cacheUpdate( wfGetDB( DB_MASTER ) );
 608                 }
 609         }
 610
 611         /**
 612          * Update the current row with the selected values
 613          */
 614         public function update() {
 615                 list( $values, $conds ) = $this->getDbParams();
 616                 $dbw = wfGetDB( DB_MASTER );
 617                 $dbw->update( 'site_stats', $values, $conds, __METHOD__ );
 618         }
 619
 620         /**
 621          * Refresh site_stats. Erase the current record and save all
 622          * the new values.
 623          */
 624         public function refresh() {
 625                 list( $values, $conds, $views ) = $this->getDbParams();
 626                 $dbw = wfGetDB( DB_MASTER );
 627                 $dbw->delete( 'site_stats', $conds, __METHOD__ );
 628                 $dbw->insert( 'site_stats', array_merge( $values, $conds, $views ), __METHOD__ );
 629         }
 630
 631         /**
 632          * Return three arrays of params for the db queries
 633          * @return Array
 634          */
 635         private function getDbParams() {
 636                 $values = array(
 637                         'ss_total_edits' => $this->mEdits,
 638                         'ss_good_articles' => $this->mArticles,
 639                         'ss_total_pages' => $this->mPages,
 640                         'ss_users' => $this->mUsers,
 641                         'ss_images' => $this->mFiles
 642                 );
 643                 $conds = array( 'ss_row_id' => 1 );
 644                 $views = array( 'ss_total_views' => $this->mViews );
 645                 return array( $values, $conds, $views );
 646         }
 647 }