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