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->lock( $lockKey, __METHOD__, 1 ) ) {
 296                                         $dbw->commit( __METHOD__ );
 297                                         $this->doUpdatePendingDeltas();
 298                                         return;
 299                                 }
 300                                 $pd = $this->getPendingDeltas();
 301                                 // Piggy-back the async deltas onto those of this stats update....
 302                                 $this->views    += ( $pd['ss_total_views']['+'] - $pd['ss_total_views']['-'] );
 303                                 $this->edits    += ( $pd['ss_total_edits']['+'] - $pd['ss_total_edits']['-'] );
 304                                 $this->articles += ( $pd['ss_good_articles']['+'] - $pd['ss_good_articles']['-'] );
 305                                 $this->pages    += ( $pd['ss_total_pages']['+'] - $pd['ss_total_pages']['-'] );
 306                                 $this->users    += ( $pd['ss_users']['+'] - $pd['ss_users']['-'] );
 307                                 $this->images   += ( $pd['ss_images']['+'] - $pd['ss_images']['-'] );
 308                         }
 309
 310                         // Build up an SQL query of deltas and apply them...
 311                         $updates = '';
 312                         $this->appendUpdate( $updates, 'ss_total_views', $this->views );
 313                         $this->appendUpdate( $updates, 'ss_total_edits', $this->edits );
 314                         $this->appendUpdate( $updates, 'ss_good_articles', $this->articles );
 315                         $this->appendUpdate( $updates, 'ss_total_pages', $this->pages );
 316                         $this->appendUpdate( $updates, 'ss_users', $this->users );
 317                         $this->appendUpdate( $updates, 'ss_images', $this->images );
 318                         if ( $updates != '' ) {
 319                                 $dbw->update( 'site_stats', array( $updates ), array(), __METHOD__ );
 320                         }
 321
 322                         if ( $rate ) {
 323                                 // Decrement the async deltas now that we applied them
 324                                 $this->removePendingDeltas( $pd );
 325                                 // Commit the updates and unlock the table
 326                                 $dbw->unlock( $lockKey, __METHOD__ );
 327                         }
 328
 329                         $dbw->commit( __METHOD__ );
 330                 }
 331         }
 332
 333         /**
 334          * @param $dbw DatabaseBase
 335          * @return bool|mixed
 336          */
 337         public static function cacheUpdate( $dbw ) {
 338                 global $wgActiveUserDays;
 339                 $dbr = wfGetDB( DB_SLAVE, array( 'SpecialStatistics', 'vslow' ) );
 340                 # Get non-bot users than did some recent action other than making accounts.
 341                 # If account creation is included, the number gets inflated ~20+ fold on enwiki.
 342                 $activeUsers = $dbr->selectField(
 343                         'recentchanges',
 344                         'COUNT( DISTINCT rc_user_text )',
 345                         array(
 346                                 'rc_user != 0',
 347                                 'rc_bot' => 0,
 348                                 "rc_log_type != 'newusers' OR rc_log_type IS NULL",
 349                                 "rc_timestamp >= '{$dbw->timestamp( wfTimestamp( TS_UNIX ) - $wgActiveUserDays*24*3600 )}'",
 350                         ),
 351                         __METHOD__
 352                 );
 353                 $dbw->update(
 354                         'site_stats',
 355                         array( 'ss_active_users' => intval( $activeUsers ) ),
 356                         array( 'ss_row_id' => 1 ),
 357                         __METHOD__
 358                 );
 359                 return $activeUsers;
 360         }
 361
 362         protected function doUpdatePendingDeltas() {
 363                 $this->adjustPending( 'ss_total_views', $this->views );
 364                 $this->adjustPending( 'ss_total_edits', $this->edits );
 365                 $this->adjustPending( 'ss_good_articles', $this->articles );
 366                 $this->adjustPending( 'ss_total_pages', $this->pages );
 367                 $this->adjustPending( 'ss_users', $this->users );
 368                 $this->adjustPending( 'ss_images', $this->images );
 369         }
 370
 371         /**
 372          * @param $sql string
 373          * @param $field string
 374          * @param $delta integer
 375          */
 376         protected function appendUpdate( &$sql, $field, $delta ) {
 377                 if ( $delta ) {
 378                         if ( $sql ) {
 379                                 $sql .= ',';
 380                         }
 381                         if ( $delta < 0 ) {
 382                                 $sql .= "$field=$field-" . abs( $delta );
 383                         } else {
 384                                 $sql .= "$field=$field+" . abs( $delta );
 385                         }
 386                 }
 387         }
 388
 389         /**
 390          * @param $type string
 391          * @param $sign string ('+' or '-')
 392          * @return void
 393          */
 394         private function getTypeCacheKey( $type, $sign ) {
 395                 return wfMemcKey( 'sitestatsupdate', 'pendingdelta', $type, $sign );
 396         }
 397
 398         /**
 399          * Adjust the pending deltas for a stat type.
 400          * Each stat type has two pending counters, one for increments and decrements
 401          * @param $type string
 402          * @param $delta integer Delta (positive or negative)
 403          * @return void
 404          */
 405         protected function adjustPending( $type, $delta ) {
 406                 global $wgMemc;
 407
 408                 if ( $delta < 0 ) { // decrement
 409                         $key = $this->getTypeCacheKey( $type, '-' );
 410                 } else { // increment
 411                         $key = $this->getTypeCacheKey( $type, '+' );
 412                 }
 413
 414                 $magnitude = abs( $delta );
 415                 if ( !$wgMemc->incr( $key, $magnitude ) ) { // not there?
 416                         if ( !$wgMemc->add( $key, $magnitude ) ) { // race?
 417                                 $wgMemc->incr( $key, $magnitude );
 418                         }
 419                 }
 420         }
 421
 422         /**
 423          * Get pending delta counters for each stat type
 424          * @return Array Positive and negative deltas for each type
 425          * @return void
 426          */
 427         protected function getPendingDeltas() {
 428                 global $wgMemc;
 429
 430                 $pending = array();
 431                 foreach ( array( 'ss_total_views', 'ss_total_edits',
 432                         'ss_good_articles', 'ss_total_pages', 'ss_users', 'ss_images' ) as $type )
 433                 {
 434                         // Get pending increments and pending decrements
 435                         $pending[$type]['+'] = (int)$wgMemc->get( $this->getTypeCacheKey( $type, '+' ) );
 436                         $pending[$type]['-'] = (int)$wgMemc->get( $this->getTypeCacheKey( $type, '-' ) );
 437                 }
 438
 439                 return $pending;
 440         }
 441
 442         /**
 443          * Reduce pending delta counters after updates have been applied
 444          * @param Array Result of getPendingDeltas(), used for DB update
 445          * @return void
 446          */
 447         protected function removePendingDeltas( array $pd ) {
 448                 global $wgMemc;
 449
 450                 foreach ( $pd as $type => $deltas ) {
 451                         foreach ( $deltas as $sign => $magnitude ) {
 452                                 // Lower the pending counter now that we applied these changes
 453                                 $wgMemc->decr( $this->getTypeCacheKey( $type, $sign ), $magnitude );
 454                         }
 455                 }
 456         }
 457 }
 458
 459 /**
 460  * Class designed for counting of stats.
 461  */
 462 class SiteStatsInit {
 463
 464         // Database connection
 465         private $db;
 466
 467         // Various stats
 468         private $mEdits, $mArticles, $mPages, $mUsers, $mViews, $mFiles = 0;
 469
 470         /**
 471          * Constructor
 472          * @param $database Boolean or DatabaseBase:
 473          * - Boolean: whether to use the master DB
 474          * - DatabaseBase: database connection to use
 475          */
 476         public function __construct( $database = false ) {
 477                 if ( $database instanceof DatabaseBase ) {
 478                         $this->db = $database;
 479                 } else {
 480                         $this->db = wfGetDB( $database ? DB_MASTER : DB_SLAVE );
 481                 }
 482         }
 483
 484         /**
 485          * Count the total number of edits
 486          * @return Integer
 487          */
 488         public function edits() {
 489                 $this->mEdits = $this->db->selectField( 'revision', 'COUNT(*)', '', __METHOD__ );
 490                 $this->mEdits += $this->db->selectField( 'archive', 'COUNT(*)', '', __METHOD__ );
 491                 return $this->mEdits;
 492         }
 493
 494         /**
 495          * Count pages in article space(s)
 496          * @return Integer
 497          */
 498         public function articles() {
 499                 global $wgArticleCountMethod;
 500
 501                 $tables = array( 'page' );
 502                 $conds = array(
 503                         'page_namespace' => MWNamespace::getContentNamespaces(),
 504                         'page_is_redirect' => 0,
 505                 );
 506
 507                 if ( $wgArticleCountMethod == 'link' ) {
 508                         $tables[] = 'pagelinks';
 509                         $conds[] = 'pl_from=page_id';
 510                 } elseif ( $wgArticleCountMethod == 'comma' ) {
 511                         // To make a correct check for this, we would need, for each page,
 512                         // to load the text, maybe uncompress it, maybe decode it and then
 513                         // check if there's one comma.
 514                         // But one thing we are sure is that if the page is empty, it can't
 515                         // contain a comma :)
 516                         $conds[] = 'page_len > 0';
 517                 }
 518
 519                 $this->mArticles = $this->db->selectField( $tables, 'COUNT(DISTINCT page_id)',
 520                         $conds, __METHOD__ );
 521                 return $this->mArticles;
 522         }
 523
 524         /**
 525          * Count total pages
 526          * @return Integer
 527          */
 528         public function pages() {
 529                 $this->mPages = $this->db->selectField( 'page', 'COUNT(*)', '', __METHOD__ );
 530                 return $this->mPages;
 531         }
 532
 533         /**
 534          * Count total users
 535          * @return Integer
 536          */
 537         public function users() {
 538                 $this->mUsers = $this->db->selectField( 'user', 'COUNT(*)', '', __METHOD__ );
 539                 return $this->mUsers;
 540         }
 541
 542         /**
 543          * Count views
 544          * @return Integer
 545          */
 546         public function views() {
 547                 $this->mViews = $this->db->selectField( 'page', 'SUM(page_counter)', '', __METHOD__ );
 548                 return $this->mViews;
 549         }
 550
 551         /**
 552          * Count total files
 553          * @return Integer
 554          */
 555         public function files() {
 556                 $this->mFiles = $this->db->selectField( 'image', 'COUNT(*)', '', __METHOD__ );
 557                 return $this->mFiles;
 558         }
 559
 560         /**
 561          * Do all updates and commit them. More or less a replacement
 562          * for the original initStats, but without output.
 563          *
 564          * @param $database DatabaseBase|bool
 565          * - Boolean: whether to use the master DB
 566          * - DatabaseBase: database connection to use
 567          * @param $options Array of options, may contain the following values
 568          * - update Boolean: whether to update the current stats (true) or write fresh (false) (default: false)
 569          * - views Boolean: when true, do not update the number of page views (default: true)
 570          * - activeUsers Boolean: whether to update the number of active users (default: false)
 571          */
 572         public static function doAllAndCommit( $database, array $options = array() ) {
 573                 $options += array( 'update' => false, 'views' => true, 'activeUsers' => false );
 574
 575                 // Grab the object and count everything
 576                 $counter = new SiteStatsInit( $database );
 577
 578                 $counter->edits();
 579                 $counter->articles();
 580                 $counter->pages();
 581                 $counter->users();
 582                 $counter->files();
 583
 584                 // Only do views if we don't want to not count them
 585                 if( $options['views'] ) {
 586                         $counter->views();
 587                 }
 588
 589                 // Update/refresh
 590                 if( $options['update'] ) {
 591                         $counter->update();
 592                 } else {
 593                         $counter->refresh();
 594                 }
 595
 596                 // Count active users if need be
 597                 if( $options['activeUsers'] ) {
 598                         SiteStatsUpdate::cacheUpdate( wfGetDB( DB_MASTER ) );
 599                 }
 600         }
 601
 602         /**
 603          * Update the current row with the selected values
 604          */
 605         public function update() {
 606                 list( $values, $conds ) = $this->getDbParams();
 607                 $dbw = wfGetDB( DB_MASTER );
 608                 $dbw->update( 'site_stats', $values, $conds, __METHOD__ );
 609         }
 610
 611         /**
 612          * Refresh site_stats. Erase the current record and save all
 613          * the new values.
 614          */
 615         public function refresh() {
 616                 list( $values, $conds, $views ) = $this->getDbParams();
 617                 $dbw = wfGetDB( DB_MASTER );
 618                 $dbw->delete( 'site_stats', $conds, __METHOD__ );
 619                 $dbw->insert( 'site_stats', array_merge( $values, $conds, $views ), __METHOD__ );
 620         }
 621
 622         /**
 623          * Return three arrays of params for the db queries
 624          * @return Array
 625          */
 626         private function getDbParams() {
 627                 $values = array(
 628                         'ss_total_edits' => $this->mEdits,
 629                         'ss_good_articles' => $this->mArticles,
 630                         'ss_total_pages' => $this->mPages,
 631                         'ss_users' => $this->mUsers,
 632                         'ss_images' => $this->mFiles
 633                 );
 634                 $conds = array( 'ss_row_id' => 1 );
 635                 $views = array( 'ss_total_views' => $this->mViews );
 636                 return array( $values, $conds, $views );
 637         }
 638 }