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 string $group 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 = array_sum( JobQueueGroup::singleton()->getQueueSizes() );
 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 ( $row === false
 230                         || $row->ss_total_pages < $row->ss_good_articles
 231                         || $row->ss_total_edits < $row->ss_total_pages
 232                         || $row->ss_users < $row->ss_active_users
 233                 ) {
 234                         return false;
 235                 }
 236                 // Now check for underflow/overflow
 237                 foreach ( array(
 238                         'ss_total_views',
 239                         'ss_total_edits',
 240                         'ss_good_articles',
 241                         'ss_total_pages',
 242                         'ss_users',
 243                         'ss_active_users',
 244                         'ss_images',
 245                 ) as $member ) {
 246                         if ( $row->$member > 2000000000 || $row->$member < 0 ) {
 247                                 return false;
 248                         }
 249                 }
 250                 return true;
 251         }
 252 }
 253
 254 /**
 255  * Class for handling updates to the site_stats table
 256  */
 257 class SiteStatsUpdate implements DeferrableUpdate {
 258         protected $views = 0;
 259         protected $edits = 0;
 260         protected $pages = 0;
 261         protected $articles = 0;
 262         protected $users = 0;
 263         protected $images = 0;
 264
 265         // @todo deprecate this constructor
 266         function __construct( $views, $edits, $good, $pages = 0, $users = 0 ) {
 267                 $this->views = $views;
 268                 $this->edits = $edits;
 269                 $this->articles = $good;
 270                 $this->pages = $pages;
 271                 $this->users = $users;
 272         }
 273
 274         /**
 275          * @param $deltas Array
 276          * @return SiteStatsUpdate
 277          */
 278         public static function factory( array $deltas ) {
 279                 $update = new self( 0, 0, 0 );
 280
 281                 $fields = array( 'views', 'edits', 'pages', 'articles', 'users', 'images' );
 282                 foreach ( $fields as $field ) {
 283                         if ( isset( $deltas[$field] ) && $deltas[$field] ) {
 284                                 $update->$field = $deltas[$field];
 285                         }
 286                 }
 287
 288                 return $update;
 289         }
 290
 291         public function doUpdate() {
 292                 global $wgSiteStatsAsyncFactor;
 293
 294                 $rate = $wgSiteStatsAsyncFactor; // convenience
 295                 // If set to do so, only do actual DB updates 1 every $rate times.
 296                 // The other times, just update "pending delta" values in memcached.
 297                 if ( $rate && ( $rate < 0 || mt_rand( 0, $rate - 1 ) != 0 ) ) {
 298                         $this->doUpdatePendingDeltas();
 299                 } else {
 300                         // Need a separate transaction because this a global lock
 301                         wfGetDB( DB_MASTER )->onTransactionIdle( array( $this, 'tryDBUpdateInternal' ) );
 302                 }
 303         }
 304
 305         /**
 306          * Do not call this outside of SiteStatsUpdate
 307          *
 308          * @return void
 309          */
 310         public function tryDBUpdateInternal() {
 311                 global $wgSiteStatsAsyncFactor;
 312
 313                 $dbw = wfGetDB( DB_MASTER );
 314                 $lockKey = wfMemcKey( 'site_stats' ); // prepend wiki ID
 315                 if ( $wgSiteStatsAsyncFactor ) {
 316                         // Lock the table so we don't have double DB/memcached updates
 317                         if ( !$dbw->lockIsFree( $lockKey, __METHOD__ )
 318                                 || !$dbw->lock( $lockKey, __METHOD__, 1 ) // 1 sec timeout
 319                         ) {
 320                                 $this->doUpdatePendingDeltas();
 321                                 return;
 322                         }
 323                         $pd = $this->getPendingDeltas();
 324                         // Piggy-back the async deltas onto those of this stats update....
 325                         $this->views += ( $pd['ss_total_views']['+'] - $pd['ss_total_views']['-'] );
 326                         $this->edits += ( $pd['ss_total_edits']['+'] - $pd['ss_total_edits']['-'] );
 327                         $this->articles += ( $pd['ss_good_articles']['+'] - $pd['ss_good_articles']['-'] );
 328                         $this->pages += ( $pd['ss_total_pages']['+'] - $pd['ss_total_pages']['-'] );
 329                         $this->users += ( $pd['ss_users']['+'] - $pd['ss_users']['-'] );
 330                         $this->images += ( $pd['ss_images']['+'] - $pd['ss_images']['-'] );
 331                 }
 332
 333                 // Build up an SQL query of deltas and apply them...
 334                 $updates = '';
 335                 $this->appendUpdate( $updates, 'ss_total_views', $this->views );
 336                 $this->appendUpdate( $updates, 'ss_total_edits', $this->edits );
 337                 $this->appendUpdate( $updates, 'ss_good_articles', $this->articles );
 338                 $this->appendUpdate( $updates, 'ss_total_pages', $this->pages );
 339                 $this->appendUpdate( $updates, 'ss_users', $this->users );
 340                 $this->appendUpdate( $updates, 'ss_images', $this->images );
 341                 if ( $updates != '' ) {
 342                         $dbw->update( 'site_stats', array( $updates ), array(), __METHOD__ );
 343                 }
 344
 345                 if ( $wgSiteStatsAsyncFactor ) {
 346                         // Decrement the async deltas now that we applied them
 347                         $this->removePendingDeltas( $pd );
 348                         // Commit the updates and unlock the table
 349                         $dbw->unlock( $lockKey, __METHOD__ );
 350                 }
 351         }
 352
 353         /**
 354          * @param $dbw DatabaseBase
 355          * @return bool|mixed
 356          */
 357         public static function cacheUpdate( $dbw ) {
 358                 global $wgActiveUserDays;
 359                 $dbr = wfGetDB( DB_SLAVE, array( 'SpecialStatistics', 'vslow' ) );
 360                 # Get non-bot users than did some recent action other than making accounts.
 361                 # If account creation is included, the number gets inflated ~20+ fold on enwiki.
 362                 $activeUsers = $dbr->selectField(
 363                         'recentchanges',
 364                         'COUNT( DISTINCT rc_user_text )',
 365                         array(
 366                                 'rc_user != 0',
 367                                 'rc_bot' => 0,
 368                                 'rc_log_type != ' . $dbr->addQuotes( 'newusers' ) . ' OR rc_log_type IS NULL',
 369                                 'rc_timestamp >= ' . $dbr->addQuotes( $dbr->timestamp( wfTimestamp( TS_UNIX ) - $wgActiveUserDays * 24 * 3600 ) ),
 370                         ),
 371                         __METHOD__
 372                 );
 373                 $dbw->update(
 374                         'site_stats',
 375                         array( 'ss_active_users' => intval( $activeUsers ) ),
 376                         array( 'ss_row_id' => 1 ),
 377                         __METHOD__
 378                 );
 379                 return $activeUsers;
 380         }
 381
 382         protected function doUpdatePendingDeltas() {
 383                 $this->adjustPending( 'ss_total_views', $this->views );
 384                 $this->adjustPending( 'ss_total_edits', $this->edits );
 385                 $this->adjustPending( 'ss_good_articles', $this->articles );
 386                 $this->adjustPending( 'ss_total_pages', $this->pages );
 387                 $this->adjustPending( 'ss_users', $this->users );
 388                 $this->adjustPending( 'ss_images', $this->images );
 389         }
 390
 391         /**
 392          * @param $sql string
 393          * @param $field string
 394          * @param $delta integer
 395          */
 396         protected function appendUpdate( &$sql, $field, $delta ) {
 397                 if ( $delta ) {
 398                         if ( $sql ) {
 399                                 $sql .= ',';
 400                         }
 401                         if ( $delta < 0 ) {
 402                                 $sql .= "$field=$field-" . abs( $delta );
 403                         } else {
 404                                 $sql .= "$field=$field+" . abs( $delta );
 405                         }
 406                 }
 407         }
 408
 409         /**
 410          * @param $type string
 411          * @param string $sign ('+' or '-')
 412          * @return string
 413          */
 414         private function getTypeCacheKey( $type, $sign ) {
 415                 return wfMemcKey( 'sitestatsupdate', 'pendingdelta', $type, $sign );
 416         }
 417
 418         /**
 419          * Adjust the pending deltas for a stat type.
 420          * Each stat type has two pending counters, one for increments and decrements
 421          * @param $type string
 422          * @param $delta integer Delta (positive or negative)
 423          * @return void
 424          */
 425         protected function adjustPending( $type, $delta ) {
 426                 global $wgMemc;
 427
 428                 if ( $delta < 0 ) { // decrement
 429                         $key = $this->getTypeCacheKey( $type, '-' );
 430                 } else { // increment
 431                         $key = $this->getTypeCacheKey( $type, '+' );
 432                 }
 433
 434                 $magnitude = abs( $delta );
 435                 if ( !$wgMemc->incr( $key, $magnitude ) ) { // not there?
 436                         if ( !$wgMemc->add( $key, $magnitude ) ) { // race?
 437                                 $wgMemc->incr( $key, $magnitude );
 438                         }
 439                 }
 440         }
 441
 442         /**
 443          * Get pending delta counters for each stat type
 444          * @return Array Positive and negative deltas for each type
 445          * @return void
 446          */
 447         protected function getPendingDeltas() {
 448                 global $wgMemc;
 449
 450                 $pending = array();
 451                 foreach ( array( 'ss_total_views', 'ss_total_edits',
 452                         'ss_good_articles', 'ss_total_pages', 'ss_users', 'ss_images' ) as $type )
 453                 {
 454                         // Get pending increments and pending decrements
 455                         $pending[$type]['+'] = (int)$wgMemc->get( $this->getTypeCacheKey( $type, '+' ) );
 456                         $pending[$type]['-'] = (int)$wgMemc->get( $this->getTypeCacheKey( $type, '-' ) );
 457                 }
 458
 459                 return $pending;
 460         }
 461
 462         /**
 463          * Reduce pending delta counters after updates have been applied
 464          * @param array $pd Result of getPendingDeltas(), used for DB update
 465          * @return void
 466          */
 467         protected function removePendingDeltas( array $pd ) {
 468                 global $wgMemc;
 469
 470                 foreach ( $pd as $type => $deltas ) {
 471                         foreach ( $deltas as $sign => $magnitude ) {
 472                                 // Lower the pending counter now that we applied these changes
 473                                 $wgMemc->decr( $this->getTypeCacheKey( $type, $sign ), $magnitude );
 474                         }
 475                 }
 476         }
 477 }
 478
 479 /**
 480  * Class designed for counting of stats.
 481  */
 482 class SiteStatsInit {
 483
 484         // Database connection
 485         private $db;
 486
 487         // Various stats
 488         private $mEdits, $mArticles, $mPages, $mUsers, $mViews, $mFiles = 0;
 489
 490         /**
 491          * Constructor
 492          * @param $database Boolean or DatabaseBase:
 493          * - Boolean: whether to use the master DB
 494          * - DatabaseBase: database connection to use
 495          */
 496         public function __construct( $database = false ) {
 497                 if ( $database instanceof DatabaseBase ) {
 498                         $this->db = $database;
 499                 } else {
 500                         $this->db = wfGetDB( $database ? DB_MASTER : DB_SLAVE );
 501                 }
 502         }
 503
 504         /**
 505          * Count the total number of edits
 506          * @return Integer
 507          */
 508         public function edits() {
 509                 $this->mEdits = $this->db->selectField( 'revision', 'COUNT(*)', '', __METHOD__ );
 510                 $this->mEdits += $this->db->selectField( 'archive', 'COUNT(*)', '', __METHOD__ );
 511                 return $this->mEdits;
 512         }
 513
 514         /**
 515          * Count pages in article space(s)
 516          * @return Integer
 517          */
 518         public function articles() {
 519                 global $wgArticleCountMethod;
 520
 521                 $tables = array( 'page' );
 522                 $conds = array(
 523                         'page_namespace' => MWNamespace::getContentNamespaces(),
 524                         'page_is_redirect' => 0,
 525                 );
 526
 527                 if ( $wgArticleCountMethod == 'link' ) {
 528                         $tables[] = 'pagelinks';
 529                         $conds[] = 'pl_from=page_id';
 530                 } elseif ( $wgArticleCountMethod == 'comma' ) {
 531                         // To make a correct check for this, we would need, for each page,
 532                         // to load the text, maybe uncompress it, maybe decode it and then
 533                         // check if there's one comma.
 534                         // But one thing we are sure is that if the page is empty, it can't
 535                         // contain a comma :)
 536                         $conds[] = 'page_len > 0';
 537                 }
 538
 539                 $this->mArticles = $this->db->selectField( $tables, 'COUNT(DISTINCT page_id)',
 540                         $conds, __METHOD__ );
 541                 return $this->mArticles;
 542         }
 543
 544         /**
 545          * Count total pages
 546          * @return Integer
 547          */
 548         public function pages() {
 549                 $this->mPages = $this->db->selectField( 'page', 'COUNT(*)', '', __METHOD__ );
 550                 return $this->mPages;
 551         }
 552
 553         /**
 554          * Count total users
 555          * @return Integer
 556          */
 557         public function users() {
 558                 $this->mUsers = $this->db->selectField( 'user', 'COUNT(*)', '', __METHOD__ );
 559                 return $this->mUsers;
 560         }
 561
 562         /**
 563          * Count views
 564          * @return Integer
 565          */
 566         public function views() {
 567                 $this->mViews = $this->db->selectField( 'page', 'SUM(page_counter)', '', __METHOD__ );
 568                 return $this->mViews;
 569         }
 570
 571         /**
 572          * Count total files
 573          * @return Integer
 574          */
 575         public function files() {
 576                 $this->mFiles = $this->db->selectField( 'image', 'COUNT(*)', '', __METHOD__ );
 577                 return $this->mFiles;
 578         }
 579
 580         /**
 581          * Do all updates and commit them. More or less a replacement
 582          * for the original initStats, but without output.
 583          *
 584          * @param $database DatabaseBase|bool
 585          * - Boolean: whether to use the master DB
 586          * - DatabaseBase: database connection to use
 587          * @param array $options of options, may contain the following values
 588          * - update Boolean: whether to update the current stats (true) or write fresh (false) (default: false)
 589          * - views Boolean: when true, do not update the number of page views (default: true)
 590          * - activeUsers Boolean: whether to update the number of active users (default: false)
 591          */
 592         public static function doAllAndCommit( $database, array $options = array() ) {
 593                 $options += array( 'update' => false, 'views' => true, 'activeUsers' => false );
 594
 595                 // Grab the object and count everything
 596                 $counter = new SiteStatsInit( $database );
 597
 598                 $counter->edits();
 599                 $counter->articles();
 600                 $counter->pages();
 601                 $counter->users();
 602                 $counter->files();
 603
 604                 // Only do views if we don't want to not count them
 605                 if ( $options['views'] ) {
 606                         $counter->views();
 607                 }
 608
 609                 // Update/refresh
 610                 if ( $options['update'] ) {
 611                         $counter->update();
 612                 } else {
 613                         $counter->refresh();
 614                 }
 615
 616                 // Count active users if need be
 617                 if ( $options['activeUsers'] ) {
 618                         SiteStatsUpdate::cacheUpdate( wfGetDB( DB_MASTER ) );
 619                 }
 620         }
 621
 622         /**
 623          * Update the current row with the selected values
 624          */
 625         public function update() {
 626                 list( $values, $conds ) = $this->getDbParams();
 627                 $dbw = wfGetDB( DB_MASTER );
 628                 $dbw->update( 'site_stats', $values, $conds, __METHOD__ );
 629         }
 630
 631         /**
 632          * Refresh site_stats. Erase the current record and save all
 633          * the new values.
 634          */
 635         public function refresh() {
 636                 list( $values, $conds, $views ) = $this->getDbParams();
 637                 $dbw = wfGetDB( DB_MASTER );
 638                 $dbw->delete( 'site_stats', $conds, __METHOD__ );
 639                 $dbw->insert( 'site_stats', array_merge( $values, $conds, $views ), __METHOD__ );
 640         }
 641
 642         /**
 643          * Return three arrays of params for the db queries
 644          * @return Array
 645          */
 646         private function getDbParams() {
 647                 $values = array(
 648                         'ss_total_edits' => $this->mEdits,
 649                         'ss_good_articles' => $this->mArticles,
 650                         'ss_total_pages' => $this->mPages,
 651                         'ss_users' => $this->mUsers,
 652                         'ss_images' => $this->mFiles
 653                 );
 654                 $conds = array( 'ss_row_id' => 1 );
 655                 $views = array( 'ss_total_views' => $this->mViews );
 656                 return array( $values, $conds, $views );
 657         }
 658 }