includes/LinkCache.php

   1 <?php
   2 /**
   3  * Cache for article titles (prefixed DB keys) and ids linked from one source
   4  * @package MediaWiki
   5  * @subpackage Cache
   6  */
   7
   8 /**
   9  *
  10  */
  11 # These are used in incrementalSetup()
  12 define ('LINKCACHE_GOOD', 0);
  13 define ('LINKCACHE_BAD', 1);
  14 define ('LINKCACHE_IMAGE', 2);
  15 define ('LINKCACHE_PAGE', 3);
  16
  17 /**
  18  * @package MediaWiki
  19  * @subpackage Cache
  20  */
  21 class LinkCache {
  22         // Increment $mClassVer whenever old serialized versions of this class
  23         // becomes incompatible with the new version.
  24         /* private */ var $mClassVer = 3;
  25
  26         /* private */ var $mPageLinks;
  27         /* private */ var $mGoodLinks, $mBadLinks, $mActive;
  28         /* private */ var $mImageLinks, $mCategoryLinks;
  29         /* private */ var $mPreFilled, $mOldGoodLinks, $mOldBadLinks;
  30         /* private */ var $mForUpdate;
  31
  32         /* private */ function getKey( $title ) {
  33                 global $wgDBname;
  34                 return $wgDBname.':lc:title:'.$title;
  35         }
  36
  37         function LinkCache() {
  38                 $this->mActive = true;
  39                 $this->mPreFilled = false;
  40                 $this->mForUpdate = false;
  41                 $this->mPageLinks = array();
  42                 $this->mGoodLinks = array();
  43                 $this->mBadLinks = array();
  44                 $this->mImageLinks = array();
  45                 $this->mCategoryLinks = array();
  46                 $this->mOldGoodLinks = array();
  47                 $this->mOldBadLinks = array();
  48                 $this->mOldPageLinks = array();
  49         }
  50
  51         /**
  52          * General accessor to get/set whether SELECT FOR UPDATE should be used
  53          */
  54         function forUpdate( $update = NULL ) {
  55                 return wfSetVar( $this->mForUpdate, $update );
  56         }
  57
  58         function getGoodLinkID( $title ) {
  59                 if ( array_key_exists( $title, $this->mGoodLinks ) ) {
  60                         return $this->mGoodLinks[$title];
  61                 } else {
  62                         return 0;
  63                 }
  64         }
  65
  66         function isBadLink( $title ) {
  67                 return array_key_exists( $title, $this->mBadLinks );
  68         }
  69
  70         function addGoodLinkObj( $id, $title ) {
  71                 if ( $this->mActive ) {
  72                         $dbkey = $title->getPrefixedDbKey();
  73                         $this->mGoodLinks[$dbkey] = $id;
  74                         $this->mPageLinks[$dbkey] = $title;
  75                 }
  76         }
  77
  78         function addBadLinkObj( $title ) {
  79                 $dbkey = $title->getPrefixedDbKey();
  80                 if ( $this->mActive && ( ! $this->isBadLink( $dbkey ) ) ) {
  81                         $this->mBadLinks[$dbkey] = 1;
  82                         $this->mPageLinks[$dbkey] = $title;
  83                 }
  84         }
  85
  86         function addImageLink( $title ) {
  87                 if ( $this->mActive ) { $this->mImageLinks[$title] = 1; }
  88         }
  89
  90         function addImageLinkObj( $nt ) {
  91                 if ( $this->mActive ) { $this->mImageLinks[$nt->getDBkey()] = 1; }
  92         }
  93
  94         function addCategoryLink( $title, $sortkey ) {
  95                 if ( $this->mActive ) { $this->mCategoryLinks[$title] = $sortkey; }
  96         }
  97
  98         function addCategoryLinkObj( &$nt, $sortkey ) {
  99                 $this->addCategoryLink( $nt->getDBkey(), $sortkey );
 100         }
 101
 102         function clearBadLink( $title ) {
 103                 unset( $this->mBadLinks[$title] );
 104                 $this->clearLink( $title );
 105         }
 106
 107         function clearLink( $title ) {
 108                 global $wgMemc, $wgLinkCacheMemcached;
 109                 if( $wgLinkCacheMemcached )
 110                         $wgMemc->delete( $this->getKey( $title ) );
 111         }
 112
 113         function suspend() { $this->mActive = false; }
 114         function resume() { $this->mActive = true; }
 115         function getPageLinks() { return $this->mPageLinks; }
 116         function getGoodLinks() { return $this->mGoodLinks; }
 117         function getBadLinks() { return array_keys( $this->mBadLinks ); }
 118         function getImageLinks() { return $this->mImageLinks; }
 119         function getCategoryLinks() { return $this->mCategoryLinks; }
 120
 121         function addLink( $title ) {
 122                 $nt = Title::newFromDBkey( $title );
 123                 if( $nt ) {
 124                         return $this->addLinkObj( $nt );
 125                 } else {
 126                         return 0;
 127                 }
 128         }
 129
 130         function addLinkObj( &$nt ) {
 131                 global $wgMemc, $wgLinkCacheMemcached, $wgAntiLockFlags;
 132                 $title = $nt->getPrefixedDBkey();
 133                 if ( $this->isBadLink( $title ) ) { return 0; }
 134                 $id = $this->getGoodLinkID( $title );
 135                 if ( 0 != $id ) { return $id; }
 136
 137                 $fname = 'LinkCache::addLinkObj';
 138                 wfProfileIn( $fname );
 139
 140                 $ns = $nt->getNamespace();
 141                 $t = $nt->getDBkey();
 142
 143                 if ( '' == $title ) {
 144                         wfProfileOut( $fname );
 145                         return 0;
 146                 }
 147
 148                 $id = NULL;
 149                 if( $wgLinkCacheMemcached )
 150                         $id = $wgMemc->get( $key = $this->getKey( $title ) );
 151                 if( ! is_integer( $id ) ) {
 152                         if ( $this->mForUpdate ) {
 153                                 $db =& wfGetDB( DB_MASTER );
 154                                 if ( !( $wgAntiLockFlags & ALF_NO_LINK_LOCK ) ) {
 155                                         $options = array( 'FOR UPDATE' );
 156                                 }
 157                         } else {
 158                                 $db =& wfGetDB( DB_SLAVE );
 159                                 $options = array();
 160                         }
 161
 162                         $id = $db->selectField( 'page', 'page_id', array( 'page_namespace' => $ns, 'page_title' => $t ), $fname, $options );
 163                         if ( !$id ) {
 164                                 $id = 0;
 165                         }
 166                         if( $wgLinkCacheMemcached )
 167                                 $wgMemc->add( $key, $id, 3600*24 );
 168                 }
 169
 170                 if( 0 == $id ) {
 171                         $this->addBadLinkObj( $nt );
 172                 } else {
 173                         $this->addGoodLinkObj( $id, $nt );
 174                 }
 175                 wfProfileOut( $fname );
 176                 return $id;
 177         }
 178
 179         /**
 180          * Bulk-check the pagelinks and page arrays for existence info.
 181          * @param Title $fromtitle
 182          */
 183         function preFill( &$fromtitle ) {
 184                 global $wgEnablePersistentLC;
 185
 186                 $fname = 'LinkCache::preFill';
 187                 wfProfileIn( $fname );
 188
 189                 $this->suspend();
 190                 $id = $fromtitle->getArticleID();
 191                 $this->resume();
 192
 193                 if( $id == 0 ) {
 194                         wfDebug( "$fname - got id 0 for title '" . $fromtitle->getPrefixedDBkey() . "'\n" );
 195                         wfProfileOut( $fname );
 196                         return;
 197                 }
 198
 199                 if ( $wgEnablePersistentLC ) {
 200                         if( $this->fillFromLinkscc( $id ) ){
 201                                 wfProfileOut( $fname );
 202                                 return;
 203                         }
 204                 }
 205                 if ( $this->mForUpdate ) {
 206                         $db =& wfGetDB( DB_MASTER );
 207                         $options = 'FOR UPDATE';
 208                 } else {
 209                         $db =& wfGetDB( DB_SLAVE );
 210                         $options = '';
 211                 }
 212
 213                 $page = $db->tableName( 'page' );
 214                 $pagelinks = $db->tableName( 'pagelinks' );
 215
 216                 $sql = "SELECT page_id,pl_namespace,pl_title
 217                         FROM $pagelinks
 218                         LEFT JOIN $page
 219                         ON pl_namespace=page_namespace AND pl_title=page_title
 220                         WHERE pl_from=$id $options";
 221                 $res = $db->query( $sql, $fname );
 222                 while( $s = $db->fetchObject( $res ) ) {
 223                         $title = Title::makeTitle( $s->pl_namespace, $s->pl_title );
 224                         if( $s->page_id ) {
 225                                 $this->addGoodLinkObj( $s->page_id, $title );
 226                         } else {
 227                                 $this->addBadLinkObj( $title );
 228                         }
 229                 }
 230                 $this->mPreFilled = true;
 231
 232                 if ( $wgEnablePersistentLC ) {
 233                         $this->saveToLinkscc( $id );
 234                 }
 235                 wfProfileOut( $fname );
 236         }
 237
 238         function getGoodAdditions() {
 239                 return array_diff( $this->mGoodLinks, $this->mOldGoodLinks );
 240         }
 241
 242         function getBadAdditions() {
 243                 #wfDebug( "mOldBadLinks: " . implode( ', ', array_keys( $this->mOldBadLinks ) ) . "\n" );
 244                 #wfDebug( "mBadLinks: " . implode( ', ', array_keys( $this->mBadLinks ) ) . "\n" );
 245                 return array_values( array_diff( array_keys( $this->mBadLinks ), array_keys( $this->mOldBadLinks ) ) );
 246         }
 247
 248         function getImageAdditions() {
 249                 return array_diff_assoc( $this->mImageLinks, $this->mOldImageLinks );
 250         }
 251
 252         function getGoodDeletions() {
 253                 return array_diff( $this->mOldGoodLinks, $this->mGoodLinks );
 254         }
 255
 256         function getBadDeletions() {
 257                 return array_values( array_diff( array_keys( $this->mOldBadLinks ), array_keys( $this->mBadLinks ) ));
 258         }
 259
 260         function getImageDeletions() {
 261                 return array_diff_assoc( $this->mOldImageLinks, $this->mImageLinks );
 262         }
 263
 264         function getPageAdditions() {
 265                 $set = array_diff( array_keys( $this->mPageLinks ), array_keys( $this->mOldPageLinks ) );
 266                 $out = array();
 267                 foreach( $set as $key ) {
 268                         $out[$key] = $this->mPageLinks[$key];
 269                 }
 270                 return $out;
 271         }
 272
 273         function getPageDeletions() {
 274                 $set = array_diff( array_keys( $this->mOldPageLinks ), array_keys( $this->mPageLinks ) );
 275                 $out = array();
 276                 foreach( $set as $key ) {
 277                         $out[$key] = $this->mOldPageLinks[$key];
 278                 }
 279                 return $out;
 280         }
 281
 282         /**
 283          * Parameters:
 284          * @param $which is one of the LINKCACHE_xxx constants
 285          * @param $del,$add are the incremental update arrays which will be filled.
 286          *
 287          * @return Returns whether or not it's worth doing the incremental version.
 288          *
 289          * For example, if [[List of mathematical topics]] was blanked,
 290          * it would take a long, long time to do incrementally.
 291          */
 292         function incrementalSetup( $which, &$del, &$add ) {
 293                 if ( ! $this->mPreFilled ) {
 294                         return false;
 295                 }
 296
 297                 switch ( $which ) {
 298                         case LINKCACHE_GOOD:
 299                                 $old =& $this->mOldGoodLinks;
 300                                 $cur =& $this->mGoodLinks;
 301                                 $del = $this->getGoodDeletions();
 302                                 $add = $this->getGoodAdditions();
 303                                 break;
 304                         case LINKCACHE_BAD:
 305                                 $old =& $this->mOldBadLinks;
 306                                 $cur =& $this->mBadLinks;
 307                                 $del = $this->getBadDeletions();
 308                                 $add = $this->getBadAdditions();
 309                                 break;
 310                         case LINKCACHE_PAGE:
 311                                 $old =& $this->mOldPageLinks;
 312                                 $cur =& $this->mPageLinks;
 313                                 $del = $this->getPageDeletions();
 314                                 $add = $this->getPageAdditions();
 315                                 break;
 316                         default: # LINKCACHE_IMAGE
 317                                 return false;
 318                 }
 319
 320                 return true;
 321         }
 322
 323         /**
 324          * Clears cache
 325          */
 326         function clear() {
 327                 $this->mPageLinks = array();
 328                 $this->mGoodLinks = array();
 329                 $this->mBadLinks = array();
 330                 $this->mImageLinks = array();
 331                 $this->mCategoryLinks = array();
 332                 $this->mOldGoodLinks = array();
 333                 $this->mOldBadLinks = array();
 334                 $this->mOldPageLinks = array();
 335         }
 336
 337         /**
 338          * @access private
 339          */
 340         function fillFromLinkscc( $id ){
 341                 $fname = 'LinkCache::fillFromLinkscc';
 342
 343                 $id = IntVal( $id );
 344                 if ( $this->mForUpdate ) {
 345                         $db =& wfGetDB( DB_MASTER );
 346                         $options = 'FOR UPDATE';
 347                 } else {
 348                         $db =& wfGetDB( DB_SLAVE );
 349                         $options = '';
 350                 }
 351                 $raw = $db->selectField( 'linkscc', 'lcc_cacheobj', array( 'lcc_pageid' => $id ), $fname, $options );
 352                 if ( $raw === false ) {
 353                         return false;
 354                 }
 355
 356                 $cacheobj = false;
 357                 if( function_exists( 'gzuncompress' ) )
 358                         $cacheobj = @gzuncompress( $raw );
 359
 360                 if($cacheobj == FALSE){
 361                         $cacheobj = $raw;
 362                 }
 363                 $cc = @unserialize( $cacheobj );
 364                 if( isset( $cc->mClassVer ) and ($cc->mClassVer == $this->mClassVer ) ){
 365                         $this->mOldPageLinks = $this->mPageLinks = $cc->mPageLinks;
 366                         $this->mOldGoodLinks = $this->mGoodLinks = $cc->mGoodLinks;
 367                         $this->mOldBadLinks = $this->mBadLinks = $cc->mBadLinks;
 368                         $this->mPreFilled = true;
 369                         return TRUE;
 370                 } else {
 371                         return FALSE;
 372                 }
 373
 374         }
 375
 376         /**
 377          * @access private
 378          */
 379         function saveToLinkscc( $pid ){
 380                 global $wgCompressedPersistentLC;
 381                 if( $wgCompressedPersistentLC and function_exists( 'gzcompress' ) ) {
 382                         $ser = gzcompress( serialize( $this ), 3 );
 383                 } else {
 384                         $ser = serialize( $this );
 385                 }
 386                 $db =& wfGetDB( DB_MASTER );
 387                 $db->replace( 'linkscc', array( 'lcc_pageid' ), array( 'lcc_pageid' => $pid, 'lcc_cacheobj' => $ser ) );
 388         }
 389
 390         /**
 391          * Delete linkscc rows which link to here
 392          * @param $title The page linked to
 393          * @static
 394          */
 395         function linksccClearLinksTo( $title ){
 396                 global $wgEnablePersistentLC;
 397                 if ( $wgEnablePersistentLC ) {
 398                         $fname = 'LinkCache::linksccClearLinksTo';
 399                         $pid = intval( $pid );
 400                         $dbw =& wfGetDB( DB_MASTER );
 401                         # Delete linkscc rows which link to here
 402                         $dbw->deleteJoin( 'linkscc', 'pagelinks', 'lcc_pageid', 'pl_from',
 403                                 array(
 404                                         'pl_namespace' => $title->getNamespace(),
 405                                         'pl-title'     => $title->getDbKey() ),
 406                                 $fname );
 407                         # Delete linkscc row representing this page
 408                         $dbw->delete( 'linkscc', array( 'lcc_pageid' => $pid ), $fname);
 409                 }
 410
 411         }
 412
 413         /**
 414          * @param $pid is a page id
 415          * @static
 416          */
 417         function linksccClearPage( $pid ){
 418                 global $wgEnablePersistentLC;
 419                 if ( $wgEnablePersistentLC ) {
 420                         $pid = intval( $pid );
 421                         $dbw =& wfGetDB( DB_MASTER );
 422                         $dbw->delete( 'linkscc', array( 'lcc_pageid' => $pid ) );
 423                 }
 424         }
 425
 426         /**
 427          * Swaps old and current link registers
 428          */
 429         function swapRegisters() {
 430                 swap( $this->mGoodLinks, $this->mOldGoodLinks );
 431                 swap( $this->mBadLinks, $this->mOldBadLinks );
 432                 swap( $this->mImageLinks, $this->mOldImageLinks );
 433         }
 434 }
 435
 436 /**
 437  * Class representing a list of titles
 438  * The execute() method checks them all for existence and adds them to a LinkCache object
 439  +
 440  * @package MediaWikki
 441  * @subpackage Cache
 442  */
 443 class LinkBatch {
 444         /**
 445          * 2-d array, first index namespace, second index dbkey, value arbitrary
 446          */
 447         var $data = array();
 448
 449         function LinkBatch( $arr = array() ) {
 450                 foreach( $arr as $item ) {
 451                         $this->addObj( $item );
 452                 }
 453         }
 454
 455         function addObj( $title ) {
 456                 $this->add( $title->getNamespace(), $title->getDBkey() );
 457         }
 458
 459         function add( $ns, $dbkey ) {
 460                 if ( $ns < 0 ) {
 461                         return;
 462                 }
 463                 if ( !array_key_exists( $ns, $this->data ) ) {
 464                         $this->data[$ns] = array();
 465                 }
 466
 467                 $this->data[$ns][$dbkey] = 1;
 468         }
 469
 470         function execute( &$cache ) {
 471                 $fname = 'LinkBatch::execute';
 472                 $namespaces = array();
 473
 474                 if ( !count( $this->data ) ) {
 475                         return;
 476                 }
 477
 478                 wfProfileIn( $fname );
 479
 480                 // Construct query
 481                 // This is very similar to Parser::replaceLinkHolders
 482                 $dbr = wfGetDB( DB_SLAVE );
 483                 $page = $dbr->tableName( 'page' );
 484                 $sql = "SELECT page_id, page_namespace, page_title FROM $page WHERE "
 485                         . $this->constructSet( 'page', $dbr );
 486
 487                 // Do query
 488                 $res = $dbr->query( $sql, $fname );
 489
 490                 // Process results
 491                 // For each returned entry, add it to the list of good links, and remove it from $remaining
 492
 493                 $remaining = $this->data;
 494                 while ( $row = $dbr->fetchObject( $res ) ) {
 495                         $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 496                         $cache->addGoodLinkObj( $row->page_id, $title );
 497                         unset( $remaining[$row->page_namespace][$row->page_title] );
 498                 }
 499                 $dbr->freeResult( $res );
 500
 501                 // The remaining links in $data are bad links, register them as such
 502                 foreach ( $remaining as $ns => $dbkeys ) {
 503                         foreach ( $dbkeys as $dbkey => $nothing ) {
 504                                 $title = Title::makeTitle( $ns, $dbkey );
 505                                 $cache->addBadLinkObj( $title );
 506                         }
 507                 }
 508
 509                 wfProfileOut( $fname );
 510         }
 511
 512         /**
 513          * Construct a WHERE clause which will match all the given titles.
 514          * Give the appropriate table's field name prefix ('page', 'pl', etc).
 515          *
 516          * @param string $prefix
 517          * @return string
 518          * @access public
 519          */
 520         function constructSet( $prefix, $db ) {
 521                 $first = true;
 522                 $sql = '';
 523                 foreach ( $this->data as $ns => $dbkeys ) {
 524                         if ( !count( $dbkeys ) ) {
 525                                 continue;
 526                         }
 527
 528                         if ( $first ) {
 529                                 $first = false;
 530                         } else {
 531                                 $sql .= ' OR ';
 532                         }
 533                         $sql .= "({$prefix}_namespace=$ns AND {$prefix}_title IN (";
 534
 535                         $firstTitle = true;
 536                         foreach( $dbkeys as $dbkey => $nothing ) {
 537                                 if ( $firstTitle ) {
 538                                         $firstTitle = false;
 539                                 } else {
 540                                         $sql .= ',';
 541                                 }
 542                                 $sql .= $db->addQuotes( $dbkey );
 543                         }
 544
 545                         $sql .= '))';
 546                 }
 547                 return $sql;
 548         }
 549 }
 550
 551 ?>