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