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