includes/LinksUpdate.php

   1 <?php
   2 /**
   3  * See deferred.txt
   4  * @package MediaWiki
   5  */
   6
   7 /**
   8  * @todo document
   9  * @package MediaWiki
  10  */
  11 class LinksUpdate {
  12
  13         /**@{{
  14          * @private
  15          */
  16         private
  17                 $mId,            //!< Page ID of the article linked from
  18                 $mTitle,         //!< Title object of the article linked from
  19                 $mLinks,         //!< Map of title strings to IDs for the links in the document
  20                 $mImages,        //!< DB keys of the images used, in the array key only
  21                 $mTemplates,     //!< Map of title strings to IDs for the template references, including broken ones
  22                 $mExternals,     //!< URLs of external links, array key only
  23                 $mCategories,    //!< Map of category names to sort keys
  24                 $mInterlangs,    //!< Map of language codes to titles
  25                 $mDb,            //!< Database connection reference
  26                 $mOptions,       //!< SELECT options to be used (array)
  27                 $mRecursive;     //!< Whether to queue jobs for recursive updates
  28         /**@}}*/
  29
  30         /**
  31          * Constructor
  32          * Initialize private variables
  33          * @param $title Integer: FIXME
  34          * @param $parserOutput FIXME
  35          * @param $recursive Boolean: FIXME, default 'true'.
  36          */
  37         function LinksUpdate( $title, $parserOutput, $recursive = true ) {
  38                 global $wgAntiLockFlags;
  39
  40                 if ( $wgAntiLockFlags & ALF_NO_LINK_LOCK ) {
  41                         $this->mOptions = array();
  42                 } else {
  43                         $this->mOptions = array( 'FOR UPDATE' );
  44                 }
  45                 $this->mDb =& wfGetDB( DB_MASTER );
  46
  47                 if ( !is_object( $title ) ) {
  48                         wfDebugDieBacktrace( "The calling convention to LinksUpdate::LinksUpdate() has changed. " .
  49                                 "Please see Article::editUpdates() for an invocation example.\n" );
  50                 }
  51                 $this->mTitle = $title;
  52                 $this->mId = $title->getArticleID();
  53
  54                 $this->mLinks = $parserOutput->getLinks();
  55                 $this->mImages = $parserOutput->getImages();
  56                 $this->mTemplates = $parserOutput->getTemplates();
  57                 $this->mExternals = $parserOutput->getExternalLinks();
  58                 $this->mCategories = $parserOutput->getCategories();
  59
  60                 # Convert the format of the interlanguage links
  61                 # I didn't want to change it in the ParserOutput, because that array is passed all
  62                 # the way back to the skin, so either a skin API break would be required, or an
  63                 # inefficient back-conversion.
  64                 $ill = $parserOutput->getLanguageLinks();
  65                 $this->mInterlangs = array();
  66                 foreach ( $ill as $link ) {
  67                         list( $key, $title ) = explode( ':', $link, 2 );
  68                         $this->mInterlangs[$key] = $title;
  69                 }
  70
  71                 $this->mRecursive = $recursive;
  72         }
  73
  74         /**
  75          * Update link tables with outgoing links from an updated article
  76          */
  77         function doUpdate() {
  78                 global $wgUseDumbLinkUpdate;
  79                 if ( $wgUseDumbLinkUpdate ) {
  80                         $this->doDumbUpdate();
  81                 } else {
  82                         $this->doIncrementalUpdate();
  83                 }
  84         }
  85
  86         function doIncrementalUpdate() {
  87                 $fname = 'LinksUpdate::doIncrementalUpdate';
  88                 wfProfileIn( $fname );
  89
  90                 # Page links
  91                 $existing = $this->getExistingLinks();
  92                 $this->incrTableUpdate( 'pagelinks', 'pl', $this->getLinkDeletions( $existing ),
  93                         $this->getLinkInsertions( $existing ) );
  94
  95                 # Image links
  96                 $existing = $this->getExistingImages();
  97                 $this->incrTableUpdate( 'imagelinks', 'il', $this->getImageDeletions( $existing ),
  98                         $this->getImageInsertions( $existing ) );
  99
 100                 # Invalidate all image description pages which had links added or removed
 101                 $imageUpdates = array_diff_key( $existing, $this->mImages ) + array_diff_key( $this->mImages, $existing );
 102                 $this->invalidateImageDescriptions( $imageUpdates );
 103
 104                 # External links
 105                 $existing = $this->getExistingExternals();
 106                 $this->incrTableUpdate( 'externallinks', 'el', $this->getExternalDeletions( $existing ),
 107                 $this->getExternalInsertions( $existing ) );
 108
 109                 # Language links
 110                 $existing = $this->getExistingInterlangs();
 111                 $this->incrTableUpdate( 'langlinks', 'll', $this->getInterlangDeletions( $existing ),
 112                         $this->getInterlangInsertions( $existing ) );
 113
 114                 # Template links
 115                 $existing = $this->getExistingTemplates();
 116                 $this->incrTableUpdate( 'templatelinks', 'tl', $this->getTemplateDeletions( $existing ),
 117                         $this->getTemplateInsertions( $existing ) );
 118
 119                 # Refresh links of all pages including this page
 120                 if ( $this->mRecursive ) {
 121                         $tlto = $this->mTitle->getTemplateLinksTo();
 122                         if ( count( $tlto ) ) {
 123                                 require_once( 'JobQueue.php' );
 124                                 Job::queueLinksJobs( $tlto );
 125                         }
 126                 }
 127
 128                 # Category links
 129                 $existing = $this->getExistingCategories();
 130                 $this->incrTableUpdate( 'categorylinks', 'cl', $this->getCategoryDeletions( $existing ),
 131                         $this->getCategoryInsertions( $existing ) );
 132
 133                 # Invalidate all categories which were added, deleted or changed (set symmetric difference)
 134                 $categoryUpdates = array_diff_assoc( $existing, $this->mCategories ) + array_diff_assoc( $this->mCategories, $existing );
 135                 $this->invalidateCategories( $categoryUpdates );
 136
 137                 wfProfileOut( $fname );
 138         }
 139
 140         /**
 141          * Link update which clears the previous entries and inserts new ones
 142          * May be slower or faster depending on level of lock contention and write speed of DB
 143          * Also useful where link table corruption needs to be repaired, e.g. in refreshLinks.php
 144          */
 145         function doDumbUpdate() {
 146                 $fname = 'LinksUpdate::doDumbUpdate';
 147                 wfProfileIn( $fname );
 148
 149                 # Refresh category pages and image description pages
 150                 $existing = $this->getExistingCategories();
 151                 $categoryUpdates = array_diff_assoc( $existing, $this->mCategories ) + array_diff_assoc( $this->mCategories, $existing );
 152                 $existing = $this->getExistingImages();
 153                 $imageUpdates = array_diff_key( $existing, $this->mImages ) + array_diff_key( $this->mImages, $existing );
 154
 155                 # Refresh links of all pages including this page
 156                 if ( $this->mRecursive ) {
 157                         $tlto = $this->mTitle->getTemplateLinksTo();
 158                         if ( count( $tlto ) ) {
 159                                 require_once( 'JobQueue.php' );
 160                                 Job::queueLinksJobs( $tlto );
 161                         }
 162                 }
 163
 164                 $this->dumbTableUpdate( 'pagelinks',     $this->getLinkInsertions(),     'pl_from' );
 165                 $this->dumbTableUpdate( 'imagelinks',    $this->getImageInsertions(),    'il_from' );
 166                 $this->dumbTableUpdate( 'categorylinks', $this->getCategoryInsertions(), 'cl_from' );
 167                 $this->dumbTableUpdate( 'templatelinks', $this->getTemplateInsertions(), 'tl_from' );
 168                 $this->dumbTableUpdate( 'externallinks', $this->getExternalInsertions(), 'el_from' );
 169                 $this->dumbTableUpdate( 'langlinks',     $this->getInterlangInsertions(), 'll_from' );
 170
 171                 # Update the cache of all the category pages and image description pages which were changed
 172                 $this->invalidateCategories( $categoryUpdates );
 173                 $this->invalidateImageDescriptions( $imageUpdates );
 174
 175                 wfProfileOut( $fname );
 176         }
 177
 178         /**
 179          * Invalidate the cache of a list of pages from a single namespace
 180          *
 181          * @param integer $namespace
 182          * @param array $dbkeys
 183          */
 184         function invalidatePages( $namespace, $dbkeys ) {
 185                 $fname = 'LinksUpdate::invalidatePages';
 186
 187                 if ( !count( $dbkeys ) ) {
 188                         return;
 189                 }
 190
 191                 /**
 192                  * Determine which pages need to be updated
 193                  * This is necessary to prevent the job queue from smashing the DB with
 194                  * large numbers of concurrent invalidations of the same page
 195                  */
 196                 $now = $this->mDb->timestamp();
 197                 $ids = array();
 198                 $res = $this->mDb->select( 'page', array( 'page_id' ),
 199                         array(
 200                                 'page_namespace' => $namespace,
 201                                 'page_title IN (' . $this->mDb->makeList( $dbkeys ) . ')',
 202                                 'page_touched < ' . $this->mDb->addQuotes( $now )
 203                         ), $fname
 204                 );
 205                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 206                         $ids[] = $row->page_id;
 207                 }
 208                 if ( !count( $ids ) ) {
 209                         return;
 210                 }
 211
 212                 /**
 213                  * Do the update
 214                  * We still need the page_touched condition, in case the row has changed since
 215                  * the non-locking select above.
 216                  */
 217                 $this->mDb->update( 'page', array( 'page_touched' => $now ),
 218                         array(
 219                                 'page_id IN (' . $this->mDb->makeList( $ids ) . ')',
 220                                 'page_touched < ' . $this->mDb->addQuotes( $now )
 221                         ), $fname
 222                 );
 223         }
 224
 225         function invalidateCategories( $cats ) {
 226                 $this->invalidatePages( NS_CATEGORY, array_keys( $cats ) );
 227         }
 228
 229         function invalidateImageDescriptions( $images ) {
 230                 $this->invalidatePages( NS_IMAGE, array_keys( $images ) );
 231         }
 232
 233         function dumbTableUpdate( $table, $insertions, $fromField ) {
 234                 $fname = 'LinksUpdate::dumbTableUpdate';
 235                 $this->mDb->delete( $table, array( $fromField => $this->mId ), $fname );
 236                 if ( count( $insertions ) ) {
 237                         # The link array was constructed without FOR UPDATE, so there may be collisions
 238                         # This may cause minor link table inconsistencies, which is better than
 239                         # crippling the site with lock contention.
 240                         $this->mDb->insert( $table, $insertions, $fname, array( 'IGNORE' ) );
 241                 }
 242         }
 243
 244         /**
 245          * Make a WHERE clause from a 2-d NS/dbkey array
 246          *
 247          * @param array $arr 2-d array indexed by namespace and DB key
 248          * @param string $prefix Field name prefix, without the underscore
 249          */
 250         function makeWhereFrom2d( &$arr, $prefix ) {
 251                 $lb = new LinkBatch;
 252                 $lb->setArray( $arr );
 253                 return $lb->constructSet( $prefix, $this->mDb );
 254         }
 255
 256         /**
 257          * Update a table by doing a delete query then an insert query
 258          * @private
 259          */
 260         function incrTableUpdate( $table, $prefix, $deletions, $insertions ) {
 261                 $fname = 'LinksUpdate::incrTableUpdate';
 262                 $where = array( "{$prefix}_from" => $this->mId );
 263                 if ( $table == 'pagelinks' || $table == 'templatelinks' ) {
 264                         $clause = $this->makeWhereFrom2d( $deletions, $prefix );
 265                         if ( $clause ) {
 266                                 $where[] = $clause;
 267                         } else {
 268                                 $where = false;
 269                         }
 270                 } else {
 271                         if ( $table == 'langlinks' ) {
 272                                 $toField = 'll_lang';
 273                         } else {
 274                                 $toField = $prefix . '_to';
 275                         }
 276                         if ( count( $deletions ) ) {
 277                                 $where[] = "$toField IN (" . $this->mDb->makeList( array_keys( $deletions ) ) . ')';
 278                         } else {
 279                                 $where = false;
 280                         }
 281                 }
 282                 if ( $where ) {
 283                         $this->mDb->delete( $table, $where, $fname );
 284                 }
 285                 if ( count( $insertions ) ) {
 286                         $this->mDb->insert( $table, $insertions, $fname, 'IGNORE' );
 287                 }
 288         }
 289
 290
 291         /**
 292          * Get an array of pagelinks insertions for passing to the DB
 293          * Skips the titles specified by the 2-D array $existing
 294          * @private
 295          */
 296         function getLinkInsertions( $existing = array() ) {
 297                 $arr = array();
 298                 foreach( $this->mLinks as $ns => $dbkeys ) {
 299                         # array_diff_key() was introduced in PHP 5.1, there is a compatibility function
 300                         # in GlobalFunctions.php
 301                         $diffs = isset( $existing[$ns] ) ? array_diff_key( $dbkeys, $existing[$ns] ) : $dbkeys;
 302                         foreach ( $diffs as $dbk => $id ) {
 303                                 $arr[] = array(
 304                                         'pl_from'      => $this->mId,
 305                                         'pl_namespace' => $ns,
 306                                         'pl_title'     => $dbk
 307                                 );
 308                         }
 309                 }
 310                 return $arr;
 311         }
 312
 313         /**
 314          * Get an array of template insertions. Like getLinkInsertions()
 315          * @private
 316          */
 317         function getTemplateInsertions( $existing = array() ) {
 318                 $arr = array();
 319                 foreach( $this->mTemplates as $ns => $dbkeys ) {
 320                         $diffs = isset( $existing[$ns] ) ? array_diff_key( $dbkeys, $existing[$ns] ) : $dbkeys;
 321                         foreach ( $diffs as $dbk => $id ) {
 322                                 $arr[] = array(
 323                                         'tl_from'      => $this->mId,
 324                                         'tl_namespace' => $ns,
 325                                         'tl_title'     => $dbk
 326                                 );
 327                         }
 328                 }
 329                 return $arr;
 330         }
 331
 332         /**
 333          * Get an array of image insertions
 334          * Skips the names specified in $existing
 335          * @private
 336          */
 337         function getImageInsertions( $existing = array() ) {
 338                 $arr = array();
 339                 $diffs = array_diff_key( $this->mImages, $existing );
 340                 foreach( $diffs as $iname => $dummy ) {
 341                         $arr[] = array(
 342                                 'il_from' => $this->mId,
 343                                 'il_to'   => $iname
 344                         );
 345                 }
 346                 return $arr;
 347         }
 348
 349         /**
 350          * Get an array of externallinks insertions. Skips the names specified in $existing
 351          * @private
 352          */
 353         function getExternalInsertions( $existing = array() ) {
 354                 $arr = array();
 355                 $diffs = array_diff_key( $this->mExternals, $existing );
 356                 foreach( $diffs as $url => $dummy ) {
 357                         $arr[] = array(
 358                                 'el_from'   => $this->mId,
 359                                 'el_to'     => $url,
 360                                 'el_index'  => wfMakeUrlIndex( $url ),
 361                         );
 362                 }
 363                 return $arr;
 364         }
 365
 366         /**
 367          * Get an array of category insertions
 368          * @param array $existing Array mapping existing category names to sort keys. If both
 369          * match a link in $this, the link will be omitted from the output
 370          * @private
 371          */
 372         function getCategoryInsertions( $existing = array() ) {
 373                 $diffs = array_diff_assoc( $this->mCategories, $existing );
 374                 $arr = array();
 375                 foreach ( $diffs as $name => $sortkey ) {
 376                         $arr[] = array(
 377                                 'cl_from'    => $this->mId,
 378                                 'cl_to'      => $name,
 379                                 'cl_sortkey' => $sortkey,
 380                                 'cl_timestamp' => $this->mDb->timestamp()
 381                         );
 382                 }
 383                 return $arr;
 384         }
 385
 386         /**
 387          * Get an array of interlanguage link insertions
 388          * @param array $existing Array mapping existing language codes to titles
 389          * @private
 390          */
 391         function getInterlangInsertions( $existing = array() ) {
 392             $diffs = array_diff_assoc( $this->mInterlangs, $existing );
 393             $arr = array();
 394             foreach( $diffs as $lang => $title ) {
 395                 $arr[] = array(
 396                     'll_from'  => $this->mId,
 397                     'll_lang'  => $lang,
 398                     'll_title' => $title
 399                 );
 400             }
 401             return $arr;
 402         }
 403
 404         /**
 405          * Given an array of existing links, returns those links which are not in $this
 406          * and thus should be deleted.
 407          * @private
 408          */
 409         function getLinkDeletions( $existing ) {
 410                 $del = array();
 411                 foreach ( $existing as $ns => $dbkeys ) {
 412                         if ( isset( $this->mLinks[$ns] ) ) {
 413                                 $del[$ns] = array_diff_key( $existing[$ns], $this->mLinks[$ns] );
 414                         } else {
 415                                 $del[$ns] = $existing[$ns];
 416                         }
 417                 }
 418                 return $del;
 419         }
 420
 421         /**
 422          * Given an array of existing templates, returns those templates which are not in $this
 423          * and thus should be deleted.
 424          * @private
 425          */
 426         function getTemplateDeletions( $existing ) {
 427                 $del = array();
 428                 foreach ( $existing as $ns => $dbkeys ) {
 429                         if ( isset( $this->mTemplates[$ns] ) ) {
 430                                 $del[$ns] = array_diff_key( $existing[$ns], $this->mTemplates[$ns] );
 431                         } else {
 432                                 $del[$ns] = $existing[$ns];
 433                         }
 434                 }
 435                 return $del;
 436         }
 437
 438         /**
 439          * Given an array of existing images, returns those images which are not in $this
 440          * and thus should be deleted.
 441          * @private
 442          */
 443         function getImageDeletions( $existing ) {
 444                 return array_diff_key( $existing, $this->mImages );
 445         }
 446
 447         /**
 448          * Given an array of existing external links, returns those links which are not
 449          * in $this and thus should be deleted.
 450          * @private
 451          */
 452         function getExternalDeletions( $existing ) {
 453                 return array_diff_key( $existing, $this->mExternals );
 454         }
 455
 456         /**
 457          * Given an array of existing categories, returns those categories which are not in $this
 458          * and thus should be deleted.
 459          * @private
 460          */
 461         function getCategoryDeletions( $existing ) {
 462                 return array_diff_assoc( $existing, $this->mCategories );
 463         }
 464
 465         /**
 466          * Given an array of existing interlanguage links, returns those links which are not
 467          * in $this and thus should be deleted.
 468          * @private
 469          */
 470         function getInterlangDeletions( $existing ) {
 471             return array_diff_assoc( $existing, $this->mInterlangs );
 472         }
 473
 474         /**
 475          * Get an array of existing links, as a 2-D array
 476          * @private
 477          */
 478         function getExistingLinks() {
 479                 $fname = 'LinksUpdate::getExistingLinks';
 480                 $res = $this->mDb->select( 'pagelinks', array( 'pl_namespace', 'pl_title' ),
 481                         array( 'pl_from' => $this->mId ), $fname, $this->mOptions );
 482                 $arr = array();
 483                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 484                         if ( !isset( $arr[$row->pl_namespace] ) ) {
 485                                 $arr[$row->pl_namespace] = array();
 486                         }
 487                         $arr[$row->pl_namespace][$row->pl_title] = 1;
 488                 }
 489                 $this->mDb->freeResult( $res );
 490                 return $arr;
 491         }
 492
 493         /**
 494          * Get an array of existing templates, as a 2-D array
 495          * @private
 496          */
 497         function getExistingTemplates() {
 498                 $fname = 'LinksUpdate::getExistingTemplates';
 499                 $res = $this->mDb->select( 'templatelinks', array( 'tl_namespace', 'tl_title' ),
 500                         array( 'tl_from' => $this->mId ), $fname, $this->mOptions );
 501                 $arr = array();
 502                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 503                         if ( !isset( $arr[$row->tl_namespace] ) ) {
 504                                 $arr[$row->tl_namespace] = array();
 505                         }
 506                         $arr[$row->tl_namespace][$row->tl_title] = 1;
 507                 }
 508                 $this->mDb->freeResult( $res );
 509                 return $arr;
 510         }
 511
 512         /**
 513          * Get an array of existing images, image names in the keys
 514          * @private
 515          */
 516         function getExistingImages() {
 517                 $fname = 'LinksUpdate::getExistingImages';
 518                 $res = $this->mDb->select( 'imagelinks', array( 'il_to' ),
 519                         array( 'il_from' => $this->mId ), $fname, $this->mOptions );
 520                 $arr = array();
 521                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 522                         $arr[$row->il_to] = 1;
 523                 }
 524                 $this->mDb->freeResult( $res );
 525                 return $arr;
 526         }
 527
 528         /**
 529          * Get an array of existing external links, URLs in the keys
 530          * @private
 531          */
 532         function getExistingExternals() {
 533                 $fname = 'LinksUpdate::getExistingExternals';
 534                 $res = $this->mDb->select( 'externallinks', array( 'el_to' ),
 535                         array( 'el_from' => $this->mId ), $fname, $this->mOptions );
 536                 $arr = array();
 537                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 538                         $arr[$row->el_to] = 1;
 539                 }
 540                 $this->mDb->freeResult( $res );
 541                 return $arr;
 542         }
 543
 544         /**
 545          * Get an array of existing categories, with the name in the key and sort key in the value.
 546          * @private
 547          */
 548         function getExistingCategories() {
 549                 $fname = 'LinksUpdate::getExistingCategories';
 550                 $res = $this->mDb->select( 'categorylinks', array( 'cl_to', 'cl_sortkey' ),
 551                         array( 'cl_from' => $this->mId ), $fname, $this->mOptions );
 552                 $arr = array();
 553                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 554                         $arr[$row->cl_to] = $row->cl_sortkey;
 555                 }
 556                 $this->mDb->freeResult( $res );
 557                 return $arr;
 558         }
 559
 560         /**
 561          * Get an array of existing interlanguage links, with the language code in the key and the
 562          * title in the value.
 563          * @private
 564          */
 565         function getExistingInterlangs() {
 566                 $fname = 'LinksUpdate::getExistingInterlangs';
 567                 $res = $this->mDb->select( 'langlinks', array( 'll_lang', 'll_title' ),
 568                         array( 'll_from' => $this->mId ), $fname, $this->mOptions );
 569                 $arr = array();
 570                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 571                         $arr[$row->ll_lang] = $row->ll_title;
 572                 }
 573                 return $arr;
 574         }
 575 }
 576 ?>