includes/LinksUpdate.php

   1 <?php
   2 /**
   3  * See docs/deferred.txt
   4  *
   5  * @todo document (e.g. one-sentence top-level class description).
   6  */
   7 class LinksUpdate {
   8
   9         /**@{{
  10          * @private
  11          */
  12         var     $mId,            //!< Page ID of the article linked from
  13                 $mTitle,         //!< Title object of the article linked from
  14                 $mLinks,         //!< Map of title strings to IDs for the links in the document
  15                 $mImages,        //!< DB keys of the images used, in the array key only
  16                 $mTemplates,     //!< Map of title strings to IDs for the template references, including broken ones
  17                 $mExternals,     //!< URLs of external links, array key only
  18                 $mCategories,    //!< Map of category names to sort keys
  19                 $mInterlangs,    //!< Map of language codes to titles
  20                 $mDb,            //!< Database connection reference
  21                 $mOptions,       //!< SELECT options to be used (array)
  22                 $mRecursive;     //!< Whether to queue jobs for recursive updates
  23         /**@}}*/
  24
  25         /**
  26          * Constructor
  27          * Initialize private variables
  28          * @param $title Integer: FIXME
  29          * @param $parserOutput FIXME
  30          * @param $recursive Boolean: FIXME, default 'true'.
  31          */
  32         function LinksUpdate( $title, $parserOutput, $recursive = true ) {
  33                 global $wgAntiLockFlags;
  34
  35                 if ( $wgAntiLockFlags & ALF_NO_LINK_LOCK ) {
  36                         $this->mOptions = array();
  37                 } else {
  38                         $this->mOptions = array( 'FOR UPDATE' );
  39                 }
  40                 $this->mDb = wfGetDB( DB_MASTER );
  41
  42                 if ( !is_object( $title ) ) {
  43                         throw new MWException( "The calling convention to LinksUpdate::LinksUpdate() has changed. " .
  44                                 "Please see Article::editUpdates() for an invocation example.\n" );
  45                 }
  46                 $this->mTitle = $title;
  47                 $this->mId = $title->getArticleID();
  48
  49                 $this->mLinks = $parserOutput->getLinks();
  50                 $this->mImages = $parserOutput->getImages();
  51                 $this->mTemplates = $parserOutput->getTemplates();
  52                 $this->mExternals = $parserOutput->getExternalLinks();
  53                 $this->mCategories = $parserOutput->getCategories();
  54
  55                 # Convert the format of the interlanguage links
  56                 # I didn't want to change it in the ParserOutput, because that array is passed all
  57                 # the way back to the skin, so either a skin API break would be required, or an
  58                 # inefficient back-conversion.
  59                 $ill = $parserOutput->getLanguageLinks();
  60                 $this->mInterlangs = array();
  61                 foreach ( $ill as $link ) {
  62                         list( $key, $title ) = explode( ':', $link, 2 );
  63                         $this->mInterlangs[$key] = $title;
  64                 }
  65
  66                 $this->mRecursive = $recursive;
  67         }
  68
  69         /**
  70          * Update link tables with outgoing links from an updated article
  71          */
  72         function doUpdate() {
  73                 global $wgUseDumbLinkUpdate;
  74                 if ( $wgUseDumbLinkUpdate ) {
  75                         $this->doDumbUpdate();
  76                 } else {
  77                         $this->doIncrementalUpdate();
  78                 }
  79                 wfRunHooks( 'TitleLinkUpdatesAfterCompletion', array( &$this->mTitle ) );
  80         }
  81
  82         function doIncrementalUpdate() {
  83                 $fname = 'LinksUpdate::doIncrementalUpdate';
  84                 wfProfileIn( $fname );
  85
  86                 # Page links
  87                 $existing = $this->getExistingLinks();
  88                 $this->incrTableUpdate( 'pagelinks', 'pl', $this->getLinkDeletions( $existing ),
  89                         $this->getLinkInsertions( $existing ) );
  90
  91                 # Image links
  92                 $existing = $this->getExistingImages();
  93                 $this->incrTableUpdate( 'imagelinks', 'il', $this->getImageDeletions( $existing ),
  94                         $this->getImageInsertions( $existing ) );
  95
  96                 # Invalidate all image description pages which had links added or removed
  97                 $imageUpdates = array_diff_key( $existing, $this->mImages ) + array_diff_key( $this->mImages, $existing );
  98                 $this->invalidateImageDescriptions( $imageUpdates );
  99
 100                 # External links
 101                 $existing = $this->getExistingExternals();
 102                 $this->incrTableUpdate( 'externallinks', 'el', $this->getExternalDeletions( $existing ),
 103                 $this->getExternalInsertions( $existing ) );
 104
 105                 # Language links
 106                 $existing = $this->getExistingInterlangs();
 107                 $this->incrTableUpdate( 'langlinks', 'll', $this->getInterlangDeletions( $existing ),
 108                         $this->getInterlangInsertions( $existing ) );
 109
 110                 # Template links
 111                 $existing = $this->getExistingTemplates();
 112                 $this->incrTableUpdate( 'templatelinks', 'tl', $this->getTemplateDeletions( $existing ),
 113                         $this->getTemplateInsertions( $existing ) );
 114
 115                 # Category links
 116                 $existing = $this->getExistingCategories();
 117                 $this->incrTableUpdate( 'categorylinks', 'cl', $this->getCategoryDeletions( $existing ),
 118                         $this->getCategoryInsertions( $existing ) );
 119
 120                 # Invalidate all categories which were added, deleted or changed (set symmetric difference)
 121                 $categoryUpdates = array_diff_assoc( $existing, $this->mCategories ) + array_diff_assoc( $this->mCategories, $existing );
 122                 $this->invalidateCategories( $categoryUpdates );
 123
 124                 # Refresh links of all pages including this page
 125                 # This will be in a separate transaction
 126                 if ( $this->mRecursive ) {
 127                         $this->queueRecursiveJobs();
 128                 }
 129
 130                 wfProfileOut( $fname );
 131         }
 132
 133         /**
 134          * Link update which clears the previous entries and inserts new ones
 135          * May be slower or faster depending on level of lock contention and write speed of DB
 136          * Also useful where link table corruption needs to be repaired, e.g. in refreshLinks.php
 137          */
 138         function doDumbUpdate() {
 139                 $fname = 'LinksUpdate::doDumbUpdate';
 140                 wfProfileIn( $fname );
 141
 142                 # Refresh category pages and image description pages
 143                 $existing = $this->getExistingCategories();
 144                 $categoryUpdates = array_diff_assoc( $existing, $this->mCategories ) + array_diff_assoc( $this->mCategories, $existing );
 145                 $existing = $this->getExistingImages();
 146                 $imageUpdates = array_diff_key( $existing, $this->mImages ) + array_diff_key( $this->mImages, $existing );
 147
 148                 $this->dumbTableUpdate( 'pagelinks',     $this->getLinkInsertions(),     'pl_from' );
 149                 $this->dumbTableUpdate( 'imagelinks',    $this->getImageInsertions(),    'il_from' );
 150                 $this->dumbTableUpdate( 'categorylinks', $this->getCategoryInsertions(), 'cl_from' );
 151                 $this->dumbTableUpdate( 'templatelinks', $this->getTemplateInsertions(), 'tl_from' );
 152                 $this->dumbTableUpdate( 'externallinks', $this->getExternalInsertions(), 'el_from' );
 153                 $this->dumbTableUpdate( 'langlinks',     $this->getInterlangInsertions(), 'll_from' );
 154
 155                 # Update the cache of all the category pages and image description pages which were changed
 156                 $this->invalidateCategories( $categoryUpdates );
 157                 $this->invalidateImageDescriptions( $imageUpdates );
 158
 159                 # Refresh links of all pages including this page
 160                 # This will be in a separate transaction
 161                 if ( $this->mRecursive ) {
 162                         $this->queueRecursiveJobs();
 163                 }
 164
 165                 wfProfileOut( $fname );
 166         }
 167
 168         function queueRecursiveJobs() {
 169                 wfProfileIn( __METHOD__ );
 170
 171                 $batchSize = 100;
 172                 $dbr = wfGetDB( DB_SLAVE );
 173                 $res = $dbr->select( array( 'templatelinks', 'page' ),
 174                         array( 'page_namespace', 'page_title' ),
 175                         array(
 176                                 'page_id=tl_from',
 177                                 'tl_namespace' => $this->mTitle->getNamespace(),
 178                                 'tl_title' => $this->mTitle->getDBkey()
 179                         ), __METHOD__
 180                 );
 181
 182                 $done = false;
 183                 while ( !$done ) {
 184                         $jobs = array();
 185                         for ( $i = 0; $i < $batchSize; $i++ ) {
 186                                 $row = $dbr->fetchObject( $res );
 187                                 if ( !$row ) {
 188                                         $done = true;
 189                                         break;
 190                                 }
 191                                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 192                                 $jobs[] = Job::factory( 'refreshLinks', $title );
 193                         }
 194                         Job::batchInsert( $jobs );
 195                 }
 196                 $dbr->freeResult( $res );
 197                 wfProfileOut( __METHOD__ );
 198         }
 199
 200         /**
 201          * Invalidate the cache of a list of pages from a single namespace
 202          *
 203          * @param integer $namespace
 204          * @param array $dbkeys
 205          */
 206         function invalidatePages( $namespace, $dbkeys ) {
 207                 $fname = 'LinksUpdate::invalidatePages';
 208
 209                 if ( !count( $dbkeys ) ) {
 210                         return;
 211                 }
 212
 213                 /**
 214                  * Determine which pages need to be updated
 215                  * This is necessary to prevent the job queue from smashing the DB with
 216                  * large numbers of concurrent invalidations of the same page
 217                  */
 218                 $now = $this->mDb->timestamp();
 219                 $ids = array();
 220                 $res = $this->mDb->select( 'page', array( 'page_id' ),
 221                         array(
 222                                 'page_namespace' => $namespace,
 223                                 'page_title IN (' . $this->mDb->makeList( $dbkeys ) . ')',
 224                                 'page_touched < ' . $this->mDb->addQuotes( $now )
 225                         ), $fname
 226                 );
 227                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 228                         $ids[] = $row->page_id;
 229                 }
 230                 if ( !count( $ids ) ) {
 231                         return;
 232                 }
 233
 234                 /**
 235                  * Do the update
 236                  * We still need the page_touched condition, in case the row has changed since
 237                  * the non-locking select above.
 238                  */
 239                 $this->mDb->update( 'page', array( 'page_touched' => $now ),
 240                         array(
 241                                 'page_id IN (' . $this->mDb->makeList( $ids ) . ')',
 242                                 'page_touched < ' . $this->mDb->addQuotes( $now )
 243                         ), $fname
 244                 );
 245         }
 246
 247         function invalidateCategories( $cats ) {
 248                 $this->invalidatePages( NS_CATEGORY, array_keys( $cats ) );
 249         }
 250
 251         function invalidateImageDescriptions( $images ) {
 252                 $this->invalidatePages( NS_IMAGE, array_keys( $images ) );
 253         }
 254
 255         function dumbTableUpdate( $table, $insertions, $fromField ) {
 256                 $fname = 'LinksUpdate::dumbTableUpdate';
 257                 $this->mDb->delete( $table, array( $fromField => $this->mId ), $fname );
 258                 if ( count( $insertions ) ) {
 259                         # The link array was constructed without FOR UPDATE, so there may be collisions
 260                         # This may cause minor link table inconsistencies, which is better than
 261                         # crippling the site with lock contention.
 262                         $this->mDb->insert( $table, $insertions, $fname, array( 'IGNORE' ) );
 263                 }
 264         }
 265
 266         /**
 267          * Make a WHERE clause from a 2-d NS/dbkey array
 268          *
 269          * @param array $arr 2-d array indexed by namespace and DB key
 270          * @param string $prefix Field name prefix, without the underscore
 271          */
 272         function makeWhereFrom2d( &$arr, $prefix ) {
 273                 $lb = new LinkBatch;
 274                 $lb->setArray( $arr );
 275                 return $lb->constructSet( $prefix, $this->mDb );
 276         }
 277
 278         /**
 279          * Update a table by doing a delete query then an insert query
 280          * @private
 281          */
 282         function incrTableUpdate( $table, $prefix, $deletions, $insertions ) {
 283                 $fname = 'LinksUpdate::incrTableUpdate';
 284                 $where = array( "{$prefix}_from" => $this->mId );
 285                 if ( $table == 'pagelinks' || $table == 'templatelinks' ) {
 286                         $clause = $this->makeWhereFrom2d( $deletions, $prefix );
 287                         if ( $clause ) {
 288                                 $where[] = $clause;
 289                         } else {
 290                                 $where = false;
 291                         }
 292                 } else {
 293                         if ( $table == 'langlinks' ) {
 294                                 $toField = 'll_lang';
 295                         } else {
 296                                 $toField = $prefix . '_to';
 297                         }
 298                         if ( count( $deletions ) ) {
 299                                 $where[] = "$toField IN (" . $this->mDb->makeList( array_keys( $deletions ) ) . ')';
 300                         } else {
 301                                 $where = false;
 302                         }
 303                 }
 304                 if ( $where ) {
 305                         $this->mDb->delete( $table, $where, $fname );
 306                 }
 307                 if ( count( $insertions ) ) {
 308                         $this->mDb->insert( $table, $insertions, $fname, 'IGNORE' );
 309                 }
 310         }
 311
 312
 313         /**
 314          * Get an array of pagelinks insertions for passing to the DB
 315          * Skips the titles specified by the 2-D array $existing
 316          * @private
 317          */
 318         function getLinkInsertions( $existing = array() ) {
 319                 $arr = array();
 320                 foreach( $this->mLinks as $ns => $dbkeys ) {
 321                         # array_diff_key() was introduced in PHP 5.1, there is a compatibility function
 322                         # in GlobalFunctions.php
 323                         $diffs = isset( $existing[$ns] ) ? array_diff_key( $dbkeys, $existing[$ns] ) : $dbkeys;
 324                         foreach ( $diffs as $dbk => $id ) {
 325                                 $arr[] = array(
 326                                         'pl_from'      => $this->mId,
 327                                         'pl_namespace' => $ns,
 328                                         'pl_title'     => $dbk
 329                                 );
 330                         }
 331                 }
 332                 return $arr;
 333         }
 334
 335         /**
 336          * Get an array of template insertions. Like getLinkInsertions()
 337          * @private
 338          */
 339         function getTemplateInsertions( $existing = array() ) {
 340                 $arr = array();
 341                 foreach( $this->mTemplates as $ns => $dbkeys ) {
 342                         $diffs = isset( $existing[$ns] ) ? array_diff_key( $dbkeys, $existing[$ns] ) : $dbkeys;
 343                         foreach ( $diffs as $dbk => $id ) {
 344                                 $arr[] = array(
 345                                         'tl_from'      => $this->mId,
 346                                         'tl_namespace' => $ns,
 347                                         'tl_title'     => $dbk
 348                                 );
 349                         }
 350                 }
 351                 return $arr;
 352         }
 353
 354         /**
 355          * Get an array of image insertions
 356          * Skips the names specified in $existing
 357          * @private
 358          */
 359         function getImageInsertions( $existing = array() ) {
 360                 $arr = array();
 361                 $diffs = array_diff_key( $this->mImages, $existing );
 362                 foreach( $diffs as $iname => $dummy ) {
 363                         $arr[] = array(
 364                                 'il_from' => $this->mId,
 365                                 'il_to'   => $iname
 366                         );
 367                 }
 368                 return $arr;
 369         }
 370
 371         /**
 372          * Get an array of externallinks insertions. Skips the names specified in $existing
 373          * @private
 374          */
 375         function getExternalInsertions( $existing = array() ) {
 376                 $arr = array();
 377                 $diffs = array_diff_key( $this->mExternals, $existing );
 378                 foreach( $diffs as $url => $dummy ) {
 379                         $arr[] = array(
 380                                 'el_from'   => $this->mId,
 381                                 'el_to'     => $url,
 382                                 'el_index'  => wfMakeUrlIndex( $url ),
 383                         );
 384                 }
 385                 return $arr;
 386         }
 387
 388         /**
 389          * Get an array of category insertions
 390          * @param array $existing Array mapping existing category names to sort keys. If both
 391          * match a link in $this, the link will be omitted from the output
 392          * @private
 393          */
 394         function getCategoryInsertions( $existing = array() ) {
 395                 $diffs = array_diff_assoc( $this->mCategories, $existing );
 396                 $arr = array();
 397                 foreach ( $diffs as $name => $sortkey ) {
 398                         $arr[] = array(
 399                                 'cl_from'    => $this->mId,
 400                                 'cl_to'      => $name,
 401                                 'cl_sortkey' => $sortkey,
 402                                 'cl_timestamp' => $this->mDb->timestamp()
 403                         );
 404                 }
 405                 return $arr;
 406         }
 407
 408         /**
 409          * Get an array of interlanguage link insertions
 410          * @param array $existing Array mapping existing language codes to titles
 411          * @private
 412          */
 413         function getInterlangInsertions( $existing = array() ) {
 414             $diffs = array_diff_assoc( $this->mInterlangs, $existing );
 415             $arr = array();
 416             foreach( $diffs as $lang => $title ) {
 417                 $arr[] = array(
 418                     'll_from'  => $this->mId,
 419                     'll_lang'  => $lang,
 420                     'll_title' => $title
 421                 );
 422             }
 423             return $arr;
 424         }
 425
 426         /**
 427          * Given an array of existing links, returns those links which are not in $this
 428          * and thus should be deleted.
 429          * @private
 430          */
 431         function getLinkDeletions( $existing ) {
 432                 $del = array();
 433                 foreach ( $existing as $ns => $dbkeys ) {
 434                         if ( isset( $this->mLinks[$ns] ) ) {
 435                                 $del[$ns] = array_diff_key( $existing[$ns], $this->mLinks[$ns] );
 436                         } else {
 437                                 $del[$ns] = $existing[$ns];
 438                         }
 439                 }
 440                 return $del;
 441         }
 442
 443         /**
 444          * Given an array of existing templates, returns those templates which are not in $this
 445          * and thus should be deleted.
 446          * @private
 447          */
 448         function getTemplateDeletions( $existing ) {
 449                 $del = array();
 450                 foreach ( $existing as $ns => $dbkeys ) {
 451                         if ( isset( $this->mTemplates[$ns] ) ) {
 452                                 $del[$ns] = array_diff_key( $existing[$ns], $this->mTemplates[$ns] );
 453                         } else {
 454                                 $del[$ns] = $existing[$ns];
 455                         }
 456                 }
 457                 return $del;
 458         }
 459
 460         /**
 461          * Given an array of existing images, returns those images which are not in $this
 462          * and thus should be deleted.
 463          * @private
 464          */
 465         function getImageDeletions( $existing ) {
 466                 return array_diff_key( $existing, $this->mImages );
 467         }
 468
 469         /**
 470          * Given an array of existing external links, returns those links which are not
 471          * in $this and thus should be deleted.
 472          * @private
 473          */
 474         function getExternalDeletions( $existing ) {
 475                 return array_diff_key( $existing, $this->mExternals );
 476         }
 477
 478         /**
 479          * Given an array of existing categories, returns those categories which are not in $this
 480          * and thus should be deleted.
 481          * @private
 482          */
 483         function getCategoryDeletions( $existing ) {
 484                 return array_diff_assoc( $existing, $this->mCategories );
 485         }
 486
 487         /**
 488          * Given an array of existing interlanguage links, returns those links which are not
 489          * in $this and thus should be deleted.
 490          * @private
 491          */
 492         function getInterlangDeletions( $existing ) {
 493             return array_diff_assoc( $existing, $this->mInterlangs );
 494         }
 495
 496         /**
 497          * Get an array of existing links, as a 2-D array
 498          * @private
 499          */
 500         function getExistingLinks() {
 501                 $fname = 'LinksUpdate::getExistingLinks';
 502                 $res = $this->mDb->select( 'pagelinks', array( 'pl_namespace', 'pl_title' ),
 503                         array( 'pl_from' => $this->mId ), $fname, $this->mOptions );
 504                 $arr = array();
 505                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 506                         if ( !isset( $arr[$row->pl_namespace] ) ) {
 507                                 $arr[$row->pl_namespace] = array();
 508                         }
 509                         $arr[$row->pl_namespace][$row->pl_title] = 1;
 510                 }
 511                 $this->mDb->freeResult( $res );
 512                 return $arr;
 513         }
 514
 515         /**
 516          * Get an array of existing templates, as a 2-D array
 517          * @private
 518          */
 519         function getExistingTemplates() {
 520                 $fname = 'LinksUpdate::getExistingTemplates';
 521                 $res = $this->mDb->select( 'templatelinks', array( 'tl_namespace', 'tl_title' ),
 522                         array( 'tl_from' => $this->mId ), $fname, $this->mOptions );
 523                 $arr = array();
 524                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 525                         if ( !isset( $arr[$row->tl_namespace] ) ) {
 526                                 $arr[$row->tl_namespace] = array();
 527                         }
 528                         $arr[$row->tl_namespace][$row->tl_title] = 1;
 529                 }
 530                 $this->mDb->freeResult( $res );
 531                 return $arr;
 532         }
 533
 534         /**
 535          * Get an array of existing images, image names in the keys
 536          * @private
 537          */
 538         function getExistingImages() {
 539                 $fname = 'LinksUpdate::getExistingImages';
 540                 $res = $this->mDb->select( 'imagelinks', array( 'il_to' ),
 541                         array( 'il_from' => $this->mId ), $fname, $this->mOptions );
 542                 $arr = array();
 543                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 544                         $arr[$row->il_to] = 1;
 545                 }
 546                 $this->mDb->freeResult( $res );
 547                 return $arr;
 548         }
 549
 550         /**
 551          * Get an array of existing external links, URLs in the keys
 552          * @private
 553          */
 554         function getExistingExternals() {
 555                 $fname = 'LinksUpdate::getExistingExternals';
 556                 $res = $this->mDb->select( 'externallinks', array( 'el_to' ),
 557                         array( 'el_from' => $this->mId ), $fname, $this->mOptions );
 558                 $arr = array();
 559                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 560                         $arr[$row->el_to] = 1;
 561                 }
 562                 $this->mDb->freeResult( $res );
 563                 return $arr;
 564         }
 565
 566         /**
 567          * Get an array of existing categories, with the name in the key and sort key in the value.
 568          * @private
 569          */
 570         function getExistingCategories() {
 571                 $fname = 'LinksUpdate::getExistingCategories';
 572                 $res = $this->mDb->select( 'categorylinks', array( 'cl_to', 'cl_sortkey' ),
 573                         array( 'cl_from' => $this->mId ), $fname, $this->mOptions );
 574                 $arr = array();
 575                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 576                         $arr[$row->cl_to] = $row->cl_sortkey;
 577                 }
 578                 $this->mDb->freeResult( $res );
 579                 return $arr;
 580         }
 581
 582         /**
 583          * Get an array of existing interlanguage links, with the language code in the key and the
 584          * title in the value.
 585          * @private
 586          */
 587         function getExistingInterlangs() {
 588                 $fname = 'LinksUpdate::getExistingInterlangs';
 589                 $res = $this->mDb->select( 'langlinks', array( 'll_lang', 'll_title' ),
 590                         array( 'll_from' => $this->mId ), $fname, $this->mOptions );
 591                 $arr = array();
 592                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 593                         $arr[$row->ll_lang] = $row->ll_title;
 594                 }
 595                 return $arr;
 596         }
 597 }
 598 ?>