includes/LinksUpdate.php

   1 <?php
   2 /**
   3  * See deferred.txt
   4  */
   5
   6 /**
   7  * @todo document
   8  */
   9 class LinksUpdate {
  10
  11         /**@{{
  12          * @private
  13          */
  14         var $mId,            //!< Page ID of the article linked from
  15                 $mTitle,         //!< Title object of the article linked from
  16                 $mLinks,         //!< Map of title strings to IDs for the links in the document
  17                 $mImages,        //!< DB keys of the images used, in the array key only
  18                 $mTemplates,     //!< Map of title strings to IDs for the template references, including broken ones
  19                 $mExternals,     //!< URLs of external links, array key only
  20                 $mCategories,    //!< Map of category names to sort keys
  21                 $mInterlangs,    //!< Map of language codes to titles
  22                 $mDb,            //!< Database connection reference
  23                 $mOptions,       //!< SELECT options to be used (array)
  24                 $mRecursive;     //!< Whether to queue jobs for recursive updates
  25         /**@}}*/
  26
  27         /**
  28          * Constructor
  29          * Initialize private variables
  30          * @param $title Integer: FIXME
  31          * @param $parserOutput FIXME
  32          * @param $recursive Boolean: FIXME, default 'true'.
  33          */
  34         function LinksUpdate( $title, $parserOutput, $recursive = true ) {
  35                 global $wgAntiLockFlags;
  36
  37                 if ( $wgAntiLockFlags & ALF_NO_LINK_LOCK ) {
  38                         $this->mOptions = array();
  39                 } else {
  40                         $this->mOptions = array( 'FOR UPDATE' );
  41                 }
  42                 $this->mDb =& wfGetDB( DB_MASTER );
  43
  44                 if ( !is_object( $title ) ) {
  45                         throw new MWException( "The calling convention to LinksUpdate::LinksUpdate() has changed. " .
  46                                 "Please see Article::editUpdates() for an invocation example.\n" );
  47                 }
  48                 $this->mTitle = $title;
  49                 $this->mId = $title->getArticleID();
  50
  51                 $this->mLinks = $parserOutput->getLinks();
  52                 $this->mImages = $parserOutput->getImages();
  53                 $this->mTemplates = $parserOutput->getTemplates();
  54                 $this->mExternals = $parserOutput->getExternalLinks();
  55                 $this->mCategories = $parserOutput->getCategories();
  56
  57                 # Convert the format of the interlanguage links
  58                 # I didn't want to change it in the ParserOutput, because that array is passed all
  59                 # the way back to the skin, so either a skin API break would be required, or an
  60                 # inefficient back-conversion.
  61                 $ill = $parserOutput->getLanguageLinks();
  62                 $this->mInterlangs = array();
  63                 foreach ( $ill as $link ) {
  64                         list( $key, $title ) = explode( ':', $link, 2 );
  65                         $this->mInterlangs[$key] = $title;
  66                 }
  67
  68                 $this->mRecursive = $recursive;
  69         }
  70
  71         /**
  72          * Update link tables with outgoing links from an updated article
  73          */
  74         function doUpdate() {
  75                 global $wgUseDumbLinkUpdate;
  76                 if ( $wgUseDumbLinkUpdate ) {
  77                         $this->doDumbUpdate();
  78                 } else {
  79                         $this->doIncrementalUpdate();
  80                 }
  81         }
  82
  83         function doIncrementalUpdate() {
  84                 $fname = 'LinksUpdate::doIncrementalUpdate';
  85                 wfProfileIn( $fname );
  86
  87                 # Page links
  88                 $existing = $this->getExistingLinks();
  89                 $this->incrTableUpdate( 'pagelinks', 'pl', $this->getLinkDeletions( $existing ),
  90                         $this->getLinkInsertions( $existing ) );
  91
  92                 # Image links
  93                 $existing = $this->getExistingImages();
  94                 $this->incrTableUpdate( 'imagelinks', 'il', $this->getImageDeletions( $existing ),
  95                         $this->getImageInsertions( $existing ) );
  96
  97                 # Invalidate all image description pages which had links added or removed
  98                 $imageUpdates = array_diff_key( $existing, $this->mImages ) + array_diff_key( $this->mImages, $existing );
  99                 $this->invalidateImageDescriptions( $imageUpdates );
 100
 101                 # External links
 102                 $existing = $this->getExistingExternals();
 103                 $this->incrTableUpdate( 'externallinks', 'el', $this->getExternalDeletions( $existing ),
 104                 $this->getExternalInsertions( $existing ) );
 105
 106                 # Language links
 107                 $existing = $this->getExistingInterlangs();
 108                 $this->incrTableUpdate( 'langlinks', 'll', $this->getInterlangDeletions( $existing ),
 109                         $this->getInterlangInsertions( $existing ) );
 110
 111                 # Template links
 112                 $existing = $this->getExistingTemplates();
 113                 $this->incrTableUpdate( 'templatelinks', 'tl', $this->getTemplateDeletions( $existing ),
 114                         $this->getTemplateInsertions( $existing ) );
 115
 116                 # Category links
 117                 $existing = $this->getExistingCategories();
 118                 $this->incrTableUpdate( 'categorylinks', 'cl', $this->getCategoryDeletions( $existing ),
 119                         $this->getCategoryInsertions( $existing ) );
 120
 121                 # Invalidate all categories which were added, deleted or changed (set symmetric difference)
 122                 $categoryUpdates = array_diff_assoc( $existing, $this->mCategories ) + array_diff_assoc( $this->mCategories, $existing );
 123                 $this->invalidateCategories( $categoryUpdates );
 124
 125                 # Refresh links of all pages including this page
 126                 # This will be in a separate transaction
 127                 if ( $this->mRecursive ) {
 128                         $this->queueRecursiveJobs();
 129                 }
 130
 131                 wfProfileOut( $fname );
 132         }
 133
 134         /**
 135          * Link update which clears the previous entries and inserts new ones
 136          * May be slower or faster depending on level of lock contention and write speed of DB
 137          * Also useful where link table corruption needs to be repaired, e.g. in refreshLinks.php
 138          */
 139         function doDumbUpdate() {
 140                 $fname = 'LinksUpdate::doDumbUpdate';
 141                 wfProfileIn( $fname );
 142
 143                 # Refresh category pages and image description pages
 144                 $existing = $this->getExistingCategories();
 145                 $categoryUpdates = array_diff_assoc( $existing, $this->mCategories ) + array_diff_assoc( $this->mCategories, $existing );
 146                 $existing = $this->getExistingImages();
 147                 $imageUpdates = array_diff_key( $existing, $this->mImages ) + array_diff_key( $this->mImages, $existing );
 148
 149                 $this->dumbTableUpdate( 'pagelinks',     $this->getLinkInsertions(),     'pl_from' );
 150                 $this->dumbTableUpdate( 'imagelinks',    $this->getImageInsertions(),    'il_from' );
 151                 $this->dumbTableUpdate( 'categorylinks', $this->getCategoryInsertions(), 'cl_from' );
 152                 $this->dumbTableUpdate( 'templatelinks', $this->getTemplateInsertions(), 'tl_from' );
 153                 $this->dumbTableUpdate( 'externallinks', $this->getExternalInsertions(), 'el_from' );
 154                 $this->dumbTableUpdate( 'langlinks',     $this->getInterlangInsertions(), 'll_from' );
 155
 156                 # Update the cache of all the category pages and image description pages which were changed
 157                 $this->invalidateCategories( $categoryUpdates );
 158                 $this->invalidateImageDescriptions( $imageUpdates );
 159
 160                 # Refresh links of all pages including this page
 161                 # This will be in a separate transaction
 162                 if ( $this->mRecursive ) {
 163                         $this->queueRecursiveJobs();
 164                 }
 165
 166                 wfProfileOut( $fname );
 167         }
 168
 169         function queueRecursiveJobs() {
 170                 wfProfileIn( __METHOD__ );
 171
 172                 $batchSize = 100;
 173                 $dbr =& wfGetDB( DB_SLAVE );
 174                 $res = $dbr->select( array( 'templatelinks', 'page' ),
 175                         array( 'page_namespace', 'page_title' ),
 176                         array(
 177                                 'page_id=tl_from',
 178                                 'tl_namespace' => $this->mTitle->getNamespace(),
 179                                 'tl_title' => $this->mTitle->getDBkey()
 180                         ), __METHOD__
 181                 );
 182
 183                 $done = false;
 184                 while ( !$done ) {
 185                         $jobs = array();
 186                         for ( $i = 0; $i < $batchSize; $i++ ) {
 187                                 $row = $dbr->fetchObject( $res );
 188                                 if ( !$row ) {
 189                                         $done = true;
 190                                         break;
 191                                 }
 192                                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 193                                 $jobs[] = Job::factory( 'refreshLinks', $title );
 194                         }
 195                         Job::batchInsert( $jobs );
 196                 }
 197                 $dbr->freeResult( $res );
 198                 wfProfileOut( __METHOD__ );
 199         }
 200
 201         /**
 202          * Invalidate the cache of a list of pages from a single namespace
 203          *
 204          * @param integer $namespace
 205          * @param array $dbkeys
 206          */
 207         function invalidatePages( $namespace, $dbkeys ) {
 208                 $fname = 'LinksUpdate::invalidatePages';
 209
 210                 if ( !count( $dbkeys ) ) {
 211                         return;
 212                 }
 213
 214                 /**
 215                  * Determine which pages need to be updated
 216                  * This is necessary to prevent the job queue from smashing the DB with
 217                  * large numbers of concurrent invalidations of the same page
 218                  */
 219                 $now = $this->mDb->timestamp();
 220                 $ids = array();
 221                 $res = $this->mDb->select( 'page', array( 'page_id' ),
 222                         array(
 223                                 'page_namespace' => $namespace,
 224                                 'page_title IN (' . $this->mDb->makeList( $dbkeys ) . ')',
 225                                 'page_touched < ' . $this->mDb->addQuotes( $now )
 226                         ), $fname
 227                 );
 228                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 229                         $ids[] = $row->page_id;
 230                 }
 231                 if ( !count( $ids ) ) {
 232                         return;
 233                 }
 234
 235                 /**
 236                  * Do the update
 237                  * We still need the page_touched condition, in case the row has changed since
 238                  * the non-locking select above.
 239                  */
 240                 $this->mDb->update( 'page', array( 'page_touched' => $now ),
 241                         array(
 242                                 'page_id IN (' . $this->mDb->makeList( $ids ) . ')',
 243                                 'page_touched < ' . $this->mDb->addQuotes( $now )
 244                         ), $fname
 245                 );
 246         }
 247
 248         function invalidateCategories( $cats ) {
 249                 $this->invalidatePages( NS_CATEGORY, array_keys( $cats ) );
 250         }
 251
 252         function invalidateImageDescriptions( $images ) {
 253                 $this->invalidatePages( NS_IMAGE, array_keys( $images ) );
 254         }
 255
 256         function dumbTableUpdate( $table, $insertions, $fromField ) {
 257                 $fname = 'LinksUpdate::dumbTableUpdate';
 258                 $this->mDb->delete( $table, array( $fromField => $this->mId ), $fname );
 259                 if ( count( $insertions ) ) {
 260                         # The link array was constructed without FOR UPDATE, so there may be collisions
 261                         # This may cause minor link table inconsistencies, which is better than
 262                         # crippling the site with lock contention.
 263                         $this->mDb->insert( $table, $insertions, $fname, array( 'IGNORE' ) );
 264                 }
 265         }
 266
 267         /**
 268          * Make a WHERE clause from a 2-d NS/dbkey array
 269          *
 270          * @param array $arr 2-d array indexed by namespace and DB key
 271          * @param string $prefix Field name prefix, without the underscore
 272          */
 273         function makeWhereFrom2d( &$arr, $prefix ) {
 274                 $lb = new LinkBatch;
 275                 $lb->setArray( $arr );
 276                 return $lb->constructSet( $prefix, $this->mDb );
 277         }
 278
 279         /**
 280          * Update a table by doing a delete query then an insert query
 281          * @private
 282          */
 283         function incrTableUpdate( $table, $prefix, $deletions, $insertions ) {
 284                 $fname = 'LinksUpdate::incrTableUpdate';
 285                 $where = array( "{$prefix}_from" => $this->mId );
 286                 if ( $table == 'pagelinks' || $table == 'templatelinks' ) {
 287                         $clause = $this->makeWhereFrom2d( $deletions, $prefix );
 288                         if ( $clause ) {
 289                                 $where[] = $clause;
 290                         } else {
 291                                 $where = false;
 292                         }
 293                 } else {
 294                         if ( $table == 'langlinks' ) {
 295                                 $toField = 'll_lang';
 296                         } else {
 297                                 $toField = $prefix . '_to';
 298                         }
 299                         if ( count( $deletions ) ) {
 300                                 $where[] = "$toField IN (" . $this->mDb->makeList( array_keys( $deletions ) ) . ')';
 301                         } else {
 302                                 $where = false;
 303                         }
 304                 }
 305                 if ( $where ) {
 306                         $this->mDb->delete( $table, $where, $fname );
 307                 }
 308                 if ( count( $insertions ) ) {
 309                         $this->mDb->insert( $table, $insertions, $fname, 'IGNORE' );
 310                 }
 311         }
 312
 313
 314         /**
 315          * Get an array of pagelinks insertions for passing to the DB
 316          * Skips the titles specified by the 2-D array $existing
 317          * @private
 318          */
 319         function getLinkInsertions( $existing = array() ) {
 320                 $arr = array();
 321                 foreach( $this->mLinks as $ns => $dbkeys ) {
 322                         # array_diff_key() was introduced in PHP 5.1, there is a compatibility function
 323                         # in GlobalFunctions.php
 324                         $diffs = isset( $existing[$ns] ) ? array_diff_key( $dbkeys, $existing[$ns] ) : $dbkeys;
 325                         foreach ( $diffs as $dbk => $id ) {
 326                                 $arr[] = array(
 327                                         'pl_from'      => $this->mId,
 328                                         'pl_namespace' => $ns,
 329                                         'pl_title'     => $dbk
 330                                 );
 331                         }
 332                 }
 333                 return $arr;
 334         }
 335
 336         /**
 337          * Get an array of template insertions. Like getLinkInsertions()
 338          * @private
 339          */
 340         function getTemplateInsertions( $existing = array() ) {
 341                 $arr = array();
 342                 foreach( $this->mTemplates as $ns => $dbkeys ) {
 343                         $diffs = isset( $existing[$ns] ) ? array_diff_key( $dbkeys, $existing[$ns] ) : $dbkeys;
 344                         foreach ( $diffs as $dbk => $id ) {
 345                                 $arr[] = array(
 346                                         'tl_from'      => $this->mId,
 347                                         'tl_namespace' => $ns,
 348                                         'tl_title'     => $dbk
 349                                 );
 350                         }
 351                 }
 352                 return $arr;
 353         }
 354
 355         /**
 356          * Get an array of image insertions
 357          * Skips the names specified in $existing
 358          * @private
 359          */
 360         function getImageInsertions( $existing = array() ) {
 361                 $arr = array();
 362                 $diffs = array_diff_key( $this->mImages, $existing );
 363                 foreach( $diffs as $iname => $dummy ) {
 364                         $arr[] = array(
 365                                 'il_from' => $this->mId,
 366                                 'il_to'   => $iname
 367                         );
 368                 }
 369                 return $arr;
 370         }
 371
 372         /**
 373          * Get an array of externallinks insertions. Skips the names specified in $existing
 374          * @private
 375          */
 376         function getExternalInsertions( $existing = array() ) {
 377                 $arr = array();
 378                 $diffs = array_diff_key( $this->mExternals, $existing );
 379                 foreach( $diffs as $url => $dummy ) {
 380                         $arr[] = array(
 381                                 'el_from'   => $this->mId,
 382                                 'el_to'     => $url,
 383                                 'el_index'  => wfMakeUrlIndex( $url ),
 384                         );
 385                 }
 386                 return $arr;
 387         }
 388
 389         /**
 390          * Get an array of category insertions
 391          * @param array $existing Array mapping existing category names to sort keys. If both
 392          * match a link in $this, the link will be omitted from the output
 393          * @private
 394          */
 395         function getCategoryInsertions( $existing = array() ) {
 396                 $diffs = array_diff_assoc( $this->mCategories, $existing );
 397                 $arr = array();
 398                 foreach ( $diffs as $name => $sortkey ) {
 399                         $arr[] = array(
 400                                 'cl_from'    => $this->mId,
 401                                 'cl_to'      => $name,
 402                                 'cl_sortkey' => $sortkey,
 403                                 'cl_timestamp' => $this->mDb->timestamp()
 404                         );
 405                 }
 406                 return $arr;
 407         }
 408
 409         /**
 410          * Get an array of interlanguage link insertions
 411          * @param array $existing Array mapping existing language codes to titles
 412          * @private
 413          */
 414         function getInterlangInsertions( $existing = array() ) {
 415             $diffs = array_diff_assoc( $this->mInterlangs, $existing );
 416             $arr = array();
 417             foreach( $diffs as $lang => $title ) {
 418                 $arr[] = array(
 419                     'll_from'  => $this->mId,
 420                     'll_lang'  => $lang,
 421                     'll_title' => $title
 422                 );
 423             }
 424             return $arr;
 425         }
 426
 427         /**
 428          * Given an array of existing links, returns those links which are not in $this
 429          * and thus should be deleted.
 430          * @private
 431          */
 432         function getLinkDeletions( $existing ) {
 433                 $del = array();
 434                 foreach ( $existing as $ns => $dbkeys ) {
 435                         if ( isset( $this->mLinks[$ns] ) ) {
 436                                 $del[$ns] = array_diff_key( $existing[$ns], $this->mLinks[$ns] );
 437                         } else {
 438                                 $del[$ns] = $existing[$ns];
 439                         }
 440                 }
 441                 return $del;
 442         }
 443
 444         /**
 445          * Given an array of existing templates, returns those templates which are not in $this
 446          * and thus should be deleted.
 447          * @private
 448          */
 449         function getTemplateDeletions( $existing ) {
 450                 $del = array();
 451                 foreach ( $existing as $ns => $dbkeys ) {
 452                         if ( isset( $this->mTemplates[$ns] ) ) {
 453                                 $del[$ns] = array_diff_key( $existing[$ns], $this->mTemplates[$ns] );
 454                         } else {
 455                                 $del[$ns] = $existing[$ns];
 456                         }
 457                 }
 458                 return $del;
 459         }
 460
 461         /**
 462          * Given an array of existing images, returns those images which are not in $this
 463          * and thus should be deleted.
 464          * @private
 465          */
 466         function getImageDeletions( $existing ) {
 467                 return array_diff_key( $existing, $this->mImages );
 468         }
 469
 470         /**
 471          * Given an array of existing external links, returns those links which are not
 472          * in $this and thus should be deleted.
 473          * @private
 474          */
 475         function getExternalDeletions( $existing ) {
 476                 return array_diff_key( $existing, $this->mExternals );
 477         }
 478
 479         /**
 480          * Given an array of existing categories, returns those categories which are not in $this
 481          * and thus should be deleted.
 482          * @private
 483          */
 484         function getCategoryDeletions( $existing ) {
 485                 return array_diff_assoc( $existing, $this->mCategories );
 486         }
 487
 488         /**
 489          * Given an array of existing interlanguage links, returns those links which are not
 490          * in $this and thus should be deleted.
 491          * @private
 492          */
 493         function getInterlangDeletions( $existing ) {
 494             return array_diff_assoc( $existing, $this->mInterlangs );
 495         }
 496
 497         /**
 498          * Get an array of existing links, as a 2-D array
 499          * @private
 500          */
 501         function getExistingLinks() {
 502                 $fname = 'LinksUpdate::getExistingLinks';
 503                 $res = $this->mDb->select( 'pagelinks', array( 'pl_namespace', 'pl_title' ),
 504                         array( 'pl_from' => $this->mId ), $fname, $this->mOptions );
 505                 $arr = array();
 506                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 507                         if ( !isset( $arr[$row->pl_namespace] ) ) {
 508                                 $arr[$row->pl_namespace] = array();
 509                         }
 510                         $arr[$row->pl_namespace][$row->pl_title] = 1;
 511                 }
 512                 $this->mDb->freeResult( $res );
 513                 return $arr;
 514         }
 515
 516         /**
 517          * Get an array of existing templates, as a 2-D array
 518          * @private
 519          */
 520         function getExistingTemplates() {
 521                 $fname = 'LinksUpdate::getExistingTemplates';
 522                 $res = $this->mDb->select( 'templatelinks', array( 'tl_namespace', 'tl_title' ),
 523                         array( 'tl_from' => $this->mId ), $fname, $this->mOptions );
 524                 $arr = array();
 525                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 526                         if ( !isset( $arr[$row->tl_namespace] ) ) {
 527                                 $arr[$row->tl_namespace] = array();
 528                         }
 529                         $arr[$row->tl_namespace][$row->tl_title] = 1;
 530                 }
 531                 $this->mDb->freeResult( $res );
 532                 return $arr;
 533         }
 534
 535         /**
 536          * Get an array of existing images, image names in the keys
 537          * @private
 538          */
 539         function getExistingImages() {
 540                 $fname = 'LinksUpdate::getExistingImages';
 541                 $res = $this->mDb->select( 'imagelinks', array( 'il_to' ),
 542                         array( 'il_from' => $this->mId ), $fname, $this->mOptions );
 543                 $arr = array();
 544                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 545                         $arr[$row->il_to] = 1;
 546                 }
 547                 $this->mDb->freeResult( $res );
 548                 return $arr;
 549         }
 550
 551         /**
 552          * Get an array of existing external links, URLs in the keys
 553          * @private
 554          */
 555         function getExistingExternals() {
 556                 $fname = 'LinksUpdate::getExistingExternals';
 557                 $res = $this->mDb->select( 'externallinks', array( 'el_to' ),
 558                         array( 'el_from' => $this->mId ), $fname, $this->mOptions );
 559                 $arr = array();
 560                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 561                         $arr[$row->el_to] = 1;
 562                 }
 563                 $this->mDb->freeResult( $res );
 564                 return $arr;
 565         }
 566
 567         /**
 568          * Get an array of existing categories, with the name in the key and sort key in the value.
 569          * @private
 570          */
 571         function getExistingCategories() {
 572                 $fname = 'LinksUpdate::getExistingCategories';
 573                 $res = $this->mDb->select( 'categorylinks', array( 'cl_to', 'cl_sortkey' ),
 574                         array( 'cl_from' => $this->mId ), $fname, $this->mOptions );
 575                 $arr = array();
 576                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 577                         $arr[$row->cl_to] = $row->cl_sortkey;
 578                 }
 579                 $this->mDb->freeResult( $res );
 580                 return $arr;
 581         }
 582
 583         /**
 584          * Get an array of existing interlanguage links, with the language code in the key and the
 585          * title in the value.
 586          * @private
 587          */
 588         function getExistingInterlangs() {
 589                 $fname = 'LinksUpdate::getExistingInterlangs';
 590                 $res = $this->mDb->select( 'langlinks', array( 'll_lang', 'll_title' ),
 591                         array( 'll_from' => $this->mId ), $fname, $this->mOptions );
 592                 $arr = array();
 593                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 594                         $arr[$row->ll_lang] = $row->ll_title;
 595                 }
 596                 return $arr;
 597         }
 598 }
 599 ?>