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