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                         throw new MWException( "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                                 Job::queueLinksJobs( $tlto );
 123                         }
 124                 }
 125
 126                 # Category links
 127                 $existing = $this->getExistingCategories();
 128                 $this->incrTableUpdate( 'categorylinks', 'cl', $this->getCategoryDeletions( $existing ),
 129                         $this->getCategoryInsertions( $existing ) );
 130
 131                 # Invalidate all categories which were added, deleted or changed (set symmetric difference)
 132                 $categoryUpdates = array_diff_assoc( $existing, $this->mCategories ) + array_diff_assoc( $this->mCategories, $existing );
 133                 $this->invalidateCategories( $categoryUpdates );
 134
 135                 wfProfileOut( $fname );
 136         }
 137
 138         /**
 139          * Link update which clears the previous entries and inserts new ones
 140          * May be slower or faster depending on level of lock contention and write speed of DB
 141          * Also useful where link table corruption needs to be repaired, e.g. in refreshLinks.php
 142          */
 143         function doDumbUpdate() {
 144                 $fname = 'LinksUpdate::doDumbUpdate';
 145                 wfProfileIn( $fname );
 146
 147                 # Refresh category pages and image description pages
 148                 $existing = $this->getExistingCategories();
 149                 $categoryUpdates = array_diff_assoc( $existing, $this->mCategories ) + array_diff_assoc( $this->mCategories, $existing );
 150                 $existing = $this->getExistingImages();
 151                 $imageUpdates = array_diff_key( $existing, $this->mImages ) + array_diff_key( $this->mImages, $existing );
 152
 153                 # Refresh links of all pages including this page
 154                 if ( $this->mRecursive ) {
 155                         $tlto = $this->mTitle->getTemplateLinksTo();
 156                         if ( count( $tlto ) ) {
 157                                 Job::queueLinksJobs( $tlto );
 158                         }
 159                 }
 160
 161                 $this->dumbTableUpdate( 'pagelinks',     $this->getLinkInsertions(),     'pl_from' );
 162                 $this->dumbTableUpdate( 'imagelinks',    $this->getImageInsertions(),    'il_from' );
 163                 $this->dumbTableUpdate( 'categorylinks', $this->getCategoryInsertions(), 'cl_from' );
 164                 $this->dumbTableUpdate( 'templatelinks', $this->getTemplateInsertions(), 'tl_from' );
 165                 $this->dumbTableUpdate( 'externallinks', $this->getExternalInsertions(), 'el_from' );
 166                 $this->dumbTableUpdate( 'langlinks',     $this->getInterlangInsertions(), 'll_from' );
 167
 168                 # Update the cache of all the category pages and image description pages which were changed
 169                 $this->invalidateCategories( $categoryUpdates );
 170                 $this->invalidateImageDescriptions( $imageUpdates );
 171
 172                 wfProfileOut( $fname );
 173         }
 174
 175         /**
 176          * Invalidate the cache of a list of pages from a single namespace
 177          *
 178          * @param integer $namespace
 179          * @param array $dbkeys
 180          */
 181         function invalidatePages( $namespace, $dbkeys ) {
 182                 $fname = 'LinksUpdate::invalidatePages';
 183
 184                 if ( !count( $dbkeys ) ) {
 185                         return;
 186                 }
 187
 188                 /**
 189                  * Determine which pages need to be updated
 190                  * This is necessary to prevent the job queue from smashing the DB with
 191                  * large numbers of concurrent invalidations of the same page
 192                  */
 193                 $now = $this->mDb->timestamp();
 194                 $ids = array();
 195                 $res = $this->mDb->select( 'page', array( 'page_id' ),
 196                         array(
 197                                 'page_namespace' => $namespace,
 198                                 'page_title IN (' . $this->mDb->makeList( $dbkeys ) . ')',
 199                                 'page_touched < ' . $this->mDb->addQuotes( $now )
 200                         ), $fname
 201                 );
 202                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 203                         $ids[] = $row->page_id;
 204                 }
 205                 if ( !count( $ids ) ) {
 206                         return;
 207                 }
 208
 209                 /**
 210                  * Do the update
 211                  * We still need the page_touched condition, in case the row has changed since
 212                  * the non-locking select above.
 213                  */
 214                 $this->mDb->update( 'page', array( 'page_touched' => $now ),
 215                         array(
 216                                 'page_id IN (' . $this->mDb->makeList( $ids ) . ')',
 217                                 'page_touched < ' . $this->mDb->addQuotes( $now )
 218                         ), $fname
 219                 );
 220         }
 221
 222         function invalidateCategories( $cats ) {
 223                 $this->invalidatePages( NS_CATEGORY, array_keys( $cats ) );
 224         }
 225
 226         function invalidateImageDescriptions( $images ) {
 227                 $this->invalidatePages( NS_IMAGE, array_keys( $images ) );
 228         }
 229
 230         function dumbTableUpdate( $table, $insertions, $fromField ) {
 231                 $fname = 'LinksUpdate::dumbTableUpdate';
 232                 $this->mDb->delete( $table, array( $fromField => $this->mId ), $fname );
 233                 if ( count( $insertions ) ) {
 234                         # The link array was constructed without FOR UPDATE, so there may be collisions
 235                         # This may cause minor link table inconsistencies, which is better than
 236                         # crippling the site with lock contention.
 237                         $this->mDb->insert( $table, $insertions, $fname, array( 'IGNORE' ) );
 238                 }
 239         }
 240
 241         /**
 242          * Make a WHERE clause from a 2-d NS/dbkey array
 243          *
 244          * @param array $arr 2-d array indexed by namespace and DB key
 245          * @param string $prefix Field name prefix, without the underscore
 246          */
 247         function makeWhereFrom2d( &$arr, $prefix ) {
 248                 $lb = new LinkBatch;
 249                 $lb->setArray( $arr );
 250                 return $lb->constructSet( $prefix, $this->mDb );
 251         }
 252
 253         /**
 254          * Update a table by doing a delete query then an insert query
 255          * @private
 256          */
 257         function incrTableUpdate( $table, $prefix, $deletions, $insertions ) {
 258                 $fname = 'LinksUpdate::incrTableUpdate';
 259                 $where = array( "{$prefix}_from" => $this->mId );
 260                 if ( $table == 'pagelinks' || $table == 'templatelinks' ) {
 261                         $clause = $this->makeWhereFrom2d( $deletions, $prefix );
 262                         if ( $clause ) {
 263                                 $where[] = $clause;
 264                         } else {
 265                                 $where = false;
 266                         }
 267                 } else {
 268                         if ( $table == 'langlinks' ) {
 269                                 $toField = 'll_lang';
 270                         } else {
 271                                 $toField = $prefix . '_to';
 272                         }
 273                         if ( count( $deletions ) ) {
 274                                 $where[] = "$toField IN (" . $this->mDb->makeList( array_keys( $deletions ) ) . ')';
 275                         } else {
 276                                 $where = false;
 277                         }
 278                 }
 279                 if ( $where ) {
 280                         $this->mDb->delete( $table, $where, $fname );
 281                 }
 282                 if ( count( $insertions ) ) {
 283                         $this->mDb->insert( $table, $insertions, $fname, 'IGNORE' );
 284                 }
 285         }
 286
 287
 288         /**
 289          * Get an array of pagelinks insertions for passing to the DB
 290          * Skips the titles specified by the 2-D array $existing
 291          * @private
 292          */
 293         function getLinkInsertions( $existing = array() ) {
 294                 $arr = array();
 295                 foreach( $this->mLinks as $ns => $dbkeys ) {
 296                         # array_diff_key() was introduced in PHP 5.1, there is a compatibility function
 297                         # in GlobalFunctions.php
 298                         $diffs = isset( $existing[$ns] ) ? array_diff_key( $dbkeys, $existing[$ns] ) : $dbkeys;
 299                         foreach ( $diffs as $dbk => $id ) {
 300                                 $arr[] = array(
 301                                         'pl_from'      => $this->mId,
 302                                         'pl_namespace' => $ns,
 303                                         'pl_title'     => $dbk
 304                                 );
 305                         }
 306                 }
 307                 return $arr;
 308         }
 309
 310         /**
 311          * Get an array of template insertions. Like getLinkInsertions()
 312          * @private
 313          */
 314         function getTemplateInsertions( $existing = array() ) {
 315                 $arr = array();
 316                 foreach( $this->mTemplates as $ns => $dbkeys ) {
 317                         $diffs = isset( $existing[$ns] ) ? array_diff_key( $dbkeys, $existing[$ns] ) : $dbkeys;
 318                         foreach ( $diffs as $dbk => $id ) {
 319                                 $arr[] = array(
 320                                         'tl_from'      => $this->mId,
 321                                         'tl_namespace' => $ns,
 322                                         'tl_title'     => $dbk
 323                                 );
 324                         }
 325                 }
 326                 return $arr;
 327         }
 328
 329         /**
 330          * Get an array of image insertions
 331          * Skips the names specified in $existing
 332          * @private
 333          */
 334         function getImageInsertions( $existing = array() ) {
 335                 $arr = array();
 336                 $diffs = array_diff_key( $this->mImages, $existing );
 337                 foreach( $diffs as $iname => $dummy ) {
 338                         $arr[] = array(
 339                                 'il_from' => $this->mId,
 340                                 'il_to'   => $iname
 341                         );
 342                 }
 343                 return $arr;
 344         }
 345
 346         /**
 347          * Get an array of externallinks insertions. Skips the names specified in $existing
 348          * @private
 349          */
 350         function getExternalInsertions( $existing = array() ) {
 351                 $arr = array();
 352                 $diffs = array_diff_key( $this->mExternals, $existing );
 353                 foreach( $diffs as $url => $dummy ) {
 354                         $arr[] = array(
 355                                 'el_from'   => $this->mId,
 356                                 'el_to'     => $url,
 357                                 'el_index'  => wfMakeUrlIndex( $url ),
 358                         );
 359                 }
 360                 return $arr;
 361         }
 362
 363         /**
 364          * Get an array of category insertions
 365          * @param array $existing Array mapping existing category names to sort keys. If both
 366          * match a link in $this, the link will be omitted from the output
 367          * @private
 368          */
 369         function getCategoryInsertions( $existing = array() ) {
 370                 $diffs = array_diff_assoc( $this->mCategories, $existing );
 371                 $arr = array();
 372                 foreach ( $diffs as $name => $sortkey ) {
 373                         $arr[] = array(
 374                                 'cl_from'    => $this->mId,
 375                                 'cl_to'      => $name,
 376                                 'cl_sortkey' => $sortkey,
 377                                 'cl_timestamp' => $this->mDb->timestamp()
 378                         );
 379                 }
 380                 return $arr;
 381         }
 382
 383         /**
 384          * Get an array of interlanguage link insertions
 385          * @param array $existing Array mapping existing language codes to titles
 386          * @private
 387          */
 388         function getInterlangInsertions( $existing = array() ) {
 389             $diffs = array_diff_assoc( $this->mInterlangs, $existing );
 390             $arr = array();
 391             foreach( $diffs as $lang => $title ) {
 392                 $arr[] = array(
 393                     'll_from'  => $this->mId,
 394                     'll_lang'  => $lang,
 395                     'll_title' => $title
 396                 );
 397             }
 398             return $arr;
 399         }
 400
 401         /**
 402          * Given an array of existing links, returns those links which are not in $this
 403          * and thus should be deleted.
 404          * @private
 405          */
 406         function getLinkDeletions( $existing ) {
 407                 $del = array();
 408                 foreach ( $existing as $ns => $dbkeys ) {
 409                         if ( isset( $this->mLinks[$ns] ) ) {
 410                                 $del[$ns] = array_diff_key( $existing[$ns], $this->mLinks[$ns] );
 411                         } else {
 412                                 $del[$ns] = $existing[$ns];
 413                         }
 414                 }
 415                 return $del;
 416         }
 417
 418         /**
 419          * Given an array of existing templates, returns those templates which are not in $this
 420          * and thus should be deleted.
 421          * @private
 422          */
 423         function getTemplateDeletions( $existing ) {
 424                 $del = array();
 425                 foreach ( $existing as $ns => $dbkeys ) {
 426                         if ( isset( $this->mTemplates[$ns] ) ) {
 427                                 $del[$ns] = array_diff_key( $existing[$ns], $this->mTemplates[$ns] );
 428                         } else {
 429                                 $del[$ns] = $existing[$ns];
 430                         }
 431                 }
 432                 return $del;
 433         }
 434
 435         /**
 436          * Given an array of existing images, returns those images which are not in $this
 437          * and thus should be deleted.
 438          * @private
 439          */
 440         function getImageDeletions( $existing ) {
 441                 return array_diff_key( $existing, $this->mImages );
 442         }
 443
 444         /**
 445          * Given an array of existing external links, returns those links which are not
 446          * in $this and thus should be deleted.
 447          * @private
 448          */
 449         function getExternalDeletions( $existing ) {
 450                 return array_diff_key( $existing, $this->mExternals );
 451         }
 452
 453         /**
 454          * Given an array of existing categories, returns those categories which are not in $this
 455          * and thus should be deleted.
 456          * @private
 457          */
 458         function getCategoryDeletions( $existing ) {
 459                 return array_diff_assoc( $existing, $this->mCategories );
 460         }
 461
 462         /**
 463          * Given an array of existing interlanguage links, returns those links which are not
 464          * in $this and thus should be deleted.
 465          * @private
 466          */
 467         function getInterlangDeletions( $existing ) {
 468             return array_diff_assoc( $existing, $this->mInterlangs );
 469         }
 470
 471         /**
 472          * Get an array of existing links, as a 2-D array
 473          * @private
 474          */
 475         function getExistingLinks() {
 476                 $fname = 'LinksUpdate::getExistingLinks';
 477                 $res = $this->mDb->select( 'pagelinks', array( 'pl_namespace', 'pl_title' ),
 478                         array( 'pl_from' => $this->mId ), $fname, $this->mOptions );
 479                 $arr = array();
 480                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 481                         if ( !isset( $arr[$row->pl_namespace] ) ) {
 482                                 $arr[$row->pl_namespace] = array();
 483                         }
 484                         $arr[$row->pl_namespace][$row->pl_title] = 1;
 485                 }
 486                 $this->mDb->freeResult( $res );
 487                 return $arr;
 488         }
 489
 490         /**
 491          * Get an array of existing templates, as a 2-D array
 492          * @private
 493          */
 494         function getExistingTemplates() {
 495                 $fname = 'LinksUpdate::getExistingTemplates';
 496                 $res = $this->mDb->select( 'templatelinks', array( 'tl_namespace', 'tl_title' ),
 497                         array( 'tl_from' => $this->mId ), $fname, $this->mOptions );
 498                 $arr = array();
 499                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 500                         if ( !isset( $arr[$row->tl_namespace] ) ) {
 501                                 $arr[$row->tl_namespace] = array();
 502                         }
 503                         $arr[$row->tl_namespace][$row->tl_title] = 1;
 504                 }
 505                 $this->mDb->freeResult( $res );
 506                 return $arr;
 507         }
 508
 509         /**
 510          * Get an array of existing images, image names in the keys
 511          * @private
 512          */
 513         function getExistingImages() {
 514                 $fname = 'LinksUpdate::getExistingImages';
 515                 $res = $this->mDb->select( 'imagelinks', array( 'il_to' ),
 516                         array( 'il_from' => $this->mId ), $fname, $this->mOptions );
 517                 $arr = array();
 518                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 519                         $arr[$row->il_to] = 1;
 520                 }
 521                 $this->mDb->freeResult( $res );
 522                 return $arr;
 523         }
 524
 525         /**
 526          * Get an array of existing external links, URLs in the keys
 527          * @private
 528          */
 529         function getExistingExternals() {
 530                 $fname = 'LinksUpdate::getExistingExternals';
 531                 $res = $this->mDb->select( 'externallinks', array( 'el_to' ),
 532                         array( 'el_from' => $this->mId ), $fname, $this->mOptions );
 533                 $arr = array();
 534                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 535                         $arr[$row->el_to] = 1;
 536                 }
 537                 $this->mDb->freeResult( $res );
 538                 return $arr;
 539         }
 540
 541         /**
 542          * Get an array of existing categories, with the name in the key and sort key in the value.
 543          * @private
 544          */
 545         function getExistingCategories() {
 546                 $fname = 'LinksUpdate::getExistingCategories';
 547                 $res = $this->mDb->select( 'categorylinks', array( 'cl_to', 'cl_sortkey' ),
 548                         array( 'cl_from' => $this->mId ), $fname, $this->mOptions );
 549                 $arr = array();
 550                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 551                         $arr[$row->cl_to] = $row->cl_sortkey;
 552                 }
 553                 $this->mDb->freeResult( $res );
 554                 return $arr;
 555         }
 556
 557         /**
 558          * Get an array of existing interlanguage links, with the language code in the key and the
 559          * title in the value.
 560          * @private
 561          */
 562         function getExistingInterlangs() {
 563                 $fname = 'LinksUpdate::getExistingInterlangs';
 564                 $res = $this->mDb->select( 'langlinks', array( 'll_lang', 'll_title' ),
 565                         array( 'll_from' => $this->mId ), $fname, $this->mOptions );
 566                 $arr = array();
 567                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 568                         $arr[$row->ll_lang] = $row->ll_title;
 569                 }
 570                 return $arr;
 571         }
 572 }
 573 ?>