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