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         function invalidateCategories( $cats ) {
 178                 $fname = 'LinksUpdate::invalidateCategories';
 179                 if ( count( $cats ) ) {
 180                         $this->mDb->update( 'page', array( 'page_touched' => $this->mDb->timestamp() ),
 181                                 array(
 182                                         'page_namespace' => NS_CATEGORY,
 183                                         'page_title IN (' . $this->mDb->makeList( array_keys( $cats ) ) . ')'
 184                                 ), $fname
 185                         );
 186                 }
 187         }
 188
 189         function invalidateImageDescriptions( $images ) {
 190                 $fname = 'LinksUpdate::invalidateImageDescriptions';
 191                 if ( count( $images ) ) {
 192                         $this->mDb->update( 'page', array( 'page_touched' => $this->mDb->timestamp() ),
 193                                 array(
 194                                         'page_namespace' => NS_IMAGE,
 195                                         'page_title IN (' . $this->mDb->makeList( array_keys( $images ) ) . ')'
 196                                 ), $fname
 197                         );
 198                 }
 199         }
 200
 201         function dumbTableUpdate( $table, $insertions, $fromField ) {
 202                 $fname = 'LinksUpdate::dumbTableUpdate';
 203                 $this->mDb->delete( $table, array( $fromField => $this->mId ), $fname );
 204                 if ( count( $insertions ) ) {
 205                         # The link array was constructed without FOR UPDATE, so there may be collisions
 206                         # This may cause minor link table inconsistencies, which is better than
 207                         # crippling the site with lock contention.
 208                         $this->mDb->insert( $table, $insertions, $fname, array( 'IGNORE' ) );
 209                 }
 210         }
 211
 212         /**
 213          * Make a WHERE clause from a 2-d NS/dbkey array
 214          *
 215          * @param array $arr 2-d array indexed by namespace and DB key
 216          * @param string $prefix Field name prefix, without the underscore
 217          */
 218         function makeWhereFrom2d( &$arr, $prefix ) {
 219                 $lb = new LinkBatch;
 220                 $lb->setArray( $arr );
 221                 return $lb->constructSet( $prefix, $this->mDb );
 222         }
 223
 224         /**
 225          * Update a table by doing a delete query then an insert query
 226          * @private
 227          */
 228         function incrTableUpdate( $table, $prefix, $deletions, $insertions ) {
 229                 $fname = 'LinksUpdate::incrTableUpdate';
 230                 $where = array( "{$prefix}_from" => $this->mId );
 231                 if ( $table == 'pagelinks' || $table == 'templatelinks' ) {
 232                         $clause = $this->makeWhereFrom2d( $deletions, $prefix );
 233                         if ( $clause ) {
 234                                 $where[] = $clause;
 235                         } else {
 236                                 $where = false;
 237                         }
 238                 } else {
 239                         if ( $table == 'langlinks' ) {
 240                                 $toField = 'll_lang';
 241                         } else {
 242                                 $toField = $prefix . '_to';
 243                         }
 244                         if ( count( $deletions ) ) {
 245                                 $where[] = "$toField IN (" . $this->mDb->makeList( array_keys( $deletions ) ) . ')';
 246                         } else {
 247                                 $where = false;
 248                         }
 249                 }
 250                 if ( $where ) {
 251                         $this->mDb->delete( $table, $where, $fname );
 252                 }
 253                 if ( count( $insertions ) ) {
 254                         $this->mDb->insert( $table, $insertions, $fname, 'IGNORE' );
 255                 }
 256         }
 257
 258
 259         /**
 260          * Get an array of pagelinks insertions for passing to the DB
 261          * Skips the titles specified by the 2-D array $existing
 262          * @private
 263          */
 264         function getLinkInsertions( $existing = array() ) {
 265                 $arr = array();
 266                 foreach( $this->mLinks as $ns => $dbkeys ) {
 267                         # array_diff_key() was introduced in PHP 5.1, there is a compatibility function
 268                         # in GlobalFunctions.php
 269                         $diffs = isset( $existing[$ns] ) ? array_diff_key( $dbkeys, $existing[$ns] ) : $dbkeys;
 270                         foreach ( $diffs as $dbk => $id ) {
 271                                 $arr[] = array(
 272                                         'pl_from'      => $this->mId,
 273                                         'pl_namespace' => $ns,
 274                                         'pl_title'     => $dbk
 275                                 );
 276                         }
 277                 }
 278                 return $arr;
 279         }
 280
 281         /**
 282          * Get an array of template insertions. Like getLinkInsertions()
 283          * @private
 284          */
 285         function getTemplateInsertions( $existing = array() ) {
 286                 $arr = array();
 287                 foreach( $this->mTemplates as $ns => $dbkeys ) {
 288                         $diffs = isset( $existing[$ns] ) ? array_diff_key( $dbkeys, $existing[$ns] ) : $dbkeys;
 289                         foreach ( $diffs as $dbk => $id ) {
 290                                 $arr[] = array(
 291                                         'tl_from'      => $this->mId,
 292                                         'tl_namespace' => $ns,
 293                                         'tl_title'     => $dbk
 294                                 );
 295                         }
 296                 }
 297                 return $arr;
 298         }
 299
 300         /**
 301          * Get an array of image insertions
 302          * Skips the names specified in $existing
 303          * @private
 304          */
 305         function getImageInsertions( $existing = array() ) {
 306                 $arr = array();
 307                 $diffs = array_diff_key( $this->mImages, $existing );
 308                 foreach( $diffs as $iname => $dummy ) {
 309                         $arr[] = array(
 310                                 'il_from' => $this->mId,
 311                                 'il_to'   => $iname
 312                         );
 313                 }
 314                 return $arr;
 315         }
 316
 317         /**
 318          * Get an array of externallinks insertions. Skips the names specified in $existing
 319          * @private
 320          */
 321         function getExternalInsertions( $existing = array() ) {
 322                 $arr = array();
 323                 $diffs = array_diff_key( $this->mExternals, $existing );
 324                 foreach( $diffs as $url => $dummy ) {
 325                         $arr[] = array(
 326                                 'el_from'   => $this->mId,
 327                                 'el_to'     => $url,
 328                                 'el_index'  => wfMakeUrlIndex( $url ),
 329                         );
 330                 }
 331                 return $arr;
 332         }
 333
 334         /**
 335          * Get an array of category insertions
 336          * @param array $existing Array mapping existing category names to sort keys. If both
 337          * match a link in $this, the link will be omitted from the output
 338          * @private
 339          */
 340         function getCategoryInsertions( $existing = array() ) {
 341                 $diffs = array_diff_assoc( $this->mCategories, $existing );
 342                 $arr = array();
 343                 foreach ( $diffs as $name => $sortkey ) {
 344                         $arr[] = array(
 345                                 'cl_from'    => $this->mId,
 346                                 'cl_to'      => $name,
 347                                 'cl_sortkey' => $sortkey,
 348                                 'cl_timestamp' => $this->mDb->timestamp()
 349                         );
 350                 }
 351                 return $arr;
 352         }
 353
 354         /**
 355          * Get an array of interlanguage link insertions
 356          * @param array $existing Array mapping existing language codes to titles
 357          * @private
 358          */
 359         function getInterlangInsertions( $existing = array() ) {
 360             $diffs = array_diff_assoc( $this->mInterlangs, $existing );
 361             $arr = array();
 362             foreach( $diffs as $lang => $title ) {
 363                 $arr[] = array(
 364                     'll_from'  => $this->mId,
 365                     'll_lang'  => $lang,
 366                     'll_title' => $title
 367                 );
 368             }
 369             return $arr;
 370         }
 371
 372         /**
 373          * Given an array of existing links, returns those links which are not in $this
 374          * and thus should be deleted.
 375          * @private
 376          */
 377         function getLinkDeletions( $existing ) {
 378                 $del = array();
 379                 foreach ( $existing as $ns => $dbkeys ) {
 380                         if ( isset( $this->mLinks[$ns] ) ) {
 381                                 $del[$ns] = array_diff_key( $existing[$ns], $this->mLinks[$ns] );
 382                         } else {
 383                                 $del[$ns] = $existing[$ns];
 384                         }
 385                 }
 386                 return $del;
 387         }
 388
 389         /**
 390          * Given an array of existing templates, returns those templates which are not in $this
 391          * and thus should be deleted.
 392          * @private
 393          */
 394         function getTemplateDeletions( $existing ) {
 395                 $del = array();
 396                 foreach ( $existing as $ns => $dbkeys ) {
 397                         if ( isset( $this->mTemplates[$ns] ) ) {
 398                                 $del[$ns] = array_diff_key( $existing[$ns], $this->mTemplates[$ns] );
 399                         } else {
 400                                 $del[$ns] = $existing[$ns];
 401                         }
 402                 }
 403                 return $del;
 404         }
 405
 406         /**
 407          * Given an array of existing images, returns those images which are not in $this
 408          * and thus should be deleted.
 409          * @private
 410          */
 411         function getImageDeletions( $existing ) {
 412                 return array_diff_key( $existing, $this->mImages );
 413         }
 414
 415         /**
 416          * Given an array of existing external links, returns those links which are not
 417          * in $this and thus should be deleted.
 418          * @private
 419          */
 420         function getExternalDeletions( $existing ) {
 421                 return array_diff_key( $existing, $this->mExternals );
 422         }
 423
 424         /**
 425          * Given an array of existing categories, returns those categories which are not in $this
 426          * and thus should be deleted.
 427          * @private
 428          */
 429         function getCategoryDeletions( $existing ) {
 430                 return array_diff_assoc( $existing, $this->mCategories );
 431         }
 432
 433         /**
 434          * Given an array of existing interlanguage links, returns those links which are not
 435          * in $this and thus should be deleted.
 436          * @private
 437          */
 438         function getInterlangDeletions( $existing ) {
 439             return array_diff_assoc( $existing, $this->mInterlangs );
 440         }
 441
 442         /**
 443          * Get an array of existing links, as a 2-D array
 444          * @private
 445          */
 446         function getExistingLinks() {
 447                 $fname = 'LinksUpdate::getExistingLinks';
 448                 $res = $this->mDb->select( 'pagelinks', array( 'pl_namespace', 'pl_title' ),
 449                         array( 'pl_from' => $this->mId ), $fname, $this->mOptions );
 450                 $arr = array();
 451                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 452                         if ( !isset( $arr[$row->pl_namespace] ) ) {
 453                                 $arr[$row->pl_namespace] = array();
 454                         }
 455                         $arr[$row->pl_namespace][$row->pl_title] = 1;
 456                 }
 457                 $this->mDb->freeResult( $res );
 458                 return $arr;
 459         }
 460
 461         /**
 462          * Get an array of existing templates, as a 2-D array
 463          * @private
 464          */
 465         function getExistingTemplates() {
 466                 $fname = 'LinksUpdate::getExistingTemplates';
 467                 $res = $this->mDb->select( 'templatelinks', array( 'tl_namespace', 'tl_title' ),
 468                         array( 'tl_from' => $this->mId ), $fname, $this->mOptions );
 469                 $arr = array();
 470                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 471                         if ( !isset( $arr[$row->tl_namespace] ) ) {
 472                                 $arr[$row->tl_namespace] = array();
 473                         }
 474                         $arr[$row->tl_namespace][$row->tl_title] = 1;
 475                 }
 476                 $this->mDb->freeResult( $res );
 477                 return $arr;
 478         }
 479
 480         /**
 481          * Get an array of existing images, image names in the keys
 482          * @private
 483          */
 484         function getExistingImages() {
 485                 $fname = 'LinksUpdate::getExistingImages';
 486                 $res = $this->mDb->select( 'imagelinks', array( 'il_to' ),
 487                         array( 'il_from' => $this->mId ), $fname, $this->mOptions );
 488                 $arr = array();
 489                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 490                         $arr[$row->il_to] = 1;
 491                 }
 492                 $this->mDb->freeResult( $res );
 493                 return $arr;
 494         }
 495
 496         /**
 497          * Get an array of existing external links, URLs in the keys
 498          * @private
 499          */
 500         function getExistingExternals() {
 501                 $fname = 'LinksUpdate::getExistingExternals';
 502                 $res = $this->mDb->select( 'externallinks', array( 'el_to' ),
 503                         array( 'el_from' => $this->mId ), $fname, $this->mOptions );
 504                 $arr = array();
 505                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 506                         $arr[$row->el_to] = 1;
 507                 }
 508                 $this->mDb->freeResult( $res );
 509                 return $arr;
 510         }
 511
 512         /**
 513          * Get an array of existing categories, with the name in the key and sort key in the value.
 514          * @private
 515          */
 516         function getExistingCategories() {
 517                 $fname = 'LinksUpdate::getExistingCategories';
 518                 $res = $this->mDb->select( 'categorylinks', array( 'cl_to', 'cl_sortkey' ),
 519                         array( 'cl_from' => $this->mId ), $fname, $this->mOptions );
 520                 $arr = array();
 521                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 522                         $arr[$row->cl_to] = $row->cl_sortkey;
 523                 }
 524                 $this->mDb->freeResult( $res );
 525                 return $arr;
 526         }
 527
 528         /**
 529          * Get an array of existing interlanguage links, with the language code in the key and the
 530          * title in the value.
 531          * @private
 532          */
 533         function getExistingInterlangs() {
 534                 $fname = 'LinksUpdate::getExistingInterlangs';
 535                 $res = $this->mDb->select( 'langlinks', array( 'll_lang', 'll_title' ),
 536                         array( 'll_from' => $this->mId ), $fname, $this->mOptions );
 537                 $arr = array();
 538                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 539                         $arr[$row->ll_lang] = $row->ll_title;
 540                 }
 541                 return $arr;
 542         }
 543 }
 544 ?>