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