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