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                 $mParserOutput,  # Parser output containing the links to be inserted into the database
  19                 $mLinks,         # Map of title strings to IDs for the links in the document
  20                 $mImages,        # DB keys of the images used, in the array key only
  21                 $mTemplates,     # Map of title strings to IDs for the template references, including broken ones
  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                 $this->mParserOutput = $parserOutput;
  50
  51                 // Shortcut aliases
  52                 $this->mLinks =& $this->mParserOutput->getLinks();
  53                 $this->mImages =& $this->mParserOutput->getImages();
  54                 $this->mTemplates =& $this->mParserOutput->getTemplates();
  55                 $this->mCategories =& $this->mParserOutput->getCategories();
  56
  57         }
  58
  59         /**
  60          * Update link tables with outgoing links from an updated article
  61          */
  62         function doUpdate() {
  63                 global $wgUseDumbLinkUpdate;
  64                 if ( $wgUseDumbLinkUpdate ) {
  65                         $this->doDumbUpdate();
  66                 } else {
  67                         $this->doIncrementalUpdate();
  68                 }
  69         }
  70
  71         function doIncrementalUpdate() {
  72                 $fname = 'LinksUpdate::doIncrementalUpdate';
  73                 wfProfileIn( $fname );
  74
  75                 # Page links
  76                 $existing = $this->getExistingLinks();
  77                 $this->incrTableUpdate( 'pagelinks', 'pl', $this->getLinkDeletions( $existing ),
  78                         $this->getLinkInsertions( $existing ) );
  79
  80                 # Template links
  81                 $existing = $this->getExistingTemplates();
  82                 $this->incrTableUpdate( 'templatelinks', 'tl', $this->getTemplateDeletions( $existing ),
  83                         $this->getTemplateInsertions( $existing ) );
  84
  85                 # Image links
  86                 $existing = $this->getExistingImages();
  87                 $this->incrTableUpdate( 'imagelinks', 'il', $this->getImageDeletions( $existing ),
  88                         $this->getImageInsertions( $existing ) );
  89
  90                 # Category links
  91                 $existing = $this->getExistingCategories();
  92                 $this->incrTableUpdate( 'categorylinks', 'cl', $this->getCategoryDeletions( $existing ),
  93                         $this->getCategoryInsertions( $existing ) );
  94
  95                 # I think this works out to a set XOR operation, the idea is to invalidate all
  96                 # categories which were added, deleted or changed
  97                 # FIXME: surely there's a more appropriate place to put this update?
  98                 $categoryUpdates = array_diff_assoc( $existing, $this->mCategories ) + array_diff_assoc( $this->mCategories, $existing );
  99                 $this->invalidateCategories( $categoryUpdates );
 100
 101                 wfProfileOut( $fname );
 102         }
 103
 104         /**
 105           * Link update which clears the previous entries and inserts new ones
 106           * May be slower or faster depending on level of lock contention and write speed of DB
 107           * Also useful where link table corruption needs to be repaired, e.g. in refreshLinks.php
 108          */
 109         function doDumbUpdate() {
 110                 $fname = 'LinksUpdate::doDumbUpdate';
 111                 wfProfileIn( $fname );
 112
 113                 $existing = $this->getExistingCategories();
 114                 $categoryUpdates = array_diff_assoc( $existing, $this->mCategories ) + array_diff_assoc( $this->mCategories, $existing );
 115
 116                 $this->dumbTableUpdate( 'pagelinks',     $this->getLinkInsertions(),     'pl_from' );
 117                 $this->dumbTableUpdate( 'imagelinks',    $this->getImageInsertions(),    'il_from' );
 118                 $this->dumbTableUpdate( 'categorylinks', $this->getCategoryInsertions(), 'cl_from' );
 119                 $this->dumbTableUpdate( 'templatelinks', $this->getTemplateInsertions(), 'tl_from' );
 120
 121                 # Update the cache of all the category pages
 122                 $this->invalidateCategories( $categoryUpdates );
 123
 124                 wfProfileOut( $fname );
 125         }
 126
 127         function invalidateCategories( $cats ) {
 128                 $fname = 'LinksUpdate::invalidateCategories';
 129                 if ( count( $cats ) ) {
 130                         $this->mDb->update( 'page', array( 'page_touched' => $this->mDb->timestamp() ),
 131                                 array(
 132                                         'page_namespace' => NS_CATEGORY,
 133                                         'page_title IN (' . $this->mDb->makeList( array_keys( $cats ) ) . ')'
 134                                 ), $fname
 135                         );
 136                 }
 137         }
 138
 139         function dumbTableUpdate( $table, $insertions, $fromField ) {
 140                 $fname = 'LinksUpdate::dumbTableUpdate';
 141                 $this->mDb->delete( $table, array( $fromField => $this->mId ), $fname );
 142                 if ( count( $insertions ) ) {
 143                         # The link array was constructed without FOR UPDATE, so there may be collisions
 144                         # Ignoring for now, I'm not sure if that causes problems or not, but I'm fairly
 145                         # sure it's better than without IGNORE
 146                         $this->mDb->insert( $table, $insertions, $fname, array( 'IGNORE' ) );
 147                 }
 148         }
 149
 150         /**
 151          * Make a WHERE clause from a 2-d NS/dbkey array
 152          *
 153          * @param array $arr 2-d array indexed by namespace and DB key
 154          * @param string $prefix Field name prefix, without the underscore
 155          */
 156         function makeWhereFrom2d( &$arr, $prefix ) {
 157                 $lb = new LinkBatch;
 158                 $lb->setArray( $arr );
 159                 return $lb->constructSet( $prefix, $this->mDb );
 160         }
 161
 162         /**
 163          * Update a table by doing a delete query then an insert query
 164          * @private
 165          */
 166         function incrTableUpdate( $table, $prefix, $deletions, $insertions ) {
 167                 $fname = 'LinksUpdate::incrTableUpdate';
 168                 $where = array( "{$prefix}_from" => $this->mId );
 169                 if ( $table == 'pagelinks' || $table == 'templatelinks' ) {
 170                         $clause = $this->makeWhereFrom2d( $deletions, $prefix );
 171                         if ( $clause ) {
 172                                 $where[] = $clause;
 173                         } else {
 174                                 $where = false;
 175                         }
 176                 } else {
 177                         if ( count( $deletions ) ) {
 178                                 $where[] = "{$prefix}_to IN (" . $this->mDb->makeList( array_keys( $deletions ) ) . ')';
 179                         } else {
 180                                 $where = false;
 181                         }
 182                 }
 183                 if ( $where ) {
 184                         $this->mDb->delete( $table, $where, $fname );
 185                 }
 186                 if ( count( $insertions ) ) {
 187                         $this->mDb->insert( $table, $insertions, $fname, 'IGNORE' );
 188                 }
 189         }
 190
 191
 192         /**
 193          * Get an array of pagelinks insertions for passing to the DB
 194          * Skips the titles specified by the 2-D array $existing
 195          * @private
 196          */
 197         function getLinkInsertions( $existing = array() ) {
 198                 $arr = array();
 199                 foreach( $this->mLinks as $ns => $dbkeys ) {
 200                         # array_diff_key() was introduced in PHP 5.1, there is a compatibility function
 201                         # in GlobalFunctions.php
 202                         $diffs = isset( $existing[$ns] ) ? array_diff_key( $dbkeys, $existing[$ns] ) : $dbkeys;
 203                         foreach ( $diffs as $dbk => $id ) {
 204                                 $arr[] = array(
 205                                         'pl_from'      => $this->mId,
 206                                         'pl_namespace' => $ns,
 207                                         'pl_title'     => $dbk
 208                                 );
 209                         }
 210                 }
 211                 return $arr;
 212         }
 213
 214         /**
 215          * Get an array of template insertions. Like getLinkInsertions()
 216          * @private
 217          */
 218         function getTemplateInsertions( $existing = array() ) {
 219                 $arr = array();
 220                 foreach( $this->mTemplates as $ns => $dbkeys ) {
 221                         $diffs = isset( $existing[$ns] ) ? array_diff_key( $dbkeys, $existing[$ns] ) : $dbkeys;
 222                         foreach ( $diffs as $dbk => $id ) {
 223                                 $arr[] = array(
 224                                         'tl_from'      => $this->mId,
 225                                         'tl_namespace' => $ns,
 226                                         'tl_title'     => $dbk
 227                                 );
 228                         }
 229                 }
 230                 return $arr;
 231         }
 232
 233         /**
 234          * Get an array of image insertions
 235          * Skips the names specified in $existing
 236          * @private
 237          */
 238         function getImageInsertions( $existing = array() ) {
 239                 $arr = array();
 240                 $diffs = array_diff_key( $this->mImages, $existing );
 241                 foreach( $diffs as $iname => $val ) {
 242                         $arr[] = array(
 243                                 'il_from' => $this->mId,
 244                                 'il_to'   => $iname
 245                         );
 246                 }
 247                 return $arr;
 248         }
 249
 250         /**
 251          * Get an array of category insertions
 252          * @param array $existing Array mapping existing category names to sort keys. If both
 253          * match a link in $this, the link will be omitted from the output
 254          * @private
 255          */
 256         function getCategoryInsertions( $existing = array() ) {
 257                 $diffs = array_diff_assoc( $this->mCategories, $existing );
 258                 $arr = array();
 259                 foreach ( $diffs as $name => $sortkey ) {
 260                         $arr[] = array(
 261                                 'cl_from'    => $this->mId,
 262                                 'cl_to'      => $name,
 263                                 'cl_sortkey' => $sortkey
 264                         );
 265                 }
 266                 return $arr;
 267         }
 268
 269         /**
 270          * Given an array of existing links, returns those links which are not in $this
 271          * and thus should be deleted.
 272          * @private
 273          */
 274         function getLinkDeletions( $existing ) {
 275                 $del = array();
 276                 foreach ( $existing as $ns => $dbkeys ) {
 277                         if ( isset( $this->mLinks[$ns] ) ) {
 278                                 $del[$ns] = array_diff_key( $existing[$ns], $this->mLinks[$ns] );
 279                         } else {
 280                                 $del[$ns] = $existing[$ns];
 281                         }
 282                 }
 283                 return $del;
 284         }
 285
 286         /**
 287          * Given an array of existing templates, returns those templates which are not in $this
 288          * and thus should be deleted.
 289          * @private
 290          */
 291         function getTemplateDeletions( $existing ) {
 292                 $del = array();
 293                 foreach ( $existing as $ns => $dbkeys ) {
 294                         if ( isset( $this->mTemplates[$ns] ) ) {
 295                                 $del[$ns] = array_diff_key( $existing[$ns], $this->mTemplates[$ns] );
 296                         } else {
 297                                 $del[$ns] = $existing[$ns];
 298                         }
 299                 }
 300                 return $del;
 301         }
 302
 303         /**
 304          * Given an array of existing images, returns those images which are not in $this
 305          * and thus should be deleted.
 306          * @private
 307          */
 308         function getImageDeletions( $existing ) {
 309                 return array_diff_key( $existing, $this->mImages );
 310         }
 311
 312         /**
 313          * Given an array of existing categories, returns those categories which are not in $this
 314          * and thus should be deleted.
 315          * @private
 316          */
 317         function getCategoryDeletions( $existing ) {
 318                 return array_diff_assoc( $existing, $this->mCategories );
 319         }
 320
 321         /**
 322          * Get an array of existing links, as a 2-D array
 323          * @private
 324          */
 325         function getExistingLinks() {
 326                 $fname = 'LinksUpdate::getExistingLinks';
 327                 $res = $this->mDb->select( 'pagelinks', array( 'pl_namespace', 'pl_title' ),
 328                         array( 'pl_from' => $this->mId ), $fname, $this->mOptions );
 329                 $arr = array();
 330                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 331                         if ( !isset( $arr[$row->pl_namespace] ) ) {
 332                                 $arr[$row->pl_namespace] = array();
 333                         }
 334                         $arr[$row->pl_namespace][$row->pl_title] = 1;
 335                 }
 336                 return $arr;
 337         }
 338
 339         /**
 340          * Get an array of existing templates, as a 2-D array
 341          * @private
 342          */
 343         function getExistingTemplates() {
 344                 $fname = 'LinksUpdate::getExistingTemplates';
 345                 $res = $this->mDb->select( 'templatelinks', array( 'tl_namespace', 'tl_title' ),
 346                         array( 'tl_from' => $this->mId ), $fname, $this->mOptions );
 347                 $arr = array();
 348                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 349                         if ( !isset( $arr[$row->tl_namespace] ) ) {
 350                                 $arr[$row->tl_namespace] = array();
 351                         }
 352                         $arr[$row->tl_namespace][$row->tl_title] = 1;
 353                 }
 354                 return $arr;
 355         }
 356
 357         /**
 358          * Get an array of existing images, image names in the keys
 359          * @private
 360          */
 361         function getExistingImages() {
 362                 $fname = 'LinksUpdate::getExistingImages';
 363                 $res = $this->mDb->select( 'imagelinks', array( 'il_to' ),
 364                         array( 'il_from' => $this->mId ), $fname, $this->mOptions );
 365                 $arr = array();
 366                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 367                         $arr[$row->il_to] = 1;
 368                 }
 369                 return $arr;
 370         }
 371
 372         /**
 373          * Get an array of existing categories, with the name in the key and sort key in the value.
 374          * @private
 375          */
 376         function getExistingCategories() {
 377                 $fname = 'LinksUpdate::getExistingCategories';
 378                 $res = $this->mDb->select( 'categorylinks', array( 'cl_to', 'cl_sortkey' ),
 379                         array( 'cl_from' => $this->mId ), $fname, $this->mOptions );
 380                 $arr = array();
 381                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 382                         $arr[$row->cl_to] = $row->cl_sortkey;
 383                 }
 384                 return $arr;
 385         }
 386 }
 387 ?>