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                 # External links
  87                 $existing = $this->getExistingExternals();
  88                 $this->incrTableUpdate( 'externallinks', 'el', $this->getExternalDeletions( $existing ),
  89                         $this->getExternalInsertions( $existing ) );
  90
  91                 # Template links
  92                 $existing = $this->getExistingTemplates();
  93                 $this->incrTableUpdate( 'templatelinks', 'tl', $this->getTemplateDeletions( $existing ),
  94                         $this->getTemplateInsertions( $existing ) );
  95
  96                 # Refresh links of all pages including this page
  97                 if ( $this->mRecursive ) {
  98                         $tlto = $this->mTitle->getTemplateLinksTo();
  99                         if ( count( $tlto ) ) {
 100                                 require_once( 'JobQueue.php' );
 101                                 Job::queueLinksJobs( $tlto );
 102                         }
 103                 }
 104
 105                 # Category links
 106                 $existing = $this->getExistingCategories();
 107                 $this->incrTableUpdate( 'categorylinks', 'cl', $this->getCategoryDeletions( $existing ),
 108                         $this->getCategoryInsertions( $existing ) );
 109
 110                 # I think this works out to a set XOR operation, the idea is to invalidate all
 111                 # categories which were added, deleted or changed
 112                 # FIXME: surely there's a more appropriate place to put this update?
 113                 $categoryUpdates = array_diff_assoc( $existing, $this->mCategories ) + array_diff_assoc( $this->mCategories, $existing );
 114                 $this->invalidateCategories( $categoryUpdates );
 115
 116                 wfProfileOut( $fname );
 117         }
 118
 119         /**
 120           * Link update which clears the previous entries and inserts new ones
 121           * May be slower or faster depending on level of lock contention and write speed of DB
 122           * Also useful where link table corruption needs to be repaired, e.g. in refreshLinks.php
 123          */
 124         function doDumbUpdate() {
 125                 $fname = 'LinksUpdate::doDumbUpdate';
 126                 wfProfileIn( $fname );
 127
 128                 # Refresh category pages
 129                 $existing = $this->getExistingCategories();
 130                 $categoryUpdates = array_diff_assoc( $existing, $this->mCategories ) + array_diff_assoc( $this->mCategories, $existing );
 131
 132                 # Refresh links of all pages including this page
 133                 if ( $this->mRecursive ) {
 134                         $tlto = $this->mTitle->getTemplateLinksTo();
 135                         if ( count( $tlto ) ) {
 136                                 require_once( 'JobQueue.php' );
 137                                 Job::queueLinksJobs( $tlto );
 138                         }
 139                 }
 140
 141                 $this->dumbTableUpdate( 'pagelinks',     $this->getLinkInsertions(),     'pl_from' );
 142                 $this->dumbTableUpdate( 'imagelinks',    $this->getImageInsertions(),    'il_from' );
 143                 $this->dumbTableUpdate( 'categorylinks', $this->getCategoryInsertions(), 'cl_from' );
 144                 $this->dumbTableUpdate( 'templatelinks', $this->getTemplateInsertions(), 'tl_from' );
 145                 $this->dumbTableUpdate( 'externallinks', $this->getExternalInsertions(), 'el_from' );
 146
 147                 # Update the cache of all the category pages
 148                 $this->invalidateCategories( $categoryUpdates );
 149
 150                 wfProfileOut( $fname );
 151         }
 152
 153         function invalidateCategories( $cats ) {
 154                 $fname = 'LinksUpdate::invalidateCategories';
 155                 if ( count( $cats ) ) {
 156                         $this->mDb->update( 'page', array( 'page_touched' => $this->mDb->timestamp() ),
 157                                 array(
 158                                         'page_namespace' => NS_CATEGORY,
 159                                         'page_title IN (' . $this->mDb->makeList( array_keys( $cats ) ) . ')'
 160                                 ), $fname
 161                         );
 162                 }
 163         }
 164
 165         function dumbTableUpdate( $table, $insertions, $fromField ) {
 166                 $fname = 'LinksUpdate::dumbTableUpdate';
 167                 $this->mDb->delete( $table, array( $fromField => $this->mId ), $fname );
 168                 if ( count( $insertions ) ) {
 169                         # The link array was constructed without FOR UPDATE, so there may be collisions
 170                         # Ignoring for now, I'm not sure if that causes problems or not, but I'm fairly
 171                         # sure it's better than without IGNORE
 172                         $this->mDb->insert( $table, $insertions, $fname, array( 'IGNORE' ) );
 173                 }
 174         }
 175
 176         /**
 177          * Make a WHERE clause from a 2-d NS/dbkey array
 178          *
 179          * @param array $arr 2-d array indexed by namespace and DB key
 180          * @param string $prefix Field name prefix, without the underscore
 181          */
 182         function makeWhereFrom2d( &$arr, $prefix ) {
 183                 $lb = new LinkBatch;
 184                 $lb->setArray( $arr );
 185                 return $lb->constructSet( $prefix, $this->mDb );
 186         }
 187
 188         /**
 189          * Update a table by doing a delete query then an insert query
 190          * @access private
 191          */
 192         function incrTableUpdate( $table, $prefix, $deletions, $insertions ) {
 193                 $fname = 'LinksUpdate::incrTableUpdate';
 194                 $where = array( "{$prefix}_from" => $this->mId );
 195                 if ( $table == 'pagelinks' || $table == 'templatelinks' ) {
 196                         $clause = $this->makeWhereFrom2d( $deletions, $prefix );
 197                         if ( $clause ) {
 198                                 $where[] = $clause;
 199                         } else {
 200                                 $where = false;
 201                         }
 202                 } else {
 203                         if ( count( $deletions ) ) {
 204                                 $where[] = "{$prefix}_to IN (" . $this->mDb->makeList( array_keys( $deletions ) ) . ')';
 205                         } else {
 206                                 $where = false;
 207                         }
 208                 }
 209                 if ( $where ) {
 210                         $this->mDb->delete( $table, $where, $fname );
 211                 }
 212                 if ( count( $insertions ) ) {
 213                         $this->mDb->insert( $table, $insertions, $fname, 'IGNORE' );
 214                 }
 215         }
 216
 217
 218         /**
 219          * Get an array of pagelinks insertions for passing to the DB
 220          * Skips the titles specified by the 2-D array $existing
 221          * @access private
 222          */
 223         function getLinkInsertions( $existing = array() ) {
 224                 $arr = array();
 225                 foreach( $this->mLinks as $ns => $dbkeys ) {
 226                         # array_diff_key() was introduced in PHP 5.1, there is a compatibility function
 227                         # in GlobalFunctions.php
 228                         $diffs = isset( $existing[$ns] ) ? array_diff_key( $dbkeys, $existing[$ns] ) : $dbkeys;
 229                         foreach ( $diffs as $dbk => $id ) {
 230                                 $arr[] = array(
 231                                         'pl_from'      => $this->mId,
 232                                         'pl_namespace' => $ns,
 233                                         'pl_title'     => $dbk
 234                                 );
 235                         }
 236                 }
 237                 return $arr;
 238         }
 239
 240         /**
 241          * Get an array of template insertions. Like getLinkInsertions()
 242          * @access private
 243          */
 244         function getTemplateInsertions( $existing = array() ) {
 245                 $arr = array();
 246                 foreach( $this->mTemplates as $ns => $dbkeys ) {
 247                         $diffs = isset( $existing[$ns] ) ? array_diff_key( $dbkeys, $existing[$ns] ) : $dbkeys;
 248                         foreach ( $diffs as $dbk => $id ) {
 249                                 $arr[] = array(
 250                                         'tl_from'      => $this->mId,
 251                                         'tl_namespace' => $ns,
 252                                         'tl_title'     => $dbk
 253                                 );
 254                         }
 255                 }
 256                 return $arr;
 257         }
 258
 259         /**
 260          * Get an array of image insertions
 261          * Skips the names specified in $existing
 262          * @access private
 263          */
 264         function getImageInsertions( $existing = array() ) {
 265                 $arr = array();
 266                 $diffs = array_diff_key( $this->mImages, $existing );
 267                 foreach( $diffs as $iname => $dummy ) {
 268                         $arr[] = array(
 269                                 'il_from' => $this->mId,
 270                                 'il_to'   => $iname
 271                         );
 272                 }
 273                 return $arr;
 274         }
 275
 276         /**
 277          * Get an array of externallinks insertions. Skips the names specified in $existing
 278          * @access private
 279          */
 280         function getExternalInsertions( $existing = array() ) {
 281                 $arr = array();
 282                 $diffs = array_diff_key( $this->mExternals, $existing );
 283                 foreach( $diffs as $url => $dummy ) {
 284                         $arr[] = array(
 285                                 'el_from'   => $this->mId,
 286                                 'el_to'     => $url,
 287                                 'el_index'  => wfMakeUrlIndex( $url ),
 288                         );
 289                 }
 290                 return $arr;
 291         }
 292
 293         /**
 294          * Get an array of category insertions
 295          * @param array $existing Array mapping existing category names to sort keys. If both
 296          * match a link in $this, the link will be omitted from the output
 297          * @access private
 298          */
 299         function getCategoryInsertions( $existing = array() ) {
 300                 $diffs = array_diff_assoc( $this->mCategories, $existing );
 301                 $arr = array();
 302                 foreach ( $diffs as $name => $sortkey ) {
 303                         $arr[] = array(
 304                                 'cl_from'    => $this->mId,
 305                                 'cl_to'      => $name,
 306                                 'cl_sortkey' => $sortkey
 307                         );
 308                 }
 309                 return $arr;
 310         }
 311
 312         /**
 313          * Given an array of existing links, returns those links which are not in $this
 314          * and thus should be deleted.
 315          * @access private
 316          */
 317         function getLinkDeletions( $existing ) {
 318                 $del = array();
 319                 foreach ( $existing as $ns => $dbkeys ) {
 320                         if ( isset( $this->mLinks[$ns] ) ) {
 321                                 $del[$ns] = array_diff_key( $existing[$ns], $this->mLinks[$ns] );
 322                         } else {
 323                                 $del[$ns] = $existing[$ns];
 324                         }
 325                 }
 326                 return $del;
 327         }
 328
 329         /**
 330          * Given an array of existing templates, returns those templates which are not in $this
 331          * and thus should be deleted.
 332          * @access private
 333          */
 334         function getTemplateDeletions( $existing ) {
 335                 $del = array();
 336                 foreach ( $existing as $ns => $dbkeys ) {
 337                         if ( isset( $this->mTemplates[$ns] ) ) {
 338                                 $del[$ns] = array_diff_key( $existing[$ns], $this->mTemplates[$ns] );
 339                         } else {
 340                                 $del[$ns] = $existing[$ns];
 341                         }
 342                 }
 343                 return $del;
 344         }
 345
 346         /**
 347          * Given an array of existing images, returns those images which are not in $this
 348          * and thus should be deleted.
 349          * @access private
 350          */
 351         function getImageDeletions( $existing ) {
 352                 return array_diff_key( $existing, $this->mImages );
 353         }
 354
 355         /**
 356          * Given an array of existing external links, returns those links which are not
 357          * in $this and thus should be deleted.
 358          * @access private
 359          */
 360         function getExternalDeletions( $existing ) {
 361                 return array_diff_key( $existing, $this->mExternals );
 362         }
 363
 364         /**
 365          * Given an array of existing categories, returns those categories which are not in $this
 366          * and thus should be deleted.
 367          * @access private
 368          */
 369         function getCategoryDeletions( $existing ) {
 370                 return array_diff_assoc( $existing, $this->mCategories );
 371         }
 372
 373         /**
 374          * Get an array of existing links, as a 2-D array
 375          * @access private
 376          */
 377         function getExistingLinks() {
 378                 $fname = 'LinksUpdate::getExistingLinks';
 379                 $res = $this->mDb->select( 'pagelinks', array( 'pl_namespace', 'pl_title' ),
 380                         array( 'pl_from' => $this->mId ), $fname, $this->mOptions );
 381                 $arr = array();
 382                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 383                         if ( !isset( $arr[$row->pl_namespace] ) ) {
 384                                 $arr[$row->pl_namespace] = array();
 385                         }
 386                         $arr[$row->pl_namespace][$row->pl_title] = 1;
 387                 }
 388                 $this->mDb->freeResult( $res );
 389                 return $arr;
 390         }
 391
 392         /**
 393          * Get an array of existing templates, as a 2-D array
 394          * @access private
 395          */
 396         function getExistingTemplates() {
 397                 $fname = 'LinksUpdate::getExistingTemplates';
 398                 $res = $this->mDb->select( 'templatelinks', array( 'tl_namespace', 'tl_title' ),
 399                         array( 'tl_from' => $this->mId ), $fname, $this->mOptions );
 400                 $arr = array();
 401                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 402                         if ( !isset( $arr[$row->tl_namespace] ) ) {
 403                                 $arr[$row->tl_namespace] = array();
 404                         }
 405                         $arr[$row->tl_namespace][$row->tl_title] = 1;
 406                 }
 407                 $this->mDb->freeResult( $res );
 408                 return $arr;
 409         }
 410
 411         /**
 412          * Get an array of existing images, image names in the keys
 413          * @access private
 414          */
 415         function getExistingImages() {
 416                 $fname = 'LinksUpdate::getExistingImages';
 417                 $res = $this->mDb->select( 'imagelinks', array( 'il_to' ),
 418                         array( 'il_from' => $this->mId ), $fname, $this->mOptions );
 419                 $arr = array();
 420                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 421                         $arr[$row->il_to] = 1;
 422                 }
 423                 $this->mDb->freeResult( $res );
 424                 return $arr;
 425         }
 426
 427         /**
 428          * Get an array of existing external links, URLs in the keys
 429          * @access private
 430          */
 431         function getExistingExternals() {
 432                 $fname = 'LinksUpdate::getExistingExternals';
 433                 $res = $this->mDb->select( 'externallinks', array( 'el_to' ),
 434                         array( 'el_from' => $this->mId ), $fname, $this->mOptions );
 435                 $arr = array();
 436                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 437                         $arr[$row->el_to] = 1;
 438                 }
 439                 $this->mDb->freeResult( $res );
 440                 return $arr;
 441         }
 442
 443         /**
 444          * Get an array of existing categories, with the name in the key and sort key in the value.
 445          * @access private
 446          */
 447         function getExistingCategories() {
 448                 $fname = 'LinksUpdate::getExistingCategories';
 449                 $res = $this->mDb->select( 'categorylinks', array( 'cl_to', 'cl_sortkey' ),
 450                         array( 'cl_from' => $this->mId ), $fname, $this->mOptions );
 451                 $arr = array();
 452                 while ( $row = $this->mDb->fetchObject( $res ) ) {
 453                         $arr[$row->cl_to] = $row->cl_sortkey;
 454                 }
 455                 $this->mDb->freeResult( $res );
 456                 return $arr;
 457         }
 458 }
 459 ?>