X-Git-Url: http://git.heureux-cyclage.org/?a=blobdiff_plain;f=includes%2FLinksUpdate.php;h=b15f684281e6238133e5d77fd821a7f952dd3f69;hb=81217f55a08341a1b3fc401d836b53b8d97e9d27;hp=3db1577ae6c3266b0de12dd1ee9300b69195097c;hpb=e349b6723c9dc43f63cc2cda107bddc680119fde;p=lhc%2Fweb%2Fwiklou.git diff --git a/includes/LinksUpdate.php b/includes/LinksUpdate.php index 3db1577ae6..b15f684281 100644 --- a/includes/LinksUpdate.php +++ b/includes/LinksUpdate.php @@ -1,6 +1,6 @@ mId = $id; + function LinksUpdate( $title, $parserOutput ) { + global $wgAntiLockFlags; + + if ( $wgAntiLockFlags & ALF_NO_LINK_LOCK ) { + $this->mOptions = array(); + } else { + $this->mOptions = array( 'FOR UPDATE' ); + } + $this->mDb =& wfGetDB( DB_MASTER ); + + if ( !is_object( $title ) ) { + wfDebugDieBacktrace( "The calling convention to LinksUpdate::LinksUpdate() has changed. " . + "Please see Article::editUpdates() for an invocation example.\n" ); + } $this->mTitle = $title; + $this->mId = $title->getArticleID(); + $this->mParserOutput = $parserOutput; + + // Shortcut aliases + $this->mLinks =& $this->mParserOutput->getLinks(); + $this->mImages =& $this->mParserOutput->getImages(); + $this->mTemplates =& $this->mParserOutput->getTemplates(); + $this->mCategories =& $this->mParserOutput->getCategories(); + } /** * Update link tables with outgoing links from an updated article - * Relies on the 'link cache' to be filled out. */ - function doUpdate() { - global $wgUseBetterLinksUpdate, $wgLinkCache, $wgDBtransactions; - global $wgEnablePersistentLC, $wgUseCategoryMagic; + global $wgUseDumbLinkUpdate; + if ( $wgUseDumbLinkUpdate ) { + $this->doDumbUpdate(); + } else { + $this->doIncrementalUpdate(); + } + } - $fname = 'LinksUpdate::doUpdate'; + function doIncrementalUpdate() { + $fname = 'LinksUpdate::doIncrementalUpdate'; wfProfileIn( $fname ); - $del = array(); - $add = array(); + # Page links + $existing = $this->getExistingLinks(); + $this->incrTableUpdate( 'pagelinks', 'pl', $this->getLinkDeletions( $existing ), + $this->getLinkInsertions( $existing ) ); + + # Template links + $existing = $this->getExistingTemplates(); + $this->incrTableUpdate( 'templatelinks', 'tl', $this->getTemplateDeletions( $existing ), + $this->getTemplateInsertions( $existing ) ); - $dbw =& wfGetDB( DB_MASTER ); - $links = $dbw->tableName( 'links' ); - $brokenlinks = $dbw->tableName( 'brokenlinks' ); - $imagelinks = $dbw->tableName( 'imagelinks' ); - $categorylinks = $dbw->tableName( 'categorylinks' ); + # Image links + $existing = $this->getExistingImages(); + $this->incrTableUpdate( 'imagelinks', 'il', $this->getImageDeletions( $existing ), + $this->getImageInsertions( $existing ) ); + + # Category links + $existing = $this->getExistingCategories(); + $this->incrTableUpdate( 'categorylinks', 'cl', $this->getCategoryDeletions( $existing ), + $this->getCategoryInsertions( $existing ) ); - #------------------------------------------------------------------------------ - # Good links + # I think this works out to a set XOR operation, the idea is to invalidate all + # categories which were added, deleted or changed + # FIXME: surely there's a more appropriate place to put this update? + $categoryUpdates = array_diff_assoc( $existing, $this->mCategories ) + array_diff_assoc( $this->mCategories, $existing ); + $this->invalidateCategories( $categoryUpdates ); + + wfProfileOut( $fname ); + } - if ( $wgLinkCache->incrementalSetup( LINKCACHE_GOOD, $del, $add ) ) { - # Delete where necessary - if ( count( $del ) ) { - $sql = "DELETE FROM $links WHERE l_from={$this->mId} AND l_to IN(". - implode( ',', $del ) . ')'; - $dbw->query( $sql, $fname ); - } - } else { - # Delete everything - $dbw->delete( 'links', array( 'l_from' => $this->mId ), $fname ); - - # Get the addition list - $add = $wgLinkCache->getGoodLinks(); + /** + * Link update which clears the previous entries and inserts new ones + * May be slower or faster depending on level of lock contention and write speed of DB + * Also useful where link table corruption needs to be repaired, e.g. in refreshLinks.php + */ + function doDumbUpdate() { + $fname = 'LinksUpdate::doDumbUpdate'; + wfProfileIn( $fname ); + + $existing = $this->getExistingCategories(); + $categoryUpdates = array_diff_assoc( $existing, $this->mCategories ) + array_diff_assoc( $this->mCategories, $existing ); + + $this->dumbTableUpdate( 'pagelinks', $this->getLinkInsertions(), 'pl_from' ); + $this->dumbTableUpdate( 'imagelinks', $this->getImageInsertions(), 'il_from' ); + $this->dumbTableUpdate( 'categorylinks', $this->getCategoryInsertions(), 'cl_from' ); + $this->dumbTableUpdate( 'templatelinks', $this->getTemplateInsertions(), 'tl_from' ); + + # Update the cache of all the category pages + $this->invalidateCategories( $categoryUpdates ); + + wfProfileOut( $fname ); + } + + function invalidateCategories( $cats ) { + $fname = 'LinksUpdate::invalidateCategories'; + if ( count( $cats ) ) { + $this->mDb->update( 'page', array( 'page_touched' => $this->mDb->timestamp() ), + array( + 'page_namespace' => NS_CATEGORY, + 'page_title IN (' . $this->mDb->makeList( array_keys( $cats ) ) . ')' + ), $fname + ); } + } - # Do the insertion - if ( 0 != count( $add ) ) { - $arr=array(); - foreach($add as $lt=>$lid) - array_push($arr,array( - 'l_from'=>$this->mId, - 'l_to'=>$lid)); - # The link cache was constructed without FOR UPDATE, so there may be collisions + function dumbTableUpdate( $table, $insertions, $fromField ) { + $fname = 'LinksUpdate::dumbTableUpdate'; + $this->mDb->delete( $table, array( $fromField => $this->mId ), $fname ); + if ( count( $insertions ) ) { + # The link array was constructed without FOR UPDATE, so there may be collisions # Ignoring for now, I'm not sure if that causes problems or not, but I'm fairly # sure it's better than without IGNORE - $dbw->insertArray('links', $arr, $fname, array('IGNORE')); + $this->mDb->insert( $table, $insertions, $fname, array( 'IGNORE' ) ); } + } - #------------------------------------------------------------------------------ - # Bad links + /** + * Make a WHERE clause from a 2-d NS/dbkey array + * + * @param array $arr 2-d array indexed by namespace and DB key + * @param string $prefix Field name prefix, without the underscore + */ + function makeWhereFrom2d( &$arr, $prefix ) { + $lb = new LinkBatch; + $lb->setArray( $arr ); + return $lb->constructSet( $prefix, $this->mDb ); + } - if ( $wgLinkCache->incrementalSetup( LINKCACHE_BAD, $del, $add ) ) { - # Delete where necessary - if ( count( $del ) ) { - $sql = "DELETE FROM $brokenlinks WHERE bl_from={$this->mId} AND bl_to IN("; - $first = true; - foreach( $del as $badTitle ) { - if ( $first ) { - $first = false; - } else { - $sql .= ','; - } - $sql .= $dbw->addQuotes( $badTitle ); - } - $sql .= ')'; - $dbw->query( $sql, $fname ); + /** + * Update a table by doing a delete query then an insert query + * @private + */ + function incrTableUpdate( $table, $prefix, $deletions, $insertions ) { + $fname = 'LinksUpdate::incrTableUpdate'; + $where = array( "{$prefix}_from" => $this->mId ); + if ( $table == 'pagelinks' || $table == 'templatelinks' ) { + $clause = $this->makeWhereFrom2d( $deletions, $prefix ); + if ( $clause ) { + $where[] = $clause; + } else { + $where = false; } } else { - # Delete all - $dbw->delete( 'brokenlinks', array( 'bl_from' => $this->mId ) ); - - # Get addition list - $add = $wgLinkCache->getBadLinks(); - } - - # Do additions - $sql = ''; - if ( 0 != count ( $add ) ) { - $arr=array(); - foreach( $add as $blt ) { - $blt = $dbw->strencode( $blt ); - array_push($arr,array( - 'bl_from'=>$this->mId, - 'bl_to'=>$blt)); + if ( count( $deletions ) ) { + $where[] = "{$prefix}_to IN (" . $this->mDb->makeList( array_keys( $deletions ) ) . ')'; + } else { + $where = false; } - $dbw->insertArray( 'brokenlinks',$arr,$fname,array('IGNORE')); } + if ( $where ) { + $this->mDb->delete( $table, $where, $fname ); + } + if ( count( $insertions ) ) { + $this->mDb->insert( $table, $insertions, $fname, 'IGNORE' ); + } + } - #------------------------------------------------------------------------------ - # Image links - $sql = "DELETE FROM $imagelinks WHERE il_from='{$this->mId}'"; - $dbw->query( $sql, $fname ); - - # Get addition list - $add = $wgLinkCache->getImageLinks(); - - # Do the insertion - $sql = ''; - $image = Namespace::getImage(); - if ( 0 != count ( $add ) ) { - $arr = array(); - foreach ($add as $iname => $val ) { - $nt = Title::makeTitle( $image, $iname ); - if( !$nt ) continue; - $nt->invalidateCache(); - $iname = $dbw->strencode( $iname ); - array_push($arr,array( - 'il_from'=>$this->mId, - 'il_to'=>$iname)); + + /** + * Get an array of pagelinks insertions for passing to the DB + * Skips the titles specified by the 2-D array $existing + * @private + */ + function getLinkInsertions( $existing = array() ) { + $arr = array(); + foreach( $this->mLinks as $ns => $dbkeys ) { + # array_diff_key() was introduced in PHP 5.1, there is a compatibility function + # in GlobalFunctions.php + $diffs = isset( $existing[$ns] ) ? array_diff_key( $dbkeys, $existing[$ns] ) : $dbkeys; + foreach ( $diffs as $dbk => $id ) { + $arr[] = array( + 'pl_from' => $this->mId, + 'pl_namespace' => $ns, + 'pl_title' => $dbk + ); } - $dbw->insertArray($imagelinks, $arr, $fname, array('IGNORE')); } + return $arr; + } - #------------------------------------------------------------------------------ - # Category links - if( $wgUseCategoryMagic ) { - $sql = "DELETE FROM $categorylinks WHERE cl_from='{$this->mId}'"; - $dbw->query( $sql, $fname ); - - # Get addition list - $add = $wgLinkCache->getCategoryLinks(); - - # Do the insertion - $sql = ''; - if ( 0 != count ( $add ) ) { - $arr=array(); - foreach( $add as $cname => $sortkey ) { - $nt = Title::makeTitle( NS_CATEGORY, $cname ); - if( !$nt ) continue; - $nt->invalidateCache(); - array_push($arr,array( - 'cl_from'=>$this->mId, - 'cl_to'=>$dbw->strencode( $cname ), - 'cl_sortkey'=>$dbw->strencode( $sortkey ))); - } - $dbw->insertArray($categorylinks,$arr,$fname,array('IGNORE')); + /** + * Get an array of template insertions. Like getLinkInsertions() + * @private + */ + function getTemplateInsertions( $existing = array() ) { + $arr = array(); + foreach( $this->mTemplates as $ns => $dbkeys ) { + $diffs = isset( $existing[$ns] ) ? array_diff_key( $dbkeys, $existing[$ns] ) : $dbkeys; + foreach ( $diffs as $dbk => $id ) { + $arr[] = array( + 'tl_from' => $this->mId, + 'tl_namespace' => $ns, + 'tl_title' => $dbk + ); } } - - $this->fixBrokenLinks(); + return $arr; + } - wfProfileOut( $fname ); + /** + * Get an array of image insertions + * Skips the names specified in $existing + * @private + */ + function getImageInsertions( $existing = array() ) { + $arr = array(); + $diffs = array_diff_key( $this->mImages, $existing ); + foreach( $diffs as $iname => $val ) { + $arr[] = array( + 'il_from' => $this->mId, + 'il_to' => $iname + ); + } + return $arr; } /** - * Old inefficient update function - * Used for rebuilding the link table - * @todo Only used by ./maintenance/refreshLinks.inc probably need to be replaced - * @deprecated + * Get an array of category insertions + * @param array $existing Array mapping existing category names to sort keys. If both + * match a link in $this, the link will be omitted from the output + * @private */ - function doDumbUpdate() { - global $wgLinkCache, $wgDBtransactions, $wgUseCategoryMagic; - $fname = 'LinksUpdate::doDumbUpdate'; - wfProfileIn( $fname ); - - - $dbw =& wfGetDB( DB_MASTER ); - $links = $dbw->tableName( 'links' ); - $brokenlinks = $dbw->tableName( 'brokenlinks' ); - $imagelinks = $dbw->tableName( 'imagelinks' ); - $categorylinks = $dbw->tableName( 'categorylinks' ); - - $sql = "DELETE FROM $links WHERE l_from={$this->mId}"; - $dbw->query( $sql, $fname ); + function getCategoryInsertions( $existing = array() ) { + $diffs = array_diff_assoc( $this->mCategories, $existing ); + $arr = array(); + foreach ( $diffs as $name => $sortkey ) { + $arr[] = array( + 'cl_from' => $this->mId, + 'cl_to' => $name, + 'cl_sortkey' => $sortkey + ); + } + return $arr; + } - $a = $wgLinkCache->getGoodLinks(); - if ( 0 != count( $a ) ) { - $arr=array(); - foreach( $a as $lt => $lid ) { - array_push($arr,array( - 'l_from'=>$this->mId, - 'l_to'=>$lid)); + /** + * Given an array of existing links, returns those links which are not in $this + * and thus should be deleted. + * @private + */ + function getLinkDeletions( $existing ) { + $del = array(); + foreach ( $existing as $ns => $dbkeys ) { + if ( isset( $this->mLinks[$ns] ) ) { + $del[$ns] = array_diff_key( $existing[$ns], $this->mLinks[$ns] ); + } else { + $del[$ns] = $existing[$ns]; } - $dbw->insertArray($links,$arr,$fname,array('IGNORE')); } + return $del; + } - $sql = "DELETE FROM $brokenlinks WHERE bl_from={$this->mId}"; - $dbw->query( $sql, $fname ); - - $a = $wgLinkCache->getBadLinks(); - if ( 0 != count ( $a ) ) { - $arr=array(); - foreach( $a as $blt ) { - $blt = $dbw->strencode( $blt ); - array_push($arr,array( - 'bl_from'=>$this->mId, - 'bl_to'=>$blt)); + /** + * Given an array of existing templates, returns those templates which are not in $this + * and thus should be deleted. + * @private + */ + function getTemplateDeletions( $existing ) { + $del = array(); + foreach ( $existing as $ns => $dbkeys ) { + if ( isset( $this->mTemplates[$ns] ) ) { + $del[$ns] = array_diff_key( $existing[$ns], $this->mTemplates[$ns] ); + } else { + $del[$ns] = $existing[$ns]; } - $dbw->insertArray($brokenlinks,$arr,$fname,array('IGNORE')); } - - $sql = "DELETE FROM $imagelinks WHERE il_from={$this->mId}"; - $dbw->query( $sql, $fname ); + return $del; + } - $a = $wgLinkCache->getImageLinks(); - $sql = ''; - if ( 0 != count ( $a ) ) { - $arr=array(); - foreach( $a as $iname => $val ) - array_push($arr,array( - 'il_from'=>$this->mId, - 'il_to'=>$dbw->strencode( $iname ))); - $dbw->insertArray($imagelinks,$arr,$fname,array('IGNORE')); - } + /** + * Given an array of existing images, returns those images which are not in $this + * and thus should be deleted. + * @private + */ + function getImageDeletions( $existing ) { + return array_diff_key( $existing, $this->mImages ); + } + + /** + * Given an array of existing categories, returns those categories which are not in $this + * and thus should be deleted. + * @private + */ + function getCategoryDeletions( $existing ) { + return array_diff_assoc( $existing, $this->mCategories ); + } - if( $wgUseCategoryMagic ) { - $sql = "DELETE FROM $categorylinks WHERE cl_from='{$this->mId}'"; - $dbw->query( $sql, $fname ); - - # Get addition list - $add = $wgLinkCache->getCategoryLinks(); - - # Do the insertion - $sql = ''; - if ( 0 != count ( $add ) ) { - $arr=array(); - foreach( $add as $cname => $sortkey ) { - # FIXME: Change all this to avoid unnecessary duplication - $nt = Title::makeTitle( NS_CATEGORY, $cname ); - if( !$nt ) continue; - $nt->invalidateCache(); - array_push($arr,array( - 'cl_from'=>$this->mId, - 'cl_to'=>$dbw->strencode( $cname ), - 'cl_sortkey'=>$dbw->strencode( $sortkey ))); - } - $dbw->insertArray($categorylinks,$arr,$fname,array('IGNORE')); + /** + * Get an array of existing links, as a 2-D array + * @private + */ + function getExistingLinks() { + $fname = 'LinksUpdate::getExistingLinks'; + $res = $this->mDb->select( 'pagelinks', array( 'pl_namespace', 'pl_title' ), + array( 'pl_from' => $this->mId ), $fname, $this->mOptions ); + $arr = array(); + while ( $row = $this->mDb->fetchObject( $res ) ) { + if ( !isset( $arr[$row->pl_namespace] ) ) { + $arr[$row->pl_namespace] = array(); } + $arr[$row->pl_namespace][$row->pl_title] = 1; } - $this->fixBrokenLinks(); - wfProfileOut( $fname ); + return $arr; } /** - * Update any brokenlinks *to* this page - * Call for a newly created page, or just to make sure state is consistent + * Get an array of existing templates, as a 2-D array + * @private */ - function fixBrokenLinks() { - $fname = 'LinksUpdate::fixBrokenLinks'; + function getExistingTemplates() { + $fname = 'LinksUpdate::getExistingTemplates'; + $res = $this->mDb->select( 'templatelinks', array( 'tl_namespace', 'tl_title' ), + array( 'tl_from' => $this->mId ), $fname, $this->mOptions ); + $arr = array(); + while ( $row = $this->mDb->fetchObject( $res ) ) { + if ( !isset( $arr[$row->tl_namespace] ) ) { + $arr[$row->tl_namespace] = array(); + } + $arr[$row->tl_namespace][$row->tl_title] = 1; + } + return $arr; + } - $dbw =& wfGetDB( DB_MASTER ); - $cur = $dbw->tableName( 'cur' ); - $links = $dbw->tableName( 'links' ); - - $res = $dbw->select( 'brokenlinks', array( 'bl_from' ), array( 'bl_to' => $this->mTitle ), - $fname, 'FOR UPDATE' ); - if ( 0 == $dbw->numRows( $res ) ) { return; } + /** + * Get an array of existing images, image names in the keys + * @private + */ + function getExistingImages() { + $fname = 'LinksUpdate::getExistingImages'; + $res = $this->mDb->select( 'imagelinks', array( 'il_to' ), + array( 'il_from' => $this->mId ), $fname, $this->mOptions ); + $arr = array(); + while ( $row = $this->mDb->fetchObject( $res ) ) { + $arr[$row->il_to] = 1; + } + return $arr; + } - # Ignore errors. If a link existed in both the brokenlinks table and the links - # table, that's an error which can be fixed at this stage by simply ignoring collisions - $arr=array(); - $now = $dbw->timestamp(); - $sql2 = "UPDATE $cur SET cur_touched='{$now}' WHERE cur_id IN ("; - $first = true; - while ( $row = $dbw->fetchObject( $res ) ) { - if ( ! $first ) { $sql2 .= ","; } - $first = false; - array_push($arr,array('l_from'=>$row->bl_from,'l_to'=>$this->mId)); - $sql2 .= $row->bl_from; + /** + * Get an array of existing categories, with the name in the key and sort key in the value. + * @private + */ + function getExistingCategories() { + $fname = 'LinksUpdate::getExistingCategories'; + $res = $this->mDb->select( 'categorylinks', array( 'cl_to', 'cl_sortkey' ), + array( 'cl_from' => $this->mId ), $fname, $this->mOptions ); + $arr = array(); + while ( $row = $this->mDb->fetchObject( $res ) ) { + $arr[$row->cl_to] = $row->cl_sortkey; } - $sql2 .= ')'; - $dbw->insertArray($links,$arr,$fname,array('IGNORE')); - $dbw->query( $sql2, $fname ); - $dbw->delete( 'brokenlinks', array( 'bl_to' => $this->mTitle ), $fname ); + return $arr; } } - ?>