3 * File for BacklinkCache class
8 * Class for fetching backlink lists, approximate backlink counts and
9 * partitions. This is a shared cache.
11 * Instances of this class should typically be fetched with the method
12 * $title->getBacklinkCache().
14 * Ideally you should only get your backlinks from here when you think
15 * there is some advantage in caching them. Otherwise it's just a waste
18 * Introduced by r47317
20 * @internal documentation reviewed on 18 Mar 2011 by hashar
22 * @author Tim Starling
23 * @copyright © 2009, Tim Starling, Domas Mituzas
24 * @copyright © 2010, Max Sem
25 * @copyright © 2011, Ashar Voultoiz
30 * Multi dimensions array representing batches. Keys are:
31 * > (string) links table name
32 * > 'numRows' : Number of rows for this link table
33 * > 'batches' : array( $start, $end )
35 * @see BacklinkCache::partitionResult()
37 * Cleared with BacklinkCache::clear()
39 protected $partitionCache = array();
42 * Contains the whole links from a database result.
43 * This is raw data that will be partitioned in $partitionCache
45 * Initialized with BacklinkCache::getLinks()
46 * Cleared with BacklinkCache::clear()
48 protected $fullResultCache = array();
51 * Local copy of a database object.
53 * Accessor: BacklinkCache::getDB()
54 * Mutator : BacklinkCache::setDB()
55 * Cleared with BacklinkCache::clear()
60 * Local copy of a Title object
64 const CACHE_EXPIRY
= 3600;
67 * Create a new BacklinkCache
68 * @param Title $title : Title object to create a backlink cache for.
70 function __construct( $title ) {
71 $this->title
= $title;
75 * Serialization handler, diasallows to serialize the database to prevent
76 * failures after this class is deserialized from cache with dead DB
80 return array( 'partitionCache', 'fullResultCache', 'title' );
84 * Clear locally stored data and database object.
86 public function clear() {
87 $this->partitionCache
= array();
88 $this->fullResultCache
= array();
93 * Set the Database object to use
95 * @param $db DatabaseBase
97 public function setDB( $db ) {
102 * Get the slave connection to the database
103 * When non existing, will initialize the connection.
104 * @return Database object
106 protected function getDB() {
107 if ( !isset( $this->db
) ) {
108 $this->db
= wfGetDB( DB_SLAVE
);
115 * Get the backlinks for a given table. Cached in process memory only.
116 * @param $table String
117 * @param $startId Integer or false
118 * @param $endId Integer or false
119 * @return TitleArrayFromResult
121 public function getLinks( $table, $startId = false, $endId = false ) {
122 wfProfileIn( __METHOD__
);
124 $fromField = $this->getPrefix( $table ) . '_from';
126 if ( $startId ||
$endId ) {
127 // Partial range, not cached
128 wfDebug( __METHOD__
. ": from DB (uncacheable range)\n" );
129 $conds = $this->getConditions( $table );
131 // Use the from field in the condition rather than the joined page_id,
132 // because databases are stupid and don't necessarily propagate indexes.
134 $conds[] = "$fromField >= " . intval( $startId );
138 $conds[] = "$fromField <= " . intval( $endId );
141 $res = $this->getDB()->select(
142 array( $table, 'page' ),
143 array( 'page_namespace', 'page_title', 'page_id' ),
148 'ORDER BY' => $fromField
150 $ta = TitleArray
::newFromResult( $res );
152 wfProfileOut( __METHOD__
);
156 // @todo FIXME: Make this a function?
157 if ( !isset( $this->fullResultCache
[$table] ) ) {
158 wfDebug( __METHOD__
. ": from DB\n" );
159 $res = $this->getDB()->select(
160 array( $table, 'page' ),
161 array( 'page_namespace', 'page_title', 'page_id' ),
162 $this->getConditions( $table ),
166 'ORDER BY' => $fromField,
168 $this->fullResultCache
[$table] = $res;
171 $ta = TitleArray
::newFromResult( $this->fullResultCache
[$table] );
173 wfProfileOut( __METHOD__
);
178 * Get the distant backtemplatelinks for the table globaltemplatelinks. Cached in process memory only.
179 * @return ResultWrapper list of distant pages that use the local title
181 public function getDistantTemplateLinks( ) {
182 global $wgGlobalDatabase, $wgLocalInterwiki;
184 $dbr = wfGetDB( DB_SLAVE
, array(), $wgGlobalDatabase );
186 array( 'globaltemplatelinks', 'globalinterwiki' ),
187 array( 'gtl_from_wiki', 'gtl_from_page', 'gtl_from_title', 'giw_prefix' ),
188 array( 'gtl_to_prefix' => $wgLocalInterwiki, 'gtl_to_title' => $this->title
->getDBkey( ) ),
191 array( 'gtl_from_wiki = giw_wikiid' )
197 * Get the field name prefix for a given table
198 * @param $table String
200 protected function getPrefix( $table ) {
201 static $prefixes = array(
203 'imagelinks' => 'il',
204 'categorylinks' => 'cl',
205 'templatelinks' => 'tl',
207 'globaltemplatelinks' => 'gtl',
210 if ( isset( $prefixes[$table] ) ) {
211 return $prefixes[$table];
214 wfRunHooks( 'BacklinkCacheGetPrefix', array( $table, &$prefix ) );
218 throw new MWException( "Invalid table \"$table\" in " . __CLASS__
);
224 * Get the SQL condition array for selecting backlinks, with a join
226 * @param $table String
228 protected function getConditions( $table ) {
229 $prefix = $this->getPrefix( $table );
231 // @todo FIXME: imagelinks and categorylinks do not rely on getNamespace,
232 // they could be moved up for nicer case statements
235 case 'templatelinks':
237 "{$prefix}_namespace" => $this->title
->getNamespace(),
238 "{$prefix}_title" => $this->title
->getDBkey(),
239 "page_id={$prefix}_from"
244 "{$prefix}_namespace" => $this->title
->getNamespace(),
245 "{$prefix}_title" => $this->title
->getDBkey(),
246 $this->getDb()->makeList( array(
247 "{$prefix}_interwiki = ''",
248 "{$prefix}_interwiki is null",
250 "page_id={$prefix}_from"
255 'il_to' => $this->title
->getDBkey(),
259 case 'categorylinks':
261 'cl_to' => $this->title
->getDBkey(),
267 wfRunHooks( 'BacklinkCacheGetConditions', array( $table, $this->title
, &$conds ) );
269 throw new MWException( "Invalid table \"$table\" in " . __CLASS__
);
276 * Get the approximate number of backlinks
277 * @param $table String
280 public function getNumLinks( $table ) {
281 if ( isset( $this->fullResultCache
[$table] ) ) {
282 return $this->fullResultCache
[$table]->numRows();
285 if ( isset( $this->partitionCache
[$table] ) ) {
286 $entry = reset( $this->partitionCache
[$table] );
287 return $entry['numRows'];
290 $titleArray = $this->getLinks( $table );
292 return $titleArray->count();
296 * Partition the backlinks into batches.
297 * Returns an array giving the start and end of each range. The first
298 * batch has a start of false, and the last batch has an end of false.
300 * @param $table String: the links table name
301 * @param $batchSize Integer
304 public function partition( $table, $batchSize ) {
306 // 1) try partition cache ...
308 if ( isset( $this->partitionCache
[$table][$batchSize] ) ) {
309 wfDebug( __METHOD__
. ": got from partition cache\n" );
310 return $this->partitionCache
[$table][$batchSize]['batches'];
313 $this->partitionCache
[$table][$batchSize] = false;
314 $cacheEntry =& $this->partitionCache
[$table][$batchSize];
316 // 2) ... then try full result cache ...
318 if ( isset( $this->fullResultCache
[$table] ) ) {
319 $cacheEntry = $this->partitionResult( $this->fullResultCache
[$table], $batchSize );
320 wfDebug( __METHOD__
. ": got from full result cache\n" );
322 return $cacheEntry['batches'];
325 // 3) ... fallback to memcached ...
329 $memcKey = wfMemcKey(
331 md5( $this->title
->getPrefixedDBkey() ),
336 $memcValue = $wgMemc->get( $memcKey );
338 if ( is_array( $memcValue ) ) {
339 $cacheEntry = $memcValue;
340 wfDebug( __METHOD__
. ": got from memcached $memcKey\n" );
342 return $cacheEntry['batches'];
346 // 4) ... finally fetch from the slow database :(
348 $this->getLinks( $table );
349 $cacheEntry = $this->partitionResult( $this->fullResultCache
[$table], $batchSize );
351 $wgMemc->set( $memcKey, $cacheEntry, self
::CACHE_EXPIRY
);
353 wfDebug( __METHOD__
. ": got from database\n" );
354 return $cacheEntry['batches'];
358 * Partition a DB result with backlinks in it into batches
359 * @param $res ResultWrapper database result
360 * @param $batchSize integer
363 protected function partitionResult( $res, $batchSize ) {
365 $numRows = $res->numRows();
366 $numBatches = ceil( $numRows / $batchSize );
368 for ( $i = 0; $i < $numBatches; $i++
) {
372 $rowNum = intval( $numRows * $i / $numBatches );
373 $res->seek( $rowNum );
374 $row = $res->fetchObject();
375 $start = $row->page_id
;
378 if ( $i == $numBatches - 1 ) {
381 $rowNum = intval( $numRows * ( $i +
1 ) / $numBatches );
382 $res->seek( $rowNum );
383 $row = $res->fetchObject();
384 $end = $row->page_id
- 1;
388 if ( $start && $end && $start > $end ) {
389 throw new MWException( __METHOD__
. ': Internal error: query result out of order' );
392 $batches[] = array( $start, $end );
395 return array( 'numRows' => $numRows, 'batches' => $batches );