1<?php
2/**
3 * Page existence cache.
4 *
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
9 *
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
14 *
15 * You should have received a copy of the GNU General Public License along
16 * with this program; if not, write to the Free Software Foundation, Inc.,
17 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
18 * http://www.gnu.org/copyleft/gpl.html
19 *
20 * @file
21 * @ingroup Cache
22 */
23
24use MediaWiki\Linker\LinkTarget;
25use MediaWiki\MediaWikiServices;
26use Wikimedia\Rdbms\Database;
27use Wikimedia\Rdbms\IDatabase;
28
29/**
30 * Cache for article titles (prefixed DB keys) and ids linked from one source
31 *
32 * @ingroup Cache
33 */
34class LinkCache {
35	/** @var MapCacheLRU */
36	private $goodLinks;
37	/** @var MapCacheLRU */
38	private $badLinks;
39	/** @var WANObjectCache */
40	private $wanCache;
41
42	/** @var bool */
43	private $mForUpdate = false;
44
45	/** @var TitleFormatter */
46	private $titleFormatter;
47
48	/** @var NamespaceInfo */
49	private $nsInfo;
50
51	/**
52	 * How many Titles to store. There are two caches, so the amount actually
53	 * stored in memory can be up to twice this.
54	 */
55	private const MAX_SIZE = 10000;
56
57	public function __construct(
58		TitleFormatter $titleFormatter,
59		WANObjectCache $cache,
60		NamespaceInfo $nsInfo = null
61	) {
62		if ( !$nsInfo ) {
63			wfDeprecated( __METHOD__ . ' with no NamespaceInfo argument', '1.34' );
64			$nsInfo = MediaWikiServices::getInstance()->getNamespaceInfo();
65		}
66		$this->goodLinks = new MapCacheLRU( self::MAX_SIZE );
67		$this->badLinks = new MapCacheLRU( self::MAX_SIZE );
68		$this->wanCache = $cache;
69		$this->titleFormatter = $titleFormatter;
70		$this->nsInfo = $nsInfo;
71	}
72
73	/**
74	 * Get an instance of this class.
75	 *
76	 * @return LinkCache
77	 * @deprecated since 1.28, use MediaWikiServices instead
78	 */
79	public static function singleton() {
80		return MediaWikiServices::getInstance()->getLinkCache();
81	}
82
83	/**
84	 * General accessor to get/set whether the master DB should be used
85	 *
86	 * This used to also set the FOR UPDATE option (locking the rows read
87	 * in order to avoid link table inconsistency), which was later removed
88	 * for performance on wikis with a high edit rate.
89	 *
90	 * @param bool|null $update
91	 * @return bool
92	 * @deprecated Since 1.34
93	 */
94	public function forUpdate( $update = null ) {
95		return wfSetVar( $this->mForUpdate, $update );
96	}
97
98	/**
99	 * @param string $title Prefixed DB key
100	 * @return int Page ID or zero
101	 */
102	public function getGoodLinkID( $title ) {
103		$info = $this->goodLinks->get( $title );
104		if ( !$info ) {
105			return 0;
106		}
107		return $info['id'];
108	}
109
110	/**
111	 * Get a field of a title object from cache.
112	 * If this link is not a cached good title, it will return NULL.
113	 * @param LinkTarget $target
114	 * @param string $field ('length','redirect','revision','model')
115	 * @return string|int|null
116	 */
117	public function getGoodLinkFieldObj( LinkTarget $target, $field ) {
118		$dbkey = $this->titleFormatter->getPrefixedDBkey( $target );
119		$info = $this->goodLinks->get( $dbkey );
120		if ( !$info ) {
121			return null;
122		}
123		return $info[$field];
124	}
125
126	/**
127	 * @param string $title Prefixed DB key
128	 * @return bool
129	 */
130	public function isBadLink( $title ) {
131		// Use get() to ensure it records as used for LRU.
132		return $this->badLinks->has( $title );
133	}
134
135	/**
136	 * Add a link for the title to the link cache
137	 *
138	 * @param int $id Page's ID
139	 * @param LinkTarget $target
140	 * @param int $len Text's length
141	 * @param int|null $redir Whether the page is a redirect
142	 * @param int $revision Latest revision's ID
143	 * @param string|null $model Latest revision's content model ID
144	 * @param string|null $lang Language code of the page, if not the content language
145	 */
146	public function addGoodLinkObj( $id, LinkTarget $target, $len = -1, $redir = null,
147		$revision = 0, $model = null, $lang = null
148	) {
149		$dbkey = $this->titleFormatter->getPrefixedDBkey( $target );
150		$this->goodLinks->set( $dbkey, [
151			'id' => (int)$id,
152			'length' => (int)$len,
153			'redirect' => (int)$redir,
154			'revision' => (int)$revision,
155			'model' => $model ? (string)$model : null,
156			'lang' => $lang ? (string)$lang : null,
157			'restrictions' => null
158		] );
159	}
160
161	/**
162	 * Same as above with better interface.
163	 * @since 1.19
164	 * @param LinkTarget $target
165	 * @param stdClass $row Object which has the fields page_id, page_is_redirect,
166	 *  page_latest and page_content_model
167	 */
168	public function addGoodLinkObjFromRow( LinkTarget $target, $row ) {
169		$dbkey = $this->titleFormatter->getPrefixedDBkey( $target );
170		$this->goodLinks->set( $dbkey, [
171			'id' => intval( $row->page_id ),
172			'length' => intval( $row->page_len ),
173			'redirect' => intval( $row->page_is_redirect ),
174			'revision' => intval( $row->page_latest ),
175			'model' => !empty( $row->page_content_model )
176				? strval( $row->page_content_model )
177				: null,
178			'lang' => !empty( $row->page_lang )
179				? strval( $row->page_lang )
180				: null,
181			'restrictions' => !empty( $row->page_restrictions )
182				? strval( $row->page_restrictions )
183				: null
184		] );
185	}
186
187	/**
188	 * @param LinkTarget $target
189	 */
190	public function addBadLinkObj( LinkTarget $target ) {
191		$dbkey = $this->titleFormatter->getPrefixedDBkey( $target );
192		if ( !$this->isBadLink( $dbkey ) ) {
193			$this->badLinks->set( $dbkey, 1 );
194		}
195	}
196
197	/**
198	 * @param string $title Prefixed DB key
199	 */
200	public function clearBadLink( $title ) {
201		$this->badLinks->clear( $title );
202	}
203
204	/**
205	 * @param LinkTarget $target
206	 */
207	public function clearLink( LinkTarget $target ) {
208		$dbkey = $this->titleFormatter->getPrefixedDBkey( $target );
209		$this->badLinks->clear( $dbkey );
210		$this->goodLinks->clear( $dbkey );
211	}
212
213	/**
214	 * Fields that LinkCache needs to select
215	 *
216	 * @since 1.28
217	 * @return array
218	 */
219	public static function getSelectFields() {
220		global $wgPageLanguageUseDB;
221
222		$fields = [
223			'page_id',
224			'page_len',
225			'page_is_redirect',
226			'page_latest',
227			'page_restrictions',
228			'page_content_model',
229		];
230
231		if ( $wgPageLanguageUseDB ) {
232			$fields[] = 'page_lang';
233		}
234
235		return $fields;
236	}
237
238	/**
239	 * Add a title to the link cache, return the page_id or zero if non-existent
240	 *
241	 * @param LinkTarget $nt LinkTarget object to add
242	 * @return int Page ID or zero
243	 */
244	public function addLinkObj( LinkTarget $nt ) {
245		$key = $this->titleFormatter->getPrefixedDBkey( $nt );
246		if ( $this->isBadLink( $key ) || $nt->isExternal() || $nt->getNamespace() < 0 ) {
247			return 0;
248		}
249		$id = $this->getGoodLinkID( $key );
250		if ( $id != 0 ) {
251			return $id;
252		}
253
254		if ( $key === '' ) {
255			return 0;
256		}
257
258		// Cache template/file pages as they are less often viewed but heavily used
259		if ( $this->mForUpdate ) {
260			$row = $this->fetchPageRow( wfGetDB( DB_MASTER ), $nt );
261		} elseif ( $this->isCacheable( $nt ) ) {
262			// These pages are often transcluded heavily, so cache them
263			$cache = $this->wanCache;
264			$row = $cache->getWithSetCallback(
265				$cache->makeKey( 'page', $nt->getNamespace(), sha1( $nt->getDBkey() ) ),
266				$cache::TTL_DAY,
267				function ( $curValue, &$ttl, array &$setOpts ) use ( $cache, $nt ) {
268					$dbr = wfGetDB( DB_REPLICA );
269					$setOpts += Database::getCacheSetOptions( $dbr );
270
271					$row = $this->fetchPageRow( $dbr, $nt );
272					$mtime = $row ? wfTimestamp( TS_UNIX, $row->page_touched ) : false;
273					$ttl = $cache->adaptiveTTL( $mtime, $ttl );
274
275					return $row;
276				}
277			);
278		} else {
279			$row = $this->fetchPageRow( wfGetDB( DB_REPLICA ), $nt );
280		}
281
282		if ( $row ) {
283			$this->addGoodLinkObjFromRow( $nt, $row );
284			$id = intval( $row->page_id );
285		} else {
286			$this->addBadLinkObj( $nt );
287			$id = 0;
288		}
289
290		return $id;
291	}
292
293	/**
294	 * @param WANObjectCache $cache
295	 * @param LinkTarget $t
296	 * @return string[]
297	 * @since 1.28
298	 */
299	public function getMutableCacheKeys( WANObjectCache $cache, LinkTarget $t ) {
300		if ( $this->isCacheable( $t ) ) {
301			return [ $cache->makeKey( 'page', $t->getNamespace(), sha1( $t->getDBkey() ) ) ];
302		}
303
304		return [];
305	}
306
307	private function isCacheable( LinkTarget $title ) {
308		$ns = $title->getNamespace();
309		if ( in_array( $ns, [ NS_TEMPLATE, NS_FILE, NS_CATEGORY, NS_MEDIAWIKI ] ) ) {
310			return true;
311		}
312		// Focus on transcluded pages more than the main content
313		if ( $this->nsInfo->isContent( $ns ) ) {
314			return false;
315		}
316		// Non-talk extension namespaces (e.g. NS_MODULE)
317		return ( $ns >= 100 && $this->nsInfo->isSubject( $ns ) );
318	}
319
320	private function fetchPageRow( IDatabase $db, LinkTarget $nt ) {
321		$fields = self::getSelectFields();
322		if ( $this->isCacheable( $nt ) ) {
323			$fields[] = 'page_touched';
324		}
325
326		return $db->selectRow(
327			'page',
328			$fields,
329			[ 'page_namespace' => $nt->getNamespace(), 'page_title' => $nt->getDBkey() ],
330			__METHOD__
331		);
332	}
333
334	/**
335	 * Purge the link cache for a title
336	 *
337	 * @param LinkTarget $title
338	 * @since 1.28
339	 */
340	public function invalidateTitle( LinkTarget $title ) {
341		if ( $this->isCacheable( $title ) ) {
342			$cache = $this->wanCache;
343			$cache->delete(
344				$cache->makeKey( 'page', $title->getNamespace(), sha1( $title->getDBkey() ) )
345			);
346		}
347	}
348
349	/**
350	 * Clears cache
351	 */
352	public function clear() {
353		$this->goodLinks->clear();
354		$this->badLinks->clear();
355	}
356}
357