Revert "Include short descriptions for extensions bundled in the release"
[lhc/web/wiklou.git] / includes / cache / LocalisationCache.php
1 <?php
2 /**
3 * Cache of the contents of localisation files.
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 */
22
23 define( 'MW_LC_VERSION', 2 );
24
25 /**
26 * Class for caching the contents of localisation files, Messages*.php
27 * and *.i18n.php.
28 *
29 * An instance of this class is available using Language::getLocalisationCache().
30 *
31 * The values retrieved from here are merged, containing items from extension
32 * files, core messages files and the language fallback sequence (e.g. zh-cn ->
33 * zh-hans -> en ). Some common errors are corrected, for example namespace
34 * names with spaces instead of underscores, but heavyweight processing, such
35 * as grammatical transformation, is done by the caller.
36 */
37 class LocalisationCache {
38 /** Configuration associative array */
39 private $conf;
40
41 /**
42 * True if recaching should only be done on an explicit call to recache().
43 * Setting this reduces the overhead of cache freshness checking, which
44 * requires doing a stat() for every extension i18n file.
45 */
46 private $manualRecache = false;
47
48 /**
49 * True to treat all files as expired until they are regenerated by this object.
50 */
51 private $forceRecache = false;
52
53 /**
54 * The cache data. 3-d array, where the first key is the language code,
55 * the second key is the item key e.g. 'messages', and the third key is
56 * an item specific subkey index. Some items are not arrays and so for those
57 * items, there are no subkeys.
58 */
59 protected $data = array();
60
61 /**
62 * The persistent store object. An instance of LCStore.
63 *
64 * @var LCStore
65 */
66 private $store;
67
68 /**
69 * A 2-d associative array, code/key, where presence indicates that the item
70 * is loaded. Value arbitrary.
71 *
72 * For split items, if set, this indicates that all of the subitems have been
73 * loaded.
74 */
75 private $loadedItems = array();
76
77 /**
78 * A 3-d associative array, code/key/subkey, where presence indicates that
79 * the subitem is loaded. Only used for the split items, i.e. messages.
80 */
81 private $loadedSubitems = array();
82
83 /**
84 * An array where presence of a key indicates that that language has been
85 * initialised. Initialisation includes checking for cache expiry and doing
86 * any necessary updates.
87 */
88 private $initialisedLangs = array();
89
90 /**
91 * An array mapping non-existent pseudo-languages to fallback languages. This
92 * is filled by initShallowFallback() when data is requested from a language
93 * that lacks a Messages*.php file.
94 */
95 private $shallowFallbacks = array();
96
97 /**
98 * An array where the keys are codes that have been recached by this instance.
99 */
100 private $recachedLangs = array();
101
102 /**
103 * All item keys
104 */
105 static public $allKeys = array(
106 'fallback', 'namespaceNames', 'bookstoreList',
107 'magicWords', 'messages', 'rtl', 'capitalizeAllNouns', 'digitTransformTable',
108 'separatorTransformTable', 'fallback8bitEncoding', 'linkPrefixExtension',
109 'linkTrail', 'linkPrefixCharset', 'namespaceAliases',
110 'dateFormats', 'datePreferences', 'datePreferenceMigrationMap',
111 'defaultDateFormat', 'extraUserToggles', 'specialPageAliases',
112 'imageFiles', 'preloadedMessages', 'namespaceGenderAliases',
113 'digitGroupingPattern', 'pluralRules', 'pluralRuleTypes', 'compiledPluralRules',
114 );
115
116 /**
117 * Keys for items which consist of associative arrays, which may be merged
118 * by a fallback sequence.
119 */
120 static public $mergeableMapKeys = array( 'messages', 'namespaceNames',
121 'dateFormats', 'imageFiles', 'preloadedMessages'
122 );
123
124 /**
125 * Keys for items which are a numbered array.
126 */
127 static public $mergeableListKeys = array( 'extraUserToggles' );
128
129 /**
130 * Keys for items which contain an array of arrays of equivalent aliases
131 * for each subitem. The aliases may be merged by a fallback sequence.
132 */
133 static public $mergeableAliasListKeys = array( 'specialPageAliases' );
134
135 /**
136 * Keys for items which contain an associative array, and may be merged if
137 * the primary value contains the special array key "inherit". That array
138 * key is removed after the first merge.
139 */
140 static public $optionalMergeKeys = array( 'bookstoreList' );
141
142 /**
143 * Keys for items that are formatted like $magicWords
144 */
145 static public $magicWordKeys = array( 'magicWords' );
146
147 /**
148 * Keys for items where the subitems are stored in the backend separately.
149 */
150 static public $splitKeys = array( 'messages' );
151
152 /**
153 * Keys which are loaded automatically by initLanguage()
154 */
155 static public $preloadedKeys = array( 'dateFormats', 'namespaceNames' );
156
157 /**
158 * Associative array of cached plural rules. The key is the language code,
159 * the value is an array of plural rules for that language.
160 */
161 private $pluralRules = null;
162
163 /**
164 * Associative array of cached plural rule types. The key is the language
165 * code, the value is an array of plural rule types for that language. For
166 * example, $pluralRuleTypes['ar'] = ['zero', 'one', 'two', 'few', 'many'].
167 * The index for each rule type matches the index for the rule in
168 * $pluralRules, thus allowing correlation between the two. The reason we
169 * don't just use the type names as the keys in $pluralRules is because
170 * Language::convertPlural applies the rules based on numeric order (or
171 * explicit numeric parameter), not based on the name of the rule type. For
172 * example, {{plural:count|wordform1|wordform2|wordform3}}, rather than
173 * {{plural:count|one=wordform1|two=wordform2|many=wordform3}}.
174 */
175 private $pluralRuleTypes = null;
176
177 private $mergeableKeys = null;
178
179 /**
180 * Constructor.
181 * For constructor parameters, see the documentation in DefaultSettings.php
182 * for $wgLocalisationCacheConf.
183 *
184 * @param $conf Array
185 * @throws MWException
186 */
187 function __construct( $conf ) {
188 global $wgCacheDirectory;
189
190 $this->conf = $conf;
191 $storeConf = array();
192 if ( !empty( $conf['storeClass'] ) ) {
193 $storeClass = $conf['storeClass'];
194 } else {
195 switch ( $conf['store'] ) {
196 case 'files':
197 case 'file':
198 $storeClass = 'LCStoreCDB';
199 break;
200 case 'db':
201 $storeClass = 'LCStoreDB';
202 break;
203 case 'accel':
204 $storeClass = 'LCStoreAccel';
205 break;
206 case 'detect':
207 $storeClass = $wgCacheDirectory ? 'LCStoreCDB' : 'LCStoreDB';
208 break;
209 default:
210 throw new MWException(
211 'Please set $wgLocalisationCacheConf[\'store\'] to something sensible.' );
212 }
213 }
214
215 wfDebug( get_class( $this ) . ": using store $storeClass\n" );
216 if ( !empty( $conf['storeDirectory'] ) ) {
217 $storeConf['directory'] = $conf['storeDirectory'];
218 }
219
220 $this->store = new $storeClass( $storeConf );
221 foreach ( array( 'manualRecache', 'forceRecache' ) as $var ) {
222 if ( isset( $conf[$var] ) ) {
223 $this->$var = $conf[$var];
224 }
225 }
226 }
227
228 /**
229 * Returns true if the given key is mergeable, that is, if it is an associative
230 * array which can be merged through a fallback sequence.
231 * @param $key
232 * @return bool
233 */
234 public function isMergeableKey( $key ) {
235 if ( $this->mergeableKeys === null ) {
236 $this->mergeableKeys = array_flip( array_merge(
237 self::$mergeableMapKeys,
238 self::$mergeableListKeys,
239 self::$mergeableAliasListKeys,
240 self::$optionalMergeKeys,
241 self::$magicWordKeys
242 ) );
243 }
244
245 return isset( $this->mergeableKeys[$key] );
246 }
247
248 /**
249 * Get a cache item.
250 *
251 * Warning: this may be slow for split items (messages), since it will
252 * need to fetch all of the subitems from the cache individually.
253 * @param $code
254 * @param $key
255 * @return mixed
256 */
257 public function getItem( $code, $key ) {
258 if ( !isset( $this->loadedItems[$code][$key] ) ) {
259 wfProfileIn( __METHOD__ . '-load' );
260 $this->loadItem( $code, $key );
261 wfProfileOut( __METHOD__ . '-load' );
262 }
263
264 if ( $key === 'fallback' && isset( $this->shallowFallbacks[$code] ) ) {
265 return $this->shallowFallbacks[$code];
266 }
267
268 return $this->data[$code][$key];
269 }
270
271 /**
272 * Get a subitem, for instance a single message for a given language.
273 * @param $code
274 * @param $key
275 * @param $subkey
276 * @return null
277 */
278 public function getSubitem( $code, $key, $subkey ) {
279 if ( !isset( $this->loadedSubitems[$code][$key][$subkey] ) &&
280 !isset( $this->loadedItems[$code][$key] )
281 ) {
282 wfProfileIn( __METHOD__ . '-load' );
283 $this->loadSubitem( $code, $key, $subkey );
284 wfProfileOut( __METHOD__ . '-load' );
285 }
286
287 if ( isset( $this->data[$code][$key][$subkey] ) ) {
288 return $this->data[$code][$key][$subkey];
289 } else {
290 return null;
291 }
292 }
293
294 /**
295 * Get the list of subitem keys for a given item.
296 *
297 * This is faster than array_keys($lc->getItem(...)) for the items listed in
298 * self::$splitKeys.
299 *
300 * Will return null if the item is not found, or false if the item is not an
301 * array.
302 * @param $code
303 * @param $key
304 * @return bool|null|string
305 */
306 public function getSubitemList( $code, $key ) {
307 if ( in_array( $key, self::$splitKeys ) ) {
308 return $this->getSubitem( $code, 'list', $key );
309 } else {
310 $item = $this->getItem( $code, $key );
311 if ( is_array( $item ) ) {
312 return array_keys( $item );
313 } else {
314 return false;
315 }
316 }
317 }
318
319 /**
320 * Load an item into the cache.
321 * @param $code
322 * @param $key
323 */
324 protected function loadItem( $code, $key ) {
325 if ( !isset( $this->initialisedLangs[$code] ) ) {
326 $this->initLanguage( $code );
327 }
328
329 // Check to see if initLanguage() loaded it for us
330 if ( isset( $this->loadedItems[$code][$key] ) ) {
331 return;
332 }
333
334 if ( isset( $this->shallowFallbacks[$code] ) ) {
335 $this->loadItem( $this->shallowFallbacks[$code], $key );
336
337 return;
338 }
339
340 if ( in_array( $key, self::$splitKeys ) ) {
341 $subkeyList = $this->getSubitem( $code, 'list', $key );
342 foreach ( $subkeyList as $subkey ) {
343 if ( isset( $this->data[$code][$key][$subkey] ) ) {
344 continue;
345 }
346 $this->data[$code][$key][$subkey] = $this->getSubitem( $code, $key, $subkey );
347 }
348 } else {
349 $this->data[$code][$key] = $this->store->get( $code, $key );
350 }
351
352 $this->loadedItems[$code][$key] = true;
353 }
354
355 /**
356 * Load a subitem into the cache
357 * @param $code
358 * @param $key
359 * @param $subkey
360 */
361 protected function loadSubitem( $code, $key, $subkey ) {
362 if ( !in_array( $key, self::$splitKeys ) ) {
363 $this->loadItem( $code, $key );
364
365 return;
366 }
367
368 if ( !isset( $this->initialisedLangs[$code] ) ) {
369 $this->initLanguage( $code );
370 }
371
372 // Check to see if initLanguage() loaded it for us
373 if ( isset( $this->loadedItems[$code][$key] ) ||
374 isset( $this->loadedSubitems[$code][$key][$subkey] )
375 ) {
376 return;
377 }
378
379 if ( isset( $this->shallowFallbacks[$code] ) ) {
380 $this->loadSubitem( $this->shallowFallbacks[$code], $key, $subkey );
381
382 return;
383 }
384
385 $value = $this->store->get( $code, "$key:$subkey" );
386 $this->data[$code][$key][$subkey] = $value;
387 $this->loadedSubitems[$code][$key][$subkey] = true;
388 }
389
390 /**
391 * Returns true if the cache identified by $code is missing or expired.
392 *
393 * @param string $code
394 *
395 * @return bool
396 */
397 public function isExpired( $code ) {
398 if ( $this->forceRecache && !isset( $this->recachedLangs[$code] ) ) {
399 wfDebug( __METHOD__ . "($code): forced reload\n" );
400
401 return true;
402 }
403
404 $deps = $this->store->get( $code, 'deps' );
405 $keys = $this->store->get( $code, 'list' );
406 $preload = $this->store->get( $code, 'preload' );
407 // Different keys may expire separately, at least in LCStoreAccel
408 if ( $deps === null || $keys === null || $preload === null ) {
409 wfDebug( __METHOD__ . "($code): cache missing, need to make one\n" );
410
411 return true;
412 }
413
414 foreach ( $deps as $dep ) {
415 // Because we're unserializing stuff from cache, we
416 // could receive objects of classes that don't exist
417 // anymore (e.g. uninstalled extensions)
418 // When this happens, always expire the cache
419 if ( !$dep instanceof CacheDependency || $dep->isExpired() ) {
420 wfDebug( __METHOD__ . "($code): cache for $code expired due to " .
421 get_class( $dep ) . "\n" );
422
423 return true;
424 }
425 }
426
427 return false;
428 }
429
430 /**
431 * Initialise a language in this object. Rebuild the cache if necessary.
432 * @param $code
433 * @throws MWException
434 */
435 protected function initLanguage( $code ) {
436 if ( isset( $this->initialisedLangs[$code] ) ) {
437 return;
438 }
439
440 $this->initialisedLangs[$code] = true;
441
442 # If the code is of the wrong form for a Messages*.php file, do a shallow fallback
443 if ( !Language::isValidBuiltInCode( $code ) ) {
444 $this->initShallowFallback( $code, 'en' );
445
446 return;
447 }
448
449 # Recache the data if necessary
450 if ( !$this->manualRecache && $this->isExpired( $code ) ) {
451 if ( file_exists( Language::getMessagesFileName( $code ) ) ) {
452 $this->recache( $code );
453 } elseif ( $code === 'en' ) {
454 throw new MWException( 'MessagesEn.php is missing.' );
455 } else {
456 $this->initShallowFallback( $code, 'en' );
457 }
458
459 return;
460 }
461
462 # Preload some stuff
463 $preload = $this->getItem( $code, 'preload' );
464 if ( $preload === null ) {
465 if ( $this->manualRecache ) {
466 // No Messages*.php file. Do shallow fallback to en.
467 if ( $code === 'en' ) {
468 throw new MWException( 'No localisation cache found for English. ' .
469 'Please run maintenance/rebuildLocalisationCache.php.' );
470 }
471 $this->initShallowFallback( $code, 'en' );
472
473 return;
474 } else {
475 throw new MWException( 'Invalid or missing localisation cache.' );
476 }
477 }
478 $this->data[$code] = $preload;
479 foreach ( $preload as $key => $item ) {
480 if ( in_array( $key, self::$splitKeys ) ) {
481 foreach ( $item as $subkey => $subitem ) {
482 $this->loadedSubitems[$code][$key][$subkey] = true;
483 }
484 } else {
485 $this->loadedItems[$code][$key] = true;
486 }
487 }
488 }
489
490 /**
491 * Create a fallback from one language to another, without creating a
492 * complete persistent cache.
493 * @param $primaryCode
494 * @param $fallbackCode
495 */
496 public function initShallowFallback( $primaryCode, $fallbackCode ) {
497 $this->data[$primaryCode] =& $this->data[$fallbackCode];
498 $this->loadedItems[$primaryCode] =& $this->loadedItems[$fallbackCode];
499 $this->loadedSubitems[$primaryCode] =& $this->loadedSubitems[$fallbackCode];
500 $this->shallowFallbacks[$primaryCode] = $fallbackCode;
501 }
502
503 /**
504 * Read a PHP file containing localisation data.
505 * @param $_fileName
506 * @param $_fileType
507 * @throws MWException
508 * @return array
509 */
510 protected function readPHPFile( $_fileName, $_fileType ) {
511 wfProfileIn( __METHOD__ );
512 // Disable APC caching
513 $_apcEnabled = ini_set( 'apc.cache_by_default', '0' );
514 include $_fileName;
515 ini_set( 'apc.cache_by_default', $_apcEnabled );
516
517 if ( $_fileType == 'core' || $_fileType == 'extension' ) {
518 $data = compact( self::$allKeys );
519 } elseif ( $_fileType == 'aliases' ) {
520 $data = compact( 'aliases' );
521 } else {
522 wfProfileOut( __METHOD__ );
523 throw new MWException( __METHOD__ . ": Invalid file type: $_fileType" );
524 }
525 wfProfileOut( __METHOD__ );
526
527 return $data;
528 }
529
530 /**
531 * Read a JSON file containing localisation messages.
532 * @param string $fileName Name of file to read
533 * @throws MWException if there is a syntax error in the JSON file
534 * @return array with a 'messages' key, or empty array if the file doesn't exist
535 */
536 protected function readJSONFile( $fileName ) {
537 wfProfileIn( __METHOD__ );
538 if ( !is_readable( $fileName ) ) {
539 return array();
540 }
541
542 $json = file_get_contents( $fileName );
543 if ( $json === false ) {
544 return array();
545 }
546 $data = FormatJson::decode( $json, true );
547 if ( $data === null ) {
548 throw new MWException( __METHOD__ . ": Invalid JSON file: $fileName" );
549 }
550 // Remove keys starting with '@', they're reserved for metadata and non-message data
551 foreach ( $data as $key => $unused ) {
552 if ( $key === '' || $key[0] === '@' ) {
553 unset( $data[$key] );
554 }
555 }
556 // The JSON format only supports messages, none of the other variables, so wrap the data
557 return array( 'messages' => $data );
558 }
559
560 /**
561 * Get the compiled plural rules for a given language from the XML files.
562 * @since 1.20
563 */
564 public function getCompiledPluralRules( $code ) {
565 $rules = $this->getPluralRules( $code );
566 if ( $rules === null ) {
567 return null;
568 }
569 try {
570 $compiledRules = CLDRPluralRuleEvaluator::compile( $rules );
571 } catch ( CLDRPluralRuleError $e ) {
572 wfDebugLog( 'l10n', $e->getMessage() . "\n" );
573
574 return array();
575 }
576
577 return $compiledRules;
578 }
579
580 /**
581 * Get the plural rules for a given language from the XML files.
582 * Cached.
583 * @since 1.20
584 */
585 public function getPluralRules( $code ) {
586 if ( $this->pluralRules === null ) {
587 $this->loadPluralFiles();
588 }
589 if ( !isset( $this->pluralRules[$code] ) ) {
590 return null;
591 } else {
592 return $this->pluralRules[$code];
593 }
594 }
595
596 /**
597 * Get the plural rule types for a given language from the XML files.
598 * Cached.
599 * @since 1.22
600 */
601 public function getPluralRuleTypes( $code ) {
602 if ( $this->pluralRuleTypes === null ) {
603 $this->loadPluralFiles();
604 }
605 if ( !isset( $this->pluralRuleTypes[$code] ) ) {
606 return null;
607 } else {
608 return $this->pluralRuleTypes[$code];
609 }
610 }
611
612 /**
613 * Load the plural XML files.
614 */
615 protected function loadPluralFiles() {
616 global $IP;
617 $cldrPlural = "$IP/languages/data/plurals.xml";
618 $mwPlural = "$IP/languages/data/plurals-mediawiki.xml";
619 // Load CLDR plural rules
620 $this->loadPluralFile( $cldrPlural );
621 if ( file_exists( $mwPlural ) ) {
622 // Override or extend
623 $this->loadPluralFile( $mwPlural );
624 }
625 }
626
627 /**
628 * Load a plural XML file with the given filename, compile the relevant
629 * rules, and save the compiled rules in a process-local cache.
630 */
631 protected function loadPluralFile( $fileName ) {
632 $doc = new DOMDocument;
633 $doc->load( $fileName );
634 $rulesets = $doc->getElementsByTagName( "pluralRules" );
635 foreach ( $rulesets as $ruleset ) {
636 $codes = $ruleset->getAttribute( 'locales' );
637 $rules = array();
638 $ruleTypes = array();
639 $ruleElements = $ruleset->getElementsByTagName( "pluralRule" );
640 foreach ( $ruleElements as $elt ) {
641 $ruleType = $elt->getAttribute( 'count' );
642 if ( $ruleType === 'other' ) {
643 // Don't record "other" rules, which have an empty condition
644 continue;
645 }
646 $rules[] = $elt->nodeValue;
647 $ruleTypes[] = $ruleType;
648 }
649 foreach ( explode( ' ', $codes ) as $code ) {
650 $this->pluralRules[$code] = $rules;
651 $this->pluralRuleTypes[$code] = $ruleTypes;
652 }
653 }
654 }
655
656 /**
657 * Read the data from the source files for a given language, and register
658 * the relevant dependencies in the $deps array. If the localisation
659 * exists, the data array is returned, otherwise false is returned.
660 */
661 protected function readSourceFilesAndRegisterDeps( $code, &$deps ) {
662 global $IP;
663 wfProfileIn( __METHOD__ );
664
665 $fileName = Language::getMessagesFileName( $code );
666 if ( !file_exists( $fileName ) ) {
667 wfProfileOut( __METHOD__ );
668
669 return false;
670 }
671
672 $deps[] = new FileDependency( $fileName );
673 $data = $this->readPHPFile( $fileName, 'core' );
674
675 # Load CLDR plural rules for JavaScript
676 $data['pluralRules'] = $this->getPluralRules( $code );
677 # And for PHP
678 $data['compiledPluralRules'] = $this->getCompiledPluralRules( $code );
679 # Load plural rule types
680 $data['pluralRuleTypes'] = $this->getPluralRuleTypes( $code );
681
682 $deps['plurals'] = new FileDependency( "$IP/languages/data/plurals.xml" );
683 $deps['plurals-mw'] = new FileDependency( "$IP/languages/data/plurals-mediawiki.xml" );
684
685 wfProfileOut( __METHOD__ );
686
687 return $data;
688 }
689
690 /**
691 * Merge two localisation values, a primary and a fallback, overwriting the
692 * primary value in place.
693 * @param $key
694 * @param $value
695 * @param $fallbackValue
696 */
697 protected function mergeItem( $key, &$value, $fallbackValue ) {
698 if ( !is_null( $value ) ) {
699 if ( !is_null( $fallbackValue ) ) {
700 if ( in_array( $key, self::$mergeableMapKeys ) ) {
701 $value = $value + $fallbackValue;
702 } elseif ( in_array( $key, self::$mergeableListKeys ) ) {
703 $value = array_unique( array_merge( $fallbackValue, $value ) );
704 } elseif ( in_array( $key, self::$mergeableAliasListKeys ) ) {
705 $value = array_merge_recursive( $value, $fallbackValue );
706 } elseif ( in_array( $key, self::$optionalMergeKeys ) ) {
707 if ( !empty( $value['inherit'] ) ) {
708 $value = array_merge( $fallbackValue, $value );
709 }
710
711 if ( isset( $value['inherit'] ) ) {
712 unset( $value['inherit'] );
713 }
714 } elseif ( in_array( $key, self::$magicWordKeys ) ) {
715 $this->mergeMagicWords( $value, $fallbackValue );
716 }
717 }
718 } else {
719 $value = $fallbackValue;
720 }
721 }
722
723 /**
724 * @param $value
725 * @param $fallbackValue
726 */
727 protected function mergeMagicWords( &$value, $fallbackValue ) {
728 foreach ( $fallbackValue as $magicName => $fallbackInfo ) {
729 if ( !isset( $value[$magicName] ) ) {
730 $value[$magicName] = $fallbackInfo;
731 } else {
732 $oldSynonyms = array_slice( $fallbackInfo, 1 );
733 $newSynonyms = array_slice( $value[$magicName], 1 );
734 $synonyms = array_values( array_unique( array_merge(
735 $newSynonyms, $oldSynonyms ) ) );
736 $value[$magicName] = array_merge( array( $fallbackInfo[0] ), $synonyms );
737 }
738 }
739 }
740
741 /**
742 * Given an array mapping language code to localisation value, such as is
743 * found in extension *.i18n.php files, iterate through a fallback sequence
744 * to merge the given data with an existing primary value.
745 *
746 * Returns true if any data from the extension array was used, false
747 * otherwise.
748 * @param $codeSequence
749 * @param $key
750 * @param $value
751 * @param $fallbackValue
752 * @return bool
753 */
754 protected function mergeExtensionItem( $codeSequence, $key, &$value, $fallbackValue ) {
755 $used = false;
756 foreach ( $codeSequence as $code ) {
757 if ( isset( $fallbackValue[$code] ) ) {
758 $this->mergeItem( $key, $value, $fallbackValue[$code] );
759 $used = true;
760 }
761 }
762
763 return $used;
764 }
765
766 /**
767 * Load localisation data for a given language for both core and extensions
768 * and save it to the persistent cache store and the process cache
769 * @param $code
770 * @throws MWException
771 */
772 public function recache( $code ) {
773 global $wgExtensionMessagesFiles, $wgMessagesDirs;
774 wfProfileIn( __METHOD__ );
775
776 if ( !$code ) {
777 wfProfileOut( __METHOD__ );
778 throw new MWException( "Invalid language code requested" );
779 }
780 $this->recachedLangs[$code] = true;
781
782 # Initial values
783 $initialData = array_combine(
784 self::$allKeys,
785 array_fill( 0, count( self::$allKeys ), null ) );
786 $coreData = $initialData;
787 $deps = array();
788
789 # Load the primary localisation from the source file
790 $data = $this->readSourceFilesAndRegisterDeps( $code, $deps );
791 if ( $data === false ) {
792 wfDebug( __METHOD__ . ": no localisation file for $code, using fallback to en\n" );
793 $coreData['fallback'] = 'en';
794 } else {
795 wfDebug( __METHOD__ . ": got localisation for $code from source\n" );
796
797 # Merge primary localisation
798 foreach ( $data as $key => $value ) {
799 $this->mergeItem( $key, $coreData[$key], $value );
800 }
801 }
802
803 # Fill in the fallback if it's not there already
804 if ( is_null( $coreData['fallback'] ) ) {
805 $coreData['fallback'] = $code === 'en' ? false : 'en';
806 }
807 if ( $coreData['fallback'] === false ) {
808 $coreData['fallbackSequence'] = array();
809 } else {
810 $coreData['fallbackSequence'] = array_map( 'trim', explode( ',', $coreData['fallback'] ) );
811 $len = count( $coreData['fallbackSequence'] );
812
813 # Ensure that the sequence ends at en
814 if ( $coreData['fallbackSequence'][$len - 1] !== 'en' ) {
815 $coreData['fallbackSequence'][] = 'en';
816 }
817
818 # Load the fallback localisation item by item and merge it
819 foreach ( $coreData['fallbackSequence'] as $fbCode ) {
820 # Load the secondary localisation from the source file to
821 # avoid infinite cycles on cyclic fallbacks
822 $fbData = $this->readSourceFilesAndRegisterDeps( $fbCode, $deps );
823 if ( $fbData === false ) {
824 continue;
825 }
826
827 foreach ( self::$allKeys as $key ) {
828 if ( !isset( $fbData[$key] ) ) {
829 continue;
830 }
831
832 if ( is_null( $coreData[$key] ) || $this->isMergeableKey( $key ) ) {
833 $this->mergeItem( $key, $coreData[$key], $fbData[$key] );
834 }
835 }
836 }
837 }
838
839 $codeSequence = array_merge( array( $code ), $coreData['fallbackSequence'] );
840
841 # Load the extension localisations
842 # This is done after the core because we know the fallback sequence now.
843 # But it has a higher precedence for merging so that we can support things
844 # like site-specific message overrides.
845 wfProfileIn( __METHOD__ . '-extensions' );
846 $allData = $initialData;
847 foreach ( $wgMessagesDirs as $dirs ) {
848 foreach ( (array)$dirs as $dir ) {
849 foreach ( $codeSequence as $csCode ) {
850 $fileName = "$dir/$csCode.json";
851 $data = $this->readJSONFile( $fileName );
852
853 foreach ( $data as $key => $item ) {
854 $this->mergeItem( $key, $allData[$key], $item );
855 }
856
857 $deps[] = new FileDependency( $fileName );
858 }
859 }
860 }
861
862 foreach ( $wgExtensionMessagesFiles as $extension => $fileName ) {
863 $data = $this->readPHPFile( $fileName, 'extension' );
864 $used = false;
865
866 foreach ( $data as $key => $item ) {
867 if ( $key === 'messages' && isset( $wgMessagesDirs[$extension] ) ) {
868 # For backwards compatibility, ignore messages from extensions in
869 # $wgExtensionMessagesFiles that are also present in $wgMessagesDirs.
870 # This allows extensions to use both and be backwards compatible.
871 # Variables other than $messages still need to be supported though.
872 continue;
873 }
874 if ( $this->mergeExtensionItem( $codeSequence, $key, $allData[$key], $item ) ) {
875 $used = true;
876 }
877 }
878
879 if ( $used ) {
880 $deps[] = new FileDependency( $fileName );
881 }
882 }
883
884 # Merge core data into extension data
885 foreach ( $coreData as $key => $item ) {
886 $this->mergeItem( $key, $allData[$key], $item );
887 }
888 wfProfileOut( __METHOD__ . '-extensions' );
889
890 # Add cache dependencies for any referenced globals
891 $deps['wgExtensionMessagesFiles'] = new GlobalDependency( 'wgExtensionMessagesFiles' );
892 $deps['wgMessagesDirs'] = new GlobalDependency( 'wgMessagesDirs' );
893 $deps['version'] = new ConstantDependency( 'MW_LC_VERSION' );
894
895 # Add dependencies to the cache entry
896 $allData['deps'] = $deps;
897
898 # Replace spaces with underscores in namespace names
899 $allData['namespaceNames'] = str_replace( ' ', '_', $allData['namespaceNames'] );
900
901 # And do the same for special page aliases. $page is an array.
902 foreach ( $allData['specialPageAliases'] as &$page ) {
903 $page = str_replace( ' ', '_', $page );
904 }
905 # Decouple the reference to prevent accidental damage
906 unset( $page );
907
908 # If there were no plural rules, return an empty array
909 if ( $allData['pluralRules'] === null ) {
910 $allData['pluralRules'] = array();
911 }
912 if ( $allData['compiledPluralRules'] === null ) {
913 $allData['compiledPluralRules'] = array();
914 }
915 # If there were no plural rule types, return an empty array
916 if ( $allData['pluralRuleTypes'] === null ) {
917 $allData['pluralRuleTypes'] = array();
918 }
919
920 # Set the list keys
921 $allData['list'] = array();
922 foreach ( self::$splitKeys as $key ) {
923 $allData['list'][$key] = array_keys( $allData[$key] );
924 }
925 # Run hooks
926 $purgeBlobs = true;
927 wfRunHooks( 'LocalisationCacheRecache', array( $this, $code, &$allData, &$purgeBlobs ) );
928
929 if ( is_null( $allData['namespaceNames'] ) ) {
930 wfProfileOut( __METHOD__ );
931 throw new MWException( __METHOD__ . ': Localisation data failed sanity check! ' .
932 'Check that your languages/messages/MessagesEn.php file is intact.' );
933 }
934
935 # Set the preload key
936 $allData['preload'] = $this->buildPreload( $allData );
937
938 # Save to the process cache and register the items loaded
939 $this->data[$code] = $allData;
940 foreach ( $allData as $key => $item ) {
941 $this->loadedItems[$code][$key] = true;
942 }
943
944 # Save to the persistent cache
945 wfProfileIn( __METHOD__ . '-write' );
946 $this->store->startWrite( $code );
947 foreach ( $allData as $key => $value ) {
948 if ( in_array( $key, self::$splitKeys ) ) {
949 foreach ( $value as $subkey => $subvalue ) {
950 $this->store->set( "$key:$subkey", $subvalue );
951 }
952 } else {
953 $this->store->set( $key, $value );
954 }
955 }
956 $this->store->finishWrite();
957 wfProfileOut( __METHOD__ . '-write' );
958
959 # Clear out the MessageBlobStore
960 # HACK: If using a null (i.e. disabled) storage backend, we
961 # can't write to the MessageBlobStore either
962 if ( $purgeBlobs && !$this->store instanceof LCStoreNull ) {
963 MessageBlobStore::clear();
964 }
965
966 wfProfileOut( __METHOD__ );
967 }
968
969 /**
970 * Build the preload item from the given pre-cache data.
971 *
972 * The preload item will be loaded automatically, improving performance
973 * for the commonly-requested items it contains.
974 * @param $data
975 * @return array
976 */
977 protected function buildPreload( $data ) {
978 $preload = array( 'messages' => array() );
979 foreach ( self::$preloadedKeys as $key ) {
980 $preload[$key] = $data[$key];
981 }
982
983 foreach ( $data['preloadedMessages'] as $subkey ) {
984 if ( isset( $data['messages'][$subkey] ) ) {
985 $subitem = $data['messages'][$subkey];
986 } else {
987 $subitem = null;
988 }
989 $preload['messages'][$subkey] = $subitem;
990 }
991
992 return $preload;
993 }
994
995 /**
996 * Unload the data for a given language from the object cache.
997 * Reduces memory usage.
998 * @param $code
999 */
1000 public function unload( $code ) {
1001 unset( $this->data[$code] );
1002 unset( $this->loadedItems[$code] );
1003 unset( $this->loadedSubitems[$code] );
1004 unset( $this->initialisedLangs[$code] );
1005 unset( $this->shallowFallbacks[$code] );
1006
1007 foreach ( $this->shallowFallbacks as $shallowCode => $fbCode ) {
1008 if ( $fbCode === $code ) {
1009 $this->unload( $shallowCode );
1010 }
1011 }
1012 }
1013
1014 /**
1015 * Unload all data
1016 */
1017 public function unloadAll() {
1018 foreach ( $this->initialisedLangs as $lang => $unused ) {
1019 $this->unload( $lang );
1020 }
1021 }
1022
1023 /**
1024 * Disable the storage backend
1025 */
1026 public function disableBackend() {
1027 $this->store = new LCStoreNull;
1028 $this->manualRecache = false;
1029 }
1030 }
1031
1032 /**
1033 * Interface for the persistence layer of LocalisationCache.
1034 *
1035 * The persistence layer is two-level hierarchical cache. The first level
1036 * is the language, the second level is the item or subitem.
1037 *
1038 * Since the data for a whole language is rebuilt in one operation, it needs
1039 * to have a fast and atomic method for deleting or replacing all of the
1040 * current data for a given language. The interface reflects this bulk update
1041 * operation. Callers writing to the cache must first call startWrite(), then
1042 * will call set() a couple of thousand times, then will call finishWrite()
1043 * to commit the operation. When finishWrite() is called, the cache is
1044 * expected to delete all data previously stored for that language.
1045 *
1046 * The values stored are PHP variables suitable for serialize(). Implementations
1047 * of LCStore are responsible for serializing and unserializing.
1048 */
1049 interface LCStore {
1050 /**
1051 * Get a value.
1052 * @param string $code Language code
1053 * @param string $key Cache key
1054 */
1055 function get( $code, $key );
1056
1057 /**
1058 * Start a write transaction.
1059 * @param string $code Language code
1060 */
1061 function startWrite( $code );
1062
1063 /**
1064 * Finish a write transaction.
1065 */
1066 function finishWrite();
1067
1068 /**
1069 * Set a key to a given value. startWrite() must be called before this
1070 * is called, and finishWrite() must be called afterwards.
1071 * @param string $key
1072 * @param mixed $value
1073 */
1074 function set( $key, $value );
1075 }
1076
1077 /**
1078 * LCStore implementation which uses PHP accelerator to store data.
1079 * This will work if one of XCache, WinCache or APC cacher is configured.
1080 * (See ObjectCache.php)
1081 */
1082 class LCStoreAccel implements LCStore {
1083 private $currentLang;
1084 private $keys;
1085
1086 public function __construct() {
1087 $this->cache = wfGetCache( CACHE_ACCEL );
1088 }
1089
1090 public function get( $code, $key ) {
1091 $k = wfMemcKey( 'l10n', $code, 'k', $key );
1092 $r = $this->cache->get( $k );
1093
1094 return $r === false ? null : $r;
1095 }
1096
1097 public function startWrite( $code ) {
1098 $k = wfMemcKey( 'l10n', $code, 'l' );
1099 $keys = $this->cache->get( $k );
1100 if ( $keys ) {
1101 foreach ( $keys as $k ) {
1102 $this->cache->delete( $k );
1103 }
1104 }
1105 $this->currentLang = $code;
1106 $this->keys = array();
1107 }
1108
1109 public function finishWrite() {
1110 if ( $this->currentLang ) {
1111 $k = wfMemcKey( 'l10n', $this->currentLang, 'l' );
1112 $this->cache->set( $k, array_keys( $this->keys ) );
1113 }
1114 $this->currentLang = null;
1115 $this->keys = array();
1116 }
1117
1118 public function set( $key, $value ) {
1119 if ( $this->currentLang ) {
1120 $k = wfMemcKey( 'l10n', $this->currentLang, 'k', $key );
1121 $this->keys[$k] = true;
1122 $this->cache->set( $k, $value );
1123 }
1124 }
1125 }
1126
1127 /**
1128 * LCStore implementation which uses the standard DB functions to store data.
1129 * This will work on any MediaWiki installation.
1130 */
1131 class LCStoreDB implements LCStore {
1132 private $currentLang;
1133 private $writesDone = false;
1134
1135 /**
1136 * @var DatabaseBase
1137 */
1138 private $dbw;
1139 private $batch;
1140 private $readOnly = false;
1141
1142 public function get( $code, $key ) {
1143 if ( $this->writesDone ) {
1144 $db = wfGetDB( DB_MASTER );
1145 } else {
1146 $db = wfGetDB( DB_SLAVE );
1147 }
1148 $row = $db->selectRow( 'l10n_cache', array( 'lc_value' ),
1149 array( 'lc_lang' => $code, 'lc_key' => $key ), __METHOD__ );
1150 if ( $row ) {
1151 return unserialize( $row->lc_value );
1152 } else {
1153 return null;
1154 }
1155 }
1156
1157 public function startWrite( $code ) {
1158 if ( $this->readOnly ) {
1159 return;
1160 }
1161
1162 if ( !$code ) {
1163 throw new MWException( __METHOD__ . ": Invalid language \"$code\"" );
1164 }
1165
1166 $this->dbw = wfGetDB( DB_MASTER );
1167 try {
1168 $this->dbw->begin( __METHOD__ );
1169 $this->dbw->delete( 'l10n_cache', array( 'lc_lang' => $code ), __METHOD__ );
1170 } catch ( DBQueryError $e ) {
1171 if ( $this->dbw->wasReadOnlyError() ) {
1172 $this->readOnly = true;
1173 $this->dbw->rollback( __METHOD__ );
1174
1175 return;
1176 } else {
1177 throw $e;
1178 }
1179 }
1180
1181 $this->currentLang = $code;
1182 $this->batch = array();
1183 }
1184
1185 public function finishWrite() {
1186 if ( $this->readOnly ) {
1187 return;
1188 }
1189
1190 if ( $this->batch ) {
1191 $this->dbw->insert( 'l10n_cache', $this->batch, __METHOD__ );
1192 }
1193
1194 $this->dbw->commit( __METHOD__ );
1195 $this->currentLang = null;
1196 $this->dbw = null;
1197 $this->batch = array();
1198 $this->writesDone = true;
1199 }
1200
1201 public function set( $key, $value ) {
1202 if ( $this->readOnly ) {
1203 return;
1204 }
1205
1206 if ( is_null( $this->currentLang ) ) {
1207 throw new MWException( __CLASS__ . ': must call startWrite() before calling set()' );
1208 }
1209
1210 $this->batch[] = array(
1211 'lc_lang' => $this->currentLang,
1212 'lc_key' => $key,
1213 'lc_value' => serialize( $value ) );
1214
1215 if ( count( $this->batch ) >= 100 ) {
1216 $this->dbw->insert( 'l10n_cache', $this->batch, __METHOD__ );
1217 $this->batch = array();
1218 }
1219 }
1220 }
1221
1222 /**
1223 * LCStore implementation which stores data as a collection of CDB files in the
1224 * directory given by $wgCacheDirectory. If $wgCacheDirectory is not set, this
1225 * will throw an exception.
1226 *
1227 * Profiling indicates that on Linux, this implementation outperforms MySQL if
1228 * the directory is on a local filesystem and there is ample kernel cache
1229 * space. The performance advantage is greater when the DBA extension is
1230 * available than it is with the PHP port.
1231 *
1232 * See Cdb.php and http://cr.yp.to/cdb.html
1233 */
1234 class LCStoreCDB implements LCStore {
1235 /** @var CdbReader[] */
1236 private $readers;
1237
1238 /** @var CdbWriter */
1239 private $writer;
1240
1241 /** @var string Current language code */
1242 private $currentLang;
1243
1244 /** @var bool|string Cache directory. False if not set */
1245 private $directory;
1246
1247 function __construct( $conf = array() ) {
1248 global $wgCacheDirectory;
1249
1250 if ( isset( $conf['directory'] ) ) {
1251 $this->directory = $conf['directory'];
1252 } else {
1253 $this->directory = $wgCacheDirectory;
1254 }
1255 }
1256
1257 public function get( $code, $key ) {
1258 if ( !isset( $this->readers[$code] ) ) {
1259 $fileName = $this->getFileName( $code );
1260
1261 $this->readers[$code] = false;
1262 if ( file_exists( $fileName ) ) {
1263 try {
1264 $this->readers[$code] = CdbReader::open( $fileName );
1265 } catch ( CdbException $e ) {
1266 wfDebug( __METHOD__ . ": unable to open cdb file for reading" );
1267 }
1268 }
1269 }
1270
1271 if ( !$this->readers[$code] ) {
1272 return null;
1273 } else {
1274 $value = false;
1275 try {
1276 $value = $this->readers[$code]->get( $key );
1277 } catch ( CdbException $e ) {
1278 wfDebug( __METHOD__ . ": CdbException caught, error message was "
1279 . $e->getMessage() );
1280 }
1281 if ( $value === false ) {
1282 return null;
1283 }
1284
1285 return unserialize( $value );
1286 }
1287 }
1288
1289 public function startWrite( $code ) {
1290 if ( !file_exists( $this->directory ) ) {
1291 if ( !wfMkdirParents( $this->directory, null, __METHOD__ ) ) {
1292 throw new MWException( "Unable to create the localisation store " .
1293 "directory \"{$this->directory}\"" );
1294 }
1295 }
1296
1297 // Close reader to stop permission errors on write
1298 if ( !empty( $this->readers[$code] ) ) {
1299 $this->readers[$code]->close();
1300 }
1301
1302 try {
1303 $this->writer = CdbWriter::open( $this->getFileName( $code ) );
1304 } catch ( CdbException $e ) {
1305 throw new MWException( $e->getMessage() );
1306 }
1307 $this->currentLang = $code;
1308 }
1309
1310 public function finishWrite() {
1311 // Close the writer
1312 try {
1313 $this->writer->close();
1314 } catch ( CdbException $e ) {
1315 throw new MWException( $e->getMessage() );
1316 }
1317 $this->writer = null;
1318 unset( $this->readers[$this->currentLang] );
1319 $this->currentLang = null;
1320 }
1321
1322 public function set( $key, $value ) {
1323 if ( is_null( $this->writer ) ) {
1324 throw new MWException( __CLASS__ . ': must call startWrite() before calling set()' );
1325 }
1326 try {
1327 $this->writer->set( $key, serialize( $value ) );
1328 } catch ( CdbException $e ) {
1329 throw new MWException( $e->getMessage() );
1330 }
1331 }
1332
1333 protected function getFileName( $code ) {
1334 if ( strval( $code ) === '' || strpos( $code, '/' ) !== false ) {
1335 throw new MWException( __METHOD__ . ": Invalid language \"$code\"" );
1336 }
1337
1338 return "{$this->directory}/l10n_cache-$code.cdb";
1339 }
1340 }
1341
1342 /**
1343 * Null store backend, used to avoid DB errors during install
1344 */
1345 class LCStoreNull implements LCStore {
1346 public function get( $code, $key ) {
1347 return null;
1348 }
1349
1350 public function startWrite( $code ) {
1351 }
1352
1353 public function finishWrite() {
1354 }
1355
1356 public function set( $key, $value ) {
1357 }
1358 }
1359
1360 /**
1361 * A localisation cache optimised for loading large amounts of data for many
1362 * languages. Used by rebuildLocalisationCache.php.
1363 */
1364 class LocalisationCacheBulkLoad extends LocalisationCache {
1365 /**
1366 * A cache of the contents of data files.
1367 * Core files are serialized to avoid using ~1GB of RAM during a recache.
1368 */
1369 private $fileCache = array();
1370
1371 /**
1372 * Most recently used languages. Uses the linked-list aspect of PHP hashtables
1373 * to keep the most recently used language codes at the end of the array, and
1374 * the language codes that are ready to be deleted at the beginning.
1375 */
1376 private $mruLangs = array();
1377
1378 /**
1379 * Maximum number of languages that may be loaded into $this->data
1380 */
1381 private $maxLoadedLangs = 10;
1382
1383 /**
1384 * @param $fileName
1385 * @param $fileType
1386 * @return array|mixed
1387 */
1388 protected function readPHPFile( $fileName, $fileType ) {
1389 $serialize = $fileType === 'core';
1390 if ( !isset( $this->fileCache[$fileName][$fileType] ) ) {
1391 $data = parent::readPHPFile( $fileName, $fileType );
1392
1393 if ( $serialize ) {
1394 $encData = serialize( $data );
1395 } else {
1396 $encData = $data;
1397 }
1398
1399 $this->fileCache[$fileName][$fileType] = $encData;
1400
1401 return $data;
1402 } elseif ( $serialize ) {
1403 return unserialize( $this->fileCache[$fileName][$fileType] );
1404 } else {
1405 return $this->fileCache[$fileName][$fileType];
1406 }
1407 }
1408
1409 /**
1410 * @param $code
1411 * @param $key
1412 * @return mixed
1413 */
1414 public function getItem( $code, $key ) {
1415 unset( $this->mruLangs[$code] );
1416 $this->mruLangs[$code] = true;
1417
1418 return parent::getItem( $code, $key );
1419 }
1420
1421 /**
1422 * @param $code
1423 * @param $key
1424 * @param $subkey
1425 * @return
1426 */
1427 public function getSubitem( $code, $key, $subkey ) {
1428 unset( $this->mruLangs[$code] );
1429 $this->mruLangs[$code] = true;
1430
1431 return parent::getSubitem( $code, $key, $subkey );
1432 }
1433
1434 /**
1435 * @param $code
1436 */
1437 public function recache( $code ) {
1438 parent::recache( $code );
1439 unset( $this->mruLangs[$code] );
1440 $this->mruLangs[$code] = true;
1441 $this->trimCache();
1442 }
1443
1444 /**
1445 * @param $code
1446 */
1447 public function unload( $code ) {
1448 unset( $this->mruLangs[$code] );
1449 parent::unload( $code );
1450 }
1451
1452 /**
1453 * Unload cached languages until there are less than $this->maxLoadedLangs
1454 */
1455 protected function trimCache() {
1456 while ( count( $this->data ) > $this->maxLoadedLangs && count( $this->mruLangs ) ) {
1457 reset( $this->mruLangs );
1458 $code = key( $this->mruLangs );
1459 wfDebug( __METHOD__ . ": unloading $code\n" );
1460 $this->unload( $code );
1461 }
1462 }
1463 }