3 * Cache of the contents of localisation files.
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.
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.
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
23 use Cdb\Exception
as CdbException
;
24 use Cdb\Reader
as CdbReader
;
25 use Cdb\Writer
as CdbWriter
;
28 * Class for caching the contents of localisation files, Messages*.php
31 * An instance of this class is available using Language::getLocalisationCache().
33 * The values retrieved from here are merged, containing items from extension
34 * files, core messages files and the language fallback sequence (e.g. zh-cn ->
35 * zh-hans -> en ). Some common errors are corrected, for example namespace
36 * names with spaces instead of underscores, but heavyweight processing, such
37 * as grammatical transformation, is done by the caller.
39 class LocalisationCache
{
42 /** Configuration associative array */
46 * True if recaching should only be done on an explicit call to recache().
47 * Setting this reduces the overhead of cache freshness checking, which
48 * requires doing a stat() for every extension i18n file.
50 private $manualRecache = false;
53 * True to treat all files as expired until they are regenerated by this object.
55 private $forceRecache = false;
58 * The cache data. 3-d array, where the first key is the language code,
59 * the second key is the item key e.g. 'messages', and the third key is
60 * an item specific subkey index. Some items are not arrays and so for those
61 * items, there are no subkeys.
63 protected $data = array();
66 * The persistent store object. An instance of LCStore.
73 * A 2-d associative array, code/key, where presence indicates that the item
74 * is loaded. Value arbitrary.
76 * For split items, if set, this indicates that all of the subitems have been
79 private $loadedItems = array();
82 * A 3-d associative array, code/key/subkey, where presence indicates that
83 * the subitem is loaded. Only used for the split items, i.e. messages.
85 private $loadedSubitems = array();
88 * An array where presence of a key indicates that that language has been
89 * initialised. Initialisation includes checking for cache expiry and doing
90 * any necessary updates.
92 private $initialisedLangs = array();
95 * An array mapping non-existent pseudo-languages to fallback languages. This
96 * is filled by initShallowFallback() when data is requested from a language
97 * that lacks a Messages*.php file.
99 private $shallowFallbacks = array();
102 * An array where the keys are codes that have been recached by this instance.
104 private $recachedLangs = array();
109 static public $allKeys = array(
110 'fallback', 'namespaceNames', 'bookstoreList',
111 'magicWords', 'messages', 'rtl', 'capitalizeAllNouns', 'digitTransformTable',
112 'separatorTransformTable', 'fallback8bitEncoding', 'linkPrefixExtension',
113 'linkTrail', 'linkPrefixCharset', 'namespaceAliases',
114 'dateFormats', 'datePreferences', 'datePreferenceMigrationMap',
115 'defaultDateFormat', 'extraUserToggles', 'specialPageAliases',
116 'imageFiles', 'preloadedMessages', 'namespaceGenderAliases',
117 'digitGroupingPattern', 'pluralRules', 'pluralRuleTypes', 'compiledPluralRules',
121 * Keys for items which consist of associative arrays, which may be merged
122 * by a fallback sequence.
124 static public $mergeableMapKeys = array( 'messages', 'namespaceNames',
125 'dateFormats', 'imageFiles', 'preloadedMessages'
129 * Keys for items which are a numbered array.
131 static public $mergeableListKeys = array( 'extraUserToggles' );
134 * Keys for items which contain an array of arrays of equivalent aliases
135 * for each subitem. The aliases may be merged by a fallback sequence.
137 static public $mergeableAliasListKeys = array( 'specialPageAliases' );
140 * Keys for items which contain an associative array, and may be merged if
141 * the primary value contains the special array key "inherit". That array
142 * key is removed after the first merge.
144 static public $optionalMergeKeys = array( 'bookstoreList' );
147 * Keys for items that are formatted like $magicWords
149 static public $magicWordKeys = array( 'magicWords' );
152 * Keys for items where the subitems are stored in the backend separately.
154 static public $splitKeys = array( 'messages' );
157 * Keys which are loaded automatically by initLanguage()
159 static public $preloadedKeys = array( 'dateFormats', 'namespaceNames' );
162 * Associative array of cached plural rules. The key is the language code,
163 * the value is an array of plural rules for that language.
165 private $pluralRules = null;
168 * Associative array of cached plural rule types. The key is the language
169 * code, the value is an array of plural rule types for that language. For
170 * example, $pluralRuleTypes['ar'] = ['zero', 'one', 'two', 'few', 'many'].
171 * The index for each rule type matches the index for the rule in
172 * $pluralRules, thus allowing correlation between the two. The reason we
173 * don't just use the type names as the keys in $pluralRules is because
174 * Language::convertPlural applies the rules based on numeric order (or
175 * explicit numeric parameter), not based on the name of the rule type. For
176 * example, {{plural:count|wordform1|wordform2|wordform3}}, rather than
177 * {{plural:count|one=wordform1|two=wordform2|many=wordform3}}.
179 private $pluralRuleTypes = null;
181 private $mergeableKeys = null;
185 * For constructor parameters, see the documentation in DefaultSettings.php
186 * for $wgLocalisationCacheConf.
189 * @throws MWException
191 function __construct( $conf ) {
192 global $wgCacheDirectory;
195 $storeConf = array();
196 if ( !empty( $conf['storeClass'] ) ) {
197 $storeClass = $conf['storeClass'];
199 switch ( $conf['store'] ) {
202 $storeClass = 'LCStoreCDB';
205 $storeClass = 'LCStoreDB';
208 $storeClass = 'LCStoreStaticArray';
211 $storeClass = $wgCacheDirectory ?
'LCStoreCDB' : 'LCStoreDB';
214 throw new MWException(
215 'Please set $wgLocalisationCacheConf[\'store\'] to something sensible.' );
219 wfDebugLog( 'caches', get_class( $this ) . ": using store $storeClass" );
220 if ( !empty( $conf['storeDirectory'] ) ) {
221 $storeConf['directory'] = $conf['storeDirectory'];
224 $this->store
= new $storeClass( $storeConf );
225 foreach ( array( 'manualRecache', 'forceRecache' ) as $var ) {
226 if ( isset( $conf[$var] ) ) {
227 $this->$var = $conf[$var];
233 * Returns true if the given key is mergeable, that is, if it is an associative
234 * array which can be merged through a fallback sequence.
238 public function isMergeableKey( $key ) {
239 if ( $this->mergeableKeys
=== null ) {
240 $this->mergeableKeys
= array_flip( array_merge(
241 self
::$mergeableMapKeys,
242 self
::$mergeableListKeys,
243 self
::$mergeableAliasListKeys,
244 self
::$optionalMergeKeys,
249 return isset( $this->mergeableKeys
[$key] );
255 * Warning: this may be slow for split items (messages), since it will
256 * need to fetch all of the subitems from the cache individually.
257 * @param string $code
261 public function getItem( $code, $key ) {
262 if ( !isset( $this->loadedItems
[$code][$key] ) ) {
263 $this->loadItem( $code, $key );
266 if ( $key === 'fallback' && isset( $this->shallowFallbacks
[$code] ) ) {
267 return $this->shallowFallbacks
[$code];
270 return $this->data
[$code][$key];
274 * Get a subitem, for instance a single message for a given language.
275 * @param string $code
277 * @param string $subkey
280 public function getSubitem( $code, $key, $subkey ) {
281 if ( !isset( $this->loadedSubitems
[$code][$key][$subkey] ) &&
282 !isset( $this->loadedItems
[$code][$key] )
284 $this->loadSubitem( $code, $key, $subkey );
287 if ( isset( $this->data
[$code][$key][$subkey] ) ) {
288 return $this->data
[$code][$key][$subkey];
295 * Get the list of subitem keys for a given item.
297 * This is faster than array_keys($lc->getItem(...)) for the items listed in
300 * Will return null if the item is not found, or false if the item is not an
302 * @param string $code
304 * @return bool|null|string
306 public function getSubitemList( $code, $key ) {
307 if ( in_array( $key, self
::$splitKeys ) ) {
308 return $this->getSubitem( $code, 'list', $key );
310 $item = $this->getItem( $code, $key );
311 if ( is_array( $item ) ) {
312 return array_keys( $item );
320 * Load an item into the cache.
321 * @param string $code
324 protected function loadItem( $code, $key ) {
325 if ( !isset( $this->initialisedLangs
[$code] ) ) {
326 $this->initLanguage( $code );
329 // Check to see if initLanguage() loaded it for us
330 if ( isset( $this->loadedItems
[$code][$key] ) ) {
334 if ( isset( $this->shallowFallbacks
[$code] ) ) {
335 $this->loadItem( $this->shallowFallbacks
[$code], $key );
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] ) ) {
346 $this->data
[$code][$key][$subkey] = $this->getSubitem( $code, $key, $subkey );
349 $this->data
[$code][$key] = $this->store
->get( $code, $key );
352 $this->loadedItems
[$code][$key] = true;
356 * Load a subitem into the cache
357 * @param string $code
359 * @param string $subkey
361 protected function loadSubitem( $code, $key, $subkey ) {
362 if ( !in_array( $key, self
::$splitKeys ) ) {
363 $this->loadItem( $code, $key );
368 if ( !isset( $this->initialisedLangs
[$code] ) ) {
369 $this->initLanguage( $code );
372 // Check to see if initLanguage() loaded it for us
373 if ( isset( $this->loadedItems
[$code][$key] ) ||
374 isset( $this->loadedSubitems
[$code][$key][$subkey] )
379 if ( isset( $this->shallowFallbacks
[$code] ) ) {
380 $this->loadSubitem( $this->shallowFallbacks
[$code], $key, $subkey );
385 $value = $this->store
->get( $code, "$key:$subkey" );
386 $this->data
[$code][$key][$subkey] = $value;
387 $this->loadedSubitems
[$code][$key][$subkey] = true;
391 * Returns true if the cache identified by $code is missing or expired.
393 * @param string $code
397 public function isExpired( $code ) {
398 if ( $this->forceRecache
&& !isset( $this->recachedLangs
[$code] ) ) {
399 wfDebug( __METHOD__
. "($code): forced reload\n" );
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 for some stores
408 if ( $deps === null ||
$keys === null ||
$preload === null ) {
409 wfDebug( __METHOD__
. "($code): cache missing, need to make one\n" );
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" );
431 * Initialise a language in this object. Rebuild the cache if necessary.
432 * @param string $code
433 * @throws MWException
435 protected function initLanguage( $code ) {
436 if ( isset( $this->initialisedLangs
[$code] ) ) {
440 $this->initialisedLangs
[$code] = true;
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( $code, 'en' );
449 # Recache the data if necessary
450 if ( !$this->manualRecache
&& $this->isExpired( $code ) ) {
451 if ( Language
::isSupportedLanguage( $code ) ) {
452 $this->recache( $code );
453 } elseif ( $code === 'en' ) {
454 throw new MWException( 'MessagesEn.php is missing.' );
456 $this->initShallowFallback( $code, 'en' );
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.' );
471 $this->initShallowFallback( $code, 'en' );
475 throw new MWException( 'Invalid or missing localisation cache.' );
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;
485 $this->loadedItems
[$code][$key] = true;
491 * Create a fallback from one language to another, without creating a
492 * complete persistent cache.
493 * @param string $primaryCode
494 * @param string $fallbackCode
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;
504 * Read a PHP file containing localisation data.
505 * @param string $_fileName
506 * @param string $_fileType
507 * @throws MWException
510 protected function readPHPFile( $_fileName, $_fileType ) {
511 // Disable APC caching
512 MediaWiki\
suppressWarnings();
513 $_apcEnabled = ini_set( 'apc.cache_by_default', '0' );
514 MediaWiki\restoreWarnings
();
518 MediaWiki\
suppressWarnings();
519 ini_set( 'apc.cache_by_default', $_apcEnabled );
520 MediaWiki\restoreWarnings
();
522 if ( $_fileType == 'core' ||
$_fileType == 'extension' ) {
523 $data = compact( self
::$allKeys );
524 } elseif ( $_fileType == 'aliases' ) {
525 $data = compact( 'aliases' );
527 throw new MWException( __METHOD__
. ": Invalid file type: $_fileType" );
534 * Read a JSON file containing localisation messages.
535 * @param string $fileName Name of file to read
536 * @throws MWException If there is a syntax error in the JSON file
537 * @return array Array with a 'messages' key, or empty array if the file doesn't exist
539 public function readJSONFile( $fileName ) {
541 if ( !is_readable( $fileName ) ) {
546 $json = file_get_contents( $fileName );
547 if ( $json === false ) {
552 $data = FormatJson
::decode( $json, true );
553 if ( $data === null ) {
555 throw new MWException( __METHOD__
. ": Invalid JSON file: $fileName" );
558 // Remove keys starting with '@', they're reserved for metadata and non-message data
559 foreach ( $data as $key => $unused ) {
560 if ( $key === '' ||
$key[0] === '@' ) {
561 unset( $data[$key] );
565 // The JSON format only supports messages, none of the other variables, so wrap the data
566 return array( 'messages' => $data );
570 * Get the compiled plural rules for a given language from the XML files.
572 * @param string $code
575 public function getCompiledPluralRules( $code ) {
576 $rules = $this->getPluralRules( $code );
577 if ( $rules === null ) {
581 $compiledRules = CLDRPluralRuleEvaluator
::compile( $rules );
582 } catch ( CLDRPluralRuleError
$e ) {
583 wfDebugLog( 'l10n', $e->getMessage() );
588 return $compiledRules;
592 * Get the plural rules for a given language from the XML files.
595 * @param string $code
598 public function getPluralRules( $code ) {
599 if ( $this->pluralRules
=== null ) {
600 $this->loadPluralFiles();
602 if ( !isset( $this->pluralRules
[$code] ) ) {
605 return $this->pluralRules
[$code];
610 * Get the plural rule types for a given language from the XML files.
613 * @param string $code
616 public function getPluralRuleTypes( $code ) {
617 if ( $this->pluralRuleTypes
=== null ) {
618 $this->loadPluralFiles();
620 if ( !isset( $this->pluralRuleTypes
[$code] ) ) {
623 return $this->pluralRuleTypes
[$code];
628 * Load the plural XML files.
630 protected function loadPluralFiles() {
632 $cldrPlural = "$IP/languages/data/plurals.xml";
633 $mwPlural = "$IP/languages/data/plurals-mediawiki.xml";
634 // Load CLDR plural rules
635 $this->loadPluralFile( $cldrPlural );
636 if ( file_exists( $mwPlural ) ) {
637 // Override or extend
638 $this->loadPluralFile( $mwPlural );
643 * Load a plural XML file with the given filename, compile the relevant
644 * rules, and save the compiled rules in a process-local cache.
646 * @param string $fileName
647 * @throws MWException
649 protected function loadPluralFile( $fileName ) {
650 // Use file_get_contents instead of DOMDocument::load (T58439)
651 $xml = file_get_contents( $fileName );
653 throw new MWException( "Unable to read plurals file $fileName" );
655 $doc = new DOMDocument
;
656 $doc->loadXML( $xml );
657 $rulesets = $doc->getElementsByTagName( "pluralRules" );
658 foreach ( $rulesets as $ruleset ) {
659 $codes = $ruleset->getAttribute( 'locales' );
661 $ruleTypes = array();
662 $ruleElements = $ruleset->getElementsByTagName( "pluralRule" );
663 foreach ( $ruleElements as $elt ) {
664 $ruleType = $elt->getAttribute( 'count' );
665 if ( $ruleType === 'other' ) {
666 // Don't record "other" rules, which have an empty condition
669 $rules[] = $elt->nodeValue
;
670 $ruleTypes[] = $ruleType;
672 foreach ( explode( ' ', $codes ) as $code ) {
673 $this->pluralRules
[$code] = $rules;
674 $this->pluralRuleTypes
[$code] = $ruleTypes;
680 * Read the data from the source files for a given language, and register
681 * the relevant dependencies in the $deps array. If the localisation
682 * exists, the data array is returned, otherwise false is returned.
684 * @param string $code
688 protected function readSourceFilesAndRegisterDeps( $code, &$deps ) {
691 // This reads in the PHP i18n file with non-messages l10n data
692 $fileName = Language
::getMessagesFileName( $code );
693 if ( !file_exists( $fileName ) ) {
696 $deps[] = new FileDependency( $fileName );
697 $data = $this->readPHPFile( $fileName, 'core' );
700 # Load CLDR plural rules for JavaScript
701 $data['pluralRules'] = $this->getPluralRules( $code );
703 $data['compiledPluralRules'] = $this->getCompiledPluralRules( $code );
704 # Load plural rule types
705 $data['pluralRuleTypes'] = $this->getPluralRuleTypes( $code );
707 $deps['plurals'] = new FileDependency( "$IP/languages/data/plurals.xml" );
708 $deps['plurals-mw'] = new FileDependency( "$IP/languages/data/plurals-mediawiki.xml" );
714 * Merge two localisation values, a primary and a fallback, overwriting the
715 * primary value in place.
717 * @param mixed $value
718 * @param mixed $fallbackValue
720 protected function mergeItem( $key, &$value, $fallbackValue ) {
721 if ( !is_null( $value ) ) {
722 if ( !is_null( $fallbackValue ) ) {
723 if ( in_array( $key, self
::$mergeableMapKeys ) ) {
724 $value = $value +
$fallbackValue;
725 } elseif ( in_array( $key, self
::$mergeableListKeys ) ) {
726 $value = array_unique( array_merge( $fallbackValue, $value ) );
727 } elseif ( in_array( $key, self
::$mergeableAliasListKeys ) ) {
728 $value = array_merge_recursive( $value, $fallbackValue );
729 } elseif ( in_array( $key, self
::$optionalMergeKeys ) ) {
730 if ( !empty( $value['inherit'] ) ) {
731 $value = array_merge( $fallbackValue, $value );
734 if ( isset( $value['inherit'] ) ) {
735 unset( $value['inherit'] );
737 } elseif ( in_array( $key, self
::$magicWordKeys ) ) {
738 $this->mergeMagicWords( $value, $fallbackValue );
742 $value = $fallbackValue;
747 * @param mixed $value
748 * @param mixed $fallbackValue
750 protected function mergeMagicWords( &$value, $fallbackValue ) {
751 foreach ( $fallbackValue as $magicName => $fallbackInfo ) {
752 if ( !isset( $value[$magicName] ) ) {
753 $value[$magicName] = $fallbackInfo;
755 $oldSynonyms = array_slice( $fallbackInfo, 1 );
756 $newSynonyms = array_slice( $value[$magicName], 1 );
757 $synonyms = array_values( array_unique( array_merge(
758 $newSynonyms, $oldSynonyms ) ) );
759 $value[$magicName] = array_merge( array( $fallbackInfo[0] ), $synonyms );
765 * Given an array mapping language code to localisation value, such as is
766 * found in extension *.i18n.php files, iterate through a fallback sequence
767 * to merge the given data with an existing primary value.
769 * Returns true if any data from the extension array was used, false
771 * @param array $codeSequence
773 * @param mixed $value
774 * @param mixed $fallbackValue
777 protected function mergeExtensionItem( $codeSequence, $key, &$value, $fallbackValue ) {
779 foreach ( $codeSequence as $code ) {
780 if ( isset( $fallbackValue[$code] ) ) {
781 $this->mergeItem( $key, $value, $fallbackValue[$code] );
790 * Gets the combined list of messages dirs from
791 * core and extensions
796 public function getMessagesDirs() {
797 global $wgMessagesDirs, $IP;
799 'core' => "$IP/languages/i18n",
800 'api' => "$IP/includes/api/i18n",
801 'oojs-ui' => "$IP/resources/lib/oojs-ui/i18n",
806 * Load localisation data for a given language for both core and extensions
807 * and save it to the persistent cache store and the process cache
808 * @param string $code
809 * @throws MWException
811 public function recache( $code ) {
812 global $wgExtensionMessagesFiles;
815 throw new MWException( "Invalid language code requested" );
817 $this->recachedLangs
[$code] = true;
820 $initialData = array_combine(
822 array_fill( 0, count( self
::$allKeys ), null ) );
823 $coreData = $initialData;
826 # Load the primary localisation from the source file
827 $data = $this->readSourceFilesAndRegisterDeps( $code, $deps );
828 if ( $data === false ) {
829 wfDebug( __METHOD__
. ": no localisation file for $code, using fallback to en\n" );
830 $coreData['fallback'] = 'en';
832 wfDebug( __METHOD__
. ": got localisation for $code from source\n" );
834 # Merge primary localisation
835 foreach ( $data as $key => $value ) {
836 $this->mergeItem( $key, $coreData[$key], $value );
840 # Fill in the fallback if it's not there already
841 if ( is_null( $coreData['fallback'] ) ) {
842 $coreData['fallback'] = $code === 'en' ?
false : 'en';
844 if ( $coreData['fallback'] === false ) {
845 $coreData['fallbackSequence'] = array();
847 $coreData['fallbackSequence'] = array_map( 'trim', explode( ',', $coreData['fallback'] ) );
848 $len = count( $coreData['fallbackSequence'] );
850 # Ensure that the sequence ends at en
851 if ( $coreData['fallbackSequence'][$len - 1] !== 'en' ) {
852 $coreData['fallbackSequence'][] = 'en';
856 $codeSequence = array_merge( array( $code ), $coreData['fallbackSequence'] );
857 $messageDirs = $this->getMessagesDirs();
859 # Load non-JSON localisation data for extensions
860 $extensionData = array_combine(
862 array_fill( 0, count( $codeSequence ), $initialData ) );
863 foreach ( $wgExtensionMessagesFiles as $extension => $fileName ) {
864 if ( isset( $messageDirs[$extension] ) ) {
865 # This extension has JSON message data; skip the PHP shim
869 $data = $this->readPHPFile( $fileName, 'extension' );
872 foreach ( $data as $key => $item ) {
873 foreach ( $codeSequence as $csCode ) {
874 if ( isset( $item[$csCode] ) ) {
875 $this->mergeItem( $key, $extensionData[$csCode][$key], $item[$csCode] );
882 $deps[] = new FileDependency( $fileName );
886 # Load the localisation data for each fallback, then merge it into the full array
887 $allData = $initialData;
888 foreach ( $codeSequence as $csCode ) {
889 $csData = $initialData;
891 # Load core messages and the extension localisations.
892 foreach ( $messageDirs as $dirs ) {
893 foreach ( (array)$dirs as $dir ) {
894 $fileName = "$dir/$csCode.json";
895 $data = $this->readJSONFile( $fileName );
897 foreach ( $data as $key => $item ) {
898 $this->mergeItem( $key, $csData[$key], $item );
901 $deps[] = new FileDependency( $fileName );
905 # Merge non-JSON extension data
906 if ( isset( $extensionData[$csCode] ) ) {
907 foreach ( $extensionData[$csCode] as $key => $item ) {
908 $this->mergeItem( $key, $csData[$key], $item );
912 if ( $csCode === $code ) {
913 # Merge core data into extension data
914 foreach ( $coreData as $key => $item ) {
915 $this->mergeItem( $key, $csData[$key], $item );
918 # Load the secondary localisation from the source file to
919 # avoid infinite cycles on cyclic fallbacks
920 $fbData = $this->readSourceFilesAndRegisterDeps( $csCode, $deps );
921 if ( $fbData !== false ) {
922 # Only merge the keys that make sense to merge
923 foreach ( self
::$allKeys as $key ) {
924 if ( !isset( $fbData[$key] ) ) {
928 if ( is_null( $coreData[$key] ) ||
$this->isMergeableKey( $key ) ) {
929 $this->mergeItem( $key, $csData[$key], $fbData[$key] );
935 # Allow extensions an opportunity to adjust the data for this
937 Hooks
::run( 'LocalisationCacheRecacheFallback', array( $this, $csCode, &$csData ) );
939 # Merge the data for this fallback into the final array
940 if ( $csCode === $code ) {
943 foreach ( self
::$allKeys as $key ) {
944 if ( !isset( $csData[$key] ) ) {
948 if ( is_null( $allData[$key] ) ||
$this->isMergeableKey( $key ) ) {
949 $this->mergeItem( $key, $allData[$key], $csData[$key] );
955 # Add cache dependencies for any referenced globals
956 $deps['wgExtensionMessagesFiles'] = new GlobalDependency( 'wgExtensionMessagesFiles' );
957 // $wgMessagesDirs is used in LocalisationCache::getMessagesDirs()
958 $deps['wgMessagesDirs'] = new GlobalDependency( 'wgMessagesDirs' );
959 $deps['version'] = new ConstantDependency( 'LocalisationCache::VERSION' );
961 # Add dependencies to the cache entry
962 $allData['deps'] = $deps;
964 # Replace spaces with underscores in namespace names
965 $allData['namespaceNames'] = str_replace( ' ', '_', $allData['namespaceNames'] );
967 # And do the same for special page aliases. $page is an array.
968 foreach ( $allData['specialPageAliases'] as &$page ) {
969 $page = str_replace( ' ', '_', $page );
971 # Decouple the reference to prevent accidental damage
974 # If there were no plural rules, return an empty array
975 if ( $allData['pluralRules'] === null ) {
976 $allData['pluralRules'] = array();
978 if ( $allData['compiledPluralRules'] === null ) {
979 $allData['compiledPluralRules'] = array();
981 # If there were no plural rule types, return an empty array
982 if ( $allData['pluralRuleTypes'] === null ) {
983 $allData['pluralRuleTypes'] = array();
987 $allData['list'] = array();
988 foreach ( self
::$splitKeys as $key ) {
989 $allData['list'][$key] = array_keys( $allData[$key] );
993 Hooks
::run( 'LocalisationCacheRecache', array( $this, $code, &$allData, &$purgeBlobs ) );
995 if ( is_null( $allData['namespaceNames'] ) ) {
996 throw new MWException( __METHOD__
. ': Localisation data failed sanity check! ' .
997 'Check that your languages/messages/MessagesEn.php file is intact.' );
1000 # Set the preload key
1001 $allData['preload'] = $this->buildPreload( $allData );
1003 # Save to the process cache and register the items loaded
1004 $this->data
[$code] = $allData;
1005 foreach ( $allData as $key => $item ) {
1006 $this->loadedItems
[$code][$key] = true;
1009 # Save to the persistent cache
1010 $this->store
->startWrite( $code );
1011 foreach ( $allData as $key => $value ) {
1012 if ( in_array( $key, self
::$splitKeys ) ) {
1013 foreach ( $value as $subkey => $subvalue ) {
1014 $this->store
->set( "$key:$subkey", $subvalue );
1017 $this->store
->set( $key, $value );
1020 $this->store
->finishWrite();
1022 # Clear out the MessageBlobStore
1023 # HACK: If using a null (i.e. disabled) storage backend, we
1024 # can't write to the MessageBlobStore either
1025 if ( $purgeBlobs && !$this->store
instanceof LCStoreNull
) {
1026 $blobStore = new MessageBlobStore();
1027 $blobStore->clear();
1033 * Build the preload item from the given pre-cache data.
1035 * The preload item will be loaded automatically, improving performance
1036 * for the commonly-requested items it contains.
1037 * @param array $data
1040 protected function buildPreload( $data ) {
1041 $preload = array( 'messages' => array() );
1042 foreach ( self
::$preloadedKeys as $key ) {
1043 $preload[$key] = $data[$key];
1046 foreach ( $data['preloadedMessages'] as $subkey ) {
1047 if ( isset( $data['messages'][$subkey] ) ) {
1048 $subitem = $data['messages'][$subkey];
1052 $preload['messages'][$subkey] = $subitem;
1059 * Unload the data for a given language from the object cache.
1060 * Reduces memory usage.
1061 * @param string $code
1063 public function unload( $code ) {
1064 unset( $this->data
[$code] );
1065 unset( $this->loadedItems
[$code] );
1066 unset( $this->loadedSubitems
[$code] );
1067 unset( $this->initialisedLangs
[$code] );
1068 unset( $this->shallowFallbacks
[$code] );
1070 foreach ( $this->shallowFallbacks
as $shallowCode => $fbCode ) {
1071 if ( $fbCode === $code ) {
1072 $this->unload( $shallowCode );
1080 public function unloadAll() {
1081 foreach ( $this->initialisedLangs
as $lang => $unused ) {
1082 $this->unload( $lang );
1087 * Disable the storage backend
1089 public function disableBackend() {
1090 $this->store
= new LCStoreNull
;
1091 $this->manualRecache
= false;
1096 * Interface for the persistence layer of LocalisationCache.
1098 * The persistence layer is two-level hierarchical cache. The first level
1099 * is the language, the second level is the item or subitem.
1101 * Since the data for a whole language is rebuilt in one operation, it needs
1102 * to have a fast and atomic method for deleting or replacing all of the
1103 * current data for a given language. The interface reflects this bulk update
1104 * operation. Callers writing to the cache must first call startWrite(), then
1105 * will call set() a couple of thousand times, then will call finishWrite()
1106 * to commit the operation. When finishWrite() is called, the cache is
1107 * expected to delete all data previously stored for that language.
1109 * The values stored are PHP variables suitable for serialize(). Implementations
1110 * of LCStore are responsible for serializing and unserializing.
1115 * @param string $code Language code
1116 * @param string $key Cache key
1118 function get( $code, $key );
1121 * Start a write transaction.
1122 * @param string $code Language code
1124 function startWrite( $code );
1127 * Finish a write transaction.
1129 function finishWrite();
1132 * Set a key to a given value. startWrite() must be called before this
1133 * is called, and finishWrite() must be called afterwards.
1134 * @param string $key
1135 * @param mixed $value
1137 function set( $key, $value );
1141 * LCStore implementation which uses the standard DB functions to store data.
1142 * This will work on any MediaWiki installation.
1144 class LCStoreDB
implements LCStore
{
1145 private $currentLang;
1146 private $writesDone = false;
1148 /** @var DatabaseBase */
1151 private $batch = array();
1153 private $readOnly = false;
1155 public function get( $code, $key ) {
1156 if ( $this->writesDone
) {
1157 $db = wfGetDB( DB_MASTER
);
1159 $db = wfGetDB( DB_SLAVE
);
1161 $row = $db->selectRow( 'l10n_cache', array( 'lc_value' ),
1162 array( 'lc_lang' => $code, 'lc_key' => $key ), __METHOD__
);
1164 return unserialize( $db->decodeBlob( $row->lc_value
) );
1170 public function startWrite( $code ) {
1171 if ( $this->readOnly
) {
1173 } elseif ( !$code ) {
1174 throw new MWException( __METHOD__
. ": Invalid language \"$code\"" );
1177 $this->dbw
= wfGetDB( DB_MASTER
);
1179 $this->currentLang
= $code;
1180 $this->batch
= array();
1183 public function finishWrite() {
1184 if ( $this->readOnly
) {
1186 } elseif ( is_null( $this->currentLang
) ) {
1187 throw new MWException( __CLASS__
. ': must call startWrite() before finishWrite()' );
1190 $this->dbw
->begin( __METHOD__
);
1192 $this->dbw
->delete( 'l10n_cache',
1193 array( 'lc_lang' => $this->currentLang
), __METHOD__
);
1194 foreach ( array_chunk( $this->batch
, 500 ) as $rows ) {
1195 $this->dbw
->insert( 'l10n_cache', $rows, __METHOD__
);
1197 $this->writesDone
= true;
1198 } catch ( DBQueryError
$e ) {
1199 if ( $this->dbw
->wasReadOnlyError() ) {
1200 $this->readOnly
= true; // just avoid site down time
1205 $this->dbw
->commit( __METHOD__
);
1207 $this->currentLang
= null;
1208 $this->batch
= array();
1211 public function set( $key, $value ) {
1212 if ( $this->readOnly
) {
1214 } elseif ( is_null( $this->currentLang
) ) {
1215 throw new MWException( __CLASS__
. ': must call startWrite() before set()' );
1218 $this->batch
[] = array(
1219 'lc_lang' => $this->currentLang
,
1221 'lc_value' => $this->dbw
->encodeBlob( serialize( $value ) ) );
1226 * LCStore implementation which stores data as a collection of CDB files in the
1227 * directory given by $wgCacheDirectory. If $wgCacheDirectory is not set, this
1228 * will throw an exception.
1230 * Profiling indicates that on Linux, this implementation outperforms MySQL if
1231 * the directory is on a local filesystem and there is ample kernel cache
1232 * space. The performance advantage is greater when the DBA extension is
1233 * available than it is with the PHP port.
1235 * See Cdb.php and http://cr.yp.to/cdb.html
1237 class LCStoreCDB
implements LCStore
{
1238 /** @var CdbReader[] */
1241 /** @var CdbWriter */
1244 /** @var string Current language code */
1245 private $currentLang;
1247 /** @var bool|string Cache directory. False if not set */
1250 function __construct( $conf = array() ) {
1251 global $wgCacheDirectory;
1253 if ( isset( $conf['directory'] ) ) {
1254 $this->directory
= $conf['directory'];
1256 $this->directory
= $wgCacheDirectory;
1260 public function get( $code, $key ) {
1261 if ( !isset( $this->readers
[$code] ) ) {
1262 $fileName = $this->getFileName( $code );
1264 $this->readers
[$code] = false;
1265 if ( file_exists( $fileName ) ) {
1267 $this->readers
[$code] = CdbReader
::open( $fileName );
1268 } catch ( CdbException
$e ) {
1269 wfDebug( __METHOD__
. ": unable to open cdb file for reading\n" );
1274 if ( !$this->readers
[$code] ) {
1279 $value = $this->readers
[$code]->get( $key );
1280 } catch ( CdbException
$e ) {
1281 wfDebug( __METHOD__
. ": CdbException caught, error message was "
1282 . $e->getMessage() . "\n" );
1284 if ( $value === false ) {
1288 return unserialize( $value );
1292 public function startWrite( $code ) {
1293 if ( !file_exists( $this->directory
) ) {
1294 if ( !wfMkdirParents( $this->directory
, null, __METHOD__
) ) {
1295 throw new MWException( "Unable to create the localisation store " .
1296 "directory \"{$this->directory}\"" );
1300 // Close reader to stop permission errors on write
1301 if ( !empty( $this->readers
[$code] ) ) {
1302 $this->readers
[$code]->close();
1306 $this->writer
= CdbWriter
::open( $this->getFileName( $code ) );
1307 } catch ( CdbException
$e ) {
1308 throw new MWException( $e->getMessage() );
1310 $this->currentLang
= $code;
1313 public function finishWrite() {
1316 $this->writer
->close();
1317 } catch ( CdbException
$e ) {
1318 throw new MWException( $e->getMessage() );
1320 $this->writer
= null;
1321 unset( $this->readers
[$this->currentLang
] );
1322 $this->currentLang
= null;
1325 public function set( $key, $value ) {
1326 if ( is_null( $this->writer
) ) {
1327 throw new MWException( __CLASS__
. ': must call startWrite() before calling set()' );
1330 $this->writer
->set( $key, serialize( $value ) );
1331 } catch ( CdbException
$e ) {
1332 throw new MWException( $e->getMessage() );
1336 protected function getFileName( $code ) {
1337 if ( strval( $code ) === '' ||
strpos( $code, '/' ) !== false ) {
1338 throw new MWException( __METHOD__
. ": Invalid language \"$code\"" );
1341 return "{$this->directory}/l10n_cache-$code.cdb";
1346 * Null store backend, used to avoid DB errors during install
1348 class LCStoreNull
implements LCStore
{
1349 public function get( $code, $key ) {
1353 public function startWrite( $code ) {
1356 public function finishWrite() {
1359 public function set( $key, $value ) {
1364 * A localisation cache optimised for loading large amounts of data for many
1365 * languages. Used by rebuildLocalisationCache.php.
1367 class LocalisationCacheBulkLoad
extends LocalisationCache
{
1369 * A cache of the contents of data files.
1370 * Core files are serialized to avoid using ~1GB of RAM during a recache.
1372 private $fileCache = array();
1375 * Most recently used languages. Uses the linked-list aspect of PHP hashtables
1376 * to keep the most recently used language codes at the end of the array, and
1377 * the language codes that are ready to be deleted at the beginning.
1379 private $mruLangs = array();
1382 * Maximum number of languages that may be loaded into $this->data
1384 private $maxLoadedLangs = 10;
1387 * @param string $fileName
1388 * @param string $fileType
1389 * @return array|mixed
1391 protected function readPHPFile( $fileName, $fileType ) {
1392 $serialize = $fileType === 'core';
1393 if ( !isset( $this->fileCache
[$fileName][$fileType] ) ) {
1394 $data = parent
::readPHPFile( $fileName, $fileType );
1397 $encData = serialize( $data );
1402 $this->fileCache
[$fileName][$fileType] = $encData;
1405 } elseif ( $serialize ) {
1406 return unserialize( $this->fileCache
[$fileName][$fileType] );
1408 return $this->fileCache
[$fileName][$fileType];
1413 * @param string $code
1414 * @param string $key
1417 public function getItem( $code, $key ) {
1418 unset( $this->mruLangs
[$code] );
1419 $this->mruLangs
[$code] = true;
1421 return parent
::getItem( $code, $key );
1425 * @param string $code
1426 * @param string $key
1427 * @param string $subkey
1430 public function getSubitem( $code, $key, $subkey ) {
1431 unset( $this->mruLangs
[$code] );
1432 $this->mruLangs
[$code] = true;
1434 return parent
::getSubitem( $code, $key, $subkey );
1438 * @param string $code
1440 public function recache( $code ) {
1441 parent
::recache( $code );
1442 unset( $this->mruLangs
[$code] );
1443 $this->mruLangs
[$code] = true;
1448 * @param string $code
1450 public function unload( $code ) {
1451 unset( $this->mruLangs
[$code] );
1452 parent
::unload( $code );
1456 * Unload cached languages until there are less than $this->maxLoadedLangs
1458 protected function trimCache() {
1459 while ( count( $this->data
) > $this->maxLoadedLangs
&& count( $this->mruLangs
) ) {
1460 reset( $this->mruLangs
);
1461 $code = key( $this->mruLangs
);
1462 wfDebug( __METHOD__
. ": unloading $code\n" );
1463 $this->unload( $code );