1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
declare( strict_types = 1 ); |
4
|
|
|
|
5
|
|
|
namespace Wikibase\Lib\Store; |
6
|
|
|
|
7
|
|
|
use Psr\SimpleCache\CacheInterface; |
8
|
|
|
use Wikibase\DataAccess\PrefetchingTermLookup; |
9
|
|
|
use Wikibase\DataModel\Entity\EntityId; |
10
|
|
|
use Wikibase\DataModel\Term\TermTypes; |
11
|
|
|
use Wikibase\Lib\ContentLanguages; |
12
|
|
|
|
13
|
|
|
/** |
14
|
|
|
* Prefetches terms from the Cache or via the provided PrefetchingTermLookup if not cached. |
15
|
|
|
* |
16
|
|
|
* Terms requested via the TermLookup methods are also buffered and cached. |
17
|
|
|
* |
18
|
|
|
* CacheInterface determines the medium of caching, and thus the availability (process, server, WAN). |
19
|
|
|
* |
20
|
|
|
* @license GPL-2.0-or-later |
21
|
|
|
*/ |
22
|
|
|
final class CachingPrefetchingTermLookup implements PrefetchingTermLookup { |
23
|
|
|
|
24
|
|
|
use TermCacheKeyBuilder; |
25
|
|
|
|
26
|
|
|
private const DEFAULT_TTL = 60; |
27
|
|
|
private const RESOLVED_KEYS = 'resolvedKeys'; |
28
|
|
|
private const UNRESOLVED_IDS = 'unresolvedIds'; |
29
|
|
|
private const KEY_PARTS_MAP = 'keyPartsMap'; |
30
|
|
|
private const UNCACHED_IDS = 'uncachedIds'; |
31
|
|
|
private const UNCACHED_TERM_TYPES = 'uncachedTermTypes'; |
32
|
|
|
private const UNCACHED_LANGUAGE_CODES = 'uncachedLanguageCodes'; |
33
|
|
|
private const ENTITY_ID = 'entityId'; |
34
|
|
|
private const TERM_TYPE = 'termType'; |
35
|
|
|
private const LANGUAGE_CODE = 'languageCode'; |
36
|
|
|
|
37
|
|
|
/** |
38
|
|
|
* @var int |
39
|
|
|
*/ |
40
|
|
|
private $cacheEntryTTL; |
41
|
|
|
|
42
|
|
|
/** |
43
|
|
|
* @var CacheInterface |
44
|
|
|
*/ |
45
|
|
|
private $cache; |
46
|
|
|
|
47
|
|
|
/** |
48
|
|
|
* @var array |
49
|
|
|
*/ |
50
|
|
|
private $prefetchedTerms; |
51
|
|
|
|
52
|
|
|
/** |
53
|
|
|
* @var PrefetchingTermLookup |
54
|
|
|
*/ |
55
|
|
|
private $lookup; |
56
|
|
|
|
57
|
|
|
/** |
58
|
|
|
* @var RedirectResolvingLatestRevisionLookup |
59
|
|
|
*/ |
60
|
|
|
private $redirectResolvingRevisionLookup; |
61
|
|
|
|
62
|
|
|
/** |
63
|
|
|
* @var ContentLanguages |
64
|
|
|
*/ |
65
|
|
|
private $termLanguages; |
66
|
|
|
|
67
|
|
|
public function __construct( |
68
|
|
|
CacheInterface $cache, |
69
|
|
|
PrefetchingTermLookup $lookup, |
70
|
|
|
RedirectResolvingLatestRevisionLookup $redirectResolvingRevisionLookup, |
71
|
|
|
ContentLanguages $termLanguages, |
72
|
|
|
?int $ttl = null |
73
|
|
|
) { |
74
|
|
|
$this->cache = $cache; |
75
|
|
|
$this->lookup = $lookup; |
76
|
|
|
$this->redirectResolvingRevisionLookup = $redirectResolvingRevisionLookup; |
77
|
|
|
$this->termLanguages = $termLanguages; |
78
|
|
|
$this->cacheEntryTTL = $ttl ?? self::DEFAULT_TTL; |
79
|
|
|
} |
80
|
|
|
|
81
|
|
|
public function prefetchTerms( array $entityIds, array $termTypes, array $languageCodes ): void { |
82
|
|
|
[ |
83
|
|
|
self::UNCACHED_IDS => $uncachedIds, |
|
|
|
|
84
|
|
|
self::UNCACHED_TERM_TYPES => $uncachedTermTypes, |
|
|
|
|
85
|
|
|
self::UNCACHED_LANGUAGE_CODES => $uncachedLanguageCodes, |
|
|
|
|
86
|
|
|
] = $this->prefetchCachedTerms( $entityIds, $termTypes, $this->filterValidTermLanguages( $languageCodes ) ); |
87
|
|
|
|
88
|
|
|
$this->prefetchAndCache( $uncachedIds, $uncachedTermTypes, $uncachedLanguageCodes ); |
|
|
|
|
89
|
|
|
} |
90
|
|
|
|
91
|
|
|
public function getPrefetchedTerm( EntityId $entityId, $termType, $languageCode ) { |
92
|
|
|
if ( isset( $this->prefetchedTerms[$entityId->getSerialization()][$termType][$languageCode] ) ) { |
93
|
|
|
return $this->prefetchedTerms[$entityId->getSerialization()][$termType][$languageCode]; |
94
|
|
|
} |
95
|
|
|
return $this->lookup->getPrefetchedTerm( $entityId, $termType, $languageCode ); |
96
|
|
|
} |
97
|
|
|
|
98
|
|
|
public function getPrefetchedAliases( EntityId $entityId, $languageCode ) { |
99
|
|
|
if ( isset( $this->prefetchedTerms[$entityId->getSerialization()][TermTypes::TYPE_ALIAS][$languageCode] ) ) { |
100
|
|
|
return $this->prefetchedTerms[$entityId->getSerialization()][TermTypes::TYPE_ALIAS][$languageCode]; |
101
|
|
|
} |
102
|
|
|
return $this->lookup->getPrefetchedAliases( $entityId, $languageCode ); |
103
|
|
|
} |
104
|
|
|
|
105
|
|
|
public function getLabel( EntityId $entityId, $languageCode ): ?string { |
106
|
|
|
if ( !$this->termLanguages->hasLanguage( $languageCode ) ) { |
107
|
|
|
return null; |
108
|
|
|
} |
109
|
|
|
|
110
|
|
|
return $this->getTerm( $entityId, $languageCode, TermTypes::TYPE_LABEL ); |
111
|
|
|
} |
112
|
|
|
|
113
|
|
|
public function getDescription( EntityId $entityId, $languageCode ): ?string { |
114
|
|
|
if ( !$this->termLanguages->hasLanguage( $languageCode ) ) { |
115
|
|
|
return null; |
116
|
|
|
} |
117
|
|
|
|
118
|
|
|
return $this->getTerm( $entityId, $languageCode, TermTypes::TYPE_DESCRIPTION ); |
119
|
|
|
} |
120
|
|
|
|
121
|
|
|
private function getTerm( EntityId $entityId, string $languageCode, string $termType ) { |
122
|
|
|
$cachedTerm = $this->getBufferedOrCachedEntry( $entityId, $termType, $languageCode ); |
123
|
|
|
if ( $cachedTerm !== null ) { |
124
|
|
|
return $cachedTerm ?: null; |
125
|
|
|
} |
126
|
|
|
|
127
|
|
|
if ( $termType === TermTypes::TYPE_LABEL ) { |
128
|
|
|
$freshTerm = $this->lookup->getLabel( $entityId, $languageCode ); |
129
|
|
|
} else { |
130
|
|
|
$freshTerm = $this->lookup->getDescription( $entityId, $languageCode ); |
131
|
|
|
} |
132
|
|
|
|
133
|
|
|
if ( $freshTerm !== null ) { |
134
|
|
|
$this->bufferAndCacheExistingTerm( $entityId, $termType, $languageCode, $freshTerm ); |
135
|
|
|
} else { |
136
|
|
|
$this->bufferAndCacheMissingTerm( $entityId, $termType, $languageCode ); |
137
|
|
|
} |
138
|
|
|
|
139
|
|
|
return $freshTerm; |
140
|
|
|
} |
141
|
|
|
|
142
|
|
|
public function getLabels( EntityId $entityId, array $languageCodes ) { |
143
|
|
|
$validTermLanguageCodes = $this->filterValidTermLanguages( $languageCodes ); |
144
|
|
|
return $this->getMultipleTermsByLanguage( $entityId, TermTypes::TYPE_LABEL, $validTermLanguageCodes ); |
145
|
|
|
} |
146
|
|
|
|
147
|
|
|
public function getDescriptions( EntityId $entityId, array $languageCodes ) { |
148
|
|
|
$validTermLanguageCodes = $this->filterValidTermLanguages( $languageCodes ); |
149
|
|
|
return $this->getMultipleTermsByLanguage( $entityId, TermTypes::TYPE_DESCRIPTION, $validTermLanguageCodes ); |
150
|
|
|
} |
151
|
|
|
|
152
|
|
|
private function prefetchCachedTerms( array $entityIds, array $termTypes, array $languageCodes ): array { |
153
|
|
|
[ |
154
|
|
|
self::RESOLVED_KEYS => $cacheKeys, |
|
|
|
|
155
|
|
|
self::UNRESOLVED_IDS => $unresolvedIds, // This is intentionally unused. |
|
|
|
|
156
|
|
|
self::KEY_PARTS_MAP => $keyPartsMap, |
|
|
|
|
157
|
|
|
] = $this->getCacheKeys( $entityIds, $termTypes, $languageCodes ); |
158
|
|
|
$cacheResults = $this->cache->getMultiple( $cacheKeys ); |
159
|
|
|
$uncachedIds = []; |
160
|
|
|
$uncachedTermTypes = []; |
161
|
|
|
$uncachedLanguageCodes = []; |
162
|
|
|
foreach ( $cacheResults as $cacheKey => $cacheResult ) { |
163
|
|
|
$entityId = $keyPartsMap[$cacheKey][self::ENTITY_ID]; |
164
|
|
|
$termType = $keyPartsMap[$cacheKey][self::TERM_TYPE]; |
165
|
|
|
$languageCode = $keyPartsMap[$cacheKey][self::LANGUAGE_CODE]; |
166
|
|
|
if ( $cacheResult === null ) { |
167
|
|
|
$uncachedIds[$entityId->getSerialization()] = $entityId; |
168
|
|
|
$uncachedTermTypes[$termType] = $termType; |
169
|
|
|
$uncachedLanguageCodes[$languageCode] = $languageCode; |
170
|
|
|
continue; |
171
|
|
|
} |
172
|
|
|
|
173
|
|
|
// we have data from the cache |
174
|
|
|
$this->setPrefetchedTermBuffer( $entityId, $termType, $languageCode, $cacheResult ); |
175
|
|
|
} |
176
|
|
|
|
177
|
|
|
return [ |
178
|
|
|
self::UNCACHED_IDS => array_values( $uncachedIds ), |
179
|
|
|
self::UNCACHED_TERM_TYPES => array_values( $uncachedTermTypes ), |
180
|
|
|
self::UNCACHED_LANGUAGE_CODES => array_values( $uncachedLanguageCodes ), |
181
|
|
|
]; |
182
|
|
|
} |
183
|
|
|
|
184
|
|
|
private function getCacheKeys( array $entityIds, array $termTypes, array $languageCodes ): array { |
185
|
|
|
$cacheKeys = []; |
186
|
|
|
$unresolvedIds = []; |
187
|
|
|
$keyPartsMap = []; |
188
|
|
|
foreach ( $entityIds as $entityId ) { |
189
|
|
|
foreach ( $termTypes as $termType ) { |
190
|
|
|
foreach ( $languageCodes as $languageCode ) { |
191
|
|
|
$key = $this->getCacheKey( $entityId, $languageCode, $termType ); |
192
|
|
|
if ( $key === null ) { |
193
|
|
|
$unresolvedIds[] = $key; |
194
|
|
|
continue 2; // problem resolving the redirect / looking up the latest revision |
195
|
|
|
} |
196
|
|
|
$cacheKeys[] = $key; |
197
|
|
|
$keyPartsMap[$key] = [ |
198
|
|
|
self::ENTITY_ID => $entityId, |
199
|
|
|
self::TERM_TYPE => $termType, |
200
|
|
|
self::LANGUAGE_CODE => $languageCode, |
201
|
|
|
]; |
202
|
|
|
} |
203
|
|
|
} |
204
|
|
|
} |
205
|
|
|
return [ |
206
|
|
|
self::RESOLVED_KEYS => $cacheKeys, |
207
|
|
|
self::UNRESOLVED_IDS => $unresolvedIds, |
208
|
|
|
self::KEY_PARTS_MAP => $keyPartsMap, |
209
|
|
|
]; |
210
|
|
|
} |
211
|
|
|
|
212
|
|
|
private function prefetchAndCache( array $uncachedIds, array $uncachedTermTypes, array $uncachedLanguageCodes ): void { |
213
|
|
|
$this->lookup->prefetchTerms( $uncachedIds, $uncachedTermTypes, $uncachedLanguageCodes ); |
214
|
|
|
$this->cache->setMultiple( $this->getPrefetchedTermsFromLookup( |
|
|
|
|
215
|
|
|
$uncachedIds, |
216
|
|
|
$uncachedLanguageCodes, |
217
|
|
|
$uncachedTermTypes |
218
|
|
|
), $this->cacheEntryTTL ); |
219
|
|
|
} |
220
|
|
|
|
221
|
|
|
private function getPrefetchedTermsFromLookup( |
222
|
|
|
array $entitiesToPrefetch, |
223
|
|
|
array $languagesToPrefetch, |
224
|
|
|
array $termTypes |
225
|
|
|
): array { |
226
|
|
|
$terms = []; |
227
|
|
|
|
228
|
|
|
foreach ( $entitiesToPrefetch as $entity ) { |
229
|
|
|
foreach ( $languagesToPrefetch as $language ) { |
230
|
|
|
foreach ( $termTypes as $termType ) { |
231
|
|
|
$cacheKey = $this->getCacheKey( $entity, $language, $termType ); |
232
|
|
|
if ( $cacheKey !== null ) { |
233
|
|
|
$terms[$cacheKey] = $this->getTermFromLookup( $entity, $termType, $language ); |
234
|
|
|
} |
235
|
|
|
} |
236
|
|
|
} |
237
|
|
|
} |
238
|
|
|
|
239
|
|
|
return $terms; |
240
|
|
|
} |
241
|
|
|
|
242
|
|
|
private function getTermFromLookup( EntityId $entity, string $termType, string $language ) { |
243
|
|
|
if ( $termType === TermTypes::TYPE_LABEL || $termType === TermTypes::TYPE_DESCRIPTION ) { |
244
|
|
|
return $this->lookup->getPrefetchedTerm( $entity, $termType, $language ); |
245
|
|
|
} |
246
|
|
|
|
247
|
|
|
return $this->lookup->getPrefetchedAliases( $entity, $language ); |
248
|
|
|
} |
249
|
|
|
|
250
|
|
|
/** |
251
|
|
|
* @param EntityId $entityId |
252
|
|
|
* @param string $termType |
253
|
|
|
* @param string $languageCode |
254
|
|
|
* @param string|string[]|false $value string for existing label or description, string[] for existing aliases, false for term known |
255
|
|
|
* to not exist. |
256
|
|
|
*/ |
257
|
|
|
private function setPrefetchedTermBuffer( EntityId $entityId, string $termType, string $languageCode, $value ): void { |
258
|
|
|
if ( !isset( $this->prefetchedTerms[$entityId->getSerialization()] ) ) { |
259
|
|
|
$this->prefetchedTerms[$entityId->getSerialization()] = []; |
260
|
|
|
} |
261
|
|
|
if ( !isset( $this->prefetchedTerms[$entityId->getSerialization()][$termType] ) ) { |
262
|
|
|
$this->prefetchedTerms[$entityId->getSerialization()][$termType] = []; |
263
|
|
|
} |
264
|
|
|
|
265
|
|
|
$this->prefetchedTerms[$entityId->getSerialization()][$termType][$languageCode] = $value; |
266
|
|
|
} |
267
|
|
|
|
268
|
|
|
/** |
269
|
|
|
* @param EntityId $entityId |
270
|
|
|
* @param string $termType |
271
|
|
|
* @param string $languageCode |
272
|
|
|
* @param string|string[] $freshTerm string for existing label or description, string[] for existing aliases |
273
|
|
|
* Should never be null or false @see bufferAndCacheMissingTerm |
274
|
|
|
*/ |
275
|
|
|
private function bufferAndCacheExistingTerm( EntityId $entityId, string $termType, string $languageCode, $freshTerm ): void { |
276
|
|
|
$this->setPrefetchedTermBuffer( $entityId, $termType, $languageCode, $freshTerm ); |
277
|
|
|
$this->cache->set( |
278
|
|
|
$this->getCacheKey( $entityId, $languageCode, $termType ), |
279
|
|
|
$freshTerm, |
280
|
|
|
$this->cacheEntryTTL |
281
|
|
|
); |
282
|
|
|
} |
283
|
|
|
|
284
|
|
|
private function bufferAndCacheMissingTerm( EntityId $entityId, string $termType, string $languageCode ): void { |
285
|
|
|
$this->setPrefetchedTermBuffer( $entityId, $termType, $languageCode, false ); |
286
|
|
|
$cacheKey = $this->getCacheKey( $entityId, $languageCode, $termType ); |
287
|
|
|
if ( $cacheKey === null ) { |
288
|
|
|
return; |
289
|
|
|
} |
290
|
|
|
$this->cache->set( |
291
|
|
|
$cacheKey, |
292
|
|
|
false, |
293
|
|
|
$this->cacheEntryTTL |
294
|
|
|
); |
295
|
|
|
} |
296
|
|
|
|
297
|
|
|
private function getCacheKey( EntityId $id, string $language, string $termType ) { |
298
|
|
|
$resolutionResult = $this->redirectResolvingRevisionLookup->lookupLatestRevisionResolvingRedirect( $id ); |
299
|
|
|
if ( $resolutionResult === null ) { |
300
|
|
|
return null; |
301
|
|
|
} |
302
|
|
|
|
303
|
|
|
return $this->buildCacheKey( $resolutionResult[1], $resolutionResult[0], $language, $termType ); |
304
|
|
|
} |
305
|
|
|
|
306
|
|
|
private function getBufferedOrCachedEntry( EntityId $entityId, string $termType, string $languageCode ) { |
307
|
|
|
// Check if it's prefetched already. |
308
|
|
|
$prefetchedTerm = $this->getPrefetchedTerm( $entityId, $termType, $languageCode ); |
309
|
|
|
if ( $prefetchedTerm !== null ) { |
310
|
|
|
return $prefetchedTerm; |
311
|
|
|
} |
312
|
|
|
|
313
|
|
|
// Try getting it from cache |
314
|
|
|
$cacheKey = $this->getCacheKey( $entityId, $languageCode, $termType ); |
315
|
|
|
return $cacheKey === null ? null : $this->cache->get( $cacheKey ); |
316
|
|
|
} |
317
|
|
|
|
318
|
|
|
private function getMultipleTermsByLanguageFromBuffer( EntityId $entityId, string $termType, array $languages ) { |
319
|
|
|
$terms = []; |
320
|
|
|
|
321
|
|
|
// Lookup in termbuffer |
322
|
|
|
foreach ( $languages as $language ) { |
323
|
|
|
$prefetchedTerm = $this->getPrefetchedTerm( $entityId, $termType, $language ); |
324
|
|
|
if ( $prefetchedTerm !== null ) { |
325
|
|
|
$terms[$language] = $prefetchedTerm; |
326
|
|
|
} |
327
|
|
|
} |
328
|
|
|
|
329
|
|
|
return $terms; |
330
|
|
|
} |
331
|
|
|
|
332
|
|
|
private function getMultipleTermsByLanguageFromCache( EntityId $entityId, string $termType, array $languages ) { |
333
|
|
|
$terms = []; |
334
|
|
|
|
335
|
|
|
$languagesToCacheKeys = []; |
336
|
|
|
|
337
|
|
|
foreach ( $languages as $language ) { |
338
|
|
|
$cacheKey = $this->getCacheKey( $entityId, $language, $termType ); |
339
|
|
|
if ( $cacheKey ) { |
340
|
|
|
$languagesToCacheKeys[$language] = $cacheKey; |
341
|
|
|
} |
342
|
|
|
} |
343
|
|
|
|
344
|
|
|
$cacheKeysToLanguages = array_flip( $languagesToCacheKeys ); |
345
|
|
|
$cacheTerms = $this->cache->getMultiple( array_values( $languagesToCacheKeys ) ); |
|
|
|
|
346
|
|
|
|
347
|
|
|
foreach ( $cacheTerms as $key => $term ) { |
348
|
|
|
if ( $term !== null ) { |
349
|
|
|
$terms[$cacheKeysToLanguages[$key]] = $term; |
350
|
|
|
$this->setPrefetchedTermBuffer( $entityId, $termType, $cacheKeysToLanguages[$key], $term ); |
351
|
|
|
} |
352
|
|
|
} |
353
|
|
|
|
354
|
|
|
return $terms; |
355
|
|
|
} |
356
|
|
|
|
357
|
|
|
private function getMultipleTermsByLanguageFromLookup( EntityId $entityId, string $termType, array $languages ) { |
358
|
|
|
if ( $termType === TermTypes::TYPE_LABEL ) { |
359
|
|
|
$freshTerms = $this->lookup->getLabels( $entityId, $languages ); |
360
|
|
|
} else { |
361
|
|
|
$freshTerms = $this->lookup->getDescriptions( $entityId, $languages ); |
362
|
|
|
} |
363
|
|
|
|
364
|
|
|
foreach ( $languages as $language ) { |
365
|
|
|
if ( !isset( $freshTerms[$language] ) || !$freshTerms[$language] ) { |
366
|
|
|
$this->bufferAndCacheMissingTerm( $entityId, $termType, $language ); |
367
|
|
|
} else { |
368
|
|
|
$this->bufferAndCacheExistingTerm( $entityId, $termType, $language, $freshTerms[$language] ); |
369
|
|
|
} |
370
|
|
|
} |
371
|
|
|
|
372
|
|
|
return $freshTerms; |
373
|
|
|
} |
374
|
|
|
|
375
|
|
|
private function getMultipleTermsByLanguage( EntityId $entityId, string $termType, array $languages ) { |
376
|
|
|
$terms = $this->getMultipleTermsByLanguageFromBuffer( $entityId, $termType, $languages ); |
377
|
|
|
|
378
|
|
|
// languages without prefetched terms |
379
|
|
|
$unbufferedLanguages = array_diff( $languages, array_keys( $terms ) ); |
380
|
|
|
if ( empty( $unbufferedLanguages ) ) { |
381
|
|
|
return array_filter( $terms ); |
382
|
|
|
} |
383
|
|
|
|
384
|
|
|
$terms = array_merge( |
385
|
|
|
$terms, |
386
|
|
|
$this->getMultipleTermsByLanguageFromCache( $entityId, $termType, $unbufferedLanguages ) |
387
|
|
|
); |
388
|
|
|
|
389
|
|
|
$unCachedAndUnbufferedLanguages = array_values( |
390
|
|
|
array_diff( $languages, array_keys( $terms ) ) |
391
|
|
|
); |
392
|
|
|
if ( empty( $unCachedAndUnbufferedLanguages ) ) { |
393
|
|
|
return array_filter( $terms ); |
394
|
|
|
} |
395
|
|
|
|
396
|
|
|
$terms = array_merge( |
397
|
|
|
$terms, |
398
|
|
|
$this->getMultipleTermsByLanguageFromLookup( $entityId, $termType, $unCachedAndUnbufferedLanguages ) |
399
|
|
|
); |
400
|
|
|
|
401
|
|
|
return array_filter( $terms ); |
402
|
|
|
} |
403
|
|
|
|
404
|
|
|
private function filterValidTermLanguages( array $languageCodes ): array { |
405
|
|
|
return array_filter( |
406
|
|
|
$languageCodes, |
407
|
|
|
function ( $languageCode ) { |
408
|
|
|
return $this->termLanguages->hasLanguage( $languageCode ); |
409
|
|
|
} |
410
|
|
|
); |
411
|
|
|
} |
412
|
|
|
|
413
|
|
|
} |
414
|
|
|
|
This check marks access to variables or properties that have not been declared yet. While PHP has no explicit notion of declaring a variable, accessing it before a value is assigned to it is most likely a bug.