Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
97.37% |
148 / 152 |
|
78.57% |
11 / 14 |
CRAP | |
0.00% |
0 / 1 |
| CachingResultsSource | |
97.37% |
148 / 152 |
|
78.57% |
11 / 14 |
49 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
11 / 11 |
|
100.00% |
1 / 1 |
1 | |||
| getResults | |
100.00% |
24 / 24 |
|
100.00% |
1 / 1 |
9 | |||
| canUseStoredResults | |
100.00% |
7 / 7 |
|
100.00% |
1 / 1 |
4 | |||
| statusSelected | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
2 | |||
| getAndStoreResults | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
3 | |||
| canStoreResults | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
3 | |||
| storeResults | |
94.74% |
18 / 19 |
|
0.00% |
0 / 1 |
6.01 | |||
| getStoredResults | |
100.00% |
22 / 22 |
|
100.00% |
1 / 1 |
4 | |||
| checkDependencyMetadata | |
96.30% |
26 / 27 |
|
0.00% |
0 / 1 |
7 | |||
| deserializeCheckResult | |
100.00% |
9 / 9 |
|
100.00% |
1 / 1 |
2 | |||
| isPossiblyStaleResult | |
100.00% |
6 / 6 |
|
100.00% |
1 / 1 |
2 | |||
| getLatestRevisionIds | |
84.62% |
11 / 13 |
|
0.00% |
0 / 1 |
4.06 | |||
| hasFalseElements | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| setMicrotimeFunction | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace WikibaseQuality\ConstraintReport\Api; |
| 4 | |
| 5 | use DataValues\TimeValue; |
| 6 | use Wikibase\DataModel\Entity\EntityId; |
| 7 | use Wikibase\DataModel\Entity\EntityIdParser; |
| 8 | use Wikibase\Lib\Store\LookupConstants; |
| 9 | use Wikibase\Lib\Store\Sql\WikiPageEntityMetaDataAccessor; |
| 10 | use WikibaseQuality\ConstraintReport\ConstraintCheck\Cache\CachedCheckResults; |
| 11 | use WikibaseQuality\ConstraintReport\ConstraintCheck\Cache\CachingMetadata; |
| 12 | use WikibaseQuality\ConstraintReport\ConstraintCheck\Cache\DependencyMetadata; |
| 13 | use WikibaseQuality\ConstraintReport\ConstraintCheck\Cache\Metadata; |
| 14 | use WikibaseQuality\ConstraintReport\ConstraintCheck\Helper\LoggingHelper; |
| 15 | use WikibaseQuality\ConstraintReport\ConstraintCheck\Helper\TimeValueComparer; |
| 16 | use WikibaseQuality\ConstraintReport\ConstraintCheck\Result\CheckResult; |
| 17 | use WikibaseQuality\ConstraintReport\ConstraintCheck\Result\CheckResultDeserializer; |
| 18 | use WikibaseQuality\ConstraintReport\ConstraintCheck\Result\CheckResultSerializer; |
| 19 | use WikibaseQuality\ConstraintReport\ConstraintCheck\Result\NullResult; |
| 20 | use Wikimedia\ObjectCache\WANObjectCache; |
| 21 | |
| 22 | /** |
| 23 | * A ResultsSource that wraps another ResultsSource, |
| 24 | * storing results in a cache |
| 25 | * and retrieving them from there if the results are still fresh. |
| 26 | * |
| 27 | * @author Lucas Werkmeister |
| 28 | * @license GPL-2.0-or-later |
| 29 | */ |
| 30 | class CachingResultsSource implements ResultsSource { |
| 31 | |
| 32 | public const CACHED_STATUSES = [ |
| 33 | CheckResult::STATUS_VIOLATION, |
| 34 | CheckResult::STATUS_WARNING, |
| 35 | CheckResult::STATUS_SUGGESTION, |
| 36 | CheckResult::STATUS_BAD_PARAMETERS, |
| 37 | ]; |
| 38 | |
| 39 | /** |
| 40 | * @var ResultsSource |
| 41 | */ |
| 42 | private $resultsSource; |
| 43 | |
| 44 | /** |
| 45 | * @var ResultsCache |
| 46 | */ |
| 47 | private $cache; |
| 48 | |
| 49 | /** |
| 50 | * @var CheckResultSerializer |
| 51 | */ |
| 52 | private $checkResultSerializer; |
| 53 | |
| 54 | /** |
| 55 | * @var CheckResultDeserializer |
| 56 | */ |
| 57 | private $checkResultDeserializer; |
| 58 | |
| 59 | /** |
| 60 | * @var WikiPageEntityMetaDataAccessor |
| 61 | */ |
| 62 | private $wikiPageEntityMetaDataAccessor; |
| 63 | |
| 64 | /** |
| 65 | * @var EntityIdParser |
| 66 | */ |
| 67 | private $entityIdParser; |
| 68 | |
| 69 | /** |
| 70 | * @var int |
| 71 | */ |
| 72 | private $ttlInSeconds; |
| 73 | |
| 74 | /** |
| 75 | * @var string[] |
| 76 | */ |
| 77 | private $possiblyStaleConstraintTypes; |
| 78 | |
| 79 | /** |
| 80 | * @var int |
| 81 | */ |
| 82 | private $maxRevisionIds; |
| 83 | |
| 84 | /** |
| 85 | * @var LoggingHelper |
| 86 | */ |
| 87 | private $loggingHelper; |
| 88 | |
| 89 | /** |
| 90 | * @var TimeValueComparer |
| 91 | */ |
| 92 | private $timeValueComparer; |
| 93 | |
| 94 | /** |
| 95 | * @var callable |
| 96 | */ |
| 97 | private $microtime = 'microtime'; |
| 98 | |
| 99 | /** |
| 100 | * @param ResultsSource $resultsSource The ResultsSource that cache misses are delegated to. |
| 101 | * @param ResultsCache $cache The cache where results can be stored. |
| 102 | * @param CheckResultSerializer $checkResultSerializer Used to serialize check results. |
| 103 | * @param CheckResultDeserializer $checkResultDeserializer Used to deserialize check results. |
| 104 | * @param WikiPageEntityMetaDataAccessor $wikiPageEntityMetaDataAccessor Used to get the latest revision ID. |
| 105 | * @param EntityIdParser $entityIdParser Used to parse entity IDs in cached objects. |
| 106 | * @param int $ttlInSeconds Time-to-live of the cached values, in seconds. |
| 107 | * @param string[] $possiblyStaleConstraintTypes item IDs of constraint types |
| 108 | * where cached results may always be stale, regardless of invalidation logic |
| 109 | * @param int $maxRevisionIds The maximum number of revision IDs to check; |
| 110 | * if a check result depends on more entity IDs than this number, it is not cached. |
| 111 | * @param LoggingHelper $loggingHelper |
| 112 | */ |
| 113 | public function __construct( |
| 114 | ResultsSource $resultsSource, |
| 115 | ResultsCache $cache, |
| 116 | CheckResultSerializer $checkResultSerializer, |
| 117 | CheckResultDeserializer $checkResultDeserializer, |
| 118 | WikiPageEntityMetaDataAccessor $wikiPageEntityMetaDataAccessor, |
| 119 | EntityIdParser $entityIdParser, |
| 120 | $ttlInSeconds, |
| 121 | array $possiblyStaleConstraintTypes, |
| 122 | $maxRevisionIds, |
| 123 | LoggingHelper $loggingHelper |
| 124 | ) { |
| 125 | $this->resultsSource = $resultsSource; |
| 126 | $this->cache = $cache; |
| 127 | $this->checkResultSerializer = $checkResultSerializer; |
| 128 | $this->checkResultDeserializer = $checkResultDeserializer; |
| 129 | $this->wikiPageEntityMetaDataAccessor = $wikiPageEntityMetaDataAccessor; |
| 130 | $this->entityIdParser = $entityIdParser; |
| 131 | $this->ttlInSeconds = $ttlInSeconds; |
| 132 | $this->possiblyStaleConstraintTypes = $possiblyStaleConstraintTypes; |
| 133 | $this->maxRevisionIds = $maxRevisionIds; |
| 134 | $this->loggingHelper = $loggingHelper; |
| 135 | $this->timeValueComparer = new TimeValueComparer(); |
| 136 | } |
| 137 | |
| 138 | public function getResults( |
| 139 | array $entityIds, |
| 140 | array $claimIds, |
| 141 | ?array $constraintIds, |
| 142 | array $statuses |
| 143 | ) { |
| 144 | $results = []; |
| 145 | $metadatas = []; |
| 146 | if ( $this->canUseStoredResults( $entityIds, $claimIds, $constraintIds, $statuses ) ) { |
| 147 | $storedEntityIds = []; |
| 148 | foreach ( $entityIds as $entityId ) { |
| 149 | $storedResults = $this->getStoredResults( $entityId ); |
| 150 | if ( $storedResults !== null ) { |
| 151 | $this->loggingHelper->logCheckConstraintsCacheHit( $entityId ); |
| 152 | foreach ( $storedResults->getArray() as $checkResult ) { |
| 153 | if ( $this->statusSelected( $statuses, $checkResult ) ) { |
| 154 | $results[] = $checkResult; |
| 155 | } |
| 156 | } |
| 157 | $metadatas[] = $storedResults->getMetadata(); |
| 158 | $storedEntityIds[] = $entityId; |
| 159 | } |
| 160 | } |
| 161 | $entityIds = array_values( array_diff( $entityIds, $storedEntityIds ) ); |
| 162 | } |
| 163 | if ( $entityIds !== [] || $claimIds !== [] ) { |
| 164 | if ( $entityIds !== [] ) { |
| 165 | $this->loggingHelper->logCheckConstraintsCacheMisses( $entityIds ); |
| 166 | } |
| 167 | $response = $this->getAndStoreResults( $entityIds, $claimIds, $constraintIds, $statuses ); |
| 168 | $results = array_merge( $results, $response->getArray() ); |
| 169 | $metadatas[] = $response->getMetadata(); |
| 170 | } |
| 171 | return new CachedCheckResults( |
| 172 | $results, |
| 173 | Metadata::merge( $metadatas ) |
| 174 | ); |
| 175 | } |
| 176 | |
| 177 | /** |
| 178 | * We can only use cached constraint results |
| 179 | * if nothing more than the problematic results of a full constraint check were requested: |
| 180 | * constraint checks for the full entity (not just individual statements), |
| 181 | * without restricting the set of constraints to check, |
| 182 | * and with no statuses other than 'violation', 'warning' and 'bad-parameters'. |
| 183 | * |
| 184 | * @param EntityId[] $entityIds |
| 185 | * @param string[] $claimIds |
| 186 | * @param string[]|null $constraintIds |
| 187 | * @param string[] $statuses |
| 188 | * @return bool |
| 189 | */ |
| 190 | private function canUseStoredResults( |
| 191 | array $entityIds, |
| 192 | array $claimIds, |
| 193 | ?array $constraintIds, |
| 194 | array $statuses |
| 195 | ) { |
| 196 | if ( $claimIds !== [] ) { |
| 197 | return false; |
| 198 | } |
| 199 | if ( $constraintIds !== null ) { |
| 200 | return false; |
| 201 | } |
| 202 | if ( array_diff( $statuses, self::CACHED_STATUSES ) !== [] ) { |
| 203 | return false; |
| 204 | } |
| 205 | return true; |
| 206 | } |
| 207 | |
| 208 | /** |
| 209 | * Check whether a check result should be used, |
| 210 | * either because it has the right status |
| 211 | * or because it is a NullResult whose metadata should be preserved. |
| 212 | * |
| 213 | * @param string[] $statuses |
| 214 | * @param CheckResult $result |
| 215 | * @return bool |
| 216 | */ |
| 217 | private function statusSelected( array $statuses, CheckResult $result ) { |
| 218 | return in_array( $result->getStatus(), $statuses, true ) || |
| 219 | $result instanceof NullResult; |
| 220 | } |
| 221 | |
| 222 | /** |
| 223 | * @param EntityId[] $entityIds |
| 224 | * @param string[] $claimIds |
| 225 | * @param string[]|null $constraintIds |
| 226 | * @param string[] $statuses |
| 227 | * @return CachedCheckResults |
| 228 | */ |
| 229 | public function getAndStoreResults( |
| 230 | array $entityIds, |
| 231 | array $claimIds, |
| 232 | ?array $constraintIds, |
| 233 | array $statuses |
| 234 | ) { |
| 235 | $results = $this->resultsSource->getResults( $entityIds, $claimIds, $constraintIds, $statuses ); |
| 236 | |
| 237 | if ( $this->canStoreResults( $entityIds, $claimIds, $constraintIds, $statuses ) ) { |
| 238 | foreach ( $entityIds as $entityId ) { |
| 239 | $this->storeResults( $entityId, $results ); |
| 240 | } |
| 241 | } |
| 242 | |
| 243 | return $results; |
| 244 | } |
| 245 | |
| 246 | /** |
| 247 | * We can only store constraint results |
| 248 | * if the set of constraints to check was not restricted |
| 249 | * and all the problematic results were requested. |
| 250 | * However, it doesn’t matter whether constraint checks on individual statements were requested: |
| 251 | * we only store results for the mentioned entity IDs, |
| 252 | * and those will be complete regardless of what’s in the statement IDs. |
| 253 | * And it also doesn’t matter whether the set of statuses requested |
| 254 | * was exactly the statuses we cache or a superset of it: |
| 255 | * as long as all the results we want to cache are there, |
| 256 | * we can filter out the extraneous ones before we serialize them. |
| 257 | * |
| 258 | * @param EntityId[] $entityIds |
| 259 | * @param string[] $claimIds |
| 260 | * @param ?string[] $constraintIds |
| 261 | * @param string[] $statuses |
| 262 | * @return bool |
| 263 | */ |
| 264 | private function canStoreResults( |
| 265 | array $entityIds, |
| 266 | array $claimIds, |
| 267 | ?array $constraintIds, |
| 268 | array $statuses |
| 269 | ) { |
| 270 | if ( $constraintIds !== null ) { |
| 271 | return false; |
| 272 | } |
| 273 | if ( array_diff( self::CACHED_STATUSES, $statuses ) !== [] ) { |
| 274 | return false; |
| 275 | } |
| 276 | return true; |
| 277 | } |
| 278 | |
| 279 | /** |
| 280 | * Store check results for the given entity ID in the cache, if possible. |
| 281 | * |
| 282 | * @param EntityId $entityId The entity ID. |
| 283 | * @param CachedCheckResults $results A collection of check results with metadata. |
| 284 | * May include check results for other entity IDs as well, |
| 285 | * or check results with statuses that we’re not interested in caching. |
| 286 | */ |
| 287 | private function storeResults( EntityId $entityId, CachedCheckResults $results ) { |
| 288 | $latestRevisionIds = $this->getLatestRevisionIds( |
| 289 | $results->getMetadata()->getDependencyMetadata()->getEntityIds() |
| 290 | ); |
| 291 | if ( $latestRevisionIds === null ) { |
| 292 | return; |
| 293 | } |
| 294 | |
| 295 | $resultSerializations = []; |
| 296 | foreach ( $results->getArray() as $checkResult ) { |
| 297 | if ( $checkResult->getContextCursor()->getEntityId() !== $entityId->getSerialization() ) { |
| 298 | continue; |
| 299 | } |
| 300 | if ( $this->statusSelected( self::CACHED_STATUSES, $checkResult ) ) { |
| 301 | $resultSerializations[] = $this->checkResultSerializer->serialize( $checkResult ); |
| 302 | } |
| 303 | } |
| 304 | |
| 305 | $value = [ |
| 306 | 'results' => $resultSerializations, |
| 307 | 'latestRevisionIds' => $latestRevisionIds, |
| 308 | ]; |
| 309 | $futureTime = $results->getMetadata()->getDependencyMetadata()->getFutureTime(); |
| 310 | if ( $futureTime !== null ) { |
| 311 | $value['futureTime'] = $futureTime->getArrayValue(); |
| 312 | } |
| 313 | |
| 314 | $this->cache->set( $entityId, $value, $this->ttlInSeconds ); |
| 315 | } |
| 316 | |
| 317 | /** |
| 318 | * @param EntityId $entityId |
| 319 | * @param int $forRevision Requested revision of $entityId |
| 320 | * If this parameter is not zero, the results are returned if this is the latest revision, |
| 321 | * otherwise null is returned, since we can't get constraints for past revisions. |
| 322 | * @return CachedCheckResults|null |
| 323 | */ |
| 324 | public function getStoredResults( |
| 325 | EntityId $entityId, |
| 326 | $forRevision = 0 |
| 327 | ) { |
| 328 | $cacheInfo = WANObjectCache::PASS_BY_REF; |
| 329 | $value = $this->cache->get( $entityId, $curTTL, [], $cacheInfo ); |
| 330 | $now = call_user_func( $this->microtime, true ); |
| 331 | |
| 332 | $dependencyMetadata = $this->checkDependencyMetadata( $value, |
| 333 | [ $entityId->getSerialization() => $forRevision ] ); |
| 334 | if ( $dependencyMetadata === null ) { |
| 335 | return null; |
| 336 | } |
| 337 | |
| 338 | // @phan-suppress-next-line PhanTypePossiblyInvalidDimOffset False positive |
| 339 | $asOf = $cacheInfo[WANObjectCache::KEY_AS_OF]; |
| 340 | $ageInSeconds = (int)ceil( $now - $asOf ); |
| 341 | $cachingMetadata = $ageInSeconds > 0 ? |
| 342 | CachingMetadata::ofMaximumAgeInSeconds( $ageInSeconds ) : |
| 343 | CachingMetadata::fresh(); |
| 344 | |
| 345 | $results = []; |
| 346 | foreach ( $value['results'] as $resultSerialization ) { |
| 347 | $results[] = $this->deserializeCheckResult( $resultSerialization, $cachingMetadata ); |
| 348 | } |
| 349 | |
| 350 | return new CachedCheckResults( |
| 351 | $results, |
| 352 | Metadata::merge( [ |
| 353 | Metadata::ofCachingMetadata( $cachingMetadata ), |
| 354 | Metadata::ofDependencyMetadata( $dependencyMetadata ), |
| 355 | ] ) |
| 356 | ); |
| 357 | } |
| 358 | |
| 359 | /** |
| 360 | * Extract the dependency metadata of $value |
| 361 | * and check that the dependency metadata does not indicate staleness. |
| 362 | * |
| 363 | * @param array|false $value |
| 364 | * @param int[] $paramRevs Revisions from parameters, id => revision |
| 365 | * These revisions are used instead of ones recorded in the metadata, |
| 366 | * so we can serve requests specifying concrete revisions, and if they are not latest, |
| 367 | * we will reject then. |
| 368 | * @return DependencyMetadata|null the dependency metadata, |
| 369 | * or null if $value should no longer be used |
| 370 | */ |
| 371 | private function checkDependencyMetadata( $value, $paramRevs ) { |
| 372 | if ( $value === false ) { |
| 373 | return null; |
| 374 | } |
| 375 | |
| 376 | if ( array_key_exists( 'futureTime', $value ) ) { |
| 377 | $futureTime = TimeValue::newFromArray( $value['futureTime'] ); |
| 378 | if ( !$this->timeValueComparer->isFutureTime( $futureTime ) ) { |
| 379 | return null; |
| 380 | } |
| 381 | $futureTimeDependencyMetadata = DependencyMetadata::ofFutureTime( $futureTime ); |
| 382 | } else { |
| 383 | $futureTimeDependencyMetadata = DependencyMetadata::blank(); |
| 384 | } |
| 385 | |
| 386 | foreach ( $paramRevs as $id => $revision ) { |
| 387 | if ( $revision > 0 ) { |
| 388 | $value['latestRevisionIds'][$id] = min( $revision, $value['latestRevisionIds'][$id] ?? PHP_INT_MAX ); |
| 389 | } |
| 390 | } |
| 391 | |
| 392 | $dependedEntityIds = array_map( |
| 393 | [ $this->entityIdParser, "parse" ], |
| 394 | array_keys( $value['latestRevisionIds'] ) |
| 395 | ); |
| 396 | |
| 397 | if ( $value['latestRevisionIds'] !== $this->getLatestRevisionIds( $dependedEntityIds ) ) { |
| 398 | return null; |
| 399 | } |
| 400 | |
| 401 | return array_reduce( |
| 402 | $dependedEntityIds, |
| 403 | static function ( DependencyMetadata $metadata, EntityId $entityId ) { |
| 404 | return DependencyMetadata::merge( [ |
| 405 | $metadata, |
| 406 | DependencyMetadata::ofEntityId( $entityId ), |
| 407 | ] ); |
| 408 | }, |
| 409 | $futureTimeDependencyMetadata |
| 410 | ); |
| 411 | } |
| 412 | |
| 413 | /** |
| 414 | * Deserialize a check result. |
| 415 | * If the result might be stale after caching |
| 416 | * (because its dependencies cannot be fully tracked in its dependency metadata), |
| 417 | * also add $cachingMetadata to it. |
| 418 | * |
| 419 | * @param array $resultSerialization |
| 420 | * @param CachingMetadata $cachingMetadata |
| 421 | * @return CheckResult |
| 422 | */ |
| 423 | private function deserializeCheckResult( |
| 424 | array $resultSerialization, |
| 425 | CachingMetadata $cachingMetadata |
| 426 | ) { |
| 427 | $result = $this->checkResultDeserializer->deserialize( $resultSerialization ); |
| 428 | if ( $this->isPossiblyStaleResult( $result ) ) { |
| 429 | $result->withMetadata( |
| 430 | Metadata::merge( [ |
| 431 | $result->getMetadata(), |
| 432 | Metadata::ofCachingMetadata( $cachingMetadata ), |
| 433 | ] ) |
| 434 | ); |
| 435 | } |
| 436 | return $result; |
| 437 | } |
| 438 | |
| 439 | /** |
| 440 | * @param CheckResult $result |
| 441 | * @return bool |
| 442 | */ |
| 443 | private function isPossiblyStaleResult( CheckResult $result ) { |
| 444 | if ( $result instanceof NullResult ) { |
| 445 | return false; |
| 446 | } |
| 447 | |
| 448 | return in_array( |
| 449 | $result->getConstraint()->getConstraintTypeItemId(), |
| 450 | $this->possiblyStaleConstraintTypes |
| 451 | ); |
| 452 | } |
| 453 | |
| 454 | /** |
| 455 | * @param EntityId[] $entityIds |
| 456 | * @return int[]|null array from entity ID serializations to revision ID, |
| 457 | * or null to indicate that not all revision IDs could be loaded |
| 458 | */ |
| 459 | private function getLatestRevisionIds( array $entityIds ) { |
| 460 | if ( $entityIds === [] ) { |
| 461 | $this->loggingHelper->logEmptyDependencyMetadata(); |
| 462 | return []; |
| 463 | } |
| 464 | if ( count( $entityIds ) > $this->maxRevisionIds ) { |
| 465 | // one of those entities will probably be edited soon, so might as well skip caching |
| 466 | $this->loggingHelper->logHugeDependencyMetadata( $entityIds, $this->maxRevisionIds ); |
| 467 | return null; |
| 468 | } |
| 469 | |
| 470 | $latestRevisionIds = $this->wikiPageEntityMetaDataAccessor->loadLatestRevisionIds( |
| 471 | $entityIds, |
| 472 | LookupConstants::LATEST_FROM_REPLICA |
| 473 | ); |
| 474 | if ( $this->hasFalseElements( $latestRevisionIds ) ) { |
| 475 | return null; |
| 476 | } |
| 477 | return $latestRevisionIds; |
| 478 | } |
| 479 | |
| 480 | /** |
| 481 | * @param array $array |
| 482 | * @return bool |
| 483 | */ |
| 484 | private function hasFalseElements( array $array ) { |
| 485 | return in_array( false, $array, true ); |
| 486 | } |
| 487 | |
| 488 | /** |
| 489 | * Set a custom function to get the current time, instead of microtime(). |
| 490 | * |
| 491 | * @param callable $microtime |
| 492 | */ |
| 493 | public function setMicrotimeFunction( callable $microtime ) { |
| 494 | $this->microtime = $microtime; |
| 495 | } |
| 496 | |
| 497 | } |