Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
79.15% |
315 / 398 |
|
63.16% |
12 / 19 |
CRAP | |
0.00% |
0 / 1 |
| SparqlHelper | |
79.15% |
315 / 398 |
|
63.16% |
12 / 19 |
138.05 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
29 / 29 |
|
100.00% |
1 / 1 |
3 | |||
| getQueryPrefixes | |
100.00% |
42 / 42 |
|
100.00% |
1 / 1 |
3 | |||
| hasType | |
85.71% |
24 / 28 |
|
0.00% |
0 / 1 |
4.05 | |||
| nestedSeparatorFilter | |
100.00% |
12 / 12 |
|
100.00% |
1 / 1 |
1 | |||
| findEntitiesWithSameStatement | |
100.00% |
16 / 16 |
|
100.00% |
1 / 1 |
2 | |||
| findEntitiesWithSameQualifierOrReference | |
100.00% |
27 / 27 |
|
100.00% |
1 / 1 |
6 | |||
| stringLiteral | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getOtherEntities | |
87.50% |
21 / 24 |
|
0.00% |
0 / 1 |
5.05 | |||
| getRdfLiteral | |
90.00% |
27 / 30 |
|
0.00% |
0 / 1 |
16.26 | |||
| matchesRegularExpression | |
0.00% |
0 / 60 |
|
0.00% |
0 / 1 |
90 | |||
| serializeConstraintParameterException | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
1 | |||
| deserializeConstraintParameterException | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
1 | |||
| matchesRegularExpressionWithSparql | |
100.00% |
13 / 13 |
|
100.00% |
1 / 1 |
2 | |||
| isTimeout | |
100.00% |
7 / 7 |
|
100.00% |
1 / 1 |
1 | |||
| getCacheMaxAge | |
100.00% |
8 / 8 |
|
100.00% |
1 / 1 |
5 | |||
| getThrottling | |
64.29% |
9 / 14 |
|
0.00% |
0 / 1 |
7.64 | |||
| getTimestampInFuture | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| runQuery | |
90.00% |
54 / 60 |
|
0.00% |
0 / 1 |
9.08 | |||
| guardAgainstTooManyRequestsError | |
88.24% |
15 / 17 |
|
0.00% |
0 / 1 |
4.03 | |||
| 1 | <?php |
| 2 | |
| 3 | declare( strict_types = 1 ); |
| 4 | |
| 5 | namespace WikibaseQuality\ConstraintReport\ConstraintCheck\Helper; |
| 6 | |
| 7 | use DataValues\DataValue; |
| 8 | use DataValues\MonolingualTextValue; |
| 9 | use DateInterval; |
| 10 | use InvalidArgumentException; |
| 11 | use MapCacheLRU; |
| 12 | use MediaWiki\Config\Config; |
| 13 | use MediaWiki\Http\HttpRequestFactory; |
| 14 | use MediaWiki\Json\FormatJson; |
| 15 | use MWHttpRequest; |
| 16 | use UnexpectedValueException; |
| 17 | use Wikibase\DataModel\Entity\EntityId; |
| 18 | use Wikibase\DataModel\Entity\EntityIdParser; |
| 19 | use Wikibase\DataModel\Entity\EntityIdParsingException; |
| 20 | use Wikibase\DataModel\Entity\EntityIdValue; |
| 21 | use Wikibase\DataModel\Entity\PropertyId; |
| 22 | use Wikibase\DataModel\Services\Lookup\PropertyDataTypeLookup; |
| 23 | use Wikibase\DataModel\Snak\PropertyValueSnak; |
| 24 | use Wikibase\DataModel\Statement\Statement; |
| 25 | use Wikibase\Repo\Rdf\RdfVocabulary; |
| 26 | use WikibaseQuality\ConstraintReport\Api\ExpiryLock; |
| 27 | use WikibaseQuality\ConstraintReport\ConstraintCheck\Cache\CachedBool; |
| 28 | use WikibaseQuality\ConstraintReport\ConstraintCheck\Cache\CachedEntityIds; |
| 29 | use WikibaseQuality\ConstraintReport\ConstraintCheck\Cache\CachedQueryResults; |
| 30 | use WikibaseQuality\ConstraintReport\ConstraintCheck\Cache\CachingMetadata; |
| 31 | use WikibaseQuality\ConstraintReport\ConstraintCheck\Cache\Metadata; |
| 32 | use WikibaseQuality\ConstraintReport\ConstraintCheck\Context\Context; |
| 33 | use WikibaseQuality\ConstraintReport\ConstraintCheck\Message\ViolationMessage; |
| 34 | use WikibaseQuality\ConstraintReport\ConstraintCheck\Message\ViolationMessageDeserializer; |
| 35 | use WikibaseQuality\ConstraintReport\ConstraintCheck\Message\ViolationMessageSerializer; |
| 36 | use WikibaseQuality\ConstraintReport\Role; |
| 37 | use Wikimedia\ObjectCache\WANObjectCache; |
| 38 | use Wikimedia\Stats\IBufferingStatsdDataFactory; |
| 39 | use Wikimedia\Timestamp\ConvertibleTimestamp; |
| 40 | |
| 41 | /** |
| 42 | * Class for running a SPARQL query on some endpoint and getting the results. |
| 43 | * |
| 44 | * @author Lucas Werkmeister |
| 45 | * @license GPL-2.0-or-later |
| 46 | */ |
| 47 | class SparqlHelper { |
| 48 | |
| 49 | private RdfVocabulary $rdfVocabulary; |
| 50 | |
| 51 | /** |
| 52 | * @var string[] |
| 53 | */ |
| 54 | private array $entityPrefixes; |
| 55 | |
| 56 | private string $prefixes; |
| 57 | |
| 58 | private EntityIdParser $entityIdParser; |
| 59 | |
| 60 | private PropertyDataTypeLookup $propertyDataTypeLookup; |
| 61 | |
| 62 | private WANObjectCache $cache; |
| 63 | |
| 64 | private ViolationMessageSerializer $violationMessageSerializer; |
| 65 | |
| 66 | private ViolationMessageDeserializer $violationMessageDeserializer; |
| 67 | |
| 68 | private IBufferingStatsdDataFactory $dataFactory; |
| 69 | |
| 70 | private LoggingHelper $loggingHelper; |
| 71 | |
| 72 | private string $defaultUserAgent; |
| 73 | |
| 74 | private ExpiryLock $throttlingLock; |
| 75 | |
| 76 | /** |
| 77 | * @var int stands for: No Retry-After header-field was sent back |
| 78 | */ |
| 79 | private const NO_RETRY_AFTER = -1; |
| 80 | /** |
| 81 | * @var int stands for: Empty Retry-After header-field was sent back |
| 82 | */ |
| 83 | private const EMPTY_RETRY_AFTER = -2; |
| 84 | /** |
| 85 | * @var int stands for: Invalid Retry-After header-field was sent back |
| 86 | * link a string |
| 87 | */ |
| 88 | private const INVALID_RETRY_AFTER = -3; |
| 89 | /** |
| 90 | * @var string ID on which the lock is applied on |
| 91 | */ |
| 92 | public const EXPIRY_LOCK_ID = 'SparqlHelper.runQuery'; |
| 93 | |
| 94 | /** |
| 95 | * @var int HTTP response code for too many requests |
| 96 | */ |
| 97 | private const HTTP_TOO_MANY_REQUESTS = 429; |
| 98 | |
| 99 | private HttpRequestFactory $requestFactory; |
| 100 | |
| 101 | private string $primaryEndpoint; |
| 102 | |
| 103 | /** |
| 104 | * @var string[] |
| 105 | */ |
| 106 | private array $additionalEndpoints; |
| 107 | |
| 108 | private int $maxQueryTimeMillis; |
| 109 | |
| 110 | private string $subclassOfId; |
| 111 | |
| 112 | private int $cacheMapSize; |
| 113 | |
| 114 | /** |
| 115 | * @var string[] |
| 116 | */ |
| 117 | private array $timeoutExceptionClasses; |
| 118 | |
| 119 | private bool $sparqlHasWikibaseSupport; |
| 120 | |
| 121 | private int $sparqlThrottlingFallbackDuration; |
| 122 | |
| 123 | public function __construct( |
| 124 | Config $config, |
| 125 | RdfVocabulary $rdfVocabulary, |
| 126 | EntityIdParser $entityIdParser, |
| 127 | PropertyDataTypeLookup $propertyDataTypeLookup, |
| 128 | WANObjectCache $cache, |
| 129 | ViolationMessageSerializer $violationMessageSerializer, |
| 130 | ViolationMessageDeserializer $violationMessageDeserializer, |
| 131 | IBufferingStatsdDataFactory $dataFactory, |
| 132 | ExpiryLock $throttlingLock, |
| 133 | LoggingHelper $loggingHelper, |
| 134 | $defaultUserAgent, |
| 135 | HttpRequestFactory $requestFactory |
| 136 | ) { |
| 137 | $this->rdfVocabulary = $rdfVocabulary; |
| 138 | $this->entityIdParser = $entityIdParser; |
| 139 | $this->propertyDataTypeLookup = $propertyDataTypeLookup; |
| 140 | $this->cache = $cache; |
| 141 | $this->violationMessageSerializer = $violationMessageSerializer; |
| 142 | $this->violationMessageDeserializer = $violationMessageDeserializer; |
| 143 | $this->dataFactory = $dataFactory; |
| 144 | $this->throttlingLock = $throttlingLock; |
| 145 | $this->loggingHelper = $loggingHelper; |
| 146 | $this->defaultUserAgent = $defaultUserAgent; |
| 147 | $this->requestFactory = $requestFactory; |
| 148 | $this->entityPrefixes = []; |
| 149 | foreach ( $rdfVocabulary->entityNamespaceNames as $namespaceName ) { |
| 150 | $this->entityPrefixes[] = $rdfVocabulary->getNamespaceURI( $namespaceName ); |
| 151 | } |
| 152 | |
| 153 | $this->primaryEndpoint = $config->get( 'WBQualityConstraintsSparqlEndpoint' ); |
| 154 | $this->additionalEndpoints = $config->get( 'WBQualityConstraintsAdditionalSparqlEndpoints' ) ?: []; |
| 155 | $this->maxQueryTimeMillis = $config->get( 'WBQualityConstraintsSparqlMaxMillis' ); |
| 156 | $this->subclassOfId = $config->get( 'WBQualityConstraintsSubclassOfId' ); |
| 157 | $this->cacheMapSize = $config->get( 'WBQualityConstraintsFormatCacheMapSize' ); |
| 158 | $this->timeoutExceptionClasses = $config->get( |
| 159 | 'WBQualityConstraintsSparqlTimeoutExceptionClasses' |
| 160 | ); |
| 161 | $this->sparqlHasWikibaseSupport = $config->get( |
| 162 | 'WBQualityConstraintsSparqlHasWikibaseSupport' |
| 163 | ); |
| 164 | $this->sparqlThrottlingFallbackDuration = (int)$config->get( |
| 165 | 'WBQualityConstraintsSparqlThrottlingFallbackDuration' |
| 166 | ); |
| 167 | |
| 168 | $this->prefixes = $this->getQueryPrefixes( $rdfVocabulary ); |
| 169 | } |
| 170 | |
| 171 | private function getQueryPrefixes( RdfVocabulary $rdfVocabulary ): string { |
| 172 | // TODO: it would probably be smarter that RdfVocabulary exposed these prefixes somehow |
| 173 | $prefixes = ''; |
| 174 | foreach ( $rdfVocabulary->entityNamespaceNames as $sourceName => $namespaceName ) { |
| 175 | $prefixes .= <<<END |
| 176 | PREFIX {$namespaceName}: <{$rdfVocabulary->getNamespaceURI( $namespaceName )}>\n |
| 177 | END; |
| 178 | } |
| 179 | $prefixes .= <<<END |
| 180 | PREFIX wds: <{$rdfVocabulary->getNamespaceURI( RdfVocabulary::NS_STATEMENT )}> |
| 181 | PREFIX wdv: <{$rdfVocabulary->getNamespaceURI( RdfVocabulary::NS_VALUE )}>\n |
| 182 | END; |
| 183 | |
| 184 | foreach ( $rdfVocabulary->propertyNamespaceNames as $sourceName => $sourceNamespaces ) { |
| 185 | $namespaceName = $sourceNamespaces[RdfVocabulary::NSP_DIRECT_CLAIM]; |
| 186 | $prefixes .= <<<END |
| 187 | PREFIX {$namespaceName}: <{$rdfVocabulary->getNamespaceURI( $namespaceName )}>\n |
| 188 | END; |
| 189 | $namespaceName = $sourceNamespaces[RdfVocabulary::NSP_CLAIM]; |
| 190 | $prefixes .= <<<END |
| 191 | PREFIX {$namespaceName}: <{$rdfVocabulary->getNamespaceURI( $namespaceName )}>\n |
| 192 | END; |
| 193 | $namespaceName = $sourceNamespaces[RdfVocabulary::NSP_CLAIM_STATEMENT]; |
| 194 | $prefixes .= <<<END |
| 195 | PREFIX {$namespaceName}: <{$rdfVocabulary->getNamespaceURI( $namespaceName )}>\n |
| 196 | END; |
| 197 | $namespaceName = $sourceNamespaces[RdfVocabulary::NSP_QUALIFIER]; |
| 198 | $prefixes .= <<<END |
| 199 | PREFIX {$namespaceName}: <{$rdfVocabulary->getNamespaceURI( $namespaceName )}>\n |
| 200 | END; |
| 201 | $namespaceName = $sourceNamespaces[RdfVocabulary::NSP_QUALIFIER_VALUE]; |
| 202 | $prefixes .= <<<END |
| 203 | PREFIX {$namespaceName}: <{$rdfVocabulary->getNamespaceURI( $namespaceName )}>\n |
| 204 | END; |
| 205 | $namespaceName = $sourceNamespaces[RdfVocabulary::NSP_REFERENCE]; |
| 206 | $prefixes .= <<<END |
| 207 | PREFIX {$namespaceName}: <{$rdfVocabulary->getNamespaceURI( $namespaceName )}>\n |
| 208 | END; |
| 209 | $namespaceName = $sourceNamespaces[RdfVocabulary::NSP_REFERENCE_VALUE]; |
| 210 | $prefixes .= <<<END |
| 211 | PREFIX {$namespaceName}: <{$rdfVocabulary->getNamespaceURI( $namespaceName )}>\n |
| 212 | END; |
| 213 | } |
| 214 | $prefixes .= <<<END |
| 215 | PREFIX wikibase: <{$rdfVocabulary->getNamespaceURI( RdfVocabulary::NS_ONTOLOGY )}>\n |
| 216 | END; |
| 217 | return $prefixes; |
| 218 | } |
| 219 | |
| 220 | /** |
| 221 | * @param string $id entity ID serialization of the entity to check |
| 222 | * @param string[] $classes entity ID serializations of the expected types |
| 223 | * |
| 224 | * @return CachedBool |
| 225 | * @throws SparqlHelperException if the query times out or some other error occurs |
| 226 | */ |
| 227 | public function hasType( string $id, array $classes ): CachedBool { |
| 228 | // TODO hint:gearing is a workaround for T168973 and can hopefully be removed eventually |
| 229 | $gearingHint = $this->sparqlHasWikibaseSupport ? |
| 230 | ' hint:Prior hint:gearing "forward".' : |
| 231 | ''; |
| 232 | |
| 233 | $metadatas = []; |
| 234 | |
| 235 | foreach ( array_chunk( $classes, 20 ) as $classesChunk ) { |
| 236 | $classesValues = implode( ' ', array_map( |
| 237 | static function ( $class ) { |
| 238 | return 'wd:' . $class; |
| 239 | }, |
| 240 | $classesChunk |
| 241 | ) ); |
| 242 | |
| 243 | $query = <<<EOF |
| 244 | ASK { |
| 245 | BIND(wd:$id AS ?item) |
| 246 | VALUES ?class { $classesValues } |
| 247 | ?item wdt:{$this->subclassOfId}* ?class.$gearingHint |
| 248 | } |
| 249 | EOF; |
| 250 | |
| 251 | $result = $this->runQuery( $query, $this->primaryEndpoint ); |
| 252 | $metadatas[] = $result->getMetadata(); |
| 253 | if ( $result->getArray()['boolean'] ) { |
| 254 | return new CachedBool( |
| 255 | true, |
| 256 | Metadata::merge( $metadatas ) |
| 257 | ); |
| 258 | } |
| 259 | } |
| 260 | |
| 261 | return new CachedBool( |
| 262 | false, |
| 263 | Metadata::merge( $metadatas ) |
| 264 | ); |
| 265 | } |
| 266 | |
| 267 | /** |
| 268 | * Helper function used by findEntitiesWithSameStatement to filter |
| 269 | * out entities with different qualifiers or no qualifier value. |
| 270 | */ |
| 271 | private function nestedSeparatorFilter( PropertyId $separator ): string { |
| 272 | $filter = <<<EOF |
| 273 | MINUS { |
| 274 | ?statement pq:$separator ?qualifier. |
| 275 | FILTER NOT EXISTS { |
| 276 | ?otherStatement pq:$separator ?qualifier. |
| 277 | } |
| 278 | } |
| 279 | MINUS { |
| 280 | ?otherStatement pq:$separator ?qualifier. |
| 281 | FILTER NOT EXISTS { |
| 282 | ?statement pq:$separator ?qualifier. |
| 283 | } |
| 284 | } |
| 285 | MINUS { |
| 286 | ?statement a wdno:$separator. |
| 287 | FILTER NOT EXISTS { |
| 288 | ?otherStatement a wdno:$separator. |
| 289 | } |
| 290 | } |
| 291 | MINUS { |
| 292 | ?otherStatement a wdno:$separator. |
| 293 | FILTER NOT EXISTS { |
| 294 | ?statement a wdno:$separator. |
| 295 | } |
| 296 | } |
| 297 | EOF; |
| 298 | return $filter; |
| 299 | } |
| 300 | |
| 301 | /** |
| 302 | * @param Statement $statement |
| 303 | * @param PropertyId[] $separators |
| 304 | * |
| 305 | * @return CachedEntityIds |
| 306 | * @throws SparqlHelperException if the query times out or some other error occurs |
| 307 | */ |
| 308 | public function findEntitiesWithSameStatement( Statement $statement, array $separators ): CachedEntityIds { |
| 309 | $pid = $statement->getPropertyId()->getSerialization(); |
| 310 | $guid = $statement->getGuid(); |
| 311 | '@phan-var string $guid'; // statement must have a non-null GUID |
| 312 | $guidForRdf = str_replace( '$', '-', $guid ); |
| 313 | |
| 314 | $separatorFilters = array_map( [ $this, 'nestedSeparatorFilter' ], $separators ); |
| 315 | $finalSeparatorFilter = implode( "\n", $separatorFilters ); |
| 316 | |
| 317 | $query = <<<EOF |
| 318 | SELECT DISTINCT ?otherEntity WHERE { |
| 319 | BIND(wds:$guidForRdf AS ?statement) |
| 320 | BIND(p:$pid AS ?p) |
| 321 | BIND(ps:$pid AS ?ps) |
| 322 | ?entity ?p ?statement. |
| 323 | ?statement ?ps ?value. |
| 324 | ?otherStatement ?ps ?value. |
| 325 | ?otherEntity ?p ?otherStatement. |
| 326 | FILTER(?otherEntity != ?entity) |
| 327 | MINUS { ?otherStatement wikibase:rank wikibase:DeprecatedRank. } |
| 328 | $finalSeparatorFilter |
| 329 | } |
| 330 | LIMIT 10 |
| 331 | EOF; |
| 332 | |
| 333 | $results = [ $this->runQuery( $query, $this->primaryEndpoint ) ]; |
| 334 | foreach ( $this->additionalEndpoints as $endpoint ) { |
| 335 | $results[] = $this->runQuery( $query, $endpoint ); |
| 336 | } |
| 337 | |
| 338 | return $this->getOtherEntities( $results ); |
| 339 | } |
| 340 | |
| 341 | /** |
| 342 | * @param EntityId $entityId The entity ID on the containing entity |
| 343 | * @param PropertyValueSnak $snak |
| 344 | * @param string $type Context::TYPE_QUALIFIER or Context::TYPE_REFERENCE |
| 345 | * @param boolean $ignoreDeprecatedStatements Whether to ignore deprecated statements or not. |
| 346 | * |
| 347 | * @return CachedEntityIds |
| 348 | * @throws SparqlHelperException if the query times out or some other error occurs |
| 349 | */ |
| 350 | public function findEntitiesWithSameQualifierOrReference( |
| 351 | EntityId $entityId, |
| 352 | PropertyValueSnak $snak, |
| 353 | string $type, |
| 354 | bool $ignoreDeprecatedStatements |
| 355 | ): CachedEntityIds { |
| 356 | $eid = $entityId->getSerialization(); |
| 357 | $pid = $snak->getPropertyId()->getSerialization(); |
| 358 | $prefix = $type === Context::TYPE_QUALIFIER ? 'pq' : 'pr'; |
| 359 | $dataValue = $snak->getDataValue(); |
| 360 | $dataType = $this->propertyDataTypeLookup->getDataTypeIdForProperty( |
| 361 | $snak->getPropertyId() |
| 362 | ); |
| 363 | [ $value, $isFullValue ] = $this->getRdfLiteral( $dataType, $dataValue ); |
| 364 | if ( $isFullValue ) { |
| 365 | $prefix .= 'v'; |
| 366 | } |
| 367 | $path = $type === Context::TYPE_QUALIFIER ? |
| 368 | "$prefix:$pid" : |
| 369 | "prov:wasDerivedFrom/$prefix:$pid"; |
| 370 | |
| 371 | $deprecatedFilter = ''; |
| 372 | if ( $ignoreDeprecatedStatements ) { |
| 373 | $deprecatedFilter = <<< EOF |
| 374 | MINUS { ?otherStatement wikibase:rank wikibase:DeprecatedRank. } |
| 375 | EOF; |
| 376 | } |
| 377 | |
| 378 | $query = <<<EOF |
| 379 | SELECT DISTINCT ?otherEntity WHERE { |
| 380 | BIND(wd:$eid AS ?entity) |
| 381 | BIND($value AS ?value) |
| 382 | ?entity ?p ?statement. |
| 383 | ?statement $path ?value. |
| 384 | ?otherStatement $path ?value. |
| 385 | ?otherEntity ?otherP ?otherStatement. |
| 386 | FILTER(?otherEntity != ?entity) |
| 387 | $deprecatedFilter |
| 388 | } |
| 389 | LIMIT 10 |
| 390 | EOF; |
| 391 | |
| 392 | $results = [ $this->runQuery( $query, $this->primaryEndpoint ) ]; |
| 393 | foreach ( $this->additionalEndpoints as $endpoint ) { |
| 394 | $results[] = $this->runQuery( $query, $endpoint ); |
| 395 | } |
| 396 | |
| 397 | return $this->getOtherEntities( $results ); |
| 398 | } |
| 399 | |
| 400 | /** |
| 401 | * Return SPARQL code for a string literal with $text as content. |
| 402 | */ |
| 403 | private function stringLiteral( string $text ): string { |
| 404 | return '"' . strtr( $text, [ '"' => '\\"', '\\' => '\\\\' ] ) . '"'; |
| 405 | } |
| 406 | |
| 407 | /** |
| 408 | * Extract and parse entity IDs from the ?otherEntity column of SPARQL query results. |
| 409 | * |
| 410 | * @param CachedQueryResults[] $results |
| 411 | * |
| 412 | * @return CachedEntityIds |
| 413 | */ |
| 414 | private function getOtherEntities( array $results ): CachedEntityIds { |
| 415 | $allResultBindings = []; |
| 416 | $metadatas = []; |
| 417 | |
| 418 | foreach ( $results as $result ) { |
| 419 | $metadatas[] = $result->getMetadata(); |
| 420 | $allResultBindings = array_merge( $allResultBindings, $result->getArray()['results']['bindings'] ); |
| 421 | } |
| 422 | |
| 423 | $entityIds = array_map( |
| 424 | function ( $resultBindings ) { |
| 425 | $entityIRI = $resultBindings['otherEntity']['value']; |
| 426 | foreach ( $this->entityPrefixes as $entityPrefix ) { |
| 427 | $entityPrefixLength = strlen( $entityPrefix ); |
| 428 | if ( substr( $entityIRI, 0, $entityPrefixLength ) === $entityPrefix ) { |
| 429 | try { |
| 430 | return $this->entityIdParser->parse( |
| 431 | substr( $entityIRI, $entityPrefixLength ) |
| 432 | ); |
| 433 | } catch ( EntityIdParsingException $e ) { |
| 434 | // fall through |
| 435 | } |
| 436 | } |
| 437 | |
| 438 | return null; |
| 439 | } |
| 440 | |
| 441 | return null; |
| 442 | }, |
| 443 | $allResultBindings |
| 444 | ); |
| 445 | |
| 446 | return new CachedEntityIds( |
| 447 | array_values( array_filter( array_unique( $entityIds ) ) ), |
| 448 | Metadata::merge( $metadatas ) |
| 449 | ); |
| 450 | } |
| 451 | |
| 452 | // phpcs:disable Generic.Metrics.CyclomaticComplexity,Squiz.WhiteSpace.FunctionSpacing |
| 453 | /** |
| 454 | * Get an RDF literal or IRI with which the given data value can be matched in a query. |
| 455 | * |
| 456 | * @return array the literal or IRI as a string in SPARQL syntax, |
| 457 | * and a boolean indicating whether it refers to a full value node or not |
| 458 | */ |
| 459 | private function getRdfLiteral( string $dataType, DataValue $dataValue ): array { |
| 460 | switch ( $dataType ) { |
| 461 | case 'string': |
| 462 | case 'external-id': |
| 463 | return [ $this->stringLiteral( $dataValue->getValue() ), false ]; |
| 464 | case 'commonsMedia': |
| 465 | $url = $this->rdfVocabulary->getMediaFileURI( $dataValue->getValue() ); |
| 466 | return [ '<' . $url . '>', false ]; |
| 467 | case 'geo-shape': |
| 468 | $url = $this->rdfVocabulary->getGeoShapeURI( $dataValue->getValue() ); |
| 469 | return [ '<' . $url . '>', false ]; |
| 470 | case 'tabular-data': |
| 471 | $url = $this->rdfVocabulary->getTabularDataURI( $dataValue->getValue() ); |
| 472 | return [ '<' . $url . '>', false ]; |
| 473 | case 'url': |
| 474 | $url = $dataValue->getValue(); |
| 475 | if ( !preg_match( '/^[^<>"{}\\\\|^`\\x00-\\x20]*$/D', $url ) ) { |
| 476 | // not a valid URL for SPARQL (see SPARQL spec, production 139 IRIREF) |
| 477 | // such an URL should never reach us, so just throw |
| 478 | throw new InvalidArgumentException( 'invalid URL: ' . $url ); |
| 479 | } |
| 480 | return [ '<' . $url . '>', false ]; |
| 481 | case 'wikibase-item': |
| 482 | case 'wikibase-property': |
| 483 | /** @var EntityIdValue $dataValue */ |
| 484 | '@phan-var EntityIdValue $dataValue'; |
| 485 | return [ 'wd:' . $dataValue->getEntityId()->getSerialization(), false ]; |
| 486 | case 'monolingualtext': |
| 487 | /** @var MonolingualTextValue $dataValue */ |
| 488 | '@phan-var MonolingualTextValue $dataValue'; |
| 489 | $lang = $dataValue->getLanguageCode(); |
| 490 | if ( !preg_match( '/^[a-zA-Z]+(-[a-zA-Z0-9]+)*$/D', $lang ) ) { |
| 491 | // not a valid language tag for SPARQL (see SPARQL spec, production 145 LANGTAG) |
| 492 | // such a language tag should never reach us, so just throw |
| 493 | throw new InvalidArgumentException( 'invalid language tag: ' . $lang ); |
| 494 | } |
| 495 | return [ $this->stringLiteral( $dataValue->getText() ) . '@' . $lang, false ]; |
| 496 | case 'globe-coordinate': |
| 497 | case 'quantity': |
| 498 | case 'time': |
| 499 | // @phan-suppress-next-line PhanUndeclaredMethod |
| 500 | return [ 'wdv:' . $dataValue->getHash(), true ]; |
| 501 | default: |
| 502 | throw new InvalidArgumentException( 'unknown data type: ' . $dataType ); |
| 503 | } |
| 504 | } |
| 505 | // phpcs:enable |
| 506 | |
| 507 | /** |
| 508 | * @throws SparqlHelperException if the query times out or some other error occurs |
| 509 | * @throws ConstraintParameterException if the $regex is invalid |
| 510 | */ |
| 511 | public function matchesRegularExpression( string $text, string $regex ): bool { |
| 512 | // caching wrapper around matchesRegularExpressionWithSparql |
| 513 | |
| 514 | $textHash = hash( 'sha256', $text ); |
| 515 | $cacheKey = $this->cache->makeKey( |
| 516 | 'WikibaseQualityConstraints', // extension |
| 517 | 'regex', // action |
| 518 | 'WDQS-Java', // regex flavor |
| 519 | hash( 'sha256', $regex ) |
| 520 | ); |
| 521 | |
| 522 | $cacheMapArray = $this->cache->getWithSetCallback( |
| 523 | $cacheKey, |
| 524 | WANObjectCache::TTL_DAY, |
| 525 | function ( $cacheMapArray ) use ( $text, $regex, $textHash ) { |
| 526 | // Initialize the cache map if not set |
| 527 | if ( $cacheMapArray === false ) { |
| 528 | $key = 'wikibase.quality.constraints.regex.cache.refresh.init'; |
| 529 | $this->dataFactory->increment( $key ); |
| 530 | return []; |
| 531 | } |
| 532 | |
| 533 | $key = 'wikibase.quality.constraints.regex.cache.refresh'; |
| 534 | $this->dataFactory->increment( $key ); |
| 535 | $cacheMap = MapCacheLRU::newFromArray( $cacheMapArray, $this->cacheMapSize ); |
| 536 | if ( $cacheMap->has( $textHash ) ) { |
| 537 | $key = 'wikibase.quality.constraints.regex.cache.refresh.hit'; |
| 538 | $this->dataFactory->increment( $key ); |
| 539 | $cacheMap->get( $textHash ); // ping cache |
| 540 | } else { |
| 541 | $key = 'wikibase.quality.constraints.regex.cache.refresh.miss'; |
| 542 | $this->dataFactory->increment( $key ); |
| 543 | try { |
| 544 | $matches = $this->matchesRegularExpressionWithSparql( $text, $regex ); |
| 545 | } catch ( ConstraintParameterException $e ) { |
| 546 | $matches = $this->serializeConstraintParameterException( $e ); |
| 547 | } catch ( SparqlHelperException $e ) { |
| 548 | // don’t cache this |
| 549 | return $cacheMap->toArray(); |
| 550 | } |
| 551 | $cacheMap->set( |
| 552 | $textHash, |
| 553 | $matches, |
| 554 | 3 / 8 |
| 555 | ); |
| 556 | } |
| 557 | |
| 558 | return $cacheMap->toArray(); |
| 559 | }, |
| 560 | [ |
| 561 | // Once map is > 1 sec old, consider refreshing |
| 562 | 'ageNew' => 1, |
| 563 | // Update 5 seconds after "ageNew" given a 1 query/sec cache check rate |
| 564 | 'hotTTR' => 5, |
| 565 | // avoid querying cache servers multiple times in a request |
| 566 | // (e. g. when checking format of a reference URL used multiple times on an entity) |
| 567 | 'pcTTL' => WANObjectCache::TTL_PROC_LONG, |
| 568 | ] |
| 569 | ); |
| 570 | |
| 571 | if ( isset( $cacheMapArray[$textHash] ) ) { |
| 572 | $key = 'wikibase.quality.constraints.regex.cache.hit'; |
| 573 | $this->dataFactory->increment( $key ); |
| 574 | $matches = $cacheMapArray[$textHash]; |
| 575 | if ( is_bool( $matches ) ) { |
| 576 | return $matches; |
| 577 | } elseif ( is_array( $matches ) && |
| 578 | $matches['type'] == ConstraintParameterException::class ) { |
| 579 | throw $this->deserializeConstraintParameterException( $matches ); |
| 580 | } else { |
| 581 | throw new UnexpectedValueException( |
| 582 | 'Value of unknown type in object cache (' . |
| 583 | 'cache key: ' . $cacheKey . ', ' . |
| 584 | 'cache map key: ' . $textHash . ', ' . |
| 585 | 'value type: ' . get_debug_type( $matches ) . ')' |
| 586 | ); |
| 587 | } |
| 588 | } else { |
| 589 | $key = 'wikibase.quality.constraints.regex.cache.miss'; |
| 590 | $this->dataFactory->increment( $key ); |
| 591 | return $this->matchesRegularExpressionWithSparql( $text, $regex ); |
| 592 | } |
| 593 | } |
| 594 | |
| 595 | private function serializeConstraintParameterException( ConstraintParameterException $cpe ): array { |
| 596 | return [ |
| 597 | 'type' => ConstraintParameterException::class, |
| 598 | 'violationMessage' => $this->violationMessageSerializer->serialize( $cpe->getViolationMessage() ), |
| 599 | ]; |
| 600 | } |
| 601 | |
| 602 | private function deserializeConstraintParameterException( array $serialization ): ConstraintParameterException { |
| 603 | $message = $this->violationMessageDeserializer->deserialize( |
| 604 | $serialization['violationMessage'] |
| 605 | ); |
| 606 | return new ConstraintParameterException( $message ); |
| 607 | } |
| 608 | |
| 609 | /** |
| 610 | * This function is only public for testing purposes; |
| 611 | * use matchesRegularExpression, which is equivalent but caches results. |
| 612 | * |
| 613 | * @throws SparqlHelperException if the query times out or some other error occurs |
| 614 | * @throws ConstraintParameterException if the $regex is invalid |
| 615 | */ |
| 616 | public function matchesRegularExpressionWithSparql( string $text, string $regex ): bool { |
| 617 | $textStringLiteral = $this->stringLiteral( $text ); |
| 618 | $regexStringLiteral = $this->stringLiteral( '^(?:' . $regex . ')$' ); |
| 619 | |
| 620 | $query = <<<EOF |
| 621 | SELECT (REGEX($textStringLiteral, $regexStringLiteral) AS ?matches) {} |
| 622 | EOF; |
| 623 | |
| 624 | $result = $this->runQuery( $query, $this->primaryEndpoint, false ); |
| 625 | |
| 626 | $vars = $result->getArray()['results']['bindings'][0]; |
| 627 | if ( array_key_exists( 'matches', $vars ) ) { |
| 628 | // true or false ⇒ regex okay, text matches or not |
| 629 | return $vars['matches']['value'] === 'true'; |
| 630 | } else { |
| 631 | // empty result: regex broken |
| 632 | throw new ConstraintParameterException( |
| 633 | ( new ViolationMessage( 'wbqc-violation-message-parameter-regex' ) ) |
| 634 | ->withInlineCode( $regex, Role::CONSTRAINT_PARAMETER_VALUE ) |
| 635 | ); |
| 636 | } |
| 637 | } |
| 638 | |
| 639 | /** |
| 640 | * Check whether the text content of an error response indicates a query timeout. |
| 641 | */ |
| 642 | public function isTimeout( string $responseContent ): bool { |
| 643 | $timeoutRegex = implode( '|', array_map( |
| 644 | static function ( $fqn ) { |
| 645 | return preg_quote( $fqn, '/' ); |
| 646 | }, |
| 647 | $this->timeoutExceptionClasses |
| 648 | ) ); |
| 649 | return (bool)preg_match( '/' . $timeoutRegex . '/', $responseContent ); |
| 650 | } |
| 651 | |
| 652 | /** |
| 653 | * Return the max-age of a cached response, |
| 654 | * or a boolean indicating whether the response was cached or not. |
| 655 | * |
| 656 | * @param array $responseHeaders see MWHttpRequest::getResponseHeaders() |
| 657 | * |
| 658 | * @return int|boolean the max-age (in seconds) |
| 659 | * or a plain boolean if no max-age can be determined |
| 660 | */ |
| 661 | public function getCacheMaxAge( array $responseHeaders ) { |
| 662 | if ( |
| 663 | array_key_exists( 'x-cache-status', $responseHeaders ) && |
| 664 | preg_match( '/^hit(?:-.*)?$/', $responseHeaders['x-cache-status'][0] ) |
| 665 | ) { |
| 666 | $maxage = []; |
| 667 | if ( |
| 668 | array_key_exists( 'cache-control', $responseHeaders ) && |
| 669 | preg_match( '/\bmax-age=(\d+)\b/', $responseHeaders['cache-control'][0], $maxage ) |
| 670 | ) { |
| 671 | return intval( $maxage[1] ); |
| 672 | } else { |
| 673 | return true; |
| 674 | } |
| 675 | } else { |
| 676 | return false; |
| 677 | } |
| 678 | } |
| 679 | |
| 680 | /** |
| 681 | * Get the delay date of a 429 headered response, which is caused by |
| 682 | * throttling of to many SPARQL-Requests. The header-format is defined |
| 683 | * in RFC7231 see: https://tools.ietf.org/html/rfc7231#section-7.1.3 |
| 684 | * |
| 685 | * @param MWHttpRequest $request |
| 686 | * |
| 687 | * @return int|ConvertibleTimestamp |
| 688 | * or SparlHelper::NO_RETRY_AFTER if there is no Retry-After header |
| 689 | * or SparlHelper::EMPTY_RETRY_AFTER if there is an empty Retry-After |
| 690 | * or SparlHelper::INVALID_RETRY_AFTER if there is something wrong with the format |
| 691 | */ |
| 692 | public function getThrottling( MWHttpRequest $request ) { |
| 693 | $retryAfterValue = $request->getResponseHeader( 'Retry-After' ); |
| 694 | if ( $retryAfterValue === null ) { |
| 695 | return self::NO_RETRY_AFTER; |
| 696 | } |
| 697 | |
| 698 | $trimmedRetryAfterValue = trim( $retryAfterValue ); |
| 699 | if ( $trimmedRetryAfterValue === '' ) { |
| 700 | return self::EMPTY_RETRY_AFTER; |
| 701 | } |
| 702 | |
| 703 | if ( is_numeric( $trimmedRetryAfterValue ) ) { |
| 704 | $delaySeconds = (int)$trimmedRetryAfterValue; |
| 705 | if ( $delaySeconds >= 0 ) { |
| 706 | return $this->getTimestampInFuture( new DateInterval( 'PT' . $delaySeconds . 'S' ) ); |
| 707 | } |
| 708 | } else { |
| 709 | $return = strtotime( $trimmedRetryAfterValue ); |
| 710 | if ( $return !== false ) { |
| 711 | return new ConvertibleTimestamp( $return ); |
| 712 | } |
| 713 | } |
| 714 | return self::INVALID_RETRY_AFTER; |
| 715 | } |
| 716 | |
| 717 | private function getTimestampInFuture( DateInterval $delta ) { |
| 718 | $now = new ConvertibleTimestamp(); |
| 719 | return new ConvertibleTimestamp( $now->timestamp->add( $delta ) ); |
| 720 | } |
| 721 | |
| 722 | /** |
| 723 | * Runs a query against the configured endpoint and returns the results. |
| 724 | * TODO: See if Sparql Client in core can be used instead of rolling our own |
| 725 | * |
| 726 | * @param string $query The query, unencoded (plain string). |
| 727 | * @param string $endpoint The endpoint to query. |
| 728 | * @param bool $needsPrefixes Whether the query requires prefixes or they can be omitted. |
| 729 | * |
| 730 | * @return CachedQueryResults |
| 731 | * |
| 732 | * @throws SparqlHelperException if the query times out or some other error occurs |
| 733 | */ |
| 734 | protected function runQuery( string $query, string $endpoint, bool $needsPrefixes = true ): CachedQueryResults { |
| 735 | if ( $this->throttlingLock->isLocked( self::EXPIRY_LOCK_ID ) ) { |
| 736 | $this->dataFactory->increment( 'wikibase.quality.constraints.sparql.throttling' ); |
| 737 | throw new TooManySparqlRequestsException(); |
| 738 | } |
| 739 | |
| 740 | if ( $this->sparqlHasWikibaseSupport ) { |
| 741 | $needsPrefixes = false; |
| 742 | } |
| 743 | |
| 744 | if ( $needsPrefixes ) { |
| 745 | $query = $this->prefixes . $query; |
| 746 | } |
| 747 | $query = "#wbqc\n" . $query; |
| 748 | |
| 749 | $url = $endpoint . '?' . http_build_query( |
| 750 | [ |
| 751 | 'query' => $query, |
| 752 | 'format' => 'json', |
| 753 | 'maxQueryTimeMillis' => $this->maxQueryTimeMillis, |
| 754 | ], |
| 755 | '', ini_get( 'arg_separator.output' ), |
| 756 | // encode spaces with %20, not + |
| 757 | PHP_QUERY_RFC3986 |
| 758 | ); |
| 759 | |
| 760 | $options = [ |
| 761 | 'method' => 'GET', |
| 762 | 'timeout' => (int)round( ( $this->maxQueryTimeMillis + 1000 ) / 1000 ), |
| 763 | 'connectTimeout' => 'default', |
| 764 | 'userAgent' => $this->defaultUserAgent, |
| 765 | ]; |
| 766 | $request = $this->requestFactory->create( $url, $options, __METHOD__ ); |
| 767 | $startTime = microtime( true ); |
| 768 | $requestStatus = $request->execute(); |
| 769 | $endTime = microtime( true ); |
| 770 | $this->dataFactory->timing( |
| 771 | 'wikibase.quality.constraints.sparql.timing', |
| 772 | ( $endTime - $startTime ) * 1000 |
| 773 | ); |
| 774 | |
| 775 | $this->guardAgainstTooManyRequestsError( $request ); |
| 776 | |
| 777 | $maxAge = $this->getCacheMaxAge( $request->getResponseHeaders() ); |
| 778 | if ( $maxAge ) { |
| 779 | $this->dataFactory->increment( 'wikibase.quality.constraints.sparql.cached' ); |
| 780 | } |
| 781 | |
| 782 | if ( $requestStatus->isOK() ) { |
| 783 | $json = $request->getContent(); |
| 784 | $jsonStatus = FormatJson::parse( $json, FormatJson::FORCE_ASSOC ); |
| 785 | if ( $jsonStatus->isOK() ) { |
| 786 | return new CachedQueryResults( |
| 787 | $jsonStatus->getValue(), |
| 788 | Metadata::ofCachingMetadata( |
| 789 | $maxAge ? |
| 790 | CachingMetadata::ofMaximumAgeInSeconds( $maxAge ) : |
| 791 | CachingMetadata::fresh() |
| 792 | ) |
| 793 | ); |
| 794 | } else { |
| 795 | $jsonErrorCode = $jsonStatus->getErrors()[0]['message']; |
| 796 | $this->dataFactory->increment( |
| 797 | "wikibase.quality.constraints.sparql.error.json.$jsonErrorCode" |
| 798 | ); |
| 799 | // fall through to general error handling |
| 800 | } |
| 801 | } else { |
| 802 | $this->dataFactory->increment( |
| 803 | "wikibase.quality.constraints.sparql.error.http.{$request->getStatus()}" |
| 804 | ); |
| 805 | // fall through to general error handling |
| 806 | } |
| 807 | |
| 808 | $this->dataFactory->increment( 'wikibase.quality.constraints.sparql.error' ); |
| 809 | |
| 810 | if ( $this->isTimeout( $request->getContent() ) ) { |
| 811 | $this->dataFactory->increment( |
| 812 | 'wikibase.quality.constraints.sparql.error.timeout' |
| 813 | ); |
| 814 | } |
| 815 | |
| 816 | throw new SparqlHelperException(); |
| 817 | } |
| 818 | |
| 819 | /** |
| 820 | * Handle a potential “too many requests” error. |
| 821 | * |
| 822 | * @param MWHttpRequest $request |
| 823 | * @throws TooManySparqlRequestsException |
| 824 | */ |
| 825 | private function guardAgainstTooManyRequestsError( MWHttpRequest $request ): void { |
| 826 | if ( $request->getStatus() !== self::HTTP_TOO_MANY_REQUESTS ) { |
| 827 | return; |
| 828 | } |
| 829 | |
| 830 | $fallbackBlockDuration = $this->sparqlThrottlingFallbackDuration; |
| 831 | |
| 832 | if ( $fallbackBlockDuration < 0 ) { |
| 833 | throw new InvalidArgumentException( 'Fallback duration must be positive int but is: ' . |
| 834 | $fallbackBlockDuration ); |
| 835 | } |
| 836 | |
| 837 | $this->dataFactory->increment( 'wikibase.quality.constraints.sparql.throttling' ); |
| 838 | $throttlingUntil = $this->getThrottling( $request ); |
| 839 | if ( !( $throttlingUntil instanceof ConvertibleTimestamp ) ) { |
| 840 | $this->loggingHelper->logSparqlHelperTooManyRequestsRetryAfterInvalid( $request ); |
| 841 | $this->throttlingLock->lock( |
| 842 | self::EXPIRY_LOCK_ID, |
| 843 | $this->getTimestampInFuture( new DateInterval( 'PT' . $fallbackBlockDuration . 'S' ) ) |
| 844 | ); |
| 845 | } else { |
| 846 | $this->loggingHelper->logSparqlHelperTooManyRequestsRetryAfterPresent( $throttlingUntil, $request ); |
| 847 | $this->throttlingLock->lock( self::EXPIRY_LOCK_ID, $throttlingUntil ); |
| 848 | } |
| 849 | throw new TooManySparqlRequestsException(); |
| 850 | } |
| 851 | |
| 852 | } |