Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
92.23% |
95 / 103 |
|
50.00% |
5 / 10 |
CRAP | |
0.00% |
0 / 1 |
| RevisionOutputCache | |
92.23% |
95 / 103 |
|
50.00% |
5 / 10 |
29.39 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
| getContentModelFromRevision | |
66.67% |
2 / 3 |
|
0.00% |
0 / 1 |
2.15 | |||
| incrementStats | |
100.00% |
7 / 7 |
|
100.00% |
1 / 1 |
2 | |||
| incrementRenderReasonStats | |
100.00% |
7 / 7 |
|
100.00% |
1 / 1 |
1 | |||
| makeParserOutputKey | |
83.33% |
5 / 6 |
|
0.00% |
0 / 1 |
2.02 | |||
| makeParserOutputKeyOptionalRevId | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
2 | |||
| get | |
95.65% |
22 / 23 |
|
0.00% |
0 / 1 |
6 | |||
| save | |
96.97% |
32 / 33 |
|
0.00% |
0 / 1 |
10 | |||
| restoreFromJson | |
100.00% |
9 / 9 |
|
100.00% |
1 / 1 |
2 | |||
| encodeAsJson | |
100.00% |
8 / 8 |
|
100.00% |
1 / 1 |
2 | |||
| 1 | <?php |
| 2 | /** |
| 3 | * Cache for outputs of the PHP parser |
| 4 | * |
| 5 | * @license GPL-2.0-or-later |
| 6 | * @file |
| 7 | * @ingroup Cache Parser |
| 8 | */ |
| 9 | |
| 10 | namespace MediaWiki\Parser; |
| 11 | |
| 12 | use InvalidArgumentException; |
| 13 | use JsonException; |
| 14 | use MediaWiki\Json\JsonCodec; |
| 15 | use MediaWiki\Revision\RevisionRecord; |
| 16 | use MediaWiki\Revision\SlotRecord; |
| 17 | use MediaWiki\Utils\MWTimestamp; |
| 18 | use Psr\Log\LoggerInterface; |
| 19 | use Wikimedia\ObjectCache\WANObjectCache; |
| 20 | use Wikimedia\Stats\StatsFactory; |
| 21 | use Wikimedia\Timestamp\TimestampFormat as TS; |
| 22 | use Wikimedia\UUID\GlobalIdGenerator; |
| 23 | |
| 24 | /** |
| 25 | * Cache for ParserOutput objects. |
| 26 | * The cache is split per ParserOptions. |
| 27 | * |
| 28 | * @since 1.36 |
| 29 | * @ingroup Cache Parser |
| 30 | */ |
| 31 | class RevisionOutputCache { |
| 32 | |
| 33 | /** @var string The name of this cache. Used as a root of the cache key. */ |
| 34 | private $name; |
| 35 | |
| 36 | /** |
| 37 | * Anything cached prior to this is invalidated |
| 38 | * |
| 39 | * @var string |
| 40 | */ |
| 41 | private $cacheEpoch; |
| 42 | |
| 43 | /** |
| 44 | * Expiry time for cache entries. |
| 45 | * |
| 46 | * @var int |
| 47 | */ |
| 48 | private $cacheExpiry; |
| 49 | |
| 50 | /** |
| 51 | * @param string $name |
| 52 | * @param WANObjectCache $cache |
| 53 | * @param int $cacheExpiry Expiry for ParserOutput in $cache. |
| 54 | * @param string $cacheEpoch Anything before this timestamp is invalidated |
| 55 | * @param JsonCodec $jsonCodec |
| 56 | * @param StatsFactory $stats |
| 57 | * @param LoggerInterface $logger |
| 58 | * @param GlobalIdGenerator $globalIdGenerator |
| 59 | */ |
| 60 | public function __construct( |
| 61 | string $name, |
| 62 | private readonly WANObjectCache $cache, |
| 63 | int $cacheExpiry, |
| 64 | string $cacheEpoch, |
| 65 | private readonly JsonCodec $jsonCodec, |
| 66 | private readonly StatsFactory $stats, |
| 67 | private readonly LoggerInterface $logger, |
| 68 | private readonly GlobalIdGenerator $globalIdGenerator, |
| 69 | ) { |
| 70 | $this->name = $name; |
| 71 | $this->cacheExpiry = $cacheExpiry; |
| 72 | $this->cacheEpoch = $cacheEpoch; |
| 73 | } |
| 74 | |
| 75 | /** |
| 76 | * @param RevisionRecord $revision |
| 77 | * @return string |
| 78 | */ |
| 79 | private function getContentModelFromRevision( RevisionRecord $revision ) { |
| 80 | if ( !$revision->hasSlot( SlotRecord::MAIN ) ) { |
| 81 | return 'missing'; |
| 82 | } |
| 83 | return str_replace( '.', '_', $revision->getMainContentModel() ); |
| 84 | } |
| 85 | |
| 86 | /** |
| 87 | * @param RevisionRecord $revision |
| 88 | * @param string $status e.g. hit, miss etc. |
| 89 | * @param string|null $reason |
| 90 | */ |
| 91 | private function incrementStats( RevisionRecord $revision, string $status, ?string $reason = null ) { |
| 92 | $contentModel = $this->getContentModelFromRevision( $revision ); |
| 93 | |
| 94 | $this->stats->getCounter( 'RevisionOutputCache_operation_total' ) |
| 95 | ->setLabel( 'name', $this->name ) |
| 96 | ->setLabel( 'contentModel', $contentModel ) |
| 97 | ->setLabel( 'status', $status ) |
| 98 | ->setLabel( 'reason', $reason ?: 'n/a' ) |
| 99 | ->increment(); |
| 100 | } |
| 101 | |
| 102 | /** |
| 103 | * @param RevisionRecord $revision |
| 104 | * @param string $renderReason |
| 105 | */ |
| 106 | private function incrementRenderReasonStats( RevisionRecord $revision, $renderReason ) { |
| 107 | $contentModel = $this->getContentModelFromRevision( $revision ); |
| 108 | $renderReason = preg_replace( '/\W+/', '_', $renderReason ); |
| 109 | |
| 110 | $this->stats->getCounter( 'RevisionOutputCache_render_total' ) |
| 111 | ->setLabel( 'name', $this->name ) |
| 112 | ->setLabel( 'contentModel', $contentModel ) |
| 113 | ->setLabel( 'reason', $renderReason ) |
| 114 | ->increment(); |
| 115 | } |
| 116 | |
| 117 | /** |
| 118 | * Get a key that will be used by this cache to store the content |
| 119 | * for a given page considering the given options and the array of |
| 120 | * used options. |
| 121 | * |
| 122 | * If there is a possibility the revision does not have a revision id, use |
| 123 | * makeParserOutputKeyOptionalRevId() instead. |
| 124 | * |
| 125 | * @warning The exact format of the key is considered internal and is subject |
| 126 | * to change, thus should not be used as storage or long-term caching key. |
| 127 | * This is intended to be used for logging or keying something transient. |
| 128 | * |
| 129 | * @param RevisionRecord $revision |
| 130 | * @param ParserOptions $options |
| 131 | * @param array|null $usedOptions currently ignored |
| 132 | * @return string |
| 133 | * @internal |
| 134 | */ |
| 135 | public function makeParserOutputKey( |
| 136 | RevisionRecord $revision, |
| 137 | ParserOptions $options, |
| 138 | ?array $usedOptions = null |
| 139 | ): string { |
| 140 | $usedOptions = $options->allCacheVaryingOptionsWithPostproc(); |
| 141 | |
| 142 | $revId = $revision->getId(); |
| 143 | if ( !$revId ) { |
| 144 | // If RevId is null, this would probably be unsafe to use as a cache key. |
| 145 | throw new InvalidArgumentException( "Revision must have an id number" ); |
| 146 | } |
| 147 | $hash = $options->optionsHash( $usedOptions ); |
| 148 | return $this->cache->makeKey( $this->name, $revId, $hash ); |
| 149 | } |
| 150 | |
| 151 | /** |
| 152 | * Get a key that will be used for locks or pool counter |
| 153 | * |
| 154 | * Similar to makeParserOutputKey except the revision id might be null, |
| 155 | * in which case it is unsafe to cache, but still needs a key for things like |
| 156 | * poolcounter. |
| 157 | * |
| 158 | * @warning The exact format of the key is considered internal and is subject |
| 159 | * to change, thus should not be used as storage or long-term caching key. |
| 160 | * This is intended to be used for logging or keying something transient. |
| 161 | * |
| 162 | * @param RevisionRecord $revision |
| 163 | * @param ParserOptions $options |
| 164 | * @param array|null $usedOptions currently ignored |
| 165 | * @return string |
| 166 | * @internal |
| 167 | */ |
| 168 | public function makeParserOutputKeyOptionalRevId( |
| 169 | RevisionRecord $revision, |
| 170 | ParserOptions $options, |
| 171 | ?array $usedOptions = null |
| 172 | ): string { |
| 173 | $usedOptions = ParserOptions::allCacheVaryingOptions(); |
| 174 | |
| 175 | // revId may be null. |
| 176 | $revId = (string)$revision->getId(); |
| 177 | $hash = $options->optionsHash( $usedOptions ); |
| 178 | return $this->cache->makeKey( $this->name, $revId, $hash ); |
| 179 | } |
| 180 | |
| 181 | /** |
| 182 | * Retrieve the ParserOutput from cache. |
| 183 | * false if not found or outdated. |
| 184 | * |
| 185 | * @param RevisionRecord $revision |
| 186 | * @param ParserOptions $parserOptions |
| 187 | * |
| 188 | * @return ParserOutput|false False on failure |
| 189 | */ |
| 190 | public function get( RevisionRecord $revision, ParserOptions $parserOptions ) { |
| 191 | if ( $this->cacheExpiry <= 0 ) { |
| 192 | // disabled |
| 193 | return false; |
| 194 | } |
| 195 | |
| 196 | if ( !$parserOptions->isSafeToCache() ) { |
| 197 | $this->incrementStats( $revision, 'miss', 'unsafe' ); |
| 198 | return false; |
| 199 | } |
| 200 | |
| 201 | $cacheKey = $this->makeParserOutputKey( $revision, $parserOptions ); |
| 202 | $json = $this->cache->get( $cacheKey ); |
| 203 | |
| 204 | if ( $json === false ) { |
| 205 | $this->incrementStats( $revision, 'miss', 'absent' ); |
| 206 | return false; |
| 207 | } |
| 208 | |
| 209 | $output = $this->restoreFromJson( $json, $cacheKey ); |
| 210 | if ( $output === null ) { |
| 211 | $this->incrementStats( $revision, 'miss', 'unserialize' ); |
| 212 | return false; |
| 213 | } |
| 214 | |
| 215 | $cacheTime = (int)MWTimestamp::convert( TS::UNIX, $output->getCacheTime() ); |
| 216 | $expiryTime = (int)MWTimestamp::convert( TS::UNIX, $this->cacheEpoch ); |
| 217 | $expiryTime = max( $expiryTime, (int)MWTimestamp::now( TS::UNIX ) - $this->cacheExpiry ); |
| 218 | |
| 219 | if ( $cacheTime < $expiryTime ) { |
| 220 | $this->incrementStats( $revision, 'miss', 'expired' ); |
| 221 | return false; |
| 222 | } |
| 223 | |
| 224 | $this->logger->debug( 'old-revision cache hit' ); |
| 225 | $this->incrementStats( $revision, 'hit' ); |
| 226 | return $output; |
| 227 | } |
| 228 | |
| 229 | /** |
| 230 | * @param ParserOutput $output |
| 231 | * @param RevisionRecord $revision |
| 232 | * @param ParserOptions $parserOptions |
| 233 | * @param string|null $cacheTime TS::MW timestamp when the output was generated |
| 234 | */ |
| 235 | public function save( |
| 236 | ParserOutput $output, |
| 237 | RevisionRecord $revision, |
| 238 | ParserOptions $parserOptions, |
| 239 | ?string $cacheTime = null |
| 240 | ) { |
| 241 | if ( !$output->hasText() ) { |
| 242 | throw new InvalidArgumentException( 'Attempt to cache a ParserOutput with no text set!' ); |
| 243 | } |
| 244 | |
| 245 | if ( $this->cacheExpiry <= 0 ) { |
| 246 | // disabled |
| 247 | return; |
| 248 | } |
| 249 | |
| 250 | $cacheKey = $this->makeParserOutputKey( $revision, $parserOptions ); |
| 251 | |
| 252 | // Ensure cache properties are set in the ParserOutput |
| 253 | // T350538: These should be turned into assertions that the |
| 254 | // properties are already present (and the $cacheTime argument |
| 255 | // removed). |
| 256 | if ( $cacheTime ) { |
| 257 | $output->setCacheTime( $cacheTime ); |
| 258 | } else { |
| 259 | $cacheTime = $output->getCacheTime(); |
| 260 | } |
| 261 | if ( !$output->getCacheRevisionId() ) { |
| 262 | $output->setCacheRevisionId( $revision->getId() ); |
| 263 | } |
| 264 | if ( !$output->getRenderId() ) { |
| 265 | $output->setRenderId( $this->globalIdGenerator->newUUIDv1() ); |
| 266 | } |
| 267 | if ( !$output->getRevisionTimestamp() ) { |
| 268 | $output->setRevisionTimestamp( $revision->getTimestamp() ); |
| 269 | } |
| 270 | |
| 271 | $msg = "Saved in RevisionOutputCache with key $cacheKey" . |
| 272 | " and timestamp $cacheTime" . |
| 273 | " and revision id {$revision->getId()}."; |
| 274 | |
| 275 | $output->addCacheMessage( $msg ); |
| 276 | |
| 277 | // The ParserOutput might be dynamic and have been marked uncacheable by the parser. |
| 278 | $output->updateCacheExpiry( $this->cacheExpiry, '$wgOldRevisionParserCacheExpireTime' ); |
| 279 | |
| 280 | $expiry = $output->getCacheExpiry(); |
| 281 | if ( $expiry <= 0 ) { |
| 282 | $this->incrementStats( $revision, 'save', 'uncacheable' ); |
| 283 | return; |
| 284 | } |
| 285 | |
| 286 | if ( !$parserOptions->isSafeToCache() ) { |
| 287 | $this->incrementStats( $revision, 'save', 'unsafe' ); |
| 288 | return; |
| 289 | } |
| 290 | |
| 291 | $json = $this->encodeAsJson( $output, $cacheKey ); |
| 292 | if ( $json === null ) { |
| 293 | $this->incrementStats( $revision, 'save', 'nonserializable' ); |
| 294 | return; |
| 295 | } |
| 296 | |
| 297 | $this->cache->set( $cacheKey, $json, $expiry ); |
| 298 | $this->incrementStats( $revision, 'save', 'success' ); |
| 299 | $this->incrementRenderReasonStats( $revision, $parserOptions->getRenderReason() ); |
| 300 | } |
| 301 | |
| 302 | /** |
| 303 | * @param string $jsonData |
| 304 | * @param string $key |
| 305 | * @return CacheTime|ParserOutput|null |
| 306 | */ |
| 307 | private function restoreFromJson( string $jsonData, string $key ) { |
| 308 | try { |
| 309 | /** @var CacheTime $obj */ |
| 310 | $obj = $this->jsonCodec->deserialize( $jsonData, ParserOutput::class ); |
| 311 | return $obj; |
| 312 | } catch ( JsonException $e ) { |
| 313 | $this->logger->error( 'Unable to deserialize JSON', [ |
| 314 | 'name' => $this->name, |
| 315 | 'cache_key' => $key, |
| 316 | 'message' => $e->getMessage() |
| 317 | ] ); |
| 318 | return null; |
| 319 | } |
| 320 | } |
| 321 | |
| 322 | /** |
| 323 | * @param CacheTime $obj |
| 324 | * @param string $key |
| 325 | * @return string|null |
| 326 | */ |
| 327 | private function encodeAsJson( CacheTime $obj, string $key ) { |
| 328 | try { |
| 329 | return $this->jsonCodec->serialize( $obj ); |
| 330 | } catch ( JsonException $e ) { |
| 331 | $this->logger->error( 'Unable to serialize JSON', [ |
| 332 | 'name' => $this->name, |
| 333 | 'cache_key' => $key, |
| 334 | 'message' => $e->getMessage(), |
| 335 | ] ); |
| 336 | return null; |
| 337 | } |
| 338 | } |
| 339 | } |