Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
79.92% |
207 / 259 |
|
74.07% |
20 / 27 |
CRAP | |
0.00% |
0 / 1 |
| SqlBlobStore | |
79.92% |
207 / 259 |
|
74.07% |
20 / 27 |
160.50 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getCacheExpiry | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| setCacheExpiry | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getCompressBlobs | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| setCompressBlobs | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getLegacyEncoding | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| setLegacyEncoding | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getUseExternalStore | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| setUseExternalStore | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getDBLoadBalancer | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getDBConnection | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| storeBlob | |
76.47% |
13 / 17 |
|
0.00% |
0 / 1 |
5.33 | |||
| getBlob | |
100.00% |
20 / 20 |
|
100.00% |
1 / 1 |
4 | |||
| getBlobBatch | |
100.00% |
8 / 8 |
|
100.00% |
1 / 1 |
3 | |||
| fetchBlobs | |
68.04% |
66 / 97 |
|
0.00% |
0 / 1 |
37.80 | |||
| getDBOptions | |
57.89% |
11 / 19 |
|
0.00% |
0 / 1 |
8.69 | |||
| getCacheKey | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
1 | |||
| getCacheOptions | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
1 | |||
| expandBlob | |
88.00% |
22 / 25 |
|
0.00% |
0 / 1 |
9.14 | |||
| compressData | |
81.82% |
9 / 11 |
|
0.00% |
0 / 1 |
4.10 | |||
| decompressData | |
100.00% |
17 / 17 |
|
100.00% |
1 / 1 |
11 | |||
| getCacheTTL | |
80.00% |
4 / 5 |
|
0.00% |
0 / 1 |
3.07 | |||
| getTextIdFromAddress | |
100.00% |
7 / 7 |
|
100.00% |
1 / 1 |
4 | |||
| makeAddressFromTextId | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| explodeFlags | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
2 | |||
| splitBlobAddress | |
100.00% |
6 / 6 |
|
100.00% |
1 / 1 |
2 | |||
| isReadOnly | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
12 | |||
| 1 | <?php |
| 2 | /** |
| 3 | * @license GPL-2.0-or-later |
| 4 | * Attribution notice: when this file was created, much of its content was taken |
| 5 | * from the Revision.php file as present in release 1.30. Refer to the history |
| 6 | * of that file for original authorship (that file was removed entirely in 1.37, |
| 7 | * but its history can still be found in prior versions of MediaWiki). |
| 8 | * |
| 9 | * @file |
| 10 | */ |
| 11 | |
| 12 | namespace MediaWiki\Storage; |
| 13 | |
| 14 | use AppendIterator; |
| 15 | use HistoryBlobUtils; |
| 16 | use InvalidArgumentException; |
| 17 | use MediaWiki\ExternalStore\ExternalStoreAccess; |
| 18 | use MediaWiki\ExternalStore\ExternalStoreException; |
| 19 | use StatusValue; |
| 20 | use Wikimedia\Assert\Assert; |
| 21 | use Wikimedia\ObjectCache\BagOStuff; |
| 22 | use Wikimedia\ObjectCache\WANObjectCache; |
| 23 | use Wikimedia\Rdbms\DBAccessObjectUtils; |
| 24 | use Wikimedia\Rdbms\IDatabase; |
| 25 | use Wikimedia\Rdbms\IDBAccessObject; |
| 26 | use Wikimedia\Rdbms\ILoadBalancer; |
| 27 | |
| 28 | /** |
| 29 | * Service for storing and loading Content objects representing revision data blobs. |
| 30 | * |
| 31 | * @since 1.31 |
| 32 | * |
| 33 | * @note This was written to act as a drop-in replacement for the corresponding |
| 34 | * static methods in the old Revision class (which was later removed in 1.37). |
| 35 | */ |
| 36 | class SqlBlobStore implements BlobStore { |
| 37 | |
| 38 | // Note: the name has been taken unchanged from the old Revision class. |
| 39 | public const TEXT_CACHE_GROUP = 'revisiontext:10'; |
| 40 | |
| 41 | /** @internal */ |
| 42 | public const DEFAULT_TTL = 7 * 24 * 3600; // 7 days |
| 43 | |
| 44 | /** |
| 45 | * @var int |
| 46 | */ |
| 47 | private $cacheExpiry = self::DEFAULT_TTL; |
| 48 | |
| 49 | /** |
| 50 | * @var bool |
| 51 | */ |
| 52 | private $compressBlobs = false; |
| 53 | |
| 54 | /** |
| 55 | * @var string|false |
| 56 | */ |
| 57 | private $legacyEncoding = false; |
| 58 | |
| 59 | /** |
| 60 | * @var bool |
| 61 | */ |
| 62 | private $useExternalStore = false; |
| 63 | |
| 64 | /** |
| 65 | * @param ILoadBalancer $dbLoadBalancer A load balancer for acquiring database connections |
| 66 | * @param ExternalStoreAccess $extStoreAccess Access layer for external storage |
| 67 | * @param WANObjectCache $cache A cache manager for caching blobs. This can be the local |
| 68 | * wiki's default instance even if $dbDomain refers to a different wiki, since |
| 69 | * makeGlobalKey() is used to construct a key that allows cached blobs from the |
| 70 | * same database to be re-used between wikis. For example, wiki A and wiki B will |
| 71 | * use the same cache keys for blobs fetched from wiki C, regardless of the |
| 72 | * wiki-specific default key space. |
| 73 | * @param bool|string $dbDomain The ID of the target wiki database. Use false for the local wiki. |
| 74 | */ |
| 75 | public function __construct( |
| 76 | private readonly ILoadBalancer $dbLoadBalancer, |
| 77 | private readonly ExternalStoreAccess $extStoreAccess, |
| 78 | private readonly WANObjectCache $cache, |
| 79 | private readonly bool|string $dbDomain = false, |
| 80 | ) { |
| 81 | } |
| 82 | |
| 83 | /** |
| 84 | * @return int Time for which blobs can be cached, in seconds |
| 85 | */ |
| 86 | public function getCacheExpiry() { |
| 87 | return $this->cacheExpiry; |
| 88 | } |
| 89 | |
| 90 | /** |
| 91 | * @param int $cacheExpiry Time for which blobs can be cached, in seconds |
| 92 | */ |
| 93 | public function setCacheExpiry( int $cacheExpiry ) { |
| 94 | $this->cacheExpiry = $cacheExpiry; |
| 95 | } |
| 96 | |
| 97 | /** |
| 98 | * @return bool Whether blobs should be compressed for storage |
| 99 | */ |
| 100 | public function getCompressBlobs() { |
| 101 | return $this->compressBlobs; |
| 102 | } |
| 103 | |
| 104 | /** |
| 105 | * @param bool $compressBlobs Whether blobs should be compressed for storage |
| 106 | */ |
| 107 | public function setCompressBlobs( $compressBlobs ) { |
| 108 | $this->compressBlobs = $compressBlobs; |
| 109 | } |
| 110 | |
| 111 | /** |
| 112 | * @return false|string The legacy encoding to assume for blobs that are not marked as utf8. |
| 113 | * False means handling of legacy encoding is disabled, and utf8 assumed. |
| 114 | */ |
| 115 | public function getLegacyEncoding() { |
| 116 | return $this->legacyEncoding; |
| 117 | } |
| 118 | |
| 119 | /** |
| 120 | * Set the legacy encoding to assume for blobs that do not have the utf-8 flag set. |
| 121 | * |
| 122 | * @note The second parameter, Language $language, was removed in 1.34. |
| 123 | * |
| 124 | * @param string $legacyEncoding The legacy encoding to assume for blobs that are |
| 125 | * not marked as utf8. |
| 126 | */ |
| 127 | public function setLegacyEncoding( string $legacyEncoding ) { |
| 128 | $this->legacyEncoding = $legacyEncoding; |
| 129 | } |
| 130 | |
| 131 | /** |
| 132 | * @return bool Whether to use the ExternalStore mechanism for storing blobs. |
| 133 | */ |
| 134 | public function getUseExternalStore() { |
| 135 | return $this->useExternalStore; |
| 136 | } |
| 137 | |
| 138 | /** |
| 139 | * @param bool $useExternalStore Whether to use the ExternalStore mechanism for storing blobs. |
| 140 | */ |
| 141 | public function setUseExternalStore( bool $useExternalStore ) { |
| 142 | $this->useExternalStore = $useExternalStore; |
| 143 | } |
| 144 | |
| 145 | /** |
| 146 | * @return ILoadBalancer |
| 147 | */ |
| 148 | private function getDBLoadBalancer() { |
| 149 | return $this->dbLoadBalancer; |
| 150 | } |
| 151 | |
| 152 | /** |
| 153 | * @param int $index A database index, like DB_PRIMARY or DB_REPLICA |
| 154 | * |
| 155 | * @return IDatabase |
| 156 | */ |
| 157 | private function getDBConnection( $index ) { |
| 158 | $lb = $this->getDBLoadBalancer(); |
| 159 | return $lb->getConnection( $index, [], $this->dbDomain ); |
| 160 | } |
| 161 | |
| 162 | /** |
| 163 | * Stores an arbitrary blob of data and returns an address that can be used with |
| 164 | * getBlob() to retrieve the same blob of data, |
| 165 | * |
| 166 | * @param string $data |
| 167 | * @param array $hints An array of hints. |
| 168 | * |
| 169 | * @throws BlobAccessException |
| 170 | * @return string an address that can be used with getBlob() to retrieve the data. |
| 171 | */ |
| 172 | public function storeBlob( $data, $hints = [] ) { |
| 173 | $flags = $this->compressData( $data ); |
| 174 | |
| 175 | # Write to external storage if required |
| 176 | if ( $this->useExternalStore ) { |
| 177 | // Store and get the URL |
| 178 | try { |
| 179 | $data = $this->extStoreAccess->insert( $data, [ 'domain' => $this->dbDomain ] ); |
| 180 | } catch ( ExternalStoreException $e ) { |
| 181 | throw new BlobAccessException( $e->getMessage(), 0, $e ); |
| 182 | } |
| 183 | if ( !$data ) { |
| 184 | throw new BlobAccessException( "Failed to store text to external storage" ); |
| 185 | } |
| 186 | if ( $flags ) { |
| 187 | return 'es:' . $data . '?flags=' . $flags; |
| 188 | } else { |
| 189 | return 'es:' . $data; |
| 190 | } |
| 191 | } else { |
| 192 | $dbw = $this->getDBConnection( DB_PRIMARY ); |
| 193 | |
| 194 | $dbw->newInsertQueryBuilder() |
| 195 | ->insertInto( 'text' ) |
| 196 | ->row( [ 'old_text' => $data, 'old_flags' => $flags ] ) |
| 197 | ->caller( __METHOD__ )->execute(); |
| 198 | |
| 199 | $textId = $dbw->insertId(); |
| 200 | |
| 201 | return self::makeAddressFromTextId( $textId ); |
| 202 | } |
| 203 | } |
| 204 | |
| 205 | /** |
| 206 | * Retrieve a blob, given an address. |
| 207 | * Currently hardcoded to the 'text' table storage engine. |
| 208 | * |
| 209 | * MCR migration note: this replaced Revision::loadText |
| 210 | * |
| 211 | * @param string $blobAddress |
| 212 | * @param int $queryFlags |
| 213 | * |
| 214 | * @throws BlobAccessException |
| 215 | * @return string |
| 216 | */ |
| 217 | public function getBlob( $blobAddress, $queryFlags = 0 ) { |
| 218 | Assert::parameterType( 'string', $blobAddress, '$blobAddress' ); |
| 219 | |
| 220 | $error = null; |
| 221 | $blob = $this->cache->getWithSetCallback( |
| 222 | $this->getCacheKey( $blobAddress ), |
| 223 | $this->getCacheTTL(), |
| 224 | function ( $unused, &$ttl, &$setOpts ) use ( $blobAddress, $queryFlags, &$error ) { |
| 225 | // Ignore $setOpts; blobs are immutable and negatives are not cached |
| 226 | [ $result, $errors ] = $this->fetchBlobs( [ $blobAddress ], $queryFlags ); |
| 227 | // No negative caching; negative hits on text rows may be due to corrupted replica DBs |
| 228 | $error = $errors[$blobAddress] ?? null; |
| 229 | if ( $error ) { |
| 230 | $ttl = WANObjectCache::TTL_UNCACHEABLE; |
| 231 | } |
| 232 | return $result[$blobAddress]; |
| 233 | }, |
| 234 | $this->getCacheOptions() |
| 235 | ); |
| 236 | |
| 237 | if ( $error ) { |
| 238 | if ( $error[0] === 'badrevision' ) { |
| 239 | throw new BadBlobException( $error[1] ); |
| 240 | } else { |
| 241 | throw new BlobAccessException( $error[1] ); |
| 242 | } |
| 243 | } |
| 244 | |
| 245 | Assert::postcondition( is_string( $blob ), 'Blob must not be null' ); |
| 246 | return $blob; |
| 247 | } |
| 248 | |
| 249 | /** |
| 250 | * A batched version of BlobStore::getBlob. |
| 251 | * |
| 252 | * @param string[] $blobAddresses An array of blob addresses. |
| 253 | * @param int $queryFlags See IDBAccessObject. |
| 254 | * @throws BlobAccessException |
| 255 | * @return StatusValue A status with a map of blobAddress => binary blob data or null |
| 256 | * if fetching the blob has failed. Fetch failures errors are the |
| 257 | * warnings in the status object. |
| 258 | * @since 1.34 |
| 259 | */ |
| 260 | public function getBlobBatch( $blobAddresses, $queryFlags = 0 ) { |
| 261 | // FIXME: All caching has temporarily been removed in I94c6f9ba7b9caeeb due to T235188. |
| 262 | // Caching behavior should be restored by reverting I94c6f9ba7b9caeeb as soon as |
| 263 | // the root cause of T235188 has been resolved. |
| 264 | |
| 265 | [ $blobsByAddress, $errors ] = $this->fetchBlobs( $blobAddresses, $queryFlags ); |
| 266 | |
| 267 | $blobsByAddress = array_map( static function ( $blob ) { |
| 268 | return $blob === false ? null : $blob; |
| 269 | }, $blobsByAddress ); |
| 270 | |
| 271 | $result = StatusValue::newGood( $blobsByAddress ); |
| 272 | foreach ( $errors as $error ) { |
| 273 | // @phan-suppress-next-line PhanParamTooFewUnpack |
| 274 | $result->warning( ...$error ); |
| 275 | } |
| 276 | return $result; |
| 277 | } |
| 278 | |
| 279 | /** |
| 280 | * MCR migration note: this corresponded to Revision::fetchText |
| 281 | * |
| 282 | * @param string[] $blobAddresses |
| 283 | * @param int $queryFlags |
| 284 | * |
| 285 | * @throws BlobAccessException |
| 286 | * @return array [ $result, $errors ] A list with the following elements: |
| 287 | * - The result: a map of blob addresses to successfully fetched blobs |
| 288 | * or false if fetch failed |
| 289 | * - Errors: a map of blob addresses to error information about the blob. |
| 290 | * On success, the relevant key will be absent. Each error is a list of |
| 291 | * parameters to be passed to StatusValue::warning(). |
| 292 | */ |
| 293 | private function fetchBlobs( $blobAddresses, $queryFlags ) { |
| 294 | $textIdToBlobAddress = []; |
| 295 | $result = []; |
| 296 | $errors = []; |
| 297 | foreach ( $blobAddresses as $blobAddress ) { |
| 298 | try { |
| 299 | [ $schema, $id, $params ] = self::splitBlobAddress( $blobAddress ); |
| 300 | } catch ( InvalidArgumentException $ex ) { |
| 301 | throw new BlobAccessException( |
| 302 | $ex->getMessage() . '. Use findBadBlobs.php to remedy.', |
| 303 | 0, |
| 304 | $ex |
| 305 | ); |
| 306 | } |
| 307 | |
| 308 | if ( $schema === 'es' ) { |
| 309 | if ( $params && isset( $params['flags'] ) ) { |
| 310 | $blob = $this->expandBlob( $id, $params['flags'] . ',external', $blobAddress ); |
| 311 | } else { |
| 312 | $blob = $this->expandBlob( $id, 'external', $blobAddress ); |
| 313 | } |
| 314 | |
| 315 | if ( $blob === false ) { |
| 316 | $errors[$blobAddress] = [ |
| 317 | 'internalerror', |
| 318 | "Bad data in external store address $id. Use findBadBlobs.php to remedy." |
| 319 | ]; |
| 320 | } |
| 321 | $result[$blobAddress] = $blob; |
| 322 | } elseif ( $schema === 'bad' ) { |
| 323 | // Database row was marked as "known bad" |
| 324 | wfDebug( |
| 325 | __METHOD__ |
| 326 | . ": loading known-bad content ($blobAddress), returning empty string" |
| 327 | ); |
| 328 | $result[$blobAddress] = ''; |
| 329 | $errors[$blobAddress] = [ |
| 330 | 'badrevision', |
| 331 | 'The content of this revision is missing or corrupted (bad schema)' |
| 332 | ]; |
| 333 | } elseif ( $schema === 'tt' ) { |
| 334 | $textId = intval( $id ); |
| 335 | |
| 336 | if ( $textId < 1 || $id !== (string)$textId ) { |
| 337 | $errors[$blobAddress] = [ |
| 338 | 'internalerror', |
| 339 | "Bad blob address: $blobAddress. Use findBadBlobs.php to remedy." |
| 340 | ]; |
| 341 | $result[$blobAddress] = false; |
| 342 | } |
| 343 | |
| 344 | $textIdToBlobAddress[$textId] = $blobAddress; |
| 345 | } else { |
| 346 | $errors[$blobAddress] = [ |
| 347 | 'internalerror', |
| 348 | "Unknown blob address schema: $schema. Use findBadBlobs.php to remedy." |
| 349 | ]; |
| 350 | $result[$blobAddress] = false; |
| 351 | } |
| 352 | } |
| 353 | |
| 354 | $textIds = array_keys( $textIdToBlobAddress ); |
| 355 | if ( !$textIds ) { |
| 356 | return [ $result, $errors ]; |
| 357 | } |
| 358 | // Callers doing updates will pass in READ_LATEST as usual. Since the text/blob tables |
| 359 | // do not normally get rows changed around, set READ_LATEST_IMMUTABLE in those cases. |
| 360 | $queryFlags |= DBAccessObjectUtils::hasFlags( $queryFlags, IDBAccessObject::READ_LATEST ) |
| 361 | ? IDBAccessObject::READ_LATEST_IMMUTABLE |
| 362 | : 0; |
| 363 | [ $index, $options, $fallbackIndex, $fallbackOptions ] = |
| 364 | self::getDBOptions( $queryFlags ); |
| 365 | // Text data is immutable; check replica DBs first. |
| 366 | $dbConnection = $this->getDBConnection( $index ); |
| 367 | $rows = $dbConnection->newSelectQueryBuilder() |
| 368 | ->select( [ 'old_id', 'old_text', 'old_flags' ] ) |
| 369 | ->from( 'text' ) |
| 370 | ->where( [ 'old_id' => $textIds ] ) |
| 371 | ->options( $options ) |
| 372 | ->caller( __METHOD__ )->fetchResultSet(); |
| 373 | $numRows = $rows->numRows(); |
| 374 | |
| 375 | // Fallback to DB_PRIMARY in some cases if not all the rows were found, using the appropriate |
| 376 | // options, such as FOR UPDATE to avoid missing rows due to REPEATABLE-READ. |
| 377 | if ( $numRows !== count( $textIds ) && $fallbackIndex !== null ) { |
| 378 | $fetchedTextIds = []; |
| 379 | foreach ( $rows as $row ) { |
| 380 | $fetchedTextIds[] = $row->old_id; |
| 381 | } |
| 382 | $missingTextIds = array_diff( $textIds, $fetchedTextIds ); |
| 383 | $dbConnection = $this->getDBConnection( $fallbackIndex ); |
| 384 | $rowsFromFallback = $dbConnection->newSelectQueryBuilder() |
| 385 | ->select( [ 'old_id', 'old_text', 'old_flags' ] ) |
| 386 | ->from( 'text' ) |
| 387 | ->where( [ 'old_id' => $missingTextIds ] ) |
| 388 | ->options( $fallbackOptions ) |
| 389 | ->caller( __METHOD__ )->fetchResultSet(); |
| 390 | $appendIterator = new AppendIterator(); |
| 391 | $appendIterator->append( $rows ); |
| 392 | $appendIterator->append( $rowsFromFallback ); |
| 393 | $rows = $appendIterator; |
| 394 | } |
| 395 | |
| 396 | foreach ( $rows as $row ) { |
| 397 | $blobAddress = $textIdToBlobAddress[$row->old_id]; |
| 398 | $blob = false; |
| 399 | if ( $row->old_text !== null ) { |
| 400 | $blob = $this->expandBlob( $row->old_text, $row->old_flags, $blobAddress ); |
| 401 | } |
| 402 | if ( $blob === false ) { |
| 403 | $errors[$blobAddress] = [ |
| 404 | 'internalerror', |
| 405 | "Bad data in text row {$row->old_id}. Use findBadBlobs.php to remedy." |
| 406 | ]; |
| 407 | } |
| 408 | $result[$blobAddress] = $blob; |
| 409 | } |
| 410 | |
| 411 | // If we're still missing some of the rows, set errors for missing blobs. |
| 412 | if ( count( $result ) !== count( $blobAddresses ) ) { |
| 413 | foreach ( $blobAddresses as $blobAddress ) { |
| 414 | if ( !isset( $result[$blobAddress ] ) ) { |
| 415 | $errors[$blobAddress] = [ |
| 416 | 'internalerror', |
| 417 | "Unable to fetch blob at $blobAddress. Use findBadBlobs.php to remedy." |
| 418 | ]; |
| 419 | $result[$blobAddress] = false; |
| 420 | } |
| 421 | } |
| 422 | } |
| 423 | return [ $result, $errors ]; |
| 424 | } |
| 425 | |
| 426 | private static function getDBOptions( int $bitfield ): array { |
| 427 | if ( DBAccessObjectUtils::hasFlags( $bitfield, IDBAccessObject::READ_LATEST_IMMUTABLE ) ) { |
| 428 | $index = DB_REPLICA; // override READ_LATEST if set |
| 429 | $fallbackIndex = DB_PRIMARY; |
| 430 | } elseif ( DBAccessObjectUtils::hasFlags( $bitfield, IDBAccessObject::READ_LATEST ) ) { |
| 431 | $index = DB_PRIMARY; |
| 432 | $fallbackIndex = null; |
| 433 | } else { |
| 434 | $index = DB_REPLICA; |
| 435 | $fallbackIndex = null; |
| 436 | } |
| 437 | |
| 438 | $lockingOptions = []; |
| 439 | if ( DBAccessObjectUtils::hasFlags( $bitfield, IDBAccessObject::READ_EXCLUSIVE ) ) { |
| 440 | $lockingOptions[] = 'FOR UPDATE'; |
| 441 | } elseif ( DBAccessObjectUtils::hasFlags( $bitfield, IDBAccessObject::READ_LOCKING ) ) { |
| 442 | $lockingOptions[] = 'LOCK IN SHARE MODE'; |
| 443 | } |
| 444 | |
| 445 | if ( $fallbackIndex !== null ) { |
| 446 | $options = []; // locks on DB_REPLICA make no sense |
| 447 | $fallbackOptions = $lockingOptions; |
| 448 | } else { |
| 449 | $options = $lockingOptions; |
| 450 | $fallbackOptions = []; // no fallback |
| 451 | } |
| 452 | |
| 453 | return [ $index, $options, $fallbackIndex, $fallbackOptions ]; |
| 454 | } |
| 455 | |
| 456 | /** |
| 457 | * Get a cache key for a given Blob address. |
| 458 | * |
| 459 | * The cache key is constructed in a way that allows cached blobs from the same database |
| 460 | * to be re-used between wikis. For example, wiki A and wiki B will use the same cache keys |
| 461 | * for blobs fetched from wiki C. |
| 462 | * |
| 463 | * @param string $blobAddress |
| 464 | * @return string |
| 465 | */ |
| 466 | private function getCacheKey( $blobAddress ) { |
| 467 | return $this->cache->makeGlobalKey( |
| 468 | 'SqlBlobStore-blob', |
| 469 | $this->dbLoadBalancer->resolveDomainID( $this->dbDomain ), |
| 470 | $blobAddress |
| 471 | ); |
| 472 | } |
| 473 | |
| 474 | /** |
| 475 | * Get the cache key options for a given Blob |
| 476 | * |
| 477 | * @return array<string,mixed> |
| 478 | */ |
| 479 | private function getCacheOptions() { |
| 480 | return [ |
| 481 | 'pcGroup' => self::TEXT_CACHE_GROUP, |
| 482 | 'pcTTL' => WANObjectCache::TTL_PROC_LONG, |
| 483 | 'segmentable' => true |
| 484 | ]; |
| 485 | } |
| 486 | |
| 487 | /** |
| 488 | * Expand a raw data blob according to the flags given. |
| 489 | * |
| 490 | * MCR migration note: this replaced Revision::getRevisionText |
| 491 | * |
| 492 | * @note direct use is deprecated, use getBlob() or SlotRecord::getContent() instead. |
| 493 | * @todo make this private, there should be no need to use this method outside this class. |
| 494 | * |
| 495 | * @param string $raw The raw blob data, to be processed according to $flags. |
| 496 | * May be the blob itself, or the blob compressed, or just the address |
| 497 | * of the actual blob, depending on $flags. |
| 498 | * @param string|string[] $flags Blob flags, such as 'external' or 'gzip'. |
| 499 | * Note that not including 'utf-8' in $flags will cause the data to be decoded |
| 500 | * according to the legacy encoding specified via setLegacyEncoding. |
| 501 | * @param string|null $blobAddress A blob address for use in the cache key. If not given, |
| 502 | * caching is disabled. |
| 503 | * |
| 504 | * @return false|string The expanded blob or false on failure |
| 505 | * @throws BlobAccessException |
| 506 | */ |
| 507 | public function expandBlob( $raw, $flags, $blobAddress = null ) { |
| 508 | if ( is_string( $flags ) ) { |
| 509 | $flags = self::explodeFlags( $flags ); |
| 510 | } |
| 511 | if ( in_array( 'error', $flags ) ) { |
| 512 | throw new BadBlobException( |
| 513 | "The content of this revision is missing or corrupted (error flag)" |
| 514 | ); |
| 515 | } |
| 516 | |
| 517 | // Use external methods for external objects, text in table is URL-only then |
| 518 | if ( in_array( 'external', $flags ) ) { |
| 519 | $url = $raw; |
| 520 | $parts = explode( '://', $url, 2 ); |
| 521 | if ( count( $parts ) == 1 || $parts[1] == '' ) { |
| 522 | return false; |
| 523 | } |
| 524 | |
| 525 | if ( $blobAddress ) { |
| 526 | // The cached value should be decompressed, so handle that and return here. |
| 527 | return $this->cache->getWithSetCallback( |
| 528 | $this->getCacheKey( $blobAddress ), |
| 529 | $this->getCacheTTL(), |
| 530 | function () use ( $url, $flags, $blobAddress ) { |
| 531 | // Ignore $setOpts; blobs are immutable and negatives are not cached |
| 532 | $blob = $this->extStoreAccess |
| 533 | ->fetchFromURL( $url, [ 'domain' => $this->dbDomain ] ); |
| 534 | |
| 535 | return $blob === false ? false : $this->decompressData( $blob, $flags, $blobAddress ); |
| 536 | }, |
| 537 | $this->getCacheOptions() |
| 538 | ); |
| 539 | } else { |
| 540 | $blob = $this->extStoreAccess->fetchFromURL( $url, [ 'domain' => $this->dbDomain ] ); |
| 541 | return $blob === false ? false : $this->decompressData( $blob, $flags, $blobAddress ); |
| 542 | } |
| 543 | } else { |
| 544 | return $this->decompressData( $raw, $flags, $blobAddress ); |
| 545 | } |
| 546 | } |
| 547 | |
| 548 | /** |
| 549 | * If $wgCompressRevisions is enabled, we will compress data. |
| 550 | * The input string is modified in place. |
| 551 | * Return value is the flags field: contains 'gzip' if the |
| 552 | * data is compressed, and 'utf-8' if we're saving in UTF-8 |
| 553 | * mode. |
| 554 | * |
| 555 | * MCR migration note: this replaced Revision::compressRevisionText |
| 556 | * |
| 557 | * @note direct use is deprecated! |
| 558 | * @todo make this private, there should be no need to use this method outside this class. |
| 559 | * |
| 560 | * @param string &$blob |
| 561 | * |
| 562 | * @return string |
| 563 | */ |
| 564 | public function compressData( &$blob ) { |
| 565 | $blobFlags = []; |
| 566 | |
| 567 | // Revisions not marked as UTF-8 will have legacy decoding applied by decompressData(). |
| 568 | // XXX: if $this->legacyEncoding is not set, we could skip this. That would however be |
| 569 | // risky, since $this->legacyEncoding being set in the future would lead to data corruption. |
| 570 | $blobFlags[] = 'utf-8'; |
| 571 | |
| 572 | if ( $this->compressBlobs ) { |
| 573 | if ( function_exists( 'gzdeflate' ) ) { |
| 574 | $deflated = gzdeflate( $blob ); |
| 575 | |
| 576 | if ( $deflated === false ) { |
| 577 | wfLogWarning( __METHOD__ . ': gzdeflate() failed' ); |
| 578 | } else { |
| 579 | $blob = $deflated; |
| 580 | $blobFlags[] = 'gzip'; |
| 581 | } |
| 582 | } else { |
| 583 | wfDebug( __METHOD__ . " -- no zlib support, not compressing" ); |
| 584 | } |
| 585 | } |
| 586 | return implode( ',', $blobFlags ); |
| 587 | } |
| 588 | |
| 589 | /** |
| 590 | * Re-converts revision text according to its flags. |
| 591 | * |
| 592 | * MCR migration note: this replaced Revision::decompressRevisionText |
| 593 | * |
| 594 | * @note direct use is deprecated, use getBlob() or SlotRecord::getContent() instead. |
| 595 | * @todo make this private, there should be no need to use this method outside this class. |
| 596 | * |
| 597 | * @param string $blob Blob in compressed/encoded form. |
| 598 | * @param array $blobFlags Compression flags, such as 'gzip'. |
| 599 | * Note that not including 'utf-8' in $blobFlags will cause the data to be decoded |
| 600 | * according to the legacy encoding specified via setLegacyEncoding. |
| 601 | * @param string|null $blobAddress Used for log message |
| 602 | * |
| 603 | * @return string|false Decompressed text, or false on failure |
| 604 | */ |
| 605 | public function decompressData( string $blob, array $blobFlags, ?string $blobAddress = null ) { |
| 606 | if ( in_array( 'error', $blobFlags ) ) { |
| 607 | // Error row, return false |
| 608 | return false; |
| 609 | } |
| 610 | |
| 611 | // Deal with optional compression of archived pages. |
| 612 | // This can be done periodically via maintenance/compressOld.php, and |
| 613 | // as pages are saved if $wgCompressRevisions is set. |
| 614 | if ( in_array( 'gzip', $blobFlags ) ) { |
| 615 | // Silence native warning in favour of more detailed warning (T380347) |
| 616 | // phpcs:ignore Generic.PHP.NoSilencedErrors.Discouraged |
| 617 | $blob = @gzinflate( $blob ); |
| 618 | if ( $blob === false ) { |
| 619 | wfWarn( __METHOD__ . ': gzinflate() failed' . |
| 620 | ( $blobAddress ? ' (at blob address ' . $blobAddress . ')' : '' ) ); |
| 621 | return false; |
| 622 | } |
| 623 | } |
| 624 | |
| 625 | if ( in_array( 'object', $blobFlags ) ) { |
| 626 | # Generic compressed storage |
| 627 | $obj = HistoryBlobUtils::unserialize( $blob ); |
| 628 | if ( !$obj ) { |
| 629 | // Invalid object |
| 630 | return false; |
| 631 | } |
| 632 | $blob = $obj->getText(); |
| 633 | } |
| 634 | |
| 635 | // Needed to support old revisions from before MW 1.5. |
| 636 | if ( $blob !== false && $this->legacyEncoding |
| 637 | && !in_array( 'utf-8', $blobFlags ) && !in_array( 'utf8', $blobFlags ) |
| 638 | ) { |
| 639 | // - Old revisions kept around in a legacy encoding? |
| 640 | // Upconvert on demand. |
| 641 | // - "utf8" checked for compatibility with some broken |
| 642 | // conversion scripts 2008-12-30. |
| 643 | // - Even with "//IGNORE" iconv can whine about illegal characters in |
| 644 | // *input* string. We just ignore those too. |
| 645 | // Ref https://bugs.php.net/bug.php?id=37166 |
| 646 | // Ref https://phabricator.wikimedia.org/T18885 |
| 647 | // |
| 648 | // phpcs:ignore Generic.PHP.NoSilencedErrors.Discouraged |
| 649 | $blob = @iconv( $this->legacyEncoding, 'UTF-8//IGNORE', $blob ); |
| 650 | } |
| 651 | |
| 652 | return $blob; |
| 653 | } |
| 654 | |
| 655 | /** |
| 656 | * Get the text cache TTL |
| 657 | * |
| 658 | * MCR migration note: this replaced Revision::getCacheTTL |
| 659 | * |
| 660 | * @return int |
| 661 | */ |
| 662 | private function getCacheTTL() { |
| 663 | $cache = $this->cache; |
| 664 | |
| 665 | if ( $cache->getQoS( BagOStuff::ATTR_DURABILITY ) >= BagOStuff::QOS_DURABILITY_RDBMS ) { |
| 666 | // Do not cache RDBMs blobs in...the RDBMs store |
| 667 | $ttl = $cache::TTL_UNCACHEABLE; |
| 668 | } else { |
| 669 | $ttl = $this->cacheExpiry ?: $cache::TTL_UNCACHEABLE; |
| 670 | } |
| 671 | |
| 672 | return $ttl; |
| 673 | } |
| 674 | |
| 675 | /** |
| 676 | * Returns an ID corresponding to the old_id field in the text table, corresponding |
| 677 | * to the given $address. |
| 678 | * |
| 679 | * Currently, $address must start with 'tt:' followed by a decimal integer representing |
| 680 | * the old_id; if $address does not start with 'tt:', null is returned. However, |
| 681 | * the implementation may change to insert rows into the text table on the fly. |
| 682 | * This implies that this method cannot be static. |
| 683 | * |
| 684 | * @note This method exists for use with the text table based storage schema. |
| 685 | * It should not be assumed that is will function with all future kinds of content addresses. |
| 686 | * |
| 687 | * @deprecated since 1.31, so don't assume that all blob addresses refer to a row in the text |
| 688 | * table. This method should become private once the relevant refactoring in WikiPage is |
| 689 | * complete. |
| 690 | * |
| 691 | * @param string $address |
| 692 | * |
| 693 | * @return int|null |
| 694 | */ |
| 695 | public function getTextIdFromAddress( $address ) { |
| 696 | [ $schema, $id, ] = self::splitBlobAddress( $address ); |
| 697 | |
| 698 | if ( $schema !== 'tt' ) { |
| 699 | return null; |
| 700 | } |
| 701 | |
| 702 | $textId = intval( $id ); |
| 703 | |
| 704 | if ( !$textId || $id !== (string)$textId ) { |
| 705 | throw new InvalidArgumentException( "Malformed text_id: $id" ); |
| 706 | } |
| 707 | |
| 708 | return $textId; |
| 709 | } |
| 710 | |
| 711 | /** |
| 712 | * Returns an address referring to content stored in the text table row with the given ID. |
| 713 | * The address schema for blobs stored in the text table is "tt:" followed by an integer |
| 714 | * that corresponds to a value of the old_id field. |
| 715 | * |
| 716 | * @internal |
| 717 | * @note This method should not be used by regular application logic. It is public so |
| 718 | * maintenance scripts can use it for bulk operations on the text table. |
| 719 | * |
| 720 | * @param int $id |
| 721 | * |
| 722 | * @return string |
| 723 | */ |
| 724 | public static function makeAddressFromTextId( $id ) { |
| 725 | return 'tt:' . $id; |
| 726 | } |
| 727 | |
| 728 | /** |
| 729 | * Split a comma-separated old_flags value into its constituent parts |
| 730 | * |
| 731 | * @param string $flagsString |
| 732 | * @return array |
| 733 | */ |
| 734 | public static function explodeFlags( string $flagsString ) { |
| 735 | return $flagsString === '' ? [] : explode( ',', $flagsString ); |
| 736 | } |
| 737 | |
| 738 | /** |
| 739 | * Splits a blob address into three parts: the schema, the ID, and parameters/flags. |
| 740 | * |
| 741 | * @since 1.33 |
| 742 | * |
| 743 | * @param string $address |
| 744 | * |
| 745 | * @return array [ $schema, $id, $parameters ], with $parameters being an assoc array. |
| 746 | */ |
| 747 | public static function splitBlobAddress( $address ) { |
| 748 | if ( !preg_match( '/^([-+.\w]+):([^\s?]+)(\?([^\s]*))?$/', $address, $m ) ) { |
| 749 | throw new InvalidArgumentException( "Bad blob address: $address" ); |
| 750 | } |
| 751 | |
| 752 | $schema = strtolower( $m[1] ); |
| 753 | $id = $m[2]; |
| 754 | $parameters = wfCgiToArray( $m[4] ?? '' ); |
| 755 | |
| 756 | return [ $schema, $id, $parameters ]; |
| 757 | } |
| 758 | |
| 759 | /** @inheritDoc */ |
| 760 | public function isReadOnly() { |
| 761 | if ( $this->useExternalStore && $this->extStoreAccess->isReadOnly() ) { |
| 762 | return true; |
| 763 | } |
| 764 | |
| 765 | return ( $this->getDBLoadBalancer()->getReadOnlyReason() !== false ); |
| 766 | } |
| 767 | } |