includes/Storage/SqlBlobStore.php

   1 <?php
   2 /**
   3  * Service for storing and loading data blobs representing revision content.
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * Attribution notice: when this file was created, much of its content was taken
  21  * from the Revision.php file as present in release 1.30. Refer to the history
  22  * of that file for original authorship.
  23  *
  24  * @file
  25  */
  26
  27 namespace MediaWiki\Storage;
  28
  29 use DBAccessObjectUtils;
  30 use ExternalStore;
  31 use IDBAccessObject;
  32 use IExpiringStore;
  33 use InvalidArgumentException;
  34 use Language;
  35 use MWException;
  36 use WANObjectCache;
  37 use Wikimedia\Assert\Assert;
  38 use Wikimedia\Rdbms\IDatabase;
  39 use Wikimedia\Rdbms\LoadBalancer;
  40
  41 /**
  42  * Service for storing and loading Content objects.
  43  *
  44  * @since 1.31
  45  *
  46  * @note This was written to act as a drop-in replacement for the corresponding
  47  *       static methods in Revision.
  48  */
  49 class SqlBlobStore implements IDBAccessObject, BlobStore {
  50
  51         // Note: the name has been taken unchanged from the Revision class.
  52         const TEXT_CACHE_GROUP = 'revisiontext:10';
  53
  54         /**
  55          * @var LoadBalancer
  56          */
  57         private $dbLoadBalancer;
  58
  59         /**
  60          * @var WANObjectCache
  61          */
  62         private $cache;
  63
  64         /**
  65          * @var bool|string Wiki ID
  66          */
  67         private $wikiId;
  68
  69         /**
  70          * @var int
  71          */
  72         private $cacheExpiry = 604800; // 7 days
  73
  74         /**
  75          * @var bool
  76          */
  77         private $compressBlobs = false;
  78
  79         /**
  80          * @var bool|string
  81          */
  82         private $legacyEncoding = false;
  83
  84         /**
  85          * @var Language|null
  86          */
  87         private $legacyEncodingConversionLang = null;
  88
  89         /**
  90          * @var boolean
  91          */
  92         private $useExternalStore = false;
  93
  94         /**
  95          * @param LoadBalancer $dbLoadBalancer A load balancer for acquiring database connections
  96          * @param WANObjectCache $cache A cache manager for caching blobs. This can be the local
  97          *        wiki's default instance even if $wikiId refers to a different wiki, since
  98          *        makeGlobalKey() is used to constructed a key that allows cached blobs from the
  99          *        same database to be re-used between wikis. For example, enwiki and frwiki will
 100          *        use the same cache keys for blobs from the wikidatawiki database, regardless of
 101          *        the cache's default key space.
 102          * @param bool|string $wikiId The ID of the target wiki database. Use false for the local wiki.
 103          */
 104         public function __construct(
 105                 LoadBalancer $dbLoadBalancer,
 106                 WANObjectCache $cache,
 107                 $wikiId = false
 108         ) {
 109                 $this->dbLoadBalancer = $dbLoadBalancer;
 110                 $this->cache = $cache;
 111                 $this->wikiId = $wikiId;
 112         }
 113
 114         /**
 115          * @return int time for which blobs can be cached, in seconds
 116          */
 117         public function getCacheExpiry() {
 118                 return $this->cacheExpiry;
 119         }
 120
 121         /**
 122          * @param int $cacheExpiry time for which blobs can be cached, in seconds
 123          */
 124         public function setCacheExpiry( $cacheExpiry ) {
 125                 Assert::parameterType( 'integer', $cacheExpiry, '$cacheExpiry' );
 126
 127                 $this->cacheExpiry = $cacheExpiry;
 128         }
 129
 130         /**
 131          * @return bool whether blobs should be compressed for storage
 132          */
 133         public function getCompressBlobs() {
 134                 return $this->compressBlobs;
 135         }
 136
 137         /**
 138          * @param bool $compressBlobs whether blobs should be compressed for storage
 139          */
 140         public function setCompressBlobs( $compressBlobs ) {
 141                 $this->compressBlobs = $compressBlobs;
 142         }
 143
 144         /**
 145          * @return false|string The legacy encoding to assume for blobs that are not marked as utf8.
 146          *         False means handling of legacy encoding is disabled, and utf8 assumed.
 147          */
 148         public function getLegacyEncoding() {
 149                 return $this->legacyEncoding;
 150         }
 151
 152         /**
 153          * @return Language|null The locale to use when decoding from a legacy encoding, or null
 154          *         if handling of legacy encoding is disabled.
 155          */
 156         public function getLegacyEncodingConversionLang() {
 157                 return $this->legacyEncodingConversionLang;
 158         }
 159
 160         /**
 161          * @param string $legacyEncoding The legacy encoding to assume for blobs that are
 162          *        not marked as utf8.
 163          * @param Language $language The locale to use when decoding from a legacy encoding.
 164          */
 165         public function setLegacyEncoding( $legacyEncoding, Language $language ) {
 166                 Assert::parameterType( 'string', $legacyEncoding, '$legacyEncoding' );
 167
 168                 $this->legacyEncoding = $legacyEncoding;
 169                 $this->legacyEncodingConversionLang = $language;
 170         }
 171
 172         /**
 173          * @return bool Whether to use the ExternalStore mechanism for storing blobs.
 174          */
 175         public function getUseExternalStore() {
 176                 return $this->useExternalStore;
 177         }
 178
 179         /**
 180          * @param bool $useExternalStore Whether to use the ExternalStore mechanism for storing blobs.
 181          */
 182         public function setUseExternalStore( $useExternalStore ) {
 183                 Assert::parameterType( 'boolean', $useExternalStore, '$useExternalStore' );
 184
 185                 $this->useExternalStore = $useExternalStore;
 186         }
 187
 188         /**
 189          * @return LoadBalancer
 190          */
 191         private function getDBLoadBalancer() {
 192                 return $this->dbLoadBalancer;
 193         }
 194
 195         /**
 196          * @param int $index A database index, like DB_MASTER or DB_REPLICA
 197          *
 198          * @return IDatabase
 199          */
 200         private function getDBConnection( $index ) {
 201                 $lb = $this->getDBLoadBalancer();
 202                 return $lb->getConnection( $index, [], $this->wikiId );
 203         }
 204
 205         /**
 206          * Stores an arbitrary blob of data and returns an address that can be used with
 207          * getBlob() to retrieve the same blob of data,
 208          *
 209          * @param string $data
 210          * @param array $hints An array of hints.
 211          *
 212          * @throws BlobAccessException
 213          * @return string an address that can be used with getBlob() to retrieve the data.
 214          */
 215         public function storeBlob( $data, $hints = [] ) {
 216                 try {
 217                         $flags = $this->compressData( $data );
 218
 219                         # Write to external storage if required
 220                         if ( $this->useExternalStore ) {
 221                                 // Store and get the URL
 222                                 $data = ExternalStore::insertToDefault( $data );
 223                                 if ( !$data ) {
 224                                         throw new BlobAccessException( "Failed to store text to external storage" );
 225                                 }
 226                                 if ( $flags ) {
 227                                         $flags .= ',';
 228                                 }
 229                                 $flags .= 'external';
 230
 231                                 // TODO: we could also return an address for the external store directly here.
 232                                 // That would mean bypassing the text table entirely when the external store is
 233                                 // used. We'll need to assess expected fallout before doing that.
 234                         }
 235
 236                         $dbw = $this->getDBConnection( DB_MASTER );
 237
 238                         $old_id = $dbw->nextSequenceValue( 'text_old_id_seq' );
 239                         $dbw->insert(
 240                                 'text',
 241                                 [
 242                                         'old_id' => $old_id,
 243                                         'old_text' => $data,
 244                                         'old_flags' => $flags,
 245                                 ],
 246                                 __METHOD__
 247                         );
 248
 249                         $textId = $dbw->insertId();
 250
 251                         return self::makeAddressFromTextId( $textId );
 252                 } catch ( MWException $e ) {
 253                         throw new BlobAccessException( $e->getMessage(), 0, $e );
 254                 }
 255         }
 256
 257         /**
 258          * Retrieve a blob, given an address.
 259          * Currently hardcoded to the 'text' table storage engine.
 260          *
 261          * MCR migration note: this replaces Revision::loadText
 262          *
 263          * @param string $blobAddress
 264          * @param int $queryFlags
 265          *
 266          * @throws BlobAccessException
 267          * @return string
 268          */
 269         public function getBlob( $blobAddress, $queryFlags = 0 ) {
 270                 Assert::parameterType( 'string', $blobAddress, '$blobAddress' );
 271
 272                 // No negative caching; negative hits on text rows may be due to corrupted replica DBs
 273                 $blob = $this->cache->getWithSetCallback(
 274                         $this->getCacheKey( $blobAddress ),
 275                         $this->getCacheTTL(),
 276                         function ( $unused, &$ttl, &$setOpts ) use ( $blobAddress, $queryFlags ) {
 277                                 // Ignore $setOpts; blobs are immutable and negatives are not cached
 278                                 return $this->fetchBlob( $blobAddress, $queryFlags );
 279                         },
 280                         [ 'pcGroup' => self::TEXT_CACHE_GROUP, 'pcTTL' => IExpiringStore::TTL_PROC_LONG ]
 281                 );
 282
 283                 if ( $blob === false ) {
 284                         throw new BlobAccessException( 'Failed to load blob from address ' . $blobAddress );
 285                 }
 286
 287                 return $blob;
 288         }
 289
 290         /**
 291          * MCR migration note: this corresponds to Revision::fetchText
 292          *
 293          * @param string $blobAddress
 294          * @param int $queryFlags
 295          *
 296          * @throws BlobAccessException
 297          * @return string|false
 298          */
 299         private function fetchBlob( $blobAddress, $queryFlags ) {
 300                 list( $schema, $id, ) = self::splitBlobAddress( $blobAddress );
 301
 302                 //TODO: MCR: also support 'ex' schema with ExternalStore URLs, plus flags encoded in the URL!
 303                 if ( $schema === 'tt' ) {
 304                         $textId = intval( $id );
 305                 } else {
 306                         // XXX: change to better exceptions! That makes migration more difficult, though.
 307                         throw new BlobAccessException( "Unknown blob address schema: $schema" );
 308                 }
 309
 310                 if ( !$textId || $id !== (string)$textId ) {
 311                         // XXX: change to better exceptions! That makes migration more difficult, though.
 312                         throw new BlobAccessException( "Bad blob address: $blobAddress" );
 313                 }
 314
 315                 // Callers doing updates will pass in READ_LATEST as usual. Since the text/blob tables
 316                 // do not normally get rows changed around, set READ_LATEST_IMMUTABLE in those cases.
 317                 $queryFlags |= DBAccessObjectUtils::hasFlags( $queryFlags, self::READ_LATEST )
 318                         ? self::READ_LATEST_IMMUTABLE
 319                         : 0;
 320
 321                 list( $index, $options, $fallbackIndex, $fallbackOptions ) =
 322                         DBAccessObjectUtils::getDBOptions( $queryFlags );
 323
 324                 // Text data is immutable; check replica DBs first.
 325                 $row = $this->getDBConnection( $index )->selectRow(
 326                         'text',
 327                         [ 'old_text', 'old_flags' ],
 328                         [ 'old_id' => $textId ],
 329                         __METHOD__,
 330                         $options
 331                 );
 332
 333                 // Fallback to DB_MASTER in some cases if the row was not found, using the appropriate
 334                 // options, such as FOR UPDATE to avoid missing rows due to REPEATABLE-READ.
 335                 if ( !$row && $fallbackIndex !== null ) {
 336                         $row = $this->getDBConnection( $fallbackIndex )->selectRow(
 337                                 'text',
 338                                 [ 'old_text', 'old_flags' ],
 339                                 [ 'old_id' => $textId ],
 340                                 __METHOD__,
 341                                 $fallbackOptions
 342                         );
 343                 }
 344
 345                 if ( !$row ) {
 346                         wfWarn( __METHOD__ . ": No text row with ID $textId." );
 347                         return false;
 348                 }
 349
 350                 $blob = $this->expandBlob( $row->old_text, $row->old_flags, $blobAddress );
 351
 352                 if ( $blob === false ) {
 353                         wfLogWarning( __METHOD__ . ": Bad data in text row $textId." );
 354                         return false;
 355                 }
 356
 357                 return $blob;
 358         }
 359
 360         /**
 361          * Get a cache key for a given Blob address.
 362          *
 363          * The cache key is constructed in a way that allows cached blobs from the same database
 364          * to be re-used between wikis. For example, enwiki and frwiki will use the same cache keys
 365          * for blobs from the wikidatawiki database.
 366          *
 367          * @param string $blobAddress
 368          * @return string
 369          */
 370         private function getCacheKey( $blobAddress ) {
 371                 return $this->cache->makeGlobalKey(
 372                         'BlobStore',
 373                         'address',
 374                         $this->dbLoadBalancer->resolveDomainID( $this->wikiId ),
 375                         $blobAddress
 376                 );
 377         }
 378
 379         /**
 380          * Expand a raw data blob according to the flags given.
 381          *
 382          * MCR migration note: this replaces Revision::getRevisionText
 383          *
 384          * @note direct use is deprecated, use getBlob() or SlotRecord::getContent() instead.
 385          * @todo make this private, there should be no need to use this method outside this class.
 386          *
 387          * @param string $raw The raw blob data, to be processed according to $flags.
 388          *        May be the blob itself, or the blob compressed, or just the address
 389          *        of the actual blob, depending on $flags.
 390          * @param string|string[] $flags Blob flags, such as 'external' or 'gzip'.
 391          *   Note that not including 'utf-8' in $flags will cause the data to be decoded
 392          *   according to the legacy encoding specified via setLegacyEncoding.
 393          * @param string|null $cacheKey A blob address for use in the cache key. If not given,
 394          *   caching is disabled.
 395          *
 396          * @return false|string The expanded blob or false on failure
 397          */
 398         public function expandBlob( $raw, $flags, $cacheKey = null ) {
 399                 if ( is_string( $flags ) ) {
 400                         $flags = explode( ',', $flags );
 401                 }
 402
 403                 // Use external methods for external objects, text in table is URL-only then
 404                 if ( in_array( 'external', $flags ) ) {
 405                         $url = $raw;
 406                         $parts = explode( '://', $url, 2 );
 407                         if ( count( $parts ) == 1 || $parts[1] == '' ) {
 408                                 return false;
 409                         }
 410
 411                         if ( $cacheKey ) {
 412                                 // The cached value should be decompressed, so handle that and return here.
 413                                 return $this->cache->getWithSetCallback(
 414                                         $this->getCacheKey( $cacheKey ),
 415                                         $this->getCacheTTL(),
 416                                         function () use ( $url, $flags ) {
 417                                                 // Ignore $setOpts; blobs are immutable and negatives are not cached
 418                                                 $blob = ExternalStore::fetchFromURL( $url, [ 'wiki' => $this->wikiId ] );
 419
 420                                                 return $blob === false ? false : $this->decompressData( $blob, $flags );
 421                                         },
 422                                         [ 'pcGroup' => self::TEXT_CACHE_GROUP, 'pcTTL' => WANObjectCache::TTL_PROC_LONG ]
 423                                 );
 424                         } else {
 425                                 $blob = ExternalStore::fetchFromURL( $url, [ 'wiki' => $this->wikiId ] );
 426                                 return $blob === false ? false : $this->decompressData( $blob, $flags );
 427                         }
 428                 } else {
 429                         return $this->decompressData( $raw, $flags );
 430                 }
 431         }
 432
 433         /**
 434          * If $wgCompressRevisions is enabled, we will compress data.
 435          * The input string is modified in place.
 436          * Return value is the flags field: contains 'gzip' if the
 437          * data is compressed, and 'utf-8' if we're saving in UTF-8
 438          * mode.
 439          *
 440          * MCR migration note: this replaces Revision::compressRevisionText
 441          *
 442          * @note direct use is deprecated!
 443          * @todo make this private, there should be no need to use this method outside this class.
 444          *
 445          * @param mixed &$blob Reference to a text
 446          *
 447          * @return string
 448          */
 449         public function compressData( &$blob ) {
 450                 $blobFlags = [];
 451
 452                 // Revisions not marked as UTF-8 will have legacy decoding applied by decompressData().
 453                 // XXX: if $this->legacyEncoding is not set, we could skip this. That would however be
 454                 // risky, since $this->legacyEncoding being set in the future would lead to data corruption.
 455                 $blobFlags[] = 'utf-8';
 456
 457                 if ( $this->compressBlobs ) {
 458                         if ( function_exists( 'gzdeflate' ) ) {
 459                                 $deflated = gzdeflate( $blob );
 460
 461                                 if ( $deflated === false ) {
 462                                         wfLogWarning( __METHOD__ . ': gzdeflate() failed' );
 463                                 } else {
 464                                         $blob = $deflated;
 465                                         $blobFlags[] = 'gzip';
 466                                 }
 467                         } else {
 468                                 wfDebug( __METHOD__ . " -- no zlib support, not compressing\n" );
 469                         }
 470                 }
 471                 return implode( ',', $blobFlags );
 472         }
 473
 474         /**
 475          * Re-converts revision text according to its flags.
 476          *
 477          * MCR migration note: this replaces Revision::decompressRevisionText
 478          *
 479          * @note direct use is deprecated, use getBlob() or SlotRecord::getContent() instead.
 480          * @todo make this private, there should be no need to use this method outside this class.
 481          *
 482          * @param string $blob Blob in compressed/encoded form.
 483          * @param array $blobFlags Compression flags, such as 'gzip'.
 484          *   Note that not including 'utf-8' in $blobFlags will cause the data to be decoded
 485          *   according to the legacy encoding specified via setLegacyEncoding.
 486          *
 487          * @return string|bool Decompressed text, or false on failure
 488          */
 489         public function decompressData( $blob, array $blobFlags ) {
 490                 // Revision::decompressRevisionText accepted false here, so defend against that
 491                 Assert::parameterType( 'string', $blob, '$blob' );
 492
 493                 if ( in_array( 'error', $blobFlags ) ) {
 494                         // Error row, return false
 495                         return false;
 496                 }
 497
 498                 if ( in_array( 'gzip', $blobFlags ) ) {
 499                         # Deal with optional compression of archived pages.
 500                         # This can be done periodically via maintenance/compressOld.php, and
 501                         # as pages are saved if $wgCompressRevisions is set.
 502                         $blob = gzinflate( $blob );
 503
 504                         if ( $blob === false ) {
 505                                 wfWarn( __METHOD__ . ': gzinflate() failed' );
 506                                 return false;
 507                         }
 508                 }
 509
 510                 if ( in_array( 'object', $blobFlags ) ) {
 511                         # Generic compressed storage
 512                         $obj = unserialize( $blob );
 513                         if ( !is_object( $obj ) ) {
 514                                 // Invalid object
 515                                 return false;
 516                         }
 517                         $blob = $obj->getText();
 518                 }
 519
 520                 // Needed to support old revisions left over from from the 1.4 / 1.5 migration.
 521                 if ( $blob !== false && $this->legacyEncoding && $this->legacyEncodingConversionLang
 522                         && !in_array( 'utf-8', $blobFlags ) && !in_array( 'utf8', $blobFlags )
 523                 ) {
 524                         # Old revisions kept around in a legacy encoding?
 525                         # Upconvert on demand.
 526                         # ("utf8" checked for compatibility with some broken
 527                         #  conversion scripts 2008-12-30)
 528                         $blob = $this->legacyEncodingConversionLang->iconv( $this->legacyEncoding, 'UTF-8', $blob );
 529                 }
 530
 531                 return $blob;
 532         }
 533
 534         /**
 535          * Get the text cache TTL
 536          *
 537          * MCR migration note: this replaces Revision::getCacheTTL
 538          *
 539          * @return int
 540          */
 541         private function getCacheTTL() {
 542                 if ( $this->cache->getQoS( WANObjectCache::ATTR_EMULATION )
 543                                 <= WANObjectCache::QOS_EMULATION_SQL
 544                 ) {
 545                         // Do not cache RDBMs blobs in...the RDBMs store
 546                         $ttl = WANObjectCache::TTL_UNCACHEABLE;
 547                 } else {
 548                         $ttl = $this->cacheExpiry ?: WANObjectCache::TTL_UNCACHEABLE;
 549                 }
 550
 551                 return $ttl;
 552         }
 553
 554         /**
 555          * Returns an ID corresponding to the old_id field in the text table, corresponding
 556          * to the given $address.
 557          *
 558          * Currently, $address must start with 'tt:' followed by a decimal integer representing
 559          * the old_id; if $address does not start with 'tt:', null is returned. However,
 560          * the implementation may change to insert rows into the text table on the fly.
 561          * This implies that this method cannot be static.
 562          *
 563          * @note This method exists for use with the text table based storage schema.
 564          * It should not be assumed that is will function with all future kinds of content addresses.
 565          *
 566          * @deprecated since 1.31, so don't assume that all blob addresses refer to a row in the text
 567          * table. This method should become private once the relevant refactoring in WikiPage is
 568          * complete.
 569          *
 570          * @param string $address
 571          *
 572          * @return int|null
 573          */
 574         public function getTextIdFromAddress( $address ) {
 575                 list( $schema, $id, ) = self::splitBlobAddress( $address );
 576
 577                 if ( $schema !== 'tt' ) {
 578                         return null;
 579                 }
 580
 581                 $textId = intval( $id );
 582
 583                 if ( !$textId || $id !== (string)$textId ) {
 584                         throw new InvalidArgumentException( "Malformed text_id: $id" );
 585                 }
 586
 587                 return $textId;
 588         }
 589
 590         /**
 591          * Returns an address referring to content stored in the text table row with the given ID.
 592          * The address schema for blobs stored in the text table is "tt:" followed by an integer
 593          * that corresponds to a value of the old_id field.
 594          *
 595          * @deprecated since 1.31. This method should become private once the relevant refactoring
 596          * in WikiPage is complete.
 597          *
 598          * @param int $id
 599          *
 600          * @return string
 601          */
 602         public static function makeAddressFromTextId( $id ) {
 603                 return 'tt:' . $id;
 604         }
 605
 606         /**
 607          * Splits a blob address into three parts: the schema, the ID, and parameters/flags.
 608          *
 609          * @since 1.33
 610          *
 611          * @param string $address
 612          *
 613          * @throws InvalidArgumentException
 614          * @return array [ $schema, $id, $parameters ], with $parameters being an assoc array.
 615          */
 616         public static function splitBlobAddress( $address ) {
 617                 if ( !preg_match( '/^(\w+):(\w+)(\?(.*))?$/', $address, $m ) ) {
 618                         throw new InvalidArgumentException( "Bad blob address: $address" );
 619                 }
 620
 621                 $schema = strtolower( $m[1] );
 622                 $id = $m[2];
 623                 $parameters = isset( $m[4] ) ? wfCgiToArray( $m[4] ) : [];
 624
 625                 return [ $schema, $id, $parameters ];
 626         }
 627
 628         public function isReadOnly() {
 629                 if ( $this->useExternalStore && ExternalStore::defaultStoresAreReadOnly() ) {
 630                         return true;
 631                 }
 632
 633                 return ( $this->getDBLoadBalancer()->getReadOnlyReason() !== false );
 634         }
 635 }