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\ILoadBalancer;
  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 ILoadBalancer
  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 ILoadBalancer $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                 ILoadBalancer $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 ILoadBalancer
 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 ( $flags ) {
 224                                         $flags .= ',';
 225                                 }
 226                                 $flags .= 'external';
 227
 228                                 // TODO: we could also return an address for the external store directly here.
 229                                 // That would mean bypassing the text table entirely when the external store is
 230                                 // used. We'll need to assess expected fallout before doing that.
 231                         }
 232
 233                         $dbw = $this->getDBConnection( DB_MASTER );
 234
 235                         $old_id = $dbw->nextSequenceValue( 'text_old_id_seq' );
 236                         $dbw->insert(
 237                                 'text',
 238                                 [
 239                                         'old_id' => $old_id,
 240                                         'old_text' => $data,
 241                                         'old_flags' => $flags,
 242                                 ],
 243                                 __METHOD__
 244                         );
 245
 246                         $textId = $dbw->insertId();
 247
 248                         return self::makeAddressFromTextId( $textId );
 249                 } catch ( MWException $e ) {
 250                         throw new BlobAccessException( $e->getMessage(), 0, $e );
 251                 }
 252         }
 253
 254         /**
 255          * Retrieve a blob, given an address.
 256          * Currently hardcoded to the 'text' table storage engine.
 257          *
 258          * MCR migration note: this replaces Revision::loadText
 259          *
 260          * @param string $blobAddress
 261          * @param int $queryFlags
 262          *
 263          * @throws BlobAccessException
 264          * @return string
 265          */
 266         public function getBlob( $blobAddress, $queryFlags = 0 ) {
 267                 Assert::parameterType( 'string', $blobAddress, '$blobAddress' );
 268
 269                 // No negative caching; negative hits on text rows may be due to corrupted replica DBs
 270                 $blob = $this->cache->getWithSetCallback(
 271                         $this->getCacheKey( $blobAddress ),
 272                         $this->getCacheTTL(),
 273                         function ( $unused, &$ttl, &$setOpts ) use ( $blobAddress, $queryFlags ) {
 274                                 // Ignore $setOpts; blobs are immutable and negatives are not cached
 275                                 return $this->fetchBlob( $blobAddress, $queryFlags );
 276                         },
 277                         [ 'pcGroup' => self::TEXT_CACHE_GROUP, 'pcTTL' => IExpiringStore::TTL_PROC_LONG ]
 278                 );
 279
 280                 if ( $blob === false ) {
 281                         throw new BlobAccessException( 'Failed to load blob from address ' . $blobAddress );
 282                 }
 283
 284                 return $blob;
 285         }
 286
 287         /**
 288          * MCR migration note: this corresponds to Revision::fetchText
 289          *
 290          * @param string $blobAddress
 291          * @param int $queryFlags
 292          *
 293          * @throws BlobAccessException
 294          * @return string|false
 295          */
 296         private function fetchBlob( $blobAddress, $queryFlags ) {
 297                 list( $schema, $id, ) = self::splitBlobAddress( $blobAddress );
 298
 299                 //TODO: MCR: also support 'ex' schema with ExternalStore URLs, plus flags encoded in the URL!
 300                 if ( $schema === 'tt' ) {
 301                         $textId = intval( $id );
 302                 } else {
 303                         // XXX: change to better exceptions! That makes migration more difficult, though.
 304                         throw new BlobAccessException( "Unknown blob address schema: $schema" );
 305                 }
 306
 307                 if ( !$textId || $id !== (string)$textId ) {
 308                         // XXX: change to better exceptions! That makes migration more difficult, though.
 309                         throw new BlobAccessException( "Bad blob address: $blobAddress" );
 310                 }
 311
 312                 // Callers doing updates will pass in READ_LATEST as usual. Since the text/blob tables
 313                 // do not normally get rows changed around, set READ_LATEST_IMMUTABLE in those cases.
 314                 $queryFlags |= DBAccessObjectUtils::hasFlags( $queryFlags, self::READ_LATEST )
 315                         ? self::READ_LATEST_IMMUTABLE
 316                         : 0;
 317
 318                 list( $index, $options, $fallbackIndex, $fallbackOptions ) =
 319                         DBAccessObjectUtils::getDBOptions( $queryFlags );
 320
 321                 // Text data is immutable; check replica DBs first.
 322                 $row = $this->getDBConnection( $index )->selectRow(
 323                         'text',
 324                         [ 'old_text', 'old_flags' ],
 325                         [ 'old_id' => $textId ],
 326                         __METHOD__,
 327                         $options
 328                 );
 329
 330                 // Fallback to DB_MASTER in some cases if the row was not found, using the appropriate
 331                 // options, such as FOR UPDATE to avoid missing rows due to REPEATABLE-READ.
 332                 if ( !$row && $fallbackIndex !== null ) {
 333                         $row = $this->getDBConnection( $fallbackIndex )->selectRow(
 334                                 'text',
 335                                 [ 'old_text', 'old_flags' ],
 336                                 [ 'old_id' => $textId ],
 337                                 __METHOD__,
 338                                 $fallbackOptions
 339                         );
 340                 }
 341
 342                 if ( !$row ) {
 343                         wfWarn( __METHOD__ . ": No text row with ID $textId." );
 344                         return false;
 345                 }
 346
 347                 $blob = $this->expandBlob( $row->old_text, $row->old_flags, $blobAddress );
 348
 349                 if ( $blob === false ) {
 350                         wfLogWarning( __METHOD__ . ": Bad data in text row $textId." );
 351                         return false;
 352                 }
 353
 354                 return $blob;
 355         }
 356
 357         /**
 358          * Get a cache key for a given Blob address.
 359          *
 360          * The cache key is constructed in a way that allows cached blobs from the same database
 361          * to be re-used between wikis. For example, enwiki and frwiki will use the same cache keys
 362          * for blobs from the wikidatawiki database.
 363          *
 364          * @param string $blobAddress
 365          * @return string
 366          */
 367         private function getCacheKey( $blobAddress ) {
 368                 return $this->cache->makeGlobalKey(
 369                         'BlobStore',
 370                         'address',
 371                         $this->dbLoadBalancer->resolveDomainID( $this->wikiId ),
 372                         $blobAddress
 373                 );
 374         }
 375
 376         /**
 377          * Expand a raw data blob according to the flags given.
 378          *
 379          * MCR migration note: this replaces Revision::getRevisionText
 380          *
 381          * @note direct use is deprecated, use getBlob() or SlotRecord::getContent() instead.
 382          * @todo make this private, there should be no need to use this method outside this class.
 383          *
 384          * @param string $raw The raw blob data, to be processed according to $flags.
 385          *        May be the blob itself, or the blob compressed, or just the address
 386          *        of the actual blob, depending on $flags.
 387          * @param string|string[] $flags Blob flags, such as 'external' or 'gzip'.
 388          *   Note that not including 'utf-8' in $flags will cause the data to be decoded
 389          *   according to the legacy encoding specified via setLegacyEncoding.
 390          * @param string|null $cacheKey A blob address for use in the cache key. If not given,
 391          *   caching is disabled.
 392          *
 393          * @return false|string The expanded blob or false on failure
 394          */
 395         public function expandBlob( $raw, $flags, $cacheKey = null ) {
 396                 if ( is_string( $flags ) ) {
 397                         $flags = explode( ',', $flags );
 398                 }
 399
 400                 // Use external methods for external objects, text in table is URL-only then
 401                 if ( in_array( 'external', $flags ) ) {
 402                         $url = $raw;
 403                         $parts = explode( '://', $url, 2 );
 404                         if ( count( $parts ) == 1 || $parts[1] == '' ) {
 405                                 return false;
 406                         }
 407
 408                         if ( $cacheKey ) {
 409                                 // The cached value should be decompressed, so handle that and return here.
 410                                 return $this->cache->getWithSetCallback(
 411                                         $this->getCacheKey( $cacheKey ),
 412                                         $this->getCacheTTL(),
 413                                         function () use ( $url, $flags ) {
 414                                                 // Ignore $setOpts; blobs are immutable and negatives are not cached
 415                                                 $blob = ExternalStore::fetchFromURL( $url, [ 'wiki' => $this->wikiId ] );
 416
 417                                                 return $blob === false ? false : $this->decompressData( $blob, $flags );
 418                                         },
 419                                         [ 'pcGroup' => self::TEXT_CACHE_GROUP, 'pcTTL' => WANObjectCache::TTL_PROC_LONG ]
 420                                 );
 421                         } else {
 422                                 $blob = ExternalStore::fetchFromURL( $url, [ 'wiki' => $this->wikiId ] );
 423                                 return $blob === false ? false : $this->decompressData( $blob, $flags );
 424                         }
 425                 } else {
 426                         return $this->decompressData( $raw, $flags );
 427                 }
 428         }
 429
 430         /**
 431          * If $wgCompressRevisions is enabled, we will compress data.
 432          * The input string is modified in place.
 433          * Return value is the flags field: contains 'gzip' if the
 434          * data is compressed, and 'utf-8' if we're saving in UTF-8
 435          * mode.
 436          *
 437          * MCR migration note: this replaces Revision::compressRevisionText
 438          *
 439          * @note direct use is deprecated!
 440          * @todo make this private, there should be no need to use this method outside this class.
 441          *
 442          * @param mixed &$blob Reference to a text
 443          *
 444          * @return string
 445          */
 446         public function compressData( &$blob ) {
 447                 $blobFlags = [];
 448
 449                 // Revisions not marked as UTF-8 will have legacy decoding applied by decompressData().
 450                 // XXX: if $this->legacyEncoding is not set, we could skip this. That would however be
 451                 // risky, since $this->legacyEncoding being set in the future would lead to data corruption.
 452                 $blobFlags[] = 'utf-8';
 453
 454                 if ( $this->compressBlobs ) {
 455                         if ( function_exists( 'gzdeflate' ) ) {
 456                                 $deflated = gzdeflate( $blob );
 457
 458                                 if ( $deflated === false ) {
 459                                         wfLogWarning( __METHOD__ . ': gzdeflate() failed' );
 460                                 } else {
 461                                         $blob = $deflated;
 462                                         $blobFlags[] = 'gzip';
 463                                 }
 464                         } else {
 465                                 wfDebug( __METHOD__ . " -- no zlib support, not compressing\n" );
 466                         }
 467                 }
 468                 return implode( ',', $blobFlags );
 469         }
 470
 471         /**
 472          * Re-converts revision text according to its flags.
 473          *
 474          * MCR migration note: this replaces Revision::decompressRevisionText
 475          *
 476          * @note direct use is deprecated, use getBlob() or SlotRecord::getContent() instead.
 477          * @todo make this private, there should be no need to use this method outside this class.
 478          *
 479          * @param string $blob Blob in compressed/encoded form.
 480          * @param array $blobFlags Compression flags, such as 'gzip'.
 481          *   Note that not including 'utf-8' in $blobFlags will cause the data to be decoded
 482          *   according to the legacy encoding specified via setLegacyEncoding.
 483          *
 484          * @return string|bool Decompressed text, or false on failure
 485          */
 486         public function decompressData( $blob, array $blobFlags ) {
 487                 // Revision::decompressRevisionText accepted false here, so defend against that
 488                 Assert::parameterType( 'string', $blob, '$blob' );
 489
 490                 if ( in_array( 'error', $blobFlags ) ) {
 491                         // Error row, return false
 492                         return false;
 493                 }
 494
 495                 if ( in_array( 'gzip', $blobFlags ) ) {
 496                         # Deal with optional compression of archived pages.
 497                         # This can be done periodically via maintenance/compressOld.php, and
 498                         # as pages are saved if $wgCompressRevisions is set.
 499                         $blob = gzinflate( $blob );
 500
 501                         if ( $blob === false ) {
 502                                 wfWarn( __METHOD__ . ': gzinflate() failed' );
 503                                 return false;
 504                         }
 505                 }
 506
 507                 if ( in_array( 'object', $blobFlags ) ) {
 508                         # Generic compressed storage
 509                         $obj = unserialize( $blob );
 510                         if ( !is_object( $obj ) ) {
 511                                 // Invalid object
 512                                 return false;
 513                         }
 514                         $blob = $obj->getText();
 515                 }
 516
 517                 // Needed to support old revisions left over from from the 1.4 / 1.5 migration.
 518                 if ( $blob !== false && $this->legacyEncoding && $this->legacyEncodingConversionLang
 519                         && !in_array( 'utf-8', $blobFlags ) && !in_array( 'utf8', $blobFlags )
 520                 ) {
 521                         # Old revisions kept around in a legacy encoding?
 522                         # Upconvert on demand.
 523                         # ("utf8" checked for compatibility with some broken
 524                         #  conversion scripts 2008-12-30)
 525                         $blob = $this->legacyEncodingConversionLang->iconv( $this->legacyEncoding, 'UTF-8', $blob );
 526                 }
 527
 528                 return $blob;
 529         }
 530
 531         /**
 532          * Get the text cache TTL
 533          *
 534          * MCR migration note: this replaces Revision::getCacheTTL
 535          *
 536          * @return int
 537          */
 538         private function getCacheTTL() {
 539                 if ( $this->cache->getQoS( WANObjectCache::ATTR_EMULATION )
 540                                 <= WANObjectCache::QOS_EMULATION_SQL
 541                 ) {
 542                         // Do not cache RDBMs blobs in...the RDBMs store
 543                         $ttl = WANObjectCache::TTL_UNCACHEABLE;
 544                 } else {
 545                         $ttl = $this->cacheExpiry ?: WANObjectCache::TTL_UNCACHEABLE;
 546                 }
 547
 548                 return $ttl;
 549         }
 550
 551         /**
 552          * Returns an ID corresponding to the old_id field in the text table, corresponding
 553          * to the given $address.
 554          *
 555          * Currently, $address must start with 'tt:' followed by a decimal integer representing
 556          * the old_id; if $address does not start with 'tt:', null is returned. However,
 557          * the implementation may change to insert rows into the text table on the fly.
 558          * This implies that this method cannot be static.
 559          *
 560          * @note This method exists for use with the text table based storage schema.
 561          * It should not be assumed that is will function with all future kinds of content addresses.
 562          *
 563          * @deprecated since 1.31, so don't assume that all blob addresses refer to a row in the text
 564          * table. This method should become private once the relevant refactoring in WikiPage is
 565          * complete.
 566          *
 567          * @param string $address
 568          *
 569          * @return int|null
 570          */
 571         public function getTextIdFromAddress( $address ) {
 572                 list( $schema, $id, ) = self::splitBlobAddress( $address );
 573
 574                 if ( $schema !== 'tt' ) {
 575                         return null;
 576                 }
 577
 578                 $textId = intval( $id );
 579
 580                 if ( !$textId || $id !== (string)$textId ) {
 581                         throw new InvalidArgumentException( "Malformed text_id: $id" );
 582                 }
 583
 584                 return $textId;
 585         }
 586
 587         /**
 588          * Returns an address referring to content stored in the text table row with the given ID.
 589          * The address schema for blobs stored in the text table is "tt:" followed by an integer
 590          * that corresponds to a value of the old_id field.
 591          *
 592          * @deprecated since 1.31. This method should become private once the relevant refactoring
 593          * in WikiPage is complete.
 594          *
 595          * @param int $id
 596          *
 597          * @return string
 598          */
 599         public static function makeAddressFromTextId( $id ) {
 600                 return 'tt:' . $id;
 601         }
 602
 603         /**
 604          * Splits a blob address into three parts: the schema, the ID, and parameters/flags.
 605          *
 606          * @since 1.33
 607          *
 608          * @param string $address
 609          *
 610          * @throws InvalidArgumentException
 611          * @return array [ $schema, $id, $parameters ], with $parameters being an assoc array.
 612          */
 613         public static function splitBlobAddress( $address ) {
 614                 if ( !preg_match( '/^(\w+):(\w+)(\?(.*))?$/', $address, $m ) ) {
 615                         throw new InvalidArgumentException( "Bad blob address: $address" );
 616                 }
 617
 618                 $schema = strtolower( $m[1] );
 619                 $id = $m[2];
 620                 $parameters = isset( $m[4] ) ? wfCgiToArray( $m[4] ) : [];
 621
 622                 return [ $schema, $id, $parameters ];
 623         }
 624
 625         public function isReadOnly() {
 626                 if ( $this->useExternalStore && ExternalStore::defaultStoresAreReadOnly() ) {
 627                         return true;
 628                 }
 629
 630                 return ( $this->getDBLoadBalancer()->getReadOnlyReason() !== false );
 631         }
 632 }