includes/libs/filebackend/FileBackend.php

   1 <?php
   2 /**
   3  * @defgroup FileBackend File backend
   4  *
   5  * File backend is used to interact with file storage systems,
   6  * such as the local file system, NFS, or cloud storage systems.
   7  */
   8
   9 /**
  10  * Base class for all file backends.
  11  *
  12  * This program is free software; you can redistribute it and/or modify
  13  * it under the terms of the GNU General Public License as published by
  14  * the Free Software Foundation; either version 2 of the License, or
  15  * (at your option) any later version.
  16  *
  17  * This program is distributed in the hope that it will be useful,
  18  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  19  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  20  * GNU General Public License for more details.
  21  *
  22  * You should have received a copy of the GNU General Public License along
  23  * with this program; if not, write to the Free Software Foundation, Inc.,
  24  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  25  * http://www.gnu.org/copyleft/gpl.html
  26  *
  27  * @file
  28  * @ingroup FileBackend
  29  */
  30 use MediaWiki\FileBackend\FSFile\TempFSFileFactory;
  31 use Psr\Log\LoggerAwareInterface;
  32 use Psr\Log\LoggerInterface;
  33 use Wikimedia\ScopedCallback;
  34 use Psr\Log\NullLogger;
  35
  36 /**
  37  * @brief Base class for all file backend classes (including multi-write backends).
  38  *
  39  * This class defines the methods as abstract that subclasses must implement.
  40  * Outside callers can assume that all backends will have these functions.
  41  *
  42  * All "storage paths" are of the format "mwstore://<backend>/<container>/<path>".
  43  * The "backend" portion is unique name for the application to refer to a backend, while
  44  * the "container" portion is a top-level directory of the backend. The "path" portion
  45  * is a relative path that uses UNIX file system (FS) notation, though any particular
  46  * backend may not actually be using a local filesystem. Therefore, the relative paths
  47  * are only virtual.
  48  *
  49  * Backend contents are stored under "domain"-specific container names by default.
  50  * A domain is simply a logical umbrella for entities, such as those belonging to a certain
  51  * application or portion of a website, for example. A domain can be local or global.
  52  * Global (qualified) backends are achieved by configuring the "domain ID" to a constant.
  53  * Global domains are simpler, but local domains can be used by choosing a domain ID based on
  54  * the current context, such as which language of a website is being used.
  55  *
  56  * For legacy reasons, the FSFileBackend class allows manually setting the paths of
  57  * containers to ones that do not respect the "domain ID".
  58  *
  59  * In key/value (object) stores, containers are the only hierarchy (the rest is emulated).
  60  * FS-based backends are somewhat more restrictive due to the existence of real
  61  * directory files; a regular file cannot have the same name as a directory. Other
  62  * backends with virtual directories may not have this limitation. Callers should
  63  * store files in such a way that no files and directories are under the same path.
  64  *
  65  * In general, this class allows for callers to access storage through the same
  66  * interface, without regard to the underlying storage system. However, calling code
  67  * must follow certain patterns and be aware of certain things to ensure compatibility:
  68  *   - a) Always call prepare() on the parent directory before trying to put a file there;
  69  *        key/value stores only need the container to exist first, but filesystems need
  70  *        all the parent directories to exist first (prepare() is aware of all this)
  71  *   - b) Always call clean() on a directory when it might become empty to avoid empty
  72  *        directory buildup on filesystems; key/value stores never have empty directories,
  73  *        so doing this helps preserve consistency in both cases
  74  *   - c) Likewise, do not rely on the existence of empty directories for anything;
  75  *        calling directoryExists() on a path that prepare() was previously called on
  76  *        will return false for key/value stores if there are no files under that path
  77  *   - d) Never alter the resulting FSFile returned from getLocalReference(), as it could
  78  *        either be a copy of the source file in /tmp or the original source file itself
  79  *   - e) Use a file layout that results in never attempting to store files over directories
  80  *        or directories over files; key/value stores allow this but filesystems do not
  81  *   - f) Use ASCII file names (e.g. base32, IDs, hashes) to avoid Unicode issues in Windows
  82  *   - g) Do not assume that move operations are atomic (difficult with key/value stores)
  83  *   - h) Do not assume that file stat or read operations always have immediate consistency;
  84  *        various methods have a "latest" flag that should always be used if up-to-date
  85  *        information is required (this trades performance for correctness as needed)
  86  *   - i) Do not assume that directory listings have immediate consistency
  87  *
  88  * Methods of subclasses should avoid throwing exceptions at all costs.
  89  * As a corollary, external dependencies should be kept to a minimum.
  90  *
  91  * @ingroup FileBackend
  92  * @since 1.19
  93  */
  94 abstract class FileBackend implements LoggerAwareInterface {
  95         /** @var string Unique backend name */
  96         protected $name;
  97
  98         /** @var string Unique domain name */
  99         protected $domainId;
 100
 101         /** @var string Read-only explanation message */
 102         protected $readOnly;
 103
 104         /** @var string When to do operations in parallel */
 105         protected $parallelize;
 106
 107         /** @var int How many operations can be done in parallel */
 108         protected $concurrency;
 109
 110         /** @var TempFSFileFactory */
 111         protected $tmpFileFactory;
 112
 113         /** @var LockManager */
 114         protected $lockManager;
 115         /** @var FileJournal */
 116         protected $fileJournal;
 117         /** @var LoggerInterface */
 118         protected $logger;
 119         /** @var callable|null */
 120         protected $profiler;
 121
 122         /** @var callable */
 123         protected $obResetFunc;
 124         /** @var callable */
 125         protected $streamMimeFunc;
 126         /** @var callable */
 127         protected $statusWrapper;
 128
 129         /** Bitfield flags for supported features */
 130         const ATTR_HEADERS = 1; // files can be tagged with standard HTTP headers
 131         const ATTR_METADATA = 2; // files can be stored with metadata key/values
 132         const ATTR_UNICODE_PATHS = 4; // files can have Unicode paths (not just ASCII)
 133
 134         /** @var false Idiom for "no info; non-existant file" (since 1.34) */
 135         const STAT_ABSENT = false;
 136
 137         /** @var null Idiom for "no info; I/O errors" (since 1.34) */
 138         const STAT_ERROR = null;
 139         /** @var null Idiom for "no file/directory list; I/O errors" (since 1.34) */
 140         const LIST_ERROR = null;
 141         /** @var null Idiom for "no temp URL; not supported or I/O errors" (since 1.34) */
 142         const TEMPURL_ERROR = null;
 143         /** @var null Idiom for "existence unknown; I/O errors" (since 1.34) */
 144         const EXISTENCE_ERROR = null;
 145
 146         /** @var false Idiom for "no timestamp; missing file or I/O errors" (since 1.34) */
 147         const TIMESTAMP_FAIL = false;
 148         /** @var false Idiom for "no content; missing file or I/O errors" (since 1.34) */
 149         const CONTENT_FAIL = false;
 150         /** @var false Idiom for "no metadata; missing file or I/O errors" (since 1.34) */
 151         const XATTRS_FAIL = false;
 152         /** @var false Idiom for "no size; missing file or I/O errors" (since 1.34) */
 153         const SIZE_FAIL = false;
 154         /** @var false Idiom for "no SHA1 hash; missing file or I/O errors" (since 1.34) */
 155         const SHA1_FAIL = false;
 156
 157         /**
 158          * Create a new backend instance from configuration.
 159          * This should only be called from within FileBackendGroup.
 160          *
 161          * @param array $config Parameters include:
 162          *   - name : The unique name of this backend.
 163          *      This should consist of alphanumberic, '-', and '_' characters.
 164          *      This name should not be changed after use (e.g. with journaling).
 165          *      Note that the name is *not* used in actual container names.
 166          *   - domainId : Prefix to container names that is unique to this backend.
 167          *      It should only consist of alphanumberic, '-', and '_' characters.
 168          *      This ID is what avoids collisions if multiple logical backends
 169          *      use the same storage system, so this should be set carefully.
 170          *   - lockManager : LockManager object to use for any file locking.
 171          *      If not provided, then no file locking will be enforced.
 172          *   - fileJournal : FileJournal object to use for logging changes to files.
 173          *      If not provided, then change journaling will be disabled.
 174          *   - readOnly : Write operations are disallowed if this is a non-empty string.
 175          *      It should be an explanation for the backend being read-only.
 176          *   - parallelize : When to do file operations in parallel (when possible).
 177          *      Allowed values are "implicit", "explicit" and "off".
 178          *   - concurrency : How many file operations can be done in parallel.
 179          *   - tmpDirectory : Directory to use for temporary files.
 180          *   - tmpFileFactory : Optional TempFSFileFactory object. Only has an effect if
 181          *      tmpDirectory is not set. If both are unset or null, then the backend will
 182          *      try to discover a usable temporary directory.
 183          *   - obResetFunc : alternative callback to clear the output buffer
 184          *   - streamMimeFunc : alternative method to determine the content type from the path
 185          *   - logger : Optional PSR logger object.
 186          *   - profiler : Optional callback that takes a section name argument and returns
 187          *      a ScopedCallback instance that ends the profile section in its destructor.
 188          * @throws InvalidArgumentException
 189          */
 190         public function __construct( array $config ) {
 191                 $this->name = $config['name'];
 192                 $this->domainId = $config['domainId'] // e.g. "my_wiki-en_"
 193                         ?? $config['wikiId']; // b/c alias
 194                 if ( !preg_match( '!^[a-zA-Z0-9-_]{1,255}$!', $this->name ) ) {
 195                         throw new InvalidArgumentException( "Backend name '{$this->name}' is invalid." );
 196                 } elseif ( !is_string( $this->domainId ) ) {
 197                         throw new InvalidArgumentException(
 198                                 "Backend domain ID not provided for '{$this->name}'." );
 199                 }
 200                 $this->lockManager = $config['lockManager'] ?? new NullLockManager( [] );
 201                 $this->fileJournal = $config['fileJournal']
 202                         ?? FileJournal::factory( [ 'class' => NullFileJournal::class ], $this->name );
 203                 $this->readOnly = isset( $config['readOnly'] )
 204                         ? (string)$config['readOnly']
 205                         : '';
 206                 $this->parallelize = isset( $config['parallelize'] )
 207                         ? (string)$config['parallelize']
 208                         : 'off';
 209                 $this->concurrency = isset( $config['concurrency'] )
 210                         ? (int)$config['concurrency']
 211                         : 50;
 212                 $this->obResetFunc = $config['obResetFunc'] ?? [ $this, 'resetOutputBuffer' ];
 213                 $this->streamMimeFunc = $config['streamMimeFunc'] ?? null;
 214                 $this->statusWrapper = $config['statusWrapper'] ?? null;
 215
 216                 $this->profiler = $config['profiler'] ?? null;
 217                 if ( !is_callable( $this->profiler ) ) {
 218                         $this->profiler = null;
 219                 }
 220                 $this->logger = $config['logger'] ?? new NullLogger();
 221                 $this->statusWrapper = $config['statusWrapper'] ?? null;
 222                 // tmpDirectory gets precedence for backward compatibility
 223                 if ( isset( $config['tmpDirectory'] ) ) {
 224                         $this->tmpFileFactory = new TempFSFileFactory( $config['tmpDirectory'] );
 225                 } else {
 226                         $this->tmpFileFactory = $config['tmpFileFactory'] ?? new TempFSFileFactory();
 227                 }
 228         }
 229
 230         public function setLogger( LoggerInterface $logger ) {
 231                 $this->logger = $logger;
 232         }
 233
 234         /**
 235          * Get the unique backend name
 236          *
 237          * We may have multiple different backends of the same type.
 238          * For example, we can have two Swift backends using different proxies.
 239          *
 240          * @return string
 241          */
 242         final public function getName() {
 243                 return $this->name;
 244         }
 245
 246         /**
 247          * Get the domain identifier used for this backend (possibly empty).
 248          *
 249          * @return string
 250          * @since 1.28
 251          */
 252         final public function getDomainId() {
 253                 return $this->domainId;
 254         }
 255
 256         /**
 257          * Alias to getDomainId()
 258          *
 259          * @return string
 260          * @since 1.20
 261          * @deprecated Since 1.34 Use getDomainId()
 262          */
 263         final public function getWikiId() {
 264                 return $this->getDomainId();
 265         }
 266
 267         /**
 268          * Check if this backend is read-only
 269          *
 270          * @return bool
 271          */
 272         final public function isReadOnly() {
 273                 return ( $this->readOnly != '' );
 274         }
 275
 276         /**
 277          * Get an explanatory message if this backend is read-only
 278          *
 279          * @return string|bool Returns false if the backend is not read-only
 280          */
 281         final public function getReadOnlyReason() {
 282                 return ( $this->readOnly != '' ) ? $this->readOnly : false;
 283         }
 284
 285         /**
 286          * Get the a bitfield of extra features supported by the backend medium
 287          *
 288          * @return int Bitfield of FileBackend::ATTR_* flags
 289          * @since 1.23
 290          */
 291         public function getFeatures() {
 292                 return self::ATTR_UNICODE_PATHS;
 293         }
 294
 295         /**
 296          * Check if the backend medium supports a field of extra features
 297          *
 298          * @param int $bitfield Bitfield of FileBackend::ATTR_* flags
 299          * @return bool
 300          * @since 1.23
 301          */
 302         final public function hasFeatures( $bitfield ) {
 303                 return ( $this->getFeatures() & $bitfield ) === $bitfield;
 304         }
 305
 306         /**
 307          * This is the main entry point into the backend for write operations.
 308          * Callers supply an ordered list of operations to perform as a transaction.
 309          * Files will be locked, the stat cache cleared, and then the operations attempted.
 310          * If any serious errors occur, all attempted operations will be rolled back.
 311          *
 312          * $ops is an array of arrays. The outer array holds a list of operations.
 313          * Each inner array is a set of key value pairs that specify an operation.
 314          *
 315          * Supported operations and their parameters. The supported actions are:
 316          *  - create
 317          *  - store
 318          *  - copy
 319          *  - move
 320          *  - delete
 321          *  - describe (since 1.21)
 322          *  - null
 323          *
 324          * FSFile/TempFSFile object support was added in 1.27.
 325          *
 326          * a) Create a new file in storage with the contents of a string
 327          * @code
 328          *     [
 329          *         'op'                  => 'create',
 330          *         'dst'                 => <storage path>,
 331          *         'content'             => <string of new file contents>,
 332          *         'overwrite'           => <boolean>,
 333          *         'overwriteSame'       => <boolean>,
 334          *         'headers'             => <HTTP header name/value map> # since 1.21
 335          *     ]
 336          * @endcode
 337          *
 338          * b) Copy a file system file into storage
 339          * @code
 340          *     [
 341          *         'op'                  => 'store',
 342          *         'src'                 => <file system path, FSFile, or TempFSFile>,
 343          *         'dst'                 => <storage path>,
 344          *         'overwrite'           => <boolean>,
 345          *         'overwriteSame'       => <boolean>,
 346          *         'headers'             => <HTTP header name/value map> # since 1.21
 347          *     ]
 348          * @endcode
 349          *
 350          * c) Copy a file within storage
 351          * @code
 352          *     [
 353          *         'op'                  => 'copy',
 354          *         'src'                 => <storage path>,
 355          *         'dst'                 => <storage path>,
 356          *         'overwrite'           => <boolean>,
 357          *         'overwriteSame'       => <boolean>,
 358          *         'ignoreMissingSource' => <boolean>, # since 1.21
 359          *         'headers'             => <HTTP header name/value map> # since 1.21
 360          *     ]
 361          * @endcode
 362          *
 363          * d) Move a file within storage
 364          * @code
 365          *     [
 366          *         'op'                  => 'move',
 367          *         'src'                 => <storage path>,
 368          *         'dst'                 => <storage path>,
 369          *         'overwrite'           => <boolean>,
 370          *         'overwriteSame'       => <boolean>,
 371          *         'ignoreMissingSource' => <boolean>, # since 1.21
 372          *         'headers'             => <HTTP header name/value map> # since 1.21
 373          *     ]
 374          * @endcode
 375          *
 376          * e) Delete a file within storage
 377          * @code
 378          *     [
 379          *         'op'                  => 'delete',
 380          *         'src'                 => <storage path>,
 381          *         'ignoreMissingSource' => <boolean>
 382          *     ]
 383          * @endcode
 384          *
 385          * f) Update metadata for a file within storage
 386          * @code
 387          *     [
 388          *         'op'                  => 'describe',
 389          *         'src'                 => <storage path>,
 390          *         'headers'             => <HTTP header name/value map>
 391          *     ]
 392          * @endcode
 393          *
 394          * g) Do nothing (no-op)
 395          * @code
 396          *     [
 397          *         'op'                  => 'null',
 398          *     ]
 399          * @endcode
 400          *
 401          * Boolean flags for operations (operation-specific):
 402          *   - ignoreMissingSource : The operation will simply succeed and do
 403          *                           nothing if the source file does not exist.
 404          *   - overwrite           : Any destination file will be overwritten.
 405          *   - overwriteSame       : If a file already exists at the destination with the
 406          *                           same contents, then do nothing to the destination file
 407          *                           instead of giving an error. This does not compare headers.
 408          *                           This option is ignored if 'overwrite' is already provided.
 409          *   - headers             : If supplied, the result of merging these headers with any
 410          *                           existing source file headers (replacing conflicting ones)
 411          *                           will be set as the destination file headers. Headers are
 412          *                           deleted if their value is set to the empty string. When a
 413          *                           file has headers they are included in responses to GET and
 414          *                           HEAD requests to the backing store for that file.
 415          *                           Header values should be no larger than 255 bytes, except for
 416          *                           Content-Disposition. The system might ignore or truncate any
 417          *                           headers that are too long to store (exact limits will vary).
 418          *                           Backends that don't support metadata ignore this. (since 1.21)
 419          *
 420          * $opts is an associative of boolean flags, including:
 421          *   - force               : Operation precondition errors no longer trigger an abort.
 422          *                           Any remaining operations are still attempted. Unexpected
 423          *                           failures may still cause remaining operations to be aborted.
 424          *   - nonLocking          : No locks are acquired for the operations.
 425          *                           This can increase performance for non-critical writes.
 426          *                           This has no effect unless the 'force' flag is set.
 427          *   - nonJournaled        : Don't log this operation batch in the file journal.
 428          *                           This limits the ability of recovery scripts.
 429          *   - parallelize         : Try to do operations in parallel when possible.
 430          *   - bypassReadOnly      : Allow writes in read-only mode. (since 1.20)
 431          *   - preserveCache       : Don't clear the process cache before checking files.
 432          *                           This should only be used if all entries in the process
 433          *                           cache were added after the files were already locked. (since 1.20)
 434          *
 435          * @note Remarks on locking:
 436          * File system paths given to operations should refer to files that are
 437          * already locked or otherwise safe from modification from other processes.
 438          * Normally these files will be new temp files, which should be adequate.
 439          *
 440          * @par Return value:
 441          *
 442          * This returns a Status, which contains all warnings and fatals that occurred
 443          * during the operation. The 'failCount', 'successCount', and 'success' members
 444          * will reflect each operation attempted.
 445          *
 446          * The StatusValue will be "OK" unless:
 447          *   - a) unexpected operation errors occurred (network partitions, disk full...)
 448          *   - b) predicted operation errors occurred and 'force' was not set
 449          *
 450          * @param array $ops List of operations to execute in order
 451          * @codingStandardsIgnoreStart
 452          * @phan-param array{ignoreMissingSource?:bool,overwrite?:bool,overwriteSame?:bool,headers?:bool} $ops
 453          * @param array $opts Batch operation options
 454          * @phan-param array{force?:bool,nonLocking?:bool,nonJournaled?:bool,parallelize?:bool,bypassReadOnly?:bool,preserveCache?:bool} $opts
 455          * @codingStandardsIgnoreEnd
 456          * @return StatusValue
 457          */
 458         final public function doOperations( array $ops, array $opts = [] ) {
 459                 if ( empty( $opts['bypassReadOnly'] ) && $this->isReadOnly() ) {
 460                         return $this->newStatus( 'backend-fail-readonly', $this->name, $this->readOnly );
 461                 }
 462                 if ( $ops === [] ) {
 463                         return $this->newStatus(); // nothing to do
 464                 }
 465
 466                 $ops = $this->resolveFSFileObjects( $ops );
 467                 if ( empty( $opts['force'] ) ) { // sanity
 468                         unset( $opts['nonLocking'] );
 469                 }
 470
 471                 /** @noinspection PhpUnusedLocalVariableInspection */
 472                 $scope = ScopedCallback::newScopedIgnoreUserAbort(); // try to ignore client aborts
 473
 474                 return $this->doOperationsInternal( $ops, $opts );
 475         }
 476
 477         /**
 478          * @see FileBackend::doOperations()
 479          * @param array $ops
 480          * @param array $opts
 481          * @return StatusValue
 482          */
 483         abstract protected function doOperationsInternal( array $ops, array $opts );
 484
 485         /**
 486          * Same as doOperations() except it takes a single operation.
 487          * If you are doing a batch of operations that should either
 488          * all succeed or all fail, then use that function instead.
 489          *
 490          * @see FileBackend::doOperations()
 491          *
 492          * @param array $op Operation
 493          * @param array $opts Operation options
 494          * @return StatusValue
 495          */
 496         final public function doOperation( array $op, array $opts = [] ) {
 497                 return $this->doOperations( [ $op ], $opts );
 498         }
 499
 500         /**
 501          * Performs a single create operation.
 502          * This sets $params['op'] to 'create' and passes it to doOperation().
 503          *
 504          * @see FileBackend::doOperation()
 505          *
 506          * @param array $params Operation parameters
 507          * @param array $opts Operation options
 508          * @return StatusValue
 509          */
 510         final public function create( array $params, array $opts = [] ) {
 511                 return $this->doOperation( [ 'op' => 'create' ] + $params, $opts );
 512         }
 513
 514         /**
 515          * Performs a single store operation.
 516          * This sets $params['op'] to 'store' and passes it to doOperation().
 517          *
 518          * @see FileBackend::doOperation()
 519          *
 520          * @param array $params Operation parameters
 521          * @param array $opts Operation options
 522          * @return StatusValue
 523          */
 524         final public function store( array $params, array $opts = [] ) {
 525                 return $this->doOperation( [ 'op' => 'store' ] + $params, $opts );
 526         }
 527
 528         /**
 529          * Performs a single copy operation.
 530          * This sets $params['op'] to 'copy' and passes it to doOperation().
 531          *
 532          * @see FileBackend::doOperation()
 533          *
 534          * @param array $params Operation parameters
 535          * @param array $opts Operation options
 536          * @return StatusValue
 537          */
 538         final public function copy( array $params, array $opts = [] ) {
 539                 return $this->doOperation( [ 'op' => 'copy' ] + $params, $opts );
 540         }
 541
 542         /**
 543          * Performs a single move operation.
 544          * This sets $params['op'] to 'move' and passes it to doOperation().
 545          *
 546          * @see FileBackend::doOperation()
 547          *
 548          * @param array $params Operation parameters
 549          * @param array $opts Operation options
 550          * @return StatusValue
 551          */
 552         final public function move( array $params, array $opts = [] ) {
 553                 return $this->doOperation( [ 'op' => 'move' ] + $params, $opts );
 554         }
 555
 556         /**
 557          * Performs a single delete operation.
 558          * This sets $params['op'] to 'delete' and passes it to doOperation().
 559          *
 560          * @see FileBackend::doOperation()
 561          *
 562          * @param array $params Operation parameters
 563          * @param array $opts Operation options
 564          * @return StatusValue
 565          */
 566         final public function delete( array $params, array $opts = [] ) {
 567                 return $this->doOperation( [ 'op' => 'delete' ] + $params, $opts );
 568         }
 569
 570         /**
 571          * Performs a single describe operation.
 572          * This sets $params['op'] to 'describe' and passes it to doOperation().
 573          *
 574          * @see FileBackend::doOperation()
 575          *
 576          * @param array $params Operation parameters
 577          * @param array $opts Operation options
 578          * @return StatusValue
 579          * @since 1.21
 580          */
 581         final public function describe( array $params, array $opts = [] ) {
 582                 return $this->doOperation( [ 'op' => 'describe' ] + $params, $opts );
 583         }
 584
 585         /**
 586          * Perform a set of independent file operations on some files.
 587          *
 588          * This does no locking, nor journaling, and possibly no stat calls.
 589          * Any destination files that already exist will be overwritten.
 590          * This should *only* be used on non-original files, like cache files.
 591          *
 592          * Supported operations and their parameters:
 593          *  - create
 594          *  - store
 595          *  - copy
 596          *  - move
 597          *  - delete
 598          *  - describe (since 1.21)
 599          *  - null
 600          *
 601          * FSFile/TempFSFile object support was added in 1.27.
 602          *
 603          * a) Create a new file in storage with the contents of a string
 604          * @code
 605          *     [
 606          *         'op'                  => 'create',
 607          *         'dst'                 => <storage path>,
 608          *         'content'             => <string of new file contents>,
 609          *         'headers'             => <HTTP header name/value map> # since 1.21
 610          *     ]
 611          * @endcode
 612          *
 613          * b) Copy a file system file into storage
 614          * @code
 615          *     [
 616          *         'op'                  => 'store',
 617          *         'src'                 => <file system path, FSFile, or TempFSFile>,
 618          *         'dst'                 => <storage path>,
 619          *         'headers'             => <HTTP header name/value map> # since 1.21
 620          *     ]
 621          * @endcode
 622          *
 623          * c) Copy a file within storage
 624          * @code
 625          *     [
 626          *         'op'                  => 'copy',
 627          *         'src'                 => <storage path>,
 628          *         'dst'                 => <storage path>,
 629          *         'ignoreMissingSource' => <boolean>, # since 1.21
 630          *         'headers'             => <HTTP header name/value map> # since 1.21
 631          *     ]
 632          * @endcode
 633          *
 634          * d) Move a file within storage
 635          * @code
 636          *     [
 637          *         'op'                  => 'move',
 638          *         'src'                 => <storage path>,
 639          *         'dst'                 => <storage path>,
 640          *         'ignoreMissingSource' => <boolean>, # since 1.21
 641          *         'headers'             => <HTTP header name/value map> # since 1.21
 642          *     ]
 643          * @endcode
 644          *
 645          * e) Delete a file within storage
 646          * @code
 647          *     [
 648          *         'op'                  => 'delete',
 649          *         'src'                 => <storage path>,
 650          *         'ignoreMissingSource' => <boolean>
 651          *     ]
 652          * @endcode
 653          *
 654          * f) Update metadata for a file within storage
 655          * @code
 656          *     [
 657          *         'op'                  => 'describe',
 658          *         'src'                 => <storage path>,
 659          *         'headers'             => <HTTP header name/value map>
 660          *     ]
 661          * @endcode
 662          *
 663          * g) Do nothing (no-op)
 664          * @code
 665          *     [
 666          *         'op'                  => 'null',
 667          *     ]
 668          * @endcode
 669          *
 670          * @par Boolean flags for operations (operation-specific):
 671          *   - ignoreMissingSource : The operation will simply succeed and do
 672          *                           nothing if the source file does not exist.
 673          *   - headers             : If supplied with a header name/value map, the backend will
 674          *                           reply with these headers when GETs/HEADs of the destination
 675          *                           file are made. Header values should be smaller than 256 bytes.
 676          *                           Content-Disposition headers can be longer, though the system
 677          *                           might ignore or truncate ones that are too long to store.
 678          *                           Existing headers will remain, but these will replace any
 679          *                           conflicting previous headers, and headers will be removed
 680          *                           if they are set to an empty string.
 681          *                           Backends that don't support metadata ignore this. (since 1.21)
 682          *
 683          * $opts is an associative of boolean flags, including:
 684          *   - bypassReadOnly      : Allow writes in read-only mode (since 1.20)
 685          *
 686          * @par Return value:
 687          * This returns a Status, which contains all warnings and fatals that occurred
 688          * during the operation. The 'failCount', 'successCount', and 'success' members
 689          * will reflect each operation attempted for the given files. The StatusValue will be
 690          * considered "OK" as long as no fatal errors occurred.
 691          *
 692          * @param array $ops Set of operations to execute
 693          * @phan-param array{ignoreMissingSource?:bool,headers?:bool} $ops
 694          * @param array $opts Batch operation options
 695          * @phan-param array{bypassReadOnly?:bool} $opts
 696          * @return StatusValue
 697          * @since 1.20
 698          */
 699         final public function doQuickOperations( array $ops, array $opts = [] ) {
 700                 if ( empty( $opts['bypassReadOnly'] ) && $this->isReadOnly() ) {
 701                         return $this->newStatus( 'backend-fail-readonly', $this->name, $this->readOnly );
 702                 }
 703                 if ( $ops === [] ) {
 704                         return $this->newStatus(); // nothing to do
 705                 }
 706
 707                 $ops = $this->resolveFSFileObjects( $ops );
 708                 foreach ( $ops as &$op ) {
 709                         $op['overwrite'] = true; // avoids RTTs in key/value stores
 710                 }
 711
 712                 /** @noinspection PhpUnusedLocalVariableInspection */
 713                 $scope = ScopedCallback::newScopedIgnoreUserAbort(); // try to ignore client aborts
 714
 715                 return $this->doQuickOperationsInternal( $ops );
 716         }
 717
 718         /**
 719          * @see FileBackend::doQuickOperations()
 720          * @param array $ops
 721          * @return StatusValue
 722          * @since 1.20
 723          */
 724         abstract protected function doQuickOperationsInternal( array $ops );
 725
 726         /**
 727          * Same as doQuickOperations() except it takes a single operation.
 728          * If you are doing a batch of operations, then use that function instead.
 729          *
 730          * @see FileBackend::doQuickOperations()
 731          *
 732          * @param array $op Operation
 733          * @return StatusValue
 734          * @since 1.20
 735          */
 736         final public function doQuickOperation( array $op ) {
 737                 return $this->doQuickOperations( [ $op ] );
 738         }
 739
 740         /**
 741          * Performs a single quick create operation.
 742          * This sets $params['op'] to 'create' and passes it to doQuickOperation().
 743          *
 744          * @see FileBackend::doQuickOperation()
 745          *
 746          * @param array $params Operation parameters
 747          * @return StatusValue
 748          * @since 1.20
 749          */
 750         final public function quickCreate( array $params ) {
 751                 return $this->doQuickOperation( [ 'op' => 'create' ] + $params );
 752         }
 753
 754         /**
 755          * Performs a single quick store operation.
 756          * This sets $params['op'] to 'store' and passes it to doQuickOperation().
 757          *
 758          * @see FileBackend::doQuickOperation()
 759          *
 760          * @param array $params Operation parameters
 761          * @return StatusValue
 762          * @since 1.20
 763          */
 764         final public function quickStore( array $params ) {
 765                 return $this->doQuickOperation( [ 'op' => 'store' ] + $params );
 766         }
 767
 768         /**
 769          * Performs a single quick copy operation.
 770          * This sets $params['op'] to 'copy' and passes it to doQuickOperation().
 771          *
 772          * @see FileBackend::doQuickOperation()
 773          *
 774          * @param array $params Operation parameters
 775          * @return StatusValue
 776          * @since 1.20
 777          */
 778         final public function quickCopy( array $params ) {
 779                 return $this->doQuickOperation( [ 'op' => 'copy' ] + $params );
 780         }
 781
 782         /**
 783          * Performs a single quick move operation.
 784          * This sets $params['op'] to 'move' and passes it to doQuickOperation().
 785          *
 786          * @see FileBackend::doQuickOperation()
 787          *
 788          * @param array $params Operation parameters
 789          * @return StatusValue
 790          * @since 1.20
 791          */
 792         final public function quickMove( array $params ) {
 793                 return $this->doQuickOperation( [ 'op' => 'move' ] + $params );
 794         }
 795
 796         /**
 797          * Performs a single quick delete operation.
 798          * This sets $params['op'] to 'delete' and passes it to doQuickOperation().
 799          *
 800          * @see FileBackend::doQuickOperation()
 801          *
 802          * @param array $params Operation parameters
 803          * @return StatusValue
 804          * @since 1.20
 805          */
 806         final public function quickDelete( array $params ) {
 807                 return $this->doQuickOperation( [ 'op' => 'delete' ] + $params );
 808         }
 809
 810         /**
 811          * Performs a single quick describe operation.
 812          * This sets $params['op'] to 'describe' and passes it to doQuickOperation().
 813          *
 814          * @see FileBackend::doQuickOperation()
 815          *
 816          * @param array $params Operation parameters
 817          * @return StatusValue
 818          * @since 1.21
 819          */
 820         final public function quickDescribe( array $params ) {
 821                 return $this->doQuickOperation( [ 'op' => 'describe' ] + $params );
 822         }
 823
 824         /**
 825          * Concatenate a list of storage files into a single file system file.
 826          * The target path should refer to a file that is already locked or
 827          * otherwise safe from modification from other processes. Normally,
 828          * the file will be a new temp file, which should be adequate.
 829          *
 830          * @param array $params Operation parameters, include:
 831          *   - srcs        : ordered source storage paths (e.g. chunk1, chunk2, ...)
 832          *   - dst         : file system path to 0-byte temp file
 833          *   - parallelize : try to do operations in parallel when possible
 834          * @return StatusValue
 835          */
 836         abstract public function concatenate( array $params );
 837
 838         /**
 839          * Prepare a storage directory for usage.
 840          * This will create any required containers and parent directories.
 841          * Backends using key/value stores only need to create the container.
 842          *
 843          * The 'noAccess' and 'noListing' parameters works the same as in secure(),
 844          * except they are only applied *if* the directory/container had to be created.
 845          * These flags should always be set for directories that have private files.
 846          * However, setting them is not guaranteed to actually do anything.
 847          * Additional server configuration may be needed to achieve the desired effect.
 848          *
 849          * @param array $params Parameters include:
 850          *   - dir            : storage directory
 851          *   - noAccess       : try to deny file access (since 1.20)
 852          *   - noListing      : try to deny file listing (since 1.20)
 853          *   - bypassReadOnly : allow writes in read-only mode (since 1.20)
 854          * @return StatusValue
 855          */
 856         final public function prepare( array $params ) {
 857                 if ( empty( $params['bypassReadOnly'] ) && $this->isReadOnly() ) {
 858                         return $this->newStatus( 'backend-fail-readonly', $this->name, $this->readOnly );
 859                 }
 860                 /** @noinspection PhpUnusedLocalVariableInspection */
 861                 $scope = ScopedCallback::newScopedIgnoreUserAbort(); // try to ignore client aborts
 862                 return $this->doPrepare( $params );
 863         }
 864
 865         /**
 866          * @see FileBackend::prepare()
 867          * @param array $params
 868          * @return StatusValue
 869          */
 870         abstract protected function doPrepare( array $params );
 871
 872         /**
 873          * Take measures to block web access to a storage directory and
 874          * the container it belongs to. FS backends might add .htaccess
 875          * files whereas key/value store backends might revoke container
 876          * access to the storage user representing end-users in web requests.
 877          *
 878          * This is not guaranteed to actually make files or listings publicly hidden.
 879          * Additional server configuration may be needed to achieve the desired effect.
 880          *
 881          * @param array $params Parameters include:
 882          *   - dir            : storage directory
 883          *   - noAccess       : try to deny file access
 884          *   - noListing      : try to deny file listing
 885          *   - bypassReadOnly : allow writes in read-only mode (since 1.20)
 886          * @return StatusValue
 887          */
 888         final public function secure( array $params ) {
 889                 if ( empty( $params['bypassReadOnly'] ) && $this->isReadOnly() ) {
 890                         return $this->newStatus( 'backend-fail-readonly', $this->name, $this->readOnly );
 891                 }
 892                 /** @noinspection PhpUnusedLocalVariableInspection */
 893                 $scope = ScopedCallback::newScopedIgnoreUserAbort(); // try to ignore client aborts
 894                 return $this->doSecure( $params );
 895         }
 896
 897         /**
 898          * @see FileBackend::secure()
 899          * @param array $params
 900          * @return StatusValue
 901          */
 902         abstract protected function doSecure( array $params );
 903
 904         /**
 905          * Remove measures to block web access to a storage directory and
 906          * the container it belongs to. FS backends might remove .htaccess
 907          * files whereas key/value store backends might grant container
 908          * access to the storage user representing end-users in web requests.
 909          * This essentially can undo the result of secure() calls.
 910          *
 911          * This is not guaranteed to actually make files or listings publicly viewable.
 912          * Additional server configuration may be needed to achieve the desired effect.
 913          *
 914          * @param array $params Parameters include:
 915          *   - dir            : storage directory
 916          *   - access         : try to allow file access
 917          *   - listing        : try to allow file listing
 918          *   - bypassReadOnly : allow writes in read-only mode (since 1.20)
 919          * @return StatusValue
 920          * @since 1.20
 921          */
 922         final public function publish( array $params ) {
 923                 if ( empty( $params['bypassReadOnly'] ) && $this->isReadOnly() ) {
 924                         return $this->newStatus( 'backend-fail-readonly', $this->name, $this->readOnly );
 925                 }
 926                 /** @noinspection PhpUnusedLocalVariableInspection */
 927                 $scope = ScopedCallback::newScopedIgnoreUserAbort(); // try to ignore client aborts
 928                 return $this->doPublish( $params );
 929         }
 930
 931         /**
 932          * @see FileBackend::publish()
 933          * @param array $params
 934          * @return StatusValue
 935          */
 936         abstract protected function doPublish( array $params );
 937
 938         /**
 939          * Delete a storage directory if it is empty.
 940          * Backends using key/value stores may do nothing unless the directory
 941          * is that of an empty container, in which case it will be deleted.
 942          *
 943          * @param array $params Parameters include:
 944          *   - dir            : storage directory
 945          *   - recursive      : recursively delete empty subdirectories first (since 1.20)
 946          *   - bypassReadOnly : allow writes in read-only mode (since 1.20)
 947          * @return StatusValue
 948          */
 949         final public function clean( array $params ) {
 950                 if ( empty( $params['bypassReadOnly'] ) && $this->isReadOnly() ) {
 951                         return $this->newStatus( 'backend-fail-readonly', $this->name, $this->readOnly );
 952                 }
 953                 /** @noinspection PhpUnusedLocalVariableInspection */
 954                 $scope = ScopedCallback::newScopedIgnoreUserAbort(); // try to ignore client aborts
 955                 return $this->doClean( $params );
 956         }
 957
 958         /**
 959          * @see FileBackend::clean()
 960          * @param array $params
 961          * @return StatusValue
 962          */
 963         abstract protected function doClean( array $params );
 964
 965         /**
 966          * Check if a file exists at a storage path in the backend.
 967          * This returns false if only a directory exists at the path.
 968          *
 969          * Callers that only care if a file is readily accessible can use non-strict
 970          * comparisons on the result. If "does not exist" and "existence is unknown"
 971          * must be distinguished, then strict comparisons to true/null should be used.
 972          *
 973          * @see FileBackend::EXISTENCE_ERROR
 974          * @see FileBackend::directoryExists()
 975          *
 976          * @param array $params Parameters include:
 977          *   - src    : source storage path
 978          *   - latest : use the latest available data
 979          * @return bool|null Whether the file exists or null (I/O error)
 980          */
 981         abstract public function fileExists( array $params );
 982
 983         /**
 984          * Get the last-modified timestamp of the file at a storage path.
 985          *
 986          * @see FileBackend::TIMESTAMP_FAIL
 987          *
 988          * @param array $params Parameters include:
 989          *   - src    : source storage path
 990          *   - latest : use the latest available data
 991          * @return string|false TS_MW timestamp or false (missing file or I/O error)
 992          */
 993         abstract public function getFileTimestamp( array $params );
 994
 995         /**
 996          * Get the contents of a file at a storage path in the backend.
 997          * This should be avoided for potentially large files.
 998          *
 999          * @see FileBackend::CONTENT_FAIL
1000          *
1001          * @param array $params Parameters include:
1002          *   - src    : source storage path
1003          *   - latest : use the latest available data
1004          * @return string|false Content string or false (missing file or I/O error)
1005          */
1006         final public function getFileContents( array $params ) {
1007                 $contents = $this->getFileContentsMulti( [ 'srcs' => [ $params['src'] ] ] + $params );
1008
1009                 return $contents[$params['src']];
1010         }
1011
1012         /**
1013          * Like getFileContents() except it takes an array of storage paths
1014          * and returns an order preserved map of storage paths to their content.
1015          *
1016          * @see FileBackend::getFileContents()
1017          *
1018          * @param array $params Parameters include:
1019          *   - srcs        : list of source storage paths
1020          *   - latest      : use the latest available data
1021          *   - parallelize : try to do operations in parallel when possible
1022          * @return string[]|false[] Map of (path name => file content or false on failure)
1023          * @since 1.20
1024          */
1025         abstract public function getFileContentsMulti( array $params );
1026
1027         /**
1028          * Get metadata about a file at a storage path in the backend.
1029          * If the file does not exist, then this returns false.
1030          * Otherwise, the result is an associative array that includes:
1031          *   - headers  : map of HTTP headers used for GET/HEAD requests (name => value)
1032          *   - metadata : map of file metadata (name => value)
1033          * Metadata keys and headers names will be returned in all lower-case.
1034          * Additional values may be included for internal use only.
1035          *
1036          * Use FileBackend::hasFeatures() to check how well this is supported.
1037          *
1038          * @see FileBackend::XATTRS_FAIL
1039          *
1040          * @param array $params
1041          * $params include:
1042          *   - src    : source storage path
1043          *   - latest : use the latest available data
1044          * @return array|false File metadata array or false (missing file or I/O error)
1045          * @since 1.23
1046          */
1047         abstract public function getFileXAttributes( array $params );
1048
1049         /**
1050          * Get the size (bytes) of a file at a storage path in the backend.
1051          *
1052          * @see FileBackend::SIZE_FAIL
1053          *
1054          * @param array $params Parameters include:
1055          *   - src    : source storage path
1056          *   - latest : use the latest available data
1057          * @return int|false File size in bytes or false (missing file or I/O error)
1058          */
1059         abstract public function getFileSize( array $params );
1060
1061         /**
1062          * Get quick information about a file at a storage path in the backend.
1063          * If the file does not exist, then this returns false.
1064          * Otherwise, the result is an associative array that includes:
1065          *   - mtime  : the last-modified timestamp (TS_MW)
1066          *   - size   : the file size (bytes)
1067          * Additional values may be included for internal use only.
1068          *
1069          * @see FileBackend::STAT_ABSENT
1070          * @see FileBackend::STAT_ERROR
1071          *
1072          * @param array $params Parameters include:
1073          *   - src    : source storage path
1074          *   - latest : use the latest available data
1075          * @return array|false|null Attribute map, false (missing file), or null (I/O error)
1076          */
1077         abstract public function getFileStat( array $params );
1078
1079         /**
1080          * Get a SHA-1 hash of the content of the file at a storage path in the backend.
1081          *
1082          * @see FileBackend::SHA1_FAIL
1083          *
1084          * @param array $params Parameters include:
1085          *   - src    : source storage path
1086          *   - latest : use the latest available data
1087          * @return string|false Hash string or false (missing file or I/O error)
1088          */
1089         abstract public function getFileSha1Base36( array $params );
1090
1091         /**
1092          * Get the properties of the content of the file at a storage path in the backend.
1093          * This gives the result of FSFile::getProps() on a local copy of the file.
1094          *
1095          * @param array $params Parameters include:
1096          *   - src    : source storage path
1097          *   - latest : use the latest available data
1098          * @return array Properties map; FSFile::placeholderProps() if file missing or on I/O error
1099          */
1100         abstract public function getFileProps( array $params );
1101
1102         /**
1103          * Stream the content of the file at a storage path in the backend.
1104          *
1105          * If the file does not exists, an HTTP 404 error will be given.
1106          * Appropriate HTTP headers (Status, Content-Type, Content-Length)
1107          * will be sent if streaming began, while none will be sent otherwise.
1108          * Implementations should flush the output buffer before sending data.
1109          *
1110          * @param array $params Parameters include:
1111          *   - src      : source storage path
1112          *   - headers  : list of additional HTTP headers to send if the file exists
1113          *   - options  : HTTP request header map with lower case keys (since 1.28). Supports:
1114          *                range             : format is "bytes=(\d*-\d*)"
1115          *                if-modified-since : format is an HTTP date
1116          *   - headless : only include the body (and headers from "headers") (since 1.28)
1117          *   - latest   : use the latest available data
1118          *   - allowOB  : preserve any output buffers (since 1.28)
1119          * @return StatusValue
1120          */
1121         abstract public function streamFile( array $params );
1122
1123         /**
1124          * Returns a file system file, identical in content to the file at a storage path.
1125          * The file returned is either:
1126          *   - a) A TempFSFile local copy of the file at a storage path in the backend.
1127          *        The temporary copy will have the same extension as the source.
1128          *        Temporary files may be purged when the file object falls out of scope.
1129          *   - b) An FSFile pointing to the original file at a storage path in the backend.
1130          *        This is applicable for backends layered directly on top of file systems.
1131          *
1132          * Never modify the returned file since it might be the original, it might be shared
1133          * among multiple callers of this method, or the backend might internally keep FSFile
1134          * references for deferred operations.
1135          *
1136          * @param array $params Parameters include:
1137          *   - src    : source storage path
1138          *   - latest : use the latest available data
1139          * @return FSFile|null Local file copy or null (missing file or I/O error)
1140          */
1141         final public function getLocalReference( array $params ) {
1142                 $fsFiles = $this->getLocalReferenceMulti( [ 'srcs' => [ $params['src'] ] ] + $params );
1143
1144                 return $fsFiles[$params['src']];
1145         }
1146
1147         /**
1148          * Like getLocalReference() except it takes an array of storage paths and
1149          * yields an order-preserved map of storage paths to temporary local file copies.
1150          *
1151          * Never modify the returned files since they might be originals, they might be shared
1152          * among multiple callers of this method, or the backend might internally keep FSFile
1153          * references for deferred operations.
1154          *
1155          * @see FileBackend::getLocalReference()
1156          *
1157          * @param array $params Parameters include:
1158          *   - srcs        : list of source storage paths
1159          *   - latest      : use the latest available data
1160          *   - parallelize : try to do operations in parallel when possible
1161          * @return array Map of (path name => FSFile or null on failure)
1162          * @since 1.20
1163          */
1164         abstract public function getLocalReferenceMulti( array $params );
1165
1166         /**
1167          * Get a local copy on disk of the file at a storage path in the backend.
1168          * The temporary copy will have the same file extension as the source.
1169          * Temporary files may be purged when the file object falls out of scope.
1170          *
1171          * Multiple calls to this method for the same path will create new copies.
1172          *
1173          * @param array $params Parameters include:
1174          *   - src    : source storage path
1175          *   - latest : use the latest available data
1176          * @return TempFSFile|null Temporary local file copy or null (missing file or I/O error)
1177          */
1178         final public function getLocalCopy( array $params ) {
1179                 $tmpFiles = $this->getLocalCopyMulti( [ 'srcs' => [ $params['src'] ] ] + $params );
1180
1181                 return $tmpFiles[$params['src']];
1182         }
1183
1184         /**
1185          * Like getLocalCopy() except it takes an array of storage paths and yields
1186          * an order preserved-map of storage paths to temporary local file copies.
1187          *
1188          * Multiple calls to this method for the same path will create new copies.
1189          *
1190          * @see FileBackend::getLocalCopy()
1191          *
1192          * @param array $params Parameters include:
1193          *   - srcs        : list of source storage paths
1194          *   - latest      : use the latest available data
1195          *   - parallelize : try to do operations in parallel when possible
1196          * @return array Map of (path name => TempFSFile or null on failure)
1197          * @since 1.20
1198          */
1199         abstract public function getLocalCopyMulti( array $params );
1200
1201         /**
1202          * Return an HTTP URL to a given file that requires no authentication to use.
1203          * The URL may be pre-authenticated (via some token in the URL) and temporary.
1204          * This will return null if the backend cannot make an HTTP URL for the file.
1205          *
1206          * This is useful for key/value stores when using scripts that seek around
1207          * large files and those scripts (and the backend) support HTTP Range headers.
1208          * Otherwise, one would need to use getLocalReference(), which involves loading
1209          * the entire file on to local disk.
1210          *
1211          * @see FileBackend::TEMPURL_ERROR
1212          *
1213          * @param array $params Parameters include:
1214          *   - src : source storage path
1215          *   - ttl : lifetime (seconds) if pre-authenticated; default is 1 day
1216          * @return string|null URL or null (not supported or I/O error)
1217          * @since 1.21
1218          */
1219         abstract public function getFileHttpUrl( array $params );
1220
1221         /**
1222          * Check if a directory exists at a given storage path
1223          *
1224          * For backends using key/value stores, a directory is said to exist whenever
1225          * there exist any files with paths using the given directory path as a prefix
1226          * followed by a forward slash. For example, if there is a file called
1227          * "mwstore://backend/container/dir/path.svg" then directories are said to exist
1228          * at "mwstore://backend/container" and "mwstore://backend/container/dir". These
1229          * can be thought of as "virtual" directories.
1230          *
1231          * Backends that directly use a filesystem layer might enumerate empty directories.
1232          * The clean() method should always be used when files are deleted or moved if this
1233          * is a concern. This is a trade-off to avoid write amplication/contention on file
1234          * changes or read amplification when calling this method.
1235          *
1236          * Storage backends with eventual consistency might return stale data.
1237          *
1238          * @see FileBackend::EXISTENCE_ERROR
1239          * @see FileBackend::clean()
1240          *
1241          * @param array $params Parameters include:
1242          *   - dir : storage directory
1243          * @return bool|null Whether a directory exists or null (I/O error)
1244          * @since 1.20
1245          */
1246         abstract public function directoryExists( array $params );
1247
1248         /**
1249          * Get an iterator to list *all* directories under a storage directory
1250          *
1251          * If the directory is of the form "mwstore://backend/container",
1252          * then all directories in the container will be listed.
1253          * If the directory is of form "mwstore://backend/container/dir",
1254          * then all directories directly under that directory will be listed.
1255          * Results will be storage directories relative to the given directory.
1256          *
1257          * Storage backends with eventual consistency might return stale data.
1258          *
1259          * Failures during iteration can result in FileBackendError exceptions (since 1.22).
1260          *
1261          * @see FileBackend::LIST_ERROR
1262          * @see FileBackend::directoryExists()
1263          *
1264          * @param array $params Parameters include:
1265          *   - dir     : storage directory
1266          *   - topOnly : only return direct child dirs of the directory
1267          * @return Traversable|array|null Directory list enumerator or null (initial I/O error)
1268          * @since 1.20
1269          */
1270         abstract public function getDirectoryList( array $params );
1271
1272         /**
1273          * Same as FileBackend::getDirectoryList() except only lists
1274          * directories that are immediately under the given directory.
1275          *
1276          * Storage backends with eventual consistency might return stale data.
1277          *
1278          * Failures during iteration can result in FileBackendError exceptions (since 1.22).
1279          *
1280          * @see FileBackend::LIST_ERROR
1281          * @see FileBackend::directoryExists()
1282          *
1283          * @param array $params Parameters include:
1284          *   - dir : storage directory
1285          * @return Traversable|array|null Directory list enumerator or null (initial I/O error)
1286          * @since 1.20
1287          */
1288         final public function getTopDirectoryList( array $params ) {
1289                 return $this->getDirectoryList( [ 'topOnly' => true ] + $params );
1290         }
1291
1292         /**
1293          * Get an iterator to list *all* stored files under a storage directory
1294          *
1295          * If the directory is of the form "mwstore://backend/container", then all
1296          * files in the container will be listed. If the directory is of form
1297          * "mwstore://backend/container/dir", then all files under that directory will
1298          * be listed. Results will be storage paths relative to the given directory.
1299          *
1300          * Storage backends with eventual consistency might return stale data.
1301          *
1302          * Failures during iteration can result in FileBackendError exceptions (since 1.22).
1303          *
1304          * @see FileBackend::LIST_ERROR
1305          *
1306          * @param array $params Parameters include:
1307          *   - dir        : storage directory
1308          *   - topOnly    : only return direct child files of the directory (since 1.20)
1309          *   - adviseStat : set to true if stat requests will be made on the files (since 1.22)
1310          * @return Traversable|array|null File list enumerator or null (initial I/O error)
1311          */
1312         abstract public function getFileList( array $params );
1313
1314         /**
1315          * Same as FileBackend::getFileList() except only lists
1316          * files that are immediately under the given directory.
1317          *
1318          * Storage backends with eventual consistency might return stale data.
1319          *
1320          * Failures during iteration can result in FileBackendError exceptions (since 1.22).
1321          *
1322          * @see FileBackend::LIST_ERROR
1323          *
1324          * @param array $params Parameters include:
1325          *   - dir        : storage directory
1326          *   - adviseStat : set to true if stat requests will be made on the files (since 1.22)
1327          * @return Traversable|array|null File list enumerator or null on failure
1328          * @since 1.20
1329          */
1330         final public function getTopFileList( array $params ) {
1331                 return $this->getFileList( [ 'topOnly' => true ] + $params );
1332         }
1333
1334         /**
1335          * Preload persistent file stat cache and property cache into in-process cache.
1336          * This should be used when stat calls will be made on a known list of a many files.
1337          *
1338          * @see FileBackend::getFileStat()
1339          *
1340          * @param array $paths Storage paths
1341          */
1342         abstract public function preloadCache( array $paths );
1343
1344         /**
1345          * Invalidate any in-process file stat and property cache.
1346          * If $paths is given, then only the cache for those files will be cleared.
1347          *
1348          * @see FileBackend::getFileStat()
1349          *
1350          * @param array|null $paths Storage paths (optional)
1351          */
1352         abstract public function clearCache( array $paths = null );
1353
1354         /**
1355          * Preload file stat information (concurrently if possible) into in-process cache.
1356          *
1357          * This should be used when stat calls will be made on a known list of a many files.
1358          * This does not make use of the persistent file stat cache.
1359          *
1360          * @see FileBackend::getFileStat()
1361          *
1362          * @param array $params Parameters include:
1363          *   - srcs        : list of source storage paths
1364          *   - latest      : use the latest available data
1365          * @return bool Whether all requests proceeded without I/O errors (since 1.24)
1366          * @since 1.23
1367          */
1368         abstract public function preloadFileStat( array $params );
1369
1370         /**
1371          * Lock the files at the given storage paths in the backend.
1372          * This will either lock all the files or none (on failure).
1373          *
1374          * Callers should consider using getScopedFileLocks() instead.
1375          *
1376          * @param array $paths Storage paths
1377          * @param int $type LockManager::LOCK_* constant
1378          * @param int $timeout Timeout in seconds (0 means non-blocking) (since 1.24)
1379          * @return StatusValue
1380          */
1381         final public function lockFiles( array $paths, $type, $timeout = 0 ) {
1382                 $paths = array_map( 'FileBackend::normalizeStoragePath', $paths );
1383
1384                 return $this->wrapStatus( $this->lockManager->lock( $paths, $type, $timeout ) );
1385         }
1386
1387         /**
1388          * Unlock the files at the given storage paths in the backend.
1389          *
1390          * @param array $paths Storage paths
1391          * @param int $type LockManager::LOCK_* constant
1392          * @return StatusValue
1393          */
1394         final public function unlockFiles( array $paths, $type ) {
1395                 $paths = array_map( 'FileBackend::normalizeStoragePath', $paths );
1396
1397                 return $this->wrapStatus( $this->lockManager->unlock( $paths, $type ) );
1398         }
1399
1400         /**
1401          * Lock the files at the given storage paths in the backend.
1402          * This will either lock all the files or none (on failure).
1403          * On failure, the StatusValue object will be updated with errors.
1404          *
1405          * Once the return value goes out scope, the locks will be released and
1406          * the StatusValue updated. Unlock fatals will not change the StatusValue "OK" value.
1407          *
1408          * @see ScopedLock::factory()
1409          *
1410          * @param array $paths List of storage paths or map of lock types to path lists
1411          * @param int|string $type LockManager::LOCK_* constant or "mixed"
1412          * @param StatusValue $status StatusValue to update on lock/unlock
1413          * @param int $timeout Timeout in seconds (0 means non-blocking) (since 1.24)
1414          * @return ScopedLock|null RAII-style self-unlocking lock or null on failure
1415          */
1416         final public function getScopedFileLocks(
1417                 array $paths, $type, StatusValue $status, $timeout = 0
1418         ) {
1419                 if ( $type === 'mixed' ) {
1420                         foreach ( $paths as &$typePaths ) {
1421                                 $typePaths = array_map( 'FileBackend::normalizeStoragePath', $typePaths );
1422                         }
1423                 } else {
1424                         $paths = array_map( 'FileBackend::normalizeStoragePath', $paths );
1425                 }
1426
1427                 return ScopedLock::factory( $this->lockManager, $paths, $type, $status, $timeout );
1428         }
1429
1430         /**
1431          * Get an array of scoped locks needed for a batch of file operations.
1432          *
1433          * Normally, FileBackend::doOperations() handles locking, unless
1434          * the 'nonLocking' param is passed in. This function is useful if you
1435          * want the files to be locked for a broader scope than just when the
1436          * files are changing. For example, if you need to update DB metadata,
1437          * you may want to keep the files locked until finished.
1438          *
1439          * @see FileBackend::doOperations()
1440          *
1441          * @param array $ops List of file operations to FileBackend::doOperations()
1442          * @param StatusValue $status StatusValue to update on lock/unlock
1443          * @return ScopedLock|null RAII-style self-unlocking lock or null on failure
1444          * @since 1.20
1445          */
1446         abstract public function getScopedLocksForOps( array $ops, StatusValue $status );
1447
1448         /**
1449          * Get the root storage path of this backend.
1450          * All container paths are "subdirectories" of this path.
1451          *
1452          * @return string Storage path
1453          * @since 1.20
1454          */
1455         final public function getRootStoragePath() {
1456                 return "mwstore://{$this->name}";
1457         }
1458
1459         /**
1460          * Get the storage path for the given container for this backend
1461          *
1462          * @param string $container Container name
1463          * @return string Storage path
1464          * @since 1.21
1465          */
1466         final public function getContainerStoragePath( $container ) {
1467                 return $this->getRootStoragePath() . "/{$container}";
1468         }
1469
1470         /**
1471          * Get the file journal object for this backend
1472          *
1473          * @return FileJournal
1474          */
1475         final public function getJournal() {
1476                 return $this->fileJournal;
1477         }
1478
1479         /**
1480          * Convert FSFile 'src' paths to string paths (with an 'srcRef' field set to the FSFile)
1481          *
1482          * The 'srcRef' field keeps any TempFSFile objects in scope for the backend to have it
1483          * around as long it needs (which may vary greatly depending on configuration)
1484          *
1485          * @param array $ops File operation batch for FileBaclend::doOperations()
1486          * @return array File operation batch
1487          */
1488         protected function resolveFSFileObjects( array $ops ) {
1489                 foreach ( $ops as &$op ) {
1490                         $src = $op['src'] ?? null;
1491                         if ( $src instanceof FSFile ) {
1492                                 $op['srcRef'] = $src;
1493                                 $op['src'] = $src->getPath();
1494                         }
1495                 }
1496                 unset( $op );
1497
1498                 return $ops;
1499         }
1500
1501         /**
1502          * Check if a given path is a "mwstore://" path.
1503          * This does not do any further validation or any existence checks.
1504          *
1505          * @param string $path
1506          * @return bool
1507          */
1508         final public static function isStoragePath( $path ) {
1509                 return ( strpos( $path, 'mwstore://' ) === 0 );
1510         }
1511
1512         /**
1513          * Split a storage path into a backend name, a container name,
1514          * and a relative file path. The relative path may be the empty string.
1515          * This does not do any path normalization or traversal checks.
1516          *
1517          * @param string $storagePath
1518          * @return array (backend, container, rel object) or (null, null, null)
1519          */
1520         final public static function splitStoragePath( $storagePath ) {
1521                 if ( self::isStoragePath( $storagePath ) ) {
1522                         // Remove the "mwstore://" prefix and split the path
1523                         $parts = explode( '/', substr( $storagePath, 10 ), 3 );
1524                         if ( count( $parts ) >= 2 && $parts[0] != '' && $parts[1] != '' ) {
1525                                 if ( count( $parts ) == 3 ) {
1526                                         return $parts; // e.g. "backend/container/path"
1527                                 } else {
1528                                         return [ $parts[0], $parts[1], '' ]; // e.g. "backend/container"
1529                                 }
1530                         }
1531                 }
1532
1533                 return [ null, null, null ];
1534         }
1535
1536         /**
1537          * Normalize a storage path by cleaning up directory separators.
1538          * Returns null if the path is not of the format of a valid storage path.
1539          *
1540          * @param string $storagePath
1541          * @return string|null Normalized storage path or null on failure
1542          */
1543         final public static function normalizeStoragePath( $storagePath ) {
1544                 list( $backend, $container, $relPath ) = self::splitStoragePath( $storagePath );
1545                 if ( $relPath !== null ) { // must be for this backend
1546                         $relPath = self::normalizeContainerPath( $relPath );
1547                         if ( $relPath !== null ) {
1548                                 return ( $relPath != '' )
1549                                         ? "mwstore://{$backend}/{$container}/{$relPath}"
1550                                         : "mwstore://{$backend}/{$container}";
1551                         }
1552                 }
1553
1554                 return null;
1555         }
1556
1557         /**
1558          * Get the parent storage directory of a storage path.
1559          * This returns a path like "mwstore://backend/container",
1560          * "mwstore://backend/container/...", or null if there is no parent.
1561          *
1562          * @param string $storagePath
1563          * @return string|null Parent storage path or null on failure
1564          */
1565         final public static function parentStoragePath( $storagePath ) {
1566                 $storagePath = dirname( $storagePath );
1567                 list( , , $rel ) = self::splitStoragePath( $storagePath );
1568
1569                 return ( $rel === null ) ? null : $storagePath;
1570         }
1571
1572         /**
1573          * Get the final extension from a storage or FS path
1574          *
1575          * @param string $path
1576          * @param string $case One of (rawcase, uppercase, lowercase) (since 1.24)
1577          * @return string
1578          */
1579         final public static function extensionFromPath( $path, $case = 'lowercase' ) {
1580                 $i = strrpos( $path, '.' );
1581                 $ext = $i ? substr( $path, $i + 1 ) : '';
1582
1583                 if ( $case === 'lowercase' ) {
1584                         $ext = strtolower( $ext );
1585                 } elseif ( $case === 'uppercase' ) {
1586                         $ext = strtoupper( $ext );
1587                 }
1588
1589                 return $ext;
1590         }
1591
1592         /**
1593          * Check if a relative path has no directory traversals
1594          *
1595          * @param string $path
1596          * @return bool
1597          * @since 1.20
1598          */
1599         final public static function isPathTraversalFree( $path ) {
1600                 return ( self::normalizeContainerPath( $path ) !== null );
1601         }
1602
1603         /**
1604          * Build a Content-Disposition header value per RFC 6266.
1605          *
1606          * @param string $type One of (attachment, inline)
1607          * @param string $filename Suggested file name (should not contain slashes)
1608          * @throws InvalidArgumentException
1609          * @return string
1610          * @since 1.20
1611          */
1612         final public static function makeContentDisposition( $type, $filename = '' ) {
1613                 $parts = [];
1614
1615                 $type = strtolower( $type );
1616                 if ( !in_array( $type, [ 'inline', 'attachment' ] ) ) {
1617                         throw new InvalidArgumentException( "Invalid Content-Disposition type '$type'." );
1618                 }
1619                 $parts[] = $type;
1620
1621                 if ( strlen( $filename ) ) {
1622                         $parts[] = "filename*=UTF-8''" . rawurlencode( basename( $filename ) );
1623                 }
1624
1625                 return implode( ';', $parts );
1626         }
1627
1628         /**
1629          * Validate and normalize a relative storage path.
1630          * Null is returned if the path involves directory traversal.
1631          * Traversal is insecure for FS backends and broken for others.
1632          *
1633          * This uses the same traversal protection as Title::secureAndSplit().
1634          *
1635          * @param string $path Storage path relative to a container
1636          * @return string|null Normalized container path or null on failure
1637          */
1638         final protected static function normalizeContainerPath( $path ) {
1639                 // Normalize directory separators
1640                 $path = strtr( $path, '\\', '/' );
1641                 // Collapse any consecutive directory separators
1642                 $path = preg_replace( '![/]{2,}!', '/', $path );
1643                 // Remove any leading directory separator
1644                 $path = ltrim( $path, '/' );
1645                 // Use the same traversal protection as Title::secureAndSplit()
1646                 if ( strpos( $path, '.' ) !== false ) {
1647                         if (
1648                                 $path === '.' ||
1649                                 $path === '..' ||
1650                                 strpos( $path, './' ) === 0 ||
1651                                 strpos( $path, '../' ) === 0 ||
1652                                 strpos( $path, '/./' ) !== false ||
1653                                 strpos( $path, '/../' ) !== false
1654                         ) {
1655                                 return null;
1656                         }
1657                 }
1658
1659                 return $path;
1660         }
1661
1662         /**
1663          * Yields the result of the status wrapper callback on either:
1664          *   - StatusValue::newGood() if this method is called without parameters
1665          *   - StatusValue::newFatal() with all parameters to this method if passed in
1666          *
1667          * @param string ...$args
1668          * @return StatusValue
1669          */
1670         final protected function newStatus( ...$args ) {
1671                 if ( count( $args ) ) {
1672                         $sv = StatusValue::newFatal( ...$args );
1673                 } else {
1674                         $sv = StatusValue::newGood();
1675                 }
1676
1677                 return $this->wrapStatus( $sv );
1678         }
1679
1680         /**
1681          * @param StatusValue $sv
1682          * @return StatusValue Modified status or StatusValue subclass
1683          */
1684         final protected function wrapStatus( StatusValue $sv ) {
1685                 return $this->statusWrapper ? call_user_func( $this->statusWrapper, $sv ) : $sv;
1686         }
1687
1688         /**
1689          * @param string $section
1690          * @return ScopedCallback|null
1691          */
1692         protected function scopedProfileSection( $section ) {
1693                 return $this->profiler ? ( $this->profiler )( $section ) : null;
1694         }
1695
1696         protected function resetOutputBuffer() {
1697                 while ( ob_get_status() ) {
1698                         if ( !ob_end_clean() ) {
1699                                 // Could not remove output buffer handler; abort now
1700                                 // to avoid getting in some kind of infinite loop.
1701                                 break;
1702                         }
1703                 }
1704         }
1705 }