includes/upload/UploadStash.php

   1 <?php
   2 /**
   3  * UploadStash is intended to accomplish a few things:
   4  *   - enable applications to temporarily stash files without publishing them to the wiki.
   5  *      - Several parts of MediaWiki do this in similar ways: UploadBase, UploadWizard, and FirefoggChunkedExtension
   6  *        And there are several that reimplement stashing from scratch, in idiosyncratic ways. The idea is to unify them all here.
   7  *        Mostly all of them are the same except for storing some custom fields, which we subsume into the data array.
   8  *   - enable applications to find said files later, as long as the session or temp files haven't been purged.
   9  *   - enable the uploading user (and *ONLY* the uploading user) to access said files, and thumbnails of said files, via a URL.
  10  *     We accomplish this by making the session serve as a URL->file mapping, on the assumption that nobody else can access
  11  *     the session, even the uploading user. See SpecialUploadStash, which implements a web interface to some files stored this way.
  12  *
  13  */
  14 class UploadStash {
  15
  16         // Format of the key for files -- has to be suitable as a filename itself (e.g. ab12cd34ef.jpg)
  17         const KEY_FORMAT_REGEX = '/^[\w-]+\.\w+$/';
  18
  19         /**
  20          * repository that this uses to store temp files
  21          * public because we sometimes need to get a LocalFile within the same repo.
  22          *
  23          * @var LocalRepo
  24          */
  25         public $repo;
  26
  27         // array of initialized objects obtained from session (lazily initialized upon getFile())
  28         private $files = array();
  29
  30         // TODO: Once UploadBase starts using this, switch to use these constants rather than UploadBase::SESSION*
  31         // const SESSION_VERSION = 2;
  32         // const SESSION_KEYNAME = 'wsUploadData';
  33
  34         /**
  35          * Represents the session which contains temporarily stored files.
  36          * Designed to be compatible with the session stashing code in UploadBase (should replace it eventually)
  37          */
  38         public function __construct( $repo ) {
  39
  40                 // this might change based on wiki's configuration.
  41                 $this->repo = $repo;
  42
  43                 if ( ! isset( $_SESSION ) ) {
  44                         throw new UploadStashNotAvailableException( 'no session variable' );
  45                 }
  46
  47                 if ( !isset( $_SESSION[UploadBase::SESSION_KEYNAME] ) ) {
  48                         $_SESSION[UploadBase::SESSION_KEYNAME] = array();
  49                 }
  50
  51         }
  52
  53         /**
  54          * Get a file and its metadata from the stash.
  55          * May throw exception if session data cannot be parsed due to schema change, or key not found.
  56          *
  57          * @param $key Integer: key
  58          * @throws UploadStashFileNotFoundException
  59          * @throws UploadStashBadVersionException
  60          * @return UploadStashFile
  61          */
  62         public function getFile( $key ) {
  63                 if ( ! preg_match( self::KEY_FORMAT_REGEX, $key ) ) {
  64                         throw new UploadStashBadPathException( "key '$key' is not in a proper format" );
  65                 }
  66
  67                 if ( !isset( $this->files[$key] ) ) {
  68                         if ( !isset( $_SESSION[UploadBase::SESSION_KEYNAME][$key] ) ) {
  69                                 throw new UploadStashFileNotFoundException( "key '$key' not found in stash" );
  70                         }
  71
  72                         $data = $_SESSION[UploadBase::SESSION_KEYNAME][$key];
  73                         // guards against PHP class changing while session data doesn't
  74                         if ($data['version'] !== UploadBase::SESSION_VERSION ) {
  75                                 throw new UploadStashBadVersionException( $data['version'] . " does not match current version " . UploadBase::SESSION_VERSION );
  76                         }
  77
  78                         // separate the stashData into the path, and then the rest of the data
  79                         $path = $data['mTempPath'];
  80                         unset( $data['mTempPath'] );
  81
  82                         $file = new UploadStashFile( $this, $this->repo, $path, $key, $data );
  83                         if ( $file->getSize() === 0 ) {
  84                                 throw new UploadStashZeroLengthFileException( "File is zero length" );
  85                         }
  86                         $this->files[$key] = $file;
  87
  88                 }
  89                 return $this->files[$key];
  90         }
  91
  92         /**
  93          * Stash a file in a temp directory and record that we did this in the session, along with other metadata.
  94          * We store data in a flat key-val namespace because that's how UploadBase did it. This also means we have to
  95          * ensure that the key-val pairs in $data do not overwrite other required fields.
  96          *
  97          * @param $path String: path to file you want stashed
  98          * @param $data Array: optional, other data you want associated with the file. Do not use 'mTempPath', 'mFileProps', 'mFileSize', or 'version' as keys here
  99          * @param $key String: optional, unique key for this file in this session. Used for directory hashing when storing, otherwise not important
 100          * @throws UploadStashBadPathException
 101          * @throws UploadStashFileException
 102          * @return UploadStashFile: file, or null on failure
 103          */
 104         public function stashFile( $path, $data = array(), $key = null ) {
 105                 if ( ! file_exists( $path ) ) {
 106                         wfDebug( "UploadStash: tried to stash file at '$path', but it doesn't exist\n" );
 107                         throw new UploadStashBadPathException( "path doesn't exist" );
 108                 }
 109                 $fileProps = File::getPropsFromPath( $path );
 110
 111                 // we will be initializing from some tmpnam files that don't have extensions.
 112                 // most of MediaWiki assumes all uploaded files have good extensions. So, we fix this.
 113                 $extension = self::getExtensionForPath( $path );
 114                 if ( ! preg_match( "/\\.\\Q$extension\\E$/", $path ) ) {
 115                         $pathWithGoodExtension = "$path.$extension";
 116                         if ( ! rename( $path, $pathWithGoodExtension ) ) {
 117                                 throw new UploadStashFileException( "couldn't rename $path to have a better extension at $pathWithGoodExtension" );
 118                         }
 119                         $path = $pathWithGoodExtension;
 120                 }
 121
 122                 // If no key was supplied, use content hash. Also has the nice property of collapsing multiple identical files
 123                 // uploaded this session, which could happen if uploads had failed.
 124                 if ( is_null( $key ) ) {
 125                         $key = $fileProps['sha1'] . "." . $extension;
 126                 }
 127
 128                 if ( ! preg_match( self::KEY_FORMAT_REGEX, $key ) ) {
 129                         throw new UploadStashBadPathException( "key '$key' is not in a proper format" );
 130                 }
 131
 132                 // if not already in a temporary area, put it there
 133                 $status = $this->repo->storeTemp( basename( $path ), $path );
 134
 135                 if( ! $status->isOK() ) {
 136                         // It is a convention in MediaWiki to only return one error per API exception, even if multiple errors
 137                         // are available. We use reset() to pick the "first" thing that was wrong, preferring errors to warnings.
 138                         // This is a bit lame, as we may have more info in the $status and we're throwing it away, but to fix it means
 139                         // redesigning API errors significantly.
 140                         // $status->value just contains the virtual URL (if anything) which is probably useless to the caller
 141                         $error = reset( $status->getErrorsArray() );
 142                         if ( ! count( $error ) ) {
 143                                 $error = reset( $status->getWarningsArray() );
 144                                 if ( ! count( $error ) ) {
 145                                         $error = array( 'unknown', 'no error recorded' );
 146                                 }
 147                         }
 148                         throw new UploadStashFileException( "error storing file in '$path': " . implode( '; ', $error ) );
 149                 }
 150                 $stashPath = $status->value;
 151
 152                 // required info we always store. Must trump any other application info in $data
 153                 // 'mTempPath', 'mFileSize', and 'mFileProps' are arbitrary names
 154                 // chosen for compatibility with UploadBase's way of doing this.
 155                 $requiredData = array(
 156                         'mTempPath' => $stashPath,
 157                         'mFileSize' => $fileProps['size'],
 158                         'mFileProps' => $fileProps,
 159                         'version' => UploadBase::SESSION_VERSION
 160                 );
 161
 162                 // now, merge required info and extra data into the session. (The extra data changes from application to application.
 163                 // UploadWizard wants different things than say FirefoggChunkedUpload.)
 164                 wfDebug( __METHOD__ . " storing under $key\n" );
 165                 $_SESSION[UploadBase::SESSION_KEYNAME][$key] = array_merge( $data, $requiredData );
 166
 167                 return $this->getFile( $key );
 168         }
 169
 170         /**
 171          * Remove all files from the stash.
 172          * Does not clean up files in the repo, just the record of them.
 173          * @return boolean: success
 174          */
 175         public function clear() {
 176                 $_SESSION[UploadBase::SESSION_KEYNAME] = array();
 177                 return true;
 178         }
 179
 180
 181         /**
 182          * Remove a particular file from the stash.
 183          * Does not clean up file in the repo, just the record of it.
 184          * @return boolean: success
 185          */
 186         public function removeFile( $key ) {
 187                 unset ( $_SESSION[UploadBase::SESSION_KEYNAME][$key] );
 188                 return true;
 189         }
 190
 191         /**
 192          * List all files in the stash.
 193          */
 194         public function listFiles() {
 195                 return array_keys( $_SESSION[UploadBase::SESSION_KEYNAME] );
 196         }
 197
 198         /**
 199          * Find or guess extension -- ensuring that our extension matches our mime type.
 200          * Since these files are constructed from php tempnames they may not start off
 201          * with an extension.
 202          * XXX this is somewhat redundant with the checks that ApiUpload.php does with incoming
 203          * uploads versus the desired filename. Maybe we can get that passed to us...
 204          */
 205         public static function getExtensionForPath( $path ) {
 206                 // Does this have an extension?
 207                 $n = strrpos( $path, '.' );
 208                 $extension = null;
 209                 if ( $n !== false ) {
 210                         $extension = $n ? substr( $path, $n + 1 ) : '';
 211                 } else {
 212                         // If not, assume that it should be related to the mime type of the original file.
 213                         $magic = MimeMagic::singleton();
 214                         $mimeType = $magic->guessMimeType( $path );
 215                         $extensions = explode( ' ', MimeMagic::singleton()->getExtensionsForType( $mimeType ) );
 216                         if ( count( $extensions ) ) {
 217                                 $extension = $extensions[0];
 218                         }
 219                 }
 220
 221                 if ( is_null( $extension ) ) {
 222                         throw new UploadStashFileException( "extension is null" );
 223                 }
 224
 225                 return File::normalizeExtension( $extension );
 226         }
 227
 228 }
 229
 230 class UploadStashFile extends UnregisteredLocalFile {
 231         private $sessionStash;
 232         private $sessionKey;
 233         private $sessionData;
 234         private $urlName;
 235         protected $url;
 236
 237         /**
 238          * A LocalFile wrapper around a file that has been temporarily stashed, so we can do things like create thumbnails for it
 239          * Arguably UnregisteredLocalFile should be handling its own file repo but that class is a bit retarded currently
 240          *
 241          * @param $stash UploadStash: useful for obtaining config, stashing transformed files
 242          * @param $repo FSRepo: repository where we should find the path
 243          * @param $path String: path to file
 244          * @param $key String: key to store the path and any stashed data under
 245          * @param $data String: any other data we want stored with this file
 246          * @throws UploadStashBadPathException
 247          * @throws UploadStashFileNotFoundException
 248          */
 249         public function __construct( $stash, $repo, $path, $key, $data ) {
 250                 $this->sessionStash = $stash;
 251                 $this->sessionKey = $key;
 252                 $this->sessionData = $data;
 253
 254                 // resolve mwrepo:// urls
 255                 if ( $repo->isVirtualUrl( $path ) ) {
 256                         $path = $repo->resolveVirtualUrl( $path );
 257                 }
 258
 259                 // check if path appears to be sane, no parent traversals, and is in this repo's temp zone.
 260                 $repoTempPath = $repo->getZonePath( 'temp' );
 261                 if ( ( ! $repo->validateFilename( $path ) ) ||
 262                                 ( strpos( $path, $repoTempPath ) !== 0 ) ) {
 263                         wfDebug( "UploadStash: tried to construct an UploadStashFile from a file that should already exist at '$path', but path is not valid\n" );
 264                         throw new UploadStashBadPathException( 'path is not valid' );
 265                 }
 266
 267                 // check if path exists! and is a plain file.
 268                 if ( ! $repo->fileExists( $path, FileRepo::FILES_ONLY ) ) {
 269                         wfDebug( "UploadStash: tried to construct an UploadStashFile from a file that should already exist at '$path', but path is not found\n" );
 270                         throw new UploadStashFileNotFoundException( 'cannot find path, or not a plain file' );
 271                 }
 272
 273                 parent::__construct( false, $repo, $path, false );
 274
 275                 $this->name = basename( $this->path );
 276         }
 277
 278         /**
 279          * A method needed by the file transforming and scaling routines in File.php
 280          * We do not necessarily care about doing the description at this point
 281          * However, we also can't return the empty string, as the rest of MediaWiki demands this (and calls to imagemagick
 282          * convert require it to be there)
 283          *
 284          * @return String: dummy value
 285          */
 286         public function getDescriptionUrl() {
 287                 return $this->getUrl();
 288         }
 289
 290         /**
 291          * Get the path for the thumbnail (actually any transformation of this file)
 292          * The actual argument is the result of thumbName although we seem to have
 293          * buggy code elsewhere that expects a boolean 'suffix'
 294          *
 295          * @param $thumbName String: name of thumbnail (e.g. "120px-123456.jpg" ), or false to just get the path
 296          * @return String: path thumbnail should take on filesystem, or containing directory if thumbname is false
 297          */
 298         public function getThumbPath( $thumbName = false ) {
 299                 $path = dirname( $this->path );
 300                 if ( $thumbName !== false ) {
 301                         $path .= "/$thumbName";
 302                 }
 303                 return $path;
 304         }
 305
 306         /**
 307          * Return the file/url base name of a thumbnail with the specified parameters.
 308          * We override this because we want to use the pretty url name instead of the
 309          * ugly file name.
 310          *
 311          * @param $params Array: handler-specific parameters
 312          * @return String: base name for URL, like '120px-12345.jpg', or null if there is no handler
 313          */
 314         function thumbName( $params ) {
 315                 return $this->generateThumbName( $this->getUrlName(), $params );
 316         }
 317
 318         /**
 319          * Helper function -- given a 'subpage', return the local URL e.g. /wiki/Special:UploadStash/subpage
 320          * @param {String} $subPage
 321          * @return {String} local URL for this subpage in the Special:UploadStash space.
 322          */
 323         private function getSpecialUrl( $subPage ) {
 324                 return SpecialPage::getTitleFor( 'UploadStash', $subPage )->getLocalURL();
 325         }
 326
 327         /**
 328          * Get a URL to access the thumbnail
 329          * This is required because the model of how files work requires that
 330          * the thumbnail urls be predictable. However, in our model the URL is not based on the filename
 331          * (that's hidden in the session)
 332          *
 333          * @param $thumbName String: basename of thumbnail file -- however, we don't want to use the file exactly
 334          * @return String: URL to access thumbnail, or URL with partial path
 335          */
 336         public function getThumbUrl( $thumbName = false ) {
 337                 wfDebug( __METHOD__ . " getting for $thumbName \n" );
 338                 return $this->getSpecialUrl( 'thumb/' . $this->getUrlName() . '/' . $thumbName );
 339         }
 340
 341         /**
 342          * The basename for the URL, which we want to not be related to the filename.
 343          * Will also be used as the lookup key for a thumbnail file.
 344          *
 345          * @return String: base url name, like '120px-123456.jpg'
 346          */
 347         public function getUrlName() {
 348                 if ( ! $this->urlName ) {
 349                         $this->urlName = $this->sessionKey;
 350                 }
 351                 return $this->urlName;
 352         }
 353
 354         /**
 355          * Return the URL of the file, if for some reason we wanted to download it
 356          * We tend not to do this for the original file, but we do want thumb icons
 357          *
 358          * @return String: url
 359          */
 360         public function getUrl() {
 361                 if ( !isset( $this->url ) ) {
 362                         $this->url = $this->getSpecialUrl( 'file/' . $this->getUrlName() );
 363                 }
 364                 return $this->url;
 365         }
 366
 367         /**
 368          * Parent classes use this method, for no obvious reason, to return the path (relative to wiki root, I assume).
 369          * But with this class, the URL is unrelated to the path.
 370          *
 371          * @return String: url
 372          */
 373         public function getFullUrl() {
 374                 return $this->getUrl();
 375         }
 376
 377         /**
 378          * Getter for session key (the session-unique id by which this file's location & metadata is stored in the session)
 379          *
 380          * @return String: session key
 381          */
 382         public function getSessionKey() {
 383                 return $this->sessionKey;
 384         }
 385
 386         /**
 387          * Remove the associated temporary file
 388          * @return Status: success
 389          */
 390         public function remove() {
 391                 return $this->repo->freeTemp( $this->path );
 392         }
 393
 394 }
 395
 396 class UploadStashNotAvailableException extends MWException {};
 397 class UploadStashFileNotFoundException extends MWException {};
 398 class UploadStashBadPathException extends MWException {};
 399 class UploadStashBadVersionException extends MWException {};
 400 class UploadStashFileException extends MWException {};
 401 class UploadStashZeroLengthFileException extends MWException {};
 402