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         // Format of the key for files -- has to be suitable as a filename itself in some cases.
  16         // This should encompass a sha1 content hash in hex (new style), or an integer (old style),
  17         // and also thumbnails with prepended strings like "120px-".
  18         // The file extension should not be part of the key.
  19         const KEY_FORMAT_REGEX = '/^[\w-]+$/';
  20
  21         // repository that this uses to store temp files
  22         protected $repo;
  23
  24         // array of initialized objects obtained from session (lazily initialized upon getFile())
  25         private $files = array();
  26
  27         // the base URL for files in the stash
  28         private $baseUrl;
  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          * @param $repo FileRepo: optional -- repo in which to store files. Will choose LocalRepo if not supplied.
  39          */
  40         public function __construct( $repo = null ) {
  41
  42                 if ( is_null( $repo ) ) {
  43                         $repo = RepoGroup::singleton()->getLocalRepo();
  44                 }
  45
  46                 $this->repo = $repo;
  47
  48                 if ( ! isset( $_SESSION ) ) {
  49                         throw new UploadStashNotAvailableException( 'no session variable' );
  50                 }
  51
  52                 if ( !isset( $_SESSION[UploadBase::SESSION_KEYNAME] ) ) {
  53                         $_SESSION[UploadBase::SESSION_KEYNAME] = array();
  54                 }
  55
  56                 $this->baseUrl = SpecialPage::getTitleFor( 'UploadStash' )->getLocalURL();
  57         }
  58
  59         /**
  60          * Get the base of URLs by which one can access the files
  61          *
  62          * @return String: url
  63          */
  64         public function getBaseUrl() {
  65                 return $this->baseUrl;
  66         }
  67
  68         /**
  69          * Get a file and its metadata from the stash.
  70          * May throw exception if session data cannot be parsed due to schema change, or key not found.
  71          *
  72          * @param $key Integer: key
  73          * @throws UploadStashFileNotFoundException
  74          * @throws UploadStashBadVersionException
  75          * @return UploadStashFile
  76          */
  77         public function getFile( $key ) {
  78                 if ( ! preg_match( self::KEY_FORMAT_REGEX, $key ) ) {
  79                         throw new UploadStashBadPathException( "key '$key' is not in a proper format" );
  80                 }
  81
  82                 if ( !isset( $this->files[$key] ) ) {
  83                         if ( !isset( $_SESSION[UploadBase::SESSION_KEYNAME][$key] ) ) {
  84                                 throw new UploadStashFileNotFoundException( "key '$key' not found in stash" );
  85                         }
  86
  87                         $data = $_SESSION[UploadBase::SESSION_KEYNAME][$key];
  88                         // guards against PHP class changing while session data doesn't
  89                         if ($data['version'] !== UploadBase::SESSION_VERSION ) {
  90                                 throw new UploadStashBadVersionException( $data['version'] . " does not match current version " . UploadBase::SESSION_VERSION );
  91                         }
  92
  93                         // separate the stashData into the path, and then the rest of the data
  94                         $path = $data['mTempPath'];
  95                         unset( $data['mTempPath'] );
  96
  97                         $file = new UploadStashFile( $this, $this->repo, $path, $key, $data );
  98
  99                         $this->files[$key] = $file;
 100
 101                 }
 102                 return $this->files[$key];
 103         }
 104
 105         /**
 106          * Stash a file in a temp directory and record that we did this in the session, along with other metadata.
 107          * We store data in a flat key-val namespace because that's how UploadBase did it. This also means we have to
 108          * ensure that the key-val pairs in $data do not overwrite other required fields.
 109          *
 110          * @param $path String: path to file you want stashed
 111          * @param $data Array: optional, other data you want associated with the file. Do not use 'mTempPath', 'mFileProps', 'mFileSize', or 'version' as keys here
 112          * @param $key String: optional, unique key for this file in this session. Used for directory hashing when storing, otherwise not important
 113          * @throws UploadStashBadPathException
 114          * @throws UploadStashFileException
 115          * @return UploadStashFile: file, or null on failure
 116          */
 117         public function stashFile( $path, $data = array(), $key = null ) {
 118                 if ( ! file_exists( $path ) ) {
 119                         throw new UploadStashBadPathException( "path doesn't exist" );
 120                 }
 121                 $fileProps = File::getPropsFromPath( $path );
 122
 123                 // If no key was supplied, use content hash. Also has the nice property of collapsing multiple identical files
 124                 // uploaded this session, which could happen if uploads had failed.
 125                 if ( is_null( $key ) ) {
 126                         $key = $fileProps['sha1'];
 127                 }
 128
 129                 if ( ! preg_match( self::KEY_FORMAT_REGEX, $key ) ) {
 130                         throw new UploadStashBadPathException( "key '$key' is not in a proper format" );
 131                 }
 132
 133                 // if not already in a temporary area, put it there
 134                 $status = $this->repo->storeTemp( basename( $path ), $path );
 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                 $_SESSION[UploadBase::SESSION_KEYNAME][$key] = array_merge( $data, $requiredData );
 165
 166                 return $this->getFile( $key );
 167         }
 168
 169 }
 170
 171 class UploadStashFile extends UnregisteredLocalFile {
 172         private $sessionStash;
 173         private $sessionKey;
 174         private $sessionData;
 175         private $urlName;
 176
 177         /**
 178          * A LocalFile wrapper around a file that has been temporarily stashed, so we can do things like create thumbnails for it
 179          * Arguably UnregisteredLocalFile should be handling its own file repo but that class is a bit retarded currently
 180          *
 181          * @param $stash UploadStash: useful for obtaining config, stashing transformed files
 182          * @param $repo FileRepo: repository where we should find the path
 183          * @param $path String: path to file
 184          * @param $key String: key to store the path and any stashed data under
 185          * @param $data String: any other data we want stored with this file
 186          * @throws UploadStashBadPathException
 187          * @throws UploadStashFileNotFoundException
 188          */
 189         public function __construct( $stash, $repo, $path, $key, $data ) {
 190                 $this->sessionStash = $stash;
 191                 $this->sessionKey = $key;
 192                 $this->sessionData = $data;
 193
 194                 // resolve mwrepo:// urls
 195                 if ( $repo->isVirtualUrl( $path ) ) {
 196                         $path = $repo->resolveVirtualUrl( $path );
 197                 }
 198
 199                 // check if path appears to be sane, no parent traversals, and is in this repo's temp zone.
 200                 $repoTempPath = $repo->getZonePath( 'temp' );
 201                 if ( ( ! $repo->validateFilename( $path ) ) ||
 202                                 ( strpos( $path, $repoTempPath ) !== 0 ) ) {
 203                         throw new UploadStashBadPathException( 'path is not valid' );
 204                 }
 205
 206                 // check if path exists! and is a plain file.
 207                 if ( ! $repo->fileExists( $path, FileRepo::FILES_ONLY ) ) {
 208                         throw new UploadStashFileNotFoundException( 'cannot find path, or not a plain file' );
 209                 }
 210
 211                 parent::__construct( false, $repo, $path, false );
 212
 213                 // we will be initializing from some tmpnam files that don't have extensions.
 214                 // most of MediaWiki assumes all uploaded files have good extensions. So, we fix this.
 215                 $this->name = basename( $this->path );
 216                 $this->setExtension();
 217
 218         }
 219
 220         /**
 221          * A method needed by the file transforming and scaling routines in File.php
 222          * We do not necessarily care about doing the description at this point
 223          * However, we also can't return the empty string, as the rest of MediaWiki demands this (and calls to imagemagick
 224          * convert require it to be there)
 225          *
 226          * @return String: dummy value
 227          */
 228         public function getDescriptionUrl() {
 229                 return $this->getUrl();
 230         }
 231
 232         /**
 233          * Find or guess extension -- ensuring that our extension matches our mime type.
 234          * Since these files are constructed from php tempnames they may not start off
 235          * with an extension.
 236          * This does not override getExtension() because things like getMimeType() already call getExtension(),
 237          * and that results in infinite recursion. So, we preemptively *set* the extension so getExtension() can find it.
 238          * For obvious reasons this should be called as early as possible, as part of initialization
 239          */
 240         public function setExtension() {
 241                 // Does this have an extension?
 242                 $n = strrpos( $this->path, '.' );
 243                 $extension = null;
 244                 if ( $n !== false ) {
 245                         $extension = $n ? substr( $this->path, $n + 1 ) : '';
 246                 } else {
 247                         // If not, assume that it should be related to the mime type of the original file.
 248                         //
 249                         // This entire thing is backwards -- we *should* just create an extension based on
 250                         // the mime type of the transformed file, *after* transformation.  But File.php demands
 251                         // to know the name of the transformed file before creating it.
 252                         $mimeType = $this->getMimeType();
 253                         $extensions = explode( ' ', MimeMagic::singleton()->getExtensionsForType( $mimeType ) );
 254                         if ( count( $extensions ) ) {
 255                                 $extension = $extensions[0];
 256                         }
 257                 }
 258
 259                 if ( is_null( $extension ) ) {
 260                         throw new UploadStashFileException( "extension is null" );
 261                 }
 262
 263                 $this->extension = parent::normalizeExtension( $extension );
 264         }
 265
 266         /**
 267          * Get the path for the thumbnail (actually any transformation of this file)
 268          * The actual argument is the result of thumbName although we seem to have
 269          * buggy code elsewhere that expects a boolean 'suffix'
 270          *
 271          * @param $thumbName String: name of thumbnail (e.g. "120px-123456.jpg" ), or false to just get the path
 272          * @return String: path thumbnail should take on filesystem, or containing directory if thumbname is false
 273          */
 274         public function getThumbPath( $thumbName = false ) {
 275                 $path = dirname( $this->path );
 276                 if ( $thumbName !== false ) {
 277                         $path .= "/$thumbName";
 278                 }
 279                 return $path;
 280         }
 281
 282         /**
 283          * Return the file/url base name of a thumbnail with the specified parameters
 284          *
 285          * @param $params Array: handler-specific parameters
 286          * @return String: base name for URL, like '120px-12345.jpg', or null if there is no handler
 287          */
 288         function thumbName( $params ) {
 289                 if ( !$this->getHandler() ) {
 290                         return null;
 291                 }
 292                 $extension = $this->getExtension();
 293                 list( $thumbExt, $thumbMime ) = $this->handler->getThumbType( $extension, $this->getMimeType(), $params );
 294                 $thumbName = $this->getHandler()->makeParamString( $params ) . '-' . $this->getUrlName();
 295                 if ( $thumbExt != $extension ) {
 296                         $thumbName .= ".$thumbExt";
 297                 }
 298                 return $thumbName;
 299         }
 300
 301         /**
 302          * Get a URL to access the thumbnail
 303          * This is required because the model of how files work requires that
 304          * the thumbnail urls be predictable. However, in our model the URL is not based on the filename
 305          * (that's hidden in the session)
 306          *
 307          * @param $thumbName String: basename of thumbnail file -- however, we don't want to use the file exactly
 308          * @return String: URL to access thumbnail, or URL with partial path
 309          */
 310         public function getThumbUrl( $thumbName = false ) {
 311                 $path = $this->sessionStash->getBaseUrl();
 312                 if ( $thumbName !== false ) {
 313                         $path .= '/' . rawurlencode( $thumbName );
 314                 }
 315                 return $path;
 316         }
 317
 318         /**
 319          * The basename for the URL, which we want to not be related to the filename.
 320          * Will also be used as the lookup key for a thumbnail file.
 321          *
 322          * @return String: base url name, like '120px-123456.jpg'
 323          */
 324         public function getUrlName() {
 325                 if ( ! $this->urlName ) {
 326                         $this->urlName = $this->sessionKey . '.' . $this->getExtension();
 327                 }
 328                 return $this->urlName;
 329         }
 330
 331         /**
 332          * Return the URL of the file, if for some reason we wanted to download it
 333          * We tend not to do this for the original file, but we do want thumb icons
 334          *
 335          * @return String: url
 336          */
 337         public function getUrl() {
 338                 if ( !isset( $this->url ) ) {
 339                         $this->url = $this->sessionStash->getBaseUrl() . '/' . $this->getUrlName();
 340                 }
 341                 return $this->url;
 342         }
 343
 344         /**
 345          * Parent classes use this method, for no obvious reason, to return the path (relative to wiki root, I assume).
 346          * But with this class, the URL is unrelated to the path.
 347          *
 348          * @return String: url
 349          */
 350         public function getFullUrl() {
 351                 return $this->getUrl();
 352         }
 353
 354
 355         /**
 356          * Getter for session key (the session-unique id by which this file's location & metadata is stored in the session)
 357          *
 358          * @return String: session key
 359          */
 360         public function getSessionKey() {
 361                 return $this->sessionKey;
 362         }
 363
 364         /**
 365          * Typically, transform() returns a ThumbnailImage, which you can think of as being the exact
 366          * equivalent of an HTML thumbnail on Wikipedia. So its URL is the full-size file, not the thumbnail's URL.
 367          *
 368          * Here we override transform() to stash the thumbnail file, and then
 369          * provide a way to get at the stashed thumbnail file to extract properties such as its URL
 370          *
 371          * @param $params Array: parameters suitable for File::transform()
 372          * @param $flags Integer: bitmask, flags suitable for File::transform()
 373          * @return ThumbnailImage: with additional File thumbnailFile property
 374          */
 375         public function transform( $params, $flags = 0 ) {
 376
 377                 // force it to get a thumbnail right away
 378                 $flags |= self::RENDER_NOW;
 379
 380                 // returns a ThumbnailImage object containing the url and path. Note. NOT A FILE OBJECT.
 381                 $thumb = parent::transform( $params, $flags );
 382                 $key = $this->thumbName($params);
 383
 384                 // remove extension, so it's stored in the session under '120px-123456'
 385                 // this makes it uniform with the other session key for the original, '123456'
 386                 $n = strrpos( $key, '.' );
 387                 if ( $n !== false ) {
 388                         $key = substr( $key, 0, $n );
 389                 }
 390
 391                 // stash the thumbnail File, and provide our caller with a way to get at its properties
 392                 $stashedThumbFile = $this->sessionStash->stashFile( $thumb->getPath(), array(), $key );
 393                 $thumb->thumbnailFile = $stashedThumbFile;
 394
 395                 return $thumb;
 396
 397         }
 398
 399         /**
 400          * Remove the associated temporary file
 401          * @return Status: success
 402          */
 403         public function remove() {
 404                 return $this->repo->freeTemp( $this->path );
 405         }
 406
 407 }
 408
 409 class UploadStashNotAvailableException extends MWException {};
 410 class UploadStashFileNotFoundException extends MWException {};
 411 class UploadStashBadPathException extends MWException {};
 412 class UploadStashBadVersionException extends MWException {};
 413 class UploadStashFileException extends MWException {};
 414