includes/media/MediaHandler.php

   1 <?php
   2 /**
   3  * Media-handling base classes and generic functionality.
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  * @ingroup Media
  22  */
  23
  24 /**
  25  * Base media handler class
  26  *
  27  * @ingroup Media
  28  */
  29 abstract class MediaHandler {
  30         const TRANSFORM_LATER = 1;
  31         const METADATA_GOOD = true;
  32         const METADATA_BAD = false;
  33         const METADATA_COMPATIBLE = 2; // for old but backwards compatible.
  34
  35         /**
  36          * Instance cache
  37          */
  38         static $handlers = array();
  39
  40         /**
  41          * Get a MediaHandler for a given MIME type from the instance cache
  42          *
  43          * @param string $type
  44          *
  45          * @return MediaHandler
  46          */
  47         static function getHandler( $type ) {
  48                 global $wgMediaHandlers;
  49                 if ( !isset( $wgMediaHandlers[$type] ) ) {
  50                         wfDebug( __METHOD__ . ": no handler found for $type.\n" );
  51
  52                         return false;
  53                 }
  54                 $class = $wgMediaHandlers[$type];
  55                 if ( !isset( self::$handlers[$class] ) ) {
  56                         self::$handlers[$class] = new $class;
  57                         if ( !self::$handlers[$class]->isEnabled() ) {
  58                                 self::$handlers[$class] = false;
  59                         }
  60                 }
  61
  62                 return self::$handlers[$class];
  63         }
  64
  65         /**
  66          * Get an associative array mapping magic word IDs to parameter names.
  67          * Will be used by the parser to identify parameters.
  68          */
  69         abstract function getParamMap();
  70
  71         /**
  72          * Validate a thumbnail parameter at parse time.
  73          * Return true to accept the parameter, and false to reject it.
  74          * If you return false, the parser will do something quiet and forgiving.
  75          *
  76          * @param string $name
  77          * @param $value
  78          */
  79         abstract function validateParam( $name, $value );
  80
  81         /**
  82          * Merge a parameter array into a string appropriate for inclusion in filenames
  83          *
  84          * @param array $params Array of parameters that have been through normaliseParams.
  85          * @return string
  86          */
  87         abstract function makeParamString( $params );
  88
  89         /**
  90          * Parse a param string made with makeParamString back into an array
  91          *
  92          * @param string $str The parameter string without file name (e.g. 122px)
  93          * @return array|bool Array of parameters or false on failure.
  94          */
  95         abstract function parseParamString( $str );
  96
  97         /**
  98          * Changes the parameter array as necessary, ready for transformation.
  99          * Should be idempotent.
 100          * Returns false if the parameters are unacceptable and the transform should fail
 101          * @param $image
 102          * @param $params
 103          */
 104         abstract function normaliseParams( $image, &$params );
 105
 106         /**
 107          * Get an image size array like that returned by getimagesize(), or false if it
 108          * can't be determined.
 109          *
 110          * @param File $image The image object, or false if there isn't one
 111          * @param string $path the filename
 112          * @return array Follow the format of PHP getimagesize() internal function. See http://www.php.net/getimagesize
 113          */
 114         abstract function getImageSize( $image, $path );
 115
 116         /**
 117          * Get handler-specific metadata which will be saved in the img_metadata field.
 118          *
 119          * @param File $image The image object, or false if there isn't one.
 120          *   Warning, FSFile::getPropsFromPath might pass an (object)array() instead (!)
 121          * @param string $path The filename
 122          * @return string
 123          */
 124         function getMetadata( $image, $path ) {
 125                 return '';
 126         }
 127
 128         /**
 129          * Get metadata version.
 130          *
 131          * This is not used for validating metadata, this is used for the api when returning
 132          * metadata, since api content formats should stay the same over time, and so things
 133          * using ForiegnApiRepo can keep backwards compatibility
 134          *
 135          * All core media handlers share a common version number, and extensions can
 136          * use the GetMetadataVersion hook to append to the array (they should append a unique
 137          * string so not to get confusing). If there was a media handler named 'foo' with metadata
 138          * version 3 it might add to the end of the array the element 'foo=3'. if the core metadata
 139          * version is 2, the end version string would look like '2;foo=3'.
 140          *
 141          * @return string Version string
 142          */
 143         static function getMetadataVersion() {
 144                 $version = array( '2' ); // core metadata version
 145                 wfRunHooks( 'GetMetadataVersion', array( &$version ) );
 146
 147                 return implode( ';', $version );
 148         }
 149
 150         /**
 151          * Convert metadata version.
 152          *
 153          * By default just returns $metadata, but can be used to allow
 154          * media handlers to convert between metadata versions.
 155          *
 156          * @param mixed|string|array $metadata Metadata array (serialized if string)
 157          * @param int $version Target version
 158          * @return array Serialized metadata in specified version, or $metadata on fail.
 159          */
 160         function convertMetadataVersion( $metadata, $version = 1 ) {
 161                 if ( !is_array( $metadata ) ) {
 162
 163                         //unserialize to keep return parameter consistent.
 164                         wfSuppressWarnings();
 165                         $ret = unserialize( $metadata );
 166                         wfRestoreWarnings();
 167
 168                         return $ret;
 169                 }
 170
 171                 return $metadata;
 172         }
 173
 174         /**
 175          * Get a string describing the type of metadata, for display purposes.
 176          * @param $image
 177          * @return string
 178          */
 179         function getMetadataType( $image ) {
 180                 return false;
 181         }
 182
 183         /**
 184          * Check if the metadata string is valid for this handler.
 185          * If it returns MediaHandler::METADATA_BAD (or false), Image
 186          * will reload the metadata from the file and update the database.
 187          * MediaHandler::METADATA_GOOD for if the metadata is a-ok,
 188          * MediaHanlder::METADATA_COMPATIBLE if metadata is old but backwards
 189          * compatible (which may or may not trigger a metadata reload).
 190          * @return bool
 191          */
 192         function isMetadataValid( $image, $metadata ) {
 193                 return self::METADATA_GOOD;
 194         }
 195
 196         /**
 197          * Get an array of standard (FormatMetadata type) metadata values.
 198          *
 199          * The returned data is largely the same as that from getMetadata(),
 200          * but formatted in a standard, stable, handler-independent way.
 201          * The idea being that some values like ImageDescription or Artist
 202          * are universal and should be retrievable in a handler generic way.
 203          *
 204          * The specific properties are the type of properties that can be
 205          * handled by the FormatMetadata class. These values are exposed to the
 206          * user via the filemetadata parser function.
 207          *
 208          * Details of the response format of this function can be found at
 209          * https://www.mediawiki.org/wiki/Manual:File_metadata_handling
 210          * tl/dr: the response is an associative array of
 211          * properties keyed by name, but the value can be complex. You probably
 212          * want to call one of the FormatMetadata::flatten* functions on the
 213          * property values before using them, or call
 214          * FormatMetadata::getFormattedData() on the full response array, which
 215          * transforms all values into prettified, human-readable text.
 216          *
 217          * Subclasses overriding this function must return a value which is a
 218          * valid API response fragment (all associative array keys are valid
 219          * XML tagnames).
 220          *
 221          * Note, if the file simply has no metadata, but the handler supports
 222          * this interface, it should return an empty array, not false.
 223          *
 224          * @param File $file
 225          *
 226          * @return array|bool False if interface not supported
 227          * @since 1.23
 228          */
 229         public function getCommonMetaArray( File $file ) {
 230                 return false;
 231         }
 232
 233         /**
 234          * Get a MediaTransformOutput object representing an alternate of the transformed
 235          * output which will call an intermediary thumbnail assist script.
 236          *
 237          * Used when the repository has a thumbnailScriptUrl option configured.
 238          *
 239          * Return false to fall back to the regular getTransform().
 240          * @return bool
 241          */
 242         function getScriptedTransform( $image, $script, $params ) {
 243                 return false;
 244         }
 245
 246         /**
 247          * Get a MediaTransformOutput object representing the transformed output. Does not
 248          * actually do the transform.
 249          *
 250          * @param File $image The image object
 251          * @param string $dstPath filesystem destination path
 252          * @param string $dstUrl Destination URL to use in output HTML
 253          * @param array $params Arbitrary set of parameters validated by $this->validateParam()
 254          * @return MediaTransformOutput
 255          */
 256         final function getTransform( $image, $dstPath, $dstUrl, $params ) {
 257                 return $this->doTransform( $image, $dstPath, $dstUrl, $params, self::TRANSFORM_LATER );
 258         }
 259
 260         /**
 261          * Get a MediaTransformOutput object representing the transformed output. Does the
 262          * transform unless $flags contains self::TRANSFORM_LATER.
 263          *
 264          * @param File $image The image object
 265          * @param string $dstPath filesystem destination path
 266          * @param string $dstUrl destination URL to use in output HTML
 267          * @param array $params arbitrary set of parameters validated by $this->validateParam()
 268          *   Note: These parameters have *not* gone through $this->normaliseParams()
 269          * @param int $flags A bitfield, may contain self::TRANSFORM_LATER
 270          *
 271          * @return MediaTransformOutput
 272          */
 273         abstract function doTransform( $image, $dstPath, $dstUrl, $params, $flags = 0 );
 274
 275         /**
 276          * Get the thumbnail extension and MIME type for a given source MIME type
 277          *
 278          * @param string $ext Extension of original file
 279          * @param string $mime Mime type of original file
 280          * @param array $params Handler specific rendering parameters
 281          * @return array thumbnail extension and MIME type
 282          */
 283         function getThumbType( $ext, $mime, $params = null ) {
 284                 $magic = MimeMagic::singleton();
 285                 if ( !$ext || $magic->isMatchingExtension( $ext, $mime ) === false ) {
 286                         // The extension is not valid for this mime type and we do
 287                         // recognize the mime type
 288                         $extensions = $magic->getExtensionsForType( $mime );
 289                         if ( $extensions ) {
 290                                 return array( strtok( $extensions, ' ' ), $mime );
 291                         }
 292                 }
 293
 294                 // The extension is correct (true) or the mime type is unknown to
 295                 // MediaWiki (null)
 296                 return array( $ext, $mime );
 297         }
 298
 299         /**
 300          * Get useful response headers for GET/HEAD requests for a file with the given metadata
 301          * @param mixed $metadata Result of the getMetadata() function of this handler for a file
 302          * @return array
 303          */
 304         public function getStreamHeaders( $metadata ) {
 305                 return array();
 306         }
 307
 308         /**
 309          * True if the handled types can be transformed
 310          * @return bool
 311          */
 312         function canRender( $file ) {
 313                 return true;
 314         }
 315
 316         /**
 317          * True if handled types cannot be displayed directly in a browser
 318          * but can be rendered
 319          * @return bool
 320          */
 321         function mustRender( $file ) {
 322                 return false;
 323         }
 324
 325         /**
 326          * True if the type has multi-page capabilities
 327          * @return bool
 328          */
 329         function isMultiPage( $file ) {
 330                 return false;
 331         }
 332
 333         /**
 334          * Page count for a multi-page document, false if unsupported or unknown
 335          * @return bool
 336          */
 337         function pageCount( $file ) {
 338                 return false;
 339         }
 340
 341         /**
 342          * The material is vectorized and thus scaling is lossless
 343          * @return bool
 344          */
 345         function isVectorized( $file ) {
 346                 return false;
 347         }
 348
 349         /**
 350          * The material is an image, and is animated.
 351          * In particular, video material need not return true.
 352          * @note Before 1.20, this was a method of ImageHandler only
 353          * @return bool
 354          */
 355         function isAnimatedImage( $file ) {
 356                 return false;
 357         }
 358
 359         /**
 360          * If the material is animated, we can animate the thumbnail
 361          * @since 1.20
 362          * @return bool If material is not animated, handler may return any value.
 363          */
 364         function canAnimateThumbnail( $file ) {
 365                 return true;
 366         }
 367
 368         /**
 369          * False if the handler is disabled for all files
 370          * @return bool
 371          */
 372         function isEnabled() {
 373                 return true;
 374         }
 375
 376         /**
 377          * Get an associative array of page dimensions
 378          * Currently "width" and "height" are understood, but this might be
 379          * expanded in the future.
 380          * Returns false if unknown.
 381          *
 382          * It is expected that handlers for paged media (e.g. DjVuHandler)
 383          * will override this method so that it gives the correct results
 384          * for each specific page of the file, using the $page argument.
 385          *
 386          * @note For non-paged media, use getImageSize.
 387          *
 388          * @param File $image
 389          * @param int $page What page to get dimensions of
 390          * @return array|bool
 391          */
 392         function getPageDimensions( $image, $page ) {
 393                 $gis = $this->getImageSize( $image, $image->getLocalRefPath() );
 394                 if ( $gis ) {
 395                         return array(
 396                                 'width' => $gis[0],
 397                                 'height' => $gis[1]
 398                         );
 399                 } else {
 400                         return false;
 401                 }
 402         }
 403
 404         /**
 405          * Generic getter for text layer.
 406          * Currently overloaded by PDF and DjVu handlers
 407          * @return bool
 408          */
 409         function getPageText( $image, $page ) {
 410                 return false;
 411         }
 412
 413         /**
 414          * Get an array structure that looks like this:
 415          *
 416          * array(
 417          *    'visible' => array(
 418          *       'Human-readable name' => 'Human readable value',
 419          *       ...
 420          *    ),
 421          *    'collapsed' => array(
 422          *       'Human-readable name' => 'Human readable value',
 423          *       ...
 424          *    )
 425          * )
 426          * The UI will format this into a table where the visible fields are always
 427          * visible, and the collapsed fields are optionally visible.
 428          *
 429          * The function should return false if there is no metadata to display.
 430          */
 431
 432         /**
 433          * @todo FIXME: I don't really like this interface, it's not very flexible
 434          * I think the media handler should generate HTML instead. It can do
 435          * all the formatting according to some standard. That makes it possible
 436          * to do things like visual indication of grouped and chained streams
 437          * in ogg container files.
 438          * @return bool
 439          */
 440         function formatMetadata( $image ) {
 441                 return false;
 442         }
 443
 444         /** sorts the visible/invisible field.
 445          * Split off from ImageHandler::formatMetadata, as used by more than
 446          * one type of handler.
 447          *
 448          * This is used by the media handlers that use the FormatMetadata class
 449          *
 450          * @param array $metadataArray metadata array
 451          * @return array for use displaying metadata.
 452          */
 453         function formatMetadataHelper( $metadataArray ) {
 454                 $result = array(
 455                         'visible' => array(),
 456                         'collapsed' => array()
 457                 );
 458
 459                 $formatted = FormatMetadata::getFormattedData( $metadataArray );
 460                 // Sort fields into visible and collapsed
 461                 $visibleFields = $this->visibleMetadataFields();
 462                 foreach ( $formatted as $name => $value ) {
 463                         $tag = strtolower( $name );
 464                         self::addMeta( $result,
 465                                 in_array( $tag, $visibleFields ) ? 'visible' : 'collapsed',
 466                                 'exif',
 467                                 $tag,
 468                                 $value
 469                         );
 470                 }
 471
 472                 return $result;
 473         }
 474
 475         /**
 476          * Get a list of metadata items which should be displayed when
 477          * the metadata table is collapsed.
 478          *
 479          * @return array of strings
 480          * @access protected
 481          */
 482         function visibleMetadataFields() {
 483                 return FormatMetadata::getVisibleFields();
 484         }
 485
 486         /**
 487          * This is used to generate an array element for each metadata value
 488          * That array is then used to generate the table of metadata values
 489          * on the image page
 490          *
 491          * @param &$array Array An array containing elements for each type of visibility
 492          * and each of those elements being an array of metadata items. This function adds
 493          * a value to that array.
 494          * @param string $visibility ('visible' or 'collapsed') if this value is hidden
 495          * by default.
 496          * @param string $type type of metadata tag (currently always 'exif')
 497          * @param string $id the name of the metadata tag (like 'artist' for example).
 498          * its name in the table displayed is the message "$type-$id" (Ex exif-artist ).
 499          * @param string $value thingy goes into a wikitext table; it used to be escaped but
 500          * that was incompatible with previous practise of customized display
 501          * with wikitext formatting via messages such as 'exif-model-value'.
 502          * So the escaping is taken back out, but generally this seems a confusing
 503          * interface.
 504          * @param string $param value to pass to the message for the name of the field
 505          * as $1. Currently this parameter doesn't seem to ever be used.
 506          *
 507          * Note, everything here is passed through the parser later on (!)
 508          */
 509         protected static function addMeta( &$array, $visibility, $type, $id, $value, $param = false ) {
 510                 $msg = wfMessage( "$type-$id", $param );
 511                 if ( $msg->exists() ) {
 512                         $name = $msg->text();
 513                 } else {
 514                         // This is for future compatibility when using instant commons.
 515                         // So as to not display as ugly a name if a new metadata
 516                         // property is defined that we don't know about
 517                         // (not a major issue since such a property would be collapsed
 518                         // by default).
 519                         wfDebug( __METHOD__ . ' Unknown metadata name: ' . $id . "\n" );
 520                         $name = wfEscapeWikiText( $id );
 521                 }
 522                 $array[$visibility][] = array(
 523                         'id' => "$type-$id",
 524                         'name' => $name,
 525                         'value' => $value
 526                 );
 527         }
 528
 529         /**
 530          * Used instead of getLongDesc if there is no handler registered for file.
 531          *
 532          * @param $file File
 533          * @return string
 534          */
 535         function getShortDesc( $file ) {
 536                 global $wgLang;
 537
 538                 return htmlspecialchars( $wgLang->formatSize( $file->getSize() ) );
 539         }
 540
 541         /**
 542          * Short description. Shown on Special:Search results.
 543          *
 544          * @param $file File
 545          * @return string
 546          */
 547         function getLongDesc( $file ) {
 548                 global $wgLang;
 549
 550                 return wfMessage( 'file-info', htmlspecialchars( $wgLang->formatSize( $file->getSize() ) ),
 551                         $file->getMimeType() )->parse();
 552         }
 553
 554         /**
 555          * Long description. Shown under image on image description page surounded by ().
 556          *
 557          * @param $file File
 558          * @return string
 559          */
 560         static function getGeneralShortDesc( $file ) {
 561                 global $wgLang;
 562
 563                 return $wgLang->formatSize( $file->getSize() );
 564         }
 565
 566         /**
 567          * Used instead of getShortDesc if there is no handler registered for file.
 568          *
 569          * @param $file File
 570          * @return string
 571          */
 572         static function getGeneralLongDesc( $file ) {
 573                 global $wgLang;
 574
 575                 return wfMessage( 'file-info', $wgLang->formatSize( $file->getSize() ),
 576                         $file->getMimeType() )->parse();
 577         }
 578
 579         /**
 580          * Calculate the largest thumbnail width for a given original file size
 581          * such that the thumbnail's height is at most $maxHeight.
 582          * @param $boxWidth Integer Width of the thumbnail box.
 583          * @param $boxHeight Integer Height of the thumbnail box.
 584          * @param $maxHeight Integer Maximum height expected for the thumbnail.
 585          * @return Integer.
 586          */
 587         public static function fitBoxWidth( $boxWidth, $boxHeight, $maxHeight ) {
 588                 $idealWidth = $boxWidth * $maxHeight / $boxHeight;
 589                 $roundedUp = ceil( $idealWidth );
 590                 if ( round( $roundedUp * $boxHeight / $boxWidth ) > $maxHeight ) {
 591                         return floor( $idealWidth );
 592                 } else {
 593                         return $roundedUp;
 594                 }
 595         }
 596
 597         /**
 598          * Shown in file history box on image description page.
 599          *
 600          * @param File $file
 601          * @return String Dimensions
 602          */
 603         function getDimensionsString( $file ) {
 604                 return '';
 605         }
 606
 607         /**
 608          * Modify the parser object post-transform.
 609          *
 610          * This is often used to do $parser->addOutputHook(),
 611          * in order to add some javascript to render a viewer.
 612          * See TimedMediaHandler or OggHandler for an example.
 613          *
 614          * @param Parser $parser
 615          * @param File $file
 616          */
 617         function parserTransformHook( $parser, $file ) {
 618         }
 619
 620         /**
 621          * File validation hook called on upload.
 622          *
 623          * If the file at the given local path is not valid, or its MIME type does not
 624          * match the handler class, a Status object should be returned containing
 625          * relevant errors.
 626          *
 627          * @param string $fileName The local path to the file.
 628          * @return Status object
 629          */
 630         function verifyUpload( $fileName ) {
 631                 return Status::newGood();
 632         }
 633
 634         /**
 635          * Check for zero-sized thumbnails. These can be generated when
 636          * no disk space is available or some other error occurs
 637          *
 638          * @param string $dstPath The location of the suspect file
 639          * @param int $retval Return value of some shell process, file will be deleted if this is non-zero
 640          * @return bool True if removed, false otherwise
 641          */
 642         function removeBadFile( $dstPath, $retval = 0 ) {
 643                 if ( file_exists( $dstPath ) ) {
 644                         $thumbstat = stat( $dstPath );
 645                         if ( $thumbstat['size'] == 0 || $retval != 0 ) {
 646                                 $result = unlink( $dstPath );
 647
 648                                 if ( $result ) {
 649                                         wfDebugLog( 'thumbnail',
 650                                                 sprintf( 'Removing bad %d-byte thumbnail "%s". unlink() succeeded',
 651                                                         $thumbstat['size'], $dstPath ) );
 652                                 } else {
 653                                         wfDebugLog( 'thumbnail',
 654                                                 sprintf( 'Removing bad %d-byte thumbnail "%s". unlink() failed',
 655                                                         $thumbstat['size'], $dstPath ) );
 656                                 }
 657
 658                                 return true;
 659                         }
 660                 }
 661
 662                 return false;
 663         }
 664
 665         /**
 666          * Remove files from the purge list.
 667          *
 668          * This is used by some video handlers to prevent ?action=purge
 669          * from removing a transcoded video, which is expensive to
 670          * regenerate.
 671          *
 672          * @see LocalFile::purgeThumbnails
 673          *
 674          * @param array $files
 675          * @param array $options Purge options. Currently will always be
 676          *  an array with a single key 'forThumbRefresh' set to true.
 677          */
 678         public function filterThumbnailPurgeList( &$files, $options ) {
 679                 // Do nothing
 680         }
 681
 682         /*
 683          * True if the handler can rotate the media
 684          * @since 1.21
 685          * @return bool
 686          */
 687         public static function canRotate() {
 688                 return false;
 689         }
 690
 691         /**
 692          * On supporting image formats, try to read out the low-level orientation
 693          * of the file and return the angle that the file needs to be rotated to
 694          * be viewed.
 695          *
 696          * This information is only useful when manipulating the original file;
 697          * the width and height we normally work with is logical, and will match
 698          * any produced output views.
 699          *
 700          * For files we don't know, we return 0.
 701          *
 702          * @param $file File
 703          * @return int 0, 90, 180 or 270
 704          */
 705         public function getRotation( $file ) {
 706                 return 0;
 707         }
 708 }