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         /** @var MediaHandler[] Instance cache with array of MediaHandler */
  36         protected static $handlers = array();
  37
  38         /**
  39          * Get a MediaHandler for a given MIME type from the instance cache
  40          *
  41          * @param string $type
  42          *
  43          * @return MediaHandler
  44          */
  45         static function getHandler( $type ) {
  46                 global $wgMediaHandlers;
  47                 if ( !isset( $wgMediaHandlers[$type] ) ) {
  48                         wfDebug( __METHOD__ . ": no handler found for $type.\n" );
  49
  50                         return false;
  51                 }
  52                 $class = $wgMediaHandlers[$type];
  53                 if ( !isset( self::$handlers[$class] ) ) {
  54                         self::$handlers[$class] = new $class;
  55                         if ( !self::$handlers[$class]->isEnabled() ) {
  56                                 self::$handlers[$class] = false;
  57                         }
  58                 }
  59
  60                 return self::$handlers[$class];
  61         }
  62
  63         /**
  64          * Get an associative array mapping magic word IDs to parameter names.
  65          * Will be used by the parser to identify parameters.
  66          */
  67         abstract function getParamMap();
  68
  69         /**
  70          * Validate a thumbnail parameter at parse time.
  71          * Return true to accept the parameter, and false to reject it.
  72          * If you return false, the parser will do something quiet and forgiving.
  73          *
  74          * @param string $name
  75          * @param $value
  76          */
  77         abstract function validateParam( $name, $value );
  78
  79         /**
  80          * Merge a parameter array into a string appropriate for inclusion in filenames
  81          *
  82          * @param array $params Array of parameters that have been through normaliseParams.
  83          * @return string
  84          */
  85         abstract function makeParamString( $params );
  86
  87         /**
  88          * Parse a param string made with makeParamString back into an array
  89          *
  90          * @param string $str The parameter string without file name (e.g. 122px)
  91          * @return array|bool Array of parameters or false on failure.
  92          */
  93         abstract function parseParamString( $str );
  94
  95         /**
  96          * Changes the parameter array as necessary, ready for transformation.
  97          * Should be idempotent.
  98          * Returns false if the parameters are unacceptable and the transform should fail
  99          * @param $image
 100          * @param $params
 101          */
 102         abstract function normaliseParams( $image, &$params );
 103
 104         /**
 105          * Get an image size array like that returned by getimagesize(), or false if it
 106          * can't be determined.
 107          *
 108          * @param File $image The image object, or false if there isn't one
 109          * @param string $path the filename
 110          * @return array Follow the format of PHP getimagesize() internal function. See http://www.php.net/getimagesize
 111          */
 112         abstract function getImageSize( $image, $path );
 113
 114         /**
 115          * Get handler-specific metadata which will be saved in the img_metadata field.
 116          *
 117          * @param File $image The image object, or false if there isn't one.
 118          *   Warning, FSFile::getPropsFromPath might pass an (object)array() instead (!)
 119          * @param string $path The filename
 120          * @return string
 121          */
 122         function getMetadata( $image, $path ) {
 123                 return '';
 124         }
 125
 126         /**
 127          * Get metadata version.
 128          *
 129          * This is not used for validating metadata, this is used for the api when returning
 130          * metadata, since api content formats should stay the same over time, and so things
 131          * using ForiegnApiRepo can keep backwards compatibility
 132          *
 133          * All core media handlers share a common version number, and extensions can
 134          * use the GetMetadataVersion hook to append to the array (they should append a unique
 135          * string so not to get confusing). If there was a media handler named 'foo' with metadata
 136          * version 3 it might add to the end of the array the element 'foo=3'. if the core metadata
 137          * version is 2, the end version string would look like '2;foo=3'.
 138          *
 139          * @return string Version string
 140          */
 141         static function getMetadataVersion() {
 142                 $version = array( '2' ); // core metadata version
 143                 wfRunHooks( 'GetMetadataVersion', array( &$version ) );
 144
 145                 return implode( ';', $version );
 146         }
 147
 148         /**
 149          * Convert metadata version.
 150          *
 151          * By default just returns $metadata, but can be used to allow
 152          * media handlers to convert between metadata versions.
 153          *
 154          * @param mixed|string|array $metadata Metadata array (serialized if string)
 155          * @param int $version Target version
 156          * @return array Serialized metadata in specified version, or $metadata on fail.
 157          */
 158         function convertMetadataVersion( $metadata, $version = 1 ) {
 159                 if ( !is_array( $metadata ) ) {
 160
 161                         //unserialize to keep return parameter consistent.
 162                         wfSuppressWarnings();
 163                         $ret = unserialize( $metadata );
 164                         wfRestoreWarnings();
 165
 166                         return $ret;
 167                 }
 168
 169                 return $metadata;
 170         }
 171
 172         /**
 173          * Get a string describing the type of metadata, for display purposes.
 174          * @param $image
 175          * @return string
 176          */
 177         function getMetadataType( $image ) {
 178                 return false;
 179         }
 180
 181         /**
 182          * Check if the metadata string is valid for this handler.
 183          * If it returns MediaHandler::METADATA_BAD (or false), Image
 184          * will reload the metadata from the file and update the database.
 185          * MediaHandler::METADATA_GOOD for if the metadata is a-ok,
 186          * MediaHanlder::METADATA_COMPATIBLE if metadata is old but backwards
 187          * compatible (which may or may not trigger a metadata reload).
 188          * @return bool
 189          */
 190         function isMetadataValid( $image, $metadata ) {
 191                 return self::METADATA_GOOD;
 192         }
 193
 194         /**
 195          * Get an array of standard (FormatMetadata type) metadata values.
 196          *
 197          * The returned data is largely the same as that from getMetadata(),
 198          * but formatted in a standard, stable, handler-independent way.
 199          * The idea being that some values like ImageDescription or Artist
 200          * are universal and should be retrievable in a handler generic way.
 201          *
 202          * The specific properties are the type of properties that can be
 203          * handled by the FormatMetadata class. These values are exposed to the
 204          * user via the filemetadata parser function.
 205          *
 206          * Details of the response format of this function can be found at
 207          * https://www.mediawiki.org/wiki/Manual:File_metadata_handling
 208          * tl/dr: the response is an associative array of
 209          * properties keyed by name, but the value can be complex. You probably
 210          * want to call one of the FormatMetadata::flatten* functions on the
 211          * property values before using them, or call
 212          * FormatMetadata::getFormattedData() on the full response array, which
 213          * transforms all values into prettified, human-readable text.
 214          *
 215          * Subclasses overriding this function must return a value which is a
 216          * valid API response fragment (all associative array keys are valid
 217          * XML tagnames).
 218          *
 219          * Note, if the file simply has no metadata, but the handler supports
 220          * this interface, it should return an empty array, not false.
 221          *
 222          * @param File $file
 223          *
 224          * @return array|bool False if interface not supported
 225          * @since 1.23
 226          */
 227         public function getCommonMetaArray( File $file ) {
 228                 return false;
 229         }
 230
 231         /**
 232          * Get a MediaTransformOutput object representing an alternate of the transformed
 233          * output which will call an intermediary thumbnail assist script.
 234          *
 235          * Used when the repository has a thumbnailScriptUrl option configured.
 236          *
 237          * Return false to fall back to the regular getTransform().
 238          * @return bool
 239          */
 240         function getScriptedTransform( $image, $script, $params ) {
 241                 return false;
 242         }
 243
 244         /**
 245          * Get a MediaTransformOutput object representing the transformed output. Does not
 246          * actually do the transform.
 247          *
 248          * @param File $image The image object
 249          * @param string $dstPath filesystem destination path
 250          * @param string $dstUrl Destination URL to use in output HTML
 251          * @param array $params Arbitrary set of parameters validated by $this->validateParam()
 252          * @return MediaTransformOutput
 253          */
 254         final function getTransform( $image, $dstPath, $dstUrl, $params ) {
 255                 return $this->doTransform( $image, $dstPath, $dstUrl, $params, self::TRANSFORM_LATER );
 256         }
 257
 258         /**
 259          * Get a MediaTransformOutput object representing the transformed output. Does the
 260          * transform unless $flags contains self::TRANSFORM_LATER.
 261          *
 262          * @param File $image The image object
 263          * @param string $dstPath filesystem destination path
 264          * @param string $dstUrl destination URL to use in output HTML
 265          * @param array $params arbitrary set of parameters validated by $this->validateParam()
 266          *   Note: These parameters have *not* gone through $this->normaliseParams()
 267          * @param int $flags A bitfield, may contain self::TRANSFORM_LATER
 268          *
 269          * @return MediaTransformOutput
 270          */
 271         abstract function doTransform( $image, $dstPath, $dstUrl, $params, $flags = 0 );
 272
 273         /**
 274          * Get the thumbnail extension and MIME type for a given source MIME type
 275          *
 276          * @param string $ext Extension of original file
 277          * @param string $mime Mime type of original file
 278          * @param array $params Handler specific rendering parameters
 279          * @return array thumbnail extension and MIME type
 280          */
 281         function getThumbType( $ext, $mime, $params = null ) {
 282                 $magic = MimeMagic::singleton();
 283                 if ( !$ext || $magic->isMatchingExtension( $ext, $mime ) === false ) {
 284                         // The extension is not valid for this mime type and we do
 285                         // recognize the mime type
 286                         $extensions = $magic->getExtensionsForType( $mime );
 287                         if ( $extensions ) {
 288                                 return array( strtok( $extensions, ' ' ), $mime );
 289                         }
 290                 }
 291
 292                 // The extension is correct (true) or the mime type is unknown to
 293                 // MediaWiki (null)
 294                 return array( $ext, $mime );
 295         }
 296
 297         /**
 298          * Get useful response headers for GET/HEAD requests for a file with the given metadata
 299          * @param mixed $metadata Result of the getMetadata() function of this handler for a file
 300          * @return array
 301          */
 302         public function getStreamHeaders( $metadata ) {
 303                 return array();
 304         }
 305
 306         /**
 307          * True if the handled types can be transformed
 308          * @return bool
 309          */
 310         function canRender( $file ) {
 311                 return true;
 312         }
 313
 314         /**
 315          * True if handled types cannot be displayed directly in a browser
 316          * but can be rendered
 317          * @return bool
 318          */
 319         function mustRender( $file ) {
 320                 return false;
 321         }
 322
 323         /**
 324          * True if the type has multi-page capabilities
 325          * @return bool
 326          */
 327         function isMultiPage( $file ) {
 328                 return false;
 329         }
 330
 331         /**
 332          * Page count for a multi-page document, false if unsupported or unknown
 333          * @return bool
 334          */
 335         function pageCount( $file ) {
 336                 return false;
 337         }
 338
 339         /**
 340          * The material is vectorized and thus scaling is lossless
 341          * @return bool
 342          */
 343         function isVectorized( $file ) {
 344                 return false;
 345         }
 346
 347         /**
 348          * The material is an image, and is animated.
 349          * In particular, video material need not return true.
 350          * @note Before 1.20, this was a method of ImageHandler only
 351          * @return bool
 352          */
 353         function isAnimatedImage( $file ) {
 354                 return false;
 355         }
 356
 357         /**
 358          * If the material is animated, we can animate the thumbnail
 359          * @since 1.20
 360          * @return bool If material is not animated, handler may return any value.
 361          */
 362         function canAnimateThumbnail( $file ) {
 363                 return true;
 364         }
 365
 366         /**
 367          * False if the handler is disabled for all files
 368          * @return bool
 369          */
 370         function isEnabled() {
 371                 return true;
 372         }
 373
 374         /**
 375          * Get an associative array of page dimensions
 376          * Currently "width" and "height" are understood, but this might be
 377          * expanded in the future.
 378          * Returns false if unknown.
 379          *
 380          * It is expected that handlers for paged media (e.g. DjVuHandler)
 381          * will override this method so that it gives the correct results
 382          * for each specific page of the file, using the $page argument.
 383          *
 384          * @note For non-paged media, use getImageSize.
 385          *
 386          * @param File $image
 387          * @param int $page What page to get dimensions of
 388          * @return array|bool
 389          */
 390         function getPageDimensions( $image, $page ) {
 391                 $gis = $this->getImageSize( $image, $image->getLocalRefPath() );
 392                 if ( $gis ) {
 393                         return array(
 394                                 'width' => $gis[0],
 395                                 'height' => $gis[1]
 396                         );
 397                 } else {
 398                         return false;
 399                 }
 400         }
 401
 402         /**
 403          * Generic getter for text layer.
 404          * Currently overloaded by PDF and DjVu handlers
 405          * @return bool
 406          */
 407         function getPageText( $image, $page ) {
 408                 return false;
 409         }
 410
 411         /**
 412          * Get an array structure that looks like this:
 413          *
 414          * array(
 415          *    'visible' => array(
 416          *       'Human-readable name' => 'Human readable value',
 417          *       ...
 418          *    ),
 419          *    'collapsed' => array(
 420          *       'Human-readable name' => 'Human readable value',
 421          *       ...
 422          *    )
 423          * )
 424          * The UI will format this into a table where the visible fields are always
 425          * visible, and the collapsed fields are optionally visible.
 426          *
 427          * The function should return false if there is no metadata to display.
 428          */
 429
 430         /**
 431          * @todo FIXME: I don't really like this interface, it's not very flexible
 432          * I think the media handler should generate HTML instead. It can do
 433          * all the formatting according to some standard. That makes it possible
 434          * to do things like visual indication of grouped and chained streams
 435          * in ogg container files.
 436          * @return bool
 437          */
 438         function formatMetadata( $image ) {
 439                 return false;
 440         }
 441
 442         /** sorts the visible/invisible field.
 443          * Split off from ImageHandler::formatMetadata, as used by more than
 444          * one type of handler.
 445          *
 446          * This is used by the media handlers that use the FormatMetadata class
 447          *
 448          * @param array $metadataArray metadata array
 449          * @return array for use displaying metadata.
 450          */
 451         function formatMetadataHelper( $metadataArray ) {
 452                 $result = array(
 453                         'visible' => array(),
 454                         'collapsed' => array()
 455                 );
 456
 457                 $formatted = FormatMetadata::getFormattedData( $metadataArray );
 458                 // Sort fields into visible and collapsed
 459                 $visibleFields = $this->visibleMetadataFields();
 460                 foreach ( $formatted as $name => $value ) {
 461                         $tag = strtolower( $name );
 462                         self::addMeta( $result,
 463                                 in_array( $tag, $visibleFields ) ? 'visible' : 'collapsed',
 464                                 'exif',
 465                                 $tag,
 466                                 $value
 467                         );
 468                 }
 469
 470                 return $result;
 471         }
 472
 473         /**
 474          * Get a list of metadata items which should be displayed when
 475          * the metadata table is collapsed.
 476          *
 477          * @return array of strings
 478          * @access protected
 479          */
 480         function visibleMetadataFields() {
 481                 return FormatMetadata::getVisibleFields();
 482         }
 483
 484         /**
 485          * This is used to generate an array element for each metadata value
 486          * That array is then used to generate the table of metadata values
 487          * on the image page
 488          *
 489          * @param &$array Array An array containing elements for each type of visibility
 490          * and each of those elements being an array of metadata items. This function adds
 491          * a value to that array.
 492          * @param string $visibility ('visible' or 'collapsed') if this value is hidden
 493          * by default.
 494          * @param string $type type of metadata tag (currently always 'exif')
 495          * @param string $id the name of the metadata tag (like 'artist' for example).
 496          * its name in the table displayed is the message "$type-$id" (Ex exif-artist ).
 497          * @param string $value thingy goes into a wikitext table; it used to be escaped but
 498          * that was incompatible with previous practise of customized display
 499          * with wikitext formatting via messages such as 'exif-model-value'.
 500          * So the escaping is taken back out, but generally this seems a confusing
 501          * interface.
 502          * @param string $param value to pass to the message for the name of the field
 503          * as $1. Currently this parameter doesn't seem to ever be used.
 504          *
 505          * Note, everything here is passed through the parser later on (!)
 506          */
 507         protected static function addMeta( &$array, $visibility, $type, $id, $value, $param = false ) {
 508                 $msg = wfMessage( "$type-$id", $param );
 509                 if ( $msg->exists() ) {
 510                         $name = $msg->text();
 511                 } else {
 512                         // This is for future compatibility when using instant commons.
 513                         // So as to not display as ugly a name if a new metadata
 514                         // property is defined that we don't know about
 515                         // (not a major issue since such a property would be collapsed
 516                         // by default).
 517                         wfDebug( __METHOD__ . ' Unknown metadata name: ' . $id . "\n" );
 518                         $name = wfEscapeWikiText( $id );
 519                 }
 520                 $array[$visibility][] = array(
 521                         'id' => "$type-$id",
 522                         'name' => $name,
 523                         'value' => $value
 524                 );
 525         }
 526
 527         /**
 528          * Used instead of getLongDesc if there is no handler registered for file.
 529          *
 530          * @param $file File
 531          * @return string
 532          */
 533         function getShortDesc( $file ) {
 534                 global $wgLang;
 535
 536                 return htmlspecialchars( $wgLang->formatSize( $file->getSize() ) );
 537         }
 538
 539         /**
 540          * Short description. Shown on Special:Search results.
 541          *
 542          * @param $file File
 543          * @return string
 544          */
 545         function getLongDesc( $file ) {
 546                 global $wgLang;
 547
 548                 return wfMessage( 'file-info', htmlspecialchars( $wgLang->formatSize( $file->getSize() ) ),
 549                         $file->getMimeType() )->parse();
 550         }
 551
 552         /**
 553          * Long description. Shown under image on image description page surounded by ().
 554          *
 555          * @param $file File
 556          * @return string
 557          */
 558         static function getGeneralShortDesc( $file ) {
 559                 global $wgLang;
 560
 561                 return $wgLang->formatSize( $file->getSize() );
 562         }
 563
 564         /**
 565          * Used instead of getShortDesc if there is no handler registered for file.
 566          *
 567          * @param $file File
 568          * @return string
 569          */
 570         static function getGeneralLongDesc( $file ) {
 571                 global $wgLang;
 572
 573                 return wfMessage( 'file-info', $wgLang->formatSize( $file->getSize() ),
 574                         $file->getMimeType() )->parse();
 575         }
 576
 577         /**
 578          * Calculate the largest thumbnail width for a given original file size
 579          * such that the thumbnail's height is at most $maxHeight.
 580          * @param $boxWidth Integer Width of the thumbnail box.
 581          * @param $boxHeight Integer Height of the thumbnail box.
 582          * @param $maxHeight Integer Maximum height expected for the thumbnail.
 583          * @return Integer.
 584          */
 585         public static function fitBoxWidth( $boxWidth, $boxHeight, $maxHeight ) {
 586                 $idealWidth = $boxWidth * $maxHeight / $boxHeight;
 587                 $roundedUp = ceil( $idealWidth );
 588                 if ( round( $roundedUp * $boxHeight / $boxWidth ) > $maxHeight ) {
 589                         return floor( $idealWidth );
 590                 } else {
 591                         return $roundedUp;
 592                 }
 593         }
 594
 595         /**
 596          * Shown in file history box on image description page.
 597          *
 598          * @param File $file
 599          * @return String Dimensions
 600          */
 601         function getDimensionsString( $file ) {
 602                 return '';
 603         }
 604
 605         /**
 606          * Modify the parser object post-transform.
 607          *
 608          * This is often used to do $parser->addOutputHook(),
 609          * in order to add some javascript to render a viewer.
 610          * See TimedMediaHandler or OggHandler for an example.
 611          *
 612          * @param Parser $parser
 613          * @param File $file
 614          */
 615         function parserTransformHook( $parser, $file ) {
 616         }
 617
 618         /**
 619          * File validation hook called on upload.
 620          *
 621          * If the file at the given local path is not valid, or its MIME type does not
 622          * match the handler class, a Status object should be returned containing
 623          * relevant errors.
 624          *
 625          * @param string $fileName The local path to the file.
 626          * @return Status object
 627          */
 628         function verifyUpload( $fileName ) {
 629                 return Status::newGood();
 630         }
 631
 632         /**
 633          * Check for zero-sized thumbnails. These can be generated when
 634          * no disk space is available or some other error occurs
 635          *
 636          * @param string $dstPath The location of the suspect file
 637          * @param int $retval Return value of some shell process, file will be deleted if this is non-zero
 638          * @return bool True if removed, false otherwise
 639          */
 640         function removeBadFile( $dstPath, $retval = 0 ) {
 641                 if ( file_exists( $dstPath ) ) {
 642                         $thumbstat = stat( $dstPath );
 643                         if ( $thumbstat['size'] == 0 || $retval != 0 ) {
 644                                 $result = unlink( $dstPath );
 645
 646                                 if ( $result ) {
 647                                         wfDebugLog( 'thumbnail',
 648                                                 sprintf( 'Removing bad %d-byte thumbnail "%s". unlink() succeeded',
 649                                                         $thumbstat['size'], $dstPath ) );
 650                                 } else {
 651                                         wfDebugLog( 'thumbnail',
 652                                                 sprintf( 'Removing bad %d-byte thumbnail "%s". unlink() failed',
 653                                                         $thumbstat['size'], $dstPath ) );
 654                                 }
 655
 656                                 return true;
 657                         }
 658                 }
 659
 660                 return false;
 661         }
 662
 663         /**
 664          * Remove files from the purge list.
 665          *
 666          * This is used by some video handlers to prevent ?action=purge
 667          * from removing a transcoded video, which is expensive to
 668          * regenerate.
 669          *
 670          * @see LocalFile::purgeThumbnails
 671          *
 672          * @param array $files
 673          * @param array $options Purge options. Currently will always be
 674          *  an array with a single key 'forThumbRefresh' set to true.
 675          */
 676         public function filterThumbnailPurgeList( &$files, $options ) {
 677                 // Do nothing
 678         }
 679
 680         /*
 681          * True if the handler can rotate the media
 682          * @since 1.21
 683          * @return bool
 684          */
 685         public static function canRotate() {
 686                 return false;
 687         }
 688
 689         /**
 690          * On supporting image formats, try to read out the low-level orientation
 691          * of the file and return the angle that the file needs to be rotated to
 692          * be viewed.
 693          *
 694          * This information is only useful when manipulating the original file;
 695          * the width and height we normally work with is logical, and will match
 696          * any produced output views.
 697          *
 698          * For files we don't know, we return 0.
 699          *
 700          * @param $file File
 701          * @return int 0, 90, 180 or 270
 702          */
 703         public function getRotation( $file ) {
 704                 return 0;
 705         }
 706
 707         /**
 708          * Log an error that occurred in an external process
 709          *
 710          * Moved from BitmapHandler to MediaHandler with MediaWiki 1.23
 711          *
 712          * @since 1.23
 713          * @param $retval int
 714          * @param $err int
 715          * @param $cmd string
 716          */
 717         protected function logErrorForExternalProcess( $retval, $err, $cmd ) {
 718                 wfDebugLog( 'thumbnail',
 719                         sprintf( 'thumbnail failed on %s: error %d "%s" from "%s"',
 720                                         wfHostname(), $retval, trim( $err ), $cmd ) );
 721         }
 722
 723 }