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