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 $type string
  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 $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 $params array 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 $str string The parameter string without file name (e.g. 122px)
  90          * @return Array|Boolean 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 $image File: 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 $image File: 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 $metadata Mixed String or Array metadata array (serialized if string)
 153          * @param $version Integer 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          *
 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 a MediaTransformOutput object representing an alternate of the transformed
 192          * output which will call an intermediary thumbnail assist script.
 193          *
 194          * Used when the repository has a thumbnailScriptUrl option configured.
 195          *
 196          * Return false to fall back to the regular getTransform().
 197          * @return bool
 198          */
 199         function getScriptedTransform( $image, $script, $params ) {
 200                 return false;
 201         }
 202
 203         /**
 204          * Get a MediaTransformOutput object representing the transformed output. Does not
 205          * actually do the transform.
 206          *
 207          * @param $image File: the image object
 208          * @param string $dstPath filesystem destination path
 209          * @param string $dstUrl Destination URL to use in output HTML
 210          * @param array $params Arbitrary set of parameters validated by $this->validateParam()
 211          * @return MediaTransformOutput
 212          */
 213         final function getTransform( $image, $dstPath, $dstUrl, $params ) {
 214                 return $this->doTransform( $image, $dstPath, $dstUrl, $params, self::TRANSFORM_LATER );
 215         }
 216
 217         /**
 218          * Get a MediaTransformOutput object representing the transformed output. Does the
 219          * transform unless $flags contains self::TRANSFORM_LATER.
 220          *
 221          * @param $image File: the image object
 222          * @param string $dstPath filesystem destination path
 223          * @param string $dstUrl destination URL to use in output HTML
 224          * @param array $params arbitrary set of parameters validated by $this->validateParam()
 225          *   Note: These parameters have *not* gone through $this->normaliseParams()
 226          * @param $flags Integer: a bitfield, may contain self::TRANSFORM_LATER
 227          *
 228          * @return MediaTransformOutput
 229          */
 230         abstract function doTransform( $image, $dstPath, $dstUrl, $params, $flags = 0 );
 231
 232         /**
 233          * Get the thumbnail extension and MIME type for a given source MIME type
 234          *
 235          * @param String $ext Extension of original file
 236          * @param String $mime Mime type of original file
 237          * @param Array $params Handler specific rendering parameters
 238          * @return array thumbnail extension and MIME type
 239          */
 240         function getThumbType( $ext, $mime, $params = null ) {
 241                 $magic = MimeMagic::singleton();
 242                 if ( !$ext || $magic->isMatchingExtension( $ext, $mime ) === false ) {
 243                         // The extension is not valid for this mime type and we do
 244                         // recognize the mime type
 245                         $extensions = $magic->getExtensionsForType( $mime );
 246                         if ( $extensions ) {
 247                                 return array( strtok( $extensions, ' ' ), $mime );
 248                         }
 249                 }
 250
 251                 // The extension is correct (true) or the mime type is unknown to
 252                 // MediaWiki (null)
 253                 return array( $ext, $mime );
 254         }
 255
 256         /**
 257          * Get useful response headers for GET/HEAD requests for a file with the given metadata
 258          * @param $metadata mixed Result of the getMetadata() function of this handler for a file
 259          * @return Array
 260          */
 261         public function getStreamHeaders( $metadata ) {
 262                 return array();
 263         }
 264
 265         /**
 266          * True if the handled types can be transformed
 267          * @return bool
 268          */
 269         function canRender( $file ) {
 270                 return true;
 271         }
 272
 273         /**
 274          * True if handled types cannot be displayed directly in a browser
 275          * but can be rendered
 276          * @return bool
 277          */
 278         function mustRender( $file ) {
 279                 return false;
 280         }
 281
 282         /**
 283          * True if the type has multi-page capabilities
 284          * @return bool
 285          */
 286         function isMultiPage( $file ) {
 287                 return false;
 288         }
 289
 290         /**
 291          * Page count for a multi-page document, false if unsupported or unknown
 292          * @return bool
 293          */
 294         function pageCount( $file ) {
 295                 return false;
 296         }
 297
 298         /**
 299          * The material is vectorized and thus scaling is lossless
 300          * @return bool
 301          */
 302         function isVectorized( $file ) {
 303                 return false;
 304         }
 305
 306         /**
 307          * The material is an image, and is animated.
 308          * In particular, video material need not return true.
 309          * @note Before 1.20, this was a method of ImageHandler only
 310          * @return bool
 311          */
 312         function isAnimatedImage( $file ) {
 313                 return false;
 314         }
 315
 316         /**
 317          * If the material is animated, we can animate the thumbnail
 318          * @since 1.20
 319          * @return bool If material is not animated, handler may return any value.
 320          */
 321         function canAnimateThumbnail( $file ) {
 322                 return true;
 323         }
 324
 325         /**
 326          * False if the handler is disabled for all files
 327          * @return bool
 328          */
 329         function isEnabled() {
 330                 return true;
 331         }
 332
 333         /**
 334          * Get an associative array of page dimensions
 335          * Currently "width" and "height" are understood, but this might be
 336          * expanded in the future.
 337          * Returns false if unknown.
 338          *
 339          * It is expected that handlers for paged media (e.g. DjVuHandler)
 340          * will override this method so that it gives the correct results
 341          * for each specific page of the file, using the $page argument.
 342          *
 343          * @note For non-paged media, use getImageSize.
 344          *
 345          * @param $image File
 346          * @param $page What page to get dimensions of
 347          * @return array|bool
 348          */
 349         function getPageDimensions( $image, $page ) {
 350                 $gis = $this->getImageSize( $image, $image->getLocalRefPath() );
 351                 if ( $gis ) {
 352                         return array(
 353                                 'width' => $gis[0],
 354                                 'height' => $gis[1]
 355                         );
 356                 } else {
 357                         return false;
 358                 }
 359         }
 360
 361         /**
 362          * Generic getter for text layer.
 363          * Currently overloaded by PDF and DjVu handlers
 364          * @return bool
 365          */
 366         function getPageText( $image, $page ) {
 367                 return false;
 368         }
 369
 370         /**
 371          * Get an array structure that looks like this:
 372          *
 373          * array(
 374          *    'visible' => array(
 375          *       'Human-readable name' => 'Human readable value',
 376          *       ...
 377          *    ),
 378          *    'collapsed' => array(
 379          *       'Human-readable name' => 'Human readable value',
 380          *       ...
 381          *    )
 382          * )
 383          * The UI will format this into a table where the visible fields are always
 384          * visible, and the collapsed fields are optionally visible.
 385          *
 386          * The function should return false if there is no metadata to display.
 387          */
 388
 389         /**
 390          * @todo FIXME: I don't really like this interface, it's not very flexible
 391          * I think the media handler should generate HTML instead. It can do
 392          * all the formatting according to some standard. That makes it possible
 393          * to do things like visual indication of grouped and chained streams
 394          * in ogg container files.
 395          * @return bool
 396          */
 397         function formatMetadata( $image ) {
 398                 return false;
 399         }
 400
 401         /** sorts the visible/invisible field.
 402          * Split off from ImageHandler::formatMetadata, as used by more than
 403          * one type of handler.
 404          *
 405          * This is used by the media handlers that use the FormatMetadata class
 406          *
 407          * @param array $metadataArray metadata array
 408          * @return array for use displaying metadata.
 409          */
 410         function formatMetadataHelper( $metadataArray ) {
 411                 $result = array(
 412                         'visible' => array(),
 413                         'collapsed' => array()
 414                 );
 415
 416                 $formatted = FormatMetadata::getFormattedData( $metadataArray );
 417                 // Sort fields into visible and collapsed
 418                 $visibleFields = $this->visibleMetadataFields();
 419                 foreach ( $formatted as $name => $value ) {
 420                         $tag = strtolower( $name );
 421                         self::addMeta( $result,
 422                                 in_array( $tag, $visibleFields ) ? 'visible' : 'collapsed',
 423                                 'exif',
 424                                 $tag,
 425                                 $value
 426                         );
 427                 }
 428                 return $result;
 429         }
 430
 431         /**
 432          * Get a list of metadata items which should be displayed when
 433          * the metadata table is collapsed.
 434          *
 435          * @return array of strings
 436          * @access protected
 437          */
 438         function visibleMetadataFields() {
 439                 $fields = array();
 440                 $lines = explode( "\n", wfMessage( 'metadata-fields' )->inContentLanguage()->text() );
 441                 foreach ( $lines as $line ) {
 442                         $matches = array();
 443                         if ( preg_match( '/^\\*\s*(.*?)\s*$/', $line, $matches ) ) {
 444                                 $fields[] = $matches[1];
 445                         }
 446                 }
 447                 $fields = array_map( 'strtolower', $fields );
 448                 return $fields;
 449         }
 450
 451         /**
 452          * This is used to generate an array element for each metadata value
 453          * That array is then used to generate the table of metadata values
 454          * on the image page
 455          *
 456          * @param &$array Array An array containing elements for each type of visibility
 457          * and each of those elements being an array of metadata items. This function adds
 458          * a value to that array.
 459          * @param string $visibility ('visible' or 'collapsed') if this value is hidden
 460          * by default.
 461          * @param string $type type of metadata tag (currently always 'exif')
 462          * @param string $id the name of the metadata tag (like 'artist' for example).
 463          * its name in the table displayed is the message "$type-$id" (Ex exif-artist ).
 464          * @param string $value thingy goes into a wikitext table; it used to be escaped but
 465          * that was incompatible with previous practise of customized display
 466          * with wikitext formatting via messages such as 'exif-model-value'.
 467          * So the escaping is taken back out, but generally this seems a confusing
 468          * interface.
 469          * @param string $param value to pass to the message for the name of the field
 470          * as $1. Currently this parameter doesn't seem to ever be used.
 471          *
 472          * Note, everything here is passed through the parser later on (!)
 473          */
 474         protected static function addMeta( &$array, $visibility, $type, $id, $value, $param = false ) {
 475                 $msg = wfMessage( "$type-$id", $param );
 476                 if ( $msg->exists() ) {
 477                         $name = $msg->text();
 478                 } else {
 479                         // This is for future compatibility when using instant commons.
 480                         // So as to not display as ugly a name if a new metadata
 481                         // property is defined that we don't know about
 482                         // (not a major issue since such a property would be collapsed
 483                         // by default).
 484                         wfDebug( __METHOD__ . ' Unknown metadata name: ' . $id . "\n" );
 485                         $name = wfEscapeWikiText( $id );
 486                 }
 487                 $array[$visibility][] = array(
 488                         'id' => "$type-$id",
 489                         'name' => $name,
 490                         'value' => $value
 491                 );
 492         }
 493
 494         /**
 495          * Used instead of getLongDesc if there is no handler registered for file.
 496          *
 497          * @param $file File
 498          * @return string
 499          */
 500         function getShortDesc( $file ) {
 501                 global $wgLang;
 502                 return htmlspecialchars( $wgLang->formatSize( $file->getSize() ) );
 503         }
 504
 505         /**
 506          * Short description. Shown on Special:Search results.
 507          *
 508          * @param $file File
 509          * @return string
 510          */
 511         function getLongDesc( $file ) {
 512                 global $wgLang;
 513                 return wfMessage( 'file-info', htmlspecialchars( $wgLang->formatSize( $file->getSize() ) ),
 514                         $file->getMimeType() )->parse();
 515         }
 516
 517         /**
 518          * Long description. Shown under image on image description page surounded by ().
 519          *
 520          * @param $file File
 521          * @return string
 522          */
 523         static function getGeneralShortDesc( $file ) {
 524                 global $wgLang;
 525                 return $wgLang->formatSize( $file->getSize() );
 526         }
 527
 528         /**
 529          * Used instead of getShortDesc if there is no handler registered for file.
 530          *
 531          * @param $file File
 532          * @return string
 533          */
 534         static function getGeneralLongDesc( $file ) {
 535                 global $wgLang;
 536                 return wfMessage( 'file-info', $wgLang->formatSize( $file->getSize() ),
 537                         $file->getMimeType() )->parse();
 538         }
 539
 540         /**
 541          * Calculate the largest thumbnail width for a given original file size
 542          * such that the thumbnail's height is at most $maxHeight.
 543          * @param $boxWidth Integer Width of the thumbnail box.
 544          * @param $boxHeight Integer Height of the thumbnail box.
 545          * @param $maxHeight Integer Maximum height expected for the thumbnail.
 546          * @return Integer.
 547          */
 548         public static function fitBoxWidth( $boxWidth, $boxHeight, $maxHeight ) {
 549                 $idealWidth = $boxWidth * $maxHeight / $boxHeight;
 550                 $roundedUp = ceil( $idealWidth );
 551                 if ( round( $roundedUp * $boxHeight / $boxWidth ) > $maxHeight ) {
 552                         return floor( $idealWidth );
 553                 } else {
 554                         return $roundedUp;
 555                 }
 556         }
 557
 558         /**
 559          * Shown in file history box on image description page.
 560          *
 561          * @param File $file
 562          * @return String Dimensions
 563          */
 564         function getDimensionsString( $file ) {
 565                 return '';
 566         }
 567
 568         /**
 569          * Modify the parser object post-transform.
 570          *
 571          * This is often used to do $parser->addOutputHook(),
 572          * in order to add some javascript to render a viewer.
 573          * See TimedMediaHandler or OggHandler for an example.
 574          *
 575          * @param Parser $parser
 576          * @param File $file
 577          */
 578         function parserTransformHook( $parser, $file ) {}
 579
 580         /**
 581          * File validation hook called on upload.
 582          *
 583          * If the file at the given local path is not valid, or its MIME type does not
 584          * match the handler class, a Status object should be returned containing
 585          * relevant errors.
 586          *
 587          * @param string $fileName The local path to the file.
 588          * @return Status object
 589          */
 590         function verifyUpload( $fileName ) {
 591                 return Status::newGood();
 592         }
 593
 594         /**
 595          * Check for zero-sized thumbnails. These can be generated when
 596          * no disk space is available or some other error occurs
 597          *
 598          * @param string $dstPath The location of the suspect file
 599          * @param int $retval Return value of some shell process, file will be deleted if this is non-zero
 600          * @return bool True if removed, false otherwise
 601          */
 602         function removeBadFile( $dstPath, $retval = 0 ) {
 603                 if ( file_exists( $dstPath ) ) {
 604                         $thumbstat = stat( $dstPath );
 605                         if ( $thumbstat['size'] == 0 || $retval != 0 ) {
 606                                 $result = unlink( $dstPath );
 607
 608                                 if ( $result ) {
 609                                         wfDebugLog( 'thumbnail',
 610                                                 sprintf( 'Removing bad %d-byte thumbnail "%s". unlink() succeeded',
 611                                                         $thumbstat['size'], $dstPath ) );
 612                                 } else {
 613                                         wfDebugLog( 'thumbnail',
 614                                                 sprintf( 'Removing bad %d-byte thumbnail "%s". unlink() failed',
 615                                                         $thumbstat['size'], $dstPath ) );
 616                                 }
 617                                 return true;
 618                         }
 619                 }
 620                 return false;
 621         }
 622
 623         /**
 624          * Remove files from the purge list.
 625          *
 626          * This is used by some video handlers to prevent ?action=purge
 627          * from removing a transcoded video, which is expensive to
 628          * regenerate.
 629          *
 630          * @see LocalFile::purgeThumbnails
 631          *
 632          * @param array $files
 633          * @param array $options Purge options. Currently will always be
 634          *  an array with a single key 'forThumbRefresh' set to true.
 635          */
 636         public function filterThumbnailPurgeList( &$files, $options ) {
 637                 // Do nothing
 638         }
 639
 640         /*
 641          * True if the handler can rotate the media
 642          * @since 1.21
 643          * @return bool
 644          */
 645         public static function canRotate() {
 646                 return false;
 647         }
 648
 649         /**
 650          * On supporting image formats, try to read out the low-level orientation
 651          * of the file and return the angle that the file needs to be rotated to
 652          * be viewed.
 653          *
 654          * This information is only useful when manipulating the original file;
 655          * the width and height we normally work with is logical, and will match
 656          * any produced output views.
 657          *
 658          * For files we don't know, we return 0.
 659          *
 660          * @param $file File
 661          * @return int 0, 90, 180 or 270
 662          */
 663         public function getRotation( $file ) {
 664                 return 0;
 665         }
 666
 667 }