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          * Max length of error logged by logErrorForExternalProcess()
  36          */
  37         const MAX_ERR_LOG_SIZE = 65535;
  38
  39         /** @var MediaHandler[] Instance cache with array of MediaHandler */
  40         protected static $handlers = array();
  41
  42         /**
  43          * Get a MediaHandler for a given MIME type from the instance cache
  44          *
  45          * @param string $type
  46          *
  47          * @return MediaHandler
  48          */
  49         static function getHandler( $type ) {
  50                 global $wgMediaHandlers;
  51                 if ( !isset( $wgMediaHandlers[$type] ) ) {
  52                         wfDebug( __METHOD__ . ": no handler found for $type.\n" );
  53
  54                         return false;
  55                 }
  56                 $class = $wgMediaHandlers[$type];
  57                 if ( !isset( self::$handlers[$class] ) ) {
  58                         self::$handlers[$class] = new $class;
  59                         if ( !self::$handlers[$class]->isEnabled() ) {
  60                                 self::$handlers[$class] = false;
  61                         }
  62                 }
  63
  64                 return self::$handlers[$class];
  65         }
  66
  67         /**
  68          * Get an associative array mapping magic word IDs to parameter names.
  69          * Will be used by the parser to identify parameters.
  70          */
  71         abstract function getParamMap();
  72
  73         /**
  74          * Validate a thumbnail parameter at parse time.
  75          * Return true to accept the parameter, and false to reject it.
  76          * If you return false, the parser will do something quiet and forgiving.
  77          *
  78          * @param string $name
  79          * @param $value
  80          */
  81         abstract function validateParam( $name, $value );
  82
  83         /**
  84          * Merge a parameter array into a string appropriate for inclusion in filenames
  85          *
  86          * @param array $params Array of parameters that have been through normaliseParams.
  87          * @return string
  88          */
  89         abstract function makeParamString( $params );
  90
  91         /**
  92          * Parse a param string made with makeParamString back into an array
  93          *
  94          * @param string $str The parameter string without file name (e.g. 122px)
  95          * @return array|bool Array of parameters or false on failure.
  96          */
  97         abstract function parseParamString( $str );
  98
  99         /**
 100          * Changes the parameter array as necessary, ready for transformation.
 101          * Should be idempotent.
 102          * Returns false if the parameters are unacceptable and the transform should fail
 103          * @param $image
 104          * @param $params
 105          */
 106         abstract function normaliseParams( $image, &$params );
 107
 108         /**
 109          * Get an image size array like that returned by getimagesize(), or false if it
 110          * can't be determined.
 111          *
 112          * @param File $image The image object, or false if there isn't one
 113          * @param string $path the filename
 114          * @return array Follow the format of PHP getimagesize() internal function. See http://www.php.net/getimagesize
 115          */
 116         abstract function getImageSize( $image, $path );
 117
 118         /**
 119          * Get handler-specific metadata which will be saved in the img_metadata field.
 120          *
 121          * @param File $image The image object, or false if there isn't one.
 122          *   Warning, FSFile::getPropsFromPath might pass an (object)array() instead (!)
 123          * @param string $path The filename
 124          * @return string
 125          */
 126         function getMetadata( $image, $path ) {
 127                 return '';
 128         }
 129
 130         /**
 131          * Get metadata version.
 132          *
 133          * This is not used for validating metadata, this is used for the api when returning
 134          * metadata, since api content formats should stay the same over time, and so things
 135          * using ForiegnApiRepo can keep backwards compatibility
 136          *
 137          * All core media handlers share a common version number, and extensions can
 138          * use the GetMetadataVersion hook to append to the array (they should append a unique
 139          * string so not to get confusing). If there was a media handler named 'foo' with metadata
 140          * version 3 it might add to the end of the array the element 'foo=3'. if the core metadata
 141          * version is 2, the end version string would look like '2;foo=3'.
 142          *
 143          * @return string Version string
 144          */
 145         static function getMetadataVersion() {
 146                 $version = array( '2' ); // core metadata version
 147                 wfRunHooks( 'GetMetadataVersion', array( &$version ) );
 148
 149                 return implode( ';', $version );
 150         }
 151
 152         /**
 153          * Convert metadata version.
 154          *
 155          * By default just returns $metadata, but can be used to allow
 156          * media handlers to convert between metadata versions.
 157          *
 158          * @param mixed|string|array $metadata Metadata array (serialized if string)
 159          * @param int $version Target version
 160          * @return array Serialized metadata in specified version, or $metadata on fail.
 161          */
 162         function convertMetadataVersion( $metadata, $version = 1 ) {
 163                 if ( !is_array( $metadata ) ) {
 164
 165                         //unserialize to keep return parameter consistent.
 166                         wfSuppressWarnings();
 167                         $ret = unserialize( $metadata );
 168                         wfRestoreWarnings();
 169
 170                         return $ret;
 171                 }
 172
 173                 return $metadata;
 174         }
 175
 176         /**
 177          * Get a string describing the type of metadata, for display purposes.
 178          * @param $image
 179          * @return string
 180          */
 181         function getMetadataType( $image ) {
 182                 return false;
 183         }
 184
 185         /**
 186          * Check if the metadata string is valid for this handler.
 187          * If it returns MediaHandler::METADATA_BAD (or false), Image
 188          * will reload the metadata from the file and update the database.
 189          * MediaHandler::METADATA_GOOD for if the metadata is a-ok,
 190          * MediaHanlder::METADATA_COMPATIBLE if metadata is old but backwards
 191          * compatible (which may or may not trigger a metadata reload).
 192          * @return bool
 193          */
 194         function isMetadataValid( $image, $metadata ) {
 195                 return self::METADATA_GOOD;
 196         }
 197
 198         /**
 199          * Get an array of standard (FormatMetadata type) metadata values.
 200          *
 201          * The returned data is largely the same as that from getMetadata(),
 202          * but formatted in a standard, stable, handler-independent way.
 203          * The idea being that some values like ImageDescription or Artist
 204          * are universal and should be retrievable in a handler generic way.
 205          *
 206          * The specific properties are the type of properties that can be
 207          * handled by the FormatMetadata class. These values are exposed to the
 208          * user via the filemetadata parser function.
 209          *
 210          * Details of the response format of this function can be found at
 211          * https://www.mediawiki.org/wiki/Manual:File_metadata_handling
 212          * tl/dr: the response is an associative array of
 213          * properties keyed by name, but the value can be complex. You probably
 214          * want to call one of the FormatMetadata::flatten* functions on the
 215          * property values before using them, or call
 216          * FormatMetadata::getFormattedData() on the full response array, which
 217          * transforms all values into prettified, human-readable text.
 218          *
 219          * Subclasses overriding this function must return a value which is a
 220          * valid API response fragment (all associative array keys are valid
 221          * XML tagnames).
 222          *
 223          * Note, if the file simply has no metadata, but the handler supports
 224          * this interface, it should return an empty array, not false.
 225          *
 226          * @param File $file
 227          *
 228          * @return array|bool False if interface not supported
 229          * @since 1.23
 230          */
 231         public function getCommonMetaArray( File $file ) {
 232                 return false;
 233         }
 234
 235         /**
 236          * Get a MediaTransformOutput object representing an alternate of the transformed
 237          * output which will call an intermediary thumbnail assist script.
 238          *
 239          * Used when the repository has a thumbnailScriptUrl option configured.
 240          *
 241          * Return false to fall back to the regular getTransform().
 242          * @return bool
 243          */
 244         function getScriptedTransform( $image, $script, $params ) {
 245                 return false;
 246         }
 247
 248         /**
 249          * Get a MediaTransformOutput object representing the transformed output. Does not
 250          * actually do the transform.
 251          *
 252          * @param File $image The image object
 253          * @param string $dstPath filesystem destination path
 254          * @param string $dstUrl Destination URL to use in output HTML
 255          * @param array $params Arbitrary set of parameters validated by $this->validateParam()
 256          * @return MediaTransformOutput
 257          */
 258         final function getTransform( $image, $dstPath, $dstUrl, $params ) {
 259                 return $this->doTransform( $image, $dstPath, $dstUrl, $params, self::TRANSFORM_LATER );
 260         }
 261
 262         /**
 263          * Get a MediaTransformOutput object representing the transformed output. Does the
 264          * transform unless $flags contains self::TRANSFORM_LATER.
 265          *
 266          * @param File $image The image object
 267          * @param string $dstPath filesystem destination path
 268          * @param string $dstUrl destination URL to use in output HTML
 269          * @param array $params arbitrary set of parameters validated by $this->validateParam()
 270          *   Note: These parameters have *not* gone through $this->normaliseParams()
 271          * @param int $flags A bitfield, may contain self::TRANSFORM_LATER
 272          *
 273          * @return MediaTransformOutput
 274          */
 275         abstract function doTransform( $image, $dstPath, $dstUrl, $params, $flags = 0 );
 276
 277         /**
 278          * Get the thumbnail extension and MIME type for a given source MIME type
 279          *
 280          * @param string $ext Extension of original file
 281          * @param string $mime Mime type of original file
 282          * @param array $params Handler specific rendering parameters
 283          * @return array thumbnail extension and MIME type
 284          */
 285         function getThumbType( $ext, $mime, $params = null ) {
 286                 $magic = MimeMagic::singleton();
 287                 if ( !$ext || $magic->isMatchingExtension( $ext, $mime ) === false ) {
 288                         // The extension is not valid for this mime type and we do
 289                         // recognize the mime type
 290                         $extensions = $magic->getExtensionsForType( $mime );
 291                         if ( $extensions ) {
 292                                 return array( strtok( $extensions, ' ' ), $mime );
 293                         }
 294                 }
 295
 296                 // The extension is correct (true) or the mime type is unknown to
 297                 // MediaWiki (null)
 298                 return array( $ext, $mime );
 299         }
 300
 301         /**
 302          * Get useful response headers for GET/HEAD requests for a file with the given metadata
 303          * @param mixed $metadata Result of the getMetadata() function of this handler for a file
 304          * @return array
 305          */
 306         public function getStreamHeaders( $metadata ) {
 307                 return array();
 308         }
 309
 310         /**
 311          * True if the handled types can be transformed
 312          * @return bool
 313          */
 314         function canRender( $file ) {
 315                 return true;
 316         }
 317
 318         /**
 319          * True if handled types cannot be displayed directly in a browser
 320          * but can be rendered
 321          * @return bool
 322          */
 323         function mustRender( $file ) {
 324                 return false;
 325         }
 326
 327         /**
 328          * True if the type has multi-page capabilities
 329          * @return bool
 330          */
 331         function isMultiPage( $file ) {
 332                 return false;
 333         }
 334
 335         /**
 336          * Page count for a multi-page document, false if unsupported or unknown
 337          * @return bool
 338          */
 339         function pageCount( $file ) {
 340                 return false;
 341         }
 342
 343         /**
 344          * The material is vectorized and thus scaling is lossless
 345          * @return bool
 346          */
 347         function isVectorized( $file ) {
 348                 return false;
 349         }
 350
 351         /**
 352          * The material is an image, and is animated.
 353          * In particular, video material need not return true.
 354          * @note Before 1.20, this was a method of ImageHandler only
 355          * @return bool
 356          */
 357         function isAnimatedImage( $file ) {
 358                 return false;
 359         }
 360
 361         /**
 362          * If the material is animated, we can animate the thumbnail
 363          * @since 1.20
 364          * @return bool If material is not animated, handler may return any value.
 365          */
 366         function canAnimateThumbnail( $file ) {
 367                 return true;
 368         }
 369
 370         /**
 371          * False if the handler is disabled for all files
 372          * @return bool
 373          */
 374         function isEnabled() {
 375                 return true;
 376         }
 377
 378         /**
 379          * Get an associative array of page dimensions
 380          * Currently "width" and "height" are understood, but this might be
 381          * expanded in the future.
 382          * Returns false if unknown.
 383          *
 384          * It is expected that handlers for paged media (e.g. DjVuHandler)
 385          * will override this method so that it gives the correct results
 386          * for each specific page of the file, using the $page argument.
 387          *
 388          * @note For non-paged media, use getImageSize.
 389          *
 390          * @param File $image
 391          * @param int $page What page to get dimensions of
 392          * @return array|bool
 393          */
 394         function getPageDimensions( $image, $page ) {
 395                 $gis = $this->getImageSize( $image, $image->getLocalRefPath() );
 396                 if ( $gis ) {
 397                         return array(
 398                                 'width' => $gis[0],
 399                                 'height' => $gis[1]
 400                         );
 401                 } else {
 402                         return false;
 403                 }
 404         }
 405
 406         /**
 407          * Generic getter for text layer.
 408          * Currently overloaded by PDF and DjVu handlers
 409          * @return bool
 410          */
 411         function getPageText( $image, $page ) {
 412                 return false;
 413         }
 414
 415         /**
 416          * Get an array structure that looks like this:
 417          *
 418          * array(
 419          *    'visible' => array(
 420          *       'Human-readable name' => 'Human readable value',
 421          *       ...
 422          *    ),
 423          *    'collapsed' => array(
 424          *       'Human-readable name' => 'Human readable value',
 425          *       ...
 426          *    )
 427          * )
 428          * The UI will format this into a table where the visible fields are always
 429          * visible, and the collapsed fields are optionally visible.
 430          *
 431          * The function should return false if there is no metadata to display.
 432          */
 433
 434         /**
 435          * @todo FIXME: I don't really like this interface, it's not very flexible
 436          * I think the media handler should generate HTML instead. It can do
 437          * all the formatting according to some standard. That makes it possible
 438          * to do things like visual indication of grouped and chained streams
 439          * in ogg container files.
 440          * @return bool
 441          */
 442         function formatMetadata( $image ) {
 443                 return false;
 444         }
 445
 446         /** sorts the visible/invisible field.
 447          * Split off from ImageHandler::formatMetadata, as used by more than
 448          * one type of handler.
 449          *
 450          * This is used by the media handlers that use the FormatMetadata class
 451          *
 452          * @param array $metadataArray metadata array
 453          * @return array for use displaying metadata.
 454          */
 455         function formatMetadataHelper( $metadataArray ) {
 456                 $result = array(
 457                         'visible' => array(),
 458                         'collapsed' => array()
 459                 );
 460
 461                 $formatted = FormatMetadata::getFormattedData( $metadataArray );
 462                 // Sort fields into visible and collapsed
 463                 $visibleFields = $this->visibleMetadataFields();
 464                 foreach ( $formatted as $name => $value ) {
 465                         $tag = strtolower( $name );
 466                         self::addMeta( $result,
 467                                 in_array( $tag, $visibleFields ) ? 'visible' : 'collapsed',
 468                                 'exif',
 469                                 $tag,
 470                                 $value
 471                         );
 472                 }
 473
 474                 return $result;
 475         }
 476
 477         /**
 478          * Get a list of metadata items which should be displayed when
 479          * the metadata table is collapsed.
 480          *
 481          * @return array of strings
 482          * @access protected
 483          */
 484         function visibleMetadataFields() {
 485                 return FormatMetadata::getVisibleFields();
 486         }
 487
 488         /**
 489          * This is used to generate an array element for each metadata value
 490          * That array is then used to generate the table of metadata values
 491          * on the image page
 492          *
 493          * @param &$array Array An array containing elements for each type of visibility
 494          * and each of those elements being an array of metadata items. This function adds
 495          * a value to that array.
 496          * @param string $visibility ('visible' or 'collapsed') if this value is hidden
 497          * by default.
 498          * @param string $type type of metadata tag (currently always 'exif')
 499          * @param string $id the name of the metadata tag (like 'artist' for example).
 500          * its name in the table displayed is the message "$type-$id" (Ex exif-artist ).
 501          * @param string $value thingy goes into a wikitext table; it used to be escaped but
 502          * that was incompatible with previous practise of customized display
 503          * with wikitext formatting via messages such as 'exif-model-value'.
 504          * So the escaping is taken back out, but generally this seems a confusing
 505          * interface.
 506          * @param string $param value to pass to the message for the name of the field
 507          * as $1. Currently this parameter doesn't seem to ever be used.
 508          *
 509          * Note, everything here is passed through the parser later on (!)
 510          */
 511         protected static function addMeta( &$array, $visibility, $type, $id, $value, $param = false ) {
 512                 $msg = wfMessage( "$type-$id", $param );
 513                 if ( $msg->exists() ) {
 514                         $name = $msg->text();
 515                 } else {
 516                         // This is for future compatibility when using instant commons.
 517                         // So as to not display as ugly a name if a new metadata
 518                         // property is defined that we don't know about
 519                         // (not a major issue since such a property would be collapsed
 520                         // by default).
 521                         wfDebug( __METHOD__ . ' Unknown metadata name: ' . $id . "\n" );
 522                         $name = wfEscapeWikiText( $id );
 523                 }
 524                 $array[$visibility][] = array(
 525                         'id' => "$type-$id",
 526                         'name' => $name,
 527                         'value' => $value
 528                 );
 529         }
 530
 531         /**
 532          * Used instead of getLongDesc if there is no handler registered for file.
 533          *
 534          * @param $file File
 535          * @return string
 536          */
 537         function getShortDesc( $file ) {
 538                 global $wgLang;
 539
 540                 return htmlspecialchars( $wgLang->formatSize( $file->getSize() ) );
 541         }
 542
 543         /**
 544          * Short description. Shown on Special:Search results.
 545          *
 546          * @param $file File
 547          * @return string
 548          */
 549         function getLongDesc( $file ) {
 550                 global $wgLang;
 551
 552                 return wfMessage( 'file-info', htmlspecialchars( $wgLang->formatSize( $file->getSize() ) ),
 553                         $file->getMimeType() )->parse();
 554         }
 555
 556         /**
 557          * Long description. Shown under image on image description page surounded by ().
 558          *
 559          * @param $file File
 560          * @return string
 561          */
 562         static function getGeneralShortDesc( $file ) {
 563                 global $wgLang;
 564
 565                 return $wgLang->formatSize( $file->getSize() );
 566         }
 567
 568         /**
 569          * Used instead of getShortDesc if there is no handler registered for file.
 570          *
 571          * @param $file File
 572          * @return string
 573          */
 574         static function getGeneralLongDesc( $file ) {
 575                 global $wgLang;
 576
 577                 return wfMessage( 'file-info', $wgLang->formatSize( $file->getSize() ),
 578                         $file->getMimeType() )->parse();
 579         }
 580
 581         /**
 582          * Calculate the largest thumbnail width for a given original file size
 583          * such that the thumbnail's height is at most $maxHeight.
 584          * @param $boxWidth Integer Width of the thumbnail box.
 585          * @param $boxHeight Integer Height of the thumbnail box.
 586          * @param $maxHeight Integer Maximum height expected for the thumbnail.
 587          * @return Integer.
 588          */
 589         public static function fitBoxWidth( $boxWidth, $boxHeight, $maxHeight ) {
 590                 $idealWidth = $boxWidth * $maxHeight / $boxHeight;
 591                 $roundedUp = ceil( $idealWidth );
 592                 if ( round( $roundedUp * $boxHeight / $boxWidth ) > $maxHeight ) {
 593                         return floor( $idealWidth );
 594                 } else {
 595                         return $roundedUp;
 596                 }
 597         }
 598
 599         /**
 600          * Shown in file history box on image description page.
 601          *
 602          * @param File $file
 603          * @return String Dimensions
 604          */
 605         function getDimensionsString( $file ) {
 606                 return '';
 607         }
 608
 609         /**
 610          * Modify the parser object post-transform.
 611          *
 612          * This is often used to do $parser->addOutputHook(),
 613          * in order to add some javascript to render a viewer.
 614          * See TimedMediaHandler or OggHandler for an example.
 615          *
 616          * @param Parser $parser
 617          * @param File $file
 618          */
 619         function parserTransformHook( $parser, $file ) {
 620         }
 621
 622         /**
 623          * File validation hook called on upload.
 624          *
 625          * If the file at the given local path is not valid, or its MIME type does not
 626          * match the handler class, a Status object should be returned containing
 627          * relevant errors.
 628          *
 629          * @param string $fileName The local path to the file.
 630          * @return Status object
 631          */
 632         function verifyUpload( $fileName ) {
 633                 return Status::newGood();
 634         }
 635
 636         /**
 637          * Check for zero-sized thumbnails. These can be generated when
 638          * no disk space is available or some other error occurs
 639          *
 640          * @param string $dstPath The location of the suspect file
 641          * @param int $retval Return value of some shell process, file will be deleted if this is non-zero
 642          * @return bool True if removed, false otherwise
 643          */
 644         function removeBadFile( $dstPath, $retval = 0 ) {
 645                 if ( file_exists( $dstPath ) ) {
 646                         $thumbstat = stat( $dstPath );
 647                         if ( $thumbstat['size'] == 0 || $retval != 0 ) {
 648                                 $result = unlink( $dstPath );
 649
 650                                 if ( $result ) {
 651                                         wfDebugLog( 'thumbnail',
 652                                                 sprintf( 'Removing bad %d-byte thumbnail "%s". unlink() succeeded',
 653                                                         $thumbstat['size'], $dstPath ) );
 654                                 } else {
 655                                         wfDebugLog( 'thumbnail',
 656                                                 sprintf( 'Removing bad %d-byte thumbnail "%s". unlink() failed',
 657                                                         $thumbstat['size'], $dstPath ) );
 658                                 }
 659
 660                                 return true;
 661                         }
 662                 }
 663
 664                 return false;
 665         }
 666
 667         /**
 668          * Remove files from the purge list.
 669          *
 670          * This is used by some video handlers to prevent ?action=purge
 671          * from removing a transcoded video, which is expensive to
 672          * regenerate.
 673          *
 674          * @see LocalFile::purgeThumbnails
 675          *
 676          * @param array $files
 677          * @param array $options Purge options. Currently will always be
 678          *  an array with a single key 'forThumbRefresh' set to true.
 679          */
 680         public function filterThumbnailPurgeList( &$files, $options ) {
 681                 // Do nothing
 682         }
 683
 684         /*
 685          * True if the handler can rotate the media
 686          * @since 1.21
 687          * @return bool
 688          */
 689         public static function canRotate() {
 690                 return false;
 691         }
 692
 693         /**
 694          * On supporting image formats, try to read out the low-level orientation
 695          * of the file and return the angle that the file needs to be rotated to
 696          * be viewed.
 697          *
 698          * This information is only useful when manipulating the original file;
 699          * the width and height we normally work with is logical, and will match
 700          * any produced output views.
 701          *
 702          * For files we don't know, we return 0.
 703          *
 704          * @param $file File
 705          * @return int 0, 90, 180 or 270
 706          */
 707         public function getRotation( $file ) {
 708                 return 0;
 709         }
 710
 711         /**
 712          * Log an error that occurred in an external process
 713          *
 714          * Moved from BitmapHandler to MediaHandler with MediaWiki 1.23
 715          *
 716          * @since 1.23
 717          * @param $retval int
 718          * @param $err string Error reported by command. Anything longer than
 719          * MediaHandler::MAX_ERR_LOG_SIZE is stripped off.
 720          * @param $cmd string
 721          */
 722         protected function logErrorForExternalProcess( $retval, $err, $cmd ) {
 723                 # Keep error output limited (bug 57985)
 724                 $errMessage = trim( substr( $err, 0, self::MAX_ERR_LOG_SIZE ) );
 725
 726                 wfDebugLog( 'thumbnail',
 727                         sprintf( 'thumbnail failed on %s: error %d "%s" from "%s"',
 728                                         wfHostname(), $retval, $errMessage, $cmd ) );
 729         }
 730
 731 }