includes/media/Generic.php

   1 <?php
   2 /**
   3  * Media-handling base classes and generic functionality
   4  *
   5  * @file
   6  * @ingroup Media
   7  */
   8
   9 /**
  10  * Base media handler class
  11  *
  12  * @ingroup Media
  13  */
  14 abstract class MediaHandler {
  15         const TRANSFORM_LATER = 1;
  16         const METADATA_GOOD = true;
  17         const METADATA_BAD = false;
  18         const METADATA_COMPATIBLE = 2; // for old but backwards compatible.
  19         /**
  20          * Instance cache
  21          */
  22         static $handlers = array();
  23
  24         /**
  25          * Get a MediaHandler for a given MIME type from the instance cache
  26          *
  27          * @return MediaHandler
  28          */
  29         static function getHandler( $type ) {
  30                 global $wgMediaHandlers;
  31                 if ( !isset( $wgMediaHandlers[$type] ) ) {
  32                         wfDebug( __METHOD__ . ": no handler found for $type.\n");
  33                         return false;
  34                 }
  35                 $class = $wgMediaHandlers[$type];
  36                 if ( !isset( self::$handlers[$class] ) ) {
  37                         self::$handlers[$class] = new $class;
  38                         if ( !self::$handlers[$class]->isEnabled() ) {
  39                                 self::$handlers[$class] = false;
  40                         }
  41                 }
  42                 return self::$handlers[$class];
  43         }
  44
  45         /**
  46          * Get an associative array mapping magic word IDs to parameter names.
  47          * Will be used by the parser to identify parameters.
  48          */
  49         abstract function getParamMap();
  50
  51         /**
  52          * Validate a thumbnail parameter at parse time.
  53          * Return true to accept the parameter, and false to reject it.
  54          * If you return false, the parser will do something quiet and forgiving.
  55          */
  56         abstract function validateParam( $name, $value );
  57
  58         /**
  59          * Merge a parameter array into a string appropriate for inclusion in filenames
  60          */
  61         abstract function makeParamString( $params );
  62
  63         /**
  64          * Parse a param string made with makeParamString back into an array
  65          */
  66         abstract function parseParamString( $str );
  67
  68         /**
  69          * Changes the parameter array as necessary, ready for transformation.
  70          * Should be idempotent.
  71          * Returns false if the parameters are unacceptable and the transform should fail
  72          */
  73         abstract function normaliseParams( $image, &$params );
  74
  75         /**
  76          * Get an image size array like that returned by getimagesize(), or false if it
  77          * can't be determined.
  78          *
  79          * @param $image File: the image object, or false if there isn't one
  80          * @param $path String: the filename
  81          * @return Array
  82          */
  83         abstract function getImageSize( $image, $path );
  84
  85         /**
  86          * Get handler-specific metadata which will be saved in the img_metadata field.
  87          *
  88          * @param $image File: the image object, or false if there isn't one.
  89          *   Warning, File::getPropsFromPath might pass an (object)array() instead (!)
  90          * @param $path String: the filename
  91          * @return String
  92          */
  93         function getMetadata( $image, $path ) { return ''; }
  94
  95         /**
  96         * Get metadata version.
  97         *
  98         * This is not used for validating metadata, this is used for the api when returning
  99         * metadata, since api content formats should stay the same over time, and so things
 100         * using ForiegnApiRepo can keep backwards compatibility
 101         *
 102         * All core media handlers share a common version number, and extensions can
 103         * use the GetMetadataVersion hook to append to the array (they should append a unique
 104         * string so not to get confusing). If there was a media handler named 'foo' with metadata
 105         * version 3 it might add to the end of the array the element 'foo=3'. if the core metadata
 106         * version is 2, the end version string would look like '2;foo=3'.
 107         *
 108         * @return string version string
 109         */
 110         static function getMetadataVersion () {
 111                 $version = Array( '2' ); // core metadata version
 112                 wfRunHooks('GetMetadataVersion', Array(&$version));
 113                 return implode( ';', $version);
 114          }
 115
 116         /**
 117         * Convert metadata version.
 118         *
 119         * By default just returns $metadata, but can be used to allow
 120         * media handlers to convert between metadata versions.
 121         *
 122         * @param $metadata Mixed String or Array metadata array (serialized if string)
 123         * @param $version Integer target version
 124         * @return Array serialized metadata in specified version, or $metadata on fail.
 125         */
 126         function convertMetadataVersion( $metadata, $version = 1 ) {
 127                 if ( !is_array( $metadata ) ) {
 128
 129                         //unserialize to keep return parameter consistent.
 130                         wfSuppressWarnings();
 131                         $ret = unserialize( $metadata );
 132                         wfRestoreWarnings();
 133                         return $ret;
 134                 }
 135                 return $metadata;
 136         }
 137
 138         /**
 139          * Get a string describing the type of metadata, for display purposes.
 140          */
 141         function getMetadataType( $image ) { return false; }
 142
 143         /**
 144          * Check if the metadata string is valid for this handler.
 145          * If it returns MediaHandler::METADATA_BAD (or false), Image
 146          * will reload the metadata from the file and update the database.
 147          * MediaHandler::METADATA_GOOD for if the metadata is a-ok,
 148          * MediaHanlder::METADATA_COMPATIBLE if metadata is old but backwards
 149          * compatible (which may or may not trigger a metadata reload).
 150          */
 151         function isMetadataValid( $image, $metadata ) {
 152                 return self::METADATA_GOOD;
 153         }
 154
 155
 156         /**
 157          * Get a MediaTransformOutput object representing an alternate of the transformed
 158          * output which will call an intermediary thumbnail assist script.
 159          *
 160          * Used when the repository has a thumbnailScriptUrl option configured.
 161          *
 162          * Return false to fall back to the regular getTransform().
 163          */
 164         function getScriptedTransform( $image, $script, $params ) {
 165                 return false;
 166         }
 167
 168         /**
 169          * Get a MediaTransformOutput object representing the transformed output. Does not
 170          * actually do the transform.
 171          *
 172          * @param $image File: the image object
 173          * @param $dstPath String: filesystem destination path
 174          * @param $dstUrl String: Destination URL to use in output HTML
 175          * @param $params Array: Arbitrary set of parameters validated by $this->validateParam()
 176          */
 177         function getTransform( $image, $dstPath, $dstUrl, $params ) {
 178                 return $this->doTransform( $image, $dstPath, $dstUrl, $params, self::TRANSFORM_LATER );
 179         }
 180
 181         /**
 182          * Get a MediaTransformOutput object representing the transformed output. Does the
 183          * transform unless $flags contains self::TRANSFORM_LATER.
 184          *
 185          * @param $image File: the image object
 186          * @param $dstPath String: filesystem destination path
 187          * @param $dstUrl String: destination URL to use in output HTML
 188          * @param $params Array: arbitrary set of parameters validated by $this->validateParam()
 189          * @param $flags Integer: a bitfield, may contain self::TRANSFORM_LATER
 190          */
 191         abstract function doTransform( $image, $dstPath, $dstUrl, $params, $flags = 0 );
 192
 193         /**
 194          * Get the thumbnail extension and MIME type for a given source MIME type
 195          * @return array thumbnail extension and MIME type
 196          */
 197         function getThumbType( $ext, $mime, $params = null ) {
 198                 $magic = MimeMagic::singleton();
 199                 if ( !$ext || $magic->isMatchingExtension( $ext, $mime ) === false ) {
 200                         // The extension is not valid for this mime type and we do
 201                         // recognize the mime type
 202                         $extensions = $magic->getExtensionsForType( $mime );
 203                         if ( $extensions ) {
 204                                 return array( strtok( $extensions, ' ' ), $mime );
 205                         }
 206                 }
 207
 208                 // The extension is correct (true) or the mime type is unknown to
 209                 // MediaWiki (null)
 210                 return array( $ext, $mime );
 211         }
 212
 213         /**
 214          * True if the handled types can be transformed
 215          */
 216         function canRender( $file ) { return true; }
 217         /**
 218          * True if handled types cannot be displayed directly in a browser
 219          * but can be rendered
 220          */
 221         function mustRender( $file ) { return false; }
 222         /**
 223          * True if the type has multi-page capabilities
 224          */
 225         function isMultiPage( $file ) { return false; }
 226         /**
 227          * Page count for a multi-page document, false if unsupported or unknown
 228          */
 229         function pageCount( $file ) { return false; }
 230         /**
 231          * The material is vectorized and thus scaling is lossless
 232          */
 233         function isVectorized( $file ) { return false; }
 234         /**
 235          * False if the handler is disabled for all files
 236          */
 237         function isEnabled() { return true; }
 238
 239         /**
 240          * Get an associative array of page dimensions
 241          * Currently "width" and "height" are understood, but this might be
 242          * expanded in the future.
 243          * Returns false if unknown or if the document is not multi-page.
 244          *
 245          * @param $image File
 246          */
 247         function getPageDimensions( $image, $page ) {
 248                 $gis = $this->getImageSize( $image, $image->getPath() );
 249                 return array(
 250                         'width' => $gis[0],
 251                         'height' => $gis[1]
 252                 );
 253         }
 254
 255         /**
 256          * Generic getter for text layer.
 257          * Currently overloaded by PDF and DjVu handlers
 258          */
 259         function getPageText( $image, $page ) {
 260                 return false;
 261         }
 262
 263         /**
 264          * Get an array structure that looks like this:
 265          *
 266          * array(
 267          *    'visible' => array(
 268          *       'Human-readable name' => 'Human readable value',
 269          *       ...
 270          *    ),
 271          *    'collapsed' => array(
 272          *       'Human-readable name' => 'Human readable value',
 273          *       ...
 274          *    )
 275          * )
 276          * The UI will format this into a table where the visible fields are always
 277          * visible, and the collapsed fields are optionally visible.
 278          *
 279          * The function should return false if there is no metadata to display.
 280          */
 281
 282         /**
 283          * @todo FIXME: I don't really like this interface, it's not very flexible
 284          * I think the media handler should generate HTML instead. It can do
 285          * all the formatting according to some standard. That makes it possible
 286          * to do things like visual indication of grouped and chained streams
 287          * in ogg container files.
 288          */
 289         function formatMetadata( $image ) {
 290                 return false;
 291         }
 292
 293         /** sorts the visible/invisible field.
 294          * Split off from ImageHandler::formatMetadata, as used by more than
 295          * one type of handler.
 296          *
 297          * This is used by the media handlers that use the FormatMetadata class
 298          *
 299          * @param $metadataArray Array metadata array
 300          * @return array for use displaying metadata.
 301          */
 302         function formatMetadataHelper( $metadataArray ) {
 303                  $result = array(
 304                         'visible' => array(),
 305                         'collapsed' => array()
 306                 );
 307
 308                 $formatted = FormatMetadata::getFormattedData( $metadataArray );
 309                 // Sort fields into visible and collapsed
 310                 $visibleFields = $this->visibleMetadataFields();
 311                 foreach ( $formatted as $name => $value ) {
 312                         $tag = strtolower( $name );
 313                         self::addMeta( $result,
 314                                 in_array( $tag, $visibleFields ) ? 'visible' : 'collapsed',
 315                                 'exif',
 316                                 $tag,
 317                                 $value
 318                         );
 319                 }
 320                 return $result;
 321         }
 322
 323         /**
 324          * Get a list of metadata items which should be displayed when
 325          * the metadata table is collapsed.
 326          *
 327          * @return array of strings
 328          * @access protected
 329          */
 330         function visibleMetadataFields() {
 331                 $fields = array();
 332                 $lines = explode( "\n", wfMsgForContent( 'metadata-fields' ) );
 333                 foreach( $lines as $line ) {
 334                         $matches = array();
 335                         if( preg_match( '/^\\*\s*(.*?)\s*$/', $line, $matches ) ) {
 336                                 $fields[] = $matches[1];
 337                         }
 338                 }
 339                 $fields = array_map( 'strtolower', $fields );
 340                 return $fields;
 341         }
 342
 343
 344         /**
 345          * This is used to generate an array element for each metadata value
 346          * That array is then used to generate the table of metadata values
 347          * on the image page
 348          *
 349          * @param &$array Array An array containing elements for each type of visibility
 350          * and each of those elements being an array of metadata items. This function adds
 351          * a value to that array.
 352          * @param $visibility string ('visible' or 'collapsed') if this value is hidden
 353          * by default.
 354          * @param $type String type of metadata tag (currently always 'exif')
 355          * @param $id String the name of the metadata tag (like 'artist' for example).
 356          * its name in the table displayed is the message "$type-$id" (Ex exif-artist ).
 357          * @param $value String thingy goes into a wikitext table; it used to be escaped but
 358          * that was incompatible with previous practise of customized display
 359          * with wikitext formatting via messages such as 'exif-model-value'.
 360          * So the escaping is taken back out, but generally this seems a confusing
 361          * interface.
 362          * @param $param String value to pass to the message for the name of the field
 363          * as $1. Currently this parameter doesn't seem to ever be used.
 364          *
 365          * Note, everything here is passed through the parser later on (!)
 366          */
 367         protected static function addMeta( &$array, $visibility, $type, $id, $value, $param = false ) {
 368                 $msgName = "$type-$id";
 369                 if ( wfEmptyMsg( $msgName ) ) {
 370                         // This is for future compatibility when using instant commons.
 371                         // So as to not display as ugly a name if a new metadata
 372                         // property is defined that we don't know about
 373                         // (not a major issue since such a property would be collapsed
 374                         // by default).
 375                         wfDebug( __METHOD__ . ' Unknown metadata name: ' . $id . "\n" );
 376                         $name = wfEscapeWikiText( $id );
 377                 } else {
 378                         $name = wfMsg( $msgName, $param );
 379                 }
 380                 $array[$visibility][] = array(
 381                         'id' => "$type-$id",
 382                         'name' => $name,
 383                         'value' => $value
 384                 );
 385         }
 386
 387         /**
 388          * @param $file File
 389          * @return string
 390          */
 391         function getShortDesc( $file ) {
 392                 global $wgLang;
 393                 $nbytes = wfMsgExt( 'nbytes', array( 'parsemag', 'escape' ),
 394                         $wgLang->formatNum( $file->getSize() ) );
 395                 return "$nbytes";
 396         }
 397
 398         /**
 399          * @param $file File
 400          * @return string
 401          */
 402         function getLongDesc( $file ) {
 403                 global $wgUser;
 404                 $sk = $wgUser->getSkin();
 405                 return wfMsgExt( 'file-info', 'parseinline',
 406                         $sk->formatSize( $file->getSize() ),
 407                         $file->getMimeType() );
 408         }
 409
 410         /**
 411          * @param $file File
 412          * @return string
 413          */
 414         static function getGeneralShortDesc( $file ) {
 415                 global $wgLang;
 416                 $nbytes = wfMsgExt( 'nbytes', array( 'parsemag', 'escape' ),
 417                         $wgLang->formatNum( $file->getSize() ) );
 418                 return "$nbytes";
 419         }
 420
 421         /**
 422          * @param $file File
 423          * @return string
 424          */
 425         static function getGeneralLongDesc( $file ) {
 426                 global $wgUser;
 427                 $sk = $wgUser->getSkin();
 428                 return wfMsgExt( 'file-info', 'parseinline',
 429                         $sk->formatSize( $file->getSize() ),
 430                         $file->getMimeType() );
 431         }
 432
 433         function getDimensionsString( $file ) {
 434                 return '';
 435         }
 436
 437         /**
 438          * Modify the parser object post-transform
 439          */
 440         function parserTransformHook( $parser, $file ) {}
 441
 442         /**
 443          * File validation hook called on upload.
 444          *
 445          * If the file at the given local path is not valid, or its MIME type does not
 446          * match the handler class, a Status object should be returned containing
 447          * relevant errors.
 448          *
 449          * @param $fileName The local path to the file.
 450          * @return Status object
 451          */
 452         function verifyUpload( $fileName ) {
 453                 return Status::newGood();
 454         }
 455
 456         /**
 457          * Check for zero-sized thumbnails. These can be generated when
 458          * no disk space is available or some other error occurs
 459          *
 460          * @param $dstPath The location of the suspect file
 461          * @param $retval Return value of some shell process, file will be deleted if this is non-zero
 462          * @return true if removed, false otherwise
 463          */
 464         function removeBadFile( $dstPath, $retval = 0 ) {
 465                 if( file_exists( $dstPath ) ) {
 466                         $thumbstat = stat( $dstPath );
 467                         if( $thumbstat['size'] == 0 || $retval != 0 ) {
 468                                 wfDebugLog( 'thumbnail',
 469                                         sprintf( 'Removing bad %d-byte thumbnail "%s"',
 470                                                 $thumbstat['size'], $dstPath ) );
 471                                 unlink( $dstPath );
 472                                 return true;
 473                         }
 474                 }
 475                 return false;
 476         }
 477 }
 478
 479 /**
 480  * Media handler abstract base class for images
 481  *
 482  * @ingroup Media
 483  */
 484 abstract class ImageHandler extends MediaHandler {
 485
 486         /**
 487          * @param $file File
 488          * @return bool
 489          */
 490         function canRender( $file ) {
 491                 return ( $file->getWidth() && $file->getHeight() );
 492         }
 493
 494         function getParamMap() {
 495                 return array( 'img_width' => 'width' );
 496         }
 497
 498         function validateParam( $name, $value ) {
 499                 if ( in_array( $name, array( 'width', 'height' ) ) ) {
 500                         if ( $value <= 0 ) {
 501                                 return false;
 502                         } else {
 503                                 return true;
 504                         }
 505                 } else {
 506                         return false;
 507                 }
 508         }
 509
 510         function makeParamString( $params ) {
 511                 if ( isset( $params['physicalWidth'] ) ) {
 512                         $width = $params['physicalWidth'];
 513                 } elseif ( isset( $params['width'] ) ) {
 514                         $width = $params['width'];
 515                 } else {
 516                         throw new MWException( 'No width specified to '.__METHOD__ );
 517                 }
 518                 # Removed for ProofreadPage
 519                 #$width = intval( $width );
 520                 return "{$width}px";
 521         }
 522
 523         function parseParamString( $str ) {
 524                 $m = false;
 525                 if ( preg_match( '/^(\d+)px$/', $str, $m ) ) {
 526                         return array( 'width' => $m[1] );
 527                 } else {
 528                         return false;
 529                 }
 530         }
 531
 532         function getScriptParams( $params ) {
 533                 return array( 'width' => $params['width'] );
 534         }
 535
 536         /**
 537          * @param $image File
 538          * @param  $params
 539          * @return bool
 540          */
 541         function normaliseParams( $image, &$params ) {
 542                 $mimeType = $image->getMimeType();
 543
 544                 if ( !isset( $params['width'] ) ) {
 545                         return false;
 546                 }
 547
 548                 if ( !isset( $params['page'] ) ) {
 549                         $params['page'] = 1;
 550                 } else  {
 551                         if ( $params['page'] > $image->pageCount() ) {
 552                                 $params['page'] = $image->pageCount();
 553                         }
 554
 555                         if ( $params['page'] < 1 ) {
 556                                 $params['page'] = 1;
 557                         }
 558                 }
 559
 560                 $srcWidth = $image->getWidth( $params['page'] );
 561                 $srcHeight = $image->getHeight( $params['page'] );
 562                 if ( isset( $params['height'] ) && $params['height'] != -1 ) {
 563                         if ( $params['width'] * $srcHeight > $params['height'] * $srcWidth ) {
 564                                 $params['width'] = wfFitBoxWidth( $srcWidth, $srcHeight, $params['height'] );
 565                         }
 566                 }
 567                 $params['height'] = File::scaleHeight( $srcWidth, $srcHeight, $params['width'] );
 568                 if ( !$this->validateThumbParams( $params['width'], $params['height'], $srcWidth, $srcHeight, $mimeType ) ) {
 569                         return false;
 570                 }
 571                 return true;
 572         }
 573
 574         /**
 575          * Get a transform output object without actually doing the transform
 576          */
 577         function getTransform( $image, $dstPath, $dstUrl, $params ) {
 578                 return $this->doTransform( $image, $dstPath, $dstUrl, $params, self::TRANSFORM_LATER );
 579         }
 580
 581         /**
 582          * Validate thumbnail parameters and fill in the correct height
 583          *
 584          * @param $width Integer: specified width (input/output)
 585          * @param $height Integer: height (output only)
 586          * @param $srcWidth Integer: width of the source image
 587          * @param $srcHeight Integer: height of the source image
 588          * @param $mimeType Unused
 589          * @return false to indicate that an error should be returned to the user.
 590          */
 591         function validateThumbParams( &$width, &$height, $srcWidth, $srcHeight, $mimeType ) {
 592                 $width = intval( $width );
 593
 594                 # Sanity check $width
 595                 if( $width <= 0) {
 596                         wfDebug( __METHOD__.": Invalid destination width: $width\n" );
 597                         return false;
 598                 }
 599                 if ( $srcWidth <= 0 ) {
 600                         wfDebug( __METHOD__.": Invalid source width: $srcWidth\n" );
 601                         return false;
 602                 }
 603
 604                 $height = File::scaleHeight( $srcWidth, $srcHeight, $width );
 605                 return true;
 606         }
 607
 608         /**
 609          * @param $image File
 610          * @param  $script
 611          * @param  $params
 612          * @return bool|ThumbnailImage
 613          */
 614         function getScriptedTransform( $image, $script, $params ) {
 615                 if ( !$this->normaliseParams( $image, $params ) ) {
 616                         return false;
 617                 }
 618                 $url = $script . '&' . wfArrayToCGI( $this->getScriptParams( $params ) );
 619                 $page = isset( $params['page'] ) ? $params['page'] : false;
 620
 621                 if( $image->mustRender() || $params['width'] < $image->getWidth() ) {
 622                         return new ThumbnailImage( $image, $url, $params['width'], $params['height'], $page );
 623                 }
 624         }
 625
 626         function getImageSize( $image, $path ) {
 627                 wfSuppressWarnings();
 628                 $gis = getimagesize( $path );
 629                 wfRestoreWarnings();
 630                 return $gis;
 631         }
 632
 633         function isAnimatedImage( $image ) {
 634                 return false;
 635         }
 636
 637         /**
 638          * @param $file File
 639          * @return string
 640          */
 641         function getShortDesc( $file ) {
 642                 global $wgLang;
 643                 $nbytes = wfMsgExt( 'nbytes', array( 'parsemag', 'escape' ),
 644                         $wgLang->formatNum( $file->getSize() ) );
 645                 $widthheight = wfMsgHtml( 'widthheight', $wgLang->formatNum( $file->getWidth() ) ,$wgLang->formatNum( $file->getHeight() ) );
 646
 647                 return "$widthheight ($nbytes)";
 648         }
 649
 650         /**
 651          * @param $file File
 652          * @return string
 653          */
 654         function getLongDesc( $file ) {
 655                 global $wgLang;
 656                 $pages = $file->pageCount();
 657                 if ( $pages === false || $pages <= 1 ) {
 658                         $msg = wfMsgExt('file-info-size', 'parseinline',
 659                                 $wgLang->formatNum( $file->getWidth() ),
 660                                 $wgLang->formatNum( $file->getHeight() ),
 661                                 $wgLang->formatSize( $file->getSize() ),
 662                                 $file->getMimeType() );
 663                 } else {
 664                         $msg = wfMsgExt('file-info-size-pages', 'parseinline',
 665                                 $wgLang->formatNum( $file->getWidth() ),
 666                                 $wgLang->formatNum( $file->getHeight() ),
 667                                 $wgLang->formatSize( $file->getSize() ),
 668                                 $file->getMimeType(),
 669                                 $wgLang->formatNum( $pages ) );
 670                 }
 671                 return $msg;
 672         }
 673
 674         /**
 675          * @param $file File
 676          * @return string
 677          */
 678         function getDimensionsString( $file ) {
 679                 global $wgLang;
 680                 $pages = $file->pageCount();
 681                 $width = $wgLang->formatNum( $file->getWidth() );
 682                 $height = $wgLang->formatNum( $file->getHeight() );
 683                 $pagesFmt = $wgLang->formatNum( $pages );
 684
 685                 if ( $pages > 1 ) {
 686                         return wfMsgExt( 'widthheightpage', 'parsemag', $width, $height, $pagesFmt );
 687                 } else {
 688                         return wfMsg( 'widthheight', $width, $height );
 689                 }
 690         }
 691 }