includes/media/Exif.php

   1 <?php
   2 /**
   3  * Extraction and validation of image metadata.
   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  * @ingroup Media
  21  * @author Ævar Arnfjörð Bjarmason <avarab@gmail.com>
  22  * @copyright Copyright © 2005, Ævar Arnfjörð Bjarmason, 2009 Brent Garber
  23  * @license http://www.gnu.org/copyleft/gpl.html GNU General Public License
  24  * @see http://exif.org/Exif2-2.PDF The Exif 2.2 specification
  25  * @file
  26  */
  27
  28 /**
  29  * Class to extract and validate Exif data from jpeg (and possibly tiff) files.
  30  * @ingroup Media
  31  */
  32 class Exif {
  33
  34         const BYTE      = 1;    //!< An 8-bit (1-byte) unsigned integer.
  35         const ASCII     = 2;    //!< An 8-bit byte containing one 7-bit ASCII code. The final byte is terminated with NULL.
  36         const SHORT     = 3;    //!< A 16-bit (2-byte) unsigned integer.
  37         const LONG      = 4;    //!< A 32-bit (4-byte) unsigned integer.
  38         const RATIONAL  = 5;    //!< Two LONGs. The first LONG is the numerator and the second LONG expresses the denominator
  39         const UNDEFINED = 7;    //!< An 8-bit byte that can take any value depending on the field definition
  40         const SLONG     = 9;    //!< A 32-bit (4-byte) signed integer (2's complement notation),
  41         const SRATIONAL = 10;   //!< Two SLONGs. The first SLONG is the numerator and the second SLONG is the denominator.
  42         const IGNORE    = -1;   // A fake value for things we don't want or don't support.
  43
  44         //@{
  45         /* @var array
  46          * @private
  47          */
  48
  49         /**
  50          * Exif tags grouped by category, the tagname itself is the key and the type
  51          * is the value, in the case of more than one possible value type they are
  52          * separated by commas.
  53          */
  54         var $mExifTags;
  55
  56         /**
  57          * The raw Exif data returned by exif_read_data()
  58          */
  59         var $mRawExifData;
  60
  61         /**
  62          * A Filtered version of $mRawExifData that has been pruned of invalid
  63          * tags and tags that contain content they shouldn't contain according
  64          * to the Exif specification
  65          */
  66         var $mFilteredExifData;
  67
  68         /**
  69          * Filtered and formatted Exif data, see FormatMetadata::getFormattedData()
  70          */
  71         var $mFormattedExifData;
  72
  73         //@}
  74
  75         //@{
  76         /* @var string
  77          * @private
  78          */
  79
  80         /**
  81          * The file being processed
  82          */
  83         var $file;
  84
  85         /**
  86          * The basename of the file being processed
  87          */
  88         var $basename;
  89
  90         /**
  91          * The private log to log to, e.g. 'exif'
  92          */
  93         var $log = false;
  94
  95         /**
  96          * The byte order of the file. Needed because php's
  97          * extension doesn't fully process some obscure props.
  98          */
  99         private $byteOrder;
 100         //@}
 101
 102         /**
 103          * Constructor
 104          *
 105          * @param $file String: filename.
 106          * @param $byteOrder String Type of byte ordering either 'BE' (Big Endian) or 'LE' (Little Endian). Default ''.
 107          * @todo FIXME: The following are broke:
 108          * SubjectArea. Need to test the more obscure tags.
 109          *
 110          * DigitalZoomRatio = 0/0 is rejected. need to determine if that's valid.
 111          * possibly should treat 0/0 = 0. need to read exif spec on that.
 112          */
 113         function __construct( $file, $byteOrder = '' ) {
 114                 /**
 115                  * Page numbers here refer to pages in the EXIF 2.2 standard
 116                  *
 117                  * Note, Exif::UNDEFINED is treated as a string, not as an array of bytes
 118                  * so don't put a count parameter for any UNDEFINED values.
 119                  *
 120                  * @link http://exif.org/Exif2-2.PDF The Exif 2.2 specification
 121                  */
 122                 $this->mExifTags = array(
 123                         # TIFF Rev. 6.0 Attribute Information (p22)
 124                         'IFD0' => array(
 125                                 # Tags relating to image structure
 126                                 'ImageWidth' => Exif::SHORT.','.Exif::LONG,             # Image width
 127                                 'ImageLength' => Exif::SHORT.','.Exif::LONG,            # Image height
 128                                 'BitsPerSample' => array( Exif::SHORT, 3 ),             # Number of bits per component
 129                                 # "When a primary image is JPEG compressed, this designation is not"
 130                                 # "necessary and is omitted." (p23)
 131                                 'Compression' => Exif::SHORT,                           # Compression scheme #p23
 132                                 'PhotometricInterpretation' => Exif::SHORT,             # Pixel composition #p23
 133                                 'Orientation' => Exif::SHORT,                           # Orientation of image #p24
 134                                 'SamplesPerPixel' => Exif::SHORT,                       # Number of components
 135                                 'PlanarConfiguration' => Exif::SHORT,                   # Image data arrangement #p24
 136                                 'YCbCrSubSampling' => array( Exif::SHORT, 2),           # Subsampling ratio of Y to C #p24
 137                                 'YCbCrPositioning' => Exif::SHORT,                      # Y and C positioning #p24-25
 138                                 'XResolution' => Exif::RATIONAL,                        # Image resolution in width direction
 139                                 'YResolution' => Exif::RATIONAL,                        # Image resolution in height direction
 140                                 'ResolutionUnit' => Exif::SHORT,                        # Unit of X and Y resolution #(p26)
 141
 142                                 # Tags relating to recording offset
 143                                 'StripOffsets' => Exif::SHORT.','.Exif::LONG,                   # Image data location
 144                                 'RowsPerStrip' => Exif::SHORT.','.Exif::LONG,                   # Number of rows per strip
 145                                 'StripByteCounts' => Exif::SHORT.','.Exif::LONG,                # Bytes per compressed strip
 146                                 'JPEGInterchangeFormat' => Exif::SHORT.','.Exif::LONG,          # Offset to JPEG SOI
 147                                 'JPEGInterchangeFormatLength' => Exif::SHORT.','.Exif::LONG,    # Bytes of JPEG data
 148
 149                                 # Tags relating to image data characteristics
 150                                 'TransferFunction' => Exif::IGNORE,                     # Transfer function
 151                                 'WhitePoint' => array( Exif::RATIONAL, 2),              # White point chromaticity
 152                                 'PrimaryChromaticities' => array( Exif::RATIONAL, 6),   # Chromaticities of primarities
 153                                 'YCbCrCoefficients' => array( Exif::RATIONAL, 3),       # Color space transformation matrix coefficients #p27
 154                                 'ReferenceBlackWhite' => array( Exif::RATIONAL, 6),     # Pair of black and white reference values
 155
 156                                 # Other tags
 157                                 'DateTime' => Exif::ASCII,                              # File change date and time
 158                                 'ImageDescription' => Exif::ASCII,                      # Image title
 159                                 'Make' => Exif::ASCII,                                  # Image input equipment manufacturer
 160                                 'Model' => Exif::ASCII,                                 # Image input equipment model
 161                                 'Software' => Exif::ASCII,                              # Software used
 162                                 'Artist' => Exif::ASCII,                                # Person who created the image
 163                                 'Copyright' => Exif::ASCII,                             # Copyright holder
 164                         ),
 165
 166                         # Exif IFD Attribute Information (p30-31)
 167                         'EXIF' => array(
 168                                 # TODO: NOTE: Nonexistence of this field is taken to mean nonconformance
 169                                 # to the EXIF 2.1 AND 2.2 standards
 170                                 'ExifVersion' =>  Exif::UNDEFINED,                      # Exif version
 171                                 'FlashPixVersion' => Exif::UNDEFINED,                   # Supported Flashpix version #p32
 172
 173                                 # Tags relating to Image Data Characteristics
 174                                 'ColorSpace' => Exif::SHORT,                            # Color space information #p32
 175
 176                                 # Tags relating to image configuration
 177                                 'ComponentsConfiguration' => Exif::UNDEFINED,                   # Meaning of each component #p33
 178                                 'CompressedBitsPerPixel' => Exif::RATIONAL,                     # Image compression mode
 179                                 'PixelYDimension' => Exif::SHORT.','.Exif::LONG,                # Valid image width
 180                                 'PixelXDimension' => Exif::SHORT.','.Exif::LONG,                # Valid image height
 181
 182                                 # Tags relating to related user information
 183                                 'MakerNote' => Exif::IGNORE,                            # Manufacturer notes
 184                                 'UserComment' => Exif::UNDEFINED,                       # User comments #p34
 185
 186                                 # Tags relating to related file information
 187                                 'RelatedSoundFile' => Exif::ASCII,                      # Related audio file
 188
 189                                 # Tags relating to date and time
 190                                 'DateTimeOriginal' => Exif::ASCII,                      # Date and time of original data generation #p36
 191                                 'DateTimeDigitized' => Exif::ASCII,                     # Date and time of original data generation
 192                                 'SubSecTime' => Exif::ASCII,                            # DateTime subseconds
 193                                 'SubSecTimeOriginal' => Exif::ASCII,                    # DateTimeOriginal subseconds
 194                                 'SubSecTimeDigitized' => Exif::ASCII,                   # DateTimeDigitized subseconds
 195
 196                                 # Tags relating to picture-taking conditions (p31)
 197                                 'ExposureTime' => Exif::RATIONAL,                       # Exposure time
 198                                 'FNumber' => Exif::RATIONAL,                            # F Number
 199                                 'ExposureProgram' => Exif::SHORT,                       # Exposure Program #p38
 200                                 'SpectralSensitivity' => Exif::ASCII,                   # Spectral sensitivity
 201                                 'ISOSpeedRatings' => Exif::SHORT,                       # ISO speed rating
 202                                 'OECF' => Exif::IGNORE,
 203                                 # Optoelectronic conversion factor. Note: We don't have support for this atm.
 204                                 'ShutterSpeedValue' => Exif::SRATIONAL,                 # Shutter speed
 205                                 'ApertureValue' => Exif::RATIONAL,                      # Aperture
 206                                 'BrightnessValue' => Exif::SRATIONAL,                   # Brightness
 207                                 'ExposureBiasValue' => Exif::SRATIONAL,                 # Exposure bias
 208                                 'MaxApertureValue' => Exif::RATIONAL,                   # Maximum land aperture
 209                                 'SubjectDistance' => Exif::RATIONAL,                    # Subject distance
 210                                 'MeteringMode' => Exif::SHORT,                          # Metering mode #p40
 211                                 'LightSource' => Exif::SHORT,                           # Light source #p40-41
 212                                 'Flash' => Exif::SHORT,                                 # Flash #p41-42
 213                                 'FocalLength' => Exif::RATIONAL,                        # Lens focal length
 214                                 'SubjectArea' => array( Exif::SHORT, 4 ),               # Subject area
 215                                 'FlashEnergy' => Exif::RATIONAL,                        # Flash energy
 216                                 'SpatialFrequencyResponse' => Exif::IGNORE,             # Spatial frequency response. Not supported atm.
 217                                 'FocalPlaneXResolution' => Exif::RATIONAL,              # Focal plane X resolution
 218                                 'FocalPlaneYResolution' => Exif::RATIONAL,              # Focal plane Y resolution
 219                                 'FocalPlaneResolutionUnit' => Exif::SHORT,              # Focal plane resolution unit #p46
 220                                 'SubjectLocation' => array( Exif::SHORT, 2),            # Subject location
 221                                 'ExposureIndex' => Exif::RATIONAL,                      # Exposure index
 222                                 'SensingMethod' => Exif::SHORT,                         # Sensing method #p46
 223                                 'FileSource' => Exif::UNDEFINED,                        # File source #p47
 224                                 'SceneType' => Exif::UNDEFINED,                         # Scene type #p47
 225                                 'CFAPattern' => Exif::IGNORE,                           # CFA pattern. not supported atm.
 226                                 'CustomRendered' => Exif::SHORT,                        # Custom image processing #p48
 227                                 'ExposureMode' => Exif::SHORT,                          # Exposure mode #p48
 228                                 'WhiteBalance' => Exif::SHORT,                          # White Balance #p49
 229                                 'DigitalZoomRatio' => Exif::RATIONAL,                   # Digital zoom ration
 230                                 'FocalLengthIn35mmFilm' => Exif::SHORT,                 # Focal length in 35 mm film
 231                                 'SceneCaptureType' => Exif::SHORT,                      # Scene capture type #p49
 232                                 'GainControl' => Exif::SHORT,                           # Scene control #p49-50
 233                                 'Contrast' => Exif::SHORT,                              # Contrast #p50
 234                                 'Saturation' => Exif::SHORT,                            # Saturation #p50
 235                                 'Sharpness' => Exif::SHORT,                             # Sharpness #p50
 236                                 'DeviceSettingDescription' => Exif::IGNORE,
 237                                 # Device settings description. This could maybe be supported. Need to find an
 238                                 # example file that uses this to see if it has stuff of interest in it.
 239                                 'SubjectDistanceRange' => Exif::SHORT,                  # Subject distance range #p51
 240
 241                                 'ImageUniqueID' => Exif::ASCII,                         # Unique image ID
 242                         ),
 243
 244                         # GPS Attribute Information (p52)
 245                         'GPS' => array(
 246                                 'GPSVersion' => Exif::UNDEFINED,
 247                                 # Should be an array of 4 Exif::BYTE's. However php treats it as an undefined
 248                                 # Note exif standard calls this GPSVersionID, but php doesn't like the id suffix
 249                                 'GPSLatitudeRef' => Exif::ASCII,                        # North or South Latitude #p52-53
 250                                 'GPSLatitude' => array( Exif::RATIONAL, 3 ),            # Latitude
 251                                 'GPSLongitudeRef' => Exif::ASCII,                       # East or West Longitude #p53
 252                                 'GPSLongitude' => array( Exif::RATIONAL, 3),            # Longitude
 253                                 'GPSAltitudeRef' => Exif::UNDEFINED,
 254                                 # Altitude reference. Note, the exif standard says this should be an EXIF::Byte,
 255                                 # but php seems to disagree.
 256                                 'GPSAltitude' => Exif::RATIONAL,                        # Altitude
 257                                 'GPSTimeStamp' => array( Exif::RATIONAL, 3),            # GPS time (atomic clock)
 258                                 'GPSSatellites' => Exif::ASCII,                         # Satellites used for measurement
 259                                 'GPSStatus' => Exif::ASCII,                             # Receiver status #p54
 260                                 'GPSMeasureMode' => Exif::ASCII,                        # Measurement mode #p54-55
 261                                 'GPSDOP' => Exif::RATIONAL,                             # Measurement precision
 262                                 'GPSSpeedRef' => Exif::ASCII,                           # Speed unit #p55
 263                                 'GPSSpeed' => Exif::RATIONAL,                           # Speed of GPS receiver
 264                                 'GPSTrackRef' => Exif::ASCII,                           # Reference for direction of movement #p55
 265                                 'GPSTrack' => Exif::RATIONAL,                           # Direction of movement
 266                                 'GPSImgDirectionRef' => Exif::ASCII,                    # Reference for direction of image #p56
 267                                 'GPSImgDirection' => Exif::RATIONAL,                    # Direction of image
 268                                 'GPSMapDatum' => Exif::ASCII,                           # Geodetic survey data used
 269                                 'GPSDestLatitudeRef' => Exif::ASCII,                    # Reference for latitude of destination #p56
 270                                 'GPSDestLatitude' => array( Exif::RATIONAL, 3 ),        # Latitude destination
 271                                 'GPSDestLongitudeRef' => Exif::ASCII,                   # Reference for longitude of destination #p57
 272                                 'GPSDestLongitude' => array( Exif::RATIONAL, 3 ),       # Longitude of destination
 273                                 'GPSDestBearingRef' => Exif::ASCII,                     # Reference for bearing of destination #p57
 274                                 'GPSDestBearing' => Exif::RATIONAL,                     # Bearing of destination
 275                                 'GPSDestDistanceRef' => Exif::ASCII,                    # Reference for distance to destination #p57-58
 276                                 'GPSDestDistance' => Exif::RATIONAL,                    # Distance to destination
 277                                 'GPSProcessingMethod' => Exif::UNDEFINED,               # Name of GPS processing method
 278                                 'GPSAreaInformation' => Exif::UNDEFINED,                # Name of GPS area
 279                                 'GPSDateStamp' => Exif::ASCII,                          # GPS date
 280                                 'GPSDifferential' => Exif::SHORT,                       # GPS differential correction
 281                         ),
 282                 );
 283
 284                 $this->file = $file;
 285                 $this->basename = wfBaseName( $this->file );
 286                 if ( $byteOrder === 'BE' || $byteOrder === 'LE' ) {
 287                         $this->byteOrder = $byteOrder;
 288                 } else {
 289                         // Only give a warning for b/c, since originally we didn't
 290                         // require this. The number of things affected by this is
 291                         // rather small.
 292                         wfWarn( 'Exif class did not have byte order specified. '
 293                          . 'Some properties may be decoded incorrectly.' );
 294                         $this->byteOrder = 'BE'; // BE seems about twice as popular as LE in jpg's.
 295                 }
 296
 297                 $this->debugFile( $this->basename, __FUNCTION__, true );
 298                 if( function_exists( 'exif_read_data' ) ) {
 299                         wfSuppressWarnings();
 300                         $data = exif_read_data( $this->file, 0, true );
 301                         wfRestoreWarnings();
 302                 } else {
 303                         throw new MWException( "Internal error: exif_read_data not present. \$wgShowEXIF may be incorrectly set or not checked by an extension." );
 304                 }
 305                 /**
 306                  * exif_read_data() will return false on invalid input, such as
 307                  * when somebody uploads a file called something.jpeg
 308                  * containing random gibberish.
 309                  */
 310                 $this->mRawExifData = $data ? $data : array();
 311                 $this->makeFilteredData();
 312                 $this->collapseData();
 313                 $this->debugFile( __FUNCTION__, false );
 314         }
 315
 316         /**
 317          * Make $this->mFilteredExifData
 318          */
 319         function makeFilteredData() {
 320                 $this->mFilteredExifData = Array();
 321
 322                 foreach ( array_keys( $this->mRawExifData ) as $section ) {
 323                         if ( !in_array( $section, array_keys( $this->mExifTags ) ) ) {
 324                                 $this->debug( $section , __FUNCTION__, "'$section' is not a valid Exif section" );
 325                                 continue;
 326                         }
 327
 328                         foreach ( array_keys( $this->mRawExifData[$section] ) as $tag ) {
 329                                 if ( !in_array( $tag, array_keys( $this->mExifTags[$section] ) ) ) {
 330                                         $this->debug( $tag, __FUNCTION__, "'$tag' is not a valid tag in '$section'" );
 331                                         continue;
 332                                 }
 333
 334                                 $this->mFilteredExifData[$tag] = $this->mRawExifData[$section][$tag];
 335                                 // This is ok, as the tags in the different sections do not conflict.
 336                                 // except in computed and thumbnail section, which we don't use.
 337
 338                                 $value = $this->mRawExifData[$section][$tag];
 339                                 if ( !$this->validate( $section, $tag, $value ) ) {
 340                                         $this->debug( $value, __FUNCTION__, "'$tag' contained invalid data" );
 341                                         unset( $this->mFilteredExifData[$tag] );
 342                                 }
 343                         }
 344                 }
 345         }
 346
 347         /**
 348         * Collapse some fields together.
 349         * This converts some fields from exif form, to a more friendly form.
 350         * For example GPS latitude to a single number.
 351         *
 352         * The rationale behind this is that we're storing data, not presenting to the user
 353         * For example a longitude is a single number describing how far away you are from
 354         * the prime meridian. Well it might be nice to split it up into minutes and seconds
 355         * for the user, it doesn't really make sense to split a single number into 4 parts
 356         * for storage. (degrees, minutes, second, direction vs single floating point number).
 357         *
 358         * Other things this might do (not really sure if they make sense or not):
 359         * Dates -> mediawiki date format.
 360         * convert values that can be in different units to be in one standardized unit.
 361         *
 362         * As an alternative approach, some of this could be done in the validate phase
 363         * if we make up our own types like Exif::DATE.
 364         */
 365         function collapseData( ) {
 366
 367                 $this->exifGPStoNumber( 'GPSLatitude' );
 368                 $this->exifGPStoNumber( 'GPSDestLatitude' );
 369                 $this->exifGPStoNumber( 'GPSLongitude' );
 370                 $this->exifGPStoNumber( 'GPSDestLongitude' );
 371
 372                 if ( isset( $this->mFilteredExifData['GPSAltitude'] ) && isset( $this->mFilteredExifData['GPSAltitudeRef'] ) ) {
 373                         if ( $this->mFilteredExifData['GPSAltitudeRef'] === "\1" ) {
 374                                 $this->mFilteredExifData['GPSAltitude'] *= - 1;
 375                         }
 376                         unset( $this->mFilteredExifData['GPSAltitudeRef'] );
 377                 }
 378
 379                 $this->exifPropToOrd( 'FileSource' );
 380                 $this->exifPropToOrd( 'SceneType' );
 381
 382                 $this->charCodeString( 'UserComment' );
 383                 $this->charCodeString( 'GPSProcessingMethod');
 384                 $this->charCodeString( 'GPSAreaInformation' );
 385
 386                 //ComponentsConfiguration should really be an array instead of a string...
 387                 //This turns a string of binary numbers into an array of numbers.
 388
 389                 if ( isset ( $this->mFilteredExifData['ComponentsConfiguration'] ) ) {
 390                         $val = $this->mFilteredExifData['ComponentsConfiguration'];
 391                         $ccVals = array();
 392                         for ($i = 0; $i < strlen($val); $i++) {
 393                                 $ccVals[$i] = ord( substr($val, $i, 1) );
 394                         }
 395                         $ccVals['_type'] = 'ol'; //this is for formatting later.
 396                         $this->mFilteredExifData['ComponentsConfiguration'] = $ccVals;
 397                 }
 398
 399                 //GPSVersion(ID) is treated as the wrong type by php exif support.
 400                 //Go through each byte turning it into a version string.
 401                 //For example: "\x02\x02\x00\x00" -> "2.2.0.0"
 402
 403                 //Also change exif tag name from GPSVersion (what php exif thinks it is)
 404                 //to GPSVersionID (what the exif standard thinks it is).
 405
 406                 if ( isset ( $this->mFilteredExifData['GPSVersion'] ) ) {
 407                         $val = $this->mFilteredExifData['GPSVersion'];
 408                         $newVal = '';
 409                         for ($i = 0; $i < strlen($val); $i++) {
 410                                 if ( $i !== 0 ) {
 411                                         $newVal .= '.';
 412                                 }
 413                                 $newVal .= ord( substr($val, $i, 1) );
 414                         }
 415                         if ( $this->byteOrder === 'LE' ) {
 416                                 // Need to reverse the string
 417                                 $newVal2 = '';
 418                                 for ( $i = strlen( $newVal ) - 1; $i >= 0; $i-- ) {
 419                                         $newVal2 .= substr( $newVal, $i, 1 );
 420                                 }
 421                                 $this->mFilteredExifData['GPSVersionID'] = $newVal2;
 422                         } else {
 423                                 $this->mFilteredExifData['GPSVersionID'] = $newVal;
 424                         }
 425                         unset( $this->mFilteredExifData['GPSVersion'] );
 426                 }
 427
 428         }
 429         /**
 430         * Do userComment tags and similar. See pg. 34 of exif standard.
 431         * basically first 8 bytes is charset, rest is value.
 432         * This has not been tested on any shift-JIS strings.
 433         * @param $prop String prop name.
 434         */
 435         private function charCodeString ( $prop ) {
 436                 if ( isset( $this->mFilteredExifData[$prop] ) ) {
 437
 438                         if ( strlen($this->mFilteredExifData[$prop]) <= 8 ) {
 439                                 //invalid. Must be at least 9 bytes long.
 440
 441                                 $this->debug( $this->mFilteredExifData[$prop] , __FUNCTION__, false );
 442                                 unset($this->mFilteredExifData[$prop]);
 443                                 return;
 444                         }
 445                         $charCode = substr( $this->mFilteredExifData[$prop], 0, 8);
 446                         $val = substr( $this->mFilteredExifData[$prop], 8);
 447
 448
 449                         switch ($charCode) {
 450                                 case "\x4A\x49\x53\x00\x00\x00\x00\x00":
 451                                         //JIS
 452                                         $charset = "Shift-JIS";
 453                                         break;
 454                                 case "UNICODE\x00":
 455                                         $charset = "UTF-16" . $this->byteOrder;
 456                                         break;
 457                                 default: //ascii or undefined.
 458                                         $charset = "";
 459                                         break;
 460                         }
 461                         // This could possibly check to see if iconv is really installed
 462                         // or if we're using the compatibility wrapper in globalFunctions.php
 463                         if ($charset) {
 464                                 wfSuppressWarnings();
 465                                 $val = iconv($charset, 'UTF-8//IGNORE', $val);
 466                                 wfRestoreWarnings();
 467                         } else {
 468                                 // if valid utf-8, assume that, otherwise assume windows-1252
 469                                 $valCopy = $val;
 470                                 UtfNormal::quickIsNFCVerify( $valCopy ); //validates $valCopy.
 471                                 if ( $valCopy !== $val ) {
 472                                         wfSuppressWarnings();
 473                                         $val = iconv('Windows-1252', 'UTF-8//IGNORE', $val);
 474                                         wfRestoreWarnings();
 475                                 }
 476                         }
 477
 478                         //trim and check to make sure not only whitespace.
 479                         $val = trim($val);
 480                         if ( strlen( $val ) === 0 ) {
 481                                 //only whitespace.
 482                                 $this->debug( $this->mFilteredExifData[$prop] , __FUNCTION__, "$prop: Is only whitespace" );
 483                                 unset($this->mFilteredExifData[$prop]);
 484                                 return;
 485                         }
 486
 487                         //all's good.
 488                         $this->mFilteredExifData[$prop] = $val;
 489                 }
 490         }
 491         /**
 492         * Convert an Exif::UNDEFINED from a raw binary string
 493         * to its value. This is sometimes needed depending on
 494         * the type of UNDEFINED field
 495         * @param $prop String name of property
 496         */
 497         private function exifPropToOrd ( $prop ) {
 498                 if ( isset( $this->mFilteredExifData[$prop] ) ) {
 499                         $this->mFilteredExifData[$prop] = ord( $this->mFilteredExifData[$prop] );
 500                 }
 501         }
 502         /**
 503         * Convert gps in exif form to a single floating point number
 504         * for example 10 degress 20`40`` S -> -10.34444
 505         * @param String $prop a gps coordinate exif tag name (like GPSLongitude)
 506         */
 507         private function exifGPStoNumber ( $prop ) {
 508                 $loc =& $this->mFilteredExifData[$prop];
 509                 $dir =& $this->mFilteredExifData[$prop . 'Ref'];
 510                 $res = false;
 511
 512                 if ( isset( $loc ) && isset( $dir ) && ( $dir === 'N' || $dir === 'S' || $dir === 'E' || $dir === 'W' ) ) {
 513                         list( $num, $denom ) = explode( '/', $loc[0] );
 514                         $res = $num / $denom;
 515                         list( $num, $denom ) = explode( '/', $loc[1] );
 516                         $res += ( $num / $denom ) * ( 1 / 60 );
 517                         list( $num, $denom ) = explode( '/', $loc[2] );
 518                         $res += ( $num / $denom ) * ( 1 / 3600 );
 519
 520                         if ( $dir === 'S' || $dir === 'W' ) {
 521                                 $res *= - 1; // make negative
 522                         }
 523                 }
 524
 525                 // update the exif records.
 526
 527                 if ( $res !== false ) { // using !== as $res could potentially be 0
 528                         $this->mFilteredExifData[$prop] = $res;
 529                         unset( $this->mFilteredExifData[$prop . 'Ref'] );
 530                 } else { // if invalid
 531                         unset( $this->mFilteredExifData[$prop] );
 532                         unset( $this->mFilteredExifData[$prop . 'Ref'] );
 533                 }
 534         }
 535
 536         /**
 537          * Use FormatMetadata to create formatted values for display to user
 538          * (is this ever used?)
 539          *
 540          * @deprecated since 1.18
 541          */
 542         function makeFormattedData( ) {
 543                 wfDeprecated( __METHOD__, '1.18' );
 544                 $this->mFormattedExifData = FormatMetadata::getFormattedData(
 545                         $this->mFilteredExifData );
 546         }
 547         /**#@-*/
 548
 549         /**#@+
 550          * @return array
 551          */
 552         /**
 553          * Get $this->mRawExifData
 554          * @return array
 555          */
 556         function getData() {
 557                 return $this->mRawExifData;
 558         }
 559
 560         /**
 561          * Get $this->mFilteredExifData
 562          */
 563         function getFilteredData() {
 564                 return $this->mFilteredExifData;
 565         }
 566
 567         /**
 568          * Get $this->mFormattedExifData
 569          *
 570          * This returns the data for display to user.
 571          * Its unclear if this is ever used.
 572          *
 573          * @deprecated since 1.18
 574          */
 575         function getFormattedData() {
 576                 wfDeprecated( __METHOD__, '1.18' );
 577                 if (!$this->mFormattedExifData) {
 578                         $this->makeFormattedData();
 579                 }
 580                 return $this->mFormattedExifData;
 581         }
 582         /**#@-*/
 583
 584         /**
 585          * The version of the output format
 586          *
 587          * Before the actual metadata information is saved in the database we
 588          * strip some of it since we don't want to save things like thumbnails
 589          * which usually accompany Exif data. This value gets saved in the
 590          * database along with the actual Exif data, and if the version in the
 591          * database doesn't equal the value returned by this function the Exif
 592          * data is regenerated.
 593          *
 594          * @return int
 595          */
 596         public static function version() {
 597                 return 2; // We don't need no bloddy constants!
 598         }
 599
 600         /**#@+
 601          * Validates if a tag value is of the type it should be according to the Exif spec
 602          *
 603          * @private
 604          *
 605          * @param $in Mixed: the input value to check
 606          * @return bool
 607          */
 608         private function isByte( $in ) {
 609                 if ( !is_array( $in ) && sprintf('%d', $in) == $in && $in >= 0 && $in <= 255 ) {
 610                         $this->debug( $in, __FUNCTION__, true );
 611                         return true;
 612                 } else {
 613                         $this->debug( $in, __FUNCTION__, false );
 614                         return false;
 615                 }
 616         }
 617
 618         /**
 619          * @param $in
 620          * @return bool
 621          */
 622         private function isASCII( $in ) {
 623                 if ( is_array( $in ) ) {
 624                         return false;
 625                 }
 626
 627                 if ( preg_match( "/[^\x0a\x20-\x7e]/", $in ) ) {
 628                         $this->debug( $in, __FUNCTION__, 'found a character not in our whitelist' );
 629                         return false;
 630                 }
 631
 632                 if ( preg_match( '/^\s*$/', $in ) ) {
 633                         $this->debug( $in, __FUNCTION__, 'input consisted solely of whitespace' );
 634                         return false;
 635                 }
 636
 637                 return true;
 638         }
 639
 640         /**
 641          * @param $in
 642          * @return bool
 643          */
 644         private function isShort( $in ) {
 645                 if ( !is_array( $in ) && sprintf('%d', $in) == $in && $in >= 0 && $in <= 65536 ) {
 646                         $this->debug( $in, __FUNCTION__, true );
 647                         return true;
 648                 } else {
 649                         $this->debug( $in, __FUNCTION__, false );
 650                         return false;
 651                 }
 652         }
 653
 654         /**
 655          * @param $in
 656          * @return bool
 657          */
 658         private function isLong( $in ) {
 659                 if ( !is_array( $in ) && sprintf('%d', $in) == $in && $in >= 0 && $in <= 4294967296 ) {
 660                         $this->debug( $in, __FUNCTION__, true );
 661                         return true;
 662                 } else {
 663                         $this->debug( $in, __FUNCTION__, false );
 664                         return false;
 665                 }
 666         }
 667
 668         /**
 669          * @param $in
 670          * @return bool
 671          */
 672         private function isRational( $in ) {
 673                 $m = array();
 674                 if ( !is_array( $in ) && @preg_match( '/^(\d+)\/(\d+[1-9]|[1-9]\d*)$/', $in, $m ) ) { # Avoid division by zero
 675                         return $this->isLong( $m[1] ) && $this->isLong( $m[2] );
 676                 } else {
 677                         $this->debug( $in, __FUNCTION__, 'fed a non-fraction value' );
 678                         return false;
 679                 }
 680         }
 681
 682         /**
 683          * @param $in
 684          * @return bool
 685          */
 686         private function isUndefined( $in ) {
 687                 $this->debug( $in, __FUNCTION__, true );
 688                 return true;
 689         }
 690
 691         /**
 692          * @param $in
 693          * @return bool
 694          */
 695         private function isSlong( $in ) {
 696                 if ( $this->isLong( abs( $in ) ) ) {
 697                         $this->debug( $in, __FUNCTION__, true );
 698                         return true;
 699                 } else {
 700                         $this->debug( $in, __FUNCTION__, false );
 701                         return false;
 702                 }
 703         }
 704
 705         /**
 706          * @param $in
 707          * @return bool
 708          */
 709         private function isSrational( $in ) {
 710                 $m = array();
 711                 if ( !is_array( $in ) && preg_match( '/^(-?\d+)\/(\d+[1-9]|[1-9]\d*)$/', $in, $m ) ) { # Avoid division by zero
 712                         return $this->isSlong( $m[0] ) && $this->isSlong( $m[1] );
 713                 } else {
 714                         $this->debug( $in, __FUNCTION__, 'fed a non-fraction value' );
 715                         return false;
 716                 }
 717         }
 718         /**#@-*/
 719
 720         /**
 721          * Validates if a tag has a legal value according to the Exif spec
 722          *
 723          * @private
 724          * @param $section String: section where tag is located.
 725          * @param $tag String: the tag to check.
 726          * @param $val Mixed: the value of the tag.
 727          * @param $recursive Boolean: true if called recursively for array types.
 728          * @return bool
 729          */
 730         private function validate( $section, $tag, $val, $recursive = false ) {
 731                 $debug = "tag is '$tag'";
 732                 $etype = $this->mExifTags[$section][$tag];
 733                 $ecount = 1;
 734                 if( is_array( $etype ) ) {
 735                         list( $etype, $ecount ) = $etype;
 736                         if ( $recursive )
 737                                 $ecount = 1; // checking individual elements
 738                 }
 739                 $count = count( $val );
 740                 if( $ecount != $count ) {
 741                         $this->debug( $val, __FUNCTION__, "Expected $ecount elements for $tag but got $count" );
 742                         return false;
 743                 }
 744                 if( $count > 1 ) {
 745                         foreach( $val as $v ) {
 746                                 if( !$this->validate( $section, $tag, $v, true ) ) {
 747                                         return false;
 748                                 }
 749                         }
 750                         return true;
 751                 }
 752                 // Does not work if not typecast
 753                 switch( (string)$etype ) {
 754                         case (string)Exif::BYTE:
 755                                 $this->debug( $val, __FUNCTION__, $debug );
 756                                 return $this->isByte( $val );
 757                         case (string)Exif::ASCII:
 758                                 $this->debug( $val, __FUNCTION__, $debug );
 759                                 return $this->isASCII( $val );
 760                         case (string)Exif::SHORT:
 761                                 $this->debug( $val, __FUNCTION__, $debug );
 762                                 return $this->isShort( $val );
 763                         case (string)Exif::LONG:
 764                                 $this->debug( $val, __FUNCTION__, $debug );
 765                                 return $this->isLong( $val );
 766                         case (string)Exif::RATIONAL:
 767                                 $this->debug( $val, __FUNCTION__, $debug );
 768                                 return $this->isRational( $val );
 769                         case (string)Exif::UNDEFINED:
 770                                 $this->debug( $val, __FUNCTION__, $debug );
 771                                 return $this->isUndefined( $val );
 772                         case (string)Exif::SLONG:
 773                                 $this->debug( $val, __FUNCTION__, $debug );
 774                                 return $this->isSlong( $val );
 775                         case (string)Exif::SRATIONAL:
 776                                 $this->debug( $val, __FUNCTION__, $debug );
 777                                 return $this->isSrational( $val );
 778                         case (string)Exif::SHORT.','.Exif::LONG:
 779                                 $this->debug( $val, __FUNCTION__, $debug );
 780                                 return $this->isShort( $val ) || $this->isLong( $val );
 781                         case (string)Exif::IGNORE:
 782                                 $this->debug( $val, __FUNCTION__, $debug );
 783                                 return false;
 784                         default:
 785                                 $this->debug( $val, __FUNCTION__, "The tag '$tag' is unknown" );
 786                                 return false;
 787                 }
 788         }
 789
 790         /**
 791          * Convenience function for debugging output
 792          *
 793          * @private
 794          *
 795          * @param $in Mixed:
 796          * @param $fname String:
 797          * @param $action Mixed: , default NULL.
 798          */
 799         private function debug( $in, $fname, $action = null ) {
 800                 if ( !$this->log ) {
 801                         return;
 802                 }
 803                 $type = gettype( $in );
 804                 $class = ucfirst( __CLASS__ );
 805                 if ( $type === 'array' ) {
 806                         $in = print_r( $in, true );
 807                 }
 808
 809                 if ( $action === true ) {
 810                         wfDebugLog( $this->log, "$class::$fname: accepted: '$in' (type: $type)\n");
 811                 } elseif ( $action === false ) {
 812                         wfDebugLog( $this->log, "$class::$fname: rejected: '$in' (type: $type)\n");
 813                 } elseif ( $action === null ) {
 814                         wfDebugLog( $this->log, "$class::$fname: input was: '$in' (type: $type)\n");
 815                 } else {
 816                         wfDebugLog( $this->log, "$class::$fname: $action (type: $type; content: '$in')\n");
 817                 }
 818         }
 819
 820         /**
 821          * Convenience function for debugging output
 822          *
 823          * @private
 824          *
 825          * @param $fname String: the name of the function calling this function
 826          * @param $io Boolean: Specify whether we're beginning or ending
 827          */
 828         private function debugFile( $fname, $io ) {
 829                 if ( !$this->log ) {
 830                         return;
 831                 }
 832                 $class = ucfirst( __CLASS__ );
 833                 if ( $io ) {
 834                         wfDebugLog( $this->log, "$class::$fname: begin processing: '{$this->basename}'\n" );
 835                 } else {
 836                         wfDebugLog( $this->log, "$class::$fname: end processing: '{$this->basename}'\n" );
 837                 }
 838         }
 839 }
 840